Cross-linking of collagen-based materials · CROSS-LINKING OF COLLAGEN-BASED MATERIALS PROEFSCHRIFT ter verkrijging van de graad van doctor aan de Universiteit Twente, op gezag van

CROSS-LINKING OF

COLLAGEN-BASED

MATERIALS

PROEFSCHRIFT

ter verkrijging vande graad van doctor aan de Universiteit Twente,

op gezag van de rector magnificus,prof. dr. F.A. van Vught,

volgens besluit van het College voor Promotiesin het openbaar te verdedigen

op vrijdag 6 november 1998 te 15.00 uur.

door

Raymond Zeeman

geboren op 24 juni 1970te Hengelo (O)

Dit proefschrift is goedgekeurd door:

Promotor: Prof. Dr. J. Feijen

Assistent Promotor: Dr. P.J. Dijkstra

For millions of years mankind lived just like animals.Then something happened which unleashed the power of our imagination.We learned to talk.…………..

Keep Talking - Pink FloydWords told by Stephen Hawking

Aan mijn ouders

Cross-linking of collagen-based materials/ Raymond ZeemanThesis University of Twente, Enschede, The Netherlands.With references- With summary in English, met samenvatting in het NederlandsISBN: 90 365 1207 7

Subject headings: collagen/ cross-linking/ heart valves/calcification/ biocompatibility

The research described in this thesis was financed by Medtronic Bakken Research Center B.V.,Maastricht, The Netherlands.

Een financiële bijdrage aan de drukkosten van dit proefschrift werd verleend door:Medtronic Bakken Research Center B.V., Maastricht, Nederland

Cover:Designed by: Dick Lelyveld, Hengelo (O), The Netherlands

middle: A space-filling computer model of the collagen triple-helixright under: A stentless porcine aortic heart valve bioprosthesis

(Freestyle, Medtronic)

© R. Zeeman, 1998

Press: FEBODRUK BV., Enschede, The Netherlands, 1998.All rights reserved.

Voorwoord

Gedurende de afgelopen 4 jaar heb ik met zeer veel plezier gewerkt aan het onderzoek datbeschreven staat in dit proefschrift. In dit onderzoek, in samenwerking met Medtronic BakkenResearch Center uit Maastricht, werd gewerkt aan het ontwikkelen van nieuwe crosslinkmethodenvoor collagene materialen, de hartkleppen uit varkens in het bijzonder. Dit onderzoek heb iknatuurlijk niet alleen gedaan en ik wil middels dit voorwoord een (groot) aantal mensen bedanken.Ten eerste wil ik mijn promotor, Prof Dr. Jan Feijen bedanken voor de mogelijkheid die ik kreegom bij hem te promoveren. Daar ik voor mijn promotie was afgestudeerd in de rubberwereld wasik een behoorlijke leek in de wereld van de eiwitten. Echter de term crosslinken was voor mij hetaanknopingspunt ten aanzien van mijn afstudeerwerk. Jan, ik waardeer de tijd die je hebtgestoken in het kritisch lezen van mijn soms nogal lange hoofdstukken.Daarnaast is er ook veel dank verschuldigd aan mijn assistent-promotor, Dr. Piet Dijkstra, dieondanks de soms wat warrige eerste versies van mij toch steeds weer tot de hoofdlijn van hetverhaal kon komen. Beste Piet, ik vond het prettig om met jou te hebben kunnen samenwerken enook tijdens je verblijf in Amherst (MA, USA) konden we de hoofdstukken snel en kritischdoornemen.Twee mensen van Medtronic Bakken Research Center moet ik speciaal bedanken. Ten eerste isdat mijn referent Dr. Marc Hendriks en ten tweede is dat Pat Cahalan. De altijd positievewaarderingen van jullie over de voortgang van mijn werk heb ik zeer gewaardeerd. Daarnaastheeft Marc mij vooral de eerste periode veel steun gegeven in het opstarten van hetcollageenproject. In een later stadium onderhielden we onze vele communicatie via de e-mailtjes. Iwould like to thank Pat Cahalan for his interest in my work. I appreciated the times that we haveworked together in the lab doing a large amount of cross-link experiments. I will never forget ourvisits to the MacDonalds after we had collected many heart valves from the slaughterhouse. Thankyou!Daarnaast werden de implantatietesten zeer zorgvuldig uitgevoerd aan de RijksuniversiteitGroningen door Dr. Pauline van Wachem, Dr. Marja van Luyn en Ing. Linda Brouwer. Ik hebjullie enthousiasme en jullie expertise op dit gebied als zeer leerzaam en prettig ervaren en hoopnog een tijdje met jullie te kunnen samenwerken. De discussie bij jullie resulteerde in een enorme'brainstorm' waardoor we weer voor vele jaren aan werk hadden gecreëerd.Ich möchte mich sehr herzlich bei Herrn Dr. Hoffmeister von der Firma Premium Fleisch Emsland(Lingen, Deutschland) für die Gastfreundschaft bedanken. Benötigte ich einmal wieder eine ReiheHerzklappen, konnte ich jederzeit im Schlachthaus anklopfen. Auch bei den Mitarbeiterinnen undMitarbeitern des Schlachthauses sowie des Labors möchte ich mich für deren Hilfsbereitschaftherzlich bedanken.Daarnaast was een andere 'collageenboy', Jeroen Pieper die momenteel werkzaam is als AIO aande Katholieke Universiteit Nijmegen, erg behulpzaam bij het uitvoeren van meerdereaminozuuranalyses. Bedankt Pieper.De studenten, Dirkje Bloemberg (HLO-Emmen), Amos Rudelsheim, Patrick Bos en Tom Uitslag,die tijdens mijn promotie allen met veel overgave gewerkt hebben aan het crosslinken van

zeemleder moet ik erg bedanken. Ondanks dat er relatief weinig van jullie werk direct in ditproefschrift staat heb ik veel aan jullie werk gehad en heb ik een geweldige tijd met jullie beleefd.Naast al deze mensen die in mijn projekt en onderzoek betrokken waren mag ik zeker mijnkamergenoten Peter Schrooyen alias Peterke (mijn ene paranimf), Richard van der Walle en Ypevan der Zijpp niet vergeten. De altijd opgewekte en vrolijke sfeer die de gehele dag op de kamerheerste, werkte zeker positief mee aan het voltooien van het werk. Ondanks dat we allen veel opde computer werkten (al dan niet e-mail, internet, spelletjes of echt werk), hebben we elkaar nooitin de weg gezeten.Ik wil in het bijzonder de overige 3 leden van de ‘4 musketiers’ namelijk Mark Olde Riekerink,alias MOR (mijn andere paranimf), Luuk Groenewoud en Robin Winters bedanken voor debroodnodige uitstapjes en feestjes. De termen Big Mark, Popov, Triviant, Quatro, Trans-SiberianExpres en schutting staan voor ons min of meer synoniem aan elkaar. Ik hoop dat de ‘4musketiers’ nog jaren stand kunnen houden.Het partijtje 'stoom-afblazen' oftewel tafeltennis tijdens de middagpauze was essentieel om debatterij weer op te laden. Ik wil vooral Jeroen Bezemer, Mark Olde Riekerink, Henk Stapert, WimStevels en Coen van Delden bedanken dat ze mijn tafeltennisniveau hielpen verbeteren. De eersteplaats was vaak een utopie voor mij maar het bleef een aangename uitdaging. Ook de voetbal- envolleybal partijtjes in de middag waren een welkome afwisseling voor het chemisch werk op hetlab. Ondanks dat we altijd het beste team waren in de reguliere competities konden we het nooitvoltooien in de befaamde play-offs.Het partijtje squashen op de welbekende woensdagavond was gewoon geweldig. Ik wil voor dezezeer inspannende avonden vooral Chantal van Dinteren, Alma Kuijpers, André Klomp, IwanNoordman en Wilco Zuiderduin bedanken.Naast het sporten werden ook diverse andere sociale evenementen georganiseerd zoals etentjes,barbeques, bierproefavonden, en minitriatlons. Daarnaast waren er de koffiepauzes tussen 10.00 en10.30 uur altijd gezellig. Voor deze sociale activiteiten wil ik naast bovengenoemde personen ookPaulien Harmsen, Marcel Wissink, Gert Bos (commissie triatlon), Krista Bouwma (commissietriatlon), Margie Topp, Miechel Zweers, Dirk Grijpma, Gerard Engbers, Leon Terlingen, RogerSeijger, Edwin van der Linden, Niels van der Aar, Mark Ankoné en Rob Lammertink bedanken.Verder mag ik vooral John Kooiker en Karin Hendriks niet vergeten. John was altijd behulpzaambij technische en andere niet-wetenschappelijke zaken op het lab, terwijl Karin voor alleadministratieve zaken en andere regeldingetjes zorgde. Ik kon bij haar met nog zo’n vreemdprobleem komen maar ze vond er altijd een adequaat antwoord op. Bedankt!

Tenslotte wil ik mijn familie en vooral mijn ouders bedanken voor hun interesse in mijn promotie-onderzoek en de nuchtere kijk op de hedendaagse zaken. Een bezoek aan hen bracht mij weereven met beide benen terug op de grond in de ‘echte wereld’.

- 1 -

Contents

Chapter 1 General introduction 3

Chapter 2 Cross-linking and calcification of collagen-based materials 9

Chapter 3 Cross-linking and modification of dermal sheep collagen using

1.4-butanediol diglycidyl ether 35

Chapter 4 The kinetics of 1,4-butanediol diglycidyl ether cross-linked

dermal sheep collagen 53

Chapter 5 In-vitro degradation of dermal sheep collagen cross-linked with

1,4-butanediol diglycidyl ether 73

Chapter 6 Successive epoxy and carbodiimide cross-linking of dermal sheep

collagen 91

Chapter 7 Characterization and biocompatibility of epoxy cross-linked dermal

sheep collagen 111

Chapter 8 Cross-linking and modification of porcine aortic heart valves 125

Chapter 9 Properties of cross-linked porcine aortic heart valves 149

Chapter 10 In-vivo behavior of cross-linked porcine aortic leaflets and walls

Effect of cross-linking method and CHAPS/SDS extraction method 169

Summary 191

Samenvatting 195

Curriculum Vitae 199

General ntroduct on

- 3 -

Chapter 1

General Introduction

Aortic heart valvesThe aortic valve has a fascinating structure, which is composed of three membraneous leafletswhich are anchored in the aortic wall, and three sinuses. The leaflets are the most mobile parts ofthe valve and the sinuses are the cavities behind the leaflets. Aortic heart valves permit blood flowfrom the left ventricle to the aorta but prevent backflow in the left ventricle. A valve opens andcloses about 103,000 times a day, which demands very good dynamics of the valve. The valvesustains variable pressures, undergoes complete reversal of curvature, and is subjected to a largeamount of flexion for billions of cycles and still survives. No man-made structure can meet thisachievement [1]. Leaflets are composed of the structural proteins collagen and elastin, and areclosely connected to surrounding proteoglycans and glycosaminoglycans, thus creating a uniquestructure which can easily sustain millions of cycles. Collagen, which is the most abundantcomponent of the matrix, is important for maintenance of structural integrity and function of thevalves [1-3].A variety of pathological processes can lead to heart valve malfunction. This is usually associatedwith degenerative changes of the tissue substance and requires surgical correction or replacementwith a prosthesis. Heart valve prostheses have been used successfully since 1960. The mostcommonly used basic types of prosthetic valves at present are mechanical and tissue valves. Onemajor disadvantage with the use of mechanical valves is the need for continuous anticoagulationtherapy to minimize the risk of thrombosis, whereas tissue valves can be used withoutanticoagulants. Tissue valves are constructed from porcine aortic valves or bovine pericardiumand are treated with glutaraldehyde to introduce cross-links that stabilize the valvular structuralproteins and make them more durable. It has been shown that cross-linking decreases theantigenicity of collagen and improves the resistance towards enzymatic degradation [2, 4].Almost 30 years after the introduction of valvular prostheses, between 100,000 and 200,000patients worldwide are receiving cardiac valve substitutes each year. Mechanical valves are themost commonly used, and form about 70 % of the market, although large variations exist betweencountries [5]. The other 30 % comprise tissue valves, of which the porcine bioprosthesis hasbecome a commonly accepted device for heart valve replacement [6].The long-term success of glutaraldehyde-treated bioprostheses is limited by the tendency ofdevitalized tissue leaflets to undergo degeneration, primarily calcification and/or structuralbreakdown [7, 8]. The factors and the mechanisms which are responsible for the induction and theenhancement of calcium phosphate crystal formation and growth are not fully understood, butboth the glutaraldehyde cross-linking and the presence of foreign proteins and cells in the tissueappear to play an important role in this process [9, 10].

Chapter 1

- 4 -

Different strategies have been applied to mitigate calcification of tissue heart valves. First, newcross-linking methods others than the widely applied glutaraldehyde method were developed [11-17]. Second, glutaraldehyde treated tissue was modified with for example amine-containingcomponents [18-20] and finally cells and proteins were removed from the cross-linked matrix [21,22]. Another approach was to release agents such as phophonates and trivalent metal ions in acontrolled way from an implanted matrix [23, 24].

CollagenCollagen, in the form of fibers, represents the single most abundant animal protein in mammals.One of the earliest chemical modification of collagen to use it as a (bio)material is associated withleather tanning. During the last 20 years, increased interest has emerged in the use of collagen andcollagen-containing tissues in medical devices. Two approaches can be followed in thisconnection. One involves the use of collagen-rich tissues, usually structural in nature, that aretreated chemically in order to transform them into implantable prostheses. Examples are heartvalves, vascular grafts, tendons, ligaments, and pericardium. Another approach involves the use ofpurified collagen obtained from animal tissue, processed in a variety of ways to generate a largenumber of products that not only have applications in the medical field, but also in themanufacturing of cosmetics. Collagen can be used in the form of native soluble collagen,enzymatically processed native collagen, soluble collagen of reconstituted fibers and so on.Products are used as dermal implants, implantable drug delivery vehicles, sponges, tubes andsuture [25-27].

ObjectiveThe objective of the study described in this thesis is to develop methods for the cross-linking ofcollagenous materials such as aortic heart valves, which provide improved materials compared toglutaraldehyde cross-linked collagenous materials. The cross-linked materials should bebiocompatible and should not calcify in-vivo. Furthermore, the mechanical properties should becomparable to those of the native tissue.In order to get more insight in the chemistry of cross-linking and to find correlations between thecross-linking method and the material properties such as the stability towards enzymaticdegradation, swelling, mechanical behavior, biocompatibility and the tendency to calcify, a modeltissue, dermal sheep collagen (DSC) was used [13, 14, 28]. DSC, which consists of almost 100 %fibrous collagen type I, contains about 32 amine and 120 carboxylic acid groups per 1000 aminoacids. Cross-linking methods which were successfully applied on DSC were evaluated for their usein stabilization of porcine aortic heart valves.

Survey of this thesisA literature overview of collagen and porcine aortic heart valves is given in chapter 2. Thestructure of collagen and the methods to cross-link the collagen compound in biological tissues aresummarized. Furthermore, a brief description of the porcine aortic heart valve is given followed bya survey of calcification of bioprostheses and the anti-mineralization techniques which have beenapplied.

General ntroduct on

- 5 -

Initial studies have been directed towards the cross-linking of DSC, the model tissue used, withthe bisepoxy compound, 1,4-butanediol diglycidyl ether (BDDGE). Literature data showed thatsome epoxy compounds effectively cross-link collagen based materials and inhibit calcification[12, 29, 30]. Because of the multi-functionality of the epoxy compounds used, the cross-linkedmaterials are generally ill-defined with respect to their chemical modification. The influence of thereaction conditions such as the reagent concentration, the reaction time, the solution pH and thereaction temperature on the cross-linking rate and density have been studied and described inchapter 3 [31]. Special attention has been given to the influence of the solution pH, becausedifferent cross-linking mechanisms occur at acidic and basic reaction conditions. The cross-linkdensity of DSC has been related to the increase in shrinkage or denaturation temperature and thedecrease in amine groups. Furthermore, DSC was modified with the monofunctional epoxycompound, glycidyl isopropyl ether (PGE). In chapter 4, a kinetic model for the cross-linkingreaction of DSC with BDDGE or for the reaction of DSC with PGE is proposed. Experimentaldata obtained in chapter 3 were used to fit the kinetic equation. This model enables one to predictthe material properties after cross-linking or modification and consequently the cross-linkingprocedure can be optimized to result in well-defined and stabilized collagen materials.Chapter 5 deals with the in-vitro degradation of BDDGE cross-linked collagen using eitherbacterial collagenase or pronase. In addition, the effect of degradation by enzymes on themechanical properties of the materials is evaluated. A successive epoxy and carbodiimide cross-linking method is described in chapter 6. The relations between the cross-linking methods and thematerial properties have been investigated in detail.The in-vivo biocompatibility and calcification of the materials mentioned in chapter 6 have beenstudied by subcutaneous implantation in male Albino Oxford rats and are described in chapter 7[32].Cross-linking methods that have been successfully applied for DSC were transferred to porcineaortic heart valves and are described in the second part of this thesis. The leaflets and the aorticwall are separately characterized and the cross-linking procedures are optimized for thesematerials. The results are discussed in chapter 8. The effect of the cross-linking procedure on thetissue properties such as the swelling and the in-vitro stability, are evaluated and described inchapter 9. Extraction methods have been developed to remove cellular elements and proteins fromthe (non)-cross-linked matrix. These methods provide additional information about the cross-linking reaction and the components which are involved.Finally, the in-vivo behavior and calcification of epoxy cross-linked valves after 8 weeks ofsubcutaneous implantation in weanling rats have been investigated and are compared toglutaraldehyde treated controls (chapter 10).

References1. M. Thubrikar, "The aortic valve", CRC Press, Boca Raton, Florida (1983).2. S.L. Hilbert, M. Jones, and V.J. Ferrans, "Flexible leaflet replacement heart valves", in "Encyclopedic

handbook of biomaterials and bioengineering Part B: Applications", Ed. by D.L. Wise, et al., MarcelDekker, Inc., New York. p. 1111-1152 (1995)

3. F.J. Schoen, "Aortic valve structure-function correlations: Role of elastic fibers no longer a stretch of theimagination", J. Heart Valve Dis., 6 pp. 1-6 (1997).

4. F.J. Schoen, "Cardiac valve prostheses: Review of clinical status and contemporary biomaterials issues", J.Biomed. Mat. Res.: Appl. Biomat., 21(A1) pp. 91-117 (1987).

Chapter 1

- 6 -

5. S. Nitter-Hauge, "Mechanical heart valves. Conclusions from long-term follow-up", Eur. Heart J., 18 pp.907-910 (1997).

6. M. Julien, D.R. Letoueau, Y. Marvis, A. Cardou, M.W. King, R. Guidoin, D. Chanchra, and J.M. Lee,"Shelf-life of bioprosthetic heart valves: A structural and mechanical study", Biomaterials, 18(8) pp. 605-612 (1997).

7. M.A. Flomenbaum and F.J. Schoen, "Effects of fixation back pressure and antimineralization treatment onthe morphology of porcine aortic heart valves", J. Thorac. Cardiovasc. Surg., 105 pp. 154-164 (1993).

8. F.J. Schoen, H. Harasaki, K.M. Kim, and H.C. Anderson, "Biomaterial-associated calcification: Pathology,mechanisms, and strategies for prevention", J. Biomed. Mat. Res., 22(A1) pp. 11-36 (1988).

9. M.E. Nimni, D. Myers, D. Ertl, and B. Han, "Factors which affect the calcification of tissue-derivedbioprostheses", J. Biomed. Mat. Res., 35 pp. 351-357 (1997).

10. N.R. Vyavahare, W. Chen, R.R. Joshi, C.H. Lee, D. Hirsch, J. Levy, F.J. Schoen, and R.J. Levy, "Currentprogress in anticalcification for bioprosthetic and polymeric heart valves", Cardiovasc. Pathol., 6(4) pp.219-229 (1997).

11. E. Imamura, O. Sawatani, H. Koyanagi, Y. Noishiki, and T. Miyata, "Epoxy compounds as a newcrosslinking agent for porcine aortic leaflets: subcutaneous implant studies in rats", J. Cardiac Surg., 4 pp.50-57 (1989).

12. J.M. Lee, C.A. Pereira, and L.W.K. Kan, "Effect of molecular structure of poly (glycidyl ether) reagents oncrosslinking and mechanical properties of bovine pericardial xenograft materials", J. Biomed. Mat. Res.,28 pp. 981-992 (1994).

13. L.H.H. Olde Damink, P.J. Dijkstra, M.J.A. v. Luyn, P.B. v. Wachem, P. Nieuwenhuis, and J. Feijen,"Crosslinking of dermal sheep collagen using hexamethylene diisocyanate", J. Mat. Sci.: Mat in Med.,6(7) pp. 429-434 (1995).

14. L.H.H. Olde Damink, P.J. Dijkstra, M.J.A. v. Luyn, P.B. v. Wachem, P. Nieuwenhuis, and J. Feijen,"Cross-linking of dermal sheep collagen using a water-soluble carbodiimide", Biomaterials, 17(8) pp. 765-774 (1996).

15. V. Charulatha and A. Rajaram, "Crosslinking density and resorption of dimethylsuberimidate-treatedcollagen", J. Biomed. Mat. Res., 36 pp. 478-486 (1997).

16. K.S. Weadock, E.J. Miller, E.L. Keuffel, and M.G. Dunn, "Effect of physical cross-linking methods oncollagen-fiber durability in proteolytic solutions", J. Biomed. Mat. Res., 32 pp. 221-226 (1996).

17. H. Petite, I. Rault, A. Huc, P. Menasche, and D. Herbage, "Use of the acyl azide method for cross-linkingcollagen-rich tissues such as pericardium", J. Biomed. Mat. Res., 24 pp. 179-187 (1990).

18. W. Chen, F.J. Schoen, and R.J. Levy, "Mechanism of efficacy of 2-AOA for inhibition of calcification ofglutaraldehyde pretreated porcine and bovine pericardial heart valves", Circulation, 90 pp. 323-329(1994).

19. G. Golomb and V. Ezra, "Prevention of bioprosthetic heart valve tissue calcification by chargemodification: effects of protamine binding by formaldehyde", J. Biomed. Mat. Res., 25 pp. 85-98 (1991).

20. P. Zilla, L. Fullard, P. Trescony, J. Meinhart, D. Bezuidenhout, M. Gorlitzer, P.Human, and U.v. Opell,"Glutaraldehyde detoxification of aortic wall tissue: A promising perspective for emerging bioprostheticvalve concepts", J. Heart Valve Dis., 6 pp. 510-520 (1997).

21. J. Chanda, "Anticalcification treatment of pericardial prostheses", Biomaterials, 15(6) pp. 465-469 (1994).22. N.R. Vyavahare, D. Hirsch, E. Lerner, J.Z. Baskin, R. Zand, F.J. Schoen, and R.J. Levy, "Prevention of

calcification of glutaraldehyde-crosslinked porcine aortic cusps by ethanol preincubation: Mechanisticstudies of protein structure and water-biomaterial relationships", J. Biomed. Mat. Res., 40 pp. 577-585(1998).

23. R.J. Levy, X. Qu, T. Underwood, J. Trachy, and F.J. Schoen, "Calcification of valves aortic allografts inrats: Effects of age, crosslinking, and inhibitors", J. Biomed. Mat. Res., 29 pp. 217-226 (1995).

24. C.L. Webb, N.M. Nguyen, F.J. Schoen, and R.J. Levy, "Calcification of allograft aortic wall in a ratsubdermal model", Am. J. Path., 114 pp. 487-496 (1992).

25. M.E. Nimni, "Collagen: Molecular structure and biomaterial properties", in "Encyclopedic handbook ofbiomaterials and bioengineering Part A: Materials", Ed. by D.L. Wise, et al., Marcel Dekker Inc., NewYork. p. 1229-1243 (1995)

26. S.T. Li, "Biological biomaterials: Tissue-derived biomaterials", in "The biomedical engineeringhandbook", Ed. by J.D. Bronzino, CRC Press Inc. in cooperation with IEEE Press, Boca Raton, Florida. p.627-647 (1995)

27. E.E. Sabelman, "Biology, biotechnology, and biocompatibility of collagen", in "Biocompatibility of tissueanalogs", Ed. by D.F. Williams, CRC Press Inc., Boca Raton (1985)

General ntroduct on

- 7 -

28. P.B.v. Wachem, M.J.A.v. Luyn, L.H.H. Olde Damink, P.J. Dijkstra, J. Feijen, and P. Nieuwenhuis,"Biocompatibility and tissue regenerating capacity of crosslinked dermal sheep collagen", J. Biomed. Mat.Res., 28 pp. 353-363 (1994).

29. Y. Noishiki, H. Koyanagi, T. Miyata, and M. Furuse, Bioprosthetic valve, Patent EP 0 306 256 A2 1988.30. X. Tingfei, M. Jiazhen, T. Wenhua, L. Xuehui, L. Shuhui, and X. Baosha, "Prevention of tissue

calcification on bioprosthetic heart valve by using epoxy compounds: A study of calcification tests in vitroand in vivo", J. Biomed. Mat. Res., 26 pp. 1241-1251 (1992).

31. R. Zeeman, P.J. Dijkstra, P.B. v. Wachem, M.J.A. v. Luyn, M. Hendriks, P.T. Cahalan, and J. Feijen,"Cross-linking and modification of dermal sheep collagen using 1,4-butanediol diglycidyl ether", Chapter3 of this thesis and submitted to J. Biomed Mat. Res, (1998).

32 P.B. v. Wachem, R. Zeeman, P.J. Dijkstra, M. Hendriks, P.T. Cahalan, J. Feijen, and M.J.A. v. Luyn,"Characterization and biocompatibility of epoxy crosslinked dermal sheep collagen", Chapter 7 of this thesis and submitted to J. Biomed. Mat. Res., (1998).

Cross-l nk ng and calc f cat on of collagen-based mater als

- 9 -

Chapter 2

Cross-linking and calcification of collagen-based

materials

R. Zeeman,1 P.J. Dijkstra,1 P.B. van Wachem,2 M.J.A. van Luyn,2

M. Hendriks,3 P.T. Cahalan,3 and J. Feijen1

1 University of Twente, Department of Chemical Technology, and Institute of Biomedical Technology, P.O. Box217, 7500 AE Enschede, The Netherlands; 2 University of Groningen, Faculty for Medical Sciences, Cell Biologyand Biomaterials, Bloemsingel 10/B2, 9712 KZ, Groningen, The Netherlands 3Medtronic Bakken Research CenterB.V., Endepolsdomein 5, 6229 GW Maastricht, the Netherlands;

INTRODUCTION

In the search for biomaterials that are both versatile and compatible with human-tissues,considerable interest has been maintained in collagen-based preparations for the repair andreplacement of soft body tissues such as tendons, skin, vascular grafts and heart valves. Thegeneral properties of collagen which make this protein interesting as a biomaterial include the highstrength of the fibers, low extensibility, minimal antigenicity, its suitability as a substrate for cellgrowth, and its controllable stability by chemical or physical cross-linking [1].During in vivo applications, collagen is prone to enzymatic attack which can result in rapiddegradation of the material. Therefore, collagen-based materials are frequently stabilized by cross-linking to control the rate of biodegradation. In addition, cross-linking is effective in suppressingthe antigenicity of collagen and can improve the mechanical properties [1-4]. However, cross-linking of collagen-based tissues enhances the tendency to calcify, which is the main cause offailure of for example tissue heart valves [5, 6].In this article, the structure of collagen will be discussed. Thereafter, the cross-linking methodsknown and the effects on calcification and material properties are given. A collagen-basedmaterial, the porcine aortic heart valve, is described in more detail. Finally, a survey of factorswhich induce calcification and the strategies which were carried out to prevent calcification arereviewed.

Chapter 2

- 10 -

THE STRUCTURE OF COLLAGEN

Collagen, the most abundant protein in mammalian tissues, accounts for up to 30% of all proteins,but is not evenly distributed throughout the body. In human heart valves collagen represents 50% -70% of the tissue on a dry weight basis and in elastic arteries approximately 25%. The mainfunction of collagen is mechanical reinforcement of the connective tissues of vertebrates [7, 8].The individual polypeptide chains of collagen contain 20 different amino acids and the precisecomposition varies between different tissues. The variation in specific amino acid sequence givesrise to the different types of collagen labeled as Type I, Type II up to Type XIX. The mostcommonly occurring collagens are Types I, and III, which form the long-recognized characteristicfiber bundles seen in many tissues. Type I collagen is mostly found in skin, tendon, and bone, andType III in blood vessels [9]. The various collagen types show differences in degrees ofglycosylation, which means that glucose and galactose are covalently coupled to the collagenmolecules.The lysine (Lys) and proline (Pro) residues present in the collagen are partly hydroxylated yieldingthe rare amino acids hydroxyproline (Hyp) and hydroxylysine (Hyl), respectively. Because thefiber forming collagen types are most abundant, their structure is discussed in more detail below.

The name collagen is used as a generic term to cover a wide range of protein molecules whichform supramolecular matrix structures. They share the basic texture of three individual α-chainsthat are cross-linked biosynthetically and fold to form a triple helix (tertiary structure) with amolecular weight of approximately 300.000 g/mol, a length of approximately 300 nm and adiameter of 1.5 nm [10]. This triple-helix generates a symmetrical pattern of three left-handedhelical α-chains (secondary structure), which consist of about 1000 amino acid residues, formingan additional "supercoil" with a pitch of 86 Å. The amino acids within each chain are displaced bya distance of 2.91 Å, with a relative twist of - 110 °, making the number of residues per turn 3.27and the distance between each third glycine 8.7 Å [11, 12].

G lyc in eP red om in an tly p ro l ine an d h yd rox yp ro lin e

8 .6 nm

0 .8 7 n m

1 .5 nm

Figure 1. The collagen triple-helix.

The presence of the cyclic imino acids, Pro and Hyp imparts rigidity and stability to the coil.Glycine (Gly), the smallest amino acid, must be in every third position in order to create the right-


- 11 -

handed triple helix. Furthermore, the hydroxyl groups of Hyp residues are involved in hydrogen-bonding and are important for stabilizing the triple-helix structure and two hydrogen bonds pertriplet are found. The two hydrogen bonds formed are: one between the NH-group of a glycylresidue with the CO-group of the residue in the second position of the triplet in the adjacent chain,and one via the water molecule participating in the formation of additional hydrogen bonds withthe help of the hydroxyl group of Hyp in the third position [13, 14]. Such a 'water-bridged' modelof the triple helix has been confirmed by physiochemical studies of the collagen molecule insolution and is supported by the observation that the thermal stability of the helix is dependent onthe content of Hyp and not of Pro [13, 15]. In addition, model studies showed that Gly, Hyp andPro are the triple-helix forming amino acids and that only molecules which contain the triplets Gly-Pro-Hyp were able to form a helical structure [16]. Therefore, the collagen triple helical domainshave an amino acid sequence (primary structure) that is rich in Gly, Pro and Hyp [9].

Figure 2. The molecular architecture of the fiber forming collagens.

The collagen molecules possess an axial periodicity that is visible in the electron microscope andpack into lattices with lateral symmetry (quaternary structure). This supramolecular structure iswidely accepted as the microfibril containing five collagen triple-helices [11, 17-19], with adiameter between 3.5 and 4.0 nm. Approximately 1000 microfibrils can aggregate laterally andend-to-end into a fibril having a diameter of 80-100 nm [10, 12, 18], that displays a regularbanding structure with a period of 65 nm (figure 3) [11]. About 500 fibrils form a collagen fiberwith a diameter of 1-4 µm [20]. Finally, the fibers aggregate into fiber bundles with a thicknessbetween 10 and 100 µm [10]. However, the hierarchy of the collagen is highly dependent on itsfunction. For example, the fibril and fiber diameter of collagen in skin varies between 20 - 100 nmand 0.3 - 40 µm, respectively. The diameter of the collagen fibril and fiber (fiber bundle) intendons and ligaments is 20 - 250 nm and 1 - 300 µm, respectively [21].

Chapter 2

- 12 -

C o llag en m o le c u le

F o rm a t io n o f m ic ro f ib r i lsL a te ra l ag g reg a t io n

E n d to en d ag g re g a tio n

C o llag en f ib r i l

3 .5 - 4 .0 n m

1 .5 n m

8 0 -1 0 0 n m

Figure 3. Formation of microfibrils and collagen fibrils.

In order to develop an extracellular network of collagen fibers, the cells involved in thebiosynthetic process must first synthesize a precursor known as procollagen. This moleculepossesses a long, non-interrupted triple helical region with N- and C-terminal globular extensionscalled propeptides. The molecule is later proteolytically trimmed of its propeptide domains, givingrise to a tropocollagen molecule with short non-helical ends of 15 to 25 amino acid residues (N-and C-terminal telopeptides) that spontaneously assembles into fibers in the extracellular space [9].Cross-linking renders these fibers stable and provides them with an adequate degree of tensilestrength and visco-elasticity to perform their structural role. Important are the degree of cross-linking, the number, and the density of the individual fibers [11]. Two types of natural cross-linksin collagen can be distinguished and are based on the aldehyde groups formed from(hydroxy)lysine residues in the telopeptides, by its enzymatic oxidation by lysyl oxidase, yieldingallysin [10, 14, 18, 22]. Intramolecular cross-links are formed by an aldol condensation reaction oftwo aldehyde groups. An intermolecular cross-link is formed if the aldehyde group reacts with theε-amino group of an (hydroxy)lysine residue of an adjacent helix, yielding an aldimine or a Schiffbase [11-13].

CROSS-LINKING

Collagenous tissues, obtained from the slaughter house to be used as bioprostheses, begin todegrade immediately. Therefore, in the exploitation of tissue as clinical material this deteriorationmust be arrested and deferred. The aim is to prolong the original structural and mechanicalintegrity and remove or at least neutralize the antigenic properties attributed to these materials.Methods concentrate on creating new additional chemical bonds between the collagen molecules,which reinforce the tissue to give a tough and strong but non-viable material that maintains the


- 13 -

original shape of the tissue. Ideally, the treatment of natural biomaterials should maintain much ofthe original character of the tissue, such as its flexibility and its mechanical properties [2].

GlutaraldehydeThe predominant chemical agent that has been investigated for the treatment of collagenoustissues is glutaraldehyde [3, 5, 23-25], which gives materials with the highest degree of cross-linking when compared with other known methods such as formaldehyde, epoxy compounds,cyanamide and the acyl-azide method [2, 3, 26-28]. The reactions involved during cross-linking ofproteins with glutaraldehyde have been extensively studied [3, 24, 29], but the reaction mechanismis very complex and still not completely understood. Aqueous solutions of glutaraldehyde containa mixture of free aldehyde and mono- and dihydrated glutaraldehyde and monomeric andpolymeric hemiacetals (figure 4). Because of the ease of hydration and cyclization, theconcentration of free, monomeric aldehydes in concentrated, commercial solutions is usually low.However, Olde Damink showed that the concentration of monomeric glutaraldehyde could beincreased by distillation. In addition, he calculated that the content of polymeric glutaraldehyde inthe reaction solution was rather low [24]. Glutaraldehyde solutions may contain various productsresulting from aldol condensation during storage and cyclic glutaraldehyde oligomers having atrioxane structure have been described as well.

(CH2)3

C

C

H O

OH

��

��H2O

(CH2)3

C

C

HO HOH

OH

��

H2O

(CH2)3

C

C

HO HOH

OHHHO

��-H2O

O OHHO

Free GA Monohydrate Dihydrate Cyclic hemiacetal

A) Monomeric form

B) Polymeric hemiacetals

O OHHO

2��

OHO O O OH

�� O OHHO

Dimer

O OHOH[ ]n

Polymer

C) Polymers α and β unsaturated

(CH2)3

C

C

H O

OH

�� Free GA(CH2)2

COH

C C

H OC

(CH2)3

C HOH

�� Free GA

Free GA Dimer

(CH2)3

COH

C C

H

CO

H

CH2

CCO

H

C

H(CH2)3 C

O

H

Trimer

Figure 4. Possible structures of glutaraldehyde (GA) in aqueous solutions.

Chapter 2

- 14 -

Because of the complexity of the reaction solutions, many reactions can occur during cross-linking[24]. Cheung et al. suggests that the penetration of glutaraldehyde molecules into dense tissuesuch as pericardium is slow and that primarily at the outer surfaces of the fibers are fixed. Inaddition, a polymeric network is created which hinders further cross-linking [29]. In general,aldehydes react with the amine groups of (hydroxy)lysine residues of the collagen, yielding a Schiffbase [30], which can be stabilized by a reduction reaction.

Glutaraldehyde was first applied successfully for bioprosthesis in the late 60s by Carpentier et al.[3, 5]. Porcine aortic heart valves treated with glutaraldehyde showed good heamodynamicperformance and a low antigenicity [3, 31]. However, it is now known that the durability ofglutaraldehyde fixed biological tissue is not so good as once thought [2]. Glutaraldehyde treatedmaterials calcify to a large extent [32, 33], which might be due to the cross-linking process. Forinstance, calcification is the major cause in the failure of bioprosthetic heart valves. Moreover,depolymerization of polymeric glutaraldehyde cross-links has been reported, which releasesmonomeric and highly cytotoxic glutaraldehyde into the recipient [34-39].

Researchers embarked on attempts to replace glutaraldehyde as a cross-linking agent. The obviousstrategy would be the use of other bifunctional reagents. Epoxy and diisocyanate compoundsdominate this approach. Another strategy is to activate the carboxylic acid groups of collagen,followed by a reaction with an adjacent amine group. This method is the basis for the carbodiimideand the acyl azide reactions.

Epoxy compoundsEpoxy compounds have been extensively used in the past decade for the stabilization of collagen-based materials including porcine aortic heart valves [40-45]. Generally mixtures of bi- andtrifunctional glycidyl ethers based on glycerol are applied. In addition, a broad range ofmultifunctional epoxy containing cross-linkers can be used [46]. Due to its highly strained three-membered ring, epoxide groups are susceptible to a nucleophilic attack [30]. Predominantly, areaction with the amine groups of (hydroxy)lysine residues will occur [47-49] as shown in figure5.

NH2

CH2CHCH2OCH2

CH2 OR

CH2 O CH2 CH CH2

O

O

NH2

��pH > 8.0

NH

CH2CH

OH

CH2OCH2

CH2 OR

CH2 O CH2 CH CH2

OH NH

Figure 5. Schematic representation of the cross-linking reaction of an epoxy compound withcollagen.


- 15 -

Additionally, epoxide groups can react with the secondary amine groups of histidine. Furthermore,reactions with the carboxylic acid groups of aspartic and glutamic acid exists, thereby increasingthe versatility of the cross-linking [2, 47, 50, 51].In general, biological tissues are cross-linked in basic solutions (pH > 8.0) containing relativelyhigh concentrations of epoxy compounds ranging from 1 to 5 wt%. A lower shrinkagetemperature (Ts) was obtained compared to GA cross-linked materials but the in-vitro stability ofthe cross-linked tissue was similar [52, 53].Vascular grafts cross-linked with epoxy compounds had higher tensile strengths and extensibilities,a lower stiffness and a better compliance compared to GA treated prostheses [43, 54]. In addition,they retained more the original character, whereas GA cross-linked grafts were somewhat stiffer[55, 56]. For example, porcine aortic heart valves cross-linked with glycerol polyglycidyl etherwere more pliable than their GA counterparts. Moreover, the epoxy fixed valve appeared to openmore widely [57]. Subcutaneous implant studies in rats revealed that grafts cross-linked withepoxy compounds displayed a lower calcification [58]. The cytotoxicity of several epoxide-containing compounds has been evaluated by in-vitro studies and has been shown acceptable [59].Besides epoxides, other bifunctional cross-linkers have been applied in cross-linking of collagensuch as hexamethylene diisocyanate [60, 61], dimethyl suberimidate [39, 62, 63] and bis-N-hydroxysuccinimide ester derivates [63, 64]. These methods will not be discussed in this overview.

CarbodiimidesThe carbodiimide reagent offers a method for generating crosslinks between carboxylic acid andamine groups, without itself being incorporated [2, 65, 66]. The water-soluble carbodiimide 1-ethyl-3-(3-dimethyl aminopropyl)carbodiimide (EDC) is often used for crosslinking collagen(figure 5). EDC crosslinking involves the activation of the carboxylic acid groups of Asp or Gluresidues (I ) by EDC (II ) to give O-acylisourea groups (III ). In order to suppress side reactions ofO-acylisourea groups such as hydrolysis and the N-acylshift [65], N-hydroxysuccinimide (NHS)(IV) is used to convert the O-acylisourea group into a NHS activated carboxylic acid group (V),which is very reactive towards amine groups of (hydroxy)lysine (VII) , yielding a so-called zero-length cross-link (VIII) [67]. EDC is not incorporated in the matrix but is converted to 1-ethyl-3-(3-dimethyl-aminopropyl)-urea (VI) .

Chapter 2

- 16 -

COOH + C

N

N

R'

R"

��C

O

O C

NH

R'

N

R"

��

N OH

O

ONO

O

C

O

O

+

R' NH C

O

NH R"

��

NH2

C

O

N

N

+

N OH

O

O

(I)

(II)

(III)

(IV)

(V)

(VI)

(VII)

(VIII)

R' = CH3

R"= N+HCH3

CH3Cl-

(IV)

Figure 6. Cross-linking of dermal sheep collagen with EDC and NHS [65].

Cross-linking of dermal sheep collagen resulted in materials having a higher Ts and enzymaticresistance than GA cross-linked collagen. Furthermore, rat subdermal implantation studies showedthat the EDC/NHS cross-linked collagen samples had a low tendency to calcify and a goodbiocompatibility [33, 68]. It appears that the amide crosslinks formed may be beneficial in terms ofthe anticalcification properties by limiting calcium binding sites [2]. Cross-linking of bovinepericardium with EDC/NHS led to materials with similar values of Ts and comparable in-vitrostability as GA cross-linked pericardium [69].Additionally, EDC and NHS can be used in combination with diamine or diacid compounds tointroduce 'extended' cross-links. The carboxylic acid groups of either collagen or diacid moleculeswill be activated with EDC/NHS followed by reaction with the amine groups of diamine moleculesor collagen, respectively [70, 71].

Acyl azideThe acyl azide method is another cross-linking procedure in which the acid groups becomeactivated followed by reaction with an adjacent amine group. The acyl-azide method is a multi-step reaction in which the carboxylic acid groups are first esterified with methanol under acidicconditions for 7 d. Than the methylated acid groups are converted to a hydrazide by reaction withhydrazine. Finally, the hydrazide is reacted with sodium nitrite to give the acyl azide, which cansubsequently react with an amine group of an adjacent polypeptide chain. Acyl-azide cross-linkedmaterials show very high values of Ts and very good in-vitro stabilities [72, 73].A variation on this method has been developed by Petite et al. They used diphenylphosphorylazide(DPPA) to convert the carboxylic acid group into an acyl azide group in one single step [74, 75].DPPA and the acyl-azide cross-linked materials were found to be less toxic than GA counterparts[2] and a marked reduction in calcification was obtained compared to GA cross-linked controlsafter 90 d of subcutaneous implantation in rats [73].


- 17 -

Several other cross-linking reagents have been used such as cyanuric derivatives [76], chromium[37] and ribose [77]. However, the modified materials were toxic or had a low cross-link densityand therefore these methods are hardly used to produce bioprosthetic materials.

Physical cross-linkingThe primary advantage of physical treatments is that they do not introduce chemicals that causepotential harm. Typical processes such as heating, drying and irradiation have been applied tocollagen. Short wave length UV irradiation (254 nm) can introduce cross-links in the collagen.However, chain scission may become a substantial side reaction resulting in denaturation of thecollagen molecules.Dehydrothermal treatment (DHT) increases the shrinkage temperature of collagen by removingwater. Removal of water from the collagen results in formation of interchain cross-links [78, 79].During DHT cross-linking, collagen is extensively dried in vacuo for several days at temperaturesup to 100 °C. Generally, the degree of cross-linking is considerably lower than obtained bychemical methods [80]. Sometimes, DHT treatment is followed by a chemical treatment(cyanamide) to increase the stability of the treated material [35, 78].Finally, Moore et al. used a dye-mediated photo-oxidizing method to cross-link bovine pericardialtissue. This method, which led to the modification of histidine, tryptophan, tyrosine, andmethionine, resulted in materials which were resistant to pepsin and cyanogen bromide (CNBr)treatments. Remarkably, the Ts was similar to the untreated material, which suggests that thetissue behaves like the original and that the cross-links did not influence the tissue character [81,82].

Degree of cross-linkingSeveral methods have been used to quantify the degree of cross-linking of collagen. A higherdegree of cross-linking is generally associated with a lower antigenicity. Furthermore, thebiostability and durability are related to the increase in denaturation temperature [41]. Thetemperature at which denaturation of the triple-helix structure occurs, is frequently used to assessthe degree of cross-linking of the collagen material. When collagen is heated in the hydrated state,the material will denature at a specific temperature, resulting in the shrinkage of the material toabout one-third of its original length. This shrinkage, which takes place within a narrowtemperature range of 2-3 °C, is the macroscopical manifestation of the transformation of thetriple-helices to random coils [12]. The denaturation temperature is usually referred to as theshrinkage temperature. Cross-linking of collagen increases the denaturation temperature of thematerials. Introduction of covalent cross-links will increase the stability of the helix and thusincrease the denaturation temperature [11]. However, others explain the increase of thistemperature by the degree of swelling. The melting temperature of collagen helices will bedepressed by water-uptake, i.e. the degree of swelling of the material [83]. Cross-linking of thetissue will lead to a lower degree of swelling and thus in a higher melting temperature. In addition,parameters such as the structure and the nature of the cross-links introduced, the solvent, the pHand the ionic strength will affect the degree of swelling. Several protocols have been developed tomeasure the shrinkage temperature (Ts). These protocols can be divided into three groups:

Chapter 2

- 18 -

hydrothermal isometric tension tests [84], a shrinkage test which is usually applied in the leatherindustry [85, 86], and finally differential scanning calorimetry [15, 87-89].A further tool to quantify the degree of reaction when these involve amine groups is to determinethe content of amine groups before and after cross-linking. Reaction of the primary amine groupsof the collagen material with 2,4,6-trinitrobenzene sulfonic acid (TNBS) yields a yellow coloredproduct. After hydrolysis of the sample, the content of amine groups of the solution can bedetected spectrophotometrically [24, 90]. The lower the content of amine groups after reaction,the higher the degree of cross-linking.

The influence of cross-linking on the degradation behavior of collagen has been frequently studiedin-vitro. Degradation of collagen is promoted by enzymes although chemicals such as CNBr [69]are also known to degrade collagen. While a-specific enzymes such as trypsin [80] and pronase[29] have been occasionally used, often bacterial collagenase from Clostridium histolyticum isapplied to assess the stability of the materials after cross-linking [12, 52, 69, 91].Furthermore, the mechanical properties of collagen-based materials will be affected by the cross-linking procedure. Contradicting results have been found and will be described later.

HEART VALVES

Porcine aortic valve tissueCardiac valve replacement has significantly improved the perspectives of patients with valvularheart disease. Surgical techniques and prosthetic valve design have undergone considerableevolution since the first aortic valve replacement by Harken et al. and the first mitral valvereplacement by Starr [5].Two general types of replacement valves can be distinguished: mechanical and tissue. Mechanicalvalves, generally fabricated with components from rigid, nonphysiologic biomaterials, comprisethe caged-ball, caged-disk and the tilting disk mechanical valves. [5, 92] Tissue valves includeallo- or homografts, which are valves taken from human donors, and hetero- or xenografts, whichare either porcine aortic valves or valves constructed from bovine pericardial tissue. The majoradvantages of tissue valves relative to mechanical prostheses are the pseudo-anatomic central flowand relative lack of surface induced thrombus formation, usually without anticoagulation therapy.Tissue valves can be subdivided into stented and stentless prostheses. Stented valves are mountedon a polypropylene stent, and have a fabric, usually dacron, sewing ring that surrounds the valveorifice at the base. Stentless valves contain usually a part of the aortic wall tissue in which theleaflets are anchored and a small dacron covering. Stentless valves provide superiorheamodynamics due to the lack of a rigid stent [93, 94].A new, promising group of valve prostheses are the trileaflet prostheses using synthetic polymerssuch as polyurethane or polytetrafluoroethylene [95]. The continued interest in segmentedpolyurethanes is based on the physical properties which are ideally suited for the use as a leafletcomponent. Previously, calcification limited the long-term durability of these materials, butprosthetic heart valves based on polyetherurethane-urea have potential for long-term applicationsdue to the lack of severe calcification [96].


- 19 -

MorphologyThe aortic valve opens to allow blood to flow into the aorta, and closes to prevent backflow intothe left ventricle. The valve opens and closes approximately 103,000 times each day and about 3.7billion times in its life span [97].The aortic valve is composed of three, endothelially invested membranous cusps or leaflets andaortic sinuses. The leaflets, which are the most mobile parts of the valve, are anchored in the aorticwall. The sites where the leaflets come together, are called the commissures. Between the leafletsand the aortic wall there are dilated pockets called the aortic sinuses. From two of these sinusesthe coronary arteries originate. The only anatomical difference between the human and the porcineaortic heart valve is the presence of a muscular shelf on the right coronary leaflet. The presence ofthis muscle shelf results in a delayed opening of the right coronary leaflet relative to that of the leftand the non-coronary. Along the top of each leaflet is the free edge. In the middle of it, there is ancollagenous-rich area, the Corpus Arantii (or Nodulus of Arantius), which supposedly aids in thevalve closure and reduces regurgitation [97-99].

A ort ic s inu s

L e af le tL e af le t

B lo o d f lo w

A ort ic w a l l

Figure 7. The aortic valve and the leaflets anatomy [100]

Leaflets consist of very small elastic and collagenous fibers relatively loosely arranged. Thecollagenous fibers, which are the major protein component of the leaflets, are unusually small: 300- 500 Å [97]. Collagen types I and III are predominant collagen constituents (99 %) of heart valvetissue. A low content of methionine is found, whereas a high content of hydroxylysine is present,which contributes to the formation of stable native cross-links in the tissue [101]. Furthermore, acertain extent of glycosylation of the α1 and the α2 chains is observed. The collagen fibrils andbundles are not completely straight but follow wavy courses. This arrangement, usually referred toas crimping, allows changes in geometry of collagen-containing structures without substantialincrease in tension [102]. Collagen comprises 60 % of the total dry weight of human aorticleaflets. Due to aging this content drops after 80 years to 40 % [101]. Another importantcomponent of valve tissue are elastic fibers, which are largely responsible for the elasticity of thetissue. They have two biochemically and structurally distinct components: elastin, a hydrophobic

Chapter 2

- 20 -

and amorphous protein and microfibrils which consist of several non-elastic glycoproteins such asfibrillin [99, 103].In most connective tissues, collagen is found in close association with proteoglycans. It is thoughtthat they are involved in the in-vivo collagen fibril formation. Proteoglycans are composed ofglycosaminoglycans (GAG) and core proteins such as aggregan, decorin, lumican, perlecan andmany more [103]. The GAGs are complex mucopolysaccharides and are covalently linked to thecore protein by serine and threonine ester bonds. The GAGs contain 65 % hyaluronic acid, 25 %chondroitin sulfate A/C and 10 % chondroitin sulfate B [104]. Due to the anionic nature of theGAGs which contain many carboxylate and sulfate groups, ionic interactions with the(hydroxy)lysine and arginine groups are present.Furthermore, two categories of cellular components present in heart valves can be classified: liningcells and connective tissue cells. Lining cells are endothelial cells, while connective tissue cells aremainly fibroblasts, myofibroblasts and smooth muscle cells [99].

Figure 8. Schematic cross-section of the leaflet.

The fibrosa faces the aorta and the ventricularis faces the left ventricle of the heart

In cross-section, the leaflet has three distinct layers [97], the fibrosa, spongiosa and ventricularis:- Lamina fibrosa: A very dense layer, arranged as a series of parallel tedious cords or in a rigidsheet of tissue. The collagenous fibers are mainly oriented in a circumferential direction. This layerprovides the essential strength of the leaflets.- Lamina spongiosa: A very loose, watery connective tissue of varying thickness, consisting offiber components, glycosaminoglycans (GAGs) and cells. Its sparse collagenous fibers and cellsare oriented radially. It has a negligible structural strength but appears to perform an importantrole in minimizing mechanical interaction between the two fibrous layers and in dissipating energyduring closure [105].- Lamina ventricularis: Consists of a superficial elastic layer, which is two or several fibers thick.This layer is less organized than the fibrosa. It enables the leaflet to have minimal surface areawhen it is open but stretch to form a large coaptation area when back pressure is applied .

Mechanical propertiesAortic leaflets have very specific anisotropic mechanical properties, which are reflected in a largedifference between the circumferential and radial direction. In the circumferential direction a highmodulus and a low elongation at break are obtained, whereas the opposite is true in the radialdirection. Collagen cords are oriented primarily in the circumferential direction and they dominatethe properties in that direction. Crimp of collagen fibers results in an initial low modulus. Upon


- 21 -

further application of stress, the fibers offer a high resistance to stretch and produce a highmodulus. In the radial direction, the properties are dominated by elastin, offering the material ahigh compliance.

Figure 9. A typical stress-strain plot for aortic leaflets.

The curve can be subdivided into several parts. In the initial part of the curve, the stress risesslowly as the strain increases. In the latter part, the stress rises rapidly if the strain increases [97].The material in the circumferential direction is about 6 times stronger (tensile strength of 6.3 MPavs. 1.2 MPa) and 7 times stiffer than in the radial direction (E-modulus of 54.6 MPa vs. 7.8 MPa)[25]. In general, a stress-strain curve of biomaterials can be described of being made of two linearsegments joined by a third non-linear, transitional segment [97]. The initial or elastic phase of thecurve shows a low E-modulus, the so-called pretransition modulus. A very high modulus, theposttransition modulus is obtained in the latter or inelastic phase. The leaflet breaks at a muchlower load in the radial than in the circumferential direction. Hence, the leaflet is more compliantand weaker in the radial but less compliant and stronger in the circumferential direction. Thesedifferences in stiffness, compliance and strength are related to the anisotropic leaflet structure. Theleaflet exhibits a visco-elastic behavior which means that the relationship between stress and strainis time-dependent [97].

Effect of cross-linkingGlutaraldehyde fixation of bioprosthetic tissue alters the mechanical characteristics of the tissuesignificantly. Not only the stress-strain curves were changed but the hysteresis as well.Furthermore, a reduction in relaxation and creep was obtained [106]. Lee et al. studied the effectof different cross-linking methods on the tissue properties. Cross-linking of bovine pericardiumwith glutaraldehyde or a poly epoxy compound resulted in an increase of the extensibility and areduction in stress relaxation. The ultimate tensile strength was increased from 2.8 to 5.0 MPa.Besides the tissue modules, which is defined as the posttransition modulus shown in figure 8, wasraised from 19.3 to 23.0 MPa for glutaraldehyde cross-linked materials and decreased to 12.1

Chapter 2

- 22 -

MPa for polyglycidyl ether cross-linked tissue [61]. They state that the reduced stress relaxation iscaused by the presence of interfibrillar cross-links. Further, the flexural stiffness is increased,which could be a disadvantage in preparing heart valve prostheses, where good flexural propertiesare necessary for free valve opening and low pressure gradients [61]. The material propertiescould be modified by performing the fixation reaction in an alcohol instead of an aqueous bufferedsolution. Fixation with glutaraldehyde in ethanol, butanol or propanol resulted in a higherelongation at break, and a more natural stress relaxation. The tensile strength was not affected.These effects of the solvent on the mechanical properties were ascribed to cross-link efficacy andinteractions between solvent and the polar collagen molecules [107]. Cross-linking reduced theelastic modulus at low strains. It was concluded that the drop in elastic modulus is a phenomenonassociated with cross-linking of the collagen fiber matrix. Fixation may alter the collagen crimplength and can distort the collagen fiber integrity [108]. Studies showed that collagen crimp iseliminated in many regions of the valve after pressure fixation with GA. The collagen thus alteredwas considered to be more susceptible to mechanical injury than normally crimped collagen [109].The crimp length, which is a measure of the collagen waviness, may be minimized by hydrostaticpressure applied to the tissue during fixation [110].Imamura et al. did not find an influence of GA or epoxy cross-linking on the tensile strength [58].However, cross-linking of pericardial tissue with glutaraldehyde resulted in an increase of theelongation at break from 40 to 60 %, while cross-linking with an epoxy compound increased it to120 %. The tensile strength was not affected [54]. The effect of the cross-linking method on themechanical behavior of the tissue was further demonstrated by Zhou et al., who showed thatpolyepoxy compound fixation resulted in a more flexible aortic wall as compared to GA cross-linking [56].Vesely and Noseworthy studied the two components of the leaflets namely the fibrosa andventricularis separately. They found that the fibrosa is stiffer circumferentially than radially but theextensibility was uniform. The ventricularis showed a higher stiffness in the circumferentialdirection and a higher extensibility was observed in the radial direction. Glutaraldehyde-fixationdid not affect the circumferential elastic modulus of the fibrosa, but it reduced its radial modulus.The E-modulus of the ventricularis remained unchanged. Fixation reduced the circumferentialextensibility of the ventricularis, and increased the radial extensibility of the fibrosa. It is obviousthat the two layers complement each other during fixation, and become detrimentally altered byglutaraldehyde fixation [111]. Studies showed that glutaraldehyde fixation did not alter theanisotropic behavior in stiffness and strength in the two orthogonal directions [112].Another problem which arises during fixation is the change in orifice area of the valve. Owing toan increase in flexural stiffness, the orifice area is reduced after fixation. Predilation prior tofixation resulted in a larger orifice area compared to standard fixed valves [113]. Nevertheless,fixed valves behave in another way than fresh ones, because of the reduced extensibility and thehigher stiffness of the aortic wall [56].Finally, pre-treatment of tissue is another aspect which can alter the mechanics of the tissue.Cryopreservation of porcine aortic heart valves did not change the mechanical characteristics ofthe leaflet, while extraction of lipids and cells reduced the fraction tension (force per unit width ofspecimen) from 2.6 to 2.1 kN/m (reduction in tensile strength from 5.9 to 3.3 MPa) and increasedthe extensibility from 31 to 45 %. However, the mechanics under physiological loading is


- 23 -

preserved. Fixation of extracted leaflets, resulted in a marked drop in fracture tension from 2.4 to1.3 kN/m (3.6 to 2.1 MPa) [114].

CALCIFICATION

General features of biomaterial associated calcification [115]Calcification is a normal, or physiologic, event in the formation of bone, dentin, and tooth enamel,but calcific deposits are unusual in functional soft tissues.Accumulation of crystalline calcium phosphate mineral in necrotic, injured or altered tissues, arelatively common phenomenon in cardiovascular disease, is called pathologic. This type ofcalcification can be subdivided into two groups:metastatic, which occurs in hypercalcemic hosts with otherwise normal tissues, and dystrophic,where the tissues are necrotic or otherwise altered in normocalcemic subjects.Calcification can occur within natural tissues, with purely biological substrates rendered nonviableby chemical treatment (bioprosthetic) materials, or in association with synthetic materials, such aspolyethers, and polyurethanes. Calcification occurs when a biomaterial is implanted into thecirculatory system. The mature mineral phase is apathetic, generally poorly crystalline calciumhydroxyapatite (Ca10[PO4]6[OH]2). The location of mineral nucleation may be intrinsic, i.e., withinthe biomaterial, or extrinsic to the biomaterial itself.In general, the determinants of mineralization include the next factors:- Host factors such as calcium metabolism- Implant factors such as cross-linking, local stress concentrations, presence of surface

defects, and surface-adhered organic or cellular debris [116].Biomaterials associated calcification occurs most predominantly in association with bioprostheticheart valves, and mechanical blood pumps.

Bioprosthetic heart valvesThe long-term durability of bioprosthetic heart valves, constructed from porcine aortic heartvalves or bovine pericardium, is limited by calcific degeneration. Primary tissue failure due tocuspal calcification has been registered in many cases [115]. Calcium deposition causes valvetearing, rupture, and stiffness, and results in heamodynamic stenosis or insufficiency, or both[117]. The failure rate in adults is approximately 15-25 % or higher, 7-10 years after implantation.Calcification is more likely to become severe and clinically significant in children and young adults,than in older patients. Degenerative intrinsic calcification begins primarily at the cuspalattachments, which are the sites of greatest cuspal mechanical stresses [115].Since calcification is the major pathologic feature associated with bioprosthetic valve failure,considerable amounts of work are being directed towards elucidation of the mechanisms andmethods for obviating this problem. Unfortunately, the mechanism of calcification is still not wellunderstood. Potential factors which induce calcifications may be grouped under a number ofheadings [117]:

1. Host metabolism: children’s metabolism, hyperparathyroidism, renal failure

Chapter 2

- 24 -

2. Collagen damage: physical stresses due to valve design, immune response, infection3. Chemical changes in the implanted material as a result of preservation techniques.

Ultrastructural examination shows that initial deposits are localized at transplanted connectivetissue cells. Direct collagen involvement occurs later [5, 118]. Other studies demonstrated thatmineralization of valve tissue is associated with connective tissue cells, cell debris, membranefragments as well as collagen and elastin [119-121]. Mako and Vesely determined thatbioprosthetic heart valve calcification is an intrinsic phenomenon regardless of the density of thecollagen fibers or the presence of connective tissue cells [122]. Contrary to leaflets, elastin appearsto be the major site of extracellular calcific deposits in aortic wall tissue [120]. Maxwell et al.studied the differences between allografts and xenografts. After implantation in humans, grosscalcification was found more frequently in porcine xenografts (89 %), and more extensive than inallografts (53 %). Pathological examination showed that the smallest calcific deposits were usuallyassociated with membrane debris of porcine donor fibroblasts. Allografts do not contain donor cellremnants and early calcification was arranged along collagen fibers [123]. Furthermore, clusters ofcalcific crystals were observed on the defective valve surface [124]. Therefore, alteration of thevalve surface appeared to be an important factor for inducing calcium phosphate crystal formation.Vasin et al. hypothesized that the formation of a protein-layer containing protein-calciumcomplexes on a biomaterial is the key event in calcification [125].

Studies on polyethylene glycol hydrogels showed that calcification of these materials was highlydependent on the molecular weight between cross-links [126] and a lower calcium content wasfound if the molecular weight was higher. The calcification extent of both collagenous materialand hydrophilic polyurethane was found to be in good correlation with the water absorptioncapacity of these biomaterials [116].Glutaraldehyde fixation of xenografts, which has been used to stabilize the tissue, can lead toseveral biochemical and structural determinants for calcific deposition. First, removal ofproteoglycans and glycosaminoglycans owing to autolysis and storage, may enhance calcification.Glycosaminoglycans are natural inhibitors of calcification [66, 99]. Coupling of chondroitin sulfateto a collagen network reduces mineralization [66]. Furthermore, glutaraldehyde fixation results inpendant reactive aldehyde groups and polymeric GA cross-links which can depolymerize. The freealdehyde groups seem to be the prime instigation of calcification. Storage of GA-fixed porcineaortic heart valves for more than 1 year resulted in a reduced content of free aldehyde groups anda diminished capacity of calcification [127]. Fixation of the tissue will also change the morphologyand the charges on the collagen molecules [128] which might be important parameters to complexcalcium ions. Another factor that must be considered in the pathogenesis of calcific deposits inbiomaterials is the presence of calcium-binding proteins. The affinity of these proteins for calciumis due to their content of carboxylated amino acids such as γ-carboxyglutamic acid (γ-CGA) [99].However, Hughes et al. did not find a correlation between γ-CGA and calcium deposition [129].In contrast, Shen et al. associated calcification with osteopontin, a non-collagenous bone matrixphosphoprotein, found in the calcified areas of aortic heart valves [130].Basically, two types of approaches have been tried to reduce the bioprostheses-associatedcalcification owing to chemical changes of the implant after fixation. Implant modification to


- 25 -

prevent calcification, using either cross-linking methods other than glutaraldehyde, a slightlymodified GA procedure (high, low or zero-pressure) or a pre-treatment of glutaraldehyde cross-linked bioprosthetic cusps with various agents such as phosphocitrate, diphosphonates, detergents,protamine, metal salts, and amino acids. The second approach comprises local delivery of drugs,such as diphosphonates from a polymeric matrix [131].Stress-induced collagen degeneration contributes to valve degeneration. Sachs et al. demonstratedthat the aortic valve microstructure was affected by the applied transvalvular pressure duringfixation with glutaraldehyde. Changes in fiber alignment were observed and the difference inorientation between the fibrosa and the ventricularis was almost completely indistinguishable athigher transvalvular pressures (60 mm Hg) [110] A strong correlation between holographicanomalies and calcification of porcine bioprostheses was found [132]. Thubrikar et al. implantedGA cross-linked valves in the aortic position in calves. They concluded that stresses in porcinebioprostheses were greater in the commissural region than at the base, and were compressive onthe aortic surface of the leaflet. Calcification started in the area of leaflet flexion. Earliest depositswere localized within collagen cords. Hence, calcification began in the areas of greatest stress. Inporcine xenografts, calcification was present where collagen fibers were likely to have beendamaged by compressive stresses. Therefore, calcification can be inhibited by reducing functionalstresses through the modification of design and tissue properties [133]. In order to reduce theinternal mechanical stiffness of the tissue, alterations in tissue preparation techniques, includinglow-pressure fixation, and radially altered design configurations using pericardial tissue, may beefficacious [115]. Dynamic fixation with glutaraldehyde instead of the usual static methodimproves the viscoelastic properties, and the tissue properties were more close to the original ones[134]. Pressure fixation (100 mm Hg) of pericardium resulted in a material which has stress-straincharacteristics more similar to the natural than to a non-pressure fixed material [106]. The sameresults were obtained by Mayne et al., who compared aortic valves fixed with a pressure of 4 mmHg to valves fixed at a pressure of 80-100 mm Hg [135]. On the other hand, Christie found thatGA-fixation greatly modified the mechanics of the leaflet but that this can be minimized by fixationwith no pressure difference across the closed valve. Zero pressure fixed leaflets were much softerand extensible [136].

The effect of different cross-linking methods on the extent of calcification was investigated bysubcutaneous implantation of cross-linked dermal sheep collagen in rats. Hexamethylenediisocyanate cross-linked dermal sheep collagen resulted in formation of calcific deposits, whichwere found intracellularly, in macrophages and fibroblasts. Moreover, GA cross-linked collagenshowed excessive Ca/P crystals [137]. On the other hand, dermal sheep collagen cross-linked witha water-soluble carbodiimide, EDC, or via the acyl-azide activation method displayed a moderatecalcification after 6 weeks, and hardly any after prolonged implantation times [33, 68]. Gong et al.preserved bioprosthetic valves in different ways. They compared formaldehyde and glutaraldehydecross-linked tissue with only glycerol treated valves. After subcutaneous implantation in rats for70 d, glycerol treated valves hardly show calcification, while aldehyde fixed tissue calcified.Therefore, they stated that aldehydes play an important role in calcification [117]. Other groups[118] found that the amount of GA incorporated in the tissue controls the cross-link density,which in turn directly determines tissue stability and calcification. Furthermore, the effect of

Chapter 2

- 26 -

cytotoxicity of GA is still a matter of debate. Bovine pericardium which was stabilized with GAinhibited the growth of endothelial cells on the material, whereas glycerol treated tissue showeduniform growth of cells. This difference was ascribed to the leakage of cytotoxic GA moleculesfrom the pericardial tissue [138].These results enhanced the search for alternative cross-link agents. The group of poly glycidylethers has got attention because of its good cross-linking ability and less stiffening of the tissue.Implantation of ethyleneglycol diglycidyl ether cross-linked valves in 3 week-old Spraque-Dawleymale rats, did not result in considerably lower calcium levels than glutaraldehyde-fixed controls inboth the leaflets and wall [53]. Moreover, only a moderate reduction in calcium levels (115 vs.206 µg/mg tissue ) was found by Tingfei et al. when comparing epoxy cross-linked valves withGA ones [139]. On the contrary, Shen et al. found no calcification after they implanted multi-functional poly glycerol glycidyl ether cross-linked valves in the mitral position of a juvenile sheep[57, 140]. Cross-linking of heart valves by several poly epoxy compounds resulted in a very lowcontent of calcium after subcutaneous implantation in 4 week-old rats. An average of 0.96 µg/mgtissue was found, while 140.7 µg calcium per mg tissue was found in glutaraldehyde fixed tissue[58].Other calcification studies have been carried out on chemically modified glutaraldehyde fixedtissue. Free residual aldehyde groups can be blocked by reaction with small amine containingmolecules, such as amino acids. Pretreatment of glutaraldehyde-fixed pericardium with L-lysine,L-glutamic acid, L-glutamine, or L-arginine does not appear to prevent calcification. After 60 d ofsubcutaneous implantation, all groups showed comparable high levels of calcific depositioncompared to the glutaraldehyde controls (110 vs. 100 mg Ca2+/g tissue) [141] Zilla et al.concluded that the content of free aldehyde groups and polymeric GA cross-links have to bereduced. Furthermore, they treated glutaraldehyde fixed tissue with amine containing compoundssuch as L-glutamic acid which resulted in a reduction in calcification of the aortic wall [142].Postfixation treatment of porcine aortic valves by monosodium glutamate significantly reducescalcification during a 90 d implantation in rats [143]. The use of α-amino oleic acid (AOA), whichis a potent, non-toxic and biocompatible anticalcification agent, has been shown to be the mosteffective for glutaraldehyde fixed valves as proved by implantation in rats and juvenile sheep [127,144-146]. A large reduction in calcium level of aortic cusps was found (7.7 versus 129 mg/gtissue) after implantation using a juvenile sheep model. Unfortunately, aortic wall calcification wasnot affected by AOA treatment [146].The hypothesis that an impaired balance between positively and negatively charged amino acidresidues resulted in affinity sites to Ca2+ ions was verified by Golomb and Ezra [128]. Theycovalently coupled protamine sulfate, a polybasic peptide, to pericardial tissue with formaldehyde.A significant reduction in calcific deposition compared to glutaraldehyde-fixed tissue was found.Coupling of amine containing macromolecules is another strategy to prevent formation of calciumapathetic crystals. Bovine pericardium treated with 5% sodium chloride-trypsin-glutaraldehyde-chitosan did not calcify at 12 weeks in the rat (calcium: 1.1 ± 0.27 mg/g). Complete elimination ofcellular elements is achieved by a sodium chloride-trypsin mixture. The amine groups of chitosancan react with the free aldehyde groups after glutaraldehyde cross-linking and can also act as aspace filler, reducing the availability of physical niches for crystal growth [147]. Comparableresults were obtained with heparin as a coupling and space filling agent [148]. In addition, cross-


- 27 -

linking of pericardium with alginate azide and further modification by grafting with poly(glycidylmethacrylate-butyl acrylate) reduced the calcium levels to negligible values due to the space-fillingability and hence the blocking of potential nucleation sites [149].Bis(acyl)phosphonates inhibit hydroxyapatite formation and can be of potential importance forclinical applications [150]. Covalent coupling of [3-amino-10-hydroxypropane-1,1-diphosphonicacid (3-APD)] to glutaraldehyde treated valves, inhibited the calcification of the cusps and aorticwall [32]. Consequently, pretreatment of fixed allograft aortic wall with aminopropanehydroxy-diphosphonate [APDP] reduced calcification provided that the ADPD concentration was highenough (1 mM or higher) [120].Another group of anticalfication agents are natural inhibitors of hydroxyapatite crystal growthsuch as AlCl3 and FeCl3. This inhibition may be related to cationic binding of Fe3+ and Al3+ tocalcium phosphate and hence it appears that they disturb the crystal growth [120]. Treatment ofaortic allograft tissue with 0.1 M AlCl3 or FeCl3 resulted in an almost complete inhibition ofcalcification [151]. Calcium phosphate formation was also prevented by vitamins and plateletinhibitors, possibly due to a reduction in transport of Ca2+-ions to the tissue surface [152].Sequential application of AOA and FeCl3 synergenically reduced aortic wall calcification moreeffectively than either of these agents alone [153].Removal of antigenic substances such as cellular elements and soluble proteins [131, 154] orpotential nucleation sites for hydroxyapatite crystals is an additional strategy. Selective removal of(phospho)lipids from the matrix by treating the tissue with chloroform/methanol mixturesappeared to be very effective since calcium levels of GA fixed valves were reduced from 188 to5.5 µg/mg tissue after 60 d of subcutaneous implantation in rats [155]. Sodium dodecyl sulfatetreatment of cross-linked valves inhibited mineralization [156, 157], which is most likely due tophospholipid extraction [157]. Removal of all cellular substances by trypsin/ 5% NaCl resulted inan almost complete prevention of calcification [131]. In addition, dimethylsulfoxide (DMSO) hasbeen found to mitigate calcification of biological tissue [158]. The exact reason why DMSO,which is used as a cryoprotective agent, acts as an anticalcification agent is not known, but mayinclude the extraction of antigenic phospholipids in fresh biological tissue. Moreover, DMSOappears to nullify the epitactic or nucleation sites that can attract Ca2+-ions [159]. Treatment ofglutaraldehyde-fixed bovine pericardium with ethanol completely inhibits calcification in a ratsubdermal model. Vyavahare et al. assume that the anticalcification effect of ethanol is not onlydue to lipid extraction but also to the interaction with the structural proteins leading toconformation changes and in an altered water content as well [160].Immunological processes of the host do not contribute to the mineralization and deteriorationprocess [161, 162].

The second approach of mitigating bioprosthetic calcification is local delivery of drugs.Application of chitosan beads loaded with FeCl3 as a local drug delivery system appeared to besuccessful. An inhibition of calcium deposition for 30-40 % compared to the bare implants, wasobtained after implantation of glutaraldehyde fixed pericardium, dura mater and fascialatacompared to bare implants [131]. Controlled delivery of NaEDHP or EDHP polymer from asilicone rubber matrix reduced the calcium levels from 225 to 10 µg/ mg allograft tissue [151].

Chapter 2

- 28 -

Finally, controlled release of trivalent metal ions from natural sequestering molecules such asferritin showed very promising results in diminishing calcification [163].

This overview demonstrates that the mechanism of calcification and the factors which induce thecalcium hydroxyapatite crystal formation are still not fully understood and many hypothesis areproposed. The cross-linking process and the presence of foreign proteins and cells appear to havea key role in calcification of tissue heart valves.Two strategies are followed in this thesis. The first approach was the development of new cross-link methods based on bisepoxy compounds, carbodiimides and combinations of them. Becausecollagen is the major structural component of valve tissue, a model tissue of 100 % type Icollagen, namely dermal sheep collagen, was used. The effect of the cross-linking method on thematerial properties and calcification behavior was evaluated. Analogously, porcine aortic heartvalves were cross-linked with the same procedures and the effect on the in-vitro stability, thebiocompatibility and the calcification behavior were investigated.The second approach was to remove cellular elements form the cross-linked valves by treatmentwith a detergent mixture. The influence of this extraction step on the extent of calcium depositionwas studied.

References

1. R.F. Oliver, "Collagen as a dermal implant", in "Biocompatibility of Tissue Analogs", Ed. by D.F.Williams, CRC Press, Inc., Baca Raton, Florida (1985)

2. E. Khor, "Methods for the treatment of collagenous tissues for bioprostheses", Biomaterials, 18(2) pp. 95-105 (1997).

3. A. Jayakrishnan and S.R. Jameela, "Glutaraldehyde as a fixative in bioprosthetic and drug deliverymatrices", Biomaterials, 17 pp. 471-484 (1996).

4. M.G. Dunn, P.N. Avasarala, and J.P. Zawadsky, "Optimization of extruded collagen fibers for ACLreconstruction", J. Biomed. Mat. Res., 27 pp. 1545-1552 (1993).


6. M.E. Nimni, "Collagen: Molecular structure and biomaterial properties", in "Encyclopedic handbook ofbiomaterials and bioengineering Part A: Materials", Ed. by D.L. Wise, et al., Marcel Dekker Inc., NewYork. p. 1229-1243 (1995)

7. G.W. Hastings, "Cardiovascular biomaterials", London, England: Springer-Verlag, Ltd.(1992).8. M.E. Nimni, D.T. Cheung, B. Strates, M. Kodama, and K. Sheikh, "Bioprosthesis derived from cross-

linked and chemically modified collagenous tissues", in "Collagen Volume III: Biotechnology", Ed. byM.E. Nimni, CRC Press, Inc., Boca Raton, Florida, USA (1988)

9. W.D. Comper, "Extracellular matrix Volume 2: Molecular components and interactions", Amsterdam,NL: Overseas Publishers Association, B.V.(1996).

10. J. Vaculik, "Die kollagenstruktur von Rindhaut and Rindleder bei Betrachtung mit Hilfe der REM", LederSchuhe Lederwaren, 3 pp. 132-138 (1983).

11. M.E. Nimni and R.D. Harkness, "Molecular structure and functions of collagen", in "Collagen Volume I:Biochemistry", Ed. by M.E. Nimni, CRC Press, Inc., Boca Raton, Florida, USA (1988)

12. E. Heidemann, "Fundamentals of leather manufacturing", 1st ed., Munchen, Germany: HanserPublishers.(1993).

13. K. Kadler, "The structure of collagens", Protein Profile, II pp. 508-523 (1995).14. J.B. Weiss and S. Ayed, "An introduction to collagen", in "Collagen in health and disease", Ed. by J.B.

Weiss and M.I.V. Jayson, Churchill Livingstone, Edinburgh (1982)15. C.A. Miles, T.V. Burjanadze, and A.J. Bailey, "The kinetics of the thermal denaturation of collagen in

unrestrained rat tail tendon determined by differential scanning calorimetry", J. Mol. Biol., 245 pp. 437-446 (1995).


- 29 -

16. H.P. Germann and E. Heidemann, "A synthetic model of collagen: An experimental investigation of thetriple-helix stability", Biopolym., 27 pp. 157-163 (1988).

17. I.V. Yannas, "Biologically active analogous of the extracellular matrix: Artificial skin and nerves", Angew.Chem. Int. Ed. Engl., 29 pp. 20-35 (1990).

18. T. Nemetschek, "Die Hierarchie der Kollagenstruktur", Verh. Anat. Ges., 75 pp. 47-60 (1981).19. D.J.S. Hulmes, T.J. Wess, D.J. Prockop, and P. Fratzel, "Radial packing, order, and disorder in collagen

fibrils", Biophys. J., 68 pp. 1661-1670 (1995).20. S.T. Li, "Biological biomaterials: Tissue-derived biomaterials", in "The biomedical engineering

handbook", Ed. by J.D. Bronzino, CRC Press Inc in cooperation with IEEE Press, Boca Raton, Florida. p.627-647 (1995)

21. F.H. Silver, G.D. Pins, M.-C. Wang, and D. Christiansen, "Collagenous biomaterials as models of tissue-inducing implants", in "Encyclopedic handbook of biomaterials and bioengineering", Ed. by D.L. Wise, etal., Marcel Dekker, Inc., New York. p. 1245-1266 (1995)

22. D.R. Eyre, M.A. Paz, and P.M. Gallop, "Crosslinking in collagen and elastin", Ann. Rev. Biochem., 53 pp.717-748 (1984).


24. L.H.H. Olde Damink, P.J. Dijkstra, M.J.A. v. Luyn, P.B. v. Wachem, P. Nieuwenhuis, and J. Feijen,"Glutaraldehyde as crosslinking agent for collagen based biomaterials", J. Mat. Sci.: Mat in Med., 6 pp.460-472 (1995).

25. J.M. Lee, D.W. Courtman, and D.B. Boughner, "The glutaraldehyde-stabilized porcine aortic valvexenograft. I. Tensile viscoelastic properties of the fresh leaflet material", J. Biomed. Mat. Res., 18 pp. 61-77 (1984).

26. D.M. Simmons and J.N. Kearney, "Evaluation of collagen cross-linking techniques for the stabilization oftissue matrices", Biotech. Appl. Biochem., 17 pp. 23-29 (1993).

27. I. Rault, V. Frei, D. Herbage, N. Abdul-Marak, and A. Huc, "Evaluation of different chemical methods forcross-linking collagen gels, films and sponges", J. Mat. Sci: Mat in Med., 7(4) pp. 215-222 (1996).

28. K. Tomihata, K. Burczak, K. Shirahi, and Y. Ikada, "Cross-linking and biodegradation of native anddenatured collagen", in "Polymers of biological and biomedical significance",ACS Symposium series 540

29. D.T. Cheung, N. Perelman, E.C. Ko, and M.E. Nimni, "Mechanism of cross-linking of proteins byglutaraldehyde III. Reaction with collagen in tissues", Conn. Tissue Res., 13 pp. 109-115 (1985).

30. T.W.G. Solomons, "Organic Chemistry", Fourth ed., New York, USA: John Wiley & Sons.(1988).31. M.N. Girardot, M. Torrianni, D. Dillehay, and J.M. Girardot, "Role of glutaraldehyde in calcification of

procine heart valves: comparing cusp and wall", J. Biomed. Mat. Res., 29 pp. 793-801 (1995).32. M.E. Nimni, D. Myers, D. Ertl, and B. Han, "Factors which affect the calcification of tissue-derived

bioprostheses", J. Biomed. Mat. Res., 35 pp. 351-357 (1997).33. P.B. v. Wachem, M.J.A. v. Luyn, L.H.H. Olde Damink, P.J. Dijkstra, J. Feijen, and P. Nieuwenhuis,

"Biocompatibility and tissue regenerating capacity of crosslinked dermal sheep collagen", J. Biomed. Mat.Res., 28 pp. 353-363 (1994).

34. L.L.H. Huang-Lee, D.T. Cheung, and M.E. Nimni, "Biochemical changes and cytotoxicity associated withthe degradation of polymeric glutaraldehyde derived crosslinks", J. Biomed. Mat. Res., 24 pp. 1185-1201(1990).

35. J.K. Law, J.R. Parsons, F.H. Silver, and A.B. Weiss, "An evaluation of purified reconstituted type Icollagen fibers", J. Biomed. Mat. Res., 23 pp. 961-977 (1989).

36. M. Grimm, E. Eybl, M. Grabenwöger, H. Spreitzer, W. Jäger, G. Grimm, P. Böck, M. Müller, and E.Wolner, "Glutaraldehyde affects biocompatibility of bioprosthetic heart valves", Surgery, 111 pp. 74-78(1992).

37. M.J. Morykwas, "In vitro properties of crosslinked, reconstituted collagen sheets", J. Biomed. Mat. Res.,24 pp. 1105-1110 (1990).

38. A. Simionescu, D. Simionescu, and R. Deac, "Lysine enhanced glutaraldehyde cross-linking ofcollagenous biomaterials", J. Biomed. Mat. Res., 25 pp. 1495-1505 (1991).

39. K.B. Hey, C.M. Lachs, M.J. Raxworthy, and E.J. Wood, "Crosslinked fibrous collagen for use as a dermalimplant: control of the cytotoxic effects of glutaraldehyde and dimethylsuberimidate", Biotech. Appl.Biochem., 12 pp. 85-93 (1990).

40. Y. Noishiki, H. Koyanagi, T. Miyata, and M. Furuse, Bioprosthetic valve, Patent EP 0 306 256 A2(1988).

Chapter 2

- 30 -

41. R. Tu, S.H. Shen, D. Lin, C. Hata, K. Thyagarajan, Y. Noishiki, and R.C. Quijano, "Fixation ofbioprosthetic tissues with monofunctional and multifunctional poly epoxy compounds", J. Biomed. Mat.Res., 28 pp. 677-684 (1994).


43. J.M. Lee, K. Thyagarajan, C.A. Pereira, J. McIntyre, and R. Tu, "Crosslinking of a prototype bovine arteryxenograft: Comparison of the effects of glutaraldehyde and poly epoxy compounds", Art. Org., 15 pp. 303-3 (1991).

44. Z. Tang and Y. Yue, "Crosslinkage of collagen by polyglycidyl ethers", ASAIO J., 41 pp. 72-78 (1995).45. H.-W. Sung, H.-L. Hsu, C.-C. Shih, and D.-S. Lin, "Cross-linking characteristics of biological tissue fixed

with monofunctional or multifunctional epoxy compounds", Biomaterials, 17(14) pp. 1405-1410 (1996).46. Nagase Chemicals Ltd., Catalog Denacol Epoxy compounds.47. R. Tu, R.C. Quijano, C.L. Lu, S. Shen, E. Wang, C. Hata, and D. Lin, "A preliminary study of the fixation

mechanism of collagen reaction with a polyepoxy fixative", Int. J. Art. Org., 16(7) pp. 537-544 (1993).48. E. Wang, K. Thyagarajan, R. Tu, C. Hata, S.H. Shen, and R.C. Quijano, "Evaluation of collagen

modification and surface properties of a bovine artery via polyepoxy compound fixation", Int. J. Art. Org.,16(7) pp. 530-536 (1993).

49. R. Tu, C.L. Lu, K. Thyagarajan, E. Wang, H. Nguyen, S. Shen, C. Hata, and R.C. Quijano, "Kinetic studyof collagen fixation with poly epoxy fixatives", J. Biomed. Mat. Res., 27 pp. 3-9 (1993).

50. L. Shechter and J. Wynstra, "Glycidyl ether reactions with alcohols, phenols, carboxylic acids, and acidanhydrides", Indust. Engin. Chem., 48(1) pp. 86-93 (1956).

51. L. Shechter, J. Wynstra, and R.P. Kurkjy, "Glycidyl ether reactions with amines", Indust. Engin. Chem.,48(1) pp. 94-97 (1956).

52. H.W. Sung, C.S. Hsu, and Y.S. Lee, "Physical properties of a porcine internal thoracic artery fixed with anepoxy compound", Biomaterials, 17(24) pp. 2357-2367 (1996).

53. D. Myers, "Stentless heart valves: Biocompatibility issues associated with new antimineralization andfixation agents", in "Stentless Bioprostheses", Ed. by A. Piwnica and S. Westaby, Isis Medical Media Ltd.,Oxford (1995)

54. T. Xi and F. Liu, "Effect of pretreatment with epoxy compounds on the mechanical properties of bovinepericardial bioprosthetic materials", J. Biomat. Appl., 7 pp. 61-75 (1992).

55. C. Hata, E. Wang, Y. Noishiki, R. Tu, H. Sung, and R.C. Quijano, "Evaluation of a polyepoxy compoundfixed biological vascular prosthesis and an expanded PTFE vascular graft", Art. Org, 16(3) pp. 263-266(1992).

56. J. Zhou, L.J. Quintero, M.N. Helmus, C. Lee, and R. Kafesjian, "Porcine aortic wall flexibility. Freshversus denacol fixed versus glutaraldehyde fixed", ASAIO J., 43 pp. M470-M475 (1997).

57. H.W. Sung, et al., "A newly developed porcine heart valve bioprosthesis fixed with an epoxy compound",ASAIO J., pp. 192-198 (1994).


59. C. Nishi, N. Nakajima, and Y. Ikada, "In vitro evaluation of cytotoxicity of diepoxy compounds used forbiomaterials", J. Biomed. Mat. Res., 29 pp. 829-835 (1995).


61. C.A. Pereira, J.M. Lee, and S.A. Haberer, "Effect of alternative crosslinking methods on the low strain rateviscoelastic properties of bovine pericardial material", J. Biomed. Mat. Res., 24 pp. 345-361 (1990).

62. G.E. Davies and G.R. Stark, "Use of dimethylsuberimidate, a crosslinking reagent, in studying the subunitstructure of oligomeric proteins", Proc. Nat. Acad. Sci. USA, 66 pp. 651-656 (1970).

63. R.L. Lundblad, "The chemical crosslinking of peptide chains", in "Chemical reagents for proteinmodification", Ed. by R.L. Lundblad, CRC Press, Boca Raton, Florida (1991)

64. S.S. Wong, "Reactive groups of proteins and their modifying agents", in "Chemistry of proteinconjugation and cross-linking", Ed. by S.S. Wong, CRC Press, Inc., Boca Raton, Florida. p. 7-48 (1991)



- 31 -

66. M.E. Nimni, D. Cheung, B. Strates, M. Kodana, and K. Sheikh, "Chemically modified collagen: A naturalbiomaterial for tissue replacement", J. Biomed. Mat. res., 21 pp. 741-771 (1987).

67. Z. Graberek and J. Gergely, "Zero-length cross-link procedure with the use of active esters", Analyt.Biochem., 185 pp. 131-135 (1990).

68. P.B. v. Wachem, M.J.A. v. Luyn, L.H.H. Olde Damink, P.J. Dijkstra, J. Feijen, and P. Nieuwenhuis,"Tissue regenerating capacity of carbodiimide-crosslinked dermal sheep collagen during repair of theabdominal wall", J. Artif. Organs, 17(4) pp. 230-239 (1994).

69. J.M. Lee, H.L. Edwards, C.A. Pereira, and I.S. Samii, "Cross-linking of tissue-derived biomaterials in 1-ethyl-3-(dimethylaminopropyl)-carbodiimide", J. Mat. Sci.: Mat. in Med., 7(9) pp. 531-542 (1996).

70. J.V. Staros, "N-hydroxysulfosuccinimide active esters: Bis(N-hydroxysulfosuccinimide) esters of twodicarboxylic acids are hydrophilic, membrane-impermeant, protein cross-linkers", Biochem., 22 pp. 3950-3955 (1982).

71. F. Scholnick, L.L. Liao, E.M. Brown, and S.H. Feairheller, "Crosslinking of collagen with dicarboxylicacids", J. Am. Leather Chem. Ass., 87 pp. 333-338 (1992).

72. H. Petite, I. Rault, A. Huc, P. Menasche, and D. Herbage, "Use of the acyl azide method for cross-linkingcollagen-rich tissues such as pericardium", J. Biomed. Mat. Res., 24 pp. 179-187 (1990).

73. K. Anselme, H. Petite, and D. Herbage, "Inhibition of calcification in vivo by acyl azide cross-linking of acollagen-GAG sponge", Matrix, 12 pp. 264-273 (1992).

74. Petite, Menasche, and Huc, Process for cross-linking of collagen by DPPA, the cross-linked collagenobtained therby, and biomaterials of collagen base thus cross-linked, Patent WO 9012055 France(1990).

75. H. Petite, V. Frei, A. Huc, and D. Herbage, "Use of diphenylphosphorylazide for cross-linking collagen-based materials", J. Biomed. Mat. Res, 28 pp. 159-165 (1994).

76. J.H. Bowes and C.W. Cater, "Crosslinking of collagen", J. Appl. Chem., 15 pp. 296-304 (1965).77. B. Brodsky, R.A. Berg, G. Avigad, E. Eikenberry, M. Jain, and S. Tanaka, Collagen-based matrices

ribose crosslinked, Patent 4971954 US (1990).78. K. Weadock, R.M. Olsen, and F.H. Silver, "Evaluation of collagen crosslinking techniques", Biomat. Med.

Dev. Art. Org., 11(4) pp. 293-318 (1983-84).79. I.V. Yannas and A.V. Tobolsky, "Cross-linking of gelatin by dehydration", Nature, 215 pp. 509-510

(1967).80. K.S. Weadock, E.J. Miller, E.L. Keuffel, and M.G. Dunn, "Effect of physical cross-linking methods on

collagen-fiber durability in proteolytic solutions", J. Biomed. Mat. Res., 32 pp. 221-226 (1996).81. M.A. Moore, I.K. Bohachensky, D.T. Cheung, B.D. Boyan, W.M. Chen, R.R. Bickers, and B.K. McIlroy,

"Stabilization of pericardial tissue by dye-mediated photooxidation", J. Biomed. Mat. Res., 28 pp. 611-618(1994).

82. M.A. Moore, W.M. Chen, R.E. Phillips, I.K. Bohachevsky, and B.K. McIlroy, "Shrinkage temperatureversus protein extraction as a measure of stabilization of photooxidized tissue", J. Biomed. Mat. Res., 32pp. 209-214 (1996).

83. P.J. Flory and R.R. Garrett, "Phase transitions in collagen and gelatin systems", J. Am. Chem. Soc., 80 pp.4836-4845 (1958).

84. J.M. Lee, C.A. Pereira, D. Abdulla, W.A. Naimark, and I. Crawford, "A multi-sample denaturationtemperature tester for collagenous biomaterials", Med. Eng. Phys., 17 pp. 115-121 (1995).

85. IUP/16, "Measurement of shrinkage temperature", J. Soc. Leather Techn. Chem., pp. 122-126 (1963).86. T.N.v. Gulik, "Processed sheep dermal collagen as a biomaterial", Thesis University of Amsterdam, The

Netherlands (1983).87. P.E. McClain and E.R. Wiley, "Differential scanning calorimeter studies of the thermal transitions of

collagen", J. Biol. Chem., 247(3) pp. 692-697 (1972).88. B. Twombly, A.T. Miller, and U. Hess, "Calorimetric and dynamic mechanical analysis of thermal

transitions in collagen", Mol. Cryst. Liq. Cryst., 265 pp. 171-180 (1995).89. W.K. Loke and E. Khor, "Validation of the shrinkage temperature of animal tissue for bioprosthetic heart

valve application by differential scanning calorimetry", Biomaterials, 16(3) pp. 251-258 (1995).90. M.L. Kakade and I.E. Liener, "Determination of available lysine in proteins", Analyt. Biochem., 27 pp.

273-280 (1969).91. L.H.H. Olde Damink, P.J. Dijkstra, M.J.A.v. Luyn, P.B.v. Wachem, P. Nieuwenhuis, and J. Feijen,

"Influence of ethylene oxide gas treatment on the in vitro degradation behaviour of dermal sheepcollagen", J. Biomed. Mat. Res., 29(2) pp. 149-155 (1995).

Chapter 2

- 32 -

92. M.M. Black, T. Chochrane, P.V. Lawford, H. Reul, and A. Yoganathan, "Design and flow characteristics",in "Replacement cardiac valves", Ed. by E. Bodnar and R. Frater, Pegamon Press, Inc., New York (1991)

93. B.S. Goldman, T.E. David, D.F.D. Rizzo, J. Sever, and J. Bos, "Stentless porcine bioprostheses for aorticvalve replacement", J. Cardiovasc. Surg., 35 (Suppl. 1 to No. 6) pp. 105-110 (1994).

94. D. Ross, "From homograft to stentless bioprostheses", in "Stentless bioprostheses", Ed. by A. Piwnica andS. Westaby, Isis Medical Media, Oxford (1995)

95. F.J. Schoen, R.J. Levy, B.D. Ratner, M.D. Lelah, and G.W. Christie, "Materials considerations forimproved cardiac valve prostheses", in "Replacement cardiac valves", Ed. by E. Bodnar and R. Frater,Percamon Press, Inc., New York. p. 357-382 (1991)

96. G.M. Bernacca, T.G. Mackay, R. Wilkinson, and D.J. Wheatley, "Polyurethane heart valves: Fatiguefailure, calcification and polyurethane structure", J. Biomed. Mat. Res., 34 pp. 371-379 (1997).

97. M. Thubrikar, "The aortic valve", CRC Press, Boca Raton, Florida (1983).98. S.L. Hilbert, M. Jones, and V.J. Ferrans, "Flexible leaflet replacement heart valves", in "Encyclopedic

handbook of biomaterials and bioengineering Part B: Applications", Ed. by D.L. Wise, et al., MarcelDekker, Inc., New York. p. 1111-1152 (1995)

99. V.J. Ferrans, S.L. Hilbert, and M. Jones, "Biomaterials", in "Replacement cardiac valves", Ed. by E.Bodnar and R.W.M. Frater, CRC Press, Boca Raton. Florida (1991)

100. Heart Lab Web-page, Background on heart valves and their function. Basic heart physiology (1997).101. R.I. Bashey and S.A. Jimenez, "Collagens in heart valves", in "Collagen volume I: Biochemistry", Ed. by

M.E. Nimni, CRC Press, Inc., Boca Raton, Florida. p. 257-274 (1988)102. S.L. Hilbert, V.J. Ferrans, and M. Jones, "Tissue-derived biomaterials and their use in cardiovascular

prosthetic devices", Med. Prog. through Techn., 14 pp. 115-163 (1988).103. M.A. Haralson and J.R. Hassell, "The extracellular matrix-an overview", in "Extracellular matrix. A

practical approach", Ed. by M.A. Haralson and J.R. Hassell, Oxford University Press. p. 1-30 (1995)104. P.K. Bajpai, "Immunological aspects of treated natural tissue prostheses", in "Biocompatibility of tissue

analogs", Ed. by D.F. Williams, CRC Press, Inc., Boca Raton, Florida. p. 5-25 (1985)105. M.A. Flomenbaum and F.J. Schoen, "Effects of fixation back pressure and antimineralization treatment on

the morphology of porcine aortic heart valves", J. Thrac. Cardiovasc. Surg., 105 pp. 154-164 (1993).106. J.M. Lee, D.R. Boughner, and D.W. Courtman, "The glutaraldehyde-stabilized porcine aortic valve

xenograft. II. Effect of fixation with or without pressure on the tensile viscoelastic properties of leafletmaterials", J. Biomed. Mat. Res., 18 pp. 79-88 (1984).

107. P.F. Gratzer and J.M. Lee, "Altered mechanical properties in aortic elastic tissue usingglutaraldehyde/solvent solutions of various dielectric constant", J. Biomed. Mat. Res., 37 pp. 497-507(1997).

108. I. Vesely, "A mechanism for the decrease in stiffness of bioprosthetic haert valve tissues after cross-linking", ASAIO J., 42 pp. 993-999 (1996).

109. S.L. Hilbert, L.C. Sword, K.F. Batchelder, M.K. Barrick, and V.J. Ferrans, "Simultaneous assessment ofbioprosthetic heart valve biomechanical properties and collagen crimp length", J. Biomed. Mat. Res., 31pp. 503-509 (1996).

110. M.S. Sachs, D.B. Smith, and E.D. Hiester, "The aortic valve microstucture: Effects of transvalvularpressure", J. Biomed. Mat. Res., 41 pp. 131-141 (1998).

111. I. Vesely and R. Noseworthy, "Micromechanics of the fibrosa and the ventricularis in aortic valve leaflets",J. Biomech., 25(1) pp. 101-113 (1992).

112. P. Zioupos, J.C. Barbanel, and J. Fisher, "Anisotropic elasticity and strength of glutaraldehyde fixedbovine pericardium for use in pericardial bioprostheticvalves", J. Biomed. Mat. Res., 28 pp. 49-57 (1994).

113. K.J. Lochie, J. Fisher, N.P. Juster, G.A. Davies, and K. Wattersonn, "Biomechanics of glutaraldehyde-treated porcine aortic roots and valves", J. Thorac. Cardiovasc. Surg., 108 pp. 1037-1042 (1994).

114. D.W. Courtman, C.A. Pereira, S. Omar, S.E. Langdon, J.M. Lee, and G.J. Wilson, "Biomechanical andultrastructural comparison of cryopreservation and a novel cellular extraction of porcine aortic valveleaflets", J. Biomed. Mat. Res., 29 pp. 1507-1516 (1995).

115. F.J. Schoen, H. Harasaki, K.M. Kim, and H.C. Anderson, "Biomaterial-associated calcification: Pathology,mechanisms and strategies for prevention", J. Biomed. Mat. Res., 22(A1) pp. 11-36 (1988).

116. G. Golomb, A. Barashi, D. Wagner, and O. Nachmias, "In-vitro and in-vivo models for the study of therelationship between hydrophilicity and calcification of polymeric and collageneous biomaterials", Clin.Mat., 13 pp. 61-69 (1993).

117. G. Gong, Z. Ling, E. Seifter, S.M. Factor, and R.W.M. Frater, "Aldehyde tanning: The villain inbioprosthetic calcification", Eur. J. Cardio-thorac. Surg., 5 pp. 288-293 (1991).


- 33 -

118. G. Golomb, F.J. Schoen, M.S. Smith, J. Linden, M. Dixon, and R.J. Levy, "The role of glutaraldehydeinduced cross-links in calcification of bovine pericardium used in cardiac valve bioprostheses", Am. J.Pathol., 127 pp. 122-130 (1987).

119. F.J. Schoen, D. Hirsch, R.W. Bianco, and R.J. Levy, "Onset and progression of calcification in porcineaortic bioprosthetic valves implanted as orthotopic mitral valve replacements in juvenile sheep", J. Thorac.Cardio. Surg., 108(5) pp. 880-887 (1994).


121. L.M. Biedrzychi, E. Lerner, R.J. Levy, and F.J. Schoen, "Differential calcification of cusps and aortic wallof failed stented porcine aortic valves", J. Biomed. Mat. Res., 34 pp. 411-415 (1997).

122. W.J. Mako and I. Vesely, "Regional distribution of calcification in aortic valve cusps", ASAIO J., 42 pp.M365-367 (1996).

123. L. Maxwell, J.B. Gavin, and B.G. Barratt-Boyes, "Differences between heart valve allografts andxenografts in the incidence and initiation of dystrophic calcification", Pathology, 21 pp. 5-10 (1989).

124. Y.S. Lee, "Calcification of porcine bioprostheses implanted into human heart: High resolution-SEM andSTM investigation", J. Electron. Micr., 43 pp. 131-140 (1994).

125. I. Vasin, I.B. Rosanova, and V.I. Sevastianov, "The role of proteins in the nucleation and formation ofcalcium-containing deposits on biomaterial surfaces", J. Biomed. Mat. Res., 39 pp. 491-497 (1998).

126. S.F.A. Hossainy and J.A. Hubbell, "Molecular weight dependence of calcification of PEG hydrogels",Biomaterials, 15(11) pp. 921-925 (1994).

127. M.N. Girardot, J.M. Girardot, and M. Torrianni, "Alpha-aminooleic acid (AOA) anticalcification effectson glutaraldehyde-fixed heart valves: shelf-life studies", in "New horizons and the future of heart valveprostheses", Ed. by S. Gabbay and R.W.M. Frater, Silent Partners Inc., Austin (1994)


129. H. Hughes, L.S. Lipton, D. Feuchuk, G. Prakhakar, H.Y. Aboul-Enein, and C.M.G. Duran,"Glutaraldehyde, gamma-carboxyglutamic acid and calcium in explanted bioprosthetic heart valves", J.Heart Valve Disease, 3 pp. 111-116 (1994).

130. M. Shen, P. Marie, D. Farge, S. Carpentier, C.d. Pollak, M. Hott, L. Chen, B. Martinet, and A. Carpentier,"Osteopontin is associated with bioprosthetic heart valve calcifiaction in humans", C. R. Acad. Sci. III,320(1) pp. 49-57 (1997).

131. T. Chandy, M. Mohanty, A. John, S.B. Rao, R. Sivakumar, C.P. Sharma, and M.S. Valiathan, "Structuralstudies on bovine bioprosthetic tissues and theit in vivo calcification: prevention via drug delivery",Biomaterials, 17 pp. 577-585 (1996).

132. B. Glasmacher, M. Deiwick, H. Reul, H. Knesch, D. Keus, and G. Rau, "A new in vitro test method forcalcification of bioprosthetic heart valves", Int. J. Art. Organs, 20(5) pp. 267-271 (1997).

133. M. Thubrikar, J.D. Deck, J. Aouad, and S.P. Nolan, "Role of mechanical stress in calcification of aorticbioprosthetic valves", J. Thorac. Cardiovasc. Surg., 86 pp. 115-125 (1983).

134. A.C. Duncan, D. Boughner, and I. Vesely, "Viscoelasticity of dynamically fixed bioprosthetic valves. II.Effect of glutaraldehyde concentration", J. Thorac. Cardiovasc. Surg., 113 pp. 302-310 (1997).

135. A.S. Mayne, G.W. Christie, B.H. Smaill, P.J. Hunter, and B.G. Barratt-Boyes, "An assessment of themechanical properties of leaflets from four second-generation porcine bioprostheses with biaxial testingtechniques", J. Thorac. Cardiovasc. Surg, 98 pp. 170-180 (1989).

136. G.W. Christie, "Anatomy of aortic heart valve leaflets. The influence of glutaraldehyde fixation onfunction", Eur. J. Cardio-thorac. Surg., 6 Suppl 1: pp. S25-32 (1992).

137. P.B. v. Wachem, M.J.A. v. Luyn, L.H.H. Olde. Damink, J. Feijen, and P. Nieuwenhuis, "Tissueinteractions with dermal sheep collagen implants: A transmission electron microscopical evaluation",Cells and Mat., 1(3) pp. 251-263 (1991).

138. D. Hoffman, G. Gong, K. Liao, F. Macaluso, S.D. Nikolic, and R.W.M. Frater, "Spontaneous hostendothelial growth on bioprostheses. Influence of fixation", Circulation, 86([Suppl. II]) pp. II75-II79(1992).

139. X. Tingfei, M. Jiazhen, T. Wenhua, L. Xuehui, L. Shuhui, and X. Baosha, "Prevention of tissuecalcification on bioprosthetic heart valve by using epoxy compounds: A study of calcification tests in vitroand in vivo", J. Biomed. Mat. Res., 26 pp. 1241-1251 (1992).

140. S.H. Shen, H.W. Sung, R. Tu, C. Hata, D. Lin, Y. Noishiki, and R.C. Quijano, "Characterization ofpolyepoxy compound fixed porcine heart valve bioprostheses", J. Appl. Biomat., 5 pp. 159-162 (1994).

Chapter 2

- 34 -

141. E. Jorge-Herrero, P. Fernandez, C. Escudero, J.M. Garcia-Páez, and J.L. Castillo-Olivarez, "Calcificationof pericardial tissue pretreated with different amino acids", Biomaterials, 17 pp. 571-575 (1996).

142. P. Zilla, L. Fullard, P. Trescony, J. Meinhart, D. Bezuidenhout, M. Gorlitzer, P.Human, and U.v. Opell,"Glutaraldehyde detoxification of aortic wall tissue: A promising perspective for emerging bioprostheticvalve concepts", J. Heart Valve Dis., 6 pp. 510-520 (1997).

143. K. Liao, E. Seifter, D. Hoffman, E.L. Yeller, and R.W.M. Frater, "Improved postfixation treatment ofglutaraldehyde fixed porcine aortic valves by monosodium glutamate", Art. Org., 6(3) pp. 267-272 (1990).

144. W. Chen, J.D. Kim, F.J. Schoen, and R.J. Levy, "Effect of 2-AOA exposure conditions on the inhibition ofcalcification of glutaaldehyde cross-linked porcine aortic valves", J. Biomed. Mat. Res., 28 pp. 1485-1495(1994).

145. J.P. Gott, "Calcification of porcine valves: A successful new method of antimineralization", Ann. Thorac.Surg., (1994).


147. J. Chanda, "Anticalcification treatment of pericardial prostheses", Biomaterials, 15(6) pp. 465-469 (1994).148. J. Chanda, R. Kuribayashi, and T. Abe, "Heparin in calcification prevention of porcine pericardial

bioprostheses", Biomaterials, 18(16) pp. 1109-1114 (1997).149. C. Shanthi and K.P. Rao, "New treatments using alginate in order to reduce the calcification of bovine

bioprosthetic heart valve tissue", J. Biomater. Sci. Polymer Edn., 8(12) pp. 919-930 (1997).150. G. Golomb, A. Schlossman, H. Saadeh, M. Levi, J.M. v. Gelder, and E. Breuer, "Bisacylphosphonates

inhibit hydroxyapatite formation and dissolution in vitro and dystrophic calcification in vivo",Pharmaceut. Res., 9 pp. 143-148 (1992).


152. S.C. Vasudev, T. Chandy, and C.P. Sharma, "Glutaraldehyde treated bovine pericardium: Changes incalcification due to vitamins and platelet inhibitors", Art. Org., 21(9) pp. 1007-1013 (1997).

153. W. Chen, F.J. Schoen, D.J. Myers, and R.J. Levy, "Synergistic inhibition of calcification of porcine aorticroot with preincubation in FeCl3 and AOA in a rat subdermal model", J. Biomed. Mat. Res. (Appl.

Biomat.), 38 pp. 43-48 (1997).154. J. Chanda, S.B. Rao, M. Mohanty, A.V. Lal, C.V. Muraleedharan, G.S. Bhuvaneshwar, and M.S.

Valiathan, "Prevention of calcification of tissue valves", Art. Org., 18(10) pp. 752-757 (1994).155. E. Jorge-Herrero, P. Fernandez, N. de la Torre, C. Escudero, J.M. Garciá-Páez, J. Buján, and J.L. Castillo-

Olivarez, "Inhibition of the calcification of porcine valve tissue by selective lipid removal", Biomaterials,15(10) pp. 815-820 (1994).

156. D.J. Lentz, E.M. Pollock, D.B. Olsen, and E.J. Andreus, "Calcification of extrinsic valves", Trans. Am.Soc. Artif. Intern. Org., 28 pp. 494-497 (1982).

157. D. Hirsch, J. Drader, T.J. Thomas, F.J. Schoen, J.T. Levy, and R.J. Levy, "Inhibition of calcification ofglutaraldehyde pretreated porcine aortic valve cusps with sodium dodecyl sulfate", J. Biomed. Mat. res., 27pp. 1477-1484 (1993).

158. E. Khor, A. Wee, W.K. Loke, and B.L. Tan, "Dimethylsulfoxide as an anticalcification agent forglutaraldehyde-fixed biological tissue.", J. Mat. Sci.: Mat. in Med., 7 pp. 691-693 (1996).

159. E. Khor, A. Wee, T.C. Feng, and D.C.L. Goh, "Glutaraldehyde-fixed biological tissue calcification:effectiveness of mitigation by DMSO", J. Mat. Sci.: Mat. in Med., 9 pp. 39-45 (1998).

160. N. Vyavahare, D. Hirsch, E. Lerner, J.Z. Bashin, F.J. Schoen, R. Bianco, H.S. Kruth, R. Zand, and R.J.Levy, "Prevention of bioprosthetic valve calcification by ethanol preincubation: Efficacy and mechanisms",Circulation, 95 pp. 479-488 (1997).

161. R.J. Levy, F.J. Schoen, and S.L. Howard, "Mechanism of calcification of porcine bioprosthetic aortic valvecusps: role of T-lymphocytes", Am. J. Cardiol., 52 pp. 629-631 (1983).

162. R.N. Mitchell, R.A. Jonas, and F.J. Schoen, "Pathology of explanted cryopreserved heart valves:comparison with aortic valves from orthotopic heart transplants", J. Thorac. Cardiovasc. Surg., 115 pp.118-127 (1998).

163. M.E. Ogle and R.F. Schroeder. Controlled release of trivalent metal ions via natural sequesteringmolecules as a bioprosthetic anti-calcification therapy. in Society for Biomaterials, 24th Annual Meeting,San Diego, USA (1998).

Cross-l nk ng and mod f cat on of dermal sheep collagen us ng 1,4-butaned ol d glyc dyl ether

- 35 -

Chapter 3

Cross-linking and modification of dermal sheep

collagen using 1,4-butanediol diglycidyl ether




ABSTRACT

Cross-linking of dermal sheep collagen was accomplished using 1,4-butanediol diglycidyl ether (BDDGE). Theepoxy groups of this reagent are able to react with both amine and carboxylic acid groups. Studies showed that nooligomeric cross-links could be formed. However, hydrolysis of the epoxide groups played a role in the cross-linkmechanism especially under acidic reaction conditions.The effects of concentration, pH, time and temperature were studied. Utilization of a 4 wt% BDDGE instead of 1wt%, resulted in a faster reaction but in a lower cross-link efficacy. The shrinkage temperature (Ts) of the collagencross-linked with 4 wt% BDDGE was raised from 46 to 66 °C, while the content of amine groups was reduced from32 to 6 (n/1000). Application of 1 wt% BDDGE resulted a similar Ts but about 15 (n/1000) amine groups were left.An increase in pH (basic) resulted in a faster reaction but in a reduced efficacy. The highest values of Ts were foundat pH 9.0 (70 °C) whereas collagen cross-linked at pH 10.5 exhibited a Ts of 61 °C. Finally, the reactiontemperature did accelerate the reaction without disrupting the efficacy, Cross-linking under acidic conditions (pH <6.0) evokes a different cross-link mechanism. The Ts of the collagen was raised to 64 °C whereas the content ofamine groups was hardly changed. It appears that cross-linking occurred via the carboxylic acid groups. Themacroscopic properties of these materials were completely different from the materials cross-linked under basicconditions. A flexible and soft tissue was found if cross-linking was performed at pH < 6.0, while a stiff sponge wasobtained at basic conditions. Similar results as described above were found if a monofunctional compound (glycidylisopropyl ether) was used.

Chapter 3

- 36 -

INTRODUCTION

Collagens are the major structural proteins of connective tissue like dermis, bone, cartilage,tendons and ligaments. They constitute more than one-third of the total body proteins in mammals[1-3]. The chemical, structural, and biological properties of collagen, make this material verysuitable for biomedical applications. Nowadays, collagen based biomaterials are used inreplacement surgery, including dermis, heart valves and blood vessels.Non-cross-linked collagen-based materials are rapidly degraded in-vivo. The durability of thesematerials can be prolonged by cross-linking. This results in several biochemical and structuralmodifications which are desirable in a surgical prosthesis, such as a decreased antigenicity,increased mechanical strength, reduced solubility, and a reduced rate of biodegradation [1-3].Two types of cross-linking procedures have been developed: physical treatments, such as heat,ultra-violet, and gamma irradiation and chemical treatments [4]. Traditionally, chemical cross-linking procedures involve the use of bifunctional reagents such as glutaraldehyde [5-7] andhexamethylene diisocyanate [8], which react with amine groups of (hydroxy)lysine residues of twoadjacent collagen molecules. Another procedure which received recent attention involves theactivation of the carboxylic acid groups of glutamic or aspartic acid residues followed by reactionwith an amine group of an adjacent polypeptide chain [9-11].During the past years considerable research efforts have been directed towards alternativemethods for cross-linking of biological tissues such as blood vessels, heart valves, and valvedconduits, using epoxy compounds [12, 13]. Generally, mixtures of multifunctional epoxycompounds have been used in these and later studies [14-16]. Using these reagents under basicconditions results in cross-links between the amine groups of the collagen molecules. Cross-linking with epoxy compounds yields well-stabilized tissue having comparable mechanicalproperties to glutaraldehyde cross-linked materials [15, 16]. In-vivo experiments showed lowerlevels of calcification compared to glutaraldehyde treated tissue and no signs of cytotoxicdegradation products were observed [12, 17].Cross-linking with mixtures of multifunctional cross-linking reagents will not result in a well-defined network. In order to obtain a well-defined structure and to get more insight in the cross-linking kinetics and mechanism, a pure, bifunctional epoxy compound was chosen. In this paper,the cross-linking of dermal sheep collagen with 1,4-butanediol diglycidyl ether has been described.The effects of the cross-linking time, reagent concentration, solution pH and reaction temperatureon the cross-linking rate were determined.

MATERIALS AND METHODS

Preparation of non-cross-linked dermal sheep collagenDermal sheep collagen (DSC) was obtained from the Zuid-Nederlandse Zeemlederfabriek(Oosterhout, the Netherlands). In brief, the skin was depilated and immersed in a lime-sodium


- 37 -

sulfide solution to remove the epidermis. Non-collagenous substances were removed usingproteolyic enzymes whereafter the skin was split to obtain the dermal layer. The remaining fibrouscollagen network was washed with water (4 times), with acetone (2 times) and with deionized-water (2 times) before lyophilization (N-DSC).

Cross-linkingAbout 1 g of N-DSC was immersed in 100 ml of a buffered solution containing 1 or 4 wt% of 1,4-butanediol diglycidyl ether (BDDGE, Fluka, Buchs, Switzerland, purity > 95% (GC)) or glycidylisopropyl ether (PGE, Fluka, Buchs, Switzerland). The reactions were carried out at pH 4.5, 5.0,5.5, 6.0, respectively, using a 0.1 M 2-[morpholino]ethanesulfonic acid (MES, Merck, Darmstadt,Germany) buffer adjusted with 1 M NaOH. A phosphate buffered saline solution (PBS, NPBI,Emmercompascuum, The Netherlands) was used for pH values of 7.4. The reactions carried out atpH 8.5 and 9.0 were done using a 0.025 M disodium tetraborate decahydrate buffer (Na2B4O7 �10 H2O Merck, Darmstadt, Germany) which was adjusted with 1 M HCl. Buffered solutions of pH9.5, 10.0 and 10.5 were prepared from a 0.036 M sodium carbonate/ 0.064 M sodiumhydrogencarbonate buffer (NaHCO3/Na2CO3, Merck, Darmstadt, Germany) adjusted with 1 MHCl or NaOH. Reactions were performed as function of time at 20 °C or at 30 °C if indicated.After reaction, the samples were washed with deionized water before lyophilization.Glutaraldehyde (GA, purified [6] by distilled (b.p. 80 °C, 16 mm Hg) from 25 % aqueous solutionz.S., Merck, Darmstadt, Germany) cross-linking was performed by immersing 1 g of DSC in 100ml of a 0.5 wt % GA solution in a phosphate buffer (0.054 M Na2HPO4, 0.013 M NaH2PO4, pH7.4) for 1 hour at room temperature. After cross-linking, the sample was rinsed with tap-water (15min), washed with 4 M NaCl (2 times 30 min) and deionized-water (4 times 30 min) beforelyophilization.

Characterization

Shrinkage temperatureThe degree of cross-linking of the samples was related to the increase of the shrinkage(denaturation) temperature (Ts) after cross-linking. Ts values were determined using an apparatussimilar to that described in IUP/16 [18]. Test specimens were cut, mounted and hydrated. Aheating rate of 2 °C/min was applied and the onset of the shrinkage was recorded as Ts.

Amine group contentThe primary amine group content (lysine residues) of cross-linked samples was determined using2,4,6-trinitrobenzenesulfonic acid (TNBS). To a collagen sample of 2-4 mg, subsequently 1.0 mlof a 4 wt% NaHCO3 solution (pH 9.0) and 1.0 ml of a freshly prepared 0.5 wt% TNBS (Fluka,Buchs, Switzerland) solution in distilled water was added. After reaction for 2 h at 40 °C, 3.0 mlof 6 M HCl was added and the temperature was raised to 60 °C. Solubilization of collagen wasachieved within 90 minutes. The resulting solution was diluted with 5.0 ml demi-water and theabsorbance was measured on a UVIKON 930 spectrophotometer (Kontron Instruments,Switzerland) at 345 nm. A control was prepared applying the same procedure except that HCl wasadded before the addition of TNBS. The amine group content was expressed either as the number

Chapter 3

- 38 -

of amine groups per 1000 amino acids (n/1000). An ε of 14.600 l/mol*cm was used for thecalculations.NMR measurementsHydrolysis rate: The rate of hydrolysis of the epoxide groups of butanediol diglycidyl ether wasstudied by 13C-NMR. A 5 wt% BDDGE solution was prepared either in a buffered 0.2 MNaH2PO4 at pH 4.5 or in a 0.05 M Na2B4O7 • 10 H2O at pH 9.0. To both buffers, 20 vol %deuterium oxide (D2O, purity > 99.8 %, Fluka, Buchs, Switzerland) was added. At different times,13C scans of the solutions were made on a Bruker AC 250 MHz NMR spectrometer. The spectrawere compared with the spectra of pure and a fully hydrolyzed BDDGE solution.Reaction products: NMR spectra of aliquots of the cross-link solutions at pH 9.0 and 10.5 wererecorded with a Bruker AC 250 MHz NMR spectrometer at different reaction times.

Amino acid analysisThe amino acid composition of tissue samples was analyzed by reversed-phase high performanceliquid chromatography (HPLC) with pre-column derivatization. Tissue samples were hydrolyzedwith 6 M HCl for 22 h at 110 °C. After drying for 24 h under vacuum, the hydrolyzate wasdissolved in 0.01 M HCl at a concentration of about 0.4 µmol amino acid residues/ml.Subsequently, a 15 µl aliquot was transferred to a sample vial and derivatized in a Varian 9095Auto Sampler with 9-fluorenylmethyl chloroformate (FMOC-Cl). After reaction, the mixture wasextracted with pentane/ethyl acetate to remove excess FMOC-OH. The derivatives were eluted ona Varian Model 5000 liquid chromatograph with increasing gradient and monitored with a Varian9070 fluorescence detector at 254 nm and 354 nm, using an excitation wavelength of 254 nm.

RESULTS

Cross-linkingAccording to the literature, cross-linking with an epoxy compound was carried out in a basicsolution. Under these conditions, 1,4-butanediol diglycidyl ether (BDDGE) reacts with the aminegroups of lysine or hydroxylysine residues present in the collagen resulting in bridges between thecollagen molecules which contain secondary amines and hydroxyl groups [19-21]. The change inshrinkage temperature (Ts) and the amine group content of BDDGE cross-linked DSC (DSC)samples was monitored to study the kinetics and mechanism of cross-linking.

Influence of the reagent concentration.The influence of the concentration of BDDGE on the degree of cross-linking was determined atdifferent cross-linking times.


- 39 -

0 20 40 60 80 100 120 14045

50

55

60

65

70

Sh

rin

kag

e te

mp

era

ture

[oC

]

Reaction time [h]

0 20 40 60 80 100 120 1400

5

10

15

20

25

30

35

Am

ine

gro

up

co

nte

nt

[n/1

00

0]

Reaction time [h]

Figure 1. The Ts and the content of amine groups as a function of BDDGE concentration(� = 1 wt% and � = 4 wt%) and reaction time (pH 10.0, 20 °C, 100 ml solution per g collagen).

Applying a 4 wt% BDDGE concentration showed a higher reaction rate than a 1 wt% solution asreflected in a faster increase of the shrinkage temperature and a faster reduction of the aminegroups. After 40 h of reaction, a plateau value of Ts of 65 °C was found. Upon prolonged reactiontimes the Ts was only slightly increased from 65 °C to 67 °C whereas the amine group contentdecreased from 9.6 to 4.5 (n/1000), indicating that reactions occurring during this stage do notinvolve cross-link reactions. If a 1 wt% BDDGE solution was used, a much lower initial reactionrate was observed. However, after 140 h a similar Ts of 66 °C was observed, while the content ofamine groups was 14.0 (n/1000), which is considerably higher than for DSC cross-linked with a 4wt% BDDGE solution.

45 50 55 60 65 700

10

20

30

Am

ine

grou

p co

nten

t [n/

1000

]

Shrinkage temperature [oC]

Figure 2. The content of amine groups as a function of the Ts of BDDGE cross-linked DSC usinga 1 wt% (�) or 4 wt% (�) BDDGE solution (pH 10.0, 20 °C, 100 ml/g collagen)

The relation between the Ts and the percentage of amine groups is shown in figure 2. An increaseof the BDDGE concentration led to a less efficient cross-link reaction because more amine groupshad reacted at a certain value of the Ts.

Chapter 3

- 40 -

Modification with a monofunctional glycidyl ether.As already indicated in the previous paragraph, reaction of collagen with BDDGE not only resultsin the formation of cross-links but also in one-side reactions leading to masking of amine groups.In order to study the influence of masking of amine groups on Ts and the reaction kinetics, amonofunctional agent, glycidyl isopropyl ether (PGE), was used. The same reaction conditionswere applied as described for the BDDGE cross-linking.

0 20 40 60 80 100 120 14040

45

50

55

60

65

70

Shr

inka

ge te

mpe

ratu

re [o

C]

Reaction time [h]0 20 40 60 80 100 120 140

0

10

20

30

Am

ine

gro

up c

ont

ent [

n/1

000]

Reaction time [h]

Figure 3. The Ts and the amine group content as a function of PGE concentration (� = 1 wt%and � = 4 wt%) and reaction time (pH 10.0, 20 °C, 100 ml/ g collagen).

Modification of the lysine residues present in the collagen matrix results in a small decrease (3-4°C) of Ts as shown in figure 3. The amount of amine groups reacted was dependent on the reagentconcentration. A higher PGE concentration resulted in a faster decrease of the percentage ofamine groups.

Influence of pHCross-linking of DSC using BDDGE was achieved under basic as well as acidic conditions asshown in figure 4.


- 41 -

4 6 8 10 120

5

10

15

20

25

30

35

Fre

e am

ine

grou

p co

nten

t [n

/100

0]

pH

50

55

60

65

Sh

rinkage tem

perature [ oC]

Figure 4. Cross-linking of DSC using a 4 wt% BDDGE solution in buffer as a function of pH.Reaction was carried out for 44 h at 20 °C. (O) = Ts and (�) = amine group content [n/1000].

At pH values higher than 8.0, BDDGE reacts with the amine groups present in the collagen. Itappears that a higher degree of cross-linking is observed at a higher pH as indicated by a higherTs. Cross-linking at pH values lower than 6.0 resulted in an increase in Ts and no change in thecontent of amine groups. Under acidic conditions, another cross-link mechanism is evoked andcarboxylic acid groups react with the BDDGE molecules with the formation of ester bonds [20].In order to get more insight in the cross-link kinetics, the influence of the solution pH (basicconditions) on the degree of cross-linking was determined as a function of time using a 4 wt%BDDGE concentration.

0 20 40 60 80 100 120 140 16045

50

55

60

65

70

Shr

inka

ge te

mpe

ratu

re [o

C]

Reaction time [h] 0 20 40 60 80 100 120 140 160

0

5

10

15

20

25

30

35

Am

ine

gro

up c

onte

nt [n

/100

0]

Reaction time [h]

Figure 5. The Ts and amine group content as function of the pH an reaction time. pH 8.5(�); 9.0(�); 9.5 (�); 10.0 (�); 10.5 (�) (conditions: 4 wt% BDDGE, 20 °C).

An increase in the rate of cross-linking was observed by increasing the pH of the solution. Despitethe faster reaction at a high pH, a more inefficient reaction is taking place as indicated by a largedecrease in amine groups. This is illustrated by figure 6 in which the content of amine groups isgiven as a function of the Ts.

Chapter 3

- 42 -

45 50 55 60 65 700

10

20

30

Am

ine

grou

p co

nten

t [n/

1000

]


Figure 6. The amine group content as function of the Ts of BDDGE cross-linked DSC at pH8.5(�); 9.0 (�); 9.5 (�); 10.0 (�); and 10.5 (�)

Similar values of Ts were found after cross-linking at pH 8.5 - 10.0. The difference in content ofamine groups of materials having similar values of Ts (figure 6) can be attributed to amine groupsthat have reacted with BDDGE but are not involved in cross-linking.The use of a monofunctional reagent, PGE, showed a similar trend in reaction rates as a functionof the basicity of the solution.

0 50 100 150 2000

5

10

15

20

25

30

35

Am

ine

gro

up

con

ten

t [n

/100

0]

Reaction time [h]

Figure 7. The content of amine groups of PGE modified DSC (4 wt%, 20 °C) as function of pHand reaction time. pH 8.5(�); 9.0 (�); 9.5 (�); 10.0 (�); 10.5 (�)

After 50 h of reaction at pH 10.5 about 6.0 (n/1000) amine groups are left, while 28.7 (n/1000)amine groups are present if the reaction was carried out at pH 8.5. The Ts, which decreased by 3or 4 °C, was hardly affected by the pH or the reaction time.

Under acidic conditions, the epoxide group will be protonated, and subsequently react through anucleophilic attack of a carboxylate anion of glutamic or aspartic acid with the formation of an


- 43 -

ester linkage. The influence of the solution pH on the degree of cross-linking was determined andthe results are presented in figure 8.

0 50 100 150 20045

50

55

60

65

70

Shr

inka

ge t

empe

ratu

re [

o C]

Reaction time [h]

Figure 8. The Ts of BDDGE (4 wt%, 20 °C) cross-linked DSC under acidic conditions asfunction of pH and reaction time. pH 4.5 (�); 5.0 (�); 5.5 (�); 6.0 (�)

Cross-linking took place as shown by the increase in Ts as function of the reaction time. Hardlyany amine group was involved in the reactions as indicated by the high content of amine residuesof 28 to 32 (n/1000) after cross-linking. Cross-linking under acidic conditions appeared to beslower than under basic conditions, and is still in progress after 200 h. Slightly higher cross-linkingrates were found at pH 4.5 and 5.0 while at pH values of 5.5 and 6.0, the highest values of Tswere observed.Treatment of DSC with a 4 wt% PGE solution at pH 5.0 exhibited a material with a slightly higherTs of 51°C compared to non cross-linked collagen. The amine group content was not significantlyreduced, indicating that only carboxylic acid groups reacted with PGE.Amino acid analysis revealed the groups which were involved in the cross-link reactions. Table Ishows the values of some amino acid residues before and after cross-linking. Glutaraldehydecross-linked DSC (G-DSC) was taken as a reference.

Chapter 3

- 44 -

Table IAmino acid composition of non-cross-linked, BDDGE and glutaraldehyde cross-linked

DSC. The amounts of amino acids are expressed as number per 1000 amino acids [n/1000]Amino acids: N-DSC BDDGE

pH 4.5 1)BDDGE pH 9.0 2)

BDDGEpH 10.0 3)

G-DSC 4)

aspartic acidglutamic acidglycineproline(hydroxy)lysinehistidine

5280322126319

5273319129293

5178326131153

547832813050

5380322128114

1). Cross-link conditions: 4 wt% BDDGE, 0.1 M MES buffer, pH 4.5, 168 h, 20 °C2). Cross-link conditions: 4 wt% BDDGE, 0.025 M borate buffer, pH 9.0, 168 h, 20 °C3). Cross-link conditions: 4 wt% BDDGE, 0.1 M carbonate buffer, pH 10.0, 72 h, 20 °C4) Cross-link conditions: 0.5 wt% GA, phosphate buffer, pH 7.4, 1 h, 20 °C.

Cross-linking using GA or BDDGE under basic conditions occurs mainly via the amine groups of(hydroxy)lysine. Moreover, a decrease in histidine was obtained in all cross-linked samples. Asexpected, cross-linking under acidic conditions did not result in a significant reduction in aminegroups. Furthermore, the amounts of carboxylic acid groups remained unchanged.

Influence of reaction temperatureCross-linking reactions performed at pH 5.0 or 10.0 and at a temperature of 30 °C instead of 20°C, revealed a higher reaction rate.

0 20 40 60 80 100 120 140

45

50

55

60

65

70

Sh

rinka

ge

tem

pera

ture

[oC

]

Reaction time [h]

5

10

15

20

25

30

35

Am

ine group content [n/1000]

0 50 100 150 200

45

50

55

60

65

70

Shr

inka

ge te

mpe

ratu

re [o

C]

Reaction time [h]

Figure 9. The influence of the reaction temperature on the cross-linking reaction at two differentpH values: Left figure: pH 10.0 and right figure: 5.0 (4 wt% BDDGE).Ts at 20 °C (�) and 30 °C (�), amine group content at 20 °C (�) and 30 °C (�)

Cross-linking at both pH 5.0 and 10.0 revealed a higher reaction rate if the temperature was raisedfrom 20 to 30 °C. However, the final Ts and the amine group content were independent of thereaction temperature.


- 45 -

Macroscopic appearanceCross-linking and masking affected the macroscopic appearance of the collagen material (table II).

Table IIColor and macroscopic character of the collagen sponges after treatment with

BDDGE or PGE.Reagent pH

rangecolor properties

NoBDDGEBDDGE

PGE

PGE

--8.5-10.54.5-6.0

8.5-10.5

5.0

whitewhiteyellowish

white

yellowish

flexible and soft.stiff, compact; increases with increasing pH andconcentration.flexible, soft and pliable, hardly any influence of pH andconcentration.stiff, more stiff than BDDGE treated materials. Stiffnessincreases when pH and concentration increases.flexible and pliable.

Cross-linking under acidic conditions resulted in a very soft and pliable material, while reactionunder basic conditions exhibited a stiff and compact material. The stiffness was increased if theconcentration and the solution pH were increased. Moreover, collagen with PGE masked aminegroups exhibited a stiff and rigid character. The higher the pH, which means the higher the degreeof masking, the stiffer and more rigid the collagen material. On the other hand, masking of freecarboxylic acid groups revealed a soft and pliable material.

DISCUSSION

Cross-linking of collagen-based biomaterials is usually achieved using glutaraldehyde (GA) [1-7].GA reacts with the amine groups of (hydroxy)lysine residues present in the collagen. This cross-linking reaction is very fast and after 1 h of reaction, a maximum value of the Ts was observed [6].Other bifunctional cross-linkers that react via the amine groups are hexamethylene diisocyanate[8], dimethylsuberimidate [5], and bisepoxy compounds [12, 13, 15]. Because epoxide groups arehighly strained three membered rings, they are susceptible to reaction with nucleophiles such asamines, carboxylic acids, and alcohols [20, 21]. Up to now, studies on the bisepoxide cross-linkingof collagen based biomaterials made use of ethylene glycol diglycidylether [22], a commerciallyavailable reagent with an ill-defined functionality, or of a mixture of multifunctional epoxycompounds [13-15]. The chemistry and kinetics of epoxy cross-linking has only been describedbriefly [23, 24]. To study the reactions involved in the collagen cross-linking process with epoxycompounds and the influence of reaction conditions on the cross-linking rate and the degree ofcross-linking, the compound 1,4-butanediol diglycidyl ether (BDDGE) was selected. Thiscompound is water-soluble and can be obtained in a pure form. DSC was selected as a modeltissue because it is composed of only collagen type I. An outline of the reactions that can beinvolved during cross-linking is given in scheme 1.

Chapter 3

- 46 -

Scheme 1. Cross-linking of collagen with a bisepoxy compound


- 47 -

At pH > 8.0, reaction of amine groups of (hydroxy)lysine residues (I) with epoxide groups of thediglycidyl ether (II) will lead via an intermediate (III) to a cross-link (IV) between two adjacenthelices. Base catalyzed ring opening of the epoxide groups occurs predominantly at the leasthindered carbon atom [25]. Therefore, intermediate (III) will dominate over its isomer. Hydrolysisof the free epoxide groups may be regarded as the most important side reaction that may occur intime. When the free epoxide group of (III) is hydrolyzed a pendant vic-diol group is formed,which is not reactive towards epoxide groups [26-28]. Hydrolysis of the bisepoxide used wasfollowed in time by independent 13C NMR analysis of BDDGE solutions. Hydrolysis rates ofepoxide groups are low at pH 9.0, while an enhancement of hydrolysis was observed under acidicconditions. Less than 6 % of the initial epoxide groups were hydrolyzed during cross-linking at pH9.0 and 160 h. Therefore, the reaction (III → V) will have a minor effect on the cross-linkprocess.Reaction of the hydroxyl group formed in (III) with another diglycidyl ether molecule can lead tostructure (VI). This reaction which is referred to an etherification reaction is expected only tooccur at high temperatures and in the presence of a catalyst such as tertiary amines [19, 28-31].No peaks related to products which underwent etherification were found after analysis of thereaction solutions by 13C-NMR. Therefore, structures like (VI) are not regarded to beincorporated in the collagen matrix and consequently no oligomeric cross-links are expected. Thesecondary amine as in structure (III) can also react with another reagent molecule to give (VII).However, the reaction of a primary amine with an epoxy group has been found much faster thanthe conversion of the secondary amine into a tertiary one [30, 31]. Formation of a cyclic structure(VIII) can be accomplished if two amine groups of one collagen chain react with the samediglycidyl ether molecule, leading to an intrahelical cross-link.

The increase in Ts from 46 to 66 °C of DSC upon reaction with a 4 wt% bisepoxy compoundsolution at pH 10.0 for 40 h, showed that cross-linking occurred under these conditions. Applyinga 1 wt% solution resulted in an increase in Ts to 65 °C after 130 h. Despite the lower cross-linkingrate less primary amine groups had reacted, which implies that the cross-link reaction had a higherefficacy at lower BDDGE concentrations as is indicated in figure 2. If these lines are approachedby a linear curve fit an assessment of the cross-link efficacy could be obtained. The higher thevalue of the slope the lower the efficacy of the cross-link reaction. A 1 wt% BDDGE resulted in aslope of 1.04 amine groups per °C, while usage of a 4 wt% solution lead to a slope of 1.21 aminesper °C. Earlier studies in which DSC was cross-linked with bifunctional reagents showedsomewhat more efficient reactions. For glutaraldehyde cross-linking this value is 0.90 aminegroups per °C [7], and for hexamethylene diisocyanate cross-linking 0.82 amine groups per °C [8].If a high BDDGE concentration was used, most probably more pendant epoxide groups areintroduced in the collagen. When the majority of amine groups has reacted no increase in Ts isseen (figure 2) when remaining amine groups react. A similar phenomenon has been observed inthe cross-linking of DSC with glutaraldehyde [6]. Masking of the amine groups happens when thedistance between two adjacent amine groups is too large to be bridged by the BDDGE molecule.Moreover, it is possible that the adjacent amine group has already reacted with another BDDGEmolecule. Furthermore, a small part of the epoxide end-groups were hydrolyzed, therefore

Chapter 3

- 48 -

introducing a mono-functional BDDGE molecule in the reaction solution, which can onlysubstitute a amine group.A monofunctional agent, glycidyl isopropyl ether (PGE) was used to get more information aboutthe reaction kinetics. As expected, the use of a 4 wt% PGE solution resulted in a faster decreaseof amine groups compared to a 1 wt% solution. After 40 h, 13.8 (n/1000) amines are left at a 4wt% solution, while 27.2 (n/1000) remained free if a 1 wt% solution was used. After the samereaction times, collagen treated with a 1 wt% BDDGE solution had 20.2 (n/1000) amine groupsand usage of a 4 wt% resulted in 9.6 (n/1000) remaining amine groups. These results show thatthe masking reaction is somewhat slower than the cross-link reaction, which implies that thereaction rate constant, k, of the overall cross-link reaction is higher then the rate constant of themasking reaction. Masking of amine groups leads to a decreased thermal stability (figure 3) as aresult of a local distortion or destabilization of the triple-helix structure [13, 32, 33]. The observeddecrease in Ts (3 - 4 °C) is almost independent of the amount of masked amine groups.Amino acid analysis of cross-linked collagen displayed a reduction in contents of (hydroxy)lysine,and histidine (Table I). Tu et al. found also a reduction in lysine and histidine after treatment ofcollagen with multifunctional glycidyl ethers [22]. To prove that only the amine residues of(hydroxy)lysine were involved in the cross-link reactions, an acylation reaction of the aminegroups was performed prior to cross-linking. Acylation of collagen with acetic acid N-hydroxysuccinimide ester affords a collagen matrix with 6 (n/1000) amine groups left. Despite thereduction in histidine, it can be concluded that cross-links were mainly formed between aminegroups of (hydroxy)lysine residues because attempts to cross-link this almost fully maskedmaterial with a 4 wt% BDDGE solution at pH 10.0 did not result in changes in the Ts and aminegroups.

Increasing the pH of the reaction solution increased the rate of cross-linking (figure 5) as a resultof the higher concentration of amine groups (pKa = 10.0 [34]). An increase of the pH from 8.5 upto 10.5 leads to a higher cross-linking rate as can be shown by the faster decrease in amine groupsleft and more rapid increase in Ts. After 20 h of cross-linking, a pH of 8.5 resulted in a Ts of 53°C while a pH of 10.5 resulted in a Ts of 60 °C. The content of amine groups was 26.9 (n/1000)after 20 h of reaction at pH 8.5, whereas a content of 8.6 (n/1000) was found when reacted at pH10.5. Despite the fast reaction at higher pH, a less efficient cross-linked material was obtained.This is illustrated in figure 6 in which the relation between the number of amine groups and Ts isplotted. A cross-link reaction performed at higher pH values lead to a higher conversion of theamine groups at similar values of Ts. An assessment of the cross-link efficacy can be made byapproaching the lines drawn in figure 6 by linear curve fits (table III).


- 49 -

Table IIIThe percentage of deprotonated amines and the value of slope of the linear curve fits

applied on the experimental data presented in figure 6pH [NH2]/([NH 3

+]+[NH2]) - Slope [(n/1000)/oC]8.59.09.510.010.5

0.0310.0910.240.500.76

0.530.691.101.221.61

The results reveal large differences in the cross-linking efficacy at the different pH values. Atrelatively low pH values (8.5 - 9.0) cross-linking is very efficient while at higher pH values (10.0 -10.5) it may be assumed that the fast reaction of the bisepoxide molecules with the amine groupsleads to a higher degree of masking. Similar to the BDDGE cross-link reaction, the rate ofmasking of amine groups with PGE depends on the pH (figure 7). A faster decrease in aminegroups was obtained if a higher pH was applied.In an acidic environment all amine groups are protonated and the mechanism of collagen cross-linking involves the reaction of carboxylic acid groups of glutamic and aspartic acid and theepoxide groups of the BDDGE. At pH values between 4.5 and 6.0, the majority of the carboxylicacid groups is deprotonated (pKa = 4.4 - 4.6) [34]. Cross-linking at acidic pH follows a reactionmechanism in which the epoxide groups become protonated, followed by a nucleophilic attack ofthe carboxylate anion with the formation of an ester linkage. Acid-catalyzed ring opening of theepoxide groups occurs primarily at the more substituted carbon atom [25]. Therefore, structure(X) will be formed predominantly. Analogously to the amine reaction, cross-linking and maskingreactions may occur under these reaction conditions.At lower pH values (4.5) the concentration of protonated epoxide groups will be increased and anincrease in reaction rate will be expected. Moreover, it is emphasized [26, 27] that the hydrolysisrate of the epoxides is substantial at acidic pH values. Cross-linking of collagen in the pH range of4.5 - 6.0 appeared to be relatively slow and times up to 200 h generally were necessary to increasethe Ts to a value of 66 °C. During this period ∼ 20 - 30 % of the epoxide groups were hydrolyzedas determined by 13C NMR analysis of a 5 wt% BDDGE solution at pH 4.5. The cross-linking atacidic conditions may be regarded to proceed with a somewhat lower efficacy than at basicconditions because of the high hydrolysis rates. A relatively high number of structures like (XI) areincorporated in the matrix. The rate of introducing pendant BDDGE molecules (masking reaction)will be a competition between hydrolysis and the possibility of the second epoxide group to reactwith an adjacent carboxylic acid group. Amino acid analysis of collagen cross-linked at pH 4.5(Table I) can not distinguish between free and reacted carboxylic acid groups, because the latterwere susceptible to acidic hydrolysis and therefore be transformed into carboxylic acid groups.Substitution of the carboxylic groups at pH 4.5 using PGE exhibited a material with a slightincrease of the Ts from 46 to 51 °C. This suggests that masking of carboxylic acid groups resultsin a slightly more stabilized triple helix. On the contrary, substitution of the amine groupsdestabilized the material as shown earlier.

Chapter 3

- 50 -

Another parameter which influences the cross-linking rate is the reaction temperature.An increase of the reaction temperature from 20 to 30 °C resulted in a faster cross-link reactionunder both acidic (pH 5.0) and basic (pH 10.0) conditions. This was illustrated by a faster increasein Ts. At pH of 10.0, a maximum value of Ts (66 °C) was obtained after 24 h, while 50 h wererequired to reach this value if cross-linking was carried out at 20 °C. The amine group content wasdecreased to 4.5 (n/1000) amine groups which appeared to be the minimum level of amine groupsafter reaction.

45 50 55 60 65 700

5

10

15

20

25

30

35

Am

ine

grou

p c

onte

nt [

n/10

00]


Figure 10. The amine group content as a function of Ts of BDDGE cross-linked DSC at 20 °C(�) or 30 °C (�) (conditions: pH 10.0, 4 wt% BDDGE)

The increase in reaction temperature did not lead to a different cross-link efficacy as shown infigure 10. Cross-linking under acidic conditions was only slightly affected by the higher reactiontemperature and a somewhat higher Ts was obtained (67 °C versus 64 °C) when the reaction wascarried out with a 4 wt% BDDGE solution at 30 °C at pH 5.0.

The macroscopic appearance of the material appeared to be dependent on cross-link conditions.Cross-linking under basic conditions resulted in a stiff and rigid material. The stiffness of thecollagen was enhanced with increasing pH and with increasing BDDGE concentration. Masking ofthe amine groups through PGE substitution led to a very stiff material. The higher stiffness andrigidity of materials cross-linked with BDDGE at higher pH values can therefore be explained bythe higher content of masked amine groups. On the contrary, cross-linking with BDDGE underacidic conditions resulted in a very flexible material. Masking of the carboxylic acid groups withPGE resulted in the same macroscopic properties.


- 51 -

CONCLUSIONS

Reaction of DSC with BDDGE resulted in an increase of the Ts indicating that cross-linkingoccurred. The cross-linking rate is rather slow compared to the glutaraldehyde process. Underbasic conditions, the reaction proceeds via the amine groups of (hydroxy)lysine residues. Anincrease in BDDGE concentration, in solution pH, and in reaction temperature will acceleratecross-link reaction. On the other hand, the cross-link efficacy was decreased if the BDDGEconcentration or the reaction pH were increased. Under acidic conditions, cross-linking of DSCcan be achieved via the carboxylic acid groups of aspartic or glutamic acid residues. The rate ofreaction is somewhat lower compared to reaction under basic conditions. Furthermore, reaction ofcollagen with a monofunctional reagent, PGE leads to masking of amine or carboxylic acidgroups. This masking reaction is somewhat slower than the cross-link reactions.The macroscopic properties of the material are dependent on the cross-link conditions. A flexibleand soft tissue was obtained after cross-linking at acidic pH, while a stiff and rigid material wasobtained after reaction at basic pH. Masking of amine groups exhibited a very stiff material.

References




4. K. Weadock, R.M. Olsen, and F.H. Silver, "Evaluation of collagen crosslinking techniques", Biomat. Med.Dev. Art. Org., 11(4) pp. 293-318 (1983-84).





9. J.M. Lee, H.L. Edwards, C.A. Pereira, and I.S. S, "Cross-linking of tissue-derived biomaterials in 1-ethyl-3-(dimethylaminopropyl)-carbodiimide", J. Mat. Sci.: Mat. in Med., 7(9) pp. 531-542 (1996).


11. J.-M.D. Girardot and M.-N. Girardot, "Amide cross-linking: An alternative to glutaraldehyde fixation", J.Heart Valve Dis., 5 pp. 518-525 (1996).

12. Y. Noishiki, H. Koyanagi, T. Miyata, and M. Furuse, Bioprosthetic valve, Patent EP 0 306 256 A2 1988.13. R. Tu, S.H. Shen, D. Lin, C. Hata, K. Thyagarajan, Y. Noishiki, and R.C. Quijano, "Fixation of

bioprosthetic tissues with monofunctional and multifunctional poly epoxy compounds", J. Biomed. Mat.Res., 28 pp. 677-684 (1994).

14. H.W. Sung, W.H. Chen, I.S. Chiu, H.L. Hsu, and S.A. Lin, "Studies on epoxy compound fixation", J.Biomed. Mat. Res. Appl. Biomat., 33 pp. 177-186 (1996).

Chapter 3

- 52 -



17. J.M. Lohre, J. Baclig, J. Sagartz, S. Guida, K. Thyagarajan, and R. Tu, "Evaluation of two epoxy ethercompounds for biocompatible potential", Art. Org., 16(6) (1992).

18. IUP/16, "Measurement of shrinkage temperature", J. Soc. Leather Techn. Chem., pp. 122-126 (1963).19. B.A. Rozenberg, "Kinetics, thermodynamics and mechanism of reactions of epoxy oligomers with

amines", in "Epoxy resins and composites II", Ed. by K. Dusek, Springer-Verlag, Berlin-Heidelberg. p.113-165 (1986)



22. R. Tu, R.C. Quijano, C.L. Lu, S. Shen, E. Wang, C. Hata, and D. Lin, "A preliminary study of the fixationmechanism of collagen reaction with a polyepoxy fixative", Int. J. Art. Org., 16(7) pp. 537-544 (1993).

23. H.W. Sung, C.S. Hsu, Y.S. Lee, and D.S. Lin, "Crosslinking characteristics of an epoxy-fixed porcinetendon: Effects of pH, temperature, and fixative concentrations.", J. Biomed. Mat. Res., 31 pp. 511-518(1996).


25. T.W.G. Solomons, "Organic Chemistry", Fourth ed., New York, USA: John Wiley & Sons.(1988).

26. F.A. Long and J.G. Pritchard, "Hydrolysis of substituted ethelyne oxide in H2O18 solutions", J. Am.

Chem. Soc., 78(12) pp. 2663-2667 (1956).27. J.G. Pritchard and F.A. Long, "Kinetics and mechanism of acid-catalyzed hydrolysis of substituted

ethylene oxides", J. Am. Chem. Soc., 78(12) pp. 2667-2670 (1956).28. W. Tänzer, S. Reinhardt, and M. Fedtke, "Reaction of glycidyl ethers with aliphatic alcohols in the

presence of benzyl dimethylamine", Polymer, 34(16) pp. 3520-3525 (1993).29. L. Matêjka and K. Dusek, "Mechanism and kinetics of curing of epoxides based on diglycidyl-amine with

aromatic amines. 2. The reaction of diglycidyl aniline with secondary amines", J. Org. Chem., 22 pp.2911-2917 (1989).

30. J.P. Eloundou, M. Feve, D. Harran, and J.P. Pascault, "Comparative studies of chemical kinetics of anepoxy-amine system", Angewand. Makromol. Chem., 230 pp. 13-46 (1995).

31. L. Matêjka and K. Dusek, "Mechanism and kinetics of curing of epoxides based on diglycidyl-amine witharomatic amines. 1. The reaction of diglycidyl aniline with secondary amines", Macromol., 22 pp. 2902-2910 (1989).

32. A.M. Diamond, S.D. Gorham, D.J. Etherington, J.G. Robertson, and N.D. Light, "The effect ofmodification on the susceptibility of collagen to proteolysis I. Chemical modification of amino acid sidechains", Matrix, 11 pp. 321-329 (1991).

33. C.L. Wang, T. Miyata, B. Weksler, A.L. Rubin, and K.H. Stenzel, "Collagen-induced platelett adhesionand release. I. Effects of side-chain modifications and role of arginyl residues", Biochim. Biophys. Acta,544 pp. 555-567 (1978).


The k net cs of 1,4-butaned ol d glyc dyl ether cross-l nked dermal sheep collagen

- 53 -

Chapter 4

The kinetics of 1,4-butanediol diglycidyl ether cross-

linking of dermal sheep collagen




ABSTRACT

Cross-linking of dermal sheep collagen was achieved with 1,4-butanediol diglycidyl ether (BDDGE). Furthermore,collagen could be modified by a reaction with the monofunctional, glycidyl isopropyl ether (PGE). The reduction inamine groups as a function of time was followed in order to study the overall reaction kinetics of collagen witheither BDDGE or PGE. Linearization of the experimental data resulted in a reaction order (β) of 2 with respect tothe amine groups in the PGE masking reaction, whereas a β of 2.5 was obtained in the BDDGE cross-linkingreaction. The reaction orders were independent of the pH in the range of 8.5 to 10.5 and the reagent concentration(1 - 4 wt%). The reaction order with respect to epoxide groups (α) was equal to 1 for both reagents. As expected,the reaction rate was favored by a higher reagent concentration and a higher solution pH. Because the BDDGEcross-linking reaction occurs via two distinct reaction steps, the content of pendant epoxide groups in the collagenmatrix was determined, by treating the collagen with either O-phosphoryl ethanolamine or lysine methylester. Theincrease in either phosphor or primary amine groups was related to the content of pendant groups. Cross-linking atpH 9.0 resulted in a low reaction rate but in a high cross-link efficacy especially after prolonged reaction times. Amaximum concentration of pendant epoxide groups was detected after 50 h. Reaction at pH 10.0 was faster, but alower cross-linking efficacy was obtained. The ratio between pendant epoxide groups and cross-links was almostequal to 1 during the course of the cross-linking reaction.

Chapter 4

- 54 -

INTRODUCTION

Cross-linking of collagen-based materials can be achieved by bifunctional reagents such asglutaraldehyde [1-4], hexamethylene diisocyanate [5, 6], dimethyl suberimidate [7, 8] and 1,4-butanediol diglycidyl ether [9]. These compounds, having aldehyde, isocyanate, bisimido ester andepoxide functional groups, respectively, react with the amine groups of (hydroxy)lysine residuespresent in the collagen. Despite the common use of these cross-linking agents for stabilization ofthe collagen component of biomaterials, hardly any study on reaction kinetics has been describedin literature. A few studies deal with the kinetics of collagen cross-linking using epoxy compoundsin aqueous solutions [10, 11]. On the contrary, extensive studies on amine-epoxy systems used inepoxy resins were carried out [12-14]. Several kinetic models, which describe the curing process,were proposed and tested experimentally. In general, these non-aqueous systems contain a mixtureof an aromatic diepoxy compound in the presence of primary and secondary amines. Sometimes, ahydroxyl-containing compound, such as water, an alcohol or an acid, is added to influence thecuring reactions. The hydroxyl group promotes the interaction of the epoxide group with anamine, by forming a trimolecular transition state [14]. If no catalyst was added, an initial reactionrate dependency proportional to the square of the amine concentration was measured. Addition ofa hydroxyl-containing compound decreased the order with respect to the initial amineconcentration.Eloundou et al. reacted 1,4- butanediol diglycidyl ether with 4,9-dioxane-1,12-dodecanediamine atrelatively low temperatures between 50 and 95 °C. They found two reaction orders with respect tothe amine groups of 1.20 and 0.45, for the non-catalyzed and the hydroxyl group catalyzedreaction, respectively. Moreover, it was deduced that especially at lower temperatures the reactionof a primary amine with an epoxide group is considerably faster than the reaction of a secondaryamine group with an epoxide group [13].In the present study, the kinetics of the reaction of the amine groups of a collagen matrix using themonofunctional glycidyl isopropyl ether (PGE) or the bifunctional 1,4-butanediol diglycidyl ether(BDDGE) is evaluated. A better understanding of the reaction kinetics enables one to optimize thereaction process in the development of bioprosthetic materials.Whereas the epoxy-amine resin systems described extensively in literature were homogeneoussystems, cross-linking of collagen in an aqueous solution involves a heterogeneous system. Theepoxy molecules are mobile while the amine groups are located at fixed points in the collagenmatrix. Consequently, the epoxy molecules have to diffuse into the collagen matrix followed byreaction with an amine group.The overall reaction kinetics including the reaction orders with respect to the concentration ofepoxide and amine groups were determined. In addition, the content of pendant epoxide groups asa function of the reaction time was measured, to further elucidate the reaction mechanism.


- 55 -


Dermal sheep collagenDermal sheep collagen was obtained from the Zuid-Nederlandse Zeemlederfabriek (Oosterhout,The Netherlands) and was prepared as reported previously [9]. The fibrous collagen network waswashed 4 times with deionized water, 2 times with acetone and 2 times with deionized waterbefore lyophilization.

Effect of the reagent concentration.Lyophilized dermal sheep collagen samples (0.50 g) were immersed in 50 ml of a carbonate (0.064M sodium hydrogencarbonate (NaHCO3)/ 0.036 M sodium carbonate (Na2CO3), pH 10.0)buffered solution containing 0.5, 1, 1.5, 2, 3, 4 or 5 wt% 1,4-butanediol diglycidyl ether (BDDGE)or 1, 3, 4 or 5 wt % glycidyl isopropyl ether (PGE). The reaction was allowed to proceed fordefinite times at 20 °C, followed by extensive washing with deionized water and lyophilization.

Effect of the solution pHA collagen sample (0.50 g) was immersed in 50 ml of a buffered solution containing 4.0 wt%BDDGE. The solution was buffered either with 0.025 M disodium tetraborate decahydrate(Na2B4O7•10H2O, Merck, Darmstadt, Germany) at pH 8.5 or 9.0 or with 0.064 M NaHCO3/0.036 M Na2CO3 at pH values of 9.5, 10.0 and 10.5. Cross-linking was carried out for definitetimes at 20 °C. The samples were washed with deionized water before lyophilization.

Characterization

Amine group contentThe amine group content of the collagen samples was determined spectrophotometrically [1] afterreaction of the primary amine groups with 2,4,6-trinitrobenzenesulfonic acid (TNBS) andsubsequent hydrolysis of the sample, and is expressed as the number of groups present per molcollagen [n/mol].

Determination of pendant epoxide groupsLyophilized dermal sheep collagen samples weighing 1 g were immersed in 100 ml of a bufferedsolution containing 4.0 g BDDGE or PGE. The solution was buffered with 0.025 MNa2B4O7•10H2O at pH 9.0 or with 0.064 M NaHCO3/ 0.036 M Na2CO3 at pH 10.0. Cross-linkingwas allowed to proceed for definite times at 20 °C. After reaction, the samples were extensivelywashed prior to lyophilization. Subsequently, the BDDGE cross-linked samples were eithertreated with a large excess of lysine methylester to prevent additional cross-linking or with O-phosphoryl ethanolamine following the next procedures.a). Lysine methyl ester

A sample of cross-linked collagen (0.20 g) was immersed in 20 ml of a 0.1 Mcarbonate buffered solution (adjusted with NaOH to pH 10.0) containing 0.5 M lysinemethyl ester dihydrochloride (Sigma-Chemical, St. Louis, USA). A large excess of lysine

Chapter 4

- 56 -

methyl ester was used to minimize the content of cros-linking reactions and thus topromote the amount of one-sided reactions. The reaction was allowed to proceed for 3days at 20 °C. Thereafter, the samples were thoroughly rinsed with water beforelyophilization.The amount of pendant epoxide groups was related to the increase in amine groups, asdetermined by the TNBS assay as described above, before and after lysine methyl estertreatment.

b). O-phosphorylethanolamine (O-PEA)A sample of cross-linked material (0.20 g) was immersed in 20 ml of a 0.05 MNa2B4O7•10H2O buffer (adjusted with NaOH to pH 9.0) containing 0.5 M O-PEA (SigmaChemical, St. Louis, USA). The reaction was allowed to proceed for 3 days at 20 °C,followed by extensive washing with water before lyophilization. About 50 mg of O-PEAtreated collagen was hydrolyzed in 4.0 ml 6 M HCl at 110 °C for 20 h. Exactly 0.5 ml ofthe hydrolyzate was added to 1.0 ml deionized water. Subsequently, 0.8 ml 0.08 wt%hydraziniumsulfate (z.A. Merck, Darmstadt, Germany) and 0.4 ml 5.0 wt% sodiummolybdate dihydrate (ACS, Sigma Chemical, St. Louis, USA) were added. The reactionwas carried out for at least 3 h at 60 °C. The absorbance of the blue solution was measuredat 820 nm [15]. The amount of coupled O-PEA was correlated to a calibration curvemade of hydrolyzed O-PEA solutions.

KineticsThe rate of reaction can be followed by measuring the decrease in amine groups as a function oftime. The reactions that have been taken into account are presented in scheme 1. This implies thatsome assumptions have to be made. First, hydrolysis of the epoxide groups is regarded not toaffect the reaction kinetics because the rate of hydrolysis of the reagents is low and less than 6 %of the initial epoxide groups was hydrolyzed after 160 h of reaction at pH 9.0 [9]. Furthermore,the reaction rate of epoxide groups with amine groups is supposed to be much slower than thediffusion rate of the reagents to the reaction sites as was shown by Tu et al. [11]. Finally, thereaction between secondary amines and the epoxide groups have not been taken into account.


- 57 -

[NH2] [PGE]

NH CH2 CH R

OH

k��

CH2 CH RO

+NH2

PGE masking:

NH2 + CH2 CH R CH CH2O O

�� NH CH2 CH

OH

R CH CH2O

k1

[BDDGE] [INTER][NH2]

BDDGE cross-linking:

NH2 NH CH2 CH

OH

R CH CH2O

+

��k2 NH CH2 CH

OH

R CH

OH

CH2 NH

[CROSS-LINK]

Scheme 1. Reactions of amine groups of dermal sheep collagen with PGE or BDDGE.

The overall kinetics of the cross-linking or masking process can be written as [11, 16]:

− =d NH

dtk Epoxide NH

[ ][ ] [ ]2

2α β (I)

Where [Epoxide] and [NH2] are the concentrations of the epoxide and the amine groups, t =reaction time, k = overall reaction rate constant, α = reaction order with respect to epoxidegroups, and β = reaction order with respect to the amine groups of the collagen material.When a large excess of the epoxy compound is used, the concentration of epoxide groups can beregarded as effectively constant during the course of the reaction.Rearrangement and integration yields:

− =∫ ∫d NH

NHk Epoxide dt

NH

NH tt [ ]

[ ][ ]

,

,

2

20

02 0

2

βα (II)

Chapter 4

- 58 -

Equation II will give equation III-a in case β=1 and equation III-b when β≠1.

LnNH

NHk Epoxide tt(

[ ]

[ ]) [ ]2

2 00= − α ( β = 1) (III-a)

11

2

2 0

1([ ]

[ ])( )NH

NH

Ktt β −

= + ( β ≠ 1) (III-b)

with K k Epoxide NH= −( ) [ ] [ ]β α β1 0 2 0 (IV)

When the amine concentration of the collagen is measured as a function of the reaction time, thedata should fit either equation III-a or III-b. Consequently, for a first-order reaction a plot of thelogarithm of ([NH2]t/[NH2]0) against time, t, gives a straight line with a slope of -k[Epoxide]0

α.For a reaction order of β other than 1, a plot of 1/[([NH2]t/[NH2]0)

(β-1) ] against time is a straightline with a slope K. The value of β was calculated by applying a linear regression on theexperimental data.The value of α can be deduced by varying the values of [Epoxide]0 while holding the othervariables constant. The corresponding values of K are measured, and a plot of the logarithm of Kas a function of the logarithm of the initial epoxide concentration ([Epoxide]0) yields a straight linewith a slope of α.If the reaction process is diffusion-controlled, it means that the overall reaction rate is determinedby the limiting rate of diffusion of the epoxy compound to the amine groups of the collagen:

− = −d NH

dt

D SEpoxide Epoxidesurface

[ ] *([ ] [ ] )2

δ(V)

Where D = diffusion coefficient of epoxy compound, S = surface area, δ = boundary layer,[Epoxide] = epoxide concentration of the bulk, which is equal to the initial concentration,[Epoxide]surface = epoxide concentration at the collagen surface which is equal to zero.Integration of equation (V) yields:

[ ] [ ]* * [ ]

NH NHD S Epoxide

tt2 2 00= −

δ(VI)

Equation (VI) shows that the amine group concentration decreases linearly with time, if thereaction is diffusion-controlled.

The order of the reaction gives no direct information about the mechanism of the reaction,although it may give some clues [16]. The cross-linking reaction of collagen with BDDGE mostprobably takes place via distinct sequential steps, thus forming an intermediate (INTER). Theconcentration of INTER depends on the reaction rates k1 and k2 and may also depend on theavailability of amine groups in the collagen matrix. Therefore, the concentration of INTER was


- 59 -

determined as a function of time to make an assessment of the ratio of cross-links and maskedamine groups during the course of the reaction.

RESULTS

Dermal sheep collagen was cross-linked with 1,4-butanediol diglycidyl ether (BDDGE) ormodified with glycidyl isopropyl ether, PGE [9] at a pH of 10.0. Reaction between the aminegroups of the collagen and the epoxide groups of either BDDGE or PGE took place as reflectedby a decrease in amine groups as a function of the reaction time (figure 1).

0 20 40 60 80 100 120 140

5

10

15

20

25

30

35 1 % BDDGE

4 % BDDGE 1 % PGE

4 % PGE

Am

ine

gro

up

con

tent

[n/

100

0]

Reaction time [h]

Figure 1. The amine group content as a function of the reaction time during masking with PGEor during cross-linking with BDDGE at two different concentrations (1 and 4 wt%, pH 10.0, 20°C, 1 g collagen in 100 ml reaction solution).

A higher concentration of both PGE and BDDGE resulted in a faster reduction of amine groups.In spite of the almost similar epoxide concentration in the corresponding BDDGE and the PGEreaction solution, a faster decrease in amine groups as a function of the reaction time wasobserved during BDDGE cross-linking. This implies that the second reaction step of the BDDGEcross-linking reaction (k2 in scheme 1) has a considerable contribution to the overall reaction rate.In order to determine the overall reaction rate of the BDDGE and PGE reaction, the reactionorders with respect to the amine groups of the collagen and the epoxide groups are determined.

PGE reactionThe kinetic data of the PGE reaction plotted in figure 1 were fitted to the kinetic model by a linearregression.

Chapter 4

- 60 -

Table IDetermination of ββββ by comparison of the regression coefficients obtained after a linear

regression of the kinetic data presented in figure 1β Regression coefficient, r

PGERegression coefficient, r

BDDGE1.11.52.02.22.52.73.03.54.0

0.9660.9860.9980.9970.9950.9940.9880.9740.960

0.9080.9460.9790.9880.9980.9980.9950.9930.984

It was found that a β of 2 (Table I) resulted in the best fit (figure 2) for the modification reactionwith either a 1 or a 4 wt % PGE solution.

0 20 40 60 80 100 120 140 1600

1

2

3

4

5

1/([

NH

2] t /[N

H2] 0)

Reaction time [h]

Figure 2. Fit of the kinetic data of the PGE masking reaction with a 1 wt% (�) or 4 wt% (�)solution by using second order reaction kinetics with respect to amine groups of the collagen

The pH of the reaction solution was altered to investigate if the reaction kinetics are applicable fora larger pH range.


- 61 -

0 50 100 150 2000

5

10

15

20

25

30

35

Am

ine

grou

p c

onte

nt [n

/100

0]

Reaction time [h]

0 50 100 150 2000

2

4

6

8

10

1/([

NH

2] t /[N

H2] 0)

Reaction time [h]

(a) (b)

Figure 3. The content of amine groups as a function of reaction time at different pH values (3-a).The fit of the kinetic data for second order reaction of the PGE masking reaction of dermal sheepcollagen at different pH values (3-b). � = pH 8.5; � = pH 9.0; � = pH 9.5; � = pH 10.0; �= pH 10.5. (Reaction conditions: 4 wt% PGE, 20 °C, 1 g collagen in 100 ml reaction solution)

Application of a β of 2 resulted in a good fit of the experimental data, as determined in a pH rangebetween 8.5 and 10.5. The slope in figure 3-b became steeper with increasing pH which meansthat K increased.

BDDGEAnalogously to the PGE reaction, the kinetics of the cross-linking reaction with the bifunctionalBDDGE were studied.

0 20 40 60 80 100 120 1400

4

8

12

16

1/([

NH

2] t /[N

H2] 0)1

.5

Reaction time [h]

Figure 4. Fit of the kinetic data of the BDDGE cross-linking reaction with a 1 wt% (�) or 4 wt%(�) solution by using a 2.5-th order reaction kinetics with respect to the amine groups of thecollagen

Chapter 4

- 62 -

In contrast to the PGE reaction, the best fits were obtained if a β of 2.5 (table I) was used. Inaddition, the effect of the solution pH on the cross-linking kinetics was investigated.

0 20 40 60 80 100 120 140 1600

5

10

15

20

25

30

35A

min

e g

roup

con

tent

[n/1

000]

Reaction time [h]

0 50 100 1500

5

10

15

20

25

1/([

NH

2] t/[NH

2] 0)1.5

Reaction time [h]

(a) (b)

Figure 5. The decrease in amine groups as a function of reaction time at different pH values (5-a). The 2.5-order fit of the kinetic data of the BDDGE cross-linking reaction at different pHvalues (5-b). � = pH 8.5; � = pH 9.0; � = pH 9.5; � = pH 10.0; � = pH 10.5.(Conditions: 4 wt% BDDGE, 20 °C, 1 g collagen in 100 ml reaction solution)

Similar to the reaction at pH 10.0, application of a 2.5-th order with respect to the amine groupsof the collagen resulted in the best fits of the kinetic data. The increase in the slope in figure 5-bmeans that the reaction rate was significantly higher at a higher pH.

The reaction order with respect to the epoxide group concentration was determined by measuringthe K-value at different epoxide concentrations. Some additional experiments with differentconcentrations of either PGE and BDDGE were applied to calculate the reaction order withrespect to epoxide groups.

-4 -3 -2 -1 0 1

-6

-5

-4

-3

-2

Ln K

Ln ([PGE]0)

-4 -3 -2 -1 0

-5

-4

-3

-2

-1

Ln K

Ln (2*[BDDGE]0)

Figure 6. The logarithm of K as a function of the logarithm of the initial epoxide concentrationfor the determination of the reaction order with respect to epoxide groups (α).


- 63 -

Both the reaction with PGE or BDDGE resulted in a straight line if the logarithm of K was plottedas function of the logarithm of [Epoxide]0. An α of 0.98 was found in case of the PGE maskingreaction, whereas an α of 1.08 was obtained if BDDGE was used. These results indicate thatwithin the experimental error a first order was found with respect to epoxide groups.

Finally, the effect of the solution pH on the value of K is shown in figure 6.

8.0 8.5 9.0 9.5 10.0 10.5 11.0-7

-6

-5

-4

-3

-2

Ln

K

pH8.0 8.5 9.0 9.5 10.0 10.5 11.0

-7

-6

-5

-4

-3

-2

-1

0

Ln

K

pH

PGE BDDGE

Figure 7. Plot of the logarithm of K as a function of the solution pH of the reaction of dermalsheep collagen with either PGE or BDDGE (4 wt%, 20 °C, 100 ml/g collagen)

In both experiments, a linear increase of the logarithm of K as a function of the pH was obtained.The gradient of both lines was approximately 1.8.

During reaction of collagen with a bifunctional reagent not only cross-linking but maskingreactions as well will account for the decrease in amine groups. Therefore, the content of pendantepoxide groups was measured as a function of the reaction time. The amount of cross-links in timedepends on the reaction rate of epoxide groups with the amine groups of the collagen, but also onthe availability of a second amine group for the reaction with the pendant epoxide group after thefirst reaction step. The concentration of pendant epoxide groups (INTER in scheme 1) wasdetermined by reaction with either a large excess of lysine methylester or O-phosphorylethanolamine (O-PEA). The increase in primary amine or phosphate groups, respectively, wasdetermined. The concentration of cross-links is related to the decrease in amine groups and theincrease in the free epoxide groups using: [Cross-links] = ([NH2]0-[NH2]t -[INTER])/2. Table IIsummarizes the results obtained after cross-linking of dermal sheep collagen with BDDGE at pH9.0 or 10.0

Chapter 4

- 64 -

Table II The concentrations of amines, intermediates and cross-links as a function of the treatmenttime during cross-linking of collagen by BDDGE at pH 9.0 (upper-part) or 10.0 (lower part

of the table)Treatment time [h] [NH2]

[n/mol][INTER][n/mol]

[Cross-links][n/mol]

0 32.0 0 02 31.2 0.7 0.15 28.8 1.2 1.224 25.6 3.9 1.348 20.8 5.4 2.972168

18.413.6

5.30.3

4.39.2

0 32.0 0 01 29.6 1.4 0.5

5.5 23.2 2.5 3.224 10.4 8 6.848 5.6 10.4 8.1

Reaction conditions: 1 g collagen in 100 ml reaction solution, 20 °C, 0.396 mol epoxide groups/l,

A graphical representation of the data presented in Table I is given in figure 8.

0 50 100 1500

5

10

15

20

25

30

Con

cen

trat

ion

[n/m

ol]

Reaction time [h]

0 10 20 30 40 500

5

10

15

20

25

30

Co

ncen

trat

ion

[n/m

ol]

Reaction time [h]

pH 9.0 pH 10.0

Figure 8. Concentrations of amine groups (�), intermediates (�) and cross-links (▲) as afunction of the reaction time during cross-linking of dermal sheep collagen at pH 9.0 (a) and pH10.0 (b). (4.0 wt% BDDGE, 20 °C, 100 ml/g collagen)

A faster decrease in amine groups is found at pH 10.0 than at pH 9.0. At pH 10.0, both theconcentration of pendant epoxide groups and cross-links increased with an almost similar rate as afunction of the reaction time. At pH 9.0, the increase in pendant epoxide groups and cross-linkswas much slower. Moreover, a higher initial increase in pendant epoxide groups as compared tothe increase in cross-links was observed. Contrary to the reaction at pH 10.0, a maximum


- 65 -

concentration of pendant epoxide groups was detected after approximately 50 h. Furthermore. acontinuous increase in cross-link density was detected during the course of the reaction.

Influence of the initial concentration of epoxide groups and the pH on the degree ofmasking and cross-linkingThe concentration of BDDGE and the deprotonated amine groups present in the collagen materialwill have an effect on the rate of reaction, the degree of cross-linking, and the cross-link efficacy.The contents of amine, and pendant epoxide groups as well as the amount of cross-links formedwere studied as a function of the initial BDDGE concentration or pH.

0.0 0.1 0.2 0.3 0.4 0.50

5

10

15

20

25

30

Con

cent

ratio

n [n

/mo

l]

[Epoxide] mol/l

8.0 8.5 9.0 9.5 10.0 10.5 11.00

5

10

15

20

25

30

Con

cent

ratio

n [n

/mol

]

pH

(a) (b)

Figure 9. The concentration of amine groups (■), intermediates (�) and cross-links (▲),respectively, as a function of the initial epoxide (a) concentration (pH 10.0, 24 h, 20 °C, 1 gcollagen per 100 ml reaction solution) or as a function of the pH (b, 4.0 wt % BDDGE, pH 8.5;9.0; 10.0; 10.5, 24 h, 20 °C, 1 g collagen per 100 ml reaction solution).

An increase in epoxide concentration resulted in a lower content of amine groups after 24 h ofreaction. In addition, a higher content of both intermediates and cross-links were found. Thecross-link efficacy was only slightly effected as indicated by the ratio of pendant epoxide groupsversus cross-links which is somewhat lower than 1 over the whole range of epoxideconcentrations. An increase of the solution pH from 8.5 to 10.5, demonstrated a lower content ofamine groups after 24 h of cross-linking. A linear relationship between the concentration of aminegroups left in the matrix and the solution pH was measured. Consequently, a higher concentrationin both cross-links and pendant epoxide groups was observed with increasing pH. Despite thelower content of amine groups at pH 10.5, a lower content of cross-links and a higher content ofmasked amine groups was obtained as compared to the collagen which was cross-linked at pH10.0.

Chapter 4

- 66 -

DISCUSSION

Cross-linking of tissues rich in collagen is a well-known method in the preparation of bioprostheticmaterials such as wound dressing, vascular grafts, tendons and aortic heart valves. Glutaraldehydehas been the most widely applied and most extensively studied cross-linking agent [1-4]. Duringthe past decade, other methods have been developed [17-19] to cross-link collagen-based tissues.A relatively new group of cross-linking reagents comprises the poly glycidyl ethers, and several ofthese multifunctional reagents appeared to be successful in cross-linking of collagen-basedmaterials [20-22]. Despite the large amount of research done in this area, the kinetics of cross-linking of collagen materials with epoxy compounds have only been described briefly [10,11]. Tuet al. cross-linked bovine internal thoracic arteries with the bifunctional ethylene diglycidyletheror with the trifunctional glycerol polyglycidyl ether. Cross-linking was carried out at 37 °C underbasic conditions (pH 8.5, 9.5 or 10.5) using three different epoxy compound concentrations of 0.5,1.0 and 4.0 wt%. They found for all reaction conditions that the decrease in amine groups was firstorder in the epoxy compound concentration and 2.5-th order in amine groups of the collagen-based tissue [11].Recently, the cross-linking of dermal sheep collagen (DSC) with 1,4-butanediol diglycidyl etherhas been described. Under basic conditions, this reagent will react with the amine groups of(hydroxy)lysine residues. By controlling the reaction conditions, the cross-link density and efficacycan be altered [9]. Elaboration of the reaction kinetics makes it possible to optimize the cross-linking process and to design the material properties. In order to simplify the kinetics, sidereactions such as hydrolysis, were supposed to occur at negligible rates. Scheme 1 summarizes thereactions which were expected to lead to a cross-link or to substitution of an amine group.Because the experimental data showed (figure 1) that the reaction with either BDDGE or PGEwas slow and not complete within 24 h and that no linear relationship was found between thedecrease in amine groups as a function of time, it was concluded that the reaction process is notdiffusion-controlled.The overall reaction rate equation (I) can be derived in which the reaction orders with respect tothe epoxide and amine groups have to be determined. The reaction order with respect to the aminegroups of the collagen was determined by application of equation (III-b) on the kinetic data. Thevalue of β was determined by linear regression of the experimental data and was referred to as thereaction order to the amine groups.Masking of the amine groups with PGE resulted in a β of 2, which was independent of the solutionpH in the range between 8.5 and 10.5 and the PGE concentration (1 - 4 wt%). On the contrary, aβ of 2.5 was found if the collagen was cross-linked with the bifunctional BDDGE. Analogously tothe PGE reaction, the value of β was not affected by the BDDGE concentration or the pH. Thishigher value of β suggests that the second reaction (k2), which results in the formation of a cross-link, has a considerable effect on the overall reaction rate.


- 67 -

PGE reaction:

+ C H 2 C HO

R

N H 2

N H 2

N H 2

N H 2

N H 2

N H 2

BDDGE reaction:

N H 2

N H 2

N H 2

N H C H 2 C H

O H

R ' C H C H2O

N H 2

N H 2

C H 2 C H R ' C H C H2O O

��+

H 2N

H 2N

H 2N

H 2N

H 2N

H 2N��

k 1

N H 2

N H 2

N H 2

N H 2

N H 2

N H 2

N H 2

N H 2

N H C H 2 C H

O H

R

N H 2

N H 2

N H 2

��k

k 2

k 2

Scheme 2. Schematic representation of the effect of the effective amine concentration during thePGE and BDDGE reaction.

Scheme 2 shows that an incoming PGE or BDDGE molecule can react with all the amine groupswhich are still available at that time. In addition, the second epoxide group of the BDDGEmolecule can only react with the amine groups which are closely located to the pendant group dueto spatial limitations. Therefore, the effective amine concentration for the second epoxide group isconsiderably lower than for an incoming BDDGE molecule. However, the pendant epoxide groupsare already close to an adjacent amine group, which implies that the possibility of this reaction willbe higher than the reaction of a amine group with an incoming epoxide group which still have toapproach the amine group. Because a faster decrease in amine groups is observed in the BDDGEcross-linking reaction as compared to the corresponding PGE masking reaction (figure 1), thesecond step appears to have a higher reaction rate constant than the first step.As shown in figures 3 and 5, a faster decrease in amine groups was found if the reaction wascarried out at a higher pH. The effective concentration of deprotonated amine groups is dependenton the pH according to the Henderson-Hesselbach equation.

pH pKNH

NHa= + +log[ ]

[ ]2

3

or [ ]

[ ]

[ ] /

[ ] /

NH

NH

H K

H Ka

a

2

3 1+

+

+=+

(VII)

with pKa (lysine in collagen) = 10.0 [23] and [H+] = 10-pH.

Chapter 4

- 68 -

A higher effective concentration of amine groups increases the value of K and consequently theoverall reaction rate according to equation I. A linear plot was obtained if the logarithm of K wasplotted as a function of the solution pH. The gradient of this line was 1.8 for both PGE andBDDGE, which indicates that the effect of the pH on the reaction rate was similar for bothreactions.

The reaction order with respect to the epoxide groups (α) was equal to 1 for both the PGE andthe BDDGE reaction. This suggests that the same reaction kinetics are valid for the reaction of anincoming BDDGE or PGE molecule with an amine group of the collagen.

The overall reaction rate equation does not give information about the mechanism of cross-linking.Because a pendant epoxide group is introduced in the collagen matrix after the first step, thecontent of these groups was determined as function of the reaction time. Dermal sheep collagenwhich was cross-linked with BDDGE for different times at pH 9.0 or 10.0 was either treated withan excess of lysine methylester or with O-phosphoryl ethanolamine (O-PEA) in order to quenchthe pendant epoxides. The pendant epoxide groups react predominantly with the α-amine group oflysine methyl ester because its pKa (8.8) is lower than the pKa (10.8) of the ε-amine group [23].Scheme 3 presents the reactions which occur during lysine methylester or O-PEA treatments.

+ NH2 (CH2)4 CH

NH2

C

O

CH3

O

��

CH CH2O

+��

CH CH2 NH CH2 CH2 O P

OH

OH

OOH

�� Hydrolysis in concentrated HCl

amino acids and

H2N CH2 CH2 O P

OH

O

OH

P OH

OH

HO

O

CH

OH

CH2 NH CH C O CH3

O

(CH2)4

NH2

CH CH2O

Scheme 3. Post-reactions of free epoxide groups with lysine methyl ester or o-phosphorylethanolamine.

Treatment of BDDGE cross-linked collagen with lysine methyl ester did not result in additionalcross-linking as indicated by similar values of the shrinkage temperature before and after treatment(data not shown). This implies that only one amine group of lysine methylester reacted with thependant epoxide group. Therefore, it is allowed to relate the content of epoxide groups in thematrix to the increase in primary amine groups. Furthermore, more or less the same results were


- 69 -

obtained with lysine methyl ester and o-phosphoryl ethanolamine, implying that both methods givea good indication of the concentration of pendant epoxide groups in the material.Cross-linking at pH 10.0 showed a faster decline in amine groups (figure 8) as compared to thereaction at pH 9.0. Consequently, a faster increase in the concentrations of both pendant epoxidegroups and cross-links was measured.During cross-linking a competition takes place between the reaction of the first epoxide group ofan incoming BDDGE molecule with a amine group (k1) and the reaction of the second epoxidegroup of a pendant BDDGE molecule with the same amine group (k2). A higher pH and thus ahigher effective concentration of deprotonated amine groups will increase the possibility of bothreactions. Due to the higher concentration of free BDDGE molecules in the solution as comparedto the concentration of pendant epoxide groups (Table I), the effect of a higher concentration ofdeprotonated amine groups will be more pronounced for the first reaction (k1). Hence, theformation of cross-links will be suppressed, which leads to a lower cross-link efficacy. Cross-linking at pH 9.0 showed a rapid increase in pendant epoxide groups in the first stage of thereaction while the amount of cross-links was only slightly elevated. This suggests that in the initialstage of reaction mainly masking reactions occur which are responsible for the decline in aminegroups. It might be that, due to the low concentration of deprotonated amine groups, thepossibility of the reaction of the second epoxide group with an adjacent amine group is rather low(see also scheme 2), therefore the first reaction dominates in the initial stage of the reaction. Afterapproximately 50 h, a reduction in pendant epoxide groups was recognized while the amine groupcontent was further decreased. This can be explained by the reaction of the pendant epoxide groupwith an amine group of an (hydroxy)lysine residue of an adjacent polypeptide chain. It appears thatafter 50 h, the possibility of the first reaction is somewhat reduced possibly due to a loweraccessibility of the amine groups. In this stage, the pendant epoxide groups have the ability toreact with an amine group to form cross-links. The increase of the cross-link density wassupported by an increase in the shrinkage temperature from 59 to 71 °C [9] in the time intervalbetween 50 and 168 h of cross-linking.Cross-linking at pH 10.0 resulted in a fast initial increase in both masking and cross-linkingreactions. It appears that a maximum in pendant epoxide groups was observed after 48 h. Nodecrease in these groups was observed because they cannot further react with another aminegroup to form cross-links because the content of amine groups is already reduced to a level of 4(n/1000) which seems to be the lower limit of accessible amine groups [9]. The lower value of Tsof a material cross-linked at pH 10.0 (Ts = 62 °C) instead of 9.0 (Ts = 71 °C) which was obtainedin a previous study [9], can therefore be explained by the high degree of masking and the lowdegree of cross-linking at pH 10.0.Finally, the amount of substitution and cross-linking as function of the epoxide concentration andthe pH was examined after 24 h of reaction. As expected, a higher epoxide concentration willresult in a lower content of amine groups, and in a higher amount of cross-links and pendantepoxide groups as well. The ratio between the content of pendant epoxide groups and cross-linkswas not much affected by the concentration of initial epoxide groups. In contrast, after 24 h ofcross-linking, a less efficient cross-linking reaction as judged from the ratio masking and cross-linking reactions, and a lower content of amine groups were obtained if the pH was elevated.

Chapter 4

- 70 -

CONCLUSIONS

The kinetics of the reaction of amine groups of dermal sheep collagen with mono- or bifunctionalepoxy compounds is rather complex. Because of the mobility of the reagents and the fixedpositions of the amine groups in the collagen matrix, cross-linking of collagen takes place in aheterogeneous system. The reaction rate is proportional to the square of the concentration ofamine groups in the PGE masking reaction. In contrast, a 2.5-th order dependency of the reactionrate with respect to amine groups was obtained during BDDGE cross-linking. This difference inreaction order must be due to the effect of the second reaction of pendant epoxide groups withamine groups. The reaction order with respect to the epoxide groups is equal to 1 for both thePGE and the BDDGE reaction. Finally, a linear dependency of the logarithm of K as a function ofthe solution pH was observed.The cross-link reaction between BDDGE and dermal sheep collagen occurs via a two-stepmechanism as shown by a cross-link experiment at pH 9.0, in which the content of pendantepoxide groups reaches a maximum in time. The subsequent reduction in pendant epoxide groupscan be explained by a second reaction, leading to the formation of cross-links. Cross-linking at pH10.0, did not show a maximum value of masked amine groups within 48 h. The ratio betweenmasked amine groups and cross-links was approximately 1 and was hardly dependent on thereaction time.

References1. L.H.H. Olde Damink, P.J. Dijkstra, M.J.A. v. Luyn, P.B. v. Wachem, P. Nieuwenhuis, and J. Feijen,

"Glutaraldehyde as crosslinking agent for collagen based biomaterials", J. Mat. Sci.: Mat in Med., 6 pp.460-472 (1995).


3. J.M. McPherson, S. Sawamura, and R. Armstrong, "An examination of the biological response toinjectable, glutaraldehyde cross-linked collagen implants", J. Biomed. Mat. Res., 20 pp. 93-107 (1986).

4. M.L. Salgaller and P.K. Bajpai, "Immunogenicity of glutaraldehyde treated bovine pericardial tissuexenografts in rabbits", J. Biomed. Mat. Res., 19 pp. 1-12 (1985).

5. W.A. Naimark, C.A. Pereira, K. Tsang, and J.M. Lee, "Hexamethylenediisocyanate crosslinking of bovinepericardial tissue: A potential role of the solvent environment in the design of bioprosthetic materials", J.Mat. Sci: Mat. in Med., 6 pp. 235-241 (1995).


7. V. Charulatha and A. Rajaram, "Crosslinking density and resorption of dimethylsuberimidate-treatedcollagen", J. Biomed. Mat. Res., 36 pp. 478-486 (1997).


9. R. Zeeman, P.J. Dijkstra, P.B. v. Wachem, M.J.A. v. Luyn, M. Hendriks, P.T. Cahalan, and J. Feijen,"Cross-linking and modification of dermal sheep collagen using 1,4-butanediol diglycidyl ether", Chapter3 of this thesis and submitted to J. Biomed. Mat. Res. (1998).

10. H.W. Sung, W.H. Chen, I.S. Chiu, H.L. Hsu, and S.A. Lin, "Studies on epoxy compound fixation", J.Biomed. Mat. Res. Appl. Biomat., 33 pp. 177-186 (1996).



- 71 -

12. L. Matêjka and K. Dusek, "Mechanism and kinetics of curing of epoxides based on diglycidyl-amine witharomatic amines. 1. The reaction of diglycidyl aniline with secondary amines", Macromol., 22 pp. 2902-2910 (1989).

13. J.P. Eloundou, M. Feve, D. Harran, and J.P. Pascault, "Comparative studies of chemical kinetics of anepoxy-amine system", Angewand. Makromol. Chem., 230 pp. 13-46 (1995).

14. B.A. Rozenberg, "Kinetics, thermodynamics and mechanism of reactions of epoxy oligomers withamines", in "Epoxy resins and composites II", Ed. by K. Dusek, Springer-Verlag, Berlin-Heidelberg. p.113-165 (1986)

15. M. Sitohy, J.M. Chobert, and T. Haertle, "Influence of the reaction conditions on the chemicalphosphorylation of milk proteins", Milchwissenschaft, 49(11) pp. 610-615 (1994).

16. M.J. Pilling and P.W.Seakins, "Reaction Kinetics", Oxford University Press, New York Inc.(1995).17. E. Khor, "Methods for the treatment of collagenous tissues for bioprostheses", Biomaterials, 18(2) pp. 95-

105 (1997).18. D.M. Simmons and J.N. Kearney, "Evaluation of collagen cross-linking techniques for the stabilization of

tissue matrices", Biotech. Appl. Biochem., 17 pp. 23-29 (1993).19. R. Stanescu and V. Stanescu, "In vitro protection of the articular surface by cross-linking agents", J.

Rheumatology, 15(11) pp. 1677-1682 (1988).20. J.M. Lee, C.A. Pereira, and L.W.K. Kan, "Effect of molecular structure of poly (glycidyl ether) reagents on

crosslinking and mechanical properties of bovine pericardial xenograft materials", J. Biomed. Mat. Res.,28 pp. 981-992 (1994).

21. Z. Tang and Y. Yue, "Crosslinkage of collagen by polyglycidyl ethers", ASAIO J., 41 pp. 72-78 (1995).22. H.-W. Sung, H.-L. Hsu, C.-C. Shih, and D.-S. Lin, "Cross-linking characteristics of biological tissue fixed

with monofunctional or multifunctional epoxy compounds", Biomaterials, 17(14) pp. 1405-1410 (1996).23. S.S. Wong, "Reactive groups of proteins and their modifying agents", in "Chemistry of protein conjugation

and cross-linking", Ed. by S.S. Wong, CRC Press Inc., Boca Raton, Florida. p. 7-48 (1991).

In-v tro degradat on of dermal sheep collagen cross-l nked w th 1,4-butaned ol d glyc dyl ether

- 73 -

Chapter 5

In vitro degradation of dermal sheep collagen cross-

linked with 1,4-butanediol diglycidyl ether




ABSTRACT

In-vitro degradation of dermal sheep collagen (DSC) cross-linked with 1,4-butanediol diglycidyl ether (BDDGE)for various times and at different pH's was studied with either bacterial collagenase or pronase. Cross-linking ofDSC at pH 4.5 resulted in a reaction between the carboxylic acid groups of collagen and epoxide groups ofBDDGE. Even after cross-linking for 10 d, the materials degraded slowly in a collagenase solution and somewhatfaster with pronase. Cross-linking at pH 9.0 or 10.0 involves amine groups of collagen and epoxide groups ofBDDGE. After 72 h of cross-linking at pH 9.0, stable materials were obtained. Cross-linking at pH 10.0 yieldedalready stable materials after reaction for 6 h. The rates of degradation were correlated with the shrinkagetemperature (Ts) of the materials. A high Ts corresponds with a low rate of degradation.DSC cross-linked with BDDGE at pH 4.5 for 10 d had a tensile strength of 5.4 MPa, an elongation at break of 159% and a high strain modulus of 5.2 MPa. Degradation with collagenase for 24 h did not affect the tensile strengthto a large extent, but exposure to a pronase solution for 24 h revealed that the materials had no mechanical strengthleft. BDDGE cross-linked collagen at pH 9.0 for 10 d had a tensile strength of 2.4 MPa, a low elongation at breakof 101 %, and a high strain modulus of 3.4 MPa. However, these values were hardly affected by exposure to eithercollagenase or pronase. The differences between BD45 and BD90 are explained in terms of intra- and interhelicalcross-links and by inter(micro)fibrillar cross-links. BD90 is the most suitable material in load-bearing applications,while BD45 can act as a temporary scaffold.

Chapter 5

- 74 -

INTRODUCTION

Well-known cross-linkers to stabilize collagen-based materials are glutaraldehyde [1-3],carbodiimides [2, 4, 5] and diisocyanates [6]. The stability of such cross-linked materials towardsdegradation is usually studied by in-vitro degradation tests in which the materials are exposed toenzymes such as bacterial collagenase [1, 3, 4, 7], pronase [8], pepsin [9] or chemicals such asCNBr [10] The resistance against degradation is mostly related to the weight-loss of the materialin time, measured by gravimetrical [3], or colorimetrical [4, 10] methods. Chvapil et al. usedhexamethylenediisocyanate cross-linked reconstituted collagen fibers for anterior cruciateligaments. Despite the fact that hardly any changes in residual weights were obtained, themechanical strength was almost completely lost after 8 - 10 months of implantation in goats [11].Furthermore, glutaraldehyde cross-linked dermal sheep collagen demonstrated almost no loss inweight after 24 h of degradation in a bacterial collagenase solution. However, within thisdegradation period the tensile strength was reduced to 35 % of its original value [3]. Porcineinternal thoracic arteries cross-linked with ethylene glycol diglycidyl ether or glutaraldehydeshowed also a decrease in tensile strength after exposure to a collagenase solution [12]. On theother hand, dermal sheep collagen (DSC) cross-linked with a water-soluble carbodiimide showedonly a slight reduction in tensile strength as a function of collagenase degradation time [3].Therefore, it is appropriate to monitor both changes in weight and mechanical properties of thesematerials during degradation [3, 4, 12], especially when collagen materials are used for loadbearing applications.Epoxy compounds have been used for cross-linking of collagen-based materials [13-15]. In thesestudies, mainly mixtures of bi- and trifunctional epoxy compounds were used. The use of abifunctional reagent, 1,4-butanediol diglycidyl ether (BDDGE), has been described earlier [16].Cross-linking of DSC was accomplished under either acidic (pH < 6.0) or basic (pH > 8.0)reaction conditions. Whereas at pH > 8.0 reaction between epoxide groups of BDDGE and aminegroups of the collagen takes place, at pH < 6.0 epoxide groups react with carboxylic acid groupsof collagen [16, 17].Here we describe the in-vitro degradation behavior of dermal sheep collagen cross-linked withBDDGE at pH 4.5, 9.0 or 10.0 using a bacterial collagenase or a pronase solution. In addition, themechanical properties such as tensile strength and elongation at break of cross-linked materialswere studied as a function of degradation time. The results lead to a schematic representation ofthe location of the cross-links and to a possible mechanism of degradation with collagenase andpronase.


Preparation of non-cross-linked dermal sheep collagenDermal sheep collagen was obtained from the Zuid-Nederlandse Zeemlederfabriek (Oosterhout,The Netherlands) and was prepared as reported previously [16]. The fibrous collagen network waswashed 4 times with deionized water, 2 times with acetone and 2 times with deionized waterbefore lyophilization.


- 75 -

Cross-linkingAbout 0.25 g of N-DSC was immersed in 25 ml of a buffered solution containing 4 wt% 1,4-butanediol diglycidyl ether (BDDGE, Fluka, Buchs, Switzerland). The solution was buffered eitherwith 0.05 M 2-[morpholino]ethanesulfonic acid (MES , Merck, Darmstadt, Germany) at pH 4.5,with 0.025 M disodium tetraborate decahydrate (Na2B4O7 • 10 H2O, Merck, Darmstadt,Germany) at pH 9.0 or with carbonate (0.036 M Na2CO3 (Merck, Darmstadt, Germany), 0.064 MNaHCO3 (Merck, Darmstadt, Germany)) at pH 10.0. Cross-linking was carried out for differenttime periods at 20 °C. After cross-linking, the samples were rinsed with water beforelyophilization.Samples to study the influence of the pH during cross-linking on the mechanical properties of thecollagen were obtained by reacting 2 g of collagen for 10 d in 200 ml of a buffered (0.05 M MESat pH 4.5 or 0.025 M Na2B4O7•10 H2O at pH 9.0) solution containing 4 wt% BDDGE. Aftercross-linking the sheets were washed with water before lyophilization.

Characterization

Shrinkage temperatureThe degree of cross-linking of the samples was related to the increase of the shrinkage(denaturation) temperature (Ts) after cross-linking. Ts values of (non)-cross-linked samplesimmersed in water were determined as described previously [16].

Amine group contentThe amine group content of the collagen samples was determined spectrophotometrically afterreaction of the primary amine groups with 2,4,6-trinitrobenzenesulfonic acid (TNBS) andsubsequent hydrolysis of the sample [16] and is expressed as the number of groups present per1000 amino acids [n/1000].

Enzymatic degradation

CollagenaseThe degradation of (non)-cross-linked collagen samples was performed using bacterial collagenasefrom Clostridium histolyticum (EC 3.4.24.3, Sigma-Chemical, St. Louis, USA) with a collagenaseactivity of 283 U/mg (one unit will release peptides from native collagen, equivalent in ninhydrincolor to 1.0 micromole of L-leucine in five h at pH 7.4 at 37 °C in the presence of calcium ions).In a typical experiment, 15 mg samples of (non)-cross-linked DSC were immersed in 1.0 ml of a0.1 M Tris-HCl buffer (pH = 7.4) containing 0.005 M CaCl2 and 0.05 mg/ml NaN3. After one h,1.0 ml collagenase solution (200 U/ml) in Tris-HCl buffer (37 °C) was added to give the desiredconcentration (100 U/ml).The degradation was discontinued at the desired time interval by addition of 0.2 ml 0.25 M EDTA(Titriplex III, Merck, Darmstadt, Germany) solution. The mixtures were cooled on ice. Agravimetrical method was used to determine the weight-loss of the samples. The collagenasesolution was decanted whereafter the remaining pellet was washed with Tris-HCl buffer (3 times

Chapter 5

- 76 -

for 15 min) and 3 times with deionized water before lyophilization. The changes in weight ofpartially degraded samples were expressed as the percentage of the initial weight.

PronaseA pronase stock solution was prepared by dissolving a calculated amount of pronase (fromStreptomyces grisseus) lyophilizate (Boehringer Mannheim, 7000 U/g lyophilizate) into a 0.1 MTris-HCl buffer solution (pH = 7.4) containing 0.005 M CaCl2 and 0.05 mg/ml NaN3. The finalpronase concentration was 20 U/ml. This stock solution was incubated at 37 °C for one h.In a typical experiment, 15 mg of (non)-cross-linked DSC was immersed in 5.0 ml pronasesolution (37 °C). The degradation was discontinued at the desired time interval by addition of 0.5ml 0.25 M EDTA, after which the mixtures were cooled on ice. The pronase solution wasdecanted and the remaining disk was washed with Tris-HCl buffer (3 times 15 min) and with demi-water (3 times 15 min) before lyophilization. The remaining weight of the disk was determinedgravimetrically.

Mechanical propertiesStress-strain curves of DSC samples were determined by uniaxial measurements using a Zwick(Z020) mechanical tester. Because of variations in the mechanical properties of different parts ofthe sheep skin, only samples from the IUP/2 [18] sampling area parallel to the backbone weretaken.Tensile bars (40.0 mm x 4.0 mm x 1.4 mm) were cut using a dumb-bell shaped knife and hydratedfor at least one hour in PBS at 20 °C. The thickness of the samples was measured in triplicateusing a spring-loaded type micrometer (Mitutoyo, Tokyo, Japan). An initial gauge length of 10mm was used and a crosshead speed of 5 mm/min was applied until rupture of the test specimenoccurred. A pre-load of 0.05 N was applied to pre-stretch the specimen before the realmeasurement. The tensile strength, the elongation at alignment, the elongation at break, the lowstrain modulus and the high strain modulus of the sample were calculated from five independentmeasurements.The change in mechanical properties as a function of the degradation time was correlated with theproperties of matching non-degraded controls taken from an adjacent part of the skin. Samplesweighing 0.25 g were degraded using slightly modified procedures as described above.Degradation with collagenase was performed with 25 ml collagenase solution (activity 100 U/ml).Degradation was discontinued by addition of 10 ml 0.25 M EDTA. Digestion with pronase wasperformed with 65 ml pronase solution (activity 20 U/ml). Addition of 5 ml 0.25 M EDTAterminated the degradation. The partially degraded samples were washed three times with an ice-cold 0.001 M EDTA solution to remove any remaining buffer and to inactivate any remainingcollagenase or pronase. After washing the samples were lyophilized. Stress-strain curves weredetermined after hydrating the samples in PBS containing 0.001 M EDTA at 20 °C.


- 77 -

RESULTS

Cross-linking of dermal sheep collagen (DSC) with 1,4-butanediol diglycidyl ether (BDDGE) atpH 4.5 and 9.0 at various time intervals was related to the in-vitro degradation behavior. First, theshrinkage temperature (Ts) and the number of amine groups (n/1000) of the BDDGE cross-linkedmaterials were determined to get an indication of the cross-link density. The Ts and the content ofamine groups of DSC cross-linked at pH 4.5 for various times are presented in table I.

Table I The shrinkage temperature and the content of amine groups of dermal sheep collagen

cross-linked with BDDGE at pH 4.5 as a function of cross-link time (4 wt % BDDGE, 20°C)

Cross-link time [h] Shrinkage temperature [°C] Amine group content [n/1000]0 45.5 32.06 49.7 32.023 52.0 31.748 53.5 30.772 57.0 30.5144 60.2 27.2240 67.8 26.6

At pH 4.5, DSC was slowly cross-linked, and after 10 days the Ts had increased by 22 °C. Cross-linking occurred mainly via the carboxylic acid groups of aspartic or glutamic acid residues asindicated by the high content of amine groups after cross-linking [16], although a few aminegroups appeared to be involved in the reactions, especially after prolonged times.

0 5 10 15 20 250

20

40

60

80

100

Rem

aini

ng w

eigh

t [%

]

Degradation time [h]

0 5 10 15 20 250

20

40

60

80

100

Rem

aini

ng

wei

ght [

%]


Collagenase Pronase

Figure 1. Remaining weight as a function of degradation time during exposure of BDDGEcross-linked (pH 4.5) dermal sheep collagen samples with a Ts of 49.7 °C (�), 52.0 °C (∇), 53.5

°C (▼), 57.0 °C (●), 60.2 °C (▲), 67.8 °C (�) to to bacterial collagenase (100 U/ml, pH 7.4, 37°C, n=3) or pronase (20 U/ml, pH 7.4, 37 °C, n=3).

Chapter 5

- 78 -

The remaining weight of cross-linked materials with different Ts, as a function of degradation timeusing bacterial collagenase or pronase is presented in figure 1, respectively. Materials having thehighest values of Ts show the highest resistance against enzymatic degradation. All cross-linkedmaterials were more susceptible to degradation by pronase than by collagenase. An almost lineardecrease in weight was observed for the collagen materials with different degrees of cross-linking.The rates in weight loss, determined from the slopes of linear curve fits applied on the initial datadisplayed in figure 1, as a function of the Ts are presented in figure 2.

45 50 55 60 65 700

50

100

150

200

Wei

ght

loss

[%/h

]


Figure 2. Rate of weight loss of DSC cross-linked with BDDGE at pH 4.5 during incubation in acollagenase (�, 100 U/ml, pH 7.4, 37 °C) or pronase (�, 20 U/ml, pH 7.4, 37 °C) solution asfunction of the shrinkage temperature.

An increase in shrinkage temperature results in a decreased rate of degradation. It appears that avalue of Ts of about 60 °C must be reached, above which only slow degradation is observed. Asindicated by the data in figure 3, introduction of a few cross-links (low Ts) appears to have a muchlarger effect on degradation rate with pronase than with collagenase.The mechanism of the cross-linking reaction between BDDGE and collagen functional groups isdifferent under acidic and basic conditions [16]. Contrary to carboxylic acid groups that react atpH 4.5, amine groups react at high pH. Analogously as described above, DSC cross-linked withBDDGE at pH 9.0 and 10.0 was exposed to collagenase and pronase and the rates of degradationwere determined. The initial values of Ts and amine group content are given in table II.


- 79 -

Table IIShrinkage temperature and amine and pendant epoxide group content of (non)-cross-linked

dermal sheep collagen as a function of the cross-link conditions (time and pH).pH Cross-link time [h] Shrinkage temperature

[°C]Amine group content

[n/1000]Pendant epoxide

groups [n/1000] 1.

-- 0 45.0 32.0 09.0 6 48.8 27.2 2.09.0 23 51.8 26.2 3.59.0 48 54.8 20.7 5.49.0 72 60.5 18.4 5.39.0 144 65.1 15.2 0.69.0 240 72.1 9.3 0.510.0 3.5 51.5 23.6 2.510.0 6 54.6 18.9 3.510.0 23 56.7 12.5 7.510.0 48 63.5 7.3 9.5

1. The content of pendant epoxide groups for this series was not measured but correlated to the values obtained in ananalogous series in a previous study [19].

The rate of reaction at pH 10.0 is higher than at pH 9.0 as reflected in a faster increase in Ts and afaster decrease in amine groups. However, cross-linking appears less efficient at pH 10.0 becausemore amine groups had reacted while a similar value of Ts was obtained. The Ts of the materialcross-linked at pH 9.0 or 10.0 was plotted as function of the content of cross-links, which isdetermined with: [Cross-links] = ([amines reacted]-[pendant epoxide group])/2.

45 50 55 60 65 70 750

2

4

6

8

10

12

[Cro

ss-li

nks]

[n/

1000

]

Ts [oC]

Figure 3. The content of cross-links as a function of the Ts of BDDGE cross-linked DSC samples(4 wt% BDDGE, pH 9.0 or 10.0, 20 °C)

A rather linear relationship between the content of cross-links and the Ts of the BDDGE cross-linked DSC was found. It appears that Ts is a good measure for the cross-link density of DSCcross-linked with BDDGE. The effect of the one-sided masking reactions is only small because thematerials cross-linked at pH 10.0 had a high degree of masking (table II), while the Ts of the

Chapter 5

- 80 -

material was comparable to DSC samples cross-linked at pH 9.0 which exhibited a lower degreeof masking.The in-vitro resistance towards enzymes was determined for the materials mentioned in table II.

0 5 10 15 20 250

20

40

60

80

100

Rem

aini

ng w

eigh

t [%

]

Degradation time

0 10 20 30 40 500

20

40

60

80

100

Re

mai

ning

wei

ght

[%]


Collagenase pronase

Figure 4. Remaining weight as function of degradation time of DSC samples cross-linked withBDDGE at pH 9.0, with a Ts of 48.8 °C (�), 51.8 °C (�), 54.8 °C (�), 60.5 °C (�), 65.1 °C (∆)

and 72.1 °C (�) using either collagenase (100 U/ml, pH 7.4, 37 °C) or pronase (20 U/ml, pH7.4, 37 °C).

DSC cross-linked with BDDGE at pH 9.0 for over 72 h resulted in materials with a Ts of higherthan 61 °C. These materials were hardly degrade by collagenase or pronase.

0 5 10 15 20 250

20

40

60

80

100

Rem

aini

ng w

eigh

t [%

]

Degradation time [h]0 10 20 30 40 50

0

20

40

60

80

100

Rem

aini

ng

wei

ght

[%]


Collagenase Pronase

Figure 5. Remaining weight as function of degradation time of DSC samples cross-linked withBDDGE at pH 10.0, with a Ts of 51.5 °C (�), 54.6 °C (�), 56.7 °C (�), 63.5 °C (�) usingeither collagenase (100 U/ml, pH 7.4, 37 °C) or pronase (20 U/ml, pH 7.4, 37 °C).


- 81 -

Cross-linking of DSC with BDDGE at pH 10.0 for over 6 h (Ts is 55 °C or higher) resulted inmaterials which exhibit hardly any susceptibility towards enzymatic degradation.

Mechanical propertiesThe change in mechanical properties of DSC cross-linked with BDDGE at pH 4.5 or 9.0 for 10 das function of the degradation time was studied. The Ts, the content of amine groups and themechanical properties of the collagen materials used are given in Table III.

Table IIIShrinkage temperature, amine group content and mechanical properties of (non)-cross-

linked dermal sheep collagen materialsMaterial Shrinkage

temperature[°C]

Amine groups[n/1000]

Tensilestrength [MPa]

Elongation atbreak[%]

High strainmodulus [MPa]

N-DSC 45.5 ± 0.5 32.0 ± 0.5 2.6 ± 0.1 173 ± 8 2.9 ± 0.2BD45 67.8 ± 0.3 26.7 ± 1.0 5.4 ± 0.6 159 ± 15 5.2 ± 0.2BD90 72.1 ± 1.1 9.3 ± 0.5 2.4 ± 0.2 101 ± 10 3.4 ± 0.3Cross-linking conditions: 4 wt % BDDGE, 240 h, 20 °C. DSC cross-linked at pH 4.5 or 9.0 is referred to as BD45and BD90, respectively.The amine group content of the collagen is expressed as the number of amine groups per 1000 amino acids (n=3).All mechanical properties are measured in five fold and are given as mean ± standard deviation

The reaction conditions applied afforded collagen materials that were extensively cross-linked asreflected by the high Ts values. The mechanical properties of the collagen materials depend on thecross-link conditions. Cross-linking with BDDGE at pH 4.5 resulted in materials which have anincreased tensile strength and high strain modulus and a lower elongation at break as compared toN-DSC. In contrast, reaction at pH 9.0 afforded a material with a similar tensile strength, asomewhat higher 'high' strain modulus and lower elongation at break as compared to N-DSC.

Collagenase degradationDSC cross-linked at pH 9.0 retained over 95 % of its original weight, whereas a 10 % decrease inweight was observed for collagen cross-linked at pH 4.5. During degradation, the amine groupcontent of the material cross-linked at pH 4.5 increased from 27 to 31 (n/1000) and from 9 to 11(n/1000) for DSC cross-linked at pH 9.0.

Chapter 5

- 82 -

0 5 10 15 20 250

2

4

6

8

Ten

sile

str

engt

h [M

Pa

]


0 5 10 15 20 250

50

100

150

200

Elo

ngat

ion

at b

reak

[%

]


(a) (b)

0 5 10 15 20 250

2

4

6

8

10

Hig

h st

rain

mod

ulus

[MP

a]


(c)

Figure 6. Changes in (a) tensile strength , (b) elongation at break and (c) high strain modulus ofBD45 (�) or BD90 (�) as a function of degradation time during exposure to a bacterialcollagenase solution (100 U/ml, pH 7.4, 37 °C, n=5).

The mechanical properties of cross-linked DSC were only slightly affected upon degradation withcollagenase as shown in figure 6. The tensile strength of both materials was not changed after 24 hof degradation. Furthermore, the elongation at break of BD45 was almost independent of thedegradation time, while an increase from 93 to 125 % was found for BD90. The high strainmodulus of both materials was only slightly affected.

Pronase degradationThe effect of the proteolytic enzyme, pronase, on the mechanical properties of BDDGE cross-linked DSC as function of the digestion time is presented in figure 7. Only 56 % of the weight ofBD45 was retained after 24 h, whereas 98 % of the weight of BD90 was recovered. An increase inthe amine group content from 27 to 36 (n/1000) amine groups was found for BD45, while a smallincrease from 9 to 12 (n/1000) was found for BD90.


- 83 -

0 5 10 15 20 250

2

4

6

Ten

sile

str

engt

h [M

Pa]

Degradation time [h]0 5 10 15 20 25

0

50

100

150

200

Elo

nga

tion

at b

reak

[%]


(a) (b)

0 5 10 15 20 250

2

4

6

Hig

h st

rain

mod

ulus

[MP

a]


(c)

Figure 7. Changes in (a) tensile strength, (b) elongation at break and (c) high strain modulus ofBD45 (�) or BD90 (�) as a function of degradation time during exposure to a pronase solution(20 U/ml, pH 7.4, 37 °C, n=5).

In contrast to degradation with collagenase, the mechanical properties of BD45 decreased uponpronase degradation. The tensile strength was reduced from 5.4 to 0.3 MPa, the elongation atbreak from 159 to 45 % and the high strain modulus from 5.2 to 1.2 MPa. BD90 showed a slightreduction in tensile strength from 2.4 to 1.8 MPa, an increase in elongation at break from 101 to149 % and a reduction of the high strain modulus from 3.4 to 1.9 MPa. Noticeable is thatapparently after 4 h of degradation no changes in the mechanical properties of BD90 were found.

DISCUSSION

The stability of collagen based materials towards enzymatic degradation can be improved bychemical or physical cross-linking [1-5, 7, 8]. Traditionally, glutaraldehyde (GA) has beenextensively used as a cross-linking agent for fixation of porcine and pericardial heart valves,pericardial patches, tendons and blood vessels. Epoxy compounds have also been used for

Chapter 5

- 84 -

stabilization of arteries and pericardium [12, 13, 20, 21]. In general, these studies revealed thatGA cross-linked materials had somewhat better stability towards enzymatic degradation overepoxy fixed materials. In-vitro degradation studies showed that glycerol triglycidyl ether cross-linked (pH 9 - 10) porcine aortic leaflets had a slightly lower resistance towards collagenase thanglutaraldehyde treated materials [20]. Sung et al. demonstrated that the tensile strength ofethyleneglycol diglycidyl ether cross-linked porcine internal thoracic arteries was reduced to 40 %after collagenase degradation for 24 h at 37 °C, whereas the tensile strength of its GA-fixedcounterpart was reduced by 25 % [12]. They also showed that porcine pericardium cross-linkedwith either ethyleneglycol diglycidyl ether or (poly)glycerol polyglycidyl ether (pH 10.5) degradedwith comparable rate to GA cross-linked controls upon exposure to collagenase or pronase [22],which means that epoxy compounds are suitable reagents to stabilize collagen-based tissues.In a previous paper [16], 1,4-butanediol diglycidyl ether (BDDGE) has been described as aneffective cross-linker in the fixation of dermal sheep collagen (DSC). Usually, cross-linking with(poly) epoxy compounds is performed under basic conditions [13-15]. Under these conditions,cross-links are formed through reaction of the bisepoxide with the amine groups of(hydroxy)lysine residues. However, epoxide groups are also able to react with carboxylic acidgroups at lower pH [16, 17]. Cross-links are formed between the carboxylic acid groups ofaspartic or glutamic acid residues. DSC, which was cross-linked via the carboxylic acid groups,retained the original macroscopic pliability of the non cross-linked DSC (N-DSC). On thecontrary, cross-linking via the amine groups resulted in a more rigid and stiff material with areduced pliability.The resistance against enzymatic degradation of the (non)-cross-linked DSC is usually studied byin-vitro tests, using enzymes such as bacterial collagenase and pronase. Bacterial collagenase fromClostridium histolyticum is capable of cleaving peptide bonds within the triple helical structure andhas a specificity for the Pro-X-Gly-Pro-Y region, splitting between X and Gly where X and Y arepredominantly apolar amino acid residues. Pronase from Streptomyces grisseus, which is a mixtureof unspecific endo- and exo proteases, is usually applied for the complete hydrolysis of peptidesand proteins [23]. However, pronase can only cleave bonds in the non-helical telopeptides of thetriple helix and is unable to disrupt the native collagen structure [24]. The peptides which arereleased from tropocollagen differ from native collagen in that they contain less glycine and prolineand little or no hydroxyproline. Furthermore they are acidic and rich in tyrosine. In addition,pronase may cleave inter(micro)fibrillar bonds, which might result in the loss of native triple-helices or even fragments of (micro)fibrils. Other studies showed that pronase can also hydrolyzeesters of several non-proteinogenic aliphatic and aromatic α-amino acids [25].The rate of degradation of a collagen-based biomaterial by enzymes is determined by the cross-linkdensity, the accessibility of the cleavage sites and the extent of denaturation [26]. Cross-links willsterically hinder the enzymes to get access to their specific cleavage sites, thus decreasing thedegradation rate. In addition, fragmented parts of the collagen matrix are kept together by thecross-links and consequently multiple chain scissions are needed to release peptide fragments.Degradation of collagen with collagenase or pronase is considered to be a surface erosion process[3, 4]. The rate of degradation will then be determined by the structural level at which degradationtakes place and the availability of cleavage sites. Non or slightly cross-linked DSC at pH 4.5 israpidly degraded by both collagenase and pronase (figure 1). At this point it has to be stated that


- 85 -

no clear insight has been obtained in the character of the degradation products. It cannot beexcluded that insoluble fragments, which have gone in suspension, are removed. Nevertheless thisimplies that in these materials collagenase has a good access to helical structures and pronase tonon-helical telopeptides, inter(micro)fibrillar bonds and diester cross-links.An assessment can be made at what structural level enzymes are able to degrade the collagenmaterials (figure 8). The molecular weight of collagenase is in between 68 and 125 kD [27]. In afirst approximation assuming a spherical shape, the radius of gyration will be in between 2.8 and3.9 nm [28]. This implies that collagenase probably will start at surfaces of fibrils. Olde Damink etal. [4] showed that collagenase initially adsorbs onto fibers and that during the course ofdegradation, collagenase was able to penetrate into the fiber. Furthermore, it appeared that cross-linking hindered penetration of the enzyme into the fiber. At this time no definite conclusion can bedrawn but based on the structural organization of collagen it seems feasible that both collagenaseand pronase (molecular weight of proteases is between 25 and 80 kD [23]) have early access tothe fibril surface.

Figure 8. Schematic representation of collagen hierarchy and possible degradation mechanisms.Initially, collagenase or pronase will, depending on its size and conformation, adsorb either onthe collagen fiber or at the fibril surface followed by cleavages of peptide bonds.

Cross-linking of DSC with BDDGE at pH 4.5 proceeds via the carboxylic acid groups and is aslow process (table I). After 10 d of cross-linking the Ts was increased from 46 to 68 °C. Anincrease in cross-linking time results in an elevated Ts, which implies that the cross-link densitywas increased. Furthermore, a higher resistance against degradation with collagenase and pronasewas observed (figure 1). In general, the degradation rate of DSC cross-linked with BDDGE at pH4.5 is higher with pronase, than with collagenase. The reasons for this may be that after exposureof collagen to pronase, telopeptide connections between helices may be degraded and that thediester cross-links between helices and microfibrils may be cleaved. The combination of theseprocesses will then result in the release of helical segments into the enzyme solution which will

Chapter 5

- 86 -

become a suspension. Collagenase only cleaves helical segments, which are now shielded by thediester cross-links.

Cross-linking of collagen via the amine groups is faster compared to cross-linking via the carboxylgroups as indicated by a faster increase in Ts as a function of reaction time. The reaction can beaccelerated by elevating the solution pH (table II). However, the cross-link efficacy is lowered ifthe pH is increased from 9.0 to 10.0, which results in a higher content of cross-link moleculeswhich have only reacted at one side (masking) [16, 19]. Nevertheless, the effect of the one-sidedreactions on the Ts is small because the increase in Ts is almost linearly proportional to the cross-link density (figure 3).Slight differences in rate of degradation with collagenase as compared to pronase were found forDSC cross-linked with BDDGE at pH 9.0 (figure 4). Materials with a low Ts were even fasterdegraded by collagenase as compared to pronase, which is in contrast with the results obtained forDSC cross-linked at pH 4.5. This can be explained as follows: In this material less helical segmentscan be released by pronase, because of the presence of interhelical and intermicrofibrillar cross-links, whereas collagenase can still attack the helices, due to the low degree of cross-linking.DSC cross-linked at pH 9.0 for 72 h was resistant towards enzymatic degradation, while reactionat pH 10.0 for 6 h also resulted in a material which was not degraded upon exposure to theenzymes (figure 5). The material which was cross-linked at pH 10.0 for 3.5 hours had a very lowresistance against pronase, and after 24 h of degradation no material could be retrieved. However,only 35 % of the material was digested by collagenase during the same time. The reason for thisbehavior is not known.The effect of the degree of masking and cross-linking on the degradation behavior is illustrated bycomparing DSC materials, having a similar Ts, but which were cross-linked at either pH 9.0 or10.0. DSC cross-linked at pH 9.0 for 48 h had a Ts of 54.8 °C and contained 5.4 (n/1000) pendantepoxide groups and 3.0 (n/1000) cross-links per α-chain. This material was degraded for about 20% by collagenase and for 10 % by pronase after 24 h. On the other hand, a material cross-linked atpH 10.0 for 6 h, which resulted in a similar Ts of 54.6 °C, displayed no degradation. This materialhad a low degree of masking (3.5 (n/1000)), while the amount of cross-links was 4.8 (n/1000).These results imply that either an increase in the number of cross-links or a decrease in number ofmasked groups lead to a higher stability towards enzymatic degradation. To further evaluate theseeffects, the degradation of DSC cross-linked at pH 9.0 for 72 h was compared to DSC cross-linked for 48 h. Although DSC cross-linked for 72 h had a comparable content of masked groups(5.3 (n/1000)) to DSC cross-linked for 48 h, the number of cross-links was higher (4.1 versus 3.0(n/1000)), whereas its rate of degradation was lower. This suggests that for DSC cross-linked viathe amine groups, the cross-link density is mainly determining the resistance against enzymaticdegradation. Because the exact degree of cross-linking and masking reactions is not known forDSC cross-linked via the carboxyl groups, this argument cannot be extrapolated to thesematerials.

Effect of degradation time on the mechanical properties


- 87 -

The data presented in table III demonstrate that the mechanical properties of BDDGE cross-linkedDSC can be designed by the reaction conditions. Cross-linking of DSC via the carboxylic acidgroups (BD45) resulted in a material with a significantly higher tensile strength, elongation atbreak and 'high' strain modulus than DSC cross-linked at pH 9.0 (BD90). In spite of the morefavorable mechanical properties and the pliability of BD45, the enzymatic stability of this materialis somewhat lower compared to BD90. Because the mechanical properties determine to a largeextent if a material is suitable for a load bearing application, the mechanical properties of DSCcross-linked with BDDGE at pH 4.5 and 9.0 for 10 d were determined as a function of thedegradation time.The exact causes for the change in mechanical properties of collagenous materials after cross-linking are still not known. However, for our systems it is obvious that the position of the intra-and interhelical cross-links (reaction with the amine groups versus the carboxyl groups), thepossibility for the formation of intermicrofibrillar cross-links, the degree of cross-linking, thenumber of masked groups (one-sided reactions) and the possible conversion of pendant groups(epoxide into a vic-diol at pH 4.5) may all play a role. To what extent 'decoration' of structuralelements at the level of helices, microfibrils, fibrils and fibers with chemical groups introduced bycross-linking and masking reactions will influence properties like strength, extensibility andmodulus has to be investigated further [29].The mechanical properties of BD90 and BD45 were hardly affected by collagenase degradation(figure 6). Similar results were obtained for DSC cross-linked with the water-soluble carbodiimide3-(3-dimethyl aminopropyl) carbodiimide hydrochloride (EDC) in the presence of N-hydroxysuccinimide (NHS) followed by ethylene oxide sterilization [29]. In contrast, other groupsfound a significant reduction in tensile strength of glutaraldehyde [3, 21] or epoxy compound [10,21] cross-linked (pH 8.5-10.5, 72 h) collagen materials after exposure to collagenase for 24 h,while only a few percent of the initial weight was lost. Based on the discussion above, themechanical properties of a particular cross-linked collagen material may be altered upon enzymaticdegradation depending on the site of attack and the degree of degradation.Spatial limitations determine at what level cross-links can be formed. The distance between twohelices inside a microfibril is 0.15-0.18 nm, while the distance between microfibrils is 1.3-1.7 nm[30] as discussed earlier.

Figure 9. Formation of intra- and interhelical and intermicrofibrillar cross-links.

Chapter 5

- 88 -

Reaction of collagen with EDC/NHS will generally lead to zero-length cross-links between orinside the helices, while glutaraldehyde and BDDGE can also form intermicrofibrillar cross-links(figure 9). Intra- and interhelical cross-links will enhance the resistance towards collagenasedegradation, because EDC/NHS cross-linked DSC was highly resistant against collagenase [4]. Onthe other hand, these materials were not completely resistant towards pronase and 25 % of theinitial weight was lost after 24 h (non-published results), which implies that other types of cross-links enhance the stability towards pronase.For BD45 and BD90 besides intra- and interhelical cross-links, also intermicrofibrillar cross-linksmay be formed. It is expected that these cross-links give an additional stabilization againstenzymatic attack, which was supported by the lack of degradation for BD90 upon exposure tocollagenase for 24 h and pronase for 48 h.The mechanical properties of BD90 and BD45 did not change after exposure to collagenase for 24h. A very slight increase in amine group content was found, without substantially changing the Ts(data not shown), which suggests that the triple-helix is not significantly affected [31]. Despite itshigh stability towards collagenase, BD45 was substantially degraded after 24 h exposure topronase as indicated by a large increase in the number of amine groups from 27 to 36 (n/1000) andthe residual weight of 56 %. Consequently, the tensile strength was dramatically reduced from 5.4to 0.3 MPa, and the elongation at break was decreased from 159 % to 45 % (figure 7). Asdiscussed above, degradation may take place at the level of fibrils and in case of BD45 may finallylead to loss of helical segments from the outer surface of the fibrils. In this way also the outer andinner surface of the fibers will be changed which apparently leads to the change in mechanicalproperties. The mechanical properties of BD90 were only slightly affected upon exposure topronase. The tensile strength was slightly reduced, while the elongation at break was somewhatincreased, which implies that a few modifications due to degradation by pronase had occurred asconfirmed by a slight increase in amine group content.

CONCLUSIONS

Cross-linking of DSC with 1,4-butanediol diglycidyl ether (BDDGE) can be achieved under acidic(pH 4.5) or basic (pH 9.0-10.0) conditions, resulting in completely different materials concerningmechanical properties and in-vitro stability towards enzymes. The resistance of the materialstowards degradation by pronase and collagenase is highly dependent on the cross-link density.Cross-linking of DSC with BDDGE at a pH of 4.5 afforded materials which degrade even if theywere cross-linked for 10 d. Cross-linking at high pH gave a stable material after 72 h at pH 9.0and after 6 h at pH 10.0. The differences in stability of these materials towards collagenase andpronase are explained in terms of locations of the cross-links, the types of cross-links and thespecificity of the enzymes.Upon degradation with pronase, the mechanical properties of BD45 were largely reduced.Exposure of this material to collagenase did not change the mechanical properties. Although theamine content of BD90 and BD45 after degradation with collagenase was somewhat increased noeffect on the mechanical properties was measured. BD90 appeared to be a very stable biomaterialas reflected in an almost negligible change in weight loss and mechanical properties after exposure


- 89 -

to enzymes and is therefore the most suitable material in load-bearing applications. BD45 cross-linked collagen may be suitable as a temporary scaffold.

References



3. L.H.H. Olde Damink, P.J. Dijstra, M.J.A. v. Luyn, P.B. v. Wachem, P. Nieuwenhuis, and J. Feijen,"Changes in mechanical properties of dermal sheep collagen during in-vitro degradation", J. Biomed. Mat.Res., 29 pp. 139-147 (1995).

4. L.H.H. Olde Damink, P.J. Dijstra, M.J.A. v. Luyn, P.B. v. Wachem, P. Nieuwenhuis, and J. Feijen, "In-vitro degradation of dermal sheep collagen cross-linked using a water-soluble carbodiimide", Biomaterials,17(8) pp. 765-774 (1996).

5. J.M. Lee, H.L. Edwards, C.A. Pereira, and S.I. Samii, "Cross-linking of tissue-derived biomaterials in 1-ethyl-3-(dimethylaminopropyl)-carbodiimide", J. Mat. Sci.: Mat. in Med., 7(9) pp. 531-542 (1996).





10. M.A. Moore, I.K. Bohachensky, D.T. Cheung, B.D. Boyan, W.M. Chen, R.R. Bickers, and B.K. McIlroy,"Stabilization of pericardial tissue by dye-mediated photooxidation", J. Biomed. Mat. Res., 28 pp. 611-618(1994).

11. M. Chvapil, D.P. Speer, H. Holubec, T.A. Chvapil, and D.H. King, "Collagen fibres as a temporaryscaffold for replacement of ACL in goats", J. Biomed. Mat. Res., 27 pp. 313-325 (1993).

12. H.W. Sung, C.S. Hsu, and Y.S. Lee, "Physical properties of a porcine internal thoracic artery fixed with anepoxy compound", Biomaterials, 17(24) pp. 2357-2367 (1996).




16. R. Zeeman, P.J. Dijkstra, P.B.v. Wachem, M.J.A.v. Luyn, M. Hendriks, P.T. Cahalan, and J. Feijen,"Cross-linking and modification of dermal sheep collagen using 1,4-butanediol diglycidyl ether", Chapter3 of this thesis and submitted to J. Biomed. Mat. Res. (1998).


18. IUP/2, J. Soc. Leather Trades' Chem., 44 (1960).19. R. Zeeman, P.J. Dijkstra, P.B.v. Wachem, M.J.A.v. Luyn, M. Hendriks, P.T. Cahalan, and J. Feijen, "The

kinetics of 1,4-butanediol diglycidyl ether cross-linking of dermal sheep collagen", Chapter 4 of this thesis,(1998).

20. E. Imamura, O. Sawatani, H. Koyanagi, Y. Noishiki, and T. Miyata, "Epoxy compounds as a newcrosslinking agent for porcine aortic leaflets: subcutaneous implant studies in rats.", J. Cardiac Surg., 4pp. 50-57 (1989).

21. H.-W. Sung, H.-L. Hsu, and C.-S. Hsu, "Effects of various chemical sterilization methods on thecrosslinking and enzymatic degradation characteristics of an epoxy-fixed biological tissue", J. Biomed.Mat. Res., 37 pp. 376-383 (1997).

Chapter 5

- 90 -

22. H.W. Sung, C.S. Hsu, S.P. Wang, and H.L. Hsu, "Degradation potential of biological tissues fixed withvarious fixatives: An in-vitro study", J. Biomed. Mat. Res., 35 pp. 147-155 (1997).

23. Boehringer Mannheim, "Biochemical Catalog", (1998).24. K.A. Piez, "Soluble collagen and the compounds reslting from its denaturation", in "Treatise on Collagen.

Volume I: Chemistry of collagen", Ed. by G.N. Ramachandran, Academic Press Inc. (1967)25. M. Prugniere, N. Domergue, B. Castro, and A. Previero, "Pronase in amino acid technology: Optical

resolution of nonproteinogenic alfa-amino acids", Chirality, 6 pp. 472-478 (1994).26. K.S. Weadock, E.J. Miller, E.L. Keuffel, and M.G. Dunn, "Effect of physical cross-linking methods on

collagen-fiber durability in proteolytic solutions", J. Biomed. Mat. Res., 32 pp. 221-226 (1996).27. M.D. Bond and H.E.v. Wart, "Characterization of the individual collagenases from Clostridium

Histolyticum", Biochemistry, 23 pp. 3085-3091 (1984).28. P.C. Hiemenz, "Polymer Chemistry. The basic concepts", Marcel Dekker, Inc., New York (1984).29. L.H.H. Olde Damink, Structure and properties of crosslinked dermal sheep collagen, PhD thesis,

University of Twente, The Netherlands, p. 167 (1993).30. G.N. Ramachandran, "Structure of collagen at a molecular level", in "Treatise on collagen. Volume 1.

Chemistry of collagen", Ed. by G.N. Ramachandran, Academic Press, London. p. 103-184 (1967)31. T. Hayashi and Y. Nagai, "Effect of pH on the stability of collagen molecule in solution", J. Biochem., 73

pp. 999-1006 (1973).

Success ve epoxy and carbod m de cross-l nk ng of dermal sheep collagen

- 91 -

Chapter 6

Successive epoxy and carbodiimide cross-linking of

dermal sheep collagen




ABSTRACT

Cross-linking of dermal sheep collagen (N-DSC, Ts = 46 °C, number of amine groups = 31 (n/1000)) with 1,4-butanediol diglycidyl ether (BDDGE) at pH 9.0 resulted in a material (BD90) with a high Ts (69 °C), a decreasednumber of amine groups of 15 (n/1000) and a high resistance towards collagenase and pronase degradation.Reaction of DSC with BDDGE at pH 4.5 yielded a material (BD45) with a Ts of 64 °C, hardly any reduction inamine groups and a lower stability towards enzymatic degradation as compared to BD90. The tensile strength ofBD45 (9.2 MPa) was substantially improved as compared to N-DSC (2.4 MPa), whereas the elongation at breakwas reduced from 210 to 140 %. BD90 had a tensile strength of 2.6 MPa and an elongation at break of only 93 %.To improve the resistance to enzymes and to retain the favorable tensile properties, BD45 was post-treated with 1-ethyl-3-(3-dimethyl aminopropyl) carbodiimide (EDC) in the presence of N-hydroxysuccinimide (NHS) to giveBD45EN. Additional cross-linking via the formation of amide bonds took place as indicated by the Ts of 81 °C andthe residual number of amine groups of 19 (n/1000). BD45EN was stable during exposure to both collagenase andpronase solutions. The tensile properties (tensile strength 7.2 MPa, elongation at break 100 %) were comparable tothose of BD45 and glutaraldehyde treated controls (G-DSC). Acylation of the residual amine groups of BD45 withacetic acid N-hydroxysuccinimide ester (HAc-NHS) yielded BD45HAc with a large reduction in amine groups to 10(n/1000) and a small reduction in Ts to 62 °C . The stability towards enzymatic degradation was reduced, but thetensile properties were comparable to BD45.

Chapter 6

- 92 -

INTRODUCTION

Cross-linking of collagen-based materials is an effective method to modify the stability towardsenzymatic degradation and to optimize the mechanical properties. Furthermore, the antigenicitywill be decreased affording more biocompatible materials [1-4]. Traditionally, glutaraldehyde hasbeen used as a cross-linking agent, but with the increased dissatisfaction over the performance ofglutaraldehyde cross-linking as a pre-implantation treatment of tissue-derived biomaterials [4-11],several alternative cross-linking methods have been investigated. These methods can be dividedinto two groups. The first group comprises bi- or multifunctional reagents such as diisocyanates[12] and epoxy compounds [13], which bridge amine groups between two adjacent polypeptidechains. The second group covers reagents which activate carboxylic acid groups of glutamic oraspartic acid residues to react with amine groups of another chain providing cross-links with theformation of amide bonds. Carbodiimides have been generally used to generate the active esters[14, 15].Reagents based on glycidyl ethers have gained increased attention to cross-link collagen-basedtissue. Cross-linking with such reagents results in collagen-based tissues with a good stabilitytowards enzymatic degradation in combination with excellent mechanical properties [16-20].In a previous study [21], the use of 1,4-butanediol diglycidyl ether (BDDGE) for cross-linking ofdermal sheep collagen has been evaluated. Collagen cross-linking was performed at pH values of8-10. Under these conditions, cross-linking will mainly involve amine groups. Acidic conditions(pH 4-6) during the fixation process will evoke a different reaction mechanism in which theepoxide groups react with carboxylic acid groups [21-23]. These different cross-linking conditionsresulted in large differences in material properties. The resistance towards enzymatic degradationof dermal sheep collagen (DSC) cross-linked with BDDGE at pH 8-10 was much higher thanmaterials cross-linked at pH 4-6. However, DSC cross-linked at pH 4.5 had a higher tensilestrength and elongation at break as compared to DSC cross-linked at pH 8-10. Moreover, N-DSCcross-linked at pH 4.5 was flexible and pliable, whereas DSC cross-linked at pH 8-10 was rigidand stiff.In order to retain the excellent tensile properties of collagen cross-linked at pH 4.5, and toincrease the resistance towards enzymatic breakdown, a subsequent cross-linking step using awater-soluble carbodiimide was introduced. In this paper, the two-step cross-linking procedure ofN-DSC will be described. The resulting materials will be evaluated with respect to their stability incontact with enzyme solutions and their mechanical properties.


Preparation of non-cross-linked dermal sheep collagen (N-DSC)Dermal sheep collagen was obtained from the Zuid-Nederlandse Zeemlederfabriek (Oosterhout,The Netherlands) and was prepared as reported previously. The fibrous collagen network waswashed 4 times with deionized water, 2 times with acetone and 2 times with deionized waterbefore lyophilization [21,24].


- 93 -

Cross-linkingAbout 1 g of N-DSC was immersed in 100 ml of a buffered solution containing 4 wt% 1,4-butanediol diglycidyl ether (BDDGE, Fluka, Buchs, Switzerland). The solution was buffered eitherwith 0.05 M 2-[N-morpholino]ethanesulfonic acid (MES, Merck, Darmstadt, Germany) at pH 4.5or with 0.025 M disodium tetraborate decahydrate (Na2B4O7 • 10 H2O z. A., Merck, Darmstadt,Germany) at pH 9.0. Cross-linking was performed at room temperature for 7 d. After cross-linking, the samples were extensively washed with deionized water before lyophilization.In addition, samples cross-linked at pH 4.5 were either treated with HAc-NHS (Procedure A) orwith EDC and NHS (Procedure B).

Procedure A: Amine groups were acylated by reaction with acetic acid N-hydroxysuccinimide ester (HAc-NHS, Sigma Chemicals, St. Louis). The cross-linked sample (1 g) wasimmersed in a buffered solution (0.05 M MES; pH = 6.8) containing 1.35 g of HAc-NHS (25times molar excess with respect to the collagen amine groups). After 16 h of reaction at roomtemperature, the sheet was washed with deionized-water before lyophilization.

Procedure B: A second cross-linking step was carried out by immersing a collagen sample(1 g) in 100 ml of a buffered solution (0.05 M MES; pH = 5.5) containing 1.15 g 1-ethyl-3-(3-dimethyl aminopropyl) carbodiimide hydrochloride (EDC z.S., Merck-Suchardt, Hohenbrunn,Germany) and 0.28 g N-hydroxysuccinimide (NHS z.S., Merck-Suchardt, Hohenbrunn, Germany).After cross-linking for 2 h at room temperature, the sample was washed with 0.1 M NaH2PO4 for2 h and with deionized-water before lyophilization.Glutaraldehyde (GA, purified [25] by distillation (b.p. 80 °C, 16 mm Hg) from 25 % aqueoussolution z.S., Merck, Darmstadt, Germany) cross-linking was performed by immersing 1 g of N-DSC in 100 ml of a 0.5 wt % GA solution in a phosphate buffer (0.054 M Na2HPO4, 0.013 MNaH2PO4, pH 7.4) for 1 h at room temperature. After cross-linking the sample was rinsed withtap-water (15 min), washed with 4 M NaCl (2 times 30 min) and deionized-water (4 times 30 min)before lyophilization.

Characterization

Shrinkage temperatureThe degree of cross-linking of the samples was related to the increase of the shrinkage(denaturation) temperature (Ts). Ts values were determined using an apparatus similar to thatdescribed in IUP/16 [26]. Test specimens were cut, mounted and hydrated. A heating rate of 2°C/min was applied and the onset of shrinkage was recorded as Ts.Thermal analysis of (non)-cross-linked DSC were performed on a Perkin-Elmer DSC-7Differential Scanning Calorimetry which was calibrated with Indium and Gallium. A collagensample (3-6 mg) was put in a volatile sample pan (Perkin-Elmer, stainless steel) and 50 µl of aphosphate buffered saline solution (PBS, 0.14 M NaCl, 0.01 M Na2HPO4, 0.002 M NaH2PO4, pH7.4, NPBI, Emmercompascuum, the Netherlands) was added. The reference contained 50 µl PBS.A heating rate of 2 °C/min was used and a temperature interval between 30 and 95 °C was chosen.The peak temperature, the transition enthalpy (∆Hs) and the transition interval (∆T) weredetermined from the thermograms.

Chapter 6

- 94 -

Amine group contentThe amine group content of the collagen samples was determined spectrophotometrically afterreaction of the primary amine groups with 2,4,6-trinitrobenzenesulfonic acid (TNBS) andsubsequent hydrolysis of the sample [21] and is expressed as the number of groups present per1000 amino acids [n/1000].

Determination of the amount of pendant groupsCross-linked materials weighing 0.10 g were immersed in 10 ml of a buffered solution (0.064 MNaHCO3, 0.036 M Na2CO3, pH 10.0) containing 0.5 M lysine methyl ester dihydrochloride(Sigma Chemicals, St. Louis, USA) for 72 h at 20 °C. After reaction, the materials werethoroughly washed with 1.0 M NaCl (3 times 30 min) and 5 times 30 min with deionized waterbefore lyophilization. The amount of pendant epoxide or aldehyde groups was calculated from thedifference in amine group content by applying a TNBS assay, as described above, on the materialsbefore and after reaction with lysine methyl ester.

Water-absorbing capacityA lyophilized collagen sample (10-15 mg) was accurately weighed (Wdry) and was immersed in 10ml of PBS at room temperature. After swelling for 24 h, the sample was removed from thesolution, blotted with filter-paper and weighed (Wwet). The water absorbing capacity (S) wascalculated from S = (Wwet-Wdry)/Wdry

Enzymatic degradationThe degradation of (non) cross-linked DSC was studied using bacterial collagenase, pronase, andelastase.Bacterial collagenaseThe procedure to determine the resistance of the materials towards a bacterial collagenase (fromClostridium histolyticum (Sigma Chemicals, St. Louis, activity = 315 U/mg) solution with anactivity of 100 U/ml in 0.1 M Tris-HCl buffer has been described previously [27].PronaseThe procedure to determine the resistance of the materials towards a pronase (from Streptomycesgrisseus) solution with an activity of 20 U/ml in a 0.1 M Tris-HCl buffer has been describedpreviously [27].ElastaseAn elastase (Type I, from porcine pancreas, Sigma, St. Louis, activity 100 U/mg, 1 Unit willsolubilize 1 mg of elastin in 20 min at pH 8.8 at 37 °C) stock solution having an activity of 15U/ml, was prepared by dissolving the desired amount of elastase in 0.1 M Tris-HCl buffer (pH 8.8)containing 0.005 M CaCl2 and 0.05 mg/ml NaN3. This stock solution was allowed to stand for 1 hat 37 °C before use.To a collagen sample weighing 4 to 6 mg, 1.0 ml of elastase solution in Tris-HCl buffer (37 °C)was added. Degradation was carried out for 24 h at 37 °C, followed by addition of 0.1 ml of 0.25M EDTA to terminate the degradation. N-DSC and elastin (from bovine neck ligament, Sigma, St.Louis) were used as control materials and the same procedure was used as described above.


- 95 -

The weight-loss of the dermal sheep collagen samples, expressed as the percentage of the initialweight remaining, was related to the hydroxyproline content of the supernatant. An aliquot of 0.5ml of the supernatant was transferred into 1.0 ml 6 M HCl. Hydrolysis of the peptide fragmentswas carried out at 110 °C for 20 h. After hydrolysis, 1.0 ml of 6 M NaOH was carefully added toneutralize the solution to pH ~ 7.0. The solution was diluted by addition of 7.5 ml citricacid/acetate buffer (14.25 g sodium acetate trihydrate (reinst), 9.38 g of trisodium citrate hydrate(p.A.) and 1.38 g of citric acid monohydrate (p.A.) in 100 ml iso-propanol (p.A), diluted to 250 mlwith deionized water, pH 6.0). Exactly 1.0 ml of the buffered hydrolyzate was reacted with 1.0 ml0.5 wt % Chloramine T solution in citric acid/acetate buffer for 15 min at 20 °C. Thereafter, 2.0ml of Ehrlich’s Reagent (7.0 g of dimethylamino benzaldehyde (p.A.) in 12.5 ml 70-72 %perchloric acid (all Merck, Darmstadt, Germany), diluted to 100 ml with iso-propanol) was addedand reacted for 20 min at 65 °C. The samples were cooled to 20 °C and the absorbances wereimmediately measured at 555 nm. A control was prepared by treating 1.0 ml citric acid/acetatebuffer which contained no hydroxyproline as described above. The amount of collagen which wasdegraded was calculated by using N-DSC and elastin as controls.

Mechanical propertiesStress-strain curves of DSC samples were determined by uniaxial measurements using a Zwick(Z020) mechanical tester. Because of variations in the mechanical properties of different parts ofthe sheep skin, only samples were taken from the IUP/2 [28] sampling area parallel to thebackbone.Tensile bars (40.0 mm x 4.0 mm x 1.4 mm) were cut using a dumb-bell shaped knife and hydratedfor at least one h in PBS at room temperature. The thickness of the samples was measured intriplicate using a spring-loaded type micrometer (Mitutoyo, Tokyo, Japan). An initial gauge lengthof 10 mm was used and a crosshead speed of 5 mm/min was applied until rupture of the testspecimen occurred. A pre-load of 0.05 N was applied to pre-stretch the specimen before the realmeasurement. The tensile strength, the elongation at alignment, the elongation at break, the lowstrain modulus and the high strain modulus of the sample were calculated from five independentmeasurements.

RESULTS

Cross-linkingCross-linking of N-DSC (I ) using a diglycidyl ether involves reaction of the epoxy groups withamine groups of (hydroxy)lysine residues (II ) when basic conditions are applied or with twocarboxylic acid residues (aspartic or glutamic acid) if cross-linking is performed under acidicconditions (III ). Acylation of the remaining amine groups of (III ) leads to structure (IV ). Noadditional cross-links are formed during this reaction. Cross-linking of (III ) using 1-ethyl-3-(3-dimethyl aminopropyl)carbodiimide (EDC) and N-hydroxysuccimide (NHS) will result in theformation of additional cross-links. EDC/NHS cross-linking involves the activation of the residualcarboxylic acid groups with EDC to give an O-acylisourea group. In the presence of NHS, this

Chapter 6

- 96 -

group is converted to a NHS activated carboxylic acid group, which is highly reactive towardsamine groups and a cross-link is formed through formation of an amide ('zero'-length [29]) bond.

COOH NH2COOH

NH2 COOH COOH

��

pH 9.0

COOH COOH

COOH COOH

�� pH 4.5

COOHNH2

NH2COOH

NH

CH2

CH(OH)

R

CH(OH)

CH2

NH

��HAc-NHS

COOH NH

C

CH3

O

NH

C O

CH3

COOH

��

EDC/NHS

COOH

NH2

BDDGE

BDDGE

(I)

(II)

(III) (IV)

(V)

C O

O

CH(HO)CH2

R

CH(HO)CH2

O

C O

C O

O

CH(HO)CH2

R

CH(HO)CH2

O

C O

C O

O

CH(HO)CH2

R

CH(HO)CH2

O

C O C O

NH

BDDGE = CH2 CH CH2O

O (CH2)4 O CH2 CH CH2O

HAc-NHS = N O C CH3

OO

O

EDC = CH3 CH2 N C N (CH2)3 N+HCH3

CH3Cl-

NHS = N OH

O

O

Figure 1. Cross-linking of dermal sheep collagen.


- 97 -

The shrinkage (denaturation) temperature (Ts) and the decrease in amine groups of the cross-linked materials were determined and used as a measure of the degree of cross-linking.

Table IShrinkage temperature, and the content of amine and pendant aldehyde or epoxide groups

of (non) cross-linked dermal sheep collagenMaterial Shrinkage

temperature (°C)Amine group content

[n/1000]Pendant aldehyde or

epoxide groups[n/1000]

N-DSC 46 ± 1 31.0 ± 1.0 --G-DSC 70 ± 1 9.1 ± 0.7 5.2 ± 1.5BD90 69 ± 4 14.6 ± 1.0 0.5 ± 0.3BD45 64 ± 4 27.0 ± 0.3 0.5 ± 0.1BD45HAc 62 ± 4 10.5 ± 3.0 1.5 ± 0.4BD45EN 81 ± 3 17.7 ± 2.6 1.8 ± 1.0

N-DSC = non-cross-linked collagenG-DSC = Glutaraldehyde cross-linked collagen (0.5 wt %, 1 h, phosphate buffer (pH 7.4))BD90 = butanediol diglycidylether cross-linked collagen at pH 9.0 (4 wt % solution, 7 d)BD45 = butanediol diglycidylether cross-linked collagen at pH 4.5 (4 wt % solution, 7 d)BD45HAc = butanediol diglycidylether cross-linked collagen at pH 4.5, followed by treatment with the NHS esterof acetic acid (0.05 M MES buffer, molar ratio HAc-NHS:NH2 (collagen) = 25:1, 16 h)BD45EN = butanediol diglycidylether cross-linked collagen at pH 4.5, followed by EDC/NHS cross-linking (0.05M MES buffer, molar ratio EDC:NHS:COOH(collagen) = 5;2:1, 2 h).The amine and pendant epoxide or aldehyde group contents are expressed as the number of amine groups per 1000amino acids. (n=3)

Cross-linking of the dermal sheep collagen under the different conditions applied results in anincrease of the shrinkage (denaturation) temperature. Glutaraldehyde cross-linking (G-DSC)which involves reaction with the amine groups present in the collagen results in an increase of theTs to 70 °C and in a large decrease of the content of amine groups to 9.1 per 1000 amino acidresidues. A similar Ts value was obtained upon cross-linking with the bisepoxy compound(BDDGE) at pH 9.0. The content of amine groups was reduced to 14.6 per 1000 amino acidresidues. Cross-linking with BDDGE at pH 4.5 appeared to give collagen materials with anincrease in Ts to 64 °C but only a small decrease of amine groups to 27.0 per 1000 amino acidresidues. When the collagen cross-linked at pH 4.5 was treated with the acylating agent HAc-NHS, the content of amine groups decreased to a value of 10.5 per 1000 amino acid residues. Thismodification step of the collagen resulted in a slightly lower shrinkage temperature. Applying asecond cross-linking step to the epoxy cross-linked material (BD45) using EDC and NHS resultedin a material with a high Ts (81 °C) and a decrease of amine groups from 27 to 17.7 per 1000amino acid residues. As indicated in table I, the content of free epoxide groups after cross-linkingis very low. On the contrary, G-DSC contained about 5 (n/1000) free aldehyde groups.The endothermic shrinkage transition of the collagen samples was also analyzed by differentialscanning calorimetry. In table II the results of these measurements are presented.

Chapter 6

- 98 -

Table IICharacteristics of the shrinkage endotherm of (non) cross-linked DSC.

Material: Tpeak [°C] ∆T [°C] ∆Hs [J/g tissue] Peak shapeN-DSC 39.4 12.3 15.1 non-symmetricG-DSC 68.8 10.0 10.7 non-symmetric

irregularBD90 72.8 8.8 7.6 non-symmetricBD45 69.6 4.0 16.9 symmetric

BD45HAc 66.7 5.6 16.6 symmetricBD45EN 81.2 8.5 19.3 symmetric

The abbreviations of the materials are explained in table I.∆T: width of the transition peak (figure 2)

Cross-linking not only increases the Ts, but also results in a more narrow transition. BDDGEcross-linking of N-DSC under acidic conditions (BD45) resulted in a material having a Ts with anarrow and symmetrically shaped transition peak. On the contrary, collagen cross-linked via theamine groups (G-DSC and BD90) afforded materials with a lower transition enthalpy (∆Hs) and anon-symmetric transition peak (figure 2 ).

Figure 2. Differential scanning calorimetry curves of 1,4-butanediol diglycidyl ether (BD45) andglutaraldehyde (G-DSC) cross-linked dermal sheep collagen.

Macroscopic appearanceThere were some differences in the macroscopic appearance of the materials obtained. N-DSC wasa white, soft and pliable tissue, while G-DSC was yellow, stiff and rigid. BD90 was a whitecolored material, which was stiff and rigid. On the contrary, the BD45 materials, which wereslightly yellow colored retained the macroscopic characteristics such as pliability and flexibility ofN-DSC. Treatment of BD45 with either HAc-NHS or EDC and NHS did not alter themacroscopic properties.


- 99 -

Enzymatic degradationThe degradation of the N-DSC samples was studied by exposing the materials to either acollagenase or a pronase solution. The change in weight of the different collagen samples wasdetermined as a function of the degradation time.

0 5 10 15 20 250

20

40

60

80

100

N-DSC G-DSC BD45 BD90 BD45HAC BD45ENR

emai

nin

g w

eigh

t [%

]


Figure 3. Remaining weight of collagen samples as a function of the degradation time duringexposure to a bacterial collagenase solution (pH 7.4, 37 °C, n = 3, abbreviations are explainedin table I)

0 10 20 30 40 500

20

40

60

80

100

G-DSC BD45 BD90

BD45HAC BD45EN

Rem

ain

ing

wei

ght

[%]


Figure 4. Remaining weight of collagen samples as a function of the degradation time duringexposure to a pronase solution (pH 7.4, 37 °C, n=3, abbreviations are explained in table I).

N-DSC was degraded within 5 h in the collagenase solution and within 30 min in the pronasesolution. Figures 3 and 4 show the excellent resistance against enzymatic attack of collagen cross-linked with BDDGE at pH 9.0 (BD90) and the collagen first cross-linked by BDDGE at pH 4.5and subsequently by the EDC/NHS method (BD45EN). Dermal sheep collagen cross-linked withBDDGE at pH 4.5 (BD45) afforded a less stable material, that was slightly degraded after 24 h in

Chapter 6

- 100 -

a collagenase solution and much more degraded after 48 h in a pronase solution. The resistancetowards enzymatic degradation was dramatically decreased when the remaining amine groupswere acylated (BD45HAc). Glutaraldehyde cross-linked collagen (G-DSC) has a similar stabilityas BD45.

Table IIIPercentages of degraded material of cross-linked dermal sheep collagen after

exposure to an elastase solution (15 U/ml, pH 8.8, 37 °C, 24 h, n = 3)Material Degraded amount of tissue [%]G-DSC 2.7 ± 0.5BD45 10.8 ± 0.8BD90 0.3 ± 0.1

BD45HAc 22.6 ± 2.0BD45EN 0.3 ± 0.1

The abbreviations used in this table are explained in table I.

Table III summarizes the results of degradation with elastase. Elastin was completely digestedafter 24 h. Moreover, N-DSC was fully degraded due to slow denaturation of the material at 37°C. BD90 and BD45EN appeared to be stable in the elastase medium, while especially BD45 andBD45HAc yielded a considerable amount of degraded collagen.

Water absorbing capacityThe amount of water absorbed was determined to study the influence of the cross-linking methodon the water absorbing capacity of the material.

Table IV Water absorbing capacity of the (non) cross-linked collagen at 20 and 37 °C.

Material Water absorbance at 20 °C [%] Water absorbance at 37 °C [%]N-DSC 610 ± 20 520 ± 40GDSC 600 ± 45 680 ± 10BD45 630 ± 25 650 ± 35BD90 610 ± 35 630 ± 30

BD45HAc 610 ± 35 680 ± 40BD45EN 550 ± 50 600 ± 10

The abbreviations used in this table are explained in table I.

Cross-linking of DSC does not significantly influence the water absorbing capacity. Only the two-step cross-linked collagen material (BD45EN) showed a slight decrease of the water absorbingcapacity. At 37 °C the water uptake of all cross-linked materials is slightly higher than at 20 °C.N-DSC showed a decrease of water-uptake at 37 °C due to partial shrinkage of the material at thistemperature.

Mechanical testing


- 101 -

The influence of the cross-linking process on the mechanical properties of the collagen materialwas determined by uniaxial tensile measurements (Table V).

Table VMechanical characteristics of (non) cross-linked dermal sheep collagen.

Sample: Tensilestrength[MPa]

Elongation atbreak [%]

Elongation atalignment

[%]

Low strainmodulus[MPa]

High strainmodulus[MPa]

N-DSC 2.6 ± 0.4 210 ± 40 44 ± 8 0.8 ± 0.1 2.4 ± 0.4GDSC 7.9 ± 0.2 160 ± 50 34 ± 8 2.0 ± 0.4 5.7 ± 1.0BD90 2.6 ± 0.1 90 ± 10 15 ± 2 1.8 ± 0.5 4.3 ± 0.6BD45 9.4 ± 0.2 130 ± 20 27 ± 5 2.9 ± 0.8 9.4 ± 1.0

BD45HAc 7.1 ± 0.6 140 ± 20 23 ± 6 2.3 ± 0.2 7.0 ± 0.6BD45EN 7.2 ± 0.5 100 ± 5 23 ± 2 2.4 ± 0.4 9.5 ± 1.5

Generally, cross-linking increases the tensile strength and both the low and high strain modulusand decreases the elongation at break and the elongation at alignment.There is a significant difference in mechanical properties between DSC cross-linked with BDDGEat 4.5 (BD45) or at pH 9.0 (BD90). The latter had both a low tensile strength and elongation atbreak, whereas BD45 had a very high tensile strength and a moderate elongation at break. Thetensile properties of BD45 were comparable to those of G-DSC.Successive cross-linking of BD45 with EDC/NHS to give BD45EN did not alter the mechanicalproperties to a large extent. Compared to BD45, BD45EN had a somewhat lower elongation atbreak. If the remaining amine groups of BD45 were acylated with HAc-NHS to give BD45HAc,only the tensile strength and the high strain modulus were slightly decreased.

DISCUSSION

Polyglycidyl ethers have gained much attention as cross-linking reagents for collagen basedbiomaterials during recent years [13, 16-20, 30]. It has been shown that these reagents areeffective cross-linkers for pericardium, vascular grafts and porcine aortic heart valves. Fixation ofcollagen using polyglycidyl ethers gives materials with good mechanical properties [18, 19, 31], adecreased calcification and a lower cytotoxicity [32] compared to glutaraldehyde cross-linkedtissue. However, use of these reagents results in an undefined cross-linked structure because of thepoly-functionality. Furthermore, it is likely that a considerable amount of pendant epoxide groupsare introduced, which are still able to react with (blood)proteins.Recently, cross-linking of dermal sheep collagen (N-DSC) using 1,4-butanediol diglycidyl ether(BDDGE) at pH between 8.5 and 10.5 has been described [21]. Cross-linking involves thereaction of amine groups of (hydroxy)lysine residues with epoxide groups of the BDDGEmolecules, resulting in formation of secondary amines. The enzymatic resistance of the materialswas excellent as shown by the absence of degradation after 48 h of incubation in a bacterialcollagenase or a pronase solution. Furthermore, the tensile strength was slightly increased whileelongation at break was largely reduced as compared to N-DSC [27].

Chapter 6

- 102 -

Reaction of epoxide groups with carboxylic acid groups of glutamic or aspartic acid residues of N-DSC can be achieved at pH 4.5 (BD45) resulting in formation of ester containing cross-links [21].However, the stability of this material towards collagenase and pronase is lower than of DSCcross-linked at pH 8.5-10.5. On the other hand, the mechanical properties of BD45 appeared to besuperior as reflected in a high tensile strength and a high elongation at break. Moreover, a flexiblematerial was found if cross-linking was carried out under acidic conditions, while cross-linkingunder basic conditions resulted in a material with a high stiffness in flexure [27]. In order to obtaina material with similar mechanical and macroscopic properties as BD45 but with a higherresistance towards enzymatic degradation, an additional cross-linking step was applied with 1-ethyl-3-(3-dimethyl aminopropyl) carbodiimide (EDC) in combination with N-hydroxysuccinimide(NHS). In this procedure, carboxylic acid groups are activated by EDC and NHS whichsubsequently react with adjacent amine groups forming amide bonds [14].

The denaturation temperature (Ts) and the percentage of amine groups after cross-linking usingthe above described reagents and conditions are summarized in table I. Glutaraldehyde cross-linking of N-DSC resulted in an increase of Ts from 46 to 70 °C and in a decrease in amine groupcontent from 31.0 to 9.1 (n/1000). N-DSC cross-linked with BDDGE at pH 9.0 (BD90) had acomparable Ts of 69 °C as G-DSC. However, the content of amine groups after cross-linking wasconsiderably higher, namely 14.6 (n/1000). This suggests that the efficacy of BDDGE cross-linking is higher than the GA cross-linking. Both materials were treated with an excess of lysinemethylester to block the residual epoxide or aldehyde groups. An increase in amine groups of 5.2(n/1000) was obtained for G-DSC. For B90, the same treatment resulted in an increase in aminegroups of only 0.5 (n/1000), which is in agreement with previous results [33] and suggests thathardly any pendant epoxide groups were left in the matrix. These results prove the higher cross-linking efficacy of BD90 as compared to G-DSC.BDDGE cross-linking of N-DSC under acidic conditions at pH 4.5 [21] involves carboxylic acidgroups. The high content of residual amine groups (27 n/1000) after cross-linking confirms thismechanism. Nevertheless, it appears that a few amine groups were involved in the cross-linkingreactions. A slightly lower Ts was observed compared to BD90. The additional EDC/NHS cross-linking of BD45 increased the Ts from 64 to 81 °C. Furthermore, the amine content decreasedfrom 27.0 to 17.7 (n/1000), which means that 9 (n/1000) amide cross-links [14] were introduced.Obviously, BDDGE cross-linking does not inhibit further cross-linking and two different types ofbridges will be present in the collagen matrix. Both BD45 and BD45EN contain a low amount ofpendant epoxide groups after cross-linking, probably due to the high rate of hydrolysis of epoxidegroups at pH 4.5 [21]. Blocking of the residual amine groups of BD45 with acetic acid N-hydroxysuccinimide ester (HAc-NHS), resulted in a slight decrease in Ts which is ascribed to thedestabilization effect of the acylated amines [34, 35].

The influence of the cross-linking procedure on the Ts of the cross-linked material can also bedetermined using differential scanning calorimetry. In general, cross-linking results in a narrowertransition peak [36, 37], and in a higher transition enthalpy as a result of a better organization andstabilization of the helices. Cross-linking with BDDGE at pH 4.5 gave a material showing anarrow, symmetric peak, combined with a high transition enthalpy while cross-linking at pH 9.0


- 103 -

resulted in a material exhibiting a non-symmetric and broad transition peak combined with a lowtransition enthalpy. This may indicate that cross-linking via the amines resulted in a lesshomogeneous distribution of the cross-links and in a lower degree of three-dimensionalorganization and packing of the collagen helices. The broad transition peak obtained in G-DSC,and hence the inhomogeneous distribution of cross-links is in agreement with the hypotheses ofCheung et al. [38], who stated that GA cross-linking primarily occurs at the surface of the fibers.When the amine groups of BD45 were acylated by HAc-NHS (BD45HAc) or when a secondcross-linking step was applied (BD45EN), materials were obtained which showed shrinkageendotherms with similar characteristics as for BD45. The highest transition enthalpy (19.3 J/g) wasmeasured for BD45EN, which implies that successive cross-linking resulted in a well-organizedand stabilized material.

The degradation behavior of (non)-cross-linked collagen was determined with the enzymes,collagenase and pronase. Bacterial collagenase from Clostridium histolyticum is capable ofcleaving peptide bonds within the triple helical structure and has a specificity for the Pro-X-Gly-Pro-Y region, splitting between X and Gly. This region is found about 40 times in an α-chain.Pronase, which is a mixture of unspecific endo- and exo proteases, from Streptomyces grisseuscleaves bonds in the non-helical regions of collagen molecules. Pronase might be able to cleavebonds between helices and (micro)fibrils, thus interrupting the aggregates of collagen helices andmicrofibrils. This cleaving mechanism will lead to triple-helices and fragments of microfibrils fromthe collagen structure, which either will dissolve or become suspended as insoluble fragments inthe enzyme solution.Enzymatic degradation of collagen materials is dependent on and determined by its helicalintegrity, the degree of cross-linking and the availability of the cleavage sites. Whereas N-DSCwas fully degraded within 4 h, cross-linking enhances the stability of collagen-based materialsagainst enzymatic degradation. BD90 and BD45EN appeared to be the most stable materials andhardly any degradation was observed. In contrast, glutaraldehyde cross-linked collagen (G-DSC)revealed some degradation in collagenase solution and the material was almost fully fragmented ina pronase solution. Previous studies using G-DSC showed that almost no degradation took placeafter 24 h of exposure to a collagenase solution [7, 39]. On the other hand, studies with G-DSCusing pronase also showed that a significant degradation took place [40]. BD45 appeared to bemore susceptible to enzymatic degradation than BD90. Ten percent of the initial weight was lostupon exposure to collagenase for 24 h and 70 % of the material was lost after 48 h of exposure topronase. In contrast, BD90 was not degraded in both enzyme solutions. This difference in stabilityis most probably due to the lower degree of cross-linking of BD45 and the nature (esters versussecondary amines) of the cross-links. Ester bonds can be more easily cleaved by enzymes than thesecondary amine bonds [41].It was emphasized before [34] that blocking or shielding of amine groups results in a decreasedrate of degradation because the cleavage sites were less accessible, but on the other hand, maskingcan lead to a certain loss of the helical integrity. Treatment of BD45 with acetic acid N-hydroxysuccinimide ester resulted in a reduction of the percentage of amines from 27.0 to 10.7(n/1000). This means that about 16-17 amine groups were acylated. The Ts of the material wasslightly decreased, which is a result of local destabilization or distortion of the triple-helix

Chapter 6

- 104 -

conformation of the collagen molecules or a more random molecular packing of the collagenmolecules, owing to the side branches created by acylation of the (hydroxy)lysine residues [17,35]. BD45HAc had a much lower resistance against collagenase than BD45. Therefore, it appearsthat partial distortion of the triple-helix caused by acylation is more important than shielding orblocking of specific sites for enzymes in determining the degradation rate of this material. Amethod to determine the influence of the distortion of the helix on the resistance against enzymaticdegradation is the use of elastase as a degrading enzyme. Elastase is capable of cleaving peptidebonds of denatured collagen. On the contrary, native collagen is not degraded [42]. The remainingweights of cross-linked collagen materials after exposure to an elastase solution for 24 h weremeasured (table III). Hardly any degradation was observed for G-DSC, BD90 and BD45EN,which suggests that these materials were either not denatured or that the denatured segments weretightly cross-linked. However, a low but significant degradation (10 %) of BD45 implies that somedistortion of the helical structure was present. This implies that many one-sided (masking)reactions took place as was suggested earlier [27]. BD45HAc obtained after acylation of theresidual amine groups of BD45, even shows a higher amount of dissolved collagen (22 %) afterelastase degradation. This means that acylation disrupts the helical structure to a larger extent.It can be concluded that cross-linking of N-DSC can lead to reactions which cause blocking orshielding of specific cleavage sites for enzymes and which partially distort or denature the helicalstructure enhancing degradation by specific enzymes. Furthermore it is generally observed that byincreasing the degree of cross-linking, the resistance against enzymatic attack is increased. This isillustrated nicely by the excellent resistance against enzymatic breakdown of BD45EN, which wascross-linked in two-steps versus the low resistance of BD45, only cross-linked with BDDGE.Linear relationships are observed for the remaining weight of collagen samples as function ofdegradation time using collagenase (figure 3). There is a trend that the rate of degradationdecreases with increasing Ts of the material studied. No linear correlation between the remainingweight and the digestion time was obtained for the degradation of the materials with pronase,indicating that different degradation mechanisms occur at the same time [27]. Because pronasecleaves bonds in the non-helical telo-peptides, cross-links formed in these regions would diminishthe degradation rate. However, the telo-peptides of collagen type I do not contain amine residuesand only a few carboxylic acid residues [43]. This indicates that hardly any cross-links will beformed in the telo-peptides. Therefore it is most likely that pronase mainly breaks or interruptsaggregates of collagen (micro)fibrils by cleaving native bonds. Furthermore, ester-containingcross-links can be slowly degraded by pronase [27].Lee et al. [18] distinguished between two types of cross-links. The first type are intrahelical cross-links, which are formed between two polypeptide chains in the same helix, and will mainlyinfluence properties like Ts and stress relaxation. The second type are interhelical cross-linkswhich are formed between polypeptide chains of two adjacent helices, and which influence theswelling and the apparent extensibility. However, they did not mention the effect of interhelicalcross-links on the Ts. Cross-links can also be formed between two adjacent microfibrils, if thedistance between two microfibrils (1.3 - 1.7 nm) is smaller than the length of the reagent. Forinstance, BDDGE can bridge between two chains which have a distance up to 2.1 - 2.6 nm, whileEDC/NHS can couple groups which are located within 1.0 nm from each other. Therefore it canbe assumed that BDDGE can and EDC/NHS cannot form intermicrofibrillar cross-links. Distances


- 105 -

between fibrils and fibers are at least one order of magnitude higher [44, 45], which means that nocross-links will be formed at these levels. Water uptake experiments show no influence of thecross-linking procedure on the swelling (table IV). Because of the porous structure of DSC, theinfluence of cross-linking between or within the microfibrils will be negligible. However, thewater-uptake results confirm that no cross-links were formed at a higher level of the collagenhierarchy (fiber or fiber bundles).The enzymatic degradation experiments showed that the stability of a collagen material is mainlydetermined by the intra- and interhelical cross-links. BD90, which has a high Ts, was very stabletowards enzymatic breakdown. BD45, which contained a lower degree of intra- and interhelicalcross-links as indicated by the Ts, showed a lower enzymatic resistance. An additional EDC/NHSstep improved the stability to a large extent. Because, EDC/NHS activation can only lead to intra-and interhelical cross-links, these cross-links determine the increased stability. The sharp increasein Ts confirms the introduction of intra- and interhelical cross-links. BD90 was not degraded bypronase, implying that also intermicrofibrillar cross-links were present. The susceptibility of BD45for pronase indicates that few intermicrofibrillar cross-links were present or that the diester cross-links were hydrolyzed by pronase.

Differences between the concentration of intra-, interhelical or intermicrofibrillar cross-linkspresent in BD45 and BD90 might also explain the remarkable differences in macroscopicproperties. Cross-linking via the amine groups resulted in a stiff and rigid material probably causedby interfibrillar cross-links. These cross-links also lead to some planar shrinkage of the material,which might reduce the pliability. The same is observed during glutaraldehyde cross-linking [18,46]. The flexibility and pliability of BD45 implies that few intermicrofibrillar bridges were formed.This is further substantiated by the fact that after acylation (no additional cross-linking) orEDC/NHS cross-linking (only intra- and interhelical bonds) the flexibility of the material was notchanged.

Cross-linking will affect the mechanical properties [19, 43, 45-47]. The stress-strain curve of (non)cross-linked dermal sheep collagen can be divided into four distinctive parts. Initially, the fiberbundles are randomly orientated and only low stresses are needed to straighten the bundles (lowstrain modulus). The initial orientation of the fiber bundles can be expressed as the elongation atalignment. As more and more bundles become taut, an increase in the modulus is observed. Thelinear part of the curve at high strains is called the high strain modulus. At a sufficient stress, thematerial starts to yield and finally breaks.

Chapter 6

- 106 -

Figure 5. Uniaxial stress-strain curves of (non)-cross-linked dermal sheep collagen(20 °C, distance between clamps: 1 cm, pull speed: 10 mm/min, n=5).

Stress-strain curves of the cross-linked DSC reveal an increase in tensile strength, and modulus buta reduction of both elongation at alignment and break as compared to N-DSC (table V and figure5). Contradictory results of the effect of cross-linking on the mechanical properties of collagen-based materials were obtained in literature. Porcine aortic leaflets and bovine pericardium whichwere cross-linked with epoxy compounds or glutaraldehyde had a slight increase in elongation atbreak and a similar or slightly reduced tensile strength as compared to non-fixed materials [19,47]. The higher extensibility can be ascribed to the crimping of the fiber network during cross-linking which increases the angle of weave of the fiber bundles, resulting in a higher elongation atalignment and a higher elongation at break. Fixation of pericardium with glutaraldehyde results inshrinkage of the tissue, which also results in higher extensibilities [46]. Cross-linking ofpericardium with EDC and NHS exhibited a higher elongation at break and a low stress relaxation[15], while DSC cross-linked with EDC/NHS had a similar elongation at break but a lower tensilestrength compared to non-cross-linked DSC [14]. Furthermore, the low strain modulus wasincreased which was explained by an increased modulus or stiffness of the fibers. The effect ofcross-linking on the mechanical properties of reconstituted collagen, which does not contain thishighly ordered hierarchy as found in DSC, is different. Cross-linking of these collagen fibers withdifferent physical and chemical methods always increased the tensile strength to a large extent [48-50]. Nevertheless it is still not completely clear which level of the collagen structure determinesthe mechanical properties of highly structured fibrous collagen. The interactions between thefiber/fiber or fibril/fibril surfaces are important parameters [27, 51].During the initial part of the stress-strain experiment, the collagen fibers and bundles of DSC willalign. The presence of chemical cross-links should not affect this behavior much, but as indicatedin table V, the cross-linked samples have a lower elongation at alignment and a higher low strainmodulus. It is hypothesized that 'decoration' (cross-links and one-sided reactions) of the outer


- 107 -

surface of fibrils and fibers alter the interactions between the fibril/fibril and fiber/fiber surfaces.Furthermore, the modulus or the stiffness of the fibrils and fibers is increased by cross-linking.These changes in interactions and stiffness may also account for the higher strain modulus (4.3 -9.5 MPa) of cross-linked DSC in relation to N-DSC (2.4 MPa).The significant difference in tensile properties between BD90 and BD45 may be partly explainedby the content or the nature (ester or secondary amine bonds) of intermicrofibrillar cross-links.These cross-links may hamper alignment and will hinder slippage of the fibrils and fibers,consequently creating internal stresses in the material. If the intermicrofibrillar cross-link density ishigh, a reduction in elongation at break is expected as observed in BD90. BD45 demonstrates ahigher extensibility and a much higher tensile strength probably due to the lower interfibrillarcross-link density as compared to BD90. The additional treatments performed on BD45 withHAc-NHS or EDC/NHS did not affect the mechanical properties to a large extent because noadditional intermicrofibrillar cross-links were formed. BD45EN shows a lower elongation at breakprobably because of the increased stiffness of the fibrils caused by the zero-length cross-links. Noadditional effect of the modified fibril and fiber surfaces due to acylation of amine groups of BD45with HAc-NHS was observed.Finally, it seems that there is a correlation between the mechanical properties and macroscopicalpliability of the cross-linked materials. BD90 which is stiff and rigid displays a low tensile strengthand elongation at break. BD45, BD45HAc, and BD45EN which were pliable and flexible had ahigher tensile strength and elongation at break. The properties of G-DSC are somewhat inbetween.

CONCLUSIONS

Cross-linking of dermal sheep collagen (DSC) with 1,4-butanediol diglycidyl ether can eitheroccur via the carboxylic acid groups (reaction at pH 4.5) or via the amine groups (reaction at pH9.0). Cross-linking of DSC at pH 4.5 resulted in good macroscopic and mechanical properties, butin relatively poor resistance towards collagenase and pronase. An additional EDC/NHS cross-linking step improved the enzymatic resistance without altering the other properties. Cross-linkingof DSC at pH 9.0 resulted in materials with a high degree of cross-linking and excellent enzymaticresistance, but with less desirable mechanical properties.The material properties, such as Ts, swelling, in-vitro degradation and tensile properties can becorrelated with the presence of inter-, and intrahelical and intermicrofibrillar cross-links. Cross-links between and in the helices will affect the Ts and the resistance against collagenase, while themechanical properties and the resistance towards pronase are apparently dependent on theintermicrofibrillar cross-links (degree and nature) and the type of groups available at the outersurface of (micro)fibrils and fibers.

References


Chapter 6

- 108 -



4. E.E. Sabelman, "Biology, biotechnology, and biocompatibility of collagen", in "Biocompatibility of tissueanalogs", Ed. by D.F. Williams, CRC Press Inc, Boca Raton (1985)

5. A. Cooke, R.F. Oliver, and M. Edward, "an in vitro cytotoxicity study of aldehyde treated pig dermalcollagen", Br. J. exp. Pathol., 64 pp. 172-176 (1983).


7. P.B. v. Wachem, M.J.A. v. Luyn, L.H.H. Olde Damink, P.J. Dijkstra, J. Feijen, and P. Nieuwenhuis,"Biocompatibility and tissue regenerating capacity of crosslinked dermal sheep collagen", J. Biomed. Mat.Res., 28 pp. 353-363 (1994).



10. P.B. v. Wachem, M.J.A. v. Luyn, L.H.H. Olde Damink, J. Feijen, and P. Nieuwenhuis, "Tissueinteractions with dermal sheep collagen implants: A transmission electron microscopical evaluation",Cells and Mat., 1(3) pp. 251-263 (1991).










20. Z. Tang and Y. Yue, "Crosslinkage of collagen by polyglycidyl ethers", ASAIO J., 41 pp. 72-78 (1995).21. R. Zeeman, P.J. Dijkstra, P.B. v. Wachem, M.J.A. v. Luyn, M. Hendriks, P.T. Cahalan, and J. Feijen,

"Cross-linking and modification of dermal sheep collagen using 1,4-butanediol diglycidyl ether", Chapter3 of this theis and submitted to J. Biomed. Mat. Res. (1998).



24. T.M.v. Gulik and P.J. Klopper, "The processing of sheepskin for use as a dermal collagen graft - anexperimental study", Neth. J. Surg., 39(3) pp. 90-94 (1987).


26. IUP/16, "Measurement of shrinkage temperature", J. Soc. Leather Techn. Chem., pp. 122-126 (1963).


- 109 -

27. R. Zeeman, P.J. Dijkstra, P.B. v. Wachem, M.J.A. v. Luyn, M. Hendriks, P.T. Cahalan, and J. Feijen, "In-vitro degradation of dermal sheep collagen cross-linked with 1,4-butanediol diglycidyl ether", Chapter 5 ofthis thesis, (1998).

28. IUP/2, J. Soc. Leather Trades' Chem., 44 (1960).29. Z. Graberek and J. Gergely, "Zero-length cross-link procedure with the use of active esters", Analyt.

Biochem., 185 pp. 131-135 (1990).30. H.W. Sung, W.H. Chen, I.S. Chiu, H.L. Hsu, and S.A. Lin, "Studies on epoxy compound fixation", J.

Biomed. Mat. Res. Appl. Biomat., 33 pp. 177-186 (1996).31. S.H. Shen, H.W. Sung, R. Tu, C. Hata, D. Lin, Y. Noishiki, and R.C. Quijano, "Characterization of

polyepoxy compound fixed porcine heart valve bioprostheses", J. Appl. Biomat., 5 pp. 159-162 (1994).32. C. Nishi, N. Nakajima, and Y. Ikada, "In vitro evaluation of cytotoxicity of diepoxy compounds used for

biomaterials", J. Biomed. Mat. Res., 29 pp. 829-835 (1995).33. R. Zeeman, P.J. Dijkstra, P.B. v. Wachem, M.J.A .v. Luyn, M. Hendriks, P.T. Cahalan, and J. Feijen, "The

kinetics of 1,4-butanediol diglycidyl ether cross-linking of dermal sheep collagen", Chapter 4 of this thesis,(1998).


35. H.W. Sung, C.S. Hsu, S.P. Wang, and H.L. Hsu, "Degradation potential of biological tissues fixed withvarious fixatives: An in-vitro study", J. Biomed. Mat. Res., 35 pp. 147-155 (1997).

36. W.K. Loke and E. Khor, "Validation of the shrinkage temperature of animal tissue for bioprosthetic heartvalve application by differential scanning calorimetry", Biomaterials, 16(3) pp. 251-258 (1995).



39. L.H.H. Olde Damink, P.J. Dijkstra, M.J.A. v. Luyn, P.B. v. Wachem, P. Nieuwenhuis, and J. Feijen,"Influence of ethylene oxide gas treatment on the in vitro degradation behaviour of dermal sheepcollagen", J. Biomed. Mat. Res., 29(2) pp. 149-155 (1995).

40. H.J.P. Triepels, M. Hendriks, and M.L.P.M. Verhoeven. "A study on a new crosslinking method forcollagen-based materials", in Alternate tissue fixation technology assessment & review Colloquium atMedtronic Bakken Research Center, Maastricht (1996).

41. L. Stryer, "Biochemistry", fourth ed., New York, USA: W.H. Freeman and Company.(1995).42. Boehringer Mannheim, "Biochemical Catalog", (1998).43. E. Heidemann, "Fundamentals of leather manufacturing", 1st ed., Munchen, Germany: Hanser

Publishers.(1993).44. G.N. Ramachandran, "Structure of collagen at a molecular level", in "Treatise on collagen. Volume 1.

Chemistry of collagen", Ed. by G.N. Ramachandran, Academic Press, London. p. 103-184 (1967)45. T.N. v. Gulik, Processed sheep dermal collagen as a biomaterial, PhD Thesis, University of Amsterdam,

The Netherlands (1981).46. P.F. Gratzer and J.M. Lee, "Altered mechanical properties in aortic elastic tissue using

glutaraldehyde/solvent solutions of various dielectric constant", J. Biomed. Mat. Res., 37 pp. 497-507(1997).

47. J.M. Lee, S.A. Haberer, and D.R. Boughner, "The bovine pericardial xenograft: I. Effect of fixation inaldehydes without constraint on the tensile viscoelastic properties of bovine pericardium", J. Biomed. Mat.Res., 23 pp. 457-475 (1989).

48. L.D. Bellincampi and M.G. Dunn, "Effect of cross-linking method on collagen fiber-fibroblastinteractions", J. Appl. Polym. Sci., 63 pp. 1493-1498 (1997).

49. G.D. Pins, E.K. Huang, D.L. Christiansen, and F.H. Silver, "Effects of static axial strain on the tensileproperties and failure mechanisms of self-assembled collagen fibers.", J. Appl. Polym. Sci., 63 pp. 1429-1440 (1997).

50. Y.P. Kato and F.H. Silver, "Formation of continuous fibers: Evaluation of biocompatibility and mechanicalproperties", Biomaterials, 11 pp. 169-175 (1990).

51. L.H.H. Olde Damink, Structure and properties of crosslinked dermal sheep collagen, PhD Thesis,University of Twente, The Netherlands (1993)

Character zat on and b ocompat b l ty of epoxy cross-l nked dermal sheep collagen

- 111 -

Chapter 7

Characterization and biocompatibility of epoxy

cross-linked dermal sheep collagens

P.B. van Wachem1, R. Zeeman2, P.J. Dijkstra2, J. Feijen2, M. Hendriks3, P.T.

Cahalan3, M.J.A. van Luyn1

University of Groningen, Faculty for Medical Sciences, Cell Biology and Biomaterials, Bloemsingel 10, 9712 KZGroningen, The Netherlands; 2 University of Twente, Department of Chemical Technology, and Institute ofBiomedical Technology , P.O. Box 217, 7500 AE Enschede, The Netherlands; 3Medtronic Bakken ResearchCenter B.V., P.O. 1220, 6201 MP, Maastricht, The Netherlands.

ABSTRACT

Dermal sheep collagen (DSC) cross-linked with 1,4-butanediol diglycidyl ether (BDDGE) was characterized andthe biocompatibility was evaluated after subcutaneous implantation in rats. Cross-linking at pH 4.5 (BD45)increased the shrinkage temperature (Ts), but hardly reduced the number of amine groups. Cross-linking at pH 9.0(BD90) or with a successive carbodiimide-step (BD45EN) resulted in a slightly higher Ts with clear reduction inamine groups. Acylation of the residual amine groups of BD45 yielding BD45HAc showed a reduction in aminesin combination with the lowest Ts.Evaluation of implants showed that BD45, BD90 and BD45EN were biocompatible. With BD45HAc, a high influxof granulocytes and macrophages was observed, but this subsided at day 5. At week 6, BD45 had completelydegraded, and BD45HAc was remarkably size-reduced, while BD45EN showed clear reduction in size of the outerdermal sheep collagen bundles. BD90 showed none of these features. This agreed with the observed degree ofmacrophage accumulation and giant cell formation. None of the materials was found to calcify. For the purpose ofsoft tissue replacement, BD90 was defined as the material of choice. Despite its reduced pliability as compared tothe other materials, BD90 combined biocompatibility, low cellular ingrowth, low biodegradation, and absence ofcalcification with fibroblast ingrowth and collagen new formation.

Chapter 7

- 112 -

INTRODUCTION

Collagen-based biomaterials have been utilized in a variety of human clinical applications [1, 2]such as wound dressings [3], artificial skin and nerves [4], vascular grafts [5], anterior cruciateligaments [6, 7] and aortic heart valves [8, 9]. Cross-linking of the implant is necessary to reducethe in-vivo degradation and to minimize the antigenicity [1, 2]. The most widely applied cross-linking agent is glutaraldehyde (GA) [10, 11], which provides very stable materials. However,glutaraldehyde-treated biomaterials show a high cytotoxicity [12, 13]. Furthermore, the durabilityof (collagen-based) heart valve bioprostheses is markedly reduced by the occurrence ofcalcification [8, 14, 15]. In spite of the unknown origin of collagen calcification, it is generallybelieved that GA cross-linking is one of the major determinants of calcific depositions [8, 15].Therefore, alternative cross-linking methods have been developed.In order to study the relation between the cross-linking method and the biocompatibility of theresulting material, dermal sheep collagen (DSC) was used as a model tissue. The GA cross-linkingwas optimized [16]. Furthermore, DSC was cross-linked with hexamethylenediisocyanate [17], orwith a water-soluble carbodiimide [18]. The biocompatibility of these cross-linked materials wasevaluated by subcutaneous implantation in rats [19]. GA cross-linked DSC with a reducedcytotoxicity was obtained by optimizing the cross-linking proces. However after implantation ofthis material still significant calcification was observed [19, 20]. Carbodiimide cross-linking, i.e., 1ethyl-3-(3-dimethyl aminopropyl) carbodiimide (EDC) in combination with N-hydroxysuccinimide(NHS), resulted in a very stable material which exhibited no cytotoxicity. A moderate calcificationwas observed only at 6 weeks, but not at later time periods after implantation. This EDC/NHScross-linked dermal sheep collagen was a promising biomaterial because it enabled collagenformation and functioned as a guidance for muscle overgrowth [19, 21]. These results proved thatthe cross-linking method is an important factor in determining the biocompatibility of collageneousmaterials.Because epoxy cross-linked collagen showed a large reduction in calcification compared to GA-cross-linked materials [22, 23], the use of 1,4-butanediol diglycidylether (BDDGE) wasinvestigated [24]. The in-vitro resistance to degradation of BDDGE cross-linked collagen at pH9.0 (BD90) was excellent, whereas DSC cross-linked at pH 4.5 (BD45) showed partialdegradation. On the contrary, BD45 had a much higher tensile strength and elongation at breakthan BD90 and a soft and pliable structure. Acylation of the residual amine groups of BD45 usingacetic acid-NHS (HAc-NHS) yielded a material (BD45HAc) with similar tensile properties asBD45, but with a reduced in vitro resistance against enzymatic degradation [25]. In order toretain the mechanical properties of BD45 but to improve its stability, a successive epoxy andcarbodiimide (BD45EN) cross-linking method was developed, which resulted in a material with ahigh resistance against enzymatic attack [25].In the present study, epoxy cross-linked dermal sheep collagen materials, further referred to asBD45, BD90, BD45HAc and BD45EN, were characterized and implanted in rats to study theeffects of the cross-linking procedure on the biocompatibility.


- 113 -


Preparation of non-cross-linked dermal sheep collagenDermal sheep collagen (DSC) was obtained from the Zuid-Nederlandse Zeemlederfabriek(Oosterhout, The Netherlands). In brief, the skin was depilated and immersed in a lime-sodiumsulfide solution to remove the epidermis [26]. Non-collagenous substances were removed usingproteolyic enzymes whereafter the skin was split to obtain the dermal layer. The remaining fibrouscollagen network was washed with water (4 times), with acetone (2 times) and with deionized-water (2 times) before lyophilization [16], yielding N-DSC.

Cross-linkingAbout 2 g of N-DSC was immersed in 200 ml of a buffered solution containing 8 g 1,4-butanedioldiglycidyl ether (BDDGE, Fluka, Buchs, Switzerland). The solution was buffered either with 0.05M 2-[N-morpholino]ethanesulfonic acid (MES, Merck, Darmstadt, Germany) at pH 4.5 or with0.025 M disodium tetraborate decahydrate (Na2B4O7 • 10 H2O z. A., Merck, Darmstadt,Germany) at pH 9.0. Cross-linking was performed at room temperature for 7 d. After cross-linking, the samples were rinsed with tap-water and extensively washed with deionized-waterbefore lyophilization yielding BD45 or BD90.In addition, samples cross-linked at pH 4.5 were either treated with HAc-NHS (Procedure A) orwith EDC and NHS (Procedure B).

Procedure A: Amine groups were acylated by reaction with acetic acid N-hydroxysuccinimide ester (HAc-NHS, Sigma Chemicals, St. Louis). The cross-linked sample (2 g) wasimmersed in a buffered solution (0.05 M MES; pH = 6.8) containing 2.7 g of HAc-NHS (25 timesmolar excess with respect to the collagen amine groups). After 16 h of reaction at roomtemperature, the sheet was washed with deionized-water before lyophilization yielding BD45HAc.

Procedure B: A second cross-linking step was done by immersing the collagen sample(2 g) in a buffered solution (0.05 M MES; pH = 5.5) containing 2.3 g 1-ethyl-3-(3-dimethylaminopropyl) carbodiimide hydrochloride (EDC z.S., Merck-Suchardt, Hohenbrunn, Germany)and 0.56 g N-hydroxysuccinimide (NHS z.S., Merck-Suchardt, Hohenbrunn, Germany). Aftercross-linking for 2 h at room temperature, the sample was washed with 0.1 M NaH2PO4 for 2 hand with deionized-water before lyophilization resulting in BD45EN.

CharacterizationThe degree of cross-linking of the dermal sheep collagen samples was related to the increase inshrinkage temperature (Ts). The Ts was measured using differential scanning calorimetry. Thepeak of the transition endotherm was referred to as Ts. The amine group content of the (non)-cross-linked samples was determined spectrophotometrically after reaction with 2,4,6-trinitrobenzene sulfonic acid (TNBS; Fluka, Buchs, Switzerland) and subsequent hydrolysis of thesample [24], and is expressed as the number of amine groups per 1000 amino acids [n/1000].

Sterilization

Chapter 7

- 114 -

BD45, BD90, BD45HAc, and BD45EN, were punched into discs with a diameter of 8 mm andsterilized by ethylene oxide [27]. Lyophilized collagen samples were exposed to a 100% ethyleneoxide atmosphere at a relative humidity of 70 % for 5 hours at 55 °C. Subsequently the sampleswere aerated with a warm air flow at atmospheric pressure for at least 48 h to remove residualethylene oxide.

ImplantationNational Institutes of Health guidelines for the care and use of laboratory animals (NIH 85-23Rev. 1985) were observed. Male Albino Oxford rats of approximately 3 months of age were used.Subcutaneous pockets were made to the right and left of two mid-line incisions on the back. Discswere implanted in the pockets at a distance of about 1 cm from the incisions. Implants withsurrounding tissue were carefully dissected from the subcutaneous site at 1, 2, 5, or 10 d and 3 or6 weeks.

MicroscopyImplants were immediately immersion-fixed in 2% (v/v) glutaraldehyde in 0.1 mol PBS (pH7.4).

After at least 24 h of fixation at 4 °C, specimens were cut into small blocks. The blocks were post-fixed in 1% OsO4, 1.5% K4Fe(CN)6 in PBS, dehydrated in graded alcohols, and embedded inEpon 812. Sections of 1 µm for light-microscopic evaluations were stained with toluidine blue. Toevaluate possible calcification, von Kossa staining was used with specimens from day 10, week 3and week 6.

RESULTS

CharacterizationThe Ts and the amine group content of (non)-cross-linked DSC was determined. The results aresummarized in Table I.

Table IThe shrinkage temperature (Ts) and the content of amine groups of the (non) cross-linked

materials.Sample Ts [°C] Amine groups [n/1000]N-DSC 47.2 ± 0.8 32.0 ± 1.0BD45 68.8 ± 0.4 28.4 ± 1.2BD90 72.5 ± 0.1 13.8 ± 0.5

BD45HAc 65.6 ± 0.2 14.9 ± 0.6BD45EN 79.6 ± 0.5 19.2 ± 1.1

An increase in Ts indicates that cross-linking had occurred. Upon cross-linking of DSC at pH 4.5hardly any reduction in amine groups was found, whereas at pH 9.0 the content of amine groupswas reduced from 32 to 14 (n/1000). Furthermore, the Ts of BD90 was somewhat higher (72.5°C) than of BD45 (68.8 °C). Acylation of the residual amines of BD45 was successful as indicated


- 115 -

by a reduction in amine groups from 28 to 15 (n/1000), whereas the Ts was only slightly declined.BD45EN revealed the highest Ts in combination with 19 (n/1000) amine groups.Before implantation, the materials were sterilized with ethylene oxide. Table II shows the influenceof sterilization treatment on the amount of amine groups and the shrinkage temperature (Ts).

Table IIProperties of (non)cross-linked dermal sheep collagen before and after sterilization with

ethylene oxide.Sample Shrinkage

temperature [°C]Amine groups

[n/1000]∆Ts [°C] ∆ Amine groups

[n/1000]N-DSC 41.7 ± 0.5 28.4 ± 0.5 5.5 3.6BD45 67.0 ± 0.3 19.9 ± 0.9 1.8 8.5BD90 71.0 ± 0.5 7.3 ± 0.4 1.5 6.5

BD45HAc 66.0 ± 1.0 10.5 ± 0.4 - 0.4 4.4BD45EN 77.4 ± 0.6 14.2 ± 0.4 2.2 5.0

∆ = Value before sterilization - value after sterilization

Sterilization resulted in a decrease in amine groups and a small change in Ts. N-DSC had thelargest drop in Ts of 5.5 °C, whereas a reduction of 1 - 2 °C was obtained for cross-linkedsamples. On the contrary, a reduction in amine groups of 4 (n/1000) was found for N-DSC, whilethe content of amine groups of cross-linked materials was diminished by 4 to 9 (n/1000).

Macroscopical evaluationDifferences in macroscopical structure within the group of BDDGE cross-linked materials wereobserved. BD90 had a rather stiff and rigid structure and a white color. In contrast, both BD45and BD45EN were soft and pliable materials with a yellowish color. BD45HAc also had ayellowish color, but its structure was somewhat more rigid than BD45At day 1 after implantation each of the materials was surrounded by a similar thin fragile capsule.At day 2 and 5, the capsule around BD45HAc was somewhat thicker than for the other materials.At day 10 the presence of blood vessels was more apparent for BD45HAc than for the othermaterials. After 3 weeks, BD45 became smaller, and at week 6 this material was almost fullydegraded and could not be retrieved anymore. At week 6, BD45HAc was similarly reduced in sizeas BD45 after 3 weeks. In contrast, BD90 and BD45EN still had the original size.

Light microscopy (see also Table III)The macroscopical fibrous structure of dermal sheep collagen samples consists primarily of amatrix of collagen bundles (figure 1, actually shown by the day 1-implants). Some remnants ofblood vessels, mainly elastin, were observed as white (i.e. non-stained with toluidine blue) plaques.BD45 (figure 1a) showed an alternation of sectioned collagen bundles with spaces in between.With BD90 (figure 1b), collagen bundles looked more like a network. With BD45HAc, collagenbundles were positioned nearer to each other, although the impression of a network, as withBD90, was not obtained. BD45EN looked like a collection of larger collagen bundles with internalspaces.

Chapter 7

- 116 -

Figure 1. Materials at day 1 after implantation. Magnification: 100xa) BD45: The collection of dermal sheep collagen (DSC) bundles shows differences in bundle size anddirection with relatively much space in between. At the left hand side the capsule (C) is present; from heremigration of macrophages and granulocytes occurs. Most pronounced are some accumulations ofmacrophages around the outer collagen-bundles (arrows).b) BD90: The collection of DSC bundles looks more like a network, leaving less space in between. Thecapsule (C) at the left hand side is hardly activated, showing only limited cellular infiltration.c) BD45HAc: DSC bundles were generally closer together, but do not show a network structure as withBD90. This material shows the most activated capsule (C) with the highest numbers of macrophages andgranulocytes and widespread fibrin formation. Less macrophage accumulation than with BD45 isobserved.d) BD45EN: This material looks similar to BD45, but most collagen bundles show cracks inside bundles(arrows); furthermore, differences in staining intensity are obvious. A fine fibrin (F) network is present inbetween the bundles. Capsule (C) activation and cellular infiltration are in between those observed withBD45 and BD90.

At day 1 after implantation, usually migration of macrophages and granulocytes occurs. WithBD45 (figure 1a) more cells infiltrated than with BD90 (figure 1b), which was almost empty, andwith BD45EN (figure 1d), in which fibrin networks were obvious. At the edges of BD45 someaccumulations of mainly macrophages were seen. In contrast, much more cellular activation wasobserved with BD45HAc (figure 1c). Both in the thick capsule and at the outer edge of thematerial many granulocytes and macrophages were present. Fibrin formation was widespread.


- 117 -

Figure 2. Materials at day 2 after implantation. Magnification: 100xa) BD90: Cellular infiltration occurs the least in this material; in fact, apart from small macrophageaccumulations, fibrin (F) networks are most obvious. C: capsule.b) BD45HAc: This material was clearly most reactive as is especially obvious from the thick cellularaccumulations in between collagen-bundles at the edges. White plaques (P) represent remnants of formerblood vessels. The capsule (C) also shows high activation with a lighter stained area with fibrin and tothe left of that more cells and some large blood vessels (V).Figure 3. Materials at day 5 after implantation.a) Made at 200x. BD45 shows a complete layer of giant (G) cells at the interface, while the large centralpart is more or less empty. P: white plaque; V: blood vessel in capsule.b) Made at 400x. BD45HAc contains one site with a concentration of macrophages with manyintracellular lipid droplets (arrows), and degenerating cells, present in fibrin (F) networks in betweencollagen-bundles (B).

At day 2 after implantation, usually granulocytes and macrophages still infiltrated, while the onsetof encapsulation by fibroblasts is observed. BD45 and BD45EN both show little infiltration ofmacrophages and granulocytes. The accumulation of macrophages resulted in concomitant onsetof giant cell formation at the edges. BD90 (figure 2a) was surrounded by a loose capsule whichcontained granulocytes and macrophages, but no fibroblast organization was observed yet. Insidethe material, fibrin networks were most obvious. Limited infiltration of granulocytes andmacrophages was observed in deeper areas, while small macrophage accumulations were presentat the outer edges. BD45HAc (figure 2b) was highly reactive, showing a lot of fibrin in the looseconnective tissue, high migrative activity in the capsule, and thick accumulations with fibrin-embedded (degenerating) granulocytes and macrophages in the edges of the material. Intra- and

Chapter 7

- 118 -

extracellular lipid droplets were observed both in the capsule and inside the material. Only themost central area of the disc showed no cellular infiltration. Some former blood vessels wereobserved as white plaques (Figure 2b).

Table III Cellular infiltration/ingrowth and other remarks

BD45 BD90 BD45HAc BD45ENday 1*

day 2*

day 5 #

day 10 #

week 3 #

week 6 #

+ +accum. macrophages

+giant cell formation

+ +

+ +

+ + + + +size reductionlipid/small giant cells

n.r.

±±±±

+ +

±±±±fibroblasts: ±±±±giant cell formation

+ ±±±±fibroblasts: +

+ + +fibroblasts: + +

+ + + + +fibroblasts: + + +

+ + +

+ + + + + +lipid/degeneration

+ ±±±±lipid/macrophages: ±±±±giant cell formation

+ +fibroblasts: ±±±±

+ + + + +

complete sizereductionlipid/small giant cells

+

+giant cell formation

±±±±

+ ±±±±

+ + + + +

some size reduction(especially the outercollagen bundles)

* : infiltration of macrophages and granulocytes# : size of accumulated macrophages or ingrown (giant cell) layern.r : not retrieved± = hardly; + + + + + + = many infiltrated cells or /thick ingrown layer

At day 5 after implantation ongoing giant cell and capsule formations usually occur. With BD90and BD45HAc giant cell formation had started. The giant cell layer and capsule of BD45 (figure3a) were thicker than with BD90 and BD45EN. BD90 at this time point showed ingrowth of a fewmacrophages and fibroblasts along with a fibrin network. Remnants of the previous high cellularinfiltration in BD45HAc were recognized from only one small site with a concentration ofmacrophages with high fatty degeneration (figure 3b). The capsule was thick and active, and stillshowed granulocyte migration. Furthermore, it contained a lot of fibrin and collagen.At day 10 after implantation, ingrown layers with primarily giant cells looked similar with eachmaterial (e.g. shown with BD45EN, figure 4a). BD90, which is still largely empty, showed someingrowth of fibroblasts and concomitant collagen formation. With BD45HAc a dispersed fibroblastingrowth was observed. Its fibrous capsule contained many blood vessels.At week 3 after implantation, the size of the ingrown layer had proceeded with each material, butleast so with BD90. The latter showed the least signs of phagocytosis, but fibroblasts and ratcollagen formation were most obvious. Each implant was surrounded by a small fibrous capsule.With BD45, a thick edge with mainly giant cells and fibrous tissue (fibroblasts/collagen) inbetween, was present inside the material. Confirming the macroscopic observation, the BD45-dischad clearly been reduced in size. Larger giant cells, clearly with enclosed DSC parts, werelocalized in deeper areas, while the smaller giant cells (figure 4b) at the interface contained many


- 119 -

lipid droplets. The central area was still not infiltrated apart from disperse fibroblasts andmacrophages. Similarly to BD45, the size of BD45HAc was reduced. The giant cell layer wasrelatively large showing a clear reduction in the presence of DSC bundles. Althoughmacroscopically not observed, size-reduction also appeared to occur with BD45EN, because againa thick layer of giant cells with clear size-reduction of collagen bundles and an empty central partwas present.

Figure 4.a) Made at 100x. BDEN45 at day 10 with the giant (G) cell layer at the interface collagen-bundles insidegiant cells were clearly smaller. Both intra- and extra-cellular fat was present. Otherwise, the materiallooked rather empty. It was surrounded by a small fibroblast-oriented capsule (C).b) Made at 200x. BD45 at week 3 had been size-reduced. Larger giant (G) cells enclosing dermal sheepcollagen-parts were observed deeper, while smaller (S) giant cells with lipid droplets were present next tothe fibrous capsule (C). Blood vessels (V) and capillaries had grown inside.c) Made at 200x. BD90 at week 6. DSC bundles were always surrounded by a giant cell, but the lighter-stained areas represented clear fibroblast (F) ingrowth and new (N) collagen formation.d) Made at 400x. BD45HAc at week 6 showed, similarly to BD45 at week 3, size-reduction, and largergiant (G) cells in deeper areas and smaller (S) giant cells at the fibrous capsule. The smaller cellsespecially were filled with lipid droplets and dermal sheep collagen remnants.

At week 6 after implantation, the BD45 samples could not be retrieved and only a concentration ofrat collagen, blood vessels, some mast cells, and external fat was found. With BD90 (figure 4c) thegiant cell layer and fibrous tissue had now clearly grown, although the material did not show manysigns of degradation. The disc still seemed to have its original size with an empty central area. Thesize of BD45HAc was further reduced. It was completely occupied by giant cell formations

Chapter 7

- 120 -

alternating with thin rat collagen bundles. At one site of the outer edge, smaller giant cells andprobably even macrophages, contained lipid droplets (figure 4d). BD45EN, which was thinlyencapsulated, was completely occupied with giant cells, especially at the outer edges.Consequently, the size of the collagen bundles was clearly reduced.With Von Kossa-staining all samples proved to be negative, indicating that no calcification hadoccurred.

DISCUSSION

Dermal sheep collagen (DSC) cross-linked with 1,4-butanediol diglycidyl ether (BDDGE) underdifferent conditions was characterized and the interaction with tissue was evaluated aftersubcutaneous implantation in rats. Differences prior to implantation concerned the cross-linkingchemistry and with it the shrinkage temperature, number of amine groups, in-vitro stability andmacroscopic characteristics [25, 28].DSC cross-linked with 1,4-butanediol diglycidyl ether at pH 9.0 (BD90) displayed a Ts of 73 °Cand an amine group content of 14 (n/1000), which implies that cross-linking occurred via theamine groups of (hydroxy)lysine residues. BD90 was less pliable than the original material and itfelt rigid. It seems that crimp had occurred, as judged from the network morphology and reducedvariance in thickness of the bundles. This agrees with the reported planar shrinkage of pericardialtissue after cross-linking with glutaraldehyde [29, 30] and poly epoxy compounds [31]. BDDGEcross-linking at pH 4.5 (BD45) of collagen involved the carboxylic acid residues of aspartic andglutamic acid and consequently the amine group content was hardy reduced, while the Ts waselevated to 69 °C. BD45 had a soft and pliable structure also recognized by its microscopicmorphology. Additional acylation with acetic acid N-hydroxysuccinimide ester (BD45HAc)resulted in a large reduction in amine groups to 15 (n/1000) and in a slightly lowered Ts. Thecharacteristics of BD45HAc, possibly showing a little crimp, due to some partial denaturation ofthe collagen helices [25], leaving collagen bundles nearer to each other, agree most with thosedescribed previously for hexamethylenediisocyanate cross-linked dermal sheep collagen [32]. Theadditional reaction step with EDC and NHS performed on BD45 to introduce extra crosslinks(BD45EN) resulted in the highest Ts of 77 °C of all materials whereas the content of amine groupswas in-between that of the other materials. BD45EN had a soft and pliable structure, and themicroscopic morphology showed a collection of larger collagen bundles with spaces inside andstaining variances.With each of the materials, ethylene oxide sterilization decreased the content of amine groups. Theepoxide groups react with the amine groups under formation of N-2-hydroxy ethyl groups.However, the reduction in amine groups was less dramatic than observed previously [27]. Thislower degree of masking of amine groups is rather unclear. A reduction in Ts was obtained aswell, especially for N-DSC. The decrease in thermal stability can be ascribed to partialdestabilization of the triple-helix conformation [33, 34].Evaluation of the implants showed that BD45, BD90 and BD45EN were biocompatible in thesense of non-cytotoxicity. On the other hand, BD45HAc evoked a high infiltration of granulocytesand macrophages, and a high cell degeneration was observed at day 2 after implantation. These


- 121 -

results agree with the in-vitro cytotoxicity tests [35] in which BD45HAc showed the highest cellgrowth inhibition (non-published results), resulting from early cytotoxic effects. Furthermore, areasonable cell growth and morphology were observed after 1 week, which agrees with thepresent in vivo results that showed that the early cytotoxic reaction subsided already at day 5,leaving no real traces later on. It is anticipated that an improved rinsing method will reduce theobserved cytotoxicity.Differences in biodegradation were found. At week 6, BD45 was not retrieved, BD45HAc wasremarkably reduced in size, and BD45EN still had the original size, but showed clear reduction ofthe outer collagen bundles. On the other hand, BD90 showed hardly any signs of degradation. Thispartly agreed with the previous in vitro enzymatic degradation results, which had shownconsiderable rates of degradation with BD45 and BD45HAc and a high stability of BD90 andBD45EN [25]. Cross-linking at pH 4.5 resulted in the formation of ester bonds [24] which areeasier to cleave by enzymes, compared to the amide bonds created by EDC/NHS or the secondaryamines introduced by BDDGE cross-linking at pH 9.0. The slow rate of degradation of BD90 isfurther explained by a delay of at least one day in macrophage accumulation (compare e.g. figure2a with 1a), while also giant cell formation and ingrowth were slow and not at all completed atweek 6. BD45 showed the fastest rate of degradation, which provoked the earliest onset of giantcell formation, starting from day 1 on with accumulations of macrophages at the outer edges. Thereason why BD45 attracted more macrophages and giant cells, is not completely understood. Itmight be that the release of vic-diol functionalized butanediol derivates due to hydrolysis of theester-containing cross-links might lead to an enhanced attraction of macrophages. Moreover, thismaterial has most likely the highest hydrophilicity. Cell interactions at week 3 with BD45 can beinterpreted by either assuming that the small lipid-filled macrophages and giant cells near thefibrous capsule are fresh cells phagocytosing degenerating giant cells. Moreover it can be assumedthat they have differentiated from these large cells, such as now observed in deeper areas enclosingdermal sheep collagen parts. The fate of the small giant cells is unknown; it may be that theymigrate from the interface with a possible drainage towards lymph nodes or degenerate andbecome phagocytosed by other cells. The presence of much intracellular lipid may be a normalphenomenon in the degradation process during progressing digestion, or a cytotoxic effect, as alsoconsidered previously with acyl azide-cross-linked dermal sheep collagen [19]. If it has a cytotoxicorigin, then it would be more expected with BD45HAc, in view of its early high cellular infiltrationand the similar small lipid-filled giant cells at week 6.BD45HAc showed a delay in degradation of less than 3 weeks when compared to BD45, sincesize and morphologies at week 6 looked similar to those of BD45 at week 3. These resultsrepresent a not understood contradiction to the in-vitro experiments which revealed a decrease instability towards enzymatic breakdown of BD45HAc [25]. Moreover, the observed high cellularinfiltration with high enzyme release early after implantation was expected to play an additionalrole in degradation. It appears that ethylene oxide sterilization had affected the stability of BD45and BD45HAc. Masking reduced the stability towards enzymatic degradation of the materials[25]. Because the hydrophilicity of acylated amine groups is lower than of the N-2-hydroxyethylfunctionalized amine groups, the effect of hydrophilicity of the material might be important as well.

None of the materials was found to calcify. Previously, using the same animal model, dermal sheep

Chapter 7

- 122 -

collagen cross-linked with glutaraldehyde had resulted in intense calcification [16, 19]. Cross-linking with the carbodiimide (EDC) had resulted in moderate calcification at week 6 [19].However, these experiments, as well as the present study, were performed with adult animals, thusnot with the more sensitive model of the weanling rat. The use of weanling rats as an animal modelmay be more representative for the higher rate of calcification of heart valve bioprostheses inchildren or adolescents than in adults [36].DSC cross-linked via the BD45, BD90 and BD45EN procedures resulted in biocompatiblematerials. Furthermore, BD90 was the only material which showed a good matrix function, i.e.new collagen formation. The use of an alternative cross-linking agent for the stabilization of heartvalve bioprostheses should result in a material which is non-cytotoxic, shows no biodegradation,and which had a minimized tendency to calcificy. For the application as soft tissue replacingmaterial, the ability of tissue reconstruction, i.e. of collagen new formation, is a prerequisite. DSCcross-linked with BDDGE at pH 9.0 (BD90) combines these characteristics in an optimal way, andis therefore the cross-linking method of choice.

CONCLUSIONS

Cross-linking of dermal sheep collagen with 1,4-butanediol diglycidyl ether resulted inbiocompatible materials in terms of non-cytotoxicity and non-antigenicty. Biodegradation wasdependent on the cross-linking method applied. Materials with the lowest shrinkage temperatures(BD45 and BD45HAc), displayed the highest rates of degradation and after 6 weeks no or hardlyany material was retrieved. The material cross-linked under basic conditions (BD90) did not showdegradation and exhibited a good matrix function for the formation of new collagen. Because noneof the materials was found to calcify, these epoxy-based cross-linking methods seem to bepromising for the stabilization of porcine aortic heart valves.

References

1. A. Arem, "Collagen modifications", Clin. Plast. Surg., 12(2) pp. 209-220 (1985).2. E.E. Sabelman, "Biology, biotechnology, and biocompatibility of collagen", in "Biocompatibility of tissue

analogs", Ed. by D.F. Williams, CRC Press Inc, Boca Raton (1985)3. M. Chvapil, "Considerations on manufacturing principles of a synthetic burn dressing: A review", J.

Biomed. Mat. Res., 16 pp. 245-263 (1981).4. I.V. Yannas, "Biologically active analogous of the extracellular matrix: Artificial skin and nerves", Angew.

Chem. Int. Ed. Engl., 29 pp. 20-35 (1990).5. Y. Noishiki, T. Miyata, and K. Kodeira, "Development of a small caliber vascular graft by a new

crosslinking method incorporating slow heparin release collagen and natural tissue compliance", Trans.Am. Soc. Artif. Intern. Org., 32 pp. 114-119 (1986).

6. M.G. Dunn, P.N. Avasarala, and J.P. Zawadsky, "Optimization of extruded collagen fibers for ACLreconstruction", J. Biomed. Mat. Res., 27 pp. 1545-1552 (1993).

7. M. Chvapil, D.P. Speer, H. Holubec, T.A. Chvapil, and D.H. King, "Collagen fibres as a temporaryscaffold for replacement of ACL in goats", J. Bimed. Mat. Res., 27 pp. 313-325 (1993).

8. S.L. Hilbert, V.J. Ferrans, and M. Jones, "Tissue-derived biomaterials and their use in cardiovascularprosthetic devices", Med. Prog. through Techn., 14 pp. 115-163 (1988).

9. J.M. Lee, D.W. Courtman, and D.B. Boughner, "The glutaraldehyde-stabilized porcine aortic valvexenograft. I. Tensile viscoelastic properties of the fresh leaflet material", J. Biomed. Mat. Res., 18 pp. 61-77 (1984).


- 123 -


11. J.M. Lee, D.R. Boughner, and D.W. Courtman, "The glutaraldehyde-stabilized porcine aortic valvexenograft. II. Effect of fixation with or without pressure on the tensile viscoelastic properties of leafletmaterials", J. Biomed. Mat. Res., 18 pp. 79-88 (1984).


13. A. Cooke, R.F. Oliver, and M. Edward, "an in vitro cytotoxicity study of aldehyde treated pig dermalcollagen", Br. J. exp. Pathol., 64 pp. 172-176 (1983).




17. L.H.H. Olde Damink, P.J. Dijkstra, M.J.A. v. Luyn, P.B. v. Wachem, "Crosslinking of dermal sheepcollagen using hexamethylene diisocyanate", J. Mat. Sci.: Mat in Med., 6(7) pp. 429-434 (1995).

18. L.H.H. Olde Damink, P.J. Dijstra, M.J.A. v. Luyn, P.B. v. Wachem, P. Nieuwenhuis, and J. Feijen, "In-vitro degradation of dermal sheep collagen cross-linked using a water-soluble carbodiimide", Biomaterials,17(8) pp. 765-774 (1996).

19. P.B.v. Wachem, M.J.A.v. Luyn, L.H.H. Olde Damink, P.J. Dijkstra, J. Feijen, and P. Nieuwenhuis,"Biocompatibility and tissue regenerating capacity of crosslinked dermal sheep collagen", J. Biomed. Mat.Res., 28 pp. 353-363 (1994).

20. M.J.A. v. Luyn, P.B. v. Wachem, P.J. Dijkstra, L.H.H. Olde Damink, and J. Feijen, "Calcification ofsubcutaneously implanted collagens in relation to cytotoxicity, cellular interactions and crosslinking", J.Mat. Sci.:Mat. in Med., 6 pp. 288-296 (1995).

21. P.B. v. Wachem, M.J.A. v. Luyn, L.H.H. Olde Damink, P.J. Dijkstra, J. Feijen, and P. Nieuwenhuis,"Tissue regenerating capacity of carbodiimide-crosslinked dermal sheep collagen during repair of theabdominal wall", J. Artif. Organs, 17(4) pp. 230-239 (1994).

22. Y. Noishiki, H. Koyanagi, T. Miyata, and M. Furuse, Bioprosthetic valve, Patent EP 0 306 256 A2 1988.23. E. Imamura, O. Sawatani, H. Koyanagi, Y. Noishiki, and T. Miyata, "Epoxy compounds as a new

crosslinking agent for porcine aortic leaflets: subcutaneous implant studies in rats", J. Cardiac Surg., 4 pp.50-57 (1989).

24. R. Zeeman, P.J. Dijkstra, P.B. v. Wachem, M.J.A. v. Luyn, M. Hendriks, P.T. Cahalan, and J. Feijen,"Cross-linking and modification of dermal sheep collagen using 1,4-butanediol diglycidyl ether", Chapter3 of this theis and submitted to J. Biomed. Mat. Res. (1998).

25. R. Zeeman, P.J. Dijkstra, P.B. v. Wachem, M.J.A. v. Luyn, M. Hendriks, P.T. Cahalan, and J. Feijen,"Successive epoxy and carbodiimide cross-linking of dermal sheep collagen", Chapter 6 of this thesis.

26. T.M.v. Gulik and P.J. Klopper, "The processing of sheepskin for use as a dermal collagen graft - anexperimental study", Neth. J. Surg., 39(3) pp. 90-94 (1987).

27. L.H.H. Olde Damink, P.J. Dijkstra, M.J.A. v. Luyn, P.B. v. Wachem, P. Nieuwenhuis, and J. Feijen,"Influence of ethylene oxide gas treatment on the in vitro degradation behaviour of dermal sheepcollagen", J. Biomed. Mat. Res., 29(2) pp. 149-155 (1995).

28. R. Zeeman, P.J. Dijkstra, P.B. v. Wachem, M.J.A. v. Luyn, M. Hendriks, P.T. Cahalan, and J. Feijen, "In-vitro degradation of dermal sheep collagen cross-linked with 1,4-butanediol diglycidyl ether", Chapter 5of this thesis.

29. D. Chachra, P.F. Gratzer, C.A. Pereira, and J.M. Lee, "Effect of applied uniaxial stress on rate andmechanical effects of cross-linking in tissue-derived biomaterials", Biomaterials, 17(19) pp. 1865-1875(1996).

30. I. Vesely, "A mechanism for the decrease in stiffness of bioprosthetic heart valve tissues after cross-linking", ASAIO J., 42 pp. 993-999 (1996).


Chapter 7

- 124 -

32. M.J.A. v. Luyn, P.B. v. Wachem, R. Leta, E.H. Blaauw, and P. Nieuwenhuis, "Modulation of the tissuereaction to biomaterials Part I: Biocompatibilty of crosslinked dermal sheep collagens after macrophagedepletion", J. Mat. Sci.:Mat. in Med., 5 pp. 671-678 (1994).



35. M.J.A. v. Luyn, P.B. v. Wachem, M.F. Jonkman, and P. Nieuwenhuis, "Cytotoxicity testing of wounddressings using methylcellulose cell culture", Biomaterials, 13(5) pp. 267-275 (1992).


Cross-l nk ng and mod f cat on of porc ne aort c heart valves

- 125 -

Chapter 8

Cross-linking and modification of porcine aortic heart

valves




ABSTRACT

Porcine aortic heart valves which consist of three leaflets that are anchored in the aortic wall, were characterizedwith respect to tissue and amino acid composition. The structural proteins elastin and collagen constitute less than55 % of the dry weight of the aortic wall. Other components are glycosaminoglycans, proteoglycans and smoothmuscle cells. Aortic leaflets contain a high content of collagen and elastin (72 %). Amino acid analysis revealedthat the presence of (glyco)proteins and proteoglycans resulted in relatively high contents of arginine, serine,glutamic and aspartic acid, and threonine.Glutaraldehyde cross-linking of aortic leaflets was completed in 1 h and resulted in an increase in the shrinkagetemperature (Ts) from 61 to 85 °C. Cross-links were formed between the (hydroxy)lysine residues, but afterprolonged reaction times, predominantly one-sided (masking) reactions took place. Cross-linking with 1-ethyl-3-(dimethylaminopropyl)-carbodiimide hydrochloride (EDC) and N-hydroxysuccinimide (NHS) resulted after 4 h in amaterial with a high shrinkage temperature of 88 °C and a decrease in amine groups to 58 %. Reaction of aorticleaflets with 1,4-butanediol diglycidyl ether (BDDGE) could be achieved under acidic (pH 4.5) conditions or underbasic conditions (pH > 8.0), yielding cross-links between the carboxylic acid groups or the amine groups,respectively. Under optimized conditions, a Ts of 83 °C was obtained at pH 9.0 -10.0. The rate of cross-linking andthe cross-link efficacy were dependent on the solution pH and the BDDGE concentration. The EDC/NHS activationmethods enables one to introduce extended diamine cross-links. Reaction of leaflet tissue with 3,6,9-trioxa-1,13-tridecanediamine (TTDD) yielded materials with a high Ts of 89 °C without a reduction in amine groups.Masking reactions occur during cross-linking of collagen with a bifunctional reagent. Reaction of porcine leafletswith glycidyl isopropyl ether or acetic acid N-hydroxysuccinimide ester (HAc-NHS) resulted in a large decrease inTs from 61 to 50 °C, due to destabilization of the triple-helix.Cross-linking experiments of leaflets from which elastin or glycosaminoglycans (GAG) and proteoglycans wereremoved by elastase or guanidine hydrochloride, respectively, reveal that besides collagen also GAGs,proteoglycans and elastin can participate in the cross-linking reactions.

Chapter 8

- 126 -

INTRODUCTION

Bioprostheses such as porcine aortic heart valves, have been widely used for heart valvereplacement since the first clinical introduction in the 1960s [1]. When implanted, bioprostheticvalves evoke a local inflammatory response and are degraded by enzymes. In order to controlthese undesirable effects, xenograft valve prostheses, which are mainly constructed from porcineaortic heart valves or bovine pericardium [2], are cross-linked with glutaraldehyde (GA) [3].Despite its low incidence of thrombo-embolism and its satisfactory heamodynamic performance,problems in the long-term durability have been reported [4-6]. Clinical failure due to cuspalcalcification has been registered in many cases [7]. Calcium deposition causes valve stiffness,tearing, and rupture, and results in heamodynamic stenosis or insufficiency, or both [8].Unfortunately, the mechanism of calcification is still not well understood. The causes arenumerous such as the loss of proteoglycans and glycosaminoglycans, which are natural inhibitorsfor calcification [9, 10], during the preparation and fixation procedure. Furthermore fixation of thetissue changes the morphology of the tissue, the charges on the collagen molecules [4, 11, 12],introduces free, reactive aldehyde groups and may introduce internal stresses in the material [7].Furthermore, leakage of cytotoxic monomeric GA from the tissue is reported [5, 8]. Thesephenomena may contribute to the formation of calcific deposits, and several attempts have beenmade to prevent calcification of GA cross-linked bioprosthetic valves. Jorge-Herrero et al. [12]treated cross-linked aortic valves with different amino acids to block the residual aldehyde groups.Levy et al. [2] treated glutaraldehyde fixed aortic valves with trivalent metal ion solutions such asFeCl3 or AlCl3 to disturb the calcium phosphate crystal growth. Golomb et al. [13] studied theeffect of bisphosphonates, while Chanda et al. [14] removed antigenic substances from the matrix,such as cell remnants, phospholipids and proteins. Coupling of 2-amino oleic acid to residualaldehyde groups prevented leaflet but not aortic wall calcification [15, 16].Another approach in prevention of calcification is the use of alternative cross-linking reagents [17-19]. Fixation of xenograft valves with multifunctional epoxy compounds resulted in well-stabilizedmaterials which did hardly show calcification in rat subdermal implantation studies [17, 18]. Inprevious papers, several cross-linking methods for the stabilization of dermal sheep collagen havebeen described [20-23]. Collagen cross-linked with the water-soluble carbodiimide (EDC) in thepresence of N-hydroxysuccinimide (NHS) or with 1,4-butanediol diglycidyl ether (BBDGE)resulted in materials which showed a low immunogenic reaction and hardly any calcification aftersubcutaneous implantation in rats [24, 25].In our current research, these cross-link methods were utilized to stabilize porcine aortic heartvalve tissue. The heart valves were characterized for their contents of collagen, elastin andglycosaminoglycans, and the amino acid composition of the leaflets and the aortic wall wasevaluated. Finally, the influence of the selective removal of elastin or proteoglycans andglycosaminoglycans by elastase and guanidine hydrochloride, respectively, on the cross-linkingreaction was investigated.


- 127 -


Tissue preparationFresh porcine aortic heart valves were selected and dissected at the slaughterhouse (PremiumFleish Emsland, Lingen, Germany). Residual fat and myocardium were removed as much aspossible. After harvesting, the valves were rinsed with and stored in ice-cold saline (0.9 wt%sodium chloride, NPBI-Braun, Emmercompascuum, The Netherlands). After transport to thelaboratory, the valves were washed with a 0.01 M N-[2-hydroxyethyl]piperazine-N’-[2-ethanesulfonic acid] (HEPES free acid, Sigma-Chemical, St. Louis, USA) buffered solution (pH7.4) before an overnight storage in 0.01 M HEPES at 4 °C.

Cross-linkingThe aortic valves (approximately 1.5 g of dry weight) were treated using the following methods.

Glutaraldehyde (GA)Fresh valves were immersed in 100 ml of a 0.01 M HEPES buffered solution (pH 7.4) containing0.2 wt% glutaraldehyde (GA, 25 % aqueous solution z.S., Merck, Darmstadt, Germany). Cross-linking was carried out between 1 and 24 h at 20°C. After reaction, the valves were extensivelywashed with deionized water before lyophilization.1-ethyl-3-(dimethylaminopropyl)-carbodiimide hydrochloride (EDC) and N-hydroxysuccinimide(NHS)Fresh valves were immersed in 100 ml of a 0.1 M 2-[morpholino]ethanesulfonic acid (MES,Merck, Darmstadt, Germany) buffered solution (pH 5.5) containing 0.28, 0.56 or 1.15 g of EDC(Merck-Suchardt, Hohenbrunn, Germany) and 0.14, 0.28 or 0.68 g NHS (z.S. Merck-Suchardt,Hohenbrunn, Germany), respectively, for 24 h at 20 °C. After cross-linking, the valves werewashed with deionized water before lyophilization. Subsequently, the influence of the cross-linkingtime on the shrinkage temperature and the content of amine groups was determined. Fresh valveswere immersed in a 0.1 M MES buffered solution containing 1.15 g EDC and 0.14 g NHS for 1,2, 4 and 24 h. After reaction, the valves were washed with deionized water before lyophilization.1,4-Butanediol diglycidyl ether (BDDGE)Fresh valves were immersed in 100 ml of buffered solution containing 4 g BDDGE (technicalpurity >95% (GC), Fluka, Buchs, Switzerland) for 144 h at 20 °C. The following buffers werechosen: 0.1 M MES at pH 4.5, 0.025 M di-sodium tetraborate decahydrate (Na2B4O7•10H2O z.A. Merck, Darmstadt, Germany) at pH 8.5 and 9.0 and a carbonate buffer (0.064 M NaHCO3,0.036 M Na2CO3) at pH 10.0 and 10.5. After reaction, the valves were washed with deionizedwater before lyophilization. Subsequently, the kinetics of the cross-link reaction was determinedby immersing fresh valves in buffered solutions (pH 4.5, 9.0 and 10.0) containing 4.0 wt%BDDGE for 2, 4, 24, 48 and 144 h at 20 °C. The effect of the BDDGE concentration on thecross-linking rate was determined by immersing the valves in a buffered solution (pH 4.5 or 9.0)containing 0.5, 1.0, 2.0, 4.0 or 5.0 g BDDGE per 100 ml solution for 72 h (pH 9.0) or 144 h (pH4.5) at 20 °C. After cross-linking, the valves were washed before lyophilization.

4,7,10-trioxa-1,13-tridecanediamine (TTDD)

Chapter 8

- 128 -

a). With acylation of the amine groups prior to cross-linkingTo prevent cross-linking of collagen with the activation method with EDC/NHS, the amine groupsof (hydroxy)lysine residues present in the tissue were acylated prior to cross-linking. Fresh valveswere immersed in 100 ml of a 0.1 M MES buffered (pH 6.9) solution which contained 1.35 g ofacetic acid N-hydroxysuccinimide ester (HAc-NHS, ICN Biomedicals Inc., Aurora, USA). After16 h of reaction at 20 °C, the valves were rinsed three times with deionized water. The acylatedvalves were immersed in 100 ml of a buffered 0.1 M MES (pH 5.5) solution containing 1.15 gEDC, 0.28 g NHS and 0.53 g 4,7,10-trioxa-1,13-tridecanediamine (TTDD, Fluka, Buchs,Switzerland). After 24 h of cross-linking at 20 °C, the valves were extensively washed withdeionized water before lyophilization.

b). Without acylation of the amine groups prior to cross-linkingThe aortic valves were immersed in 100 ml of a buffered (0.1 M MES, pH 5.5) solution containing1.15 g EDC, 0.28 g NHS and 0.53 g TTDD. After 24 h of cross-linking at 20 °C, the valves wereextensively washed with deionized water before lyophilization.

Modification

HAc-NHSFresh valves were immersed in 100 ml of a 0.1 M MES buffered solution (pH 6.9) containing 1.35g HAc-NHS for 1 or 24 h at 20 °C. During the 24 h-experiment an additional amount of 0.5 gHAc-NHS was added after 5 h. The valves were washed with deionized water beforelyophilization.Glycidyl isopropyl ether (PGE)Fresh valves were immersed in 100 ml of a 0.1 M MES buffered solution (pH 4.5) containing 4 gPGE (Fluka, Buchs, Switzerland) or in 100 ml of a carbonate (0.064 M NaHCO3/0.036 MNa2CO3) buffered solution (pH 10.0) containing 1 or 4 g of PGE. The reaction was allowed toproceed for 48 h at 20 °C. The tissue was washed with deionized water followed by lyophilization.

Characterization

Collagen contentFresh aortic valves were lyophilized, and a small piece of tissue (2-4 mg) was hydrolyzed in 1.0 ml6 M hydrochloric acid at 110 °C for 20 h. After hydrolysis, 1.0 ml of 6 M sodium hydroxide wascarefully added to neutralize the solution to pH ~ 7.0. The solution was diluted by addition of 7.5ml citric acid/acetate buffer (14.25 g sodium acetate trihydrate, 9.38 g of trisodium citratedihydrate and 1.38 g of citric acid monohydrate (all Merck, Darmstadt, Germany) in 100 ml iso-propanol (p.A), diluted to 250 ml with deionized water, pH 6.0).Exactly 1.0 ml of the buffered hydrolyzate was reacted with 1.0 ml 0.5 wt% chloramine T solutionin citric acid/acetate buffer for 15 min at 20 °C. Thereafter, 2.0 ml of Ehrlich’s Reagent (7.0 g ofdimethylamino benzaldehyde in 12.5 ml 70-72 % perchloric acid, diluted to 100 ml with iso-propanol (p.A. and all Merck, Darmstadt, Germany)) was added and reacted for 20 min at 65 °C.The samples were cooled to 20 °C and the absorbances were measured with an UVIKON 930


- 129 -

spectrophotometer (Kontron Instruments, Switzerland) at 555 nm. A blank was prepared bytreating 1.0 ml citric acid/acetate buffer which contained no tissue sample as described above.The amount of hydroxyproline in the hydrolyzate originates from both collagen and elastin.Because the content of elastin in the leaflet and the wall is determined using elastase (see below),the amount of hydroxyproline from collagen can be calculated and was related to hydrolyzedcontrol samples of N-DSC, which consists of 100 % collagen type I.

Elastin contentThe elastin content of lyophilized aortic valves was determined by immersion of the materials in anelastase solution. An elastase (Type I, from porcine pancreas, Sigma, St. Louis, activity 100 U/mg,1 U will solubilize 1 mg of elastin in 20 min at pH 8.8 at 37 °C) stock solution (activity 15 U/ml)was prepared by dissolving the desired amount of elastase in 0.1 M Tris-HCl (Sigma-Chemical, St.Louis, USA) buffer (pH 8.8) [26] containing 0.005 M CaCl2 and 0.05 mg/ml NaN3. This stocksolution was allowed to stand for 1 h at 37 °C before use. To a tissue sample weighing 4 to 6 mg,1.0 ml of elastase solution in Tris-HCl buffer (37 °C) was added, and degradation was carried outfor 24 h at 37 °C. The reaction was terminated by addition of 0.1 ml of 0.25 M EDTA (TitriplexIII, Merck, Darmstadt, Germany). As control materials, similar weights of N-DSC and elastin(from bovine neck ligament, Sigma, St. Louis, USA) were taken and the same procedure asdescribed above was used. Because elastin contains a small amount of hydroxyproline (1-2 %),this assay was applied to calculate the amount of elastin present in the tissue. An aliquot of thesupernatant of 0.5 ml was hydrolyzed with 0.5 ml 6 M HCl at 110 °C for 20 h. After hydrolysis,0.5 ml 6 M NaOH was added to neutralize the solution, followed by addition of 3.5 ml citricacid/acetate buffer (pH 6.0). The amount of degraded collagen was determined by performing ahydroxyproline assay on 1.0 ml of this buffered solution as described above.

Glycosaminoglycan (GAG) contentThe amount of GAGs present in the tissue was determined using a slightly modified method ofBlix [27]. A piece of tissue (2 - 4 mg) was hydrolyzed in 1.0 ml 2 M HCl at 105 °C for 20 h. Afterhydrolysis, 1.0 ml 2.0 M NaOH was carefully added to elevate the pH to ~ 7.0, followed byaddition of 3.0 ml 1.25 M Na2CO3. The solution was shaken and 2.0 ml of this solution wastransferred into a glass-tube. Subsequently, 2.0 ml of an acetylacetone solution (3 v/v % indeionized water) was added and the mixture was reacted at 96 °C for 20 min. After reaction, themixtures were cooled and 4.0 ml ethanol was poured in the tube. Detection of the freehexosamines in the solution was achieved by addition of 2.0 ml Ehrlich's Reagent (1.34 gdimethylamino benzaldehyde in 25 ml 6 M HCl and 25 ml ethanol). The reaction was allowed toproceed for 45 min at 20 °C. The absorbances of the solutions were measured at 540 nm. Areference was prepared by following the same procedure as described above but without additionof tissue. The amounts of hexosamines were related to a calibration curve of chondroitin-4-sulfate(chondroitin sulfate A, Sigma-Chemical, St. Louis, USA) ranging between 0 and 5.0 mg/ml.

Amino acid analysis

Chapter 8

- 130 -

The amino acid composition of tissue samples was analyzed by reversed-phase high performanceliquid chromatography (HPLC) as described earlier [20].

Shrinkage temperatureThermal analysis of (non) cross-linked leaflets was carried out with a Perkin-Elmer DSC7Differential Scanning Calorimeter which was calibrated with Indium and Gallium. A piece ofleaflet tissue, 2-4 mg, was put in a volatile sample pan (Perkin Elmer, stainless steel) and 50 µl ofphosphate buffered saline (PBS, pH 7.4, NPBI, Emmercompascuum, Netherlands) was added.The reference pan contained only 50 µl PBS. A heating rate of 2°C/min was applied and atemperature range between 30 and 95 °C was chosen. The peak temperature was recorded and setas the shrinkage temperature (Ts) of the tissue.

Amine groupsThe content of amine groups was carried out using a slightly modified procedure as describedbefore [20]. A weighed amount of lyophilized tissue (2 - 6 mg) was incubated in 1.0 ml NaHCO3

(4 wt%) solution. After 30 min, 1.0 ml of a 0.5 wt% 2,4,6-trinitrobenzene sulfonic acid (TNBS 1M in water, Biochemica, Fluka, Buchs, Switzerland) solution was added and the reaction wascarried out at 40 °C for 3 h. After derivatization, the solutions were decanted and the tissuesamples were washed thoroughly with deionized water to remove the residual TNBS. Then thesample was immersed in 1.0 ml 6 M HCl and hydrolysis was carried out at 110 °C for 20 h. Thehydrolyzed samples were diluted with 9.0 ml deionized water, followed by measurement of theabsorbance at 345 nm. The content of amine groups present after cross-linking was expressed aspercentage of the initial amine content (%).

Degree of denaturation of the collagenA leaflet sample weighing 4 to 6 mg was immersed in 1.0 ml elastase solution (15 U/ml) in Tris-HCl buffer (pH 8.8) as described above. Degradation was carried out for 24 h at 37 °C. Thereaction was terminated by addition of 0.1 ml of 0.25 M EDTA. An aliquot of the supernatant of0.5 ml was hydrolyzed with 0.5 ml 6 M HCl at 110 °C for 20 h. After hydrolysis, 0.5 ml 6 MNaOH was added to neutralize the solution, followed by addition of 3.5 ml citric acid/acetatebuffer (pH 6.0). The amount of degraded collagen was determined by performing a hydroxyprolineassay as described above on 1.0 ml of this buffered solution. The amount of hydroxyproline whichdescends from collagen was calculated by subtracting the amount of hydroxyproline whichoriginates from elastin from the total concentration in the hydrolyzate. The degree of denaturationwas defined as the amount of collagen degraded by elastase divided by the initial content ofcollagen present in the leaflet.

The influence of leaflet tissue components on the cross-linking reactionIn order to determine if besides collagen, elastin and glycosaminoglycans are involved in the cross-linking reactions of aortic leaflets the following experiments were performed.Leaflets were exposed to an elastase solution, to remove the elastin. Hereto, lyophilized leaflets(40 mg) were immersed in 7.5 ml elastase solution (15 U/ml, 0.1 M Tris-HCl, 0.005 M CaCl2, pH8.8) at 37 °C for 24 h. The leaflets were extensively washed with deionized water before


- 131 -

lyophilization. Parallel to this, leaflets were exposed to 10 ml of a 4 M guanidine hydrochloridesolution (Merck, Darmstadt, Germany), containing 2 wt% Triton-X-100 (Sigma-Chemical, St.Louis, USA), 0.1 M sodium acetate and a protease inhibitor cocktail (pH 5.8) to removeglycosaminoglycans and proteoglycans [28]. The solution was allowed to stand at 4 °C for 24 h,followed by washing with deionized water and lyophilization.Cross-linkingNon-treated or pre-treated leaflets were subjected to either EDC/NHS or GA cross-linking orwere kept as control. Cross-linking was performed by immersing a weighed amount of tissue in abuffered solution (2 ml/mg tissue) containing either 0.2 wt% GA solution in 0.01 M HEPES buffer(pH 7.4) or 0.0115 g EDC/ml and 0.0014 g NHS/ml in 0.1 M MES buffer (pH 5.5) at roomtemperature. Cross-linking was terminated after 24 h by extensive washing with deionized water.

RESULTS

CompositionPorcine aortic heart valves are complex structures which are built from several components suchas collagen, elastin, proteoglycans, glycosaminoglycans, and cells. In table I the dry weightcomposition of the leaflet and the aortic wall tissue is given.

Table ITissue composition of the leaflet and the aortic wall

Component Leaflet [%] 2. aortic wall [%] 2.

collagen 58 ± 2 19 ± 4elastin 13 ± 4 34 ± 4glycosaminoglycans (GAG) 14 ± 3 7 ± 2others 1 15 ± 1 40 ± 8

1. Others include lipids, proteoglycans, glycoproteins and cellular components (fibroblasts for leaflets andfibroblasts and smooth muscle cells for aortic wall tissue)2. Swelling in PBS at 20 °C: leaflet tissue 650 % and aortic wall 330 %

In general, leaflets are composed of a higher content of structural proteins (72 wt%) compared tothe aortic wall (53 wt%). Collagen is the major component of the leaflets, while this is only 20wt% of the aortic wall. Elastin is the major structural component of the aortic wall. Besides, 40 %of the aortic wall comprises smooth muscle cells, and proteoglycans. Under physiologicalconditions, the leaflets and the aortic wall contain about 83 wt% and 77 wt% of water,respectively.

Amino acid analysisThe amino acid composition of pig collagen, elastin, and porcine aortic leaflet and wall tissue ispresented in table II. There are significant differences in amino acid composition between pigcollagen, elastin, leaflet, and aortic wall material. The most abundant amino acids in leaflet tissueare glycine, proline, alanine, and glutamic acid. Compared to 100 % porcine collagen, the amountsof hydroxyproline, glycine, alanine and, proline are markedly reduced, while the amount of

Chapter 8

- 132 -

hydrophilic residues such as aspartic and glutamic acid, serine, and (hydroxy)lysine are increased.The aortic wall contains very high amounts of alanine, valine, and leucine compared to the leaflets.Furthermore, low concentrations of arginine, hydroxyproline, and aspartic acid were found.

Table II Amino acid composition of collagen [29], elastin [29], leaflet and aortic wall tissue from pig

Amino acid Pig collagen(average) [n/1000]

Pig elastin (average)[n/1000]

Leaflet[n/1000]

Aortic wall[n/1000]

Arginine 49 8 51 28Hydroxyproline 96 13 80 28

Serine 30 11 41 30Aspartic acid 47 4 63 40Glutamic acid 73 24 87 63

Threonine 18 11 24 25Glycine 346 244 262 257Alanine 98 171 100 164Proline 125 131 103 103

Methionine 6 0 9 7Valine 21 140 34 88

Phenylalanine 14 46 22 31Isoleucine 11 23 19 23Leucine 24 67 44 64Histidine 5 0 10 7

(Hydroxy)lysine 33 4 41 27Tyrosine 4 25 10 16Cysteine 0 0 0 0Des/Ides 0 4 ? ?

The amount of each amino acid is represented as the number of residues per 1000 amino acids [n/1000].

Cross-linking

glutaraldehydeGlutaraldehyde cross-linked porcine aortic leaflets were prepared to provide a reference material.The influence of the reaction time on both the shrinkage temperature and the amine grouppercentage was determined.


- 133 -

0 5 10 15 20 2560

65

70

75

80

85

90

Sh

rinka

ge te

mp

erat

ure

[o C]

Reaction time [h]

0

20

40

60

80

100

Am

ine grou

ps [%]

Figure 1. The shrinkage temperature (■) and the percentage of amine groups (�) afterglutaraldehyde cross-linking of porcine aortic leaflets (0.2 wt% GA, 0.01 M HEPES, 20 °C, 100ml, pH 7.4) as a function of the cross-linking time.

Glutaraldehyde cross-linking of leaflets is almost complete in 1 h (figure 1). The maximum valueof the shrinkage temperature is 85 °C. However, the amount of amine groups was still decreasingfrom 44 % after 1 h to 25 % after 24 h.

EDC/NHSCross-linking of collagen-based material can be achieved by utilizing the EDC/NHS activationmethod [21, 30]. Table III summarizes the results obtained after cross-linking of aortic leaflets byusing different ratios of EDC and NHS.

Table IIIInfluence of the molar amounts of EDC and NHS on the Ts and the amine group content of

porcine aortic leaflets (Conditions: 0.1 M MES, pH 5.5, 20 °C, 24 h)# EDC [g] NHS [g] EDC:NHS

[molar ratio]Ts [°C] Amine groups

[%]1 0.23 0.14 1:1 74.5 85.0 ± 3.02 0.46 0.14 2:1 80.2 75.5 ± 2.63 1.15 0.14 5:1 87.7 53.7 ± 1.94 1.15 0.28 5:2 84.3 66.8 ± 4.25 1.15 0.69 5:5 76.3 82.4 ± 5.86 0 0 -- 60.5 100

Application of an EDC/NHS ratio of 5:1 resulted in the highest cross-link density as reflected by aTs of 88 °C and a percentage of amine groups of 54 %. If the molar amount of EDC wasincreased, while the NHS concentration was kept constant, an increase in Ts and a decrease inamine groups were obtained, indicating a higher cross-link density. If the NHS concentration wasincreased, while the amount of EDC was kept constant, a lower cross-link density was obtained.

Chapter 8

- 134 -

Reaction timeThe kinetics of the cross-linking reaction of aortic leaflets using EDC/NHS at a ratio of 5:1 wasdetermined (figure 2).

0 5 10 15 20 2560

65

70

75

80

85

90

Shr

inka

ge te

mpe

ratu

re [o

C]

Reaction time [h]

0

20

40

60

80

100

Am

ine groups [%

]

Figure 2. The shrinkage temperature (■) and the percentage of amine groups (�) of aorticleaflets as a function of the reaction time using 1.15 g EDC and 0.14 g NHS (0.1 MES, pH 5.5,20 °C, 100 ml).

Cross-linking of leaflets is complete in 4 h as reflected in a maximum value of Ts of 86 °C.Furthermore, no reduction in amine groups was observed after approximately 5 h. Compared toglutaraldehyde cross-linking, EDC/NHS cross-linking is somewhat slower and results in a similarTs.

1,4-Butanediol diglycidyl ether (BDDGE)The bisepoxy compound 1,4-butanediol diglycidyl ether has proven to be an effective cross-linkerof collagen type I. It has been shown that the reaction rate and mechanisms are dependent on thesolution pH [20]. The influence of the pH on the Ts and the amine group content of porcine aorticleaflets was determined by using a 4 wt% BDDGE solution and a cross-link time of 144 h (figure3). Cross-linking of leaflets under basic conditions (pH > 8.0) evoked a reaction with the aminegroups of (hydroxy)lysine residues. Elevation of the pH from 8.5 to 10.5 resulted in a fasterdecrease in amine groups, but in a less efficient cross-link reaction (lower Ts). Cross-linking underacidic conditions (pH < 6.0) demonstrated that predominantly carboxylic acid groups wereinvolved in the cross-linking reactions.


- 135 -

4 5 6 7 8 9 10 1160

65

70

75

80

Shr

inka

ge te

mpe

ratu

re [o

C]

pH

0

20

40

60

80

100

Am

ine groups [%]

Figure 3. The shrinkage temperature (■) and the percentage of amine groups (�) as a functionof the solution pH during BDDGE cross-linking of porcine aortic leaflets. (4 wt% BDDGE, 20°C, 144 h, 100 ml)

The highest values of the Ts were obtained at a pH of 9.0 or 10.0. Cross-linking at pH 4.5 led to aTs of 75 °C without significantly altering the percentage of amine groups (91 %).

Reaction time and concentrationCross-linking of heart valve tissue with a 4 wt% BDDGE solution at pH 4.5 was slow comparedto glutaraldehyde and EDC/NHS cross-linking as described above. A kinetic study revealed thatcross-linking for 144 h did not result in a maximum value of the Ts, which indicates that thereaction was not complete. In separate experiments it was found that an increase in BDDGEconcentration accelerated the cross-link reaction and that the Ts of the leaflets could be furtherincreased to 77 °C when a 5 wt% solution was applied. The percentage of amine groups wasapproximately 91 % in all samples.The kinetics of the cross-linking reaction at pH 9.0 and 10.0 was examined. The reaction at pH10.0 was faster than de reaction at pH 9.0 as manifested by a fast initial increase in Ts andreduction in amine groups (figure 4).

Chapter 8

- 136 -

0 20 40 60 80 100 120 14060

65

70

75

80

85

90

Shr

inka

ge te

mpe

ratu

re [o

C]

Reaction time [h]

0

20

40

60

80

100

Am

ine gro

ups [%]

Figure 4. The shrinkage temperature (● pH 9.0 and ■ pH 10.0) and the amine group content(� pH 9.0 and � pH 10.0) as a function of the BDDGE (4 g/100 ml) cross-linking time duringfixation of aortic leaflets at pH 9.0 or 10.0 (0.025 M Na2B4O7

· 10 H2O or 0.064 M

NaHCO3/0.036 M Na2CO3, 100 ml solution, 20 °C)

The maximum value of Ts at pH 9.0 or 10.0 was 83 °C, followed by a slight decrease to 78-79 °C.The percentage of amine groups is still declining after 48 h, while the Ts is not increasinganymore. Therefore, a cross-linking time between 24 and 48 h proved to be sufficient.

Table IVShrinkage temperature (Ts) and percentage of amine groups of BDDGE cross-linked

porcine aortic leaflets as function of the BDDGE concentration (100 ml solution, 0.025 MNa2B4O7, 20 °C, 48 h, pH 9.0)

BDDGE concentration[wt%]

Ts [°C] Amine groups [%]

0 61.9 1000.5 70.4 86.22 78.2 78.64 81.4 57.25 81.7 54.5

A 4 wt% concentration of BDDGE at a pH of 9.0 appeared to be sufficient for optimal cross-linking (48 h) of the leaflets. A lower concentration used under the same cross-link conditionsresulted in materials with a lower Ts. Moreover, application of a higher BDDGE concentration didnot result in a higher cross-link density as judged from the Ts and the percentage of amine groupsafter reaction (table IV).

Diamine spacer cross-linkingThe EDC/NHS activation method cannot only be used to cross-link collagen-based tissue, but itcan also be applied in combination with a diamine or diacid spacer. Carboxylic acid groups of


- 137 -

either the collagen or the diacid spacer are activated by EDC/NHS followed by a reaction with anamine group of the diamine compound or of the collagen, respectively. Cross-linking of leafletswith 4,7,10-trioxa-1,13-tridecanediamine (TTDD) in the presence of EDC and NHS will introduceextended bisamine cross-links. In order to prevent the fast zero-length cross-linking and topromote the formation of bisamine cross-links, the amine groups of (hydroxy)lysine residues areacylated with acetic acid N-hydroxysuccinimide ester (HAc-NHS) prior to cross-linking. Table Vsummarizes the results obtained after both methods.

Table VShrinkage temperatures and amine group percentages of (non) cross-linked or HAc-NHS

modified leaflets.Method Ts [°C] Amine groups [%]fresh1) 63.0 100 ± 5.0HAc-NHS2) 53.0 22.9 ± 3.0EDC / NHS 3) 85.0 60.8 ± 3.0TTDD / EDC / NHS 4) 89.0 100 ± 2.01). HAc-NHS2). TTDD / EDC / NHS 5) 79.9 36.5 ± 2.5

1) Non treated leaflets2) Leaflet was only acylated with HAc-NHS (1.35 g HAc-NHS, pH 6.9, 16 h, 20 °C, 100 ml)3) Direct EDC/NHS cross-linking (1.15 g EDC and 0.28 g NHS, pH 5.5, 24 h, 20 °C, 100 ml)4) Direct TTDD cross-linking without acylation (1.15 g EDC, 0.28 g NHS and 0.53 g TTDD, pH 5.5, 24 h, 20 °C,100 ml)5) Acylated leaflet prior to cross-linking with TTDD in the presence of EDC and NHS.

Acylation of amine groups prior to cross-linking resulted in a sharp decrease in Ts from 63 to 53°C and a large reduction in amine groups to 23 %. Cross-linking of this acylated material usingTTDD in the presence of EDC and NHS gave a steep increase in Ts to 80 °C and a moderateincrease in amine groups to 37 %. Direct TTDD cross-linking led to a very high Ts (89 °C)without a reduction of the amine group content compared to fresh tissue.

MaskingMasking of reactive residues of the collagen helix resulted in a decrease of the Ts, which wasascribed to some denaturation of the triple-helix. The effect of the extent of masking of the aminegroups by acetic acid N-hydroxysuccinimide ester (HAc-NHS) or glycidyl isopropyl ether (PGE)on the Ts was studied. Furthermore, the degree of denaturation was determined by exposure ofthe modified leaflets to an elastase solution, which is able to degrade elastin and denaturedcollagen. The content of degraded collagen was determined and was correlated to the initialcollagen content of the leaflets.

Chapter 8

- 138 -

Table VI The Ts, the content of amine groups and the degree of collagen denaturation of modified

aortic heart valve tissueReagent Reaction

time [h]Amount ofreagents [g]

Ts [°C] NH2, [%] Degree ofdenaturation [%]

Fresh -- -- 61.3 100.0 ± 3.0 0HAc-NHS 1 1.35 51.8 16.5 ± 2.0 54.6HAc-NHS 24 1.35 + 0.5 1) 49.9 7.7 ± 3.5 71.8PGE pH 10 48 1.0 59.0 77.0 ± 1.5 6.0PGE pH 10 48 4.0 55.0 36.0 ± 1.5 34.6PGE pH 4.5 48 4.0 64.3 105.8 ± 5.6 01) An aortic valve was immersed in 100 ml 0.1 M MES buffered solution (pH 6.9) containing 1.35 g HAc-NHS.After 5 h, an additional 0.5 g was added to the solution

Both HAc-NHS and PGE react with the amine groups and using a higher concentration or aprolonged reaction time resulted in a larger decrease in amine groups. Furthermore, the Ts wasdramatically reduced and a higher degree of denaturation was observed when more amine groupshad reacted. Treatment with PGE under acidic conditions, resulted in a material with a slightlyincreased Ts and hardly any change in amine groups. No destabilization and denaturation of thecollagen molecules was found. The influence of the degree of amine substitution and the degree ofcollagen helix disruption on the Ts is shown in figure 5.

0

20

40

60

80

100

50 52 54 56 58 60 62


Am

ine

grou

ps [%

]

50 52 54 56 58 60 620

20

40

60

80

100

Deg

ree

of d

enat

urat

ion

[%]


Figure 5. The percentage of amine groups after reaction with HAc-NHS or PGE (�) and thedegree of denaturation of the collagen molecules (�) as a function of the shrinkage temperature.

A decrease in amine group content of aortic leaflets resulted in a decline in Ts. Furthermore, alower Ts corresponds with a higher degree of denaturation of the collagen.

Influence of selective removal of tissue components on the cross-linking reactionBecause not only collagen but elastin, (glyco)proteins and glycosaminoglycans (GAG) containreactive side groups such as amine and carboxylic acid groups, these can participate in the cross-


- 139 -

link reaction and probably effect the cross-link efficacy. Tissue components were selectivelyremoved from the matrix by either an elastase or an guanidine hydrochloride (GdCl) treatment.These pre-treated leaflets were subsequently cross-linked with either EDC/NHS or GA. Table VIIrepresents the results obtained.

Table VIIInfluence of the removal of elastin or soluble proteins and GAGs from the leaflet matrix on

the shrinkage temperature, the amine group content and the swelling after cross-linkingwith GA (0.2 wt%, pH 7.4, 24 h) or EDC/NHS (ratio 5:1, pH 5.5, 24 h) or only after

treatment with elastase (15 U/ml, pH 8.8, 37 °C, 24 h) or GdCl (4 M, pH 5.8, 4 °C, 24 h).# Pre-treatment Cross-linking Ts [°C] Amine groups [%]1 None None 61.6 100.0 ± 8.02 GdCl None 58.4 98.3 ± 3.03 Elastase None 61.1 91.1 ± 3.04 None GA 85.5 32.6 ± 0.15 GdCl GA 81.8 26.8 ± 0.26 Elastase GA 83.0 35.9 ± 1.37 None EDC/NHS 87.7 55.3 ± 0.88 GdCl EDC/NHS 86.6 30.8 ± 0.39 Elastase EDC/NHS 89.1 34.5 ± 1.4

Treatment of fresh leaflets with elastase and GdCl slightly decreased the Ts and the amine groupcontent. Cross-linking using glutaraldehyde increased the Ts to 85 °C, while the amine groupcontent was reduced to 31-35 %. The effect of either elastase or GdCl treatment on the Ts and thepercentage of amine groups was hardly significant. Pre-treatment of the leaflets prior to EDC/NHScross-linking resulted in comparable values of Ts but in a significant lower content of aminegroups after cross-linking compared to non-pre-treated valves (31-35 vs. 55 %).

DISCUSSION

The long-term clinical use of glutaraldehyde cross-linked porcine aortic valves or bovinepericardial tissues is limited by calcific degeneration. Several strategies have been proven effectivefor mitigating calcification of cusps [4-9, 16, 31]. However, many of these strategies only preventcuspal and not aortic wall calcification, which is a potentially important problem with stentlessvalves. Although the exact mechanism of calcification is still unknown, the glutaraldehyde (GA)cross-linking process seems to play a role in the enhancement of calcium deposition. Furthermore,GA can lead to severe immunogenic and toxic reactions due to leakage from the fixed tissue [31].In previous studies it has been shown that collagen type I matrices have been cross-linked with abisepoxy compound [20] or a water-soluble carbodiimide [21]. These methods afforded materialswith a high stability towards enzymatic degradation. Furthermore, subcutaneous implantation ofthese matrices in rats showed no or hardly any calcification [24,25]. The present study focuses oncharacterization of heart valve tissue and the optimization of cross-linking methods other than thewidely applied glutaraldehyde fixation.

Chapter 8

- 140 -

The structural proteins elastin and collagen constitute less than 55 % of the dry weight of theaortic wall. The other components are smooth muscle cells, proteoglycans, and other proteins[32]. Elastin confers elasticity to the aortic wall while collagen acts to stiffen the wall and to limitits extensibility [33]. The content of elastin and collagen of pig aortic wall tissue are approximately34 % and 20 %, respectively and similar ratios have been reported in literature [34]. The mostabundant amino acids in the aortic wall are glycine, alanine, valine, proline, and leucine which iscontributed to the high extent of elastin in the tissue. Furthermore, relatively high contents ofhydrophilic amino acids such as (hydroxy)lysine, glutamic and aspartic acid, serine, and threonineresidues are detected.Leaflets contain a higher content of fibrous proteins and an amount of 75-80 % was found, whichis in agreement with literature [35]. The remaining part consists mainly of soluble proteins,glycosaminoglycans, and cellular components. The amino acid composition (Table II) of leaflettissue resembles the composition of pure pig collagen quite well, which confirms that collagen isthe predominant protein in the leaflet tissue. A high content of (hydroxy)lysine of 41 (n/1000) wasdetected which is in agreement with the value found for bovine pericardium [36]. The occurrenceof elastin in the leaflets resulted in an increase of the concentration of hydrophobic residues suchas valine, phenylalanine, leucine, and isoleucine as related to pure pig collagen. Moreover, thepresence of soluble proteins and proteoglycans, which contain many hydrophilic residues, resultedin high levels of amino acids like serine, (hydroxy)lysine, aspartic and glutamic acid, and arginine.Glutaraldehyde is the most common fixative used in cross-linking and stabilization of bioprosthetictissues [31, 37]. Usually, fixation of pericardium or aortic heart valves is carried out in 0.2 or0.625 wt% buffered GA solutions for 24 h or more. Aortic leaflets were cross-linked in a 0.2 wt%glutaraldehyde solution at pH 7.4 to prepare a reference material. The effect of the cross-linkingtime on the degree of cross-linking was determined in order to prepare a fully cross-linked materialwith a high shrinkage temperature (Ts). As shown in figure 1, a plateau value of Ts of 85 °C wasobtained in 1 h of reaction, which resembles the results found by others [22, 37, 38]. A decrease inamine groups was observed if cross-linking was carried out for prolonged times, while the Tsremained constant. This implies that amine groups react with glutaraldehyde without the formationof cross-links (masking reactions), which agrees with earlier results [22]. Because it is assumedthat these groups are involved in calcification, introduction of these groups has to be minimizedand short cross-linking times have to be applied. However, the aortic wall needs much longerreaction times in order to cross-link the tissue completely [5]. This was visualized by treating walltissue which was cross-linked with GA for 1 h, with a 2,4,6-trinitrobenzene sulfonic acid solution.This reagent reacts with the primary amine groups to provide a yellow colored product. The outersurface of the wall was slightly yellow colored, whereas the bulk of the tissue was highly colored,indicating that many amine groups were still present in the bulk of the wall. Diffusion of GA intothe dense wall is very slow and might be hampered by the fast cross-linking reaction at the surface[39]. Therefore, prolonged reaction times are needed to cross-link the aortic wall sufficiently.In the past decade several cross-linking techniques have been developed for stabilization ofcollagen-based materials. The activation method which uses 1-ethyl-3-(dimethylaminopropyl)-carbodiimide hydrochloride (EDC) and N-hydroxysuccinimide (NHS) has been successfullyapplied in cross-linking of dermal sheep collagen [21] and bovine pericardium [30]. EDC/NHScross-linking of porcine aortic leaflets afforded a cross-linked material with a maximum Ts of 88


- 141 -

°C. The cross-link density of the leaflets was highly dependent on the concentrations of EDC andNHS. An increase in EDC concentration relative to the NHS concentration resulted in a higher Ts.An increase of the ratio NHS/EDC revealed a decrease in the cross-link density which is caused bya less efficient reaction [21, 30].A kinetic study of the EDC/NHS cross-link reaction (figure 2), demonstrated that the reaction wascompleted in 4 h. The leaflets had a Ts of 86 °C and a percentage of amine groups of 58 %. Inearlier work it has been shown that a linear relationship is obtained between the Ts and the contentof amine groups after cross-linking within a specific method. In figure 6 this relation has beenpresented for EDC/NHS cross-linked leaflets and dermal sheep collagen.

55 60 65 70 75 80 85 90 9540

50

60

70

80

90

100

Am

ine

grou

ps [%

]


Figure 6. The amine groups as a function of the Ts of EDC/NHS cross-linked porcine aorticleaflets (■ and solid line). The dotted line (�) represents the results of EDC/NHS cross-linkingof dermal sheep collagen (reproduced from [21]).

In contrast to the results obtained with dermal sheep collagen, no linear relation was foundbetween the amine groups and the Ts. The reason of this non-linearity is the presence of othercomponents such as elastin and glycosaminoglycans (GAG) which can be involved in the cross-link reaction. For instance, if a GAG molecule is coupled to the collagen, no increase in Ts will beobtained whereas the percentage of amine groups decreases. On the other hand, reaction of twocarboxylic acid groups of a GAG molecule with two amine groups of adjacent collagen helicesresults in a cross-link.Epoxy compounds have been used as cross-linking agents for collagen-based materials [17, 18, 20,40]. Treatment of porcine aortic leaflets with an epoxy compound for 48 h at pH > 9.0 [18] orreacting bovine pericardium with an epoxy compound for 96 h at pH 10.0 resulted in a Ts of 79°C [40]. Generally, a more soft and pliable material was obtained compared to GA fixed valves.The use of 1,4-butanediol diglycidyl ether (BDDGE) as a cross-linking agent for stabilization ofcollagen was recently evaluated [20]. Cross-linking can not only be achieved via the amine groupsof (hydroxy)lysine residues but can also be established via the carboxylic acid groups of glutamicand aspartic acid. The influence of the solution pH on the Ts and the amine group content ofporcine aortic leaflets after immersion in a 4 wt% BDDGE solution for 144 h is shown in figure 3.

Chapter 8

- 142 -

Under basic conditions a higher pH resulted in a higher reaction rate and less amine groups aftercross-linking, which is caused by the higher amount of deprotonated amine groups. However, theTs reached a maximum value of 80 °C at a pH of 10.0. The reaction rate at pH 10.5 becomes toohigh and led to reaction of most of the amine groups in the early stage of the cross-linking process.The pendant epoxide groups may not find an adjacent amine group anymore to form a cross-link.This results in considerable amount of one-sided (masking) reactions which cause a destabilizationof the triple helix and consequently a decrease in the Ts. The destabilizing effect of maskingreactions on the thermal stability has also been observed in the reaction of collagen withmonofunctional epoxy compounds [41, 42].Another cross-linking mechanism was evoked if the reaction was carried out under acidicconditions, and mostly carboxylic acid groups were involved in the cross-linking. Hardly anyreduction in amine groups was obtained while the Ts reached a value of 75 °C. A kinetic study ofcross-linking at pH 4.5 revealed that the reaction between carboxylic acid groups and epoxidegroups of BDDGE is very slow [20], and after 144 h of reaction no maximum of the shrinkagetemperature was observed which implies that the reaction was still in progress. Cross-linking viathe amine groups occurs at a higher rate and the reaction was accelerated by elevating the pH from9.0 to 10.0 (Figure 4). A maximum value of the Ts was achieved within 48 h, which confirms thatthe cross-link reaction under basic conditions is faster than under acidic conditions, which is due tothe higher nucleophilicity of the amine groups compared to the carboxylate groups [43]. Both thecross-link reactions at pH 9.0 and 10.0 yield materials with a slightly lower Ts at prolonged times.The lower cross-link efficacy becomes also apparent when high concentrations of BDDGE areused. Although the Ts is slightly increased at higher cross-linker concentrations a considerabledecrease in amine groups has been observed (Table IV).

Cross-linking of the aortic leaflets using the EDC/NHS method afforded a more stiff material thanthe native material. To overcome this problem, the introduction of extended cross-links built fromshort hydrophilic polyether chain segments was investigated. The EDC/NHS activation method inthe presence of the diamine 4,7,10-trioxa-1,13-tridecanediamine (TTDD) was regarded as apotential methodology. To prevent the direct coupling between carboxylic acid and amine groupsof the collagen, the (hydroxy)lysine residues were acylated using acetic acid N-hydroxysuccinimideester (HAc-NHS) prior to cross-linking. The reaction pathways are illustrated in scheme 1.Carboxylic acid groups were activated with EDC/NHS, which subsequently react with either theamine groups of the collagen, thus yielding zero-length cross-links, or with the primary aminegroups of TTDD (pathway 1). Both reactions occurred as indicated by the higher Ts of TTDDcross-linked collagen (89 °C) as compared to the EDC/NHS cross-linked tissue which had a Ts of85 °C. Furthermore, the content of amine groups of TTDD cross-linked leaflets remainedunchanged, which implies that besides zero-length and TTDD cross-links, one-sided or maskingreactions with TTDD molecules had occurred.Acylation of amine residues resulted in a sharp decrease in thermal stability from 63 to 53 °C, andin a large reduction of amine groups to 23 %. Treatment of acylated leaflets with TTDD (pathway2) resulted in introduction of cross-links as indicated by an increase in Ts from 53 to 80 °C, whilethe percentage of amine groups was elevated to 37 %. This suggests that besides TTDD cross-links, one-side reactions occur. In comparison to the material cross-linked via pathway 1, the


- 143 -

material which was cross-linking via pathway 2 had a considerably lower Ts which suggests thatthe latter had a lower degree of cross-linking. On the other hand, mainly short hydrophilicpolyether cross-links are incorporated in the material cross-linked via pathway 2, whereas mainlyzero-length amide type cross-links were formed in the other material. The relation between the Tsand the cross-link density will be different in both cases, which means that the Ts is not anabsolute measure for the cross-link density. Independent studies on dermal sheep collagen revealedthat the material which was cross-linked via pathway 2 was considerably less stable towardsenzymatic degradation than the material cross-linked via pathway 1, which suggests that indeed alower cross-link density was observed.

COOH

COOH COOH

NH2

��

COOH

NH

C

CH3

O

COOH

COOH

H3C C

O

O N

O

O

TTDD = H2NO

OO

NH2

(HAc-NHS)

COOH

COOH

COOH

NH

C

CH3

O

C O

NH

R

NH

C O

C O

NH

R

NH2

C O

NHC O

NH

R

NH

C O

C O

NH

R

NH2

��

EDC + NHS

H2N R NH2

(TTDD)

pathway 1

EDC + NHS

��(TTDD)

H2N R NH2

pathway 2

Scheme 1. Expected mechanism of diamine (TTDD) cross-linking in the presence of EDC andNHS of (acylated) aortic leaflets

During cross-linking of collagen with bifunctional reagents such as glutaraldehyde, BDDGE orTTDD both cross-linking and masking of the material occurs. In literature hardly any attentionwas paid to the influence of the masking reactions on the thermal stability of the collagen material.In a previous study was found that reaction of collagen with glycidyl isopropyl ether displayed areduction in the Ts [20]. Other groups found similar results after reaction of collagen with amonofunctional reagent and they concluded that masking or branching reactions destabilized thecollagen triple-helical conformation and packing [41, 42, 44]. Porcine aortic leaflets were treatedwith either HAc-NHS or glycidyl isopropyl ether to modify the amine groups. The Ts decreased

Chapter 8

- 144 -

with increasing degree of masking and denaturation (figure 5). The reduction in Ts is explained bypartial denaturation of the triple helices, which led to the unraveling of the three α-chains. Thiswas confirmed by the higher susceptibility to elastase which is able to cleave peptide bonds only innon helical domains. It appears that the degree of denaturation is directly related to the Ts.

It is expected that glycosaminoglycans, which contain many carboxylic acid groups, and elastinwhich contains 5 amine and 21 carboxylic acid groups per 800 amino acids [29] can participate inthe cross-linking reactions. Therefore, these components were selectively removed prior to cross-linking by either guanidine hydrochloride (GdCl) or elastase, respectively. Treatment of freshleaflets with elastase or GdCl resulted in a slight decrease in Ts and in a slight reduction of amines(Table VII). The slight decrease in Ts especially after GdCl treatment can be a result of somedenaturation by the chaotropic solution, which is able to break the hydrogen bonds which stabilizethe triple-helix [28]. Cross-linking of the pre-treated leaflets using GA gave materials which hadabout the same cross-link densities as the untreated leaflets. It seems that removal of elastin orsoluble proteins and GAGs did not affect the cross-link reaction to a high extent and it can beconcluded that mainly collagen is involved during GA stabilization of aortic leaflets. On thecontrary, pretreatment of leaflet tissue had an effect on the EDC/NHS reaction. Whereas the Tsvalues (86-89 °C) were not significantly altered, cross-linking of extracted leaflets resulted in alower content of amines after reaction. Under normal conditions, EDC/NHS activation includesthe carboxylic acid groups of the collagen and glycosaminoglycans. In general, GAGs are closelyconnected to the collagen in the extracellular matrix, which suggest that during EDC/NHS cross-linking, GAG branches and cross-links can be easily formed aside from the zero-length cross-links.Extraction of GAGs and other components promotes the formation of zero-length cross-links asshown by the low content of amine groups after reaction (31 - 35 %). In contrast, non-extractedleaflets contained a considerably lower content of zero-length cross-links as reflected by thecontent of amine groups of 55 %. This implies that in non-pretreated tissue only few GAG cross-links were formed, and that they have a large impact on the Ts.It can be concluded that besides the collagen molecules also GAGs and elastin participate in thecross-link reactions.

CONCLUSIONS

The aortic wall is composed of high contents of elastin and smooth muscle cells, while the leafletscontain mainly collagen. Amino acid analysis revealed that the remaining proteins such asproteoglycans contain high contents of hydrophilic amino acids.Cross-linking of porcine aortic leaflets was accomplished with the same methods as applied ondermal sheep collagen. Comparable values of the shrinkage temperature were obtained as theglutaraldehyde cross-linked controls. Cross-linking in which amide type bonds were formed(EDC/NHS or diamine cross-linking) displayed the highest values of Ts and a moderate reductionof amine groups. Reaction of aortic leaflets with bifunctional agents which react with the aminegroups (glutaraldehyde and the bisepoxide at pH 9.0) resulted in materials which had a somewhatlower Ts as compared to EDC/NHS. Cross-linking via the carboxylic acid groups (diamine and


- 145 -

bisepoxide cross-linking at pH 4.5) revealed the lowest Ts without a change in amine groups.Masking of amine groups resulted in a reduction of Ts due to partial destabilization of the triplehelix. This was confirmed by the higher susceptibility to an elastase solution, which is only able tocleave peptide bonds of denatured domains.Cross-linking of leaflets which were selectively extracted with elastase or guanidine hydrochlorideto remove elastin or proteoglycans and glycosaminoglycans, demonstrated that not only collagenparticipated in the cross-link reactions.

References

1. S.H. Shen, H.W. Sung, R. Tu, C. Hata, D. Lin, Y. Noishiki, and R.C. Quijano, "Characterization ofpolyepoxy compound fixed porcine heart valve bioprostheses", J. Appl. Biomat., 5 pp. 159-162 (1994).




5. M.N. Girardot, M. Torrianni, D. Dillehay, and J.M. Girardot, "Role of glutaraldehyde in calcification ofprocine heart valves: comparing cusp and wall", J. Biomed. Mat. Res., 29 pp. 793-801 (1995).

6. D.W. Courtman, C.A. Pereira, S. Omar, S.E. Langdon, J.M. Lee, and G.J. Wilson, "Biomechanical andultrastructural comparison of cryopreservation and a novel cellular extraction of porcine aortic valveleaflets", J. Biomed. Mat. Res., 29 pp. 1507-1516 (1995).





11. J. Chanda, S.B. Rao, M. Mohanty, A.V. Lal, C.V. Muraleedharan, G.S. Bhuvaneshwar, and M.S.Valiathan, "Prevention of calcification of tissue valves", Art. Org., 18(10) pp. 752-757 (1994).

12. E. Jorge-Herrero, P. Fernandez, C. Escudero, J.M. Garcia-Páez, and J.L. Castillo-Olivarez, "Calcificationof pericardial tissue pretreated with different amino acids", Biomaterials, 17 pp. 571-575 (1996).

13. G. Golomb, A. Schlossman, H. Saadeh, M. Levi, J.M. v. Gelder, and E. Breuer, "Bisacylphosphonatesinhibit hydroxyapatite formation and dissolution in vitro and dystrophic calcification in vivo",Pharmaceut. Res., 9 pp. 143-148 (1992).

14. T. Chandy, M. Mohanty, A. John, S.B. Rao, R. Sivakumar, C.P. Sharma, and M.S. Valiathan, "Structuralstudies on bovine bioprosthetic tissues and their in vivo calcification: prevention via drug delivery",Biomaterials, 17 pp. 577-585 (1996).

15. J.P. Gott, "Calcification of porcine valves: A successful new method of antimineralization", Ann. Thorac.Surg., (1994).


17. H.W. Sung, et al., "A newly developed porcine heart valve bioprostheses fixed with an epoxy compound",ASAIO J., pp. 192-198 (1994).


Chapter 8

- 146 -







25. P.B. v. Wachem, R. Zeeman, P.J. Dijkstra, M. Hendriks, P.T. Cahalan, J. Feijen, and M.J.A. v. Luyn,"Characterization and biocompatibility of epoxy crosslinked dermal sheep collagen", Chapter 7 of thisthesis and submitted to J. Biomed. Mat. Res. (1998).

26. Boehringer Mannheim, "Biochemical Catalog", (1998).27. G. Blix, "The determination of hexosamines according to Elson and Morgan", Acta Chem. Scandin., 2 pp.

467-473 (1948).28. V.C. Hascall, M. Yanagashita, A. Calabro, R. Midura, J.A. Rada, S. Chakravanti, and J.R. Hassell,

"Isolation and characterization of proteoglycan core protein", in "Extracellular matrix. A practicalapproach", Ed. by M.A. Haralson and J.R. Hassell, Oxford University Press (1995)

29. J.E. Eastoe, "Composition of collagen and allied proteins", in "Treatise on Collagen. Volume 1. Chemistryof collagen", Ed. by G.N. Ramachandran, Academic Press, London. p. 1-72 (1967)



32. M. Sauvage, M.P. Jacob, and M. Osborne-Pellegrin, "Aortic elastin and collagen content and synthesis intwo strains of rats with different susceptibilities to rupture of the internal elastic lamina", J. Vasc. Res., 34pp. 126-136 (1997).

33. A. Bruel and H. Oxlund, "Changes in biomechanical properties, composition of collagen and elastin, andadvanced glycation endproducts of the rat in relation to age", Atherosclerosis, 127 pp. 155-165 (1996).

34. M.A. Cattell, J.C. Anderson, and P.S. Hasleton, "Age-related changes in amounts and concentrations ofcollagen and elastin in normotensive human thoracic aorta", Clin. Chim. Acta, 245 pp. 73-84 (1996).




38. H.W. Sung, J.S. Shih, and C.S. Hsu, "Crosslinking characteristics of porcine tendons: Effects of fixationwith glutaraldehyde or epoxy", J. Biomed. Mat. Res., 30 pp. 361-367 (1996).



41. H.-W. Sung, H.-L. Hsu, C.-C. Shih, and D.-S. Lin, "Cross-linking characteristics of biological tissue fixedwith monofunctional or multifunctional epoxy compounds", Biomaterials, 17(14) pp. 1405-1410 (1996).


- 147 -


43. S.S. Wong, "Reactive groups of proteins and their modifying agents", in "Chemistry of protein conjugationand cross-linking", Ed. by S.S. Wong, CRC Press, Inc., Boca Raton, Florida. p. 7-48 (1991)


ropert es of cross-l nked porc ne aort c heart valves

- 149 -

Chapter 9

Properties of cross-linked porcine aortic heart valves




ABSTRACT

Porcine aortic heart valves were cross-linked with (a) 1 ethyl-3-(3-dimethyl aminopropyl) carbodiimide and N-hydroxysuccinimide (EDC/NHS), (b) 4,7,10-trioxa-1,13-tridecanediamine (TTDD) in the presence of EDC/NHS,(c) chondroitin sulfate (CS) in the presence of EDC and NHS, (d and e)1,4-butanediol diglycidyl ether (BDDGE) atpH 4.5 or 9.0 (BD45 or BD90), or with (f) successive BDDGE (pH 4.5) and EDC/NHS (BD45EN). The propertiesof heart valves cross-linked via these methods were compared with those of glutaraldehyde cross-linked heartvalves (GA). Shrinkage temperatures (Ts) of the cross-linked leaflets (80 - 89 °C) were comparable to GA (84 °C),except for BD45 which had a considerably lower Ts of 74 °C. The amount of amine groups after cross-linking wasdependent on the method applied but was always higher as in GA cross-linked leaflets. Cross-linking in thepresence of EDC and NHS (a-c) resulted in materials with an increased transition enthalpy (∆Hs), whereas BDDGEand GA cross-linking did not change the ∆Hs of the material.Contrary to the leaflets of the valve, the dense and more hydrophobic wall was not fully cross-linked as reflected bya high content of amine groups and the presence of three transitions in the thermograms.In general, the swelling of the leaflets and the wall was reduced due to cross-linking. It was shown that the degreeof swelling was dependent on the degree of cross-linking, the nature of the cross-links and degree of ionizablegroups of the collagen. The in-vitro resistance towards enzymes (collagenase and pronase) of materials (a-d) and (f)was comparable to GA fixed materials. The resistance towards enzymatic degradation of cross-linked aortic wallsamples was lower than its leaflet counterparts.In order to get more insight in the cross-linking reactions, the (non)-cross-linked leaflets and walls were subjectedto three different 'extraction' methods. Removal of cell remnants and small matrix proteins was achieved with thedetergents 3-(3-cholamidopropyl) dimethylammonio)-1-propanesulfonate (CHAPS) and sodium dodecyl sulfate(SDS). All components except collagen were removed from the tissue matrix upon degradation with elastase.Removal of proteoglycans, glycosaminoglycans (GAGs) and soluble proteins from the (non)-cross-linked tissue wasaccomplished using guanidine hydrochloride (GdCl). These extraction studies revealed that besides collagen, alsoelastin, GAGs, proteoglycans and possibly small proteins are involved in the cross-linking reactions. Furthermore,it was shown that it is possible to remove components in a rather selective manner from the matrix. Because thepresence of foreign cells and proteins in the valve tissue appeared to be a key determinant in calcification of thesebioprosthetic valves, extraction treatments performed on cross-linked aortic heart valves may be advantageous toextend the life-span of tissue heart valves.

Chapter 9

- 150 -

INTRODUCTION

The successful use of stentless heterografts to replace the human aortic heart valve was reportedin literature [1, 2]. However, interest in stentless valves declined due to technical difficulties ofimplantation. Stented valves were introduced, but significant failure rates were observed due tothe rigidity of the stent [1]. David et al. [3] found that stentless valves were heamodynamicallysuperior to the stented bioprostheses and therefore stentless valves will be more applied in thefuture although the implant techniques are more complex [2]. Nowadays, bioprosthetic valvesconsist either of glutaraldehyde treated porcine aortic valves or bovine pericardial valves.However, the long-term durability of tissue heart valves is still limited. Calcification is the mostfrequent cause of xenograft valve dysfunction [4]. Therefore, efforts are being made to reducexenograft calcification and to optimize the biomechanical properties by applying different cross-linking techniques.Calcification involves both intrinsic valve components such as cell debris and membranes, lipids,collagen fibrils, elastin fibers and non-collageneous proteins and extrinsic components such asthrombi [5-8]. Furthermore, the cross-linking method and technique appear to play a crucial rolein enhancement of calcific depositions. Glutaraldehyde treatment of tissue may result in free,unreacted aldehyde groups, in altered biomechanical properties, introduction of stressconcentrations, differences in charge on the collagen fibrils, and leakage of cytotoxic products [9,10]. Regarding these drawbacks, researchers are aiming for new cross-linking procedures, whichdo not evoke cytotoxicity and give a better defined network structure with less alteredbiomechanical properties. A previous study [11] showed that cross-linking of porcine aortic heartvalves could be achieved with 1,4-butanediol diglycidyl ether, or with a water-solublecarbodiimide (EDC/NHS). In addition, cross-linking can be accomplished by using natural andcalcification inhibiting compounds [12] such as chondroitin sulfate [13].Another approach which received attention is the removal of components from the GA cross-linked matrix. Extraction methods using trypsin or sodium dodecyl sulfate (SDS) where applied onGA cross-linked bovine pericardium or porcine aortic leaflets to remove the antigenic substancessuch as cell remnants and non-collageneous matrix proteins [8, 14]. Detergents such as 3-(3-cholamidopropyl) dimethylammonio)-1-propanesulfonate (CHAPS) and sodium dodecyl sulfate(SDS) are able to remove lipids and cellular membranes [15, 16]. Enzymes like elastase candegrade the non-collageneous components such as elastin and probably other proteins present inthe tissue [17, 18]. Furthermore, the chaotropic solvent guanidine hydrochloride (GdCl)solubilizes proteoglycans, glycosaminoglycans (GAGs) and small proteins [19, 20]. However,treatment of cross-linked tissue with one of these methods should not affect the Ts or the stabilitytowards enzymes to a large extent.In previous experiments, treatment of leaflets prior to cross-linking with either elastase orguanidine hydrochloride revealed that the cross-linking reaction was altered, which suggests thatnot only collagen, but elastin, proteoglycans, GAGs and other components are involved [11].In this study, porcine aortic heart valves were cross-linked with (a) 1 ethyl-3-(3-dimethylaminopropyl) carbodiimide and N-hydroxysuccinimide (EDC/NHS), (b) 4,7,10-trioxa-1,13-tridecanediamine (TTDD) or (c) chondroitin sulfate (CS) in the presence of EDC and NHS, (d,e)1,4-butanediol diglycidyl ether (BDDGE) at pH 4.5 and 9.0, and (f) with a successive BDDGE


- 151 -

and EDC/NHS procedure [11]. The influence of the cross-linking procedure on the shrinkagetemperature, the swelling and the stability towards enzymatic attack are determined and comparedto glutaraldehyde cross-linked materials. In addition, (non)-cross-linked valves were treated witheither CHAPS/SDS, elastase or GdCl to selectively remove components from the tissue matrix.Evaluation of the components which were removed by these treatments enables one to get a betterunderstanding of the cross-linking process. Furthermore, selective extraction of (antigenic)components might be a tool to improve the durability of bioprostheses and to get more insight themechanism of calcification.


Tissue preparationNon-cross-linked porcine aortic heart valves (leaflets and wall, having a dry weight ofapproximately 1.5 g) were selected and dissected as described previously [11].

Cross-linking

Glutaraldehyde (GA)Aortic valves were immersed in 100 ml of a 0.01 M N-[2-hydroxyethyl]piperazine-N’-[2-ethanesulfonic acid] (HEPES free acid, Sigma-Chemical, St. Louis, USA) buffered (pH 7.4)solution containing 0.2 wt% GA (from a 25 % solution, Merck, Darmstadt, Germany). Thereaction was allowed to proceed for 24 h at 20 °C, after which the valves were lyophilized.1-ethyl-3-(3-dimethyl aminopropyl) carbodiimide. HCl (EDC) and N-hydroxysuccinimide (NHS)Aortic valves were immersed in 100 ml of a 0.1 M 2-[morpholino]ethanesulfonic acid (MES,Merck, Darmstadt, Germany) buffered solution (pH 5.5) containing 1.15 g EDC (z.S., Merck-Suchardt, Hohenbrunn, Germany) and 0.14 g NHS (z.S., Merck-Suchardt, Hohenbrunn,Germany). Cross-linking was carried out for 24 h at room temperature. After cross-linking, thevalves were washed with water before lyophilization.4,7,10-trioxa-1,13-tridecanediamine (TTDD) or chondroitin sulfate (CS)Aortic valves were immersed in 100 ml of a 0.1 M MES buffered solution (pH 5.5) containing0.53 g TTDD (Fluka, Buchs, Switzerland) or 1.0 g chondroitin sulfate (chondroitin sulfate A,Sigma Chemicals, St. Louis, USA). After 30 min of immersion, the solution was refreshed andsubsequently 1.15 g EDC and 0.28 g NHS were added. Cross-linking was allowed to proceed for24 h at 20 °C, after which the valves were washed with deionized water before lyophilization.1,4-Butanediol diglycidyl ether (pH 4.5 or pH 9.0)The valves were immersed in 100 ml of a 0.025 M disodium tetraborate decahydrate (pH 9.0) or0.1 M MES (pH 4.5) buffered solution containing 4 wt% 1,4-butanediol diglycidyl ether(BDDGE, Fluka, Buchs, Switzerland). Cross-linking was allowed to proceed for 72 h at pH 9.0and for 144 h at pH 4.5 at room temperature. After cross-linking the valves were washed withwater before lyophilization.

Chapter 9

- 152 -

Successive epoxy and carbodiimide cross-linkingAortic valves were immersed in 100 ml of a 0.1 MES buffered solution (pH 4.5) containing 4 gBDDGE. Cross-linking was allowed to proceed for 144 h at room temperature, after which thevalve was washed 2 times for 30 min with 0.01 M HEPES (pH 7.4) buffer. A valve was immersedin 100 ml of a 0.1 M MES buffered solution (pH 5.5) containing 1.15 g EDC and 0.28 g NHS(ratio EDC:NHS = 5:2). Cross-linking was allowed to proceed for 24 h at 20 °C, followed bywashing with deionized water before lyophilization.

Characterization

Extent of cross-linkingThe extent of cross-linking of leaflets and wall samples was related to the increase in shrinkagetemperature (Ts). The Ts of (non)-cross-linked leaflets in the hydrated state was determined withDSC as described previously [11]. The content of amine groups of the tissue before and aftercross-linking was determined spectrophotometrically after reaction of the primary amine groupswith 2,4,6-trinitrobenzenesulfonic acid (TNBS) and subsequent hydrolysis of the tissue [11] and isexpressed as the percentage [%] of the initial amount of amine groups.

Glycosaminoglycan (GAG) contentThe amount of GAGs present in the tissue was determined using a slightly modified method ofBlix as described previously [11, 21].

Amino acid analysisThe amino acid composition of tissue samples was analyzed by reversed-phase high performanceliquid chromatography (HPLC) as described previously [22].

SwellingAbout 6.0 mg of lyophilized tissue (Wdry) was immersed in 5.0 ml phosphate buffered saline (PBS,NPBI, Emmercompascuum, The Netherlands; pH 7.4). After one day, the tissue was removedfrom the solution, blotted with paper tissue and weighed again (Wwet). The swelling is defined as:S = (Wwet-Wdry)/Wdry.* 100%.

In-vitro degradation

CollagenaseIn a typical experiment, 3-5 mg samples of (non) cross-linked tissue were immersed in 1.0 ml of a0.1 M Tris-HCl buffer (pH = 7.4) containing 0.005 M CaCl2 and 0.05 mg/ml NaN3. After one h,1.0 ml bacterial collagenase from Clostridium histolyticum (EC 3.4.24.3, Sigma-Chemical, St.Louis, USA) solution in Tris-HCl buffer (37 °C) was added to give the desired concentration of100 U/ml. The degradation was discontinued at the desired time interval by addition of 0.2 ml 0.25M EDTA (Titriplex III, Merck, Darmstadt, Germany). The solutions were cooled on ice. Exactly0.1 ml of the supernatant was transferred into 1.0 ml 6 M HCl followed by hydrolysis for 16 h at


- 153 -

110 °C. The weight-loss was measured by assaying the supernatant for hydroxyproline asdescribed previously [11].PronaseA pronase stock solution was prepared by dissolving a calculated amount of pronase (fromStreptomyces grisseus) lyophilizate (Boehringer Mannheim, 7000 U/g lyophilizate) into a 0.1 MTris-HCl buffer solution (pH = 7.4) containing 0.005 M CaCl2 and 0.05 mg/ml NaN3. The finalpronase concentration was 20 U/ml. This stock solution was incubated at 37 °C for one h. In atypical experiment, 3-5 mg of (non)-cross-linked tissue was immersed in 2.0 ml pronase solution(37 °C). Degradation was discontinued at the desired time interval by addition of 0.2 ml 0.25 MEDTA, after which the solutions were cooled on ice. The weight-loss was determined bymeasuring the supernatant for hydroxyproline as described above.

Extraction methods

CHAPS/SDSA valve was immersed in 100 ml of a 0.01 M HEPES buffered solution (pH 7.4) containing 0.008M 3-(3-cholamidopropyl) dimethylammonio)-1-propanesulfonate (CHAPS, Sigma-Chemical, St.Louis, USA), 1 M sodium chloride and 0.025 M EDTA. After 60 min, the valve was washed twicefor 5 min with 0.01 M HEPES. Then, the valve was transferred to 100 ml of a 0.01 M HEPESbuffered (pH 7.4) solution containing 0.0018 M sodium dodecyl sulfate (SDS, Merck, Darmstadt,Germany), 1 M NaCl and 0.025 M EDTA for 15 min. The valve was rinsed twice for 5 min with0.01 M HEPES buffer and treated with a final 100 ml of CHAPS (pH 7.4) solution. After 60 min,the valve was rinsed 5 times for 30 min with 0.01 M HEPES before lyophilization. The amount ofmaterial removed was determined gravimetrically.ElastaseAn elastase (Type I, from porcine pancreas, Sigma, St. Louis, activity 100 U/mg, 1 Unit willsolubilize 1 mg of elastin in 20 min at pH 8.8 and 37 °C) stock solution with an activity of 10 U/mlwas prepared by dissolving the desired amount of elastase in 0.1 M Tris-HCl buffer (pH 8.8)containing 0.005 M CaCl2 and 0.05 mg/ml NaN3. This stock solution was allowed to stand for 1 hat 37 °C before use. To a tissue sample weighing 20 to 30 mg, 5.0 ml of elastase solution in Tris-HCl buffer (37 °C) was added. Degradation was carried out for 24 h at 37 °C. Degradation wasterminated by addition of 0.5 ml of 0.25 M EDTA. After washing with deionized water thesamples were lyophilized. The amount of elastin removed was determined gravimetrically andcolorimetrically by a hydroxyproline assay on the enzyme solution after treatment. Exactly 0.5 mlof the remaining solution was hydrolyzed in 1.0 ml 6 M HCl for 16 h at 110 °C, followed by theprocedure as described earlier [11].Guanidine hydrochlorideThe amount of components such as proteoglycans and GAGs of the dry tissue was determined byimmersing the valves in a chaotropic solution (pH 5.8), containing 4 M guanidine hydrochloride(Merck, Darmstadt, Germany), 2 wt% Triton-X-100 (Sigma-Chemical, St. Louis, USA), 0.1 Msodium acetate and a small amount of a protease inhibitor cocktail (Sigma Chemical, St. Louis,USA) [19]. After 24 h of treatment at 20 °C, the tissue was washed thoroughly with deionizedwater before lyophilization. The amount of material removed was determined gravimetrically.

Chapter 9

- 154 -

RESULTS

Cross-linkingThe different cross-linking procedures that were applied for the stabilization of the leaflets ofporcine aortic heart valves have been described in detail previously [11], except for chondroitinsulfate (CS). Optimized cross-linking procedures have been used and the leaflets werecharacterized by the shrinkage temperature (Ts), transition enthalpy (∆Hs), and the amine groupand glycosaminoglycan (GAG) content (Table I).

Table IThe shrinkage temperature (Ts), the transition enthalpy (∆Hs ), the amine group and

glycosaminoglycan (GAG) content of (non)-cross-linked porcine aortic leaflets.Material Ts

[°C]Amine groups

[%]∆Hs[J/g]

GAG content [%]

Non-cross-linked 62.4 ± 0.8 100.0 ± 5.0 12.0 ± 2.0 14.5 ± 2.0GAEDC/NHSTTDDBD90BD45BD45ENCS

84.1 ± 1.586.3 ± 0.288.5 ± 0.281.4 ± 1.574.0 ± 3.582.9 ± 0.484.3 ± 0.3

32.6 ± 1.5 58.1 ± 2.5 97.5 ± 2.0 54.4 ± 3.5 95.4 ± 4.0 56.4 ± 6.0 78.8 ± 0.5

11.8 ± 1.017.9 ± 1.517.1 ± 1.012.8 ± 2.5 9.3 ± 1.513.1 ± 1.018.8 ± 0.5

14.1 ± 0.515.9 ± 0.515.0 ± 0.415.9 ± 1.014.5 ± 0.916.0 ± 0.719.9 ± 0.6

The next abbreviations are used:GA = Glutaraldehyde cross-linking (0.2 wt%, 24 h, pH 7.4)EDC/NHS = EDC/NHS cross-linking (1.15 g EDC and 0.14 g NHS, 24 h, pH 5.5)TTDD = 4,7,10-trioxa-1,13-tridecanediamine (TTDD) cross-linking in the presence of EDC/NHS (0.53 g TTDD,1.15 g EDC, and 0.28 g NHS, 24 h, pH 5.5)BD90 = 1,4-butanediol diglycidyl ether (BDDGE) cross-linking at pH 9.0 (4 wt% BDDGE, 72 h, pH 9.0)BD45 = BDDGE cross-linking at pH 4.5 (4 wt% BDDGE, 144 h, pH 4.5)BD45EN = successive epoxy and carbodiimide cross-linking (4 wt% BDDGE, 144 h, pH 4.5 followed by 1.15 gEDC and 0.28 g NHS, 2 h, pH 5.5)CS = Chondroitine sulfate A (CS) cross-linking in the presence of EDC/NHS (1.0 g CS, 1.15 g EDC, and 0.28 gNHS, 24 h, pH 5.5)

Cross-linking of aortic leaflets with the different methods resulted in an increase of the Ts from 62°C to 81 °C or higher, except for the bisepoxy cross-linking at pH 4.5 (BD45) which resulted in amaterial with a Ts of 74 °C. Glutaraldehyde (GA) and bisepoxy (pH 9.0; BD90) cross-linkingoccurred via the amine groups as reflected by the large decrease in percentage of amine groups.Cross-linking methods in which the EDC/NHS activation methods were used gave materials withthe highest values of Ts (84 - 88 °C). Leaflets cross-linked with 4,7,10-trioxa-1,13-tridecanediamine (TTDD) did not show a reduction in amine groups (98 %), while a reduction to79 % was found after cross-linking with chondroitin sulfate (CS) and a large decline to 58 % wasobserved for EDC/NHS cross-linked leaflets. Hardly any change in amine groups was obtained inleaflets cross-linked with BD45. An additional EDC/NHS cross-linking step performed on BD45resulted in the insertion of cross-links as judged from the higher Ts and the lower content of aminegroups.


- 155 -

The transition enthalpy of cross-linked leaflets (∆Hs) was only affected when EDC and NHS wereused leading to an increase from 12 to 17 J/g.The GAG content was not affected by the cross-linking procedure applied, except for CS cross-linking which showed a significant increase from 14.5 to 19.9 %.The amino acid composition of (non) cross-linked leaflets was determined in order to evaluate ifreactive groups other than the amine groups of (hydroxy)lysine residues were involved in thecross-linking reactions (table II).

Table IIAmino acid composition of (non)-cross-linked porcine aortic leaflets in [n/1000]

Aminoacid

Non-cross-linked

GA EDC/NHS

TTDD BD90 BD45 BD45EN CS

AspGluGlyProMet(HO)LysHis

638726210294110

67902741088158

608727010484110

65902609993912

6488271980230

6486265983418

62862791022366

7199262959307

Non-cross-linked porcine leaflets had a remarkably high content of (hydroxy)lysine residues of 41(n/1000). GA cross-linking involved the amine groups of (hydroxy)lysine residues and no changesin other amino acid residues were detected. Leaflets cross-linked with EDC/NHS or TTDD didnot display significant changes in amino acid composition. BD90 cross-linked leaflets displayednot only a decrease in (hydroxy)lysine residues but in histidine and methionine residues as well.BD45 and BD45EN showed only a significant reduction in methionine residues from 9 to 3(n/1000). Finally leaflets cross-linked with CS exhibited an unusually high amount of glutamic andaspartic acid residues, whereas the content of (hydroxy)lysine residues of 30 (n/1000) wasmarkedly low.The aortic wall is an essential component of stentless heart valves. Analogously to the leaflets, theeffects of the cross-linking procedures on the thermal properties and the content of amine groupsof the aortic wall were evaluated (Table III).

Chapter 9

- 156 -

Table III The shrinkage temperature (Ts), the shrinkage transition (∆Hs) and the percentage of

amine groups of (non) cross-linked aortic wall tissueMaterial Ts

[°C]∆Hs[J/g]

Amine group content[n/1000]

Non-cross-linkedGAENTTDDBD90BD45BD45ENCS

58.6 / 63.658.2 / 63.6 / 84.158.2 / 63.1 / 84.257.7 / 63.2 / 85.157.2 / 63.2 / 85.358.9 / 63.6 / 77.758.3 / 63.1 / 80.358.4 / 63.6 / 83.8

2.8 ± 0.52.3 ± 0.61.6 ± 0.52.5 ± 0.43.1 ± 0.72.9 ± 0.91.6 ± 0.62.3 ± 0.5

100.0 ± 3.5 60.5 ± 6.8 84.2 ± 2.5 95.5 ± 3.0 60.9 ± 2.0 89.2 ± 1.7 83.7 ± 1.7

not determinedAbbreviations are explained in table I

Thermal analysis demonstrated that the amine group content after cross-linking of the aortic wallwas significant higher than for the leaflets. For example, valves cross-linked with GA had an aminecontent of 40 % for leaflets while a value of 61 % was observed for the corresponding walls.Furthermore, aortic leaflets cross-linked with EDC/NHS had an amine group content of 58 %,whereas 84 % of amine groups remained free after cross-linking of the wall. Moreover, thermalanalysis of aortic wall tissue showed three transitions. Two were located around 60 °C, whichcorrespond to the peaks observed for non-cross-linked tissue, and one peak emerged at atemperature almost similar to the Ts of the leaflets. The value of the total transition enthalpy was2.8 ± 0.5 J/g and was not significantly changed by cross-linking.

SwellingThe swelling of (non) cross-linked leaflets and wall tissue was determined (Table IV).

Table IVThe swelling of the leaflets and the aortic wall before and after cross-linking

(24 h, 20 °C, PBS, pH 7.4)Material Swelling leaflets [%] Swelling aortic wall [%]Non-cross-linked 680 ± 90 225 ± 3GAENTTDDBD90BD45BD45ENCS

470 ± 70360 ± 30280 ± 10380 ± 80290 ± 30270 ± 20340 ± 20

233 ± 8189 ± 5178 ± 3195 ± 8164 ± 3142 ± 4223 ± 3

In general, the swelling of the tissue was reduced as a result of the chemical cross-linksintroduced. Whereas non-cross-linked tissue exhibits a swelling of 680 %, cross-linked leafletsshowed values between 270 and 470 %. The swelling was dependent on the cross-linking methodutilized. Leaflets cross-linked with GA showed the highest swelling, whereas TTDD and BD45EN


- 157 -

cross-linked leaflets had the lowest water uptake. Similar trends in swelling were observed for theaortic wall, although the differences were less pronounced.

Enzymatic degradationThe in-vitro stability towards enzymatic degradation of (non)-cross-linked porcine leaflets andwall tissue was evaluated using bacterial collagenase or pronase (Tables V and VI).

Table VRemaining weights [%] of (non) cross-linked leaflets after exposure to a bacterial

collagenase solution (100 U/ml, 48 h, 37 °C)Material Remaining weight of the

leaflets [%]Remaining weight of the

aortic wall [%]Non-cross-linked 11 ± 3 7 ± 4GAENTTDDBD90BD45BD45ENCS

96 ± 197 ± 197 ± 196 ± 164 ± 298 ± 197 ± 1

73 ± 369 ± 364 ± 175 ± 271 ± 183 ± 358 ± 5

Cross-linked leaflets demonstrated a high stability towards collagenase degradation, and onlyBD45 cross-linked leaflets had a low resistance against degradation. In contrast to the highstability of the leaflets, cross-linked aortic wall tissue displayed larger changes in weight uponcollagenase degradation and about 70 % of the initial weight was retrieved.

Table VIRemaining weights [%] of (non) cross-linked leaflets after exposure to a pronase solution

(20 U/ml, 24 h, 37 °C)Material Remaining weight of the

leaflets [%]Remaining weight of the

aortic wall [%]Non-cross-linked 0 0GAENTTDDBD90BD45BD45ENCS

97 ± 296 ± 195 ± 295 ± 147 ± 294 ± 197 ± 1

70 ± 573 ± 372 ± 279 ± 250 ± 371 ± 140 ± 4

In general, cross-linking of aortic leaflets afforded stable materials towards pronase degradation.Similar trends were found as observed in the collagenase experiments and only BD45 cross-linkedleaflets were partly degraded by pronase. Aortic wall cross-linked with either BD45 or CSexhibited a low resistance towards pronase.

Chapter 9

- 158 -

Extraction of tissue components

Non-cross-linked tissueTo determine which components were involved in the cross-linking process, selective removal ofcomponents from the tissue matrix was performed. Exposing the tissue to a detergent mixture of3-(3-cholamidopropyl) dimethylammonio)-1-propanesulfonate (CHAPS) and sodium dodecylsulfate (SDS) will remove cellular remnants and lipids. Treatment of tissue with elastase will resultin a matrix without elastin, and the use of guanidine hydrochloride (GdCl) will results in a materiallacking in proteoglycans, GAGs and (glyco)proteins. The effect of each of these treatment on theremaining weight, the thermal properties and the GAG content was investigated (table VII).

Table VIIThe remaining weight, the shrinkage temperature (Ts), the transition enthalpy (∆∆∆∆Hs), and

the glycosaminoglycan (GAG) content of leaflets as a function of the extraction methodutilized

Treatment Remainingweight [%]

Ts [°C] ∆Hs [J/g] GAG [% of dryweight]

NoneCHAPS/SDSElastaseGdCl

10066.5 ± 3.553.5 ± 2.070.7 ± 1.5

62.463.762.258.0

12.014.432.310.6

14.510.2~ 04.6

Treatment of non-cross-linked leaflets with CHAPS/SDS, elastase or guanidine hydrochlorideresulted in the loss of 30 to 45 % of the initial weight. Moreover, the variable decrease in GAGcontent implies that this component could not be removed selectively. Elastase apparentlyremoved all components except the collagen and an enormous increase in transition enthalpy from12 to 32 J/g was observed. Treatment of non-cross-linked leaflets with GdCl resulted in slightlylower values of Ts and ∆Hs.

Cross-linked aortic leaflets were also subjected to the three extraction methods. The remainingweights after extraction are presented in figure 1.


- 159 -

None GA EN TTDD BD90 BD45 BD45EN C4S50

60

70

80

90

100

Rem

aini

ng w

eigh

t [%

]

CHAPS/SDS Elastase GdCl

Figure 1. Remaining weights of (non) cross-linked porcine aortic leaflets as a function of thecross-linking method (x-axis) and the extraction or degradation method (legend)

Exposure of cross-linked leaflets to the extraction or enzyme solutions resulted in higherremaining weights compared to non-cross-linked leaflets. Whereas 67 % of the initial weight ofnon-cross-linked tissue was left after CHAPS/SDS treatment, more than 80 % of the material wasretrieved after cross-linking. Furthermore, the remaining weight after elastase degradation wasincreased from 53 to 65 % or more, except for BD45 cross-linked leaflets. Especially cross-linkingwith the diamine TTDD in the presence of EDC and NHS or with the bisepoxy compound at pH9.0 (BD90) resulted in materials which were highly stabilized with respect to degradation byelastase. In addition, cross-linking reduces the amount of components which can be extracted byGdCl from 30 % for non-cross-linked leaflets to 15 % or less for cross-linked leaflets. The Ts(data not shown) was hardly altered after each treatment and only slight increases of 1 or 2 °Cwere observed. Furthermore, the amine group content (data not shown) was slightly increased. Onthe other hand, the transition enthalpy remained unchanged after CHAPS/SDS and GdCltreatments, whereas an increase from 13-17 to 20-37 J/g was observed for cross-linked leafletsupon elastase degradation. Finally, the GAG content of the (non)-cross-linked leaflets before andafter the different treatments was determined.

Table VIII The content of GAGs of (non)-cross-linked aortic leaflets after several treatments. The

content is defined as the weight percentage of the initial (before post-treatment) dry weightMaterial No treatment CHAPS/SDS Elastase GdClNon-cross-linked 14.5 10.2 ~0 4.6GA 14.1 11.9 2.4 9.1EN 15.9 11.6 7.4 12.4TTDD 15.0 12.7 6.7 11.9BD90 15.9 12.0 n.d. 12.3BD45 14.5 16.2 3.6 10.1BD45EN 15.0 12.7 9.4 14.5C4S 19.9 17.8 5.9 16.4

Chapter 9

- 160 -

Treatment of cross-linked leaflets with CHAPS/SDS resulted in a slight reduction of the GAGcontent. Exposure of the cross-linked leaflets to an elastase solution revealed that a small contentof GAGs was left. Moreover, the content of GAGs in the matrix after exposure to guanidinehydrochloride (GdCl) was considerably higher in cross-linked than in non-cross-linked leaflets.

Analogously to the leaflets, extraction experiments with non-cross-linked aortic wall tissue werecarried out (Table IX).

Table IXThe remaining weight, the GAG content and the swelling (PBS, 24 h, 20 °C) of (non)-

extracted fresh aortic wall tissueTreatment: Remaining weight

[%]GAG content

[%]Swelling [%]

NoneCHAPS/SDSElastaseGdCl

10090.720.182.7

7.36.7~ 01.9

220 ± 10250 ± 10530 ± 40240 ± 10

A low content of material (9 - 17 %) was extracted from the tissue matrix if the aortic wall wastreated with CHAPS/SDS or GdCl, whereas 80 % was removed by elastase. The GAGs were fullyremoved upon elastase treatment, whereas 6.7 % and 1.9 % was left in the wall after CHAPS/SDSand GdCl treatments, respectively. The swelling of the tissue was only significantly affected byelastase and a steep increase from 325 to 530 % was observed. Extraction of cross-linked aorticwall tissue resulted in similar trends as described for the leaflets, although differences betweencross-linked and non-cross-linked material were less pronounced.

DISCUSSION

Stented tissue heart valves have been widely applied for over 3 decades [23]. During the past fewyears, stentless valves became more clinically applied because these valves have superiorheamodynamics and less problems regarding tearing of the leaflets due to stress concentrationsintroduced by the rigid stent. However, the long-term durability of tissue valves is limited due tocalcification [24, 25]. Although the exact mechanism of calcification is still under investigation, itis clear that several determinants play a crucial role in this process [26]. New cross-linkingprocedures were developed which proved to reduce calcification [27, 28]. In addition, washingand extraction methods are utilized to remove antigenic and calcium binding components from thecross-linked tissue [8, 24, 29].Based on previous work on the cross-linking of dermal sheep collagen, the in-vivo behavior andcalcification of the materials obtained [22, 30, 31], and the work done on the optimization ofcross-linking procedures for porcine aortic heart valves [11], several methods were selected.Porcine aortic heart valves were cross-linked with 1-ethyl-3-(3-dimethylaminopropyl)carbodiimide and N-hydroxysuccinimide (EDC/NHS), 4,7,10-trioxa-1,13-tridecanediamine(TTDD) in the presence of EDC/NHS, 1,4-butanediol diglycidyl ether (BDDGE) at pH 4.5 or 9.0


- 161 -

(BD45 or BD90), or with a successive BDDGE (pH 4.5) and EDC/NHS procedure (BD45EN).Furthermore, chondroitin sulfate in the presence of EDC/NHS proved to be an effective cross-linker for dermal sheep collagen, resulting in a very high stability towards enzymatic degradation(unpublished results). Therefore, this method was also applied for the stabilization of heart valves(CS). The properties of the cross-linked materials were compared with glutaraldehyde cross-linkedcontrols (GA).Cross-linking of porcine aortic leaflets was accomplished as indicated by an increase in shrinkagetemperature (Ts). Furthermore, the majority of the methods applied led to a decline in aminegroups indicating that reactions had occurred. GA cross-linking of leaflets resulted in an increaseof Ts from 61 to 84 °C and in the largest reduction in amine groups to 33 %. This implies thatcross-links were formed between the (hydroxy)lysine residues. Cross-linking of leaflets with EDCand NHS resulted in materials with a high Ts of 86 °C, whereas the content of amine groups wasdecreased to 58 %. Leaflets which were cross-linked with EDC/NHS in the presence of TTDD,displayed the highest Ts (89 °C), while the content of amine groups was hardly reduced. Thisindicates that besides the insertion of zero-length and TTDD cross-links, one-sided (masking)reactions occurred, leading to pendant primary amine groups. Cross-linking of aortic leaflets withthe bisepoxy at pH 9.0 (BD90) occurred via the amine groups. In contrast, bisepoxy cross-linkingat pH 4.5 (BD45) resulted in the formation of cross-links between the carboxylic acid groups. Amaterial with a low cross-link density is obtained, as indicated by the low Ts of 74 °C. Anadditional EDC/NHS cross-linking step was performed on BD45 (BD45EN), which resulted in anincrease in Ts from 74 to 82 °C due to the introduction of amide-type cross-links. Leaflets cross-linked with chondroitin sulfate in the presence of EDC and NHS (CS) , exhibited a high Ts incombination with a high content of amine groups of 79 %. This suggests that few cross-links wereformed and that the chondroitin sulfate cross-links have a large effect on the Ts of the material aswas observed earlier probably due to strong ionic interactions between GAGs and collagen.No significant change in transition enthalpy (∆Hs) was detected in leaflets cross-linked with GA,BD90 or BD45EN. On the contrary, cross-linking of leaflets with the EDC and NHS activationmethod (EDC/NHS, TTDD and CS) showed a significant increase in transition enthalpy form 12to 17-19 J/g. This might be ascribed to the zero-length cross-links, which occur intra- orinterhelically [32]. This indicates that the helices are highly stabilized and that probably a betterthree-dimensional packing of the individual collagen molecules was established. BDDGE and GAform 'extended' cross-links and also result in one-sided masking reactions and hence result in a lesstight organization of the triple-helices [33, 34].The methods applied do not significantly change the content of GAGs. In contrast, other groupsassert that GAGs such as dermatan sulfate and chondroitin sulfate are removed during cross-linking and hence the natural inhibiting effect on calcification is lost [12, 35]. Reaction of leafletswith chondroitin sulfate was achieved as reflected by an increase in GAGs from 14.5 to 19.9 %.

Amino acid analysis of cross-linked leaflets reveals that GA only reacted with the amine groups of(hydroxy)lysine. Cross-linking methods in which the carboxylic acid groups were involved did notshow a change in the amino acid composition due to hydrolysis, during the analytical method, ofthe amide and ester bonds containing cross-links. Leaflets which were cross-linked with EDC andNHS did not show any changes in amino acid composition, suggesting that only amide type bonds

Chapter 9

- 162 -

were formed. Porcine leaflets cross-linked with BDDGE at pH 4.5 or 9.0 showed a ratherunexpected decrease in methionine (table II). Methionine can act as a nucleophile and alkylationreactions with α-haloacids are reported. However, methionine is generally embedded in thehydrophobic regions of proteins, resulting in a low reactivity [20, 36]. Epoxide groups of BDDGEcan react with the secondary amine groups of histidine under basic conditions (Table II), whichagrees with earlier studies [22, 37].

The aortic wall, which is an essential part of stentless valves, has a composition which iscompletely different from the leaflets [11]. A high content of elastin is present (34 %) whichresults in a much more hydrophobic material than the leaflets. In addition, the material is muchthicker and more dense, which reduces the cross-linking rate due to a slower penetration of thereagents into the tissue. Cross-linking of aortic wall tissue was therefore incomplete as reflected bythe appearance of three transitions in the thermograms (Table III). The first two transitions in thethermograms are related to the shrinkage endotherms of the non-cross-linked tissue, whereas thelast peak reflects the cross-linked portion of the wall. The Ts of the cross-linked areas resemblesthe values found for cross-linked leaflets. The incompleteness of the cross-linking reaction wasconfirmed by the higher percentage of amine groups of cross-linked aortic wall samples comparedto the corresponding leaflets. For example, the amine group content of GA cross-linked leafletswas reduced to 33 %, while a value of 61 % was measured for the corresponding GA cross-linkedwall. The incomplete cross-linking was visualized after a reaction of the wall with a 2,4,6-trinitrobenzene sulfonic acid solution as described previously [11]. Addition of the ∆Hs values ofeach peak resulted in a much lower transition enthalpy (2.8 ± 0.3 J/g) compared to the leaflets(12.0 ± 2.0 J/g). Despite the lower content of collagen in the wall (19 %) as compared to theleaflets (58 %), the ∆Hs, if expressed as J/g of collagen in the sample, is still lower (14.7 versus20.7 J/g collagen).

Cross-linking of aortic heart valves decreases the swelling of the material (table III). Surprisingly,large differences in swelling were observed for cross-linked materials, which indicates thatswelling is not only determined by the cross-link density. The swelling of a poly-electrolyte isdetermined by the degree of ionization, the degree of cross-linking, the pH and the ionic strengthof the swelling medium, and the hydrophilic/hydrophobic balance [38, 39]. As an example, poly(acrylic acid-co-hydroxyethyl methacrylate) cross-linked hydrogels showed a decrease in swellingat constant pH if the content of the non-ionizable (HEMA) monomer was increased or the degreeof cross-linking was raised [39]. Proteins are polyampholytes and differences in content ofionizable groups will affect the swelling [40]. Another parameter which might influence theswelling behavior of the valve tissue, is the length and nature of the cross-links. This can beillustrated by figure 2, which shows that a material with a high Ts does not automatically have alow degree of swelling.


- 163 -

60 65 70 75 80 85 90200

300

400

500

600

700

800

Sw

ellin

g [%

]


Figure 2. The swelling of (non) cross-linked leaflets as a function of the shrinkage temperature.non-cross-linked (�), tissue cross-linked (�) via the amine groups (GA, BD90), (▲) via amidebonds (EDC/NHS), (�) via carboxylic acid groups (BD45), (�) via amide bonds and carboxylicacid groups (TTDD, BD45EN), and (�) via amide bonds and amine groups (CS).

Leaflets cross-linked only through their amine groups, as in GA or BD90, showed high swellingratios of 380 - 470 % (see � in figure 3). On the contrary, tissue cross-linked via the carboxylicacid groups such as in BD45 (�), exhibited a very low swelling of 290 %. Because the cross-linkdensity of BD45 is rather low as reflected by its low Ts (74 °C), the steep reduction in swellingwill be predominantly determined by changes in the degree of ionization. Reaction of carboxylicacid groups with BDDGE reduces the negative charge on the collagen helices, which results inless repulsion between the negatively charged molecules. Furthermore, the hydrophilicity of thematerial is somewhat decreased. Consequently, the swelling is reduced and a more densely packedstructure is obtained. Independent measurements showed that masking of the carboxylic acidgroups with glycidyl isopropyl ether (pH 4.5) also resulted in a decrease of the swelling, whichconfirms that the content of ionizable groups affects the degree of swelling. On the contrary,masking of amine groups using acetic acid N-hydroxysuccinimide or glycidyl isopropyl ether underbasic conditions, demonstrated a somewhat higher swelling of the leaflets as compared to nontreated leaflets. Because amine groups are mainly positively charged under physiologicalconditions (pKa of ε-amine groups is 10.0 [36]), the collagen molecules will be more negativelycharged after masking of the amines. The increase in swelling is therefore determined by the higherrepulsion between the helices and the insertion of branches. Hence, the reduced swelling of leafletscross-linked via their amine groups with respect to non-cross-linked leaflets is mainly caused bythe cross-link density and the nature of cross-links.Tissue cross-linked with EDC/NHS resulted in a reduction of both amine and carboxylic acidgroups. The balance between ionizable residues is expected only to be slightly affected. Despitethe high amount of cross-links (Ts = 86 °C), the swelling was higher than for example BD45which had a much lower Ts. This confirms that both the increase in cross-link density and thedecrease in content of ionizable residues determine the swelling behavior of leaflets.

Chapter 9

- 164 -

Materials cross-linked with TTDD or CS show a swelling which is comparable to EDC/NHScross-linked leaflets. This suggests that the majority of the inserted cross-links were zero-lengthones. Furthermore, it is emphasized that introduction of chondroitin sulfate molecules in thematrix either as a branch or a cross-link, would result in a high swelling because CS can beconceived as a polyelectrolyte due to the high content of negatively charged carboxylate andsulfate groups [41]. For example, encapsulation of hyaluronan and collagen (type I) inpolyacrylamide gels showed an increased isothermal swelling with an increasing amount ofbiopolymer [42]. Because the swelling was not significantly increased with respect to EDC/NHScross-linked leaflets, it was assumed that mainly zero-length cross-links were formed. The increasein CS content (table I) shows that CS is incorporated in the leaflet matrix and it is assumed thatCS molecules are coupled on and between the collagen fibers or fibrils.Because the aortic wall is composed of only 19 % collagen and 34 % of elastin, the effect of eachcross-linking procedure on the swelling will be less pronounced as compared to the leaflets (tableIV). However, similar trends were observed as for the leaflets.

The stability of cross-linked collagen-based materials towards enzymatic degradation was assessedusing a collagen specific enzyme, collagenase, and a collagen a-specific enzyme, pronase. Ingeneral, cross-linking improved the stability of the leaflets to a large extent, and only low amountsof collagen and elastin (3-5 %) were degraded, which was comparable to the GA cross-linkedcontrol leaflets. Only leaflets which were cross-linked with the bisepoxy at pH 4.5 (BD45)exhibited a lower resistance towards enzymatic degradation probably due to its lower cross-linkdensity and its ester-containing cross-links which can be hydrolyzed by enzymes [32].On the contrary, different trends were found in cross-linked aortic wall tissue. Although cross-linking stabilized the tissue, a considerably higher loss of collagen and elastin between 20 and 30% was observed in the cross-linked wall tissue, probably due to the less homogeneous cross-linking of the aortic wall. The stability towards enzymes of the walls cross-linked via the newroutes was comparable to the GA treated controls, except for the BD45 and CS cross-linkedwalls, which showed a poor resistance towards pronase. The low stability of the CS cross-linkedwall is explained by the low penetration depth of the large CS molecules in the wall matrix.

Aortic valves are very complex structured materials, and besides collagen, also elastin,proteoglycans, GAGs, non-collageneous proteins and cellular components are present andinterconnected with each other. Not only collagen but elastin [43] and GAGs [41] contain amineor carboxylic acid residues as well, and hence they can be involved in the cross-link reaction.Selective extraction of the tissue components after cross-linking can give information about theproteins which were involved in the cross-linking reactions. In addition, removal of thesecomponents might be advantageous for the long-term durability of stentless valve prostheses,because several researchers have concluded that the presence of cellular remnants, phospholipids,(glyco)proteins and elastine induces or enhances calcification.The first extraction method comprises a detergent treatment with 3-(3-cholamidopropyl)dimethylammonio)-1-propanesulfonate (CHAPS), which is a non-denaturing zwitterionicdetergent for solubilization of membrane proteins [16, 17] and sodium dodecyl sulfate (SDS),which denatures proteins. The second treatment is carried out with elastase, which is able to


- 165 -

degrade elastin and other non-helical proteins [17, 18]. The third method comprises the use ofguanidine hydrochloride (GdCl), a chaotropic solvent, which denatures proteins and dissociateshydrogen bonds and other non-covalent interactions [19]. Consequently, proteoglycans, GAGsand matrix proteins are extracted.Treatment of non-cross-linked leaflets using the methods described above resulted in a reductionof the remaining weights (table VII). CHAPS/SDS removed about 30 % of the material. Elastaseremoved all components from the matrix except the collagen, and the remaining weight of 54 %corresponds to the average collagen content of 58 % present in leaflets. GdCl treatment resulted inthe loss of 30 % of material. Whereas all GAGs were removed by elastase, about 70 % and 30 %of the initial GAG content was extracted by GdCl and CHAPS/SDS, respectively. The incompleteremoval of GAGs after GdCl treatment implies that there is a strong (ionic) interaction betweenthe collagen and the proteoglycan-GAG complexes. The Ts of leaflets was only slightly affectedby the three procedures which suggests that the helical integrity of the collagen was maintained[44]. Furthermore, the transition enthalpy was increased, especially after elastase treatment, whichis mainly due to the higher relative content of collagen in the leaflet after extraction.Cross-linking of aortic leaflets with each of the methods described resulted in lower amounts ofextractable material as reflected in figure 1. Whereas 67 % of material remained in non-cross-linked leaflets after a CHAPS/SDS treatment, more than 80 % was left in cross-linked leaflets.The extraction solutions were analyzed for the presence of proteins (TNBS assay), GAGs(hexosamine assay) or collagen and elastin (hydroxyproline). These analyses demonstrate thatCHAPS/SDS was not able to remove elastin or collagen. Moreover, only low amounts of GAGsand proteins were present in the extraction solution, which suggests that mainly cellularcomponents were released from the matrix. Similar to CHAPS/SDS treatment, less componentscould be removed from the cross-linked tissue by elastase as compared to the non-cross-linkedleaflets. The increase in remaining weight of extracted leaflets from 54 % to 60 - 75 % suggeststhat elastin participated in the cross-link reactions. Furthermore, the higher GAG content (TableVIII) suggests that also GAGs were covalently coupled to the collagen molecules. Analysis of theelastase extraction solutions demonstrated the presence of proteins, elastin and GAGs.Analogously to the other treatments, less material could be extracted from the cross-linked leafletmatrix upon exposure to a GdCl solution as compared to non-cross-linked leaflets. Analysis of theremaining solution revealed that no collagen and elastin and low amounts of GAGs and proteinswere removed from the leaflets.The effect of each 'extraction' method on the Ts was almost negligible. Furthermore, the aminegroup content after extraction was slightly elevated due to the higher relative content of collagenin the tissue. In contrast, the transition enthalpy was changed considerably. Whereas CHAPS/SDSand guanidine hydrochloride did not influence the ∆Hs to a large extent, a steep increase wasobserved after exposure to an elastase solution due to a higher relative collagen content in theresidual leaflets.Extraction experiments were also carried out on (non) cross-linked aortic walls. Again, elastaseremoved all tissue components from the non-cross-linked wall except the collagen. The remainingweight of non-cross-linked aortic walls treated with either CHAPS/SDS or guanidinehydrochloride was very high, probably due to the low penetration depth of these reagents into thedense wall. Similar trends as with the leaflets were obtained if cross-linked aortic walls were

Chapter 9

- 166 -

exposed to the three treatments as mentioned before. However, the effect of the cross-linkingprocedure on the amount of material which was extracted was very small, and the remainingweight of non-cross-linked walls was comparable to cross-linked ones. This suggests that duringthe cross-linking of the wall mainly collagen was involved. In addition, due to the incompletenessof the cross-linking reaction, differences between the non-cross-linked and the cross-linked tissueare small.

The presence of cell remnants, phospholipids, non-collagenous proteins and elastin and theutilization of the glutaraldehyde cross-linking process might be key determinants in inducingcalcification. Therefore, new cross-linking methods which stabilize the valve tissue similar as theglutaraldehyde process have been developed and tested. The use of extraction methods asdescribed in this paper in order to remove the components which are antigenic and possiblenucleation sites for calcium phosphate crystals from the cross-linked tissue, may result in materialswhich show a low tendency to calcify. Furthermore, selective extraction of components from thecross-linked tissue may be a tool to elucidate the mechanism of calcium deposition in soft tissues.Application of new cross-linking methods in combination with the selective removal ofcomponents possibly enables one to manufacture tissue heart valves with prolonged durability.

CONCLUSIONS

Cross-linking of porcine aortic heart leaflets was achieved with several methods and wascompared with glutaraldehyde cross-linked leaflets. The shrinkage temperatures of cross-linkedleaflets were in the same range (81 - 89 °C) as for GA cross-linked controls (85 °C), except forleaflets which were cross-linked with the bisepoxy compound (BDDGE) at pH 4.5. The transition(shrinkage) enthalpy of cross-linked leaflets was increased as compared to the non-cross-linkedleaflets if EDC and NHS were involved in the reactions, resulting in amide type cross-links whichled to highly stabilized and well-organized triple-helices. In contrast, extended cross-links as in GAand BDDGE cross-linked leaflets, resulted in a less tight organization of triple-helices. Theswelling ratio, which was reduced after cross-linking, was not proportional to the Ts, whichindicated that not only the cross-link density (nature and concentration) but the charges on thecollagen determine the swelling as well. In-vitro enzymatic degradation tests demonstrated that thecross-linking methods resulted in leaflets with a comparable stability towards enzymes as GAtreated controls. Extraction of (non)-cross-linked leaflets revealed that cross-linking of porcineaortic heart valve tissue includes mainly collagen, but also elastin, GAGs and likely other proteins.It was shown that these components can be removed from the matrix.Contrary to the leaflets, the aortic wall was incompletely cross-linked as reflected by threedifferent transitions in the thermograms. This resulted in materials which were less resistanttowards enzymes as compared to their leaflet counterparts.


- 167 -

References

1. B.S. Goldman, T.E. David, D.F. Del Rizzo, J. Sever, and J. Bos, "Stentless porcine bioprostheses for aorticvalve replacement", J. Cardiovasc. Surg., 35(Suppl. 1 to No. 6) pp. 105-110 (1994).

2. S. Westaby, H.A. Huysmans, and T.E. David, "Stentless aortic bioprostheses: Compelling data from thesecond international symposium", Ann. Thorac. Surg., 65 pp. 235-240 (1998).

3. D.F. Del Rizzo, B.S. Goldman, and T.E. David, "Aortic valve replacement with a stentless porcinebioprostheses: Multicentre trial", Can. J. Cardiovas., 11(7) pp. 597-603 (1995).





8. T. Chandy, M. Mohanty, A. John, S.B. Rao, R. Sivakumar, C.P. Sharma, and M.S. Valiathan, "Structuralstudies on bovine bioprosthetic tissues and their in vivo calcification: prevention via drug delivery",Biomaterials, 17 pp. 577-585 (1996).



11. R. Zeeman, P.J. Dijkstra, P.B. v. Wachem, M.J.A. v. Luyn, M. Hendriks, P.T. Cahalan, and J. Feijen,"Cross-linking and modification of porcine aortic heart valves", Chapter 8 of this thesis (1998).


13. I.V. Yannas, J.F. Burke, P.L. Gordon, C. Huang, and R.H. Rubenstein, "Design of an artificial skin. II.Control of chemical composition", J. Biomed. Mat. Res., 14 pp. 107-131 (1980).

14. D. Hirsch, J. Drader, T.J. Thomas, F.J. Schoen, J.T. Levy, and R.J. Levy, "Inhibition of calcification ofglutaraldehyde pretreated porcine aortic valve cusps with sodium dodecyl sulfate", J. Biomed. Mat. res., 27pp. 1477-1484 (1993).

15. J. Cladera, J.-L. Rigaud, J. Villaverde, and M. Dunach, "Liposome solubilization and mebrane proteinreconstitution using CHAPS and CHAPSO", Eur. J. Biochem., 243 pp. 798-804 (1997).

16. T. Schroeder and T. Schuerholz, "Bimodal solubilization of phospholipid-cholesterol vesicles in 3-[(3-cholamidopropyl)dimethylammonio]-1-propanesulfonate (CHAPS) solutions. Formation of filamentousand helical microstructures depends on negatively charged lipids", Eur. Biophys. J., 25 pp. 67-73 (1996).

17. Boehringer Mannheim, "Biochemical Catalog", (1998).18. R. Kittelberger, T.J. Neale, K.T. Francky, N.S. Greenhill, and G.J. Gibson, "Cleavage of type VIII

collagen by human neutrophil elastase", Biochim. Biophys. Acta, 1139 pp. 295-299 (1992).19. V.C. Hascall, M. Yanagashita, A. Calabro, R. Midura, J.A. Rada, S. Chakravanti, and J.R. Hassell,

"Isolation and characterization of proteoglycan core protein", in "Extracellular matrix. A practicalapproach", Ed. by M.A. Haralson and J.R. Hassell, Oxford University Press (1995)

20. T.E. Creighton, "Proteins. Structures and molecular properties", W.H. Freeman and Company, New York(1983).

21. G. Blix, "The determination of hexosamines according to Elson and Morgan", Acta Chem. Scandin., 2 pp.467-473 (1948).



Chapter 9

- 168 -

24. N.R. Vyavahare, D. Hirsch, E. Lerner, J.Z. Baskin, R. Zand, F.J. Schoen, and R.J. Levy, "Prevention ofcalcification of glutaraldehyde-crosslinked porcine aortic cusps by ethanol preincubation: Mechanisticstudies of protein structure and water-biomaterial relationships", J. Biomed. Mat. Res., 40 pp. 577-585(1998).

25. F.J. Schoen, H. Harasaki, K.M. Kim, and H.C. Anderson, "Biomaterial-associated calcification: Pathology,mechanisms. and strategies for prevention", J. Biomed. Mat. Res., 22(A1) pp. 11-36 (1988).



28. Y. Noishiki, H. Koyanagi, T. Miyata, and M. Furuse, Bioprosthetic valve, Patent EP 0 306 256 A2 1988.29. E. Khor, A. Wee, T.C. Feng, and D.C.L. Goh, "Glutaraldehyde-fixed biological tissue calcification:

effectiveness of mitigation by DMSO", J. Mat. Sci.: Mat. in Med., 9 pp. 39-45 (1998).30. L.H.H. Olde Damink, P.J. Dijkstra, M.J.A.v. Luyn, P.B. v. Wachem, P. Nieuwenhuis, and J. Feijen,

"Cross-linking of dermal sheep collagen using a water-soluble carbodiimide", Biomaterials, 17(8) pp. 765-774 (1996).

31. R. Zeeman, P.J. Dijkstra, P.B. v. Wachem, M.J.A. v. Luyn, M. Hendriks, P.T. Cahalan, and J. Feijen,"Successive epoxy and carbodiimide cross-linking of dermal sheep collagen", Chapter 6 of this thesis,(1998).

32. R. Zeeman, P.J. Dijkstra, P.B.v. Wachem, M.J.A.v. Luyn, M. Hendriks, P.T. Cahalan, and J. Feijen, "In-vitro degradation of dermal sheep collagen cross-linked with 1,4-butanediol diglycidyl ether", Chapter 5 ofthis thesis, (1998).


34. C.A. Miles, T.V. Burjanadze, and A.J. Bailey, "The kinetics of the thermal denaturation of collagen inunrestrained rat tail tendon determined by differential scanning calorimetry", J. Mol. Biol., 245 pp. 437-446 (1995).




38. A. Veis, "Intact collagen", in "Treatise on collagen. Volume 1. Chemistry of collagen", Ed. by G.N.Ramachandran, Academic Press, London. p. 367-440 (1967)

39. M.T. AmEnde and N.A. Peppas, "Transport of ionizable drugs and proteins in cross-linked PAA andPAA-co-PHEMA hydrogels. I. Polymer characterization", J. Appl. Polym. Sci., 59 pp. 673-685 (1996).

40. N.A. Peppas and A.R. Khare, "Preparation, structure and diffusional behavior of hydrogels in controlledrelease", Adv. Drug Delivery Rev., 11 pp. 1-35 (1993).

41. L. Stryer, "Biochemistry", fourth ed., W.H. Freeman and Company, New York, USA (1995).42. C. Nagorski, D. Opalecky, and F.A. Bettelheim, "A study of collagen-hyaluronic interaction through

swelling in polyacrylamide gels", Res. Comm. Mol. Pathol. Pharma., 89(2) pp. 179-188 (1995).43. J.E. Eastoe, "Composition of collagen and allied proteins", in "Treatise on Collagen. Volume 1. Chemistry

of collagen", Ed. by G.N. Ramachandran, Academic Press, London. p. 1-72 (1967)44. T. Hayashi and Y. Nagai, "Effect of pH on the stability of collagen molecule in solution", J. Biochem., 73

pp. 999-1006 (1973).

In-v vo behav or of cross-l nked porc ne heart valve leaflets and walls

- 169 -

Chapter 10

In vivo behavior of cross-linked porcine aortic leaflets

and walls

Effect of cross-linking method and CHAPS/SDS extraction method

P.B. van Wachem1, L.A. Brouwer1, R. Zeeman2, P.J. Dijkstra2, J. Feijen2,

M. Hendriks3, P.T. Cahalan3, M.J.A. van Luyn1


ABSTRACT

Calcification limits the long term durability of xenograft heart valves cross-linked with glutaraldehyde (GA). In thisstudy, carbodiimide (EDC/NHS) and bisepoxy (BDDGE) cross-linked porcine aortic valve tissue was evaluatedafter subcutaneous implantation in weanling rats. Tissue kept in the storage medium or cross-linked withglutaraldehyde functioned as controls. In a second approach, cellular elements, phospholipids and small solubleproteins, which are known to act as possible nucleation sites for calcification, were extracted from the cross-linkedmatrix using the detergents 3-[(3-cholamidopropyl) dimethylammonio]-1-propanesulfonate (CHAPS) and sodiumdodecylsulfate (SDS).Porcine leaflets cross-linked with EDC/NHS or BDDGE or by a combination of both had comparable or somewhatlower shrinkage temperatures (Ts) than the GA controls. Aortic wall tissue cross-linked with the same methodsresulted in a lower Ts and a higher content of amine groups as the corresponding leaflets. Both (non)-cross-linkedleaflets and aortic walls were implanted in 25 day-old Sprague Dawley rats. After an 8 week-implantation, non-cross-linked leaflets could not be retrieved. Leaflets cross-linked with EDC/NHS and bisepoxy elicit a somewhatlower inflammatory and immune reaction as compared to GA cross-linked controls. The levels of calcium,measured by von Kossa-staining and by quantitative calcium-analysis, were only slightly lower than in the GAtreated leaflets. The non-cross-linked walls were more degraded than the cross-linked walls. Furthermore, thenumbers of lymphocytes and plasma cells in capsules of non-cross-linked walls were clearly the highest, also ascompared to cross-linked leaflets. Cross-linked leaflets calcified to a higher extent (144 - 194 mg/g tissue) thancross-linked wall samples (79 - 113 mg/g tissue), but remarkably, the non-cross-linked walls exhibited the highestcalcium levels (155 mg/g). It is concluded that epoxy cross-linked valve tissue induced an almost similar immuneor foreign body reaction as the GA cross-linked controls. Furthermore, the tendency to calcify of the cross-linkedtissue was only slightly reduced.CHAPS/SDS treatment of BD45EN and GA cross-linked heart valves resulted in reduced calcification, especiallyclear at the edges of both leaflet and aortic wall tissue. Data concerning the immunological reaction varied slightly,but were in the same range as for the GA-cross-linked controls.

Chapter 10

- 170 -

INTRODUCTION

Porcine aortic heart valves are often considered for heart valve replacement in elderly patients [1].To suppress immunogenicity, to reduce the inflammatory reaction [2] and to minimizedegradation, the porcine tissue is cross-linked with glutaraldehyde [3, 4]. Furthermore, in thefabrication process a stent is added to simplify implantation [5]. In spite of this, thesebioprostheses have serious long-term failure rates. The relative rigidity of the stent is thought toenhance primary prosthetic failure [6]. Stentless valves, which contain more aortic wall tissue,have therefore gained more attention during the last decade [7]. Failure of valves also occurs fromcalcification [8, 9] of both leaflet and aortic wall tissue [10]. Though the exact mechanism ofcalcification is still under investigation, the presence of foreign elastin and collagen and of highlyantigenic cellular remnants and soluble proteins appear to play an important role in this process[11-15]. In addition, the cross-linking procedure has a crucial role too. In relation to non-fixedtissue, GA-cross-linked valves calcify to a larger extent [16-19].To reduce the occurrence of calcification, alternative cross-linking methods are investigated.Previously, studies on dermal sheep collagen (DSC) as a model material have been described.Materials cross-linked with 1-ethyl-3-(3-dimethyl aminopropyl) carbodiimide (EDC) in thepresence of N-hydroxysuccinimide (NHS) [20], showed a moderate calcification at week 6 aftersubcutaneous implantation in rats [21]. Because epoxy-cross-linked collagen materials had showna large reduction in calcification compared to GA-cross-linked materials [22, 23], DSC was alsocross-linked with 1,4-butanediol diglycidyl ether (BDDGE), either at pH 9.0 (BD90), or at pH 4.5(BD45) or successively with BDDGE and EDC/NHS (BD45EN) [24, 25]. After subcutaneousimplantation in rats, the BDDGE cross-linked DSC materials were found to be biocompatible,without induction of calcification [26].Porcine aortic heart valves were also cross-linked with EDC/NHS, BD90, BD45 or BD45EN [27,28]. The increase in shrinkage temperature was comparable to GA cross-linked tissue [27], but theamount of amine groups after cross-linking was highly dependent on the method applied. In vitro-degradation experiments revealed that epoxy cross-linked leaflets and walls had a similar stabilitytowards enzymatic degradation as the glutaraldehyde-treated controls [28].A second strategy to mitigate calcification is the removal of antigenic and cellular remnants fromthe cross-linked matrix. An extraction procedure with 3-[(3-cholamidopropyl) dimethylammonio]-1-propanesulfonate (CHAPS) and sodium dodecyl sulfate (SDS) was utilized. SDS, which is ananionic detergent, can be applied to extract various components such as lipids and proteins [12,29, 30]. A reduction in calcification was obtained after SDS treatment of glutaraldehyde cross-linked valves. CHAPS, which is a non-denaturing zwitterionic detergent [31, 32], was specificallydesigned for membrane biochemistry and is often employed for protein and phospholipidsolubilization [32]. In an effort to develop a biologically active, non-cross-linked bioprostheticvalve, Vesely et al. developd a cellular extraction process with CHAPSO, a detergent almostsimilar to CHAPS. They found that cell extraction significantly reduced the propensity of thematerial to calcify in-vivo [33].In this paper, the biocompatibility of EDC/NHS, BD90, BD45 and BD45EN cross-linked porcineaortic heart valves was evaluated by implantation of both leaflets and aortic wall tissue in 25 days


- 171 -

old rats. Samples were explanted after 8 weeks and cellular responses and calcification levels wereevaluated. The effect of the cross-linking procedure in relation to the standard GA process wasstudied. In addition, non-cross-linked tissue was used as a control material. Secondly,glutaraldehyde and the successive epoxy and carbodiimide cross-linked valves were subsequentlytreated with CHAPS and SDS. The effect of this extraction step on the calcification behavior wasevaluated.


Tissue preparationFresh porcine aortic heart valves (leaflets and wall) were selected and dissected as describedpreviously [27].

Cross-linkingFreshAfter the overnight immersion in 0.01 M HEPES buffer at 4°C, the valve was transferred to thestorage medium, a 0.01 M HEPES buffer (pH 7.4) containing 20 % (v/v) isopropanol (IPA, z.A.Merck, Darmstadt, Germany).Glutaraldehyde (GA)A valve (dry weight approximately 1.5 g) was immersed in 100 ml of a 0.01 M N-[2-hydroxyethyl]piperazine-N’-[2-ethanesulfonic acid] (HEPES free acid, Sigma-Chemical, St. Louis,USA) buffered (pH 7.4) solution containing 0.2 wt% GA (from a 25 % solution, Merck,Darmstadt, Germany). The reaction was allowed to proceed for 24 h at 20 °C, after which thevalve was stored in a 0.01 M HEPES buffered solution (pH 7.4) containing 0.2 wt % GA.1-ethyl-3-(3-dimethyl aminopropyl) carbodiimide. HCl (EDC) and N-hydroxysuccinimide (NHS)A valve was immersed in 100 ml of a 0.1 M 2-[morpholino]ethanesulfonic acid (MES, Merck,Darmstadt, Germany) buffered solution (pH 5.5) containing 1.15 g EDC (z.S., Merck-Suchardt,Hohenbrunn, Germany) and 0.14 g NHS (z.S., Merck-Suchardt, Hohenbrunn, Germany). Cross-linking was carried out for 24 h at 20 °C. After cross-linking, the valves were washed with waterbefore storage in a 0.01 M HEPES buffered solution containing 20 (v/v) % IPA (pH 7.4).1,4-Butanediol diglycidyl ether (pH 4.5 or pH 9.0)A valve was immersed in 100 ml of a 0.025 M disodium tetraborate decahydrate (pH 9.0) or 0.1M MES (pH 4.5) buffered solution containing 4 wt% 1,4-butanediol diglycidyl ether (BDDGE,Fluka, Buchs, Switzerland). Cross-linking was allowed to proceed for 3 d at pH 9.0 and for 6 d atpH 4.5 at 20 °C. After cross-linking, the valves were washed with water before storage in a 0.01M HEPES buffered solution containing 20 (v/v) % IPA (pH 7.4).Successive epoxy and carbodiimide cross-linking (BD45EN)A valve was immersed in 100 ml of a 0.1 MES buffered solution (pH 4.5) containing 4 g BDDGE.Cross-linking was allowed to proceed for 6 d at 20 °C, after which the valve was washed 2 timesfor 30 min with 0.01 M HEPES (pH 7.4) buffer. A valve was immersed in 100 ml of a 0.1 M MESbuffered solution (pH 5.5) containing 1.15 g EDC and 0.28 g NHS (ratio EDC:NHS = 5:2).Cross-linking was allowed to proceed for 24 h at 20 °C, after which the valve was washed with

Chapter 10

- 172 -

deionized water before storage in a 0.01 M HEPES buffered solution containing 20 (v/v) % IPA(pH 7.4).

CHAPS/SDS treatmentGlutaraldehyde and BD45EN cross-linked valves were treated with 3-[(3-cholamidopropyl)dimethylammonio]-1-propanesulfonate (CHAPS) and sodium dodecyl sulfate (SDS) following theprocedure as described previously [28].

Extent of cross-linkingThe extent of cross-linking of tissue samples was related to the increase in shrinkage temperature(Ts). The Ts of (non)-cross-linked leaflets in the hydrated state was determined with differentialscanning calorimetry [27]. The content of amine groups of the tissue before and after cross-linkingwas determined spectrophotometrically after reaction of the primary amine groups with 2,4,6-trinitrobenzenesulfonic acid (TNBS) and subsequent hydrolysis of the sample [27], and isexpressed as the percentage [%] of the initial amount of amine groups.

Animal modelOf each valve, three discs (8 mm diameter) were punched from the aortic wall and the leaflets. Allmaterials were washed 3 times 2 min with normal saline (0.9 %, NPBI, Emmercompascuum, TheNetherlands) prior to implantation. National Institutes of Health guidelines for the care and use oflaboratory animals (NIH 85-23 Rev. 1985) were observed. Male, 25 d old Sprague-Dawley rats(CD-strain) were used. Their backs were shaved and disinfected, a mid-line incision was made inthe skin and two subcutaneous pockets were created on each side of the spine. One sample wasinserted per pocket and the skin was closed with one suture. Samples were explanted after 8weeks.

MicroscopyAfter 8 weeks, samples were carefully explanted and cut into two pieces. From one half, thesurrounding capsule was removed; it was stored in 0.01 M HEPES/ 20 % IPA for quantitativecalcium analysis. The other half was immediately immersion-fixed in 2% (v/v) GA in 0.1 mol PBS(pH 7.4) and, after at least 24 hr at 4 °C and dehydration in graded alcohols, a mid part wasembedded in glycol methacrylate (GMA) for light microscopic (LM) evaluation. As controls, non-implanted, EDC/NHS, and BD45 cross-linked leaflets and walls were also fixed in 2% GA andembedded in GMA. Sections (2µm) of GMA embedded materials were routinely stained withtoluidin blue. To evaluate calcification, von Kossa-staining was used.Numbers of lymphocytes and the extent of calcification per sample were estimated by threepersons.

Quantitative calcium analysisBoth non-implanted cross-linked valves and explanted samples, immersed in 10mM HEPES/20%IPA, were frozen at -70 °C for 60 min, after which they were freeze-dried for 12 h. The extent ofcalcification was determined by inductively coupled plasma atomic emission spectroscopy on aPerkin-Elmer radial Optima 3000 (DSM-Research, PAN-AN, Geleen, The Netherlands).


- 173 -

Explanted and freeze-dried samples (2- 30 mg) were hydrolyzed in 15 ml of 6 M HCl for 24 h,followed by addition of 10 ml demineralized water. A blank was prepared by omitting the sample.The solution obtained was nebulized in an inductively coupled argon plasma. The signal intensityof Ca with the aid of an atomic emission spectrometer using emission lines of 317.933 nm and422.673 nm was determined. The concentration of Ca2+ per dry weight of tissue was calculatedusing a Ca2+ calibration curve.

RESULTS

Cross-linking of porcine aortic heart valves was successful as reflected by an increase in shrinkagetemperature (Table I).

Table IThe shrinkage temperatures (Ts) and amine group content of (non) cross-linked leaflets

Material leaflets wallTs [°C] Amine groups

[%]Ts [°C] Amine groups

[%]FreshGAEDC/NHSBD90BD45BD45EN

61.685.085.981.472.582.0

10032.660.858.595.456.4

61.183.381.177.172.679.8

10060.584.260.994.283.7

The highest Ts values were obtained after GA or EDC/NHS cross-linking, whereas BD45 had aTs of 73 °C. GA cross-linked leaflets had the lowest content of amine groups (33 %), while hardlyany change in amine groups was encountered in BD45, due to cross-linking via the carboxylic acidgroups. In general, The Ts of the aortic wall is somewhat lower and the amine group contenthigher than of the corresponding leaflets.

Non-implanted controlsSome macroscopic differences were visible. The non-cross-linked, the EDC/NHS and BD90cross-linked tissues were white. GA cross-linked tissue had a yellow color, and BD45 andBD45EN had a slightly yellowish appearance.The original morphology of the leaflets, with fibrosa (aorta side), spongiosa and ventricularis(cardiac side) [34], and of the wall, with intima (luminal side), media and adventitia, were easilyrecognized at the light microscopical (LM) level with toluidin blue. Cells, i.e. fibroblasts, withround (degenerated) nuclei, were most obvious in the fibrosa of non-treated leaflets (figure 1a).The spongiosa was less stained and endothelial cells (EC) were not obvious. EDC/NHS and BD45(figure 1b) leaflets had similar morphologies, although fibroblasts were preserved in a lessdegenerated way. Furthermore, higher numbers of fibroblasts were present and endothelial cellswere more obvious.Of the wall, the media with its thick layers of alternating smooth muscle cells and extracellularmatrix (mainly elastin) was most obvious. The smooth muscle cells of non-cross-linked wall

Chapter 10

- 174 -

samples showed round nuclei and a regular alternation with elastin (figure 1c). Only very vagueremnants of endothelial cells were present at the luminal side on top of a thick wavingcollagenous/elastinous subendothelium. In general, the elastin looked irregular. At the other sideof the media remnants of the almost fully removed adventitia were recognized from the looseconnective tissue.

Figure 1. Non-implanted controls (made at 10x)a: Fresh leaflet, showing the fibrosa (F) with fibroblasts (the black dots are nuclei) at the left hand side.

Fibroblasts are hardly recognized in the light-stained spongiosa (S) and the ventricularis (V).b: BD45 cross-linked leaflet (a thinner part). Fibroblasts are observed all through the three layers

(F=fibrosa, S=spongiosa, V=ventricularis). Some endothelial cell-ghosts are pointed out (arrows).c: Fresh wall, showing the clear alternation of dark gray-stained smooth muscle cells (with dark-stained

nuclei) and light-stained extracellular matrix of the media, and the thick collagenous/elasticsubendothelium (SE) without endothelial cells.

d: BD45 cross-linked wall, which is lighter-stained than the non-cross-linked wall. A row of endothelialcell-ghosts (arrows) is present on the subendothelium (SE).

The EDC/NHS cross-linked wall showed a more regularly preserved luminal side and adventitia.Endothelial cells were only present as lightly stained remnants. The media showed an irregularpreservation of smooth muscle cells, with more stretched cells at the luminal side, more shrunkencells at the adventitia side and both morphologies, generally degenerated, in between. Theirregularly preserved elastin seemed to dominate. Vacuoles appeared to be present near the manyshrunken nuclei.


- 175 -

The BD45 cross-linked wall (figure 1d) was generally light-stained. The central part appearedlightest and most degenerated, but still some stretched and probably well-preserved smooth musclecells were present. The luminal side showed a row of endothelial cell-ghosts on the non-wavingsubendothelium.

With all non-implanted samples, von Kossa-staining showed that calcification was not presentbefore implantation.

ImplantsAfter 8 weeks of implantation all samples were retrieved except for the fresh leaflets. Tissueexplanted from the implantation site still showed effects of the former presence of the non-cross-linked leaflets, e.g. with several areas or accumulations with lymphocytes and plasma cells. Thenon-fixed walls looked macroscopically the most degenerated of all wall samples. Microscopically,non-fixed walls (figure 2a) showed a fibrous capsule with cellular ingrowth consisting of giantcells, fibroblasts, but most importantly, large numbers and accumulations of lymphocytes (TableII). Plasma cells were dominantly present (figure 2b and 2c). Calcification determined via vonKossa (figure 2d) varied from intense at the rims (the punched sides and the adventitia and luminalside) to intense all over. In general, it was probably mostly related to the extracellular matrix, butalso to smooth muscle cells. Precise identification was difficult, at the least intense areas, wherefocal nucleation occurred, because of the very light-staining of the tissue with von Kossa. Proof ofextracellular matrix related calcification was found in the calcifying elastin from lamina elastica ofan artery in the former adventitia. The calcium analysis exhibited a mean of 155 mg calcium per gtissue.

Chapter 10

- 176 -

Figure 2. The fresh wall explanted after 8 weeksa: (overview made at 4x): Showing the wall with the former EC-side at the left-hand side and the thick

media (M). At both sides, fibrous capsules are observed, the larger one at the former adventitia (A)side. At both rims, white areas (arrows) indicate calcification. (Frameworks: figures 2b and c)

b: (made at 20x): Showing a detail at the adventitia side. The capsule consists of a giant (G) cell layer,a layer with blood vessels (V) with lymphocyte accumulations in between, and a fibrous (F) layer.Arrow: a former blood vessel, which is mainly recognized from its lamina elastica. (Framework:figure 2c)

c: (made at 100x): Showing a further detail with non-stained calcification (C), giant cells (G), a bloodvessel (V), and lymphocytes. Most of these concern plasma cells (arrows).

d: (made at 20x): Showing the same detail as figure 2b. Von Kossa staining is used to point outcalcification, which is present in thick layers at the interface, getting a more focal appearance higherup. The lamina elastica (arrow) of the former blood vessel is also calcified.

Glutaraldehyde (GA) cross-linking resulted in severely calcified leaflets and a value of 194 mg/gtissue was found. Calcification was related to cells, i.e. fibroblasts, and the extracellular matrix(figures 3a and b). Especially in the ventricularis it was present as a clear rim. In general,lymphocyte presence was clearly lower than with non-cross-linked walls. The GA cross-linkedwalls attracted less lymphocytes than the leaflets. Ingrowth hardly occurred and some giant cellswere observed in the capsule, but not at the interface. Calcification, was lower than with theleaflets and was related to both the extracellular matrix and smooth muscle cells, primarily in therims. The calcium level of 113 mg/g of the GA cross-linked wall was lower than for the non-cross-linked wall.Leaflets cross-linked with EDC and NHS were largely surrounded by mature fibrous (figure 3c)capsules with a low lymphocyte activity (Table II). Calcification was not significantly different


- 177 -

from GA cross-linked leaflets and observed all over. EDC/NHS cross-linked walls showed asimilar calcification as the GA walls. It was mostly diffuse and intense at the rims and appearedboth extracellular matrix and smooth muscle cell related (Figures 3c and d). Areas of calcifiedtissue surrounded by giant cells were present in the capsule. In general, only a little giant cellformation was observed. In contrast to the leaflets, EDC/NHS cross-linked walls showed muchmore lymphocytes, both in a small rim all along the interface, but also in small accumulations in orjust outside the fibrous capsule.

Table IINumbers of lymphocytes and plasma cells, von Kossa staining and quantitative calcium

levels of (non) cross-linked leaflets and wallsCross-linkingmethod

Sample Number oflymphocytes andplasma cells 1.)

von Kossastaining 2.)

Calcium content[mg/ g dry tissue]

non-cross-linked leafletwall

n.r.3

n.r.3

n.r.154.6 ± 21.6

GA leafletwall

21

32

194.2 ± 15.7112.6 ± 5.0

EDC/NHS leafletwall

12

32

171.2 ± 29.6 96.4 ± 8.7

BD90 leafletwall

11

32

185.5 ± 18.6100.9 ± 6.4

BD45 leafletwall

21

22

143.9 ± 5.2 78.6 ± 16.5

BD45EN leafletwall

11

22

147.5 ± 44.8 91.1 ± 10.8

n.r. = not retrieved1.) number of cells: 1 = some or dispersed presence of lymphocytes and plasma cells, 2 = rows or smallaccumulations of lymphocytes and plasma cells, 3 = large numbers and accumulations of lymphocytes and plasmacells2.) von Kossa staining: 2 = little/ moderate to 3 = intense/severe calcification.

Leaflets cross-linked with BDDGE at pH 9.0 (BD90) had a dispersed presence and smallaccumulations of lymphocytes were observed. Plasma cells were often amongst them.Lymphocytes were also observed in ingrowing tissue in the spongiosa. Calcification was found allover, most intensely at the ventricularis (figures 4a and b). Calcium analysis showed again asignificantly higher calcification of the leaflets (186 mg/g tissue) compared to the walls (101 mg/gtissue). With the latter it was associated with extracellular matrix and smooth muscle cells (figures4c and d) and again dominated at the rims. Lymphocytes were disperse or in small accumulationspresent in the fibrous capsule at the former adventitia side, in which giant cells were also observed.

Chapter 10

- 178 -

Figure 3. GA or EDC/NHS cross-linked wall samplesa: (made at 10x ): The GA cross-linked wall shows electron-dense calcified collagen. Often, the typical

waving morphology of calcified collagen was recognized (Framework: fig. 3b).b: (made at 40x): Detail which shows several examples of fibroblast calcification (large arrows) and of

collagen calcification (small arrows).c: (made at 10x) The adventitia-side of an EDC/NHS cross-linked wall showing calcification (C) at the

rim. The capsule contained calcified areas (arrows), and is surrounded by giant cells. Lymphocytesand blood vessels are also present. V: a former blood vessel present in the media.

d: (made at 10x) The von Kossa-stained counterpart of 3c. It shows the position of the calcified chips inthe capsule, thus outside the wall.

Leaflets cross-linked with BD45 were encapsulated by fibrous tissue, showing a dispersedappearance, of small accumulations of lymphocytes and plasma cells. Only a few giant cells wereobserved. Calcification was present in the spongiosa or all over, and was related to both fibroblastsand the extracellular matrix. Calcium analysis of the hydrolyzed samples revealed the lowest datafor both the leaflets (144 mg/g tissue) and the walls (79 mg/g tissue). Calcification wasconcentrated at the rims, mainly at the adventitia and luminal side. The tissue reactions to wallswere similar to those to the leaflets, and slightly more giant cells were found at the formeradventitia.


- 179 -

Figure 4. BD90 cross-linked tissuea: (made at 4x). Overview of the leaflet, the fibrosa (F) at the top and the ventricularis at the bottom.

Ingrowth (arrow) occurred predominantly at the spongiosa. Calcification is easily recognized justabove and below the ingrowing tissue.

b: (made at 4x). The von Kossa-stained counterpart of figure 4a. It further identified calcification allthrough the leaflet, although the ventricularis (V) represents the most intense layer.

c: (made at 20x). The wall showed dark-stained focal calcification all over, but it is dominated at therims. Here, a central part is shown (Framework: figure 4d)

d: (made at 100x). A detail shows that the calcification primarily occurred in the extracellular matrix(larger brown/black dots). However, nucleation appears also to be related to smooth muscle cells(SMC), as indicated by the light-brown staining of small dots (arrows).

The leaflets cross-linked with BD45EN were surrounded by mature fibrous capsules containingonly some single lymphocytes and plasma cells (Table II). Giant cells were present at several sitesalong the interface. Fibrous ingrowth was observed in the spongiosa. Calcification varied fromintense all over the ventricularis and the spongiosa to especially one site of the ventricularis. Itconcerned both fibroblasts, which was obvious at less concentrated areas, and the extracellularmatrix. Two of the BD45EN wall samples showed not much evidence of lymphocytes in e.g. themature fibrous capsule at the former luminal side. One sample contained a lining of lymphocytesand a few accumulations. At the adventitia, collagen had a compact morphology with dispersedpresence of giant cells. Giant cells were sometimes present at the punched sides, concomitant with

Chapter 10

- 180 -

a little ingrowth. Calcification was always intense at the luminal, adventitia or punched sides, andrelated to both the extracellular matrix and smooth muscle cells.

In general, epoxy cross-linked valves did not show a considerable reduction in calcium levels ifcompared to the well accepted GA cross-linked controls (Table II). The number of lymphocytesand plasma cells were somewhat lower as obtained in GA cross-linked counterparts.

CHAPS/SDS

The effect of the CHAPS/SDS extraction treatment on the shrinkage temperature (Ts), the contentof amine groups and the swelling of the tissue was rather small as already shown previously [28].

Non implanted controlsThe GA cross-linked leaflet clearly showed fibroblasts all through the three layers (figure 5a).They had a degenerated morphology with pyknotic (reduction in size) nuclei in the spongiosa, andmore swollen cells at the outer rims. Few swollen remnants of endothelial cells were observed.The wall with the smooth muscle cell-containing media looked less stretched than the non-cross-linked wall. Furthermore, the collagenous endothelium was thicker and clearly contained swollenfibroblasts and some EC (figure 5e).GA-cross-linking in combination with CHAPS/SDS resulted in a remarkable phenomenon in thefibrosa of the leaflet (figures 5c and d). Some of the cells were swollen and showed ‘dark-stained’granules, which were probably not of nuclear origin. Otherwise, this leaflet looked like the GAleaflet. The morphology of the GA-CHAPS treated walls was comparable to the GA walls (figure5f).BD45EN-CHAPS treatment of leaflets resulted in a 'washed-out' appearance in large parts of thespongiosa and fibrosa, with discontinuities between pyknotic cells and extracellular matrix (figure5b). Cell fragments as with GA-CHAPS treated leaflets were not observed. The ventricularislooked a little amorphous. Swollen endothelial cells were hardly present. The BD45EN-CHAPSwall had a limited 'washed-out' appearance only at the subendothelium side, which resulted in aless clear transition with the first layers of the media and pyknotic cells and vacuoles (figure 5g).Endothelial cells were not present. The following layer contained irregular elastin and vacuoles,but the media looked more or less normal. The tissue of the adventitia hardly showed anystructure.


- 181 -

Figure 5. Non-implanted materialsa: (10x) GA cross-linked leaflet, showing the ventricularis (V), spongiosa (S) and fibrosa (F). Small dots

represent nuclei of the fibroblasts, dispersed in the tissue. Endothelial cells were not observed.b: (10x) BD45EN-CHAPS treated leaflet, showing 'washed-out' areas in spongiosa and fibrosa (arrows).

Fibroblasts look pyknotic. Some endothelial cell-ghosts are present (small arrows). (V = ventricularis,S = spongiosa, F = fibrosa).

c: (10x) GA-CHAPS treated leaflet, showing a similar appearance as the GA leaflet, except for theswollen cells with ‘dark-stained’ granules in the fibrosa (framework). (V = ventricularis, S =spongiosa, F = fibrosa).

d: (40x) GA-CHAPS treated leaflet, with detail of the swollen cells with‘dark-stained’ granules (arrows).S = spongiosa

e: (10x) GA cross-linked wall, with the media showing a clear alternation of gray layers of smooth musclecells (SMC) and non-stained extracellular matrix (ECM, mainly recognized from the non-stainedelastin). A few swollen endothelial cells (EC, arrows) are recognized. (SE = subendothelium).

f: (10x) GA-CHAPS treated wall, showing a slight crimp in the first layers at the luminal side; otherwisethis wall has a morphology which is comparable to the GA cross-linked wall. SE = subendothelium.

g: (10x) BD45EN-CHAPS treated wall, showing a limited 'washed-out' appearance only at the luminalside, resulting in a less clear transition from the subendothelium (SE) to the first layers of the media,with pyknotic cells, vacuoles and an irregular following layer. Endothelial cells are not present. Atdeeper layers, the media looks normal.

Implants (Table III)GA-cross-linked leaflets (figures 6a and b) were surrounded by small fibrous capsules with fewsmall giant cell layers. Only a few lymphocytes and plasma cells were present. Sometimes a little

Chapter 10

- 182 -

ingrowth was observed in the spongiosa. Calcification was present all over the leaflets but itappears mostly related to the cells. Still, the extracellular matrix was found to calcify as well.Similarly, the GA cross-linked walls (figures 6c and d) showed low amounts of lymphocytes andplasma cells. The calcification was predominantly localized at the rims. At the endothelial cell sidethe intense calcified rim represented the subendothelium and first layers of the media. Calciumdeposits were mostly related to the extracellular matrix, especially at the former adventitia side.

Table IIIHistological evaluation and the calcium content determined by ICP-AES of cross-linked and

CHAPS/SDS treated leaflets and wallsCross-linkingmethod

Sample Number oflymphocytes andplasma cells 1.)

von Kossastaining 2.)

Calcium content[mg/ g dry tissue]

GA leafletwall

01

22

220.3 ± 16.1120.4 ± 12.9

GA-CHAPS leafletwall

00

12

160.1 ± 29.1108.5 ± 16.8

BD45EN-CHAPS

leafletwall

11

12

--

n.r. = not retrieved1.) 1 = some or dispersed presence of lymphocytes and plasma cells, 2 = rows or small accumulations oflymphocytes and plasma cells, 3 = large numbers and accumulations of lymphocytes and plasma cells2.) von Kossa: 2 = little/ moderate to 3 = intense/severe calcification.

The additional CHAPS/SDS treatment resulted in a material which gave a tissue reaction, similarto GA cross-linked tissue. With both the leaflets and the walls, capsules and ingrowth showedsimilar numbers of lymphocytes or giant cells. Quantitatively, calcification was reduced for theleaflets from 220 to 160 mg/g tissue, and it was especially located at the edges of the fibrosa.Qualitatively (figures 6e and 6f) it did not look different, showing both cell and extracellularmatrix related staining. The walls showed a similar decrease in calcium levels, which concernedmainly the first layers at the former luminal side (figures 6g and h). A small but hardly significantdecline in calcium from 121 to 109 mg/g tissue was measured.Leaflets cross-linked with BD45EN in combination with CHAPS/SDS were surrounded by maturefibrous capsules with only few giant cells at some corners. Lymphocytes were not obvious.Calcification occurred predominantly in the spongiosa, and was both extracellular matrix andfibroblast related. The BD45EN-CHAPS treated walls usually showed a clear fibrous capsule atthe former luminal side, and a clear giant cell layer with a dispersed presence of lymphocytes at theformer adventitia side. At the punched sides some ingrowth of fibroblasts occurred, although mostcells were lymphocytes. Calcification was both related to the extracellular matrix and smoothmuscle cells, but the relative contribution of the latter was more than observed in the previousmaterials. Furthermore, the outer edges of luminal and adventitia side hardly calcified.


- 183 -

Figure 6. GA cross-linked tissuea: (20x) The fibrosa-side of a leaflet surrounded by a thin capsule (C) without giant cells. White spots in

the leaflet tissue represent calcification.b: (20x) The von Kossa-stained counterpart of 6a. Calcification was both related with fibroblasts (a few

pointed out; small arrows) and extracellular matrix (ECM, large arrows). C: the capsule.c: (10x) The adventitia-side of the wall contains a large former blood vessel (V), present next to the

fibrous capsule (C). SMC of the media (M) are observed at the right-hand side.d: (10x) The von Kossa-stained counterpart of figure 3c. Calcification is associated with both SMC and

ECM, of which elastin calcifies with certainty, as proven from the calcifying lamina elastica (arrows)of the former blood vessel. C: the capsule.

GA-CHAPS-cross-linked tissuee: (10x) The complete cross-section of the leaflet is surrounded by a thin capsule (C) without giant cells.

Calcification, here only recognized by a few white spots (arrow), is markedly reduced as compared toleaflet tissue cross-linked with GA only (compare with figure 3a). C: the capsule.

f: (20x) The von Kossa-stained counterpart of figure 4a, identifying the reduced calcification, bothfibroblast- and ECM-related (compare with figure 3b). C: the capsule.

g: (10x) The former luminal side of the wall shows an absence of calcification in the first (1) layers of themedia. In the second (2) layer an irregularity of white spots represents calcification. C: the fibrouscapsule.

h: (20x) The von Kossa-stained counterpart of figure 4c identifies only a few calcifying ECM-relatedspots (arrows) in the first layer. In the second layer, calcification was clearly associated with bothSMC (smaller dotted areas) and ECM (larger waving areas). C: the capsule.

Chapter 10

- 184 -

DISCUSSION

The long-term durability of xenograft heart valves is limited due to calcification of the tissue,which leads to stiffening, tearing and rupture of the leaflet [11, 16, 17]. The mechanism ofcalcification is still under investigation, but researchers believe that the glutaraldehyde (GA) cross-linking gives a major contribution to this process [4, 13, 18, 19, 35]. Therefore, other cross-linkersare applied on collagen-based materials. The biocompatibility and the tendency to calcify ofporcine aortic valve tissue cross-linked with a water soluble carbodiimide (EDC/NHS), thebisepoxide 1,4-butanediol diglycidyl ether (BDDGE) or with a successive epoxy and carbodiimidemethod were evaluated after 8 weeks of subcutaneous implantation in weanling rats. Non cross-linked, and GA cross-linked tissue functioned as controls. In a second approach, the effect of awashing/extraction step on GA or successive epoxy and carbodiimide (BD45EN) cross-linkedsamples with the detergents 3-[(3-cholamidopropyl) dimethylammonio]-1-propanesulfonate(CHAPS) and sodium dodecylsulfate (SDS) on the biocompatibility and the in-vivo calcificationwas evaluated.Porcine leaflets cross-linked with EDC and NHS resulted in a material with a comparable Ts as theGA controls, but in a higher content of amine groups. BDDGE cross-linking at pH 9.0 resulted ina Ts of 81 °C and a large reduction in amine groups, which indicates that analogously to GAcross-linking, bonds were formed between the (hydroxy)lysine residues. In contrast, the leastcross-linked material was observed if BDDGE cross-linking was carried out at pH 4.5 (BD45, Ts= 73 °C). Hardly any change in amine groups was observed, indicating that ester-type cross-linkswere formed [24]. Finally, an additional EDC/NHS cross-linking step performed on BD45 resultedin a material (BD45EN) with a higher Ts and a lower content of amine groups, which means thatboth ester and amide type crosslinks were introduced in the matrix.Penetration of the crosslinking agents into the aortic wall occurred in an irregular way, which ledto an inhomogeneously cross-linked wall [27] as reflected by the somewhat lower Ts and thehigher content of amine groups as compared to the corresponding leaflets (table I). This irregularpenetration was correlated with the lightest-stained and most degenerated central part of the BD45wall (figure 1b). The irregular cross-linking may also result in irregular shrinkage of the tissue.Non-cross-linked walls are more subject to shrinkage than cross-linked walls, which may explainthe irregularities of the elastin in this material (figure 1d). The irregular shrinkage of the tissue isalso attributed to the storage solution which contained iso-propanol and which results indehydration of collagen and consequently in tightening of the collagen fibrils [36]. Due to thelower swelling after cross-linking, the effect of dehydration by iso-propanol thus inducing irregularshrinkage of the tissue, will be less pronounced in cross-linked tissue [28].Generally, cross-linking resulted in an increase in flexural and bending stiffness of both the walland the leaflet, which agrees with the results obtained in GA cross-linked bovine pericardium orporcine aortic heart valves [37].After an 8-week implantation, non-cross-linked leaflets were not retrieved. Furthermore, freshwalls were clearly more degenerated than the cross-linked walls.The highest number of lymphocytes and plasma cells were obtained in the capsules around non-cross-linked walls. The resulting cellular reaction is, though only evaluated at week 8, more severe


- 185 -

than the usual inflammation/foreign body reaction to biomaterials and antigen presentation early onresulted in an additional immune reaction with activation of T- and B-lymphocytes, as proven fromthe presence at week 8 of accumulations of lymphocytes, amongst which plasma cells [38].However, this immune responses are noncontributory to calcification or valve deterioration [39].Gong et al. demonstrated that although glycerol treated bovine pericardium had a higherimmunogenicity , and exhibited a greater inflammatory reaction than GA fixed counterparts, thelatter had considerably higher calcium levels (197 vs. 0.7 mg/g) [2].Cross-linking resulted in masking of antigenicity, as reflected by a lower number of lymphocytesand plasma cells as compared to non-cross-linked tissue (Table II). However this reduction wasrather small, and since lymphocytes and plasma cells were still observed around the cross-linkedwalls, basically the same immune reactions occurred as observed with fresh walls. Lowest numbersof lymphocytes and plasma cells were found with EDC/NHS and BD45EN cross-linked leaflets,but the highest number of cells was found with the EDC/NHS cross-linked wall. Hence, norelationship between the cross-linking chemistry and its resulting tissue characteristics and theinflammatory/ foreign body reaction was detected. Since GA cross-linked porcine aortic heartvalves are well-accepted for clinical use, EDC/NHS and BDDGE cross-linking resulted inbioprostheses which may also be clinically acceptable.Calcification was measured by von Kossa-staining and by quantitative calcium analysis, and bothmeasurements proved always positive. Von Kossa-staining relates the calcification with areas inthe tissue, whereas calcium analysis give a quantification of it. Corresponding results between bothmethods were obtained. However, calcium analysis imparted that the GA cross-linked leafletsproved to be the most calcified tissue, whereas the BD45 cross-linked walls were the leastcalcified samples (Table II). Calcification is a phenomenon, which clearly occurred duringimplantation, since it was not found before implantation. The exact onset is unknown, but it mayhave been started in the first week of implantation [40]. Because of the relatively wide time span,week 8 is an inappropriate period to identify the origins of calcification. In general, calcification ofboth leaflets and wall tissue was related to the extracellular matrix (collagen, elastin and probablymatrix proteins) and with tissue-specific cells (fibroblasts, smooth muscle cells). This agrees withother reports that showed that calcification was associated with connective tissue cells, membraneand cell fragments as well as collagen and elastin [40-43].In general, BDDGE and EDC/NHS cross-linking of porcine aortic heart valves resulted in similardata as the clinically well-accepted GA fixed tissue. Calcification of epoxy and EDC/NHS cross-linked heart valve tissue is in contrast with previous results in which BDDGE or EDC/NHS cross-linked dermal sheep collagen (DSC) resulted in biocompatible materials which did show no orhardly any calcification [26, 44]. However, DSC is an entirely different material and contains nocells and other proteins than collagen. Furthermore, it gives the opportunity for good ingrowth. Inthat view it is interesting to state that calcification was not observed where ingrowth occurred,which suggests that ingrowth of host tissue prevented calcification. Furthermore, calcification atthe rims indicated that diffusion, and consequently concentration of calcium and phosphate ions isimportant as well.Cross-linked leaflets with calcium levels between 150 and 200 mg/g dry tissue were more calcifiedthan the corresponding walls (80 to 120 mg Ca2+/g tissue), which agrees with other reports [16].The relatively higher calcification of leaflets may be related to its relatively higher content of

Chapter 10

- 186 -

collagen, as compared to the walls (58 versus 19 %) [27]. Surprisingly, the non cross-linked wallsshowed the most severe calcification, probably due to the presence of many foreign cell remnants.Thus far, compared to non-fixed tissue, it was stated that GA cross-linked valves calcify to alarger extent [16, 18, 19]. It is not clear whether or not fresh leaflets were degenerated bycalcification or by other processes.

No correlation was observed between the cross-linking method and the extent of calcification. Theintroduction of either zero-length amide crosslinks by EDC/NHS or the 'extended' crosslinks byGA or BDDGE did not affect calcium deposition. Furthermore, both the materials which werecross-linked with BDDGE either via the carboxylic acid groups (BD45) or via the amine groups(BD90) did not show a significant change in calcification. These results show that the cross-linkingmethod itself is not the only determinant which causes calcification. Other factors will contributeto the formation of calcium deposits on the valve tissue.

CHAPS/SDSPorcine aortic heart valves cross-linked with either GA or with BD45EN were treated with 3-[(3-cholamidopropyl) dimethylammonio]-1-propanesulfonate (CHAPS) and sodium dodecylsulfate(SDS) in order to remove cellular elements, phopholipids and small soluble proteins [29, 31, 32].This treatment increased the Ts a little and it slightly reduced the percentage of amine groups asshown previously [28].GA cross-linked tissue which was extracted with CHAPS/SDS showed that both the leaflets andthe walls had a comparable morphology to GA cross-linked tissue itself, except for the swollencells with ‘dark-stained’ granules in the fibrosa. This partly agrees with observations ofFlomenbaum et al. [45], who treated GA cross-linked leaflets with SDS and found no definiteincremental microscopic changes. It may be that the observed deviant cell morphology in thefibrosa is the result of the additional CHAPS treatment.When BD45EN cross-linked valves were treated with CHAPS and SDS, especially the leaflets hada ‘washed-out’ appearance. This agrees with the loss of about 10 % of initial weight afterCHAPS/SDS treatment of cross-linked leaflets [28].CHAPS/SDS-treatment reduced the antigenicity of both GA and BD45EN cross-linked walls.Thus highly antigenic components appeared to have been washed out from the walls. Sinceantigenicity was not reduced with GA-CHAPS or BD45EN-CHAPS treated leaflets, this treatmentwas not effective for those. This might be caused by the presence of less antigenic substances inthe leaflets, because it contains a higher content of structural proteins [27] which are lessantigenic. Moreover, the leaflets were cross-linked more homogeneously [28], and antigenicproteins and cell remnants were linked to the collagen and cannot be extracted anymore byCHAPS/SDS treatment.Von Kossa staining demonstrated that CHAPS/SDS treatment had clearly decreased theoccurrence of calcium. Compared with GA cross-linked tissue, the additional CHAPS/SDStreatment resulted in a reduction of calcification at the edges of the fibrosa of leaflets, and at thefirst layers of the luminal side of the walls. BD45EN cross-linked leaflets demonstrated thatcalcification was found in both the ventricularis and the spongiosa of leaflets and intensely at the


- 187 -

outer edges of walls. CHAPS/SDS induced calcification in predominantly the spongiosa of leafletsand hardly at the outer edges of the walls. This agrees with the observed morphological changesand the detergent function of CHAPS/SDS.Calcification was always found to be related with both cells and ECM, which agrees with otherreports and CHAPS/SDS did not change this.

CONCLUSIONS

It is concluded that epoxy and EDC/NHS cross-linked valve tissue induced an immuno and foreignbody reaction as well as a calcification reaction similar to the clinically well accepted GA cross-linked tissue. If bisepoxide or EDC/NHS cross-linked tissue has a lower cytotoxicity than GAcross-linked tissue, which evokes a cytotoxic reaction [46, 47], and it results in materials withgood mechanical properties, these methods are promising in the fabrication of heart valvebioprostheses. However, these material calcify only somewhat less as GA cross-linked valves.An additional CHAPS/SDS treatment performed on successive epoxy and carbodiimide(BD45EN) or GA cross-linked heart valves resulted in reduced calcification and slightly varyingdata in view of antigenicity, which showed that always lower values were obtained than in non-cross-linked tissue. Due to the lower tendency to calcify this CHAPS/SDS treatment performed onporcine valves cross-linked with the methods mentioned before may be an additional improvementfor the fabrication of heart valve bioprostheses.

References1. F. Ricou, A. Brun, and R. Lerch, "Hemodynamic comparison of Medtronic intact bioprostheses and

bileaflet mechanical prostheses in aortic position", Cardiology, 87 pp. 212-215 (1996).2. G. Gong, E. Seifter, W.D. Lyman, S.M. Factor, S. Blau, and R.W.M. Frater, "Bioprosthetic cardiac valve

degeneration: Role of inflammatory and immune reaction", J. Heart Valve Dis., 2 pp. 684-693 (1993).3. R.F. Oliver, "Collagen as a dermal implant", in "Biocompatibility of Tissue Analogs", Ed. by D.F.

Williams, CRC Press, Inc., Baca Raton, Florida (1985)4. A. Jayakrishnan and S.R. Jameela, "Glutaraldehyde as a fixative in bioprosthetic and drug delivery

matrices", Biomaterials, 17 pp. 471-484 (1996).5. B.S. Goldman, T.E. David, D.F. Del Rizzo, J. Sever, and J. Bos, "Stentless porcine bioprostheses for aortic

valve replacement", J. Cardiovasc. Surg., 35(Suppl. 1 to No. 6) pp. 105-110 (1994).6. D.F. Del Rizzo, B.S. Goldman, and T.E. David, "Aortic valve replacement with a stentless porcine

bioprostheses: Multicentre trial", Can. J. Cardiovas., 11(7) pp. 597-603 (1995).7. D. Ross, "From homograft to stentless bioprostheses", in "Stentless bioprostheses", Ed. by A. Piwnica and

S. Westaby, Isis Medical Media, Oxford (1995)8. F.J. Schoen, H. Harasaki, K.M. Kim, and H.C. Anderson, "Biomaterial-associated calcification: Pathology,

mechanisms. and strategies for prevention", J. Biomed. Mat. Res., 22(A1) pp. 11-36 (1988).9. J. Chanda, S.B. Rao, M. Mohanty, A.V. Lal, C.V. Muraleedharan, G.S. Bhuvaneshwar, and M.S.

Valiathan, "Prevention of calcification of tissue valves", Art. Org., 18(10) pp. 752-757 (1994).10. M.N. Girardot, J.M. Girardot, and M. Torrianni, "Alpha-aminooleic acid (AOA) anticalcification effects

on glutaraldehyde-fixed heart valves: shelf-life studies", in "New horizons and the future of heart valveprostheses", Ed. by S. Gabbay and R.W.M. Frater, Silent Partners Inc., Austin (1994)

11. P.K. Bajpai, "Immunological aspects of treated natural tissue prostheses", in "Biocompatibility of tissueanalogs", Ed. by D.F. Williams, CRC Press, Inc., Boca Raton, Florida. p. 5-25 (1985)


Chapter 10

- 188 -


14. S.S. Srivatsa, P.J. Harrity, P.B. Maercklein, L. Kleppe, J. Velnot, W.D. Edwards, C.M. Johnson, and L.A.Fitzpatrick, "Increased cellular expression of matrix proteins that regulate mineralization is associatedwith calcification of native human and porcine xenograft bioprosthetic heart valves", J. Clin. Invest., 99pp. 996-1006 (1997).

15. M. Shen, P. Marie, D. Farge, S. Carpentier, C.d. Pollak, M. Hott, L. Chen, B. Martinet, and A. Carpentier,"Osteopontin is associated with bioprosthetic heart valve calcification in humans", C. R. Acad. Sci. III,320(1) pp. 49-57 (1997).


17. S.L. Hilbert, M. Jones, and V.J. Ferrans, "Flexible leaflet replacement heart valves", in "Encyclopedichandbook of biomaterials and bioengineering Part B: Applications", Ed. by D.L. Wise, et al., MarcelDekker, Inc., New York. p. 1111-1152 (1995)


19. R.J. Levy, F.J. Schoen, J.T. Levy, A.C. Nelson, S.L. Howard, and L.J. Oshry, "Biologic determinants ofdystrophic calcification and osteocalcin deposition in glutaraldehyde-preserved porcine aortc valve leafletsimplanted subcutaneously in rats", Am. J. Pathol., 113 pp. 143-155 (1983).



22. Y. Noishiki, H. Koyanagi, T. Miyata, and M. Furuse, Bioprosthetic valve, Patent EP 0 306 256 A2 1988.23. E. Imamura, O. Sawatani, H. Koyanagi, Y. Noishiki, and T. Miyata, "Epoxy compounds as a new

crosslinking agent for porcine aortic leaflets: subcutaneous implant studies in rats", J. Cardiac Surg., 4 pp.50-57 (1989).

24. R. Zeeman, P.J. Dijkstra, P.B. v. Wachem, M.J.A. v. Luyn, M. Hendriks, P.T. Cahalan, and J. Feijen,"Cross-linking and modification of dermal sheep collagen using 1,4-butanediol diglycidyl ether", Chapter3 of this thesis and submitted to J. Biomed. mat. Res. (1998).

25. R. Zeeman, P.J. Dijkstra, P.B. v. Wachem, M.J.A. v. Luyn, M. Hendriks, P.T. Cahalan, and J. Feijen,"Successive epoxy and carbodiimide cross-linking of dermal sheep collagen", Chapter 6 of this thesis.

26. P.B. v. Wachem, R. Zeeman, P.J. Dijkstra, M. Hendriks, P.T. Cahalan, J. Feijen, and M.J.A.v. Luyn,"Characterization and biocompatibility of epoxy crosslinked dermal sheep collagen", Chapter 7 of thisthesis and ubmitted to J. Biomed. Mat. Res. (1998).

27. R. Zeeman, P.J. Dijkstra, P.B. v. Wachem, M.J.A. v. Luyn, M. Hendriks, P.T. Cahalan, and J. Feijen,"Cross-linking and modification of porcine aortic heart valves", Chapter 8 of this thesis

28. R. Zeeman, P.J. Dijkstra, P.B. v. Wachem, M.J.A. v. Luyn, M. Hendriks, P.T. Cahalan, and J. Feijen,"Properties of cross-linked porcine aortic heart valves", Chapter 9 of this thesis

29. D. Hirsch, J. Drader, T.J. Thomas, F.J. Schoen, J.T. Levy, and R.J. Levy, "Inhibition of calcification ofglutaraldehyde pretreated porcine aortic valve cusps with sodiumdodecyl sulfate", J. Biomed. Mat. res., 27pp. 1477-1484 (1993).

30. D.J. Lentz, E.M. Pollock, D.B. Olsen, and E.J. Andreus, "Calcification of extrinsic valves", Trans. Am.Soc. Artif. Intern. Org., 28 pp. 494-497 (1982).

31. T. Schuerholz, "Critical dependence of the solubilization of lipid vescicles by the detergent CHAPS on thelipid composition. Functional reconstruction of the nicotine acetylcholine receptor into performed vesiclesabove the critical micellization concentration", Biophys. Chem., 58 pp. 87-96 (1996).

32. J. Cladera, J.-L. Rigaud, J. Villaverde, and M. Dunach, "Liposome solubilization and mebrane proteinreconstitution using CHAPS and CHAPSO", Eur. J. Biochem., 243 pp. 798-804 (1997).

33. I. Vesely, R. Noseworthy, and G. Pringle, "The hybrid xenograft/autograft bioprosthetic heart valve: Invivo evaluation of tissue extraction", Ann. Thorac. Surg., 60 pp. S359-364 (1995).


35. M.N. Girardot, M. Torrianni, D. Dillehay, and J.M. Girardot, "Role of glutaraldehyde in calcification ofprocine heart valves: comparing cusp and wall", J. Biomed. Mat. Res., 29 pp. 793-801 (1995).


- 189 -

36. P.F. Gratzer and J.M. Lee, "Altered mechanical properties in aortic elastic tissue usingglutaraldehyde/solvent solutions of various dielectric constant", J. Biomed. Mat. Res., 37 pp. 497-507(1997).

37. I. Vesely and W.J. Mako, "Comparison of the compressive buckling of porcine aortic valve cusps andbovine pericardium", J. Heart Valve Dis., 7 pp. 34-39 (1998).

38. A.K. Abbas, A.H. Lichtman, and J.S. Pober, "Cellular and molecular immunology", 3rd. ed.,Philiadelphia, PA, USA: W.B. Saunders Co.(1997).

39. R.N. Mitchell, R.A. Jonas, and F.J. Schoen, "Pathology of explanted cryopreserved heart valves:comparison with aortic valves from orthotopic heart transplants", J. Thorac. Cardiovasc. Surg., 115 pp.118-127 (1998).




43. L.M. Biedrzychi, E. Lerner, R.J. Levy, and F.J. Schoen, "Differential calcification of cusps and aortic wallof failed stented porcine aortic valves", J. Biomed. Mat. Res., 34 pp. 411-415 (1997).


45. M.A. Flomenbaum and F.J. Schoen, "Effects of fixation back pressure and antimineralization treatment onthe morphology of porcine aortic heart valves", J. Thorac. Cardiovasc. Surg., 105 pp. 154-164 (1993).



Summary

- 191 -

SUMMARY

During the past decades, collagen-based materials have been employed in replacement surgery tosubstitute missing, injured or diseased parts of the body. Collagen, in the form of fibers, representsthe single most abundant protein in mammals. The general properties of collagen which make thisprotein interesting as a biomaterial include the high strength of the fibers, low extensibility, and itssuitability as a substrate for cell growth. During in vivo applications, collagen is prone toenzymatic degradation. Therefore, collagen-based materials are frequently stabilized by cross-linking to control the rate of biodegradation. In addition, cross-linking is effective in suppressingthe antigenicity of collagen.An example of a collagen-based tissue is the aortic heart valve. A variety of pathological processescan lead to heart valve malfunction and this is usually associated with degenerative changes of thetissue. The most commonly used types of prosthetic valves are mechanical and tissue valves. Onemajor disadvantage in the use of mechanical valves is the need for continuous anticoagulationtherapy to minimize the risk of thrombosis, whereas tissue valves can be used withoutanticoagulants. Tissue valves are constructed from porcine aortic valves or bovine pericardium andare treated with glutaraldehyde to introduce cross-links. However, the durability of bioprosthesesis limited due to calcification, which leads to tearing and rupture of the material. The factors andthe mechanisms of calcification are still not fully understood. It was the objective of this study todevelop and to optimize new cross-linking methods for the stabilization of collagen-based tissuesresulting in materials which are biocompatible and non-calcifying and do not evoke cytotoxicreactions. A model tissue, dermal sheep collagen (DSC) which consists of 100 % fibrous collagentype I, was chosen to study the cross-linking process in more detail. Furthermore, the relationbetween the cross-linking conditions and the material properties obtained was studied. DSC waspreviously used in the development of new biomaterials. New cross-linking methods such as theuse of the water-soluble carbodiimide 1-ethyl-3-(3-dimethyl aminopropyl) carbodiimide (EDC) inthe presence of N-hydroxysuccinimide (NHS) were successfully developed and led tobiocompatible, non-calcifying materials.In chapter 2 the structure of collagen is described and the various known cross-linking methodsare given. Especially the effect of the cross-linking method on the stability towards enzymes andthe in-vivo calcification is evaluated. Furthermore, the morphology and the mechanical propertiesof porcine aortic heart valves are mentioned. Calcification of bioprosthetic valves and the anti-mineralization treatments carried out so far are reviewed in detail. It was concluded thatcalcification is a multi-factorial process which is induced by the glutaraldehyde cross-linking andthe presence of cells and proteins. Furthermore, epoxy compounds have proved to result in well-stabilized materials with a markedly reduced calcification. However, these studies did not describethe cross-linking process in detail and several undefined poly-functional epoxy compound mixtureswere utilized. Therefore, the use of a bifunctional epoxy compound as cross-linking agent wasstudied in more detail.The bisepoxide 1,4-butanediol diglycidyl ether (BDDGE) was selected to cross-link DSC. Theeffect of the reaction time, the BDDGE concentration, the solution pH, and the temperature on theshrinkage temperature (Ts) and the content of amine groups of the residual materials is evaluated

Summary

- 192 -

in chapter 3. Depending on the pH, cross-links were formed between the (hydroxy)lysine residuesof the collagen at pH > 8.0, while bridges were created between the carboxylic acid groups ofaspartic and glutamic acid residues at pH < 6.0. Under optimal basic conditions (pH 9.0), the Tswas raised from 46 to 70 °C, while a Ts of 62 °C was obtained after cross-linking at pH 4.5. Anincrease in BDDGE concentration or an elevation of the pH from 8.5 to 10.5, resulted in a fasterdecrease in amine groups, and in a higher degree of one-sided (masking) reactions, leading topendant epoxide groups in the DSC matrix. The effect of masking reactions on the collagenmaterial was investigated by reacting DSC with the monofunctional glycidyl isopropyl ether (PGE)under basic conditions. It was shown that masking of amine groups led to a decrease in Ts.Hydrolysis of epoxide groups became especially important during cross-linking at prolongedreaction times under acidic conditions.A kinetic model which describes the PGE masking or the BDDGE cross-linking reaction asfunction of the reaction time is proposed in chapter 4. The reaction rate, which was determined bythe decrease in amine groups, was proportional to the square of the concentration of amine groupsduring the PGE masking reaction, whereas a 2.5-th order dependency of the reaction rate withrespect to the amine groups was obtained during BDDGE cross-linking in the pH range from 8.5to 10.5. This difference in reaction order is most probably due to the effect of a second reaction ofa pendant epoxide group with an amine group, thus introducing a cross-link. The reaction orderwith respect to the epoxide groups was equal to 1 for both the PGE and the BDDGE reaction.The cross-link reaction between BDDGE and DSC occurred via a two-step mechanism as shownby a cross-link experiment at pH 9.0, in which the content of pendant epoxide groups reached amaximum. The ratio between masked amine groups and cross-links, which is a measure of thecross-link efficacy, was approximately one at pH 10.0, and was hardly dependent on the reactiontime and the initial BDDGE concentration. Reaction at pH 9.0 revealed that the ratio betweenmasking and cross-link was dependent on the cross-linking time.The in-vitro resistance towards bacterial collagenase and pronase of BDDGE cross-linked DSC isevaluated in chapter 5. Cross-linking at pH 4.5 (BD45) yielded a pliable and flexible material,which had a high tensile strength (5.4 MPa) but a moderate resistance towards enzymes, even after144 h of cross-linking. Reaction at pH 9.0 (BD90) resulted in a stiff and rigid material, which hada low tensile strength (2.6 MPa), but an excellent resistance towards enzymes after at least 72 h ofcross-linking. These differences in stability towards enzymes of DSC cross-linked at pH 4.5 or 9.0are explained by the differences in locations of the cross-links, types of cross-links and thespecificity of the enzymes.Upon degradation with pronase, the changes in mechanical properties of BD45 were moresensitive to degradation than the change in weight. Degradation of this material with collagenasedid not reduce the mechanical properties. BD90 cross-linked DSC affords a material which is notdegraded and which mechanical properties were not affected by exposure to collagenase orpronase.Because BD45 cross-linked DSC had good macroscopic and mechanical properties, but arelatively poor resistance towards enzymes, an additional EDC/NHS cross-linking step wasperformed on this material (chapter 6). This method improved the resistance towards enzymeswithout altering the other properties. The material properties, such as swelling, in-vitro enzymaticdegradation and tensile properties can be explained by the presence of inter-, and intrahelical and

Summary

- 193 -

intermicrofibrillar cross-links. Cross-links between and in the helices affects the Ts and theresistance against degradation by collagenase, while the mechanical properties and thesusceptibility towards pronase are dependent on the intermicrofibrillar cross-links.In chapter 7, the in-vivo biocompatibility and the tendency to calcify of BDDGE cross-linked DSCmaterials is determined by a subcutaneous implantation in male Albino Oxford rats. The materialswere biocompatible in terms of non-cytotoxicity and non-antigenicty. Furthermore, none of thematerials tested was found to calcify. Biodegradation was dependent on the cross-linking methodapplied. Materials with the lowest shrinkage temperatures, displayed the highest rates ofdegradation and after 6 weeks no or hardly any material was retrieved. BD90 cross-linked DSC isnot degraded and exhibited a good matrix function for the formation of new collagen.

Porcine aortic heart valves are composed of three leaflets which are anchored in the aortic wall.The aortic wall is composed of high contents of elastin and smooth muscle cells, while the leafletscontain mainly collagen as described in chapter 8. Amino acid analysis revealed that the remainingproteins such as proteoglycans contain high contents of hydrophilic amino acids. Cross-linking ofporcine aortic leaflets was accomplished with EDC/NHS, BDDGE or 4,7,10-trioxa-1,13-tridecanediamine (TTDD) which resulted in an increase in Ts from 61 to 80 °C or higher, which iscomparable to glutaraldehyde cross-linked controls (84 °C). Cross-linking in which amide typebonds were formed (EDC/NHS) displayed the highest values of Ts and a moderate reduction inamine groups to 58 %. Reaction of leaflets with BDDGE or glutaraldehyde (GA) which react withthe amine groups gave materials with a somewhat lower Ts as compared to EDC/NHS. Cross-linking via the carboxylic acid groups using TTDD or BDDGE (pH 4.5) revealed the lowest Tswithout a change in amine groups. Masking of amine groups with PGE or acetic acid N-hydroxysuccinimide ester resulted in a reduction of Ts due to partial destabilization of the triplehelix.The properties of the cross-linked valves are described in chapter 9. In addition, porcine aorticvalves were also cross-linked with a successive BDDGE and EDC/NHS method and withchondroitin sulfate (CS). GA cross-linked leaflets exhibited the lowest content of amine groupsafter cross-linking, whereas leaflets cross-linked with TTDD in the presence of EDC and NHS hadthe highest Ts of 89 °C. The swelling ratio, which was reduced after cross-linking, was notproportional to the Ts, which indicated that not only the extent of cross-linking but also the typeof cross-links and the content of ionizable groups on the collagen determine the swelling. In-vitroenzymatic degradation tests demonstrated that the cross-linking methods resulted in leaflets with acomparable stability towards enzymes as glutaraldehyde treated controls. Selective extraction ofcomponents from (non)-cross-linked leaflets using either the detergents 3-(3-cholamidopropyl)dimethylammonio)-1-propanesulfonate (CHAPS) and sodium dodecyl sulfate (SDS), elastase orguanidine hydrochloride, revealed that cross-linking of porcine aortic heart valve tissue involvesmainly collagen, but also elastin, glycosaminoglycans and likely other proteins.Contrary to the leaflets, the aortic wall is incompletely cross-linked as reflected by three differenttransitions in the thermograms. This resulted in materials which were less resistant towardsenzymes as compared to their leaflet counterparts.In chapter 10, the biocompatibility of BDDGE and/or EDC/NHS cross-linked porcine aortic heartvalves is evaluated by subcutaneous implantation of both the wall and the leaflets in Sprague-

Summary

- 194 -

Dawley rats for 8 weeks. The leaflets and the walls evoked an immuno and foreign body reactionsimilar to the clinically well accepted GA cross-linked tissue. Furthermore, the calcium levels wereonly slightly lower than GA fixed controls. An additional CHAPS/SDS treatment performed onsuccessive epoxy and carbodiimide (BD45EN) or GA cross-linked heart valves resulted in reducedcalcification and slightly varying data in view of antigenicity, which showed that always lowervalues were obtained than in non-cross-linked tissue. It is concluded that a combination ofbisepoxy cross-linking followed by CHAPS/SDS extraction may be an attractive method for thefabrication of heart valve bioprostheses.

Samenvatt ng

- 195 -

SAMENVATTING

Gedurende de afgelopen jaren zijn collagene materialen in de chirurgie gebruikt voor hetvervangen van ontbrekende, gewonde of zieke delen van het menselijk lichaam. Collageen, in devorm van vezels, is het meest voorkomende eiwit in zoogdieren. De eigenschappen van collageendie dit eiwit interessant maken voor het gebruik als een biomateriaal omvatten de hoge sterkte vande vezels, alsmede de lage rek en de geschiktheid als een substraat voor celgroei. Tijdensimplantatie is collageen echter gevoelig voor degradatie door enzymen. Om de degradatiesnelheidvan het materiaal te controleren wordt het collageen gecrosslinkt (vernet). Daarnaast verlaagt hetcrosslinkproces de antigeniciteit (mate van het opwekken van een allergische of immunogenereactie) van het materiaal.De hartklep in de aorta is een voorbeeld van een collageen materiaal. Diverse pathologischeprocessen kunnen resulteren in het slecht functioneren van deze klep en dit wordt normalitergeassocieerd met degenererende veranderingen van het klepweefsel. De huidige hartklepprothesenomvatten mechanische en weefselhartkleppen. Een groot nadeel van de mechanische hartkleppenis dat de patiënt altijd bloedverdunners nodig heeft om de risico's op trombose te minimalizeren.Daarentegen kunnen weefselhartkleppen zonder medicijnen worden gebruikt. Hartkleppen vandierlijke oorsprong worden of van aortakleppen van een varken, of van runderpericardia(membraan dat om het hart ligt) gemaakt die vervolgens worden gecrosslinkt metglutaaraldehyde. Echter, de duurzaamheid van deze hartkleppen is beperkt doordat er calcificatie(neerslaan van calciumfosfaatzouten op het zachte weefseloppervlak) optreedt, wat kan leiden tothet slijten en scheuren van de kleppen. De factoren en de mechanismen die aan dit verschijnsel tengrondslag liggen zijn nog niet volledig opgehelderd. Het doel van het onderzoek beschreven in ditproefschrift is om nieuwe crosslinkmethoden voor collagene materialen te ontwikkelen en teoptimaliseren, wat zou moeten leiden tot materialen die bloed en lichaamsvriendelijk(biocompatibel) zijn, die niet calcificeren en die geen cytotoxische (giftige) reacties opwekken.Dermaal schapencollageen (DSC), dat bestaat uit 100 % type I collageen, is gebruikt alsmodelweefsel om het crosslinkproces in detail te kunnen bestuderen. Daarnaast werd getrachtrelaties te leggen tussen de toegepaste crosslinkprocedure en de verkregenmateriaaleigenschappen. DSC is in het verleden al gebruikt voor de ontwikkeling van nieuwebiomaterialen. Crosslinkmethoden gebruik makende van het wateroplosbare 1-ethyl-3-(dimethylaminopropyl) carbodiimide (EDC) in aanwezigheid van N-hydroxysuccinimide (NHS) werdensuccesvol toegepast en leidde tot biocompatibele materialen die niet calcificeerden.In hoofdstuk 2 wordt eerst een overzicht gegeven van de struktuur van collageen en van dehuidige bekende en toegepaste crosslinkmethoden. Vooral het effect van de gebruikte methode ophet degradatiegedrag door enzymen en het calcificatiegedrag tijdens implantatie wordenbeschreven. Vervolgens worden de anatomie, de morfologie en de mechanische eigenschappen vanaortakleppen van varkens uitvoerig beschreven. Tenslotte wordt een overzicht gegeven van decalcificatie van klepprothesen en de behandelingen die tot nu toe zijn uitgevoerd om calcificatietegen te gaan. Uit dit literatuuronderdeel werd geconcludeerd dat calcificatie een proces is datdoor diverse factoren wordt bepaald, en dat het glutaaraldehyde crosslinken en de aanwezigheid

Samenvatt ng

- 196 -

van lichaamsvreemde (afkomstig van het varken) cellen en eiwitten calcificatie induceren enversterken. Daarnaast bleek dat het crosslinken van collagene materialen met epoxide-verbindingen een goed alternatief is. De verkregen materialen degradeerden niet en calcificieerdenduidelijk minder dan GA gecrosslinkte materialen. De huidige studies die gebruik maakten vanepoxide-verbindingen beschrijven echter niet of nauwelijks iets over het crosslinkmechanisme ende chemie van de reacties. Daarnaast worden diverse ongedefinieerde mengsels vanpolyfunctionele epoxide-verbindingen gebruikt. Mede hierdoor werd een bifunctionele epoxideverbinding als crosslinker gekozen om het crosslinkproces in detail te bestuderen.

De bisepoxide 1,4-butaandioldiglycidyl ether (BDDGE) werd geselecteerd om DSC te crosslinken.De invloed van de reactietijd, de BDDGE concentratie, de pH en de reactietemperatuur op dekrimptemperatuur (Ts) en het aantal aminogroepen van het verkregen materiaal werd bestudeerdin hoofdstuk 3. Crosslinks werden gevormd tussen de (hydroxy)lysine residuen van het collageenals de pH van de oplossing > 8.0 was, terwijl er bindingen tussen de carboxylzuurgroepen vanasparagine- en glutaminezuur werden gevormd als de pH < 6.0 was. Onder optimale basischereactiecondities (pH 9.0) werd de Ts verhoogd van 46 naar 70 °C, terwijl een stijging tot 62 °Cwerd waargenomen indien de reactie werd uitgevoerd bij pH 4.5. Een toename van de BDDGEconcentratie of een verhoging van de pH van 8.5 naar 10.5 resulteerde in een versnelde afname inaminogroepen en in een hogere mate van éénstandige (maskerings) reacties. Dit laatste resulteertin de introductie van een epoxide groep in de matrix. Het effect van deze éénstandige reacties opde materiaaleigenschappen werd onderzocht door DSC te laten reageren met het monofunctioneleglycidyl isopropyl ether (PGE) bij basische pH. Hieruit bleek dat maskering van aminogroepenleidt tot een verlaging in Ts.Tenslotte bleek dat hydrolyse van de functionele epoxide groepen tijdens het crosslinkenvoornamelijk van belang als de reactie voor lange tijd wordt uitgevoerd bij een zure pH.Een kinetisch model dat zowel de PGE maskeringsreactie als de BDDGE crosslinkreactie in de tijdbeschrijft werd opgesteld in hoofdstuk 4. De reactiesnelheid die uitgedrukt werd in de afname vanaminogroepen per tijdseenheid, was 2de orde in aminogroepen bij de PGE reactie. Daarentegenwerd een 2.5de orde in aminogroepen gevonden voor de BDDGE reactie. Het verschil tussen dezetwee werd toegeschreven aan het effect van de (extra) reactie van de tweede epoxide groep meteen aminogroep wat resulteert in een crosslink. De reactieorde in epoxidegroepen was zowel voorde PGE als de BDDGE reactie gelijk aan 1. Het crosslinken van DSC met BDDGE geschiedt viatwee afzonderlijke stappen zoals werd aangetoond in een experiment uitgevoerd bij pH 9.0, waarineen maximum hoeveelheid aan eindstandige epoxide groepen in de matrix werd gevonden. Deverhouding tussen de éénstandige reacties en de crosslinks, wat een maat is voor decrosslinkefficiëntie, was ongeveer 1 als de reactie werd uitgevoerd bij pH 10.0. Daarnaast bleekdeze verhouding nauwelijks afhankelijk te zijn van de reactietijd en de beginconcentratie vanBDDGE. De crosslinkefficiëntie bleek echter wel afhankelijk te zijn van de reactietijd als de reactiewerd uitgevoerd bij pH 9.0.De in-vitro weerstand tegen afbraak door collagenase en pronase van BDDGE gecrosslinkt DSCwerd geëvalueerd in hoofdstuk 5. Crosslinken bij pH 4.5 (BD45) geeft een buigzaam en flexibelmateriaal, dat een hoge treksterkte heeft (5.4 MPa), maar dat langzaam wordt gedegradeerd, ookals de crosslinkreactie meer dan 144 h heeft geduurd. Reactie bij pH 9.0 (BD90) leidt tot een stug

Samenvatt ng

- 197 -

en stijf materiaal, dat een lage treksterkte heeft (2.6 MPa), maar dat geheel niet wordtgedegradeerd door enzymen als het crosslinken langer dan 72 h heeft geduurd. Het verschil tussenhet in-vitro degradatiegedrag van DSC gecrosslinkt met BDDGE bij pH 4.5 of 9.0 wordtverklaard door de verschillen in locaties van de gevormde crosslinks, het type crosslink en despecificiteit van de gebruikte enzymen.Tijdens degradatie met pronase waren de veranderingen in mechanische eigenschappen van BD45veel groter dan de veranderingen in gewichtsverlies. Degradatie van dit materiaal met collagenaseveranderde de mechanische eigenschappen niet. BD90 werd niet gedegradeerd en de mechanischeeigenschappen werden niet veranderd door zowel collagenase als pronase.Doordat BD45 goede macroscopische en mechanische eigenschappen heeft, maar gedegradeerdwordt door enzymen is een additionele EDC/NHS crosslinkingsstap uitgevoerd (hoofdstuk 6).Deze stap zorgde voor een verhoogde stabiliteit van het materiaal zonder de overigeeigenschappen te veranderen. De materiaaleigenschappen zoals de zwelling, de in-vitro degradatieen de mechanische eigenschappen kunnen verklaard worden door de aanwezigheid van inter- enintrahelische en intermicrofibrillaire crosslinks. Crosslinks die tussen of in de helices zijn gevormdbeïnvloeden de Ts en de degradatie in collagenase, terwijl de mechanische eigenschappen en dedegradatie in pronase meer lijken te worden bepaald door de intermicrofibrillaire crosslinks.De in-vivo biocompatibiliteit en de neiging tot calcificatie van BDDGE gecrosslinkt DSCmaterialen werd bepaald middels een onderhuidse implantatie in mannelijke Albino Oxford ratten.De gebruikte materialen waren niet cytotoxisch en antigeen en konden dus als biocompatibelbestempeld worden. Daarnaast calcificeerde geen van de geteste materialen. De biodegradatie waserg afhankelijk van de crosslinkmethode die gebruikt was. Materialen die de laagstekrimptemperatuur hadden, lieten de hoogste degradatiesnelheden zien en na 6 weken kon ernauwelijks nog materiaal worden teruggevonden. BD90 degradeerde niet en bleek een goedmatrixfunctie te hebben voor de vorming van nieuw (ratten)collageen.

Varkenshartkleppen zijn opgebouwd uit drie ‘blaadjes’ (leaflets) die verankerd zijn in deaortawand. De aortawand bestaat veelal uit elastine en gladde spiercellen, terwijl de afzonderlijkeleaflets voornamelijk uit collageen bestaan (hoofdstuk 8). Aminozuur analyses lieten zien dat deoverige eiwitten zoals proteoglycanen een hydrofoob karakter hebben. Aortakleppen van varkenswerden gecrosslinkt met EDC/NHS, BDDGE of 4,7,10-trixoxa-1,13-tridecanediamine (TTDD).Een verhoging van de Ts van 61 naar 80 °C of hoger werd gemeten, wat goed overeenkomt metde waardes van glutaaraldehyde (GA) gecrosslinkte materialen (84 °C). De hoogste waardes vanTs werden gemeten indien er tijdens het crosslinken amide-bindingen werden gevormd. Hetgehalte aan aminogroepen werd gereduceerd tot 58 %. Reactie van de leaflets met BDDGE ofGA, die reageren met de aminogroepen van het collageen, resulteerden in materialen met een ietslagere Ts dan met EDC/NHS. Crosslinken via de carbozylzuurgroepen zoals met TTDD ofBDDGE bij pH 4.5 gaven de laagste waardes van Ts te zien en het gehalte aan aminogroepen wasnauwelijks veranderd. Maskeringsreacties van de aminogroepen met PGE of azijnzuur N-hydroxsuccinimide ester resulteerden in een duidelijke verlaging van de Ts, dat werdtoegeschreven aan destabilizatie van de triple-helix.De eigenschappen van de gecrosslinkte kleppen zijn beschreven in hoofdstuk 9. Devarkenshartkleppen werden ook nog gecrosslinkt met twee andere methoden namelijk een

Samenvatt ng

- 198 -

tweestaps BDDGE en EDC/NHS methode en met chondroitinesulfaat (CS). GA gecrosslinkteleaflets hadden het laagste percentage aan aminogroepen na crosslinken, terwijl de leaflets diegecrosslinkt waren met TTDD in de aanwezigheid van EDC en NHS de hoogste waarde van Ts(89 °C) lieten zien. De zwelling, die afnam na crosslinken, had geen lineair verband met degemeten Ts. Dit impliceert dat de zwelling niet alleen door de crosslinkingsgraad, maar ook doorhet type crosslink en de hoeveelheid ionizeerbare groepen van het collageen wordt bepaald. Degecrosslinkte materialen hadden een stabiliteit in een enzymoplossing die vergelijkbaar was met deGA gecrosslinkte standaardmaterialen. Selectieve extractie van componenten uit (niet)gecrosslinkte materialen werd bewerkstelligd met de detergenten (zepen) 3-(3-cholamidopropyl)dimethylammonio)-1-propaansulfonaat (CHAPS) en natrium dodecylsulfaat (SDS), elastase ofguanidine hydrochloride. Deze studies lieten zien dat het crosslinken van varkenshartkleppenmeestal plaatsvindt in de collagene component van het weefsel. Echter, ook elastine,glycosaminoglycanen (suikers) en waarschijnlijk andere eiwitten kunnen deelnemen in hetcrosslinkproces.In tegenstelling tot de leaflets wordt de aortawand niet volledig gecrosslinkt. Dit werd vooralduidelijk doordat er drie thermische overgangen werden waargenomen. De materialen waren medehierdoor veel gevoeliger voor enzymatische degradatie dan de corresponderende leaflets.De biocompatibiliteit van BDDGE en/of EDC/NHS gecrosslinkte varkensaortakleppen werdgeëvalueerd middels een onderhuidse implantatie van 8 weken in Sprague-Dawley ratten(hoofdstuk 10). Zowel de leaflets als de aortwanden gaven een zelfde immuno- en lichaamsreactiete zien als de klinisch geaccepteerde GA behandelde materialen. Alle gecrosslinkte materialenlieten echter min of meer dezelfde calcificatie zien. Het extraheren van componenten uit dekleppen die gecrosslinkt waren of via de tweestaps BDDGE en EDC/NHS methode, of met GA,werd uitgevoerd met CHAPS en SDS. Deze behandeling resulteerde in een verlaagde calcificatie.De antigeniciteit varieerde, maar was altijd lager dan de waardes verkregen bij implantatie vanniet-gecrosslinkt weefsel. Er wordt geconcludeerd dat vooral de combinatie van bisepoxidecrosslinking, gevolgd door een wasstap met CHAPS/SDS interessant is voor het vervaardigen vanhartklepprothesen.

Curr culum v tae

- 199 -

CURRICULUM VITAE

Raymond Zeeman werd geboren op 24 juni 1970 te Hengelo (O) en bracht zijn jeugd door inachtereenvolgens Hengelo (O), Ulft, Enschede en Losser. Hij volgde zijn middelbare schoolopleiding aan het "Kottenpark College" te Enschede, waar hij in 1988 het VWO-diploma behaalde.In datzelfde jaar begon hij aan de studie Chemische Technologie aan de Universiteit Twente. Naeen stage bij (toen nog) Ciba-Geigy in Basel, Zwisterland, deed hij zijn afstudeeropdracht in devakgroep van Prof. Dr. A. Bantjes onder de bezielende leiding van Dr. Jeroen Verschuuren.Tijdens dit onderzoek werd gekeken naar de invloed van een tweestaps peroxide-vulcanizatie vanroetgevulde EPDM-rubbers. In juli 1994 studeerde hij af.In september 1994 begon hij aan zijn promotieonderzoek onder leiding van Prof. Dr. J. Feijen enDr. P.J. Dijkstra. Daarnaast was Dr. Ir. Marc Hendriks van Medtronic/Bakken Research Centerdirect betrokken bij het onderzoek. De resultaten van dit onderzoek staan beschreven in hetvoorliggende proefschrift, waar de auteur op 6 november 1998 op hoopt te promoveren.

Documents

Cross-linking of collagen-based materials · CROSS-LINKING OF COLLAGEN-BASED MATERIALS PROEFSCHRIFT ter verkrijging van de graad van doctor aan de Universiteit Twente, op gezag van