Available online at www.sciencedirect.com

Acta Biomaterialia 5 (2009) 1864–1871

www.elsevier.com/locate/actabiomat

Lysine-PEG-modified polyurethane as a fibrinolytic surface:Effect of PEG chain length on protein interactions, platelet

interactions and clot lysis

Dan Li a,b, Hong Chen a,b,*, W. Glenn McClung c, John L. Brash a,c

a State Key Laboratory of Advanced Technology for Materials Synthesis and Processing, 122 Luoshi Rd., Wuhan 430070, People’s Republic of Chinab School of Materials Science and Engineering, Wuhan University of Technology, 122 Luoshi Rd., Wuhan 430070, People’s Republic of China

c School of Biomedical Engineering and Dept. of Chemical Engineering, McMaster University, Hamilton, Ont., Canada

Received 26 November 2008; received in revised form 6 February 2009; accepted 2 March 2009Available online 9 March 2009

Abstract

Fibrinolytic polyurethane surfaces were prepared by conjugating lysine to the distal terminus of surface-grafted poly(ethylene glycol)(PEG). Conjugation was through the a-amino group leaving the e-amino group free. Lysine in this form is expected to adsorb both plas-minogen and t-PA specifically from blood. It was shown in previous work that the PEG spacer, while effectively resisting nonspecificprotein adsorption, was a deterrent to the specific binding of plasminogen. In the present work, the effects of PEG spacer chain lengthon the balance of nonspecific and specific protein binding were investigated. PEG–lysine (PEG-Lys) surfaces were prepared using PEGsof different molecular weight (PEG300 and PEG1000). The lysine-derivatized surfaces with either PEG300 or PEG1000 as spacer showedgood resistance to fibrinogen in buffer. The PEG300-Lys surface adsorbed plasminogen from plasma more rapidly than the PEG1000-Lys surface. The PEG300-Lys was also more effective in lysing fibrin formed on the surface. These results suggest that the optimumspacer length for protein resistance and plasminogen binding is relatively short. Immunoblots of proteins eluted after plasma contactconfirmed that the PEG–lysine surface adsorbed plasminogen while resisting most of the other plasma proteins. The hemocompatibilityof the optimized PEG–lysine surface was further assessed in whole blood experiments in which fibrinogen adsorption and platelet adhe-sion were measured simultaneously. Platelet adhesion was shown to be strongly correlated with fibrinogen adsorption. Platelet adhesionwas very low on the PEG-containing surfaces and neither surface-bound lysine nor adsorbed plasminogen promoted platelet adhesion.� 2009 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved.

Keywords: Biocompatibility; Fibrinolytic property; Polyurethane; Protein adsorption; Platelet adhesion

1. Introduction

Major reasons for the failure of devices implanted in thebody are the rapid accumulation of proteins on the mate-rial surface and the interweaving of subsequent hostresponses, including blood coagulation, platelet activationand complement activation. Many methods of surface

1742-7061/$ - see front matter � 2009 Acta Materialia Inc. Published by Else

doi:10.1016/j.actbio.2009.03.001

* Corresponding author. Address: State Key Laboratory of AdvancedTechnology for Materials Synthesis and Processing, 122 Luoshi Rd.,Wuhan 430070, People’s Republic of China. Tel./fax: +86 27 87168305;fax: +86 27 87168305.

E-mail address: hongchen@whut.edu.cn (H. Chen).

modification have been used to improve the biocompatibil-ity of biomaterials. These can be roughly divided into twostrategies: ‘‘bioinert” and ‘‘bioactive”. Modification withpoly(ethylene glycol) (PEG) or polyethylene oxide (PEO)is the most widely used approach to bioinertness mainlybecause of the excellent resistance of PEG to nonspecificprotein adsorption and cell adhesion. PEG has also beenused as a spacer to couple bioactive moieties to surfaces,thus potentially exerting both ‘‘bioinert” and ‘‘bioactive”

functions [1–5]. In this approach PEG/PEO has the poten-tial not only to inhibit nonspecific protein adsorption butalso to move the bioactive moiety away from the surface,making it potentially more effective than if coupled directly[6]. From this perspective, the present paper develops the

vier Ltd. All rights reserved.

D. Li et al. / Acta Biomaterialia 5 (2009) 1864–1871 1865

concept of a fibrinolytic surface on which PEG is used as aspacer to conjugate lysine such that the e-NH2 is free andexposed to capture plasminogen upon exposure to blood.Plasminogen, the key zymogen of the fibrinolytic pathway,is cleaved by its physiological activator, tissue-type plas-minogen activator (t-PA), yielding the enzymatically activeform plasmin that lyses fibrin [7].

Brash and co-workers have explored the concept of aclot-lysing surface based on immobilization of lysine[8–11]. In vitro experiments indicated that a lysine-immobi-lized surface can selectively bind plasminogen from plasmaand, when activated by t-PA, the plasmin generated at thesurface can lyse fibrin. In the present work, lysine was con-jugated to the distal terminus of PEG surface-grafted onpolyurethane, such that the e-amino group was free. Itwas shown that this surface reduces nonspecific proteinadsorption efficiently while binding plasminogen fromplasma with some degree of selectivity [3]. When activatedby t-PA the adsorbed plasminogen was converted to plas-min, and the fibrinolytic activity of the surface localizedplasmin was demonstrated. However, the rate of plasmino-gen uptake was relatively slow (requiring 7 h to saturate),presumably due to the protein repellent properties of thePEG and the mobility of the terminally conjugated lysines.Thus it appeared that it might be necessary to vary thematerial properties to achieve an appropriate balance ofefficient plasminogen uptake and prevention of nonspecificprotein adsorption. It has been reported that the repellenteffect of PEG increases with increasing chain length inthe range up to a few thousand [12,13]; thus, by optimizingthe PEG chain length, it seemed that it might be possible tobalance the two effects.

Although the protein-resistant properties of PEG/PEOhave been extensively investigated, the effects of PEG prop-erties on the binding of target biomolecules to ligands atthe distal terminus when the PEG is used as a spacer havenot been considered. In the work reported here, the effectof PEG chain length on plasminogen binding to lysine atthe PEG distal terminus was investigated. Platelet adhesionand activation in whole blood under flow conditions andWestern blot analysis of adsorbed proteins on these sur-faces were also studied.

2. Materials and methods

2.1. Reagents

N,N0-disuccinimidyl carbonate (DSC, anhydrous, P95%),4,40-methylene-bis-(phenyl-isocyanate) (MDI, anhydrous,98%), trifluoroacetic acid (TFA), H-Lys(t-BOC)-OH and 4-nitrobenzaldehyde were from Sigma–Aldrich Chemical Co.and used without further purification. Poly(ethylene glycol)(Mn = 1000 and Mn = 300) was purchased from Sigma–Aldrich Chemical Co. and dried before use. All anhydroussolvents such as toluene, dimethyl formamide (DMF) andacetonitrile were obtained from EMD Chemicals Inc., Trieth-ylamine (TEA, 99%) was obtained from Alfa Aesar Co. Both

fibrinogen and plasminogen were purchased from EnzymeResearch Laboratories (South Bend, IN). Recombinant tissueplasminogen activator (t-PA) was obtained from Genentech(San Francisco, CA).

2.2. Preparation of surfaces

Tecothane polyurethane (TT-1095A) was from Ther-medics (Wilmington, MA), and was extracted (Soxhlet)for 48 h with methanol to remove impurities. Films of thismaterial were cast from a 5% (wt./vol.) solution in DMF,dried in air at 75 �C for 48 h and vacuum dried at 60 �Cfor 48 h to remove excess solvent. The polyurethane elasto-mer films were punched into discs, approximately 5 mm indiameter and 0.5 mm thick.

The modified surfaces used in this study consist of‘‘base” polyurethane (PU) with immobilized PEG towhich lysine is covalently attached at the distal terminus;the PEG may thus be seen as a spacer for the lysine andas a protein-repelling element. The details of the graftingtechniques have been described previously [3]. In brief,PU discs were immersed in a toluene solution containing7.5% (wt./vol.) MDI and 2.5% (wt./vol.) triethylamine.After stirring at 50 �C for 100 min, PEG-grafted PU sur-faces (PU-PEG) were obtained by immersing NCO-func-tionalized polyurethane surfaces (PU-NCO) in a toluenesolution containing 5% PEG (wt./vol.) at room tempera-ture for 24 h. For the covalent conjugation of e-lysine,the PU-PEG surfaces were first added to an acetonitrilesolution containing DSC (0.05 mmol ml�1) and TEA(0.05 mmol ml�1) and stirred at room temperature for6 h. The resulting surfaces (PU-PEG-NHS) were incu-bated overnight in phosphate-buffered saline (PBS) pH8.3 containing 5 mg ml�1 H-Lys(t-BOC)-OH to give aPU-PEG-Lys(P) surface. The surfaces with e-NH-t-Bocgroups were deprotected by treatment with 25% TFAfor 90 min and subsequently washed with PBS. The result-ing surface, with the e-NH2 groups of lysine exposed (PU-PEG-Lys), is of the greatest interest in terms of its fibri-nolytic potential. To study the effect of PEG spacer lengthon the balance between ‘‘repulsion” and ‘‘attraction” ofplasminogen, two PEGs, with molecular weights of 300and 1000, referred to as PEG300 and PEG1000 respec-tively, were used.

2.3. Graft density

The lysine graft densities were determined by reactingthe surface amino groups with 4-nitrobenzaldehyde toform imines, and subsequent hydrolysis to liberate the4-nitrobenzaldehyde [14]. Typically, nine discs of PU-PEG-Lys surface or PU-PEG-Lys(P) control surface wereimmersed in anhydrous ethanol (10 ml) containing 4-nitrobenzaldehyde (40 mg) and acetic acid (0.008 ml)under nitrogen at 50 �C for 3 h. The surfaces were thenwashed and sonicated in absolute ethanol for 2 min.The discs were immersed in water (1 ml) containing acetic

1866 D. Li et al. / Acta Biomaterialia 5 (2009) 1864–1871

acid (0.002 ml) and the solution was kept at 40 �C for 1 h.The hydrolysis was run in triplicate (three groups of threediscs). The 4-nitrobenzaldehyde liberated, equivalent tothe surface amino content, was determined by measuringabsorbance at 263 nm. A calibration curve was con-structed from absorbance measurements on a series of4-nitrobenzaldehyde solutions (0–0.1 lmol ml�1) in water.

2.4. Protein adsorption

Fibrinogen, labeled with 125I (ICN Pharmaceuticals,Irvine, CA) using the iodine monochloride (ICl) method,was passed through an AG 1-X4 column (Bio-Rad Labora-tories, Hercules, CA) to remove free iodide. Plasminogen(ICN Pharmaceuticals, Irvine, CA) labeled with 125I wasdialyzd against Tris-buffered saline (TBS, pH 7.4) toremove free iodide. For studies of fibrinogen adsorptionfrom buffer, labeled fibrinogen was mixed with unlabeledfibrinogen (2:98, labeled:unlabeled) at a total concentrationof 1 mg ml�1. For studies of plasminogen adsorption fromplasma, labeled protein was added to citrated platelet-poorplasma (PPP) at a concentration approximately 10% of theendogenous protein level. The surfaces, except PU-PEG-NHS, were equilibrated in TBS for 12 h prior to adsorptionexperiments.

Surfaces were incubated with protein in buffer or withplasma for 3 h at room temperature, rinsed three times(10 min each time) with TBS, wicked onto filter paperand transferred to clean tubes for radioactivity determina-tion. For studies of plasminogen adsorption kinetics, thesurfaces were incubated in plasma containing 125I plasmin-ogen for periods of 3, 5 and 7 h, then rinsed and counted.

2.5. Plasma clot lysis

Surfaces were incubated in pooled normal humanplasma (PNP) for 2 h at room temperature in microtiterplate wells. The films were removed from the wells, rinsedthree times in TBS and then placed in clean wells. Tissueplasminogen activator (t-PA) was added to the wells at aconcentration of 0.1 mg ml�1 in TBS and incubated for30 min at room temperature. The films were rinsed exten-sively with buffer to remove any unbound proteins. Thisprocedure provides surfaces bearing a layer of bound plas-min. The clot-lysing potential of the plasmin-adsorbed sur-faces was assessed using a modified plasma recalcificationassay. A 100 ll volume of PNP was added to the wells con-taining the surfaces. Following a 5 min equilibration periodat 37 �C, an equal volume of 0.025 M CaCl2 was injectedinto the wells. Absorbance at 405 nm was measured at30 s intervals over a 40 min period.

2.6. Western blot analysis

Surfaces were incubated in 100% PNP for 3 h atroom temperature. Adsorbed proteins were eluted using2% aqueous sodium dodecyl sulfate (SDS). Polyacryl-

amide gel electrophoresis (PAGE) and immunoblottingwere performed as described previously [15]. Briefly,the eluates were run on 12% reduced SDS–PAGE gelsto separate the proteins according to molecular weight.The proteins were then transferred from the gel ontoan Immobilon PVDF membrane (Millipore, Bedford,MA, USA). The membranes were cut into 3 mm stripsand blocked with 5% non-fat dry milk. The strips wereincubated with the primary antibodies (dilution of1:1000) to 21 plasma proteins selected for study andthen with the appropriate alkaline phosphatase-conju-gated second antibody (dilution of 1:1000). The sub-strate system used to develop a color reaction foralkaline phosphatase was 5-bromo-4-chloro-3-indolylphosphate (BCIP) and nitroblue tetrazolium (NBT)(both from Bio-Rad), prepared as described by thesupplier.

2.7. Platelet experiments

Platelets were harvested from fresh human whole bloodcollected into citrate anticoagulant using the method ofMustard et al. [16]. Heparinized PPP and a suspensionof red cells were prepared from the same blood sample.The platelets were labeled with sodium 51Cr-chromate[17] and fibrinogen was labeled with 125I using the IClmethod as described above. The washed 51Cr-labeledplatelets were resuspended in heparinized PPP containingapyrase (0.03 lL ml�1) and 125I-labeled fibrinogen wasadded at a concentration approximately 10% of theendogenous protein level. The final ‘‘whole blood” usedas the test fluid contained platelets at a concentration of250,000 ll�1. Washed red cells were added to give ahematocrit of 40%.

A cone-and-plate device was used to investigate theinteractions of fibrinogen and platelets with the surfacesin flowing whole blood [18]. Disks of the surfaces (40 mmdiameter) were equilibrated in Tyrodes buffer overnight.They were removed from the buffer immediately prior tothe experiment and assembled in the cone-and-plate appa-ratus. Reconstituted whole blood (1 ml), prepared asdescribed above, was then added to the wells. The coneswere lowered until the cone tips touched the surfaces androtated at 200 rpm (fluid shear rate of 300 s�1) to give flowconditions typical of those in the arterial circulation. After15 min the test surfaces were removed and rinsed withTyrodes buffer containing 0.01 M ethylenediaminetetraace-tic, followed by gamma counting or scanning electronmicroscopy (SEM).

For SEM the test surfaces with adherent plateletswere immersed in a 0.2% solution of glutaraldehyde inTyrodes buffer for 30 min at room temperature and over-night at 4 �C for fixation. They were then dehydratedthrough graded ethanol, dried in a CO2 critical pointdryer, mounted on aluminum specimen stubs, coatedwith gold and examined in the scanning electronmicroscope.

Fig. 1. Fibrinogen adsorption from buffer (3 h exposure) on modified andunmodified surfaces (mean ± SD, n = 3).

D. Li et al. / Acta Biomaterialia 5 (2009) 1864–1871 1867

3. Results and discussion

3.1. Graft density

The amounts of surface-immobilized lysine were deter-mined by measuring the 4-nitrobenzaldehyde that reactsvia Schiff-base formation and is then liberated by hydroly-sis (Table 1). This method has been used by Moon et al. fordetermining the absolute density of surface-immobilizedamino groups [14]. It should be noted that some of the iso-cyanate groups did not react during the PEG grafting pro-cess due to the steric hindrance of approaching PEGmolecules, and were ‘‘removed” by hydrolysis to formamino groups by extensive washing in water. The aminocontent values obtained for the PU-PEG-Lys(P) surfacesthus indicate the densities of amino groups on the ‘‘base”

and were used as ‘‘controls” for the PU-PEG-Lys. Theincrease in amino groups after deprotection indicates thedensity of e-amino-free lysine.

The lysine density is an important parameter that deter-mines the extent of plasminogen binding. The calculatedlysine densities on the PEG300 and PEG1000 surfacesshowed only a small difference within experimental error,so it can be concluded that lysine was immobilized at thesame density and this is reflected in the plasminogenadsorption data presented below. Typical grafting densitiesfor PEO-grafted surfaces have been found to be in therange of 0.1–0.5 chain nm�2 [19], so the values in Table 1appear reasonable. Given that the lysine (and presumablyalso the PEG) densities are similar for the two molecularweights, the PEG chain length should be the dominant con-sideration in the interactions of these materials withproteins.

3.2. Fibrinogen adsorption from buffer

Fibrinogen is a key protein of the coagulation cascadeand is known to play a leading role in mediating plateletadhesion to biomaterials. In this research, fibrinogenadsorption was measured to provide an indication of theprotein resistance of the modified surfaces.

As shown in Fig. 1, the unmodified PU surface exhibitedthe highest level of adsorption (�0.96 lg cm�2), whilemuch less protein was adsorbed on the PEG- and lysine-modified surfaces. The decrease is attributed to the proteinresistance of PEG, with the PEG300 materials clearly lesseffective than the PEG1000. This effect of chain lengthhas been observed by others [12,13]. It should be pointed

Table 1Densities of surface-immobilized lysine (mean ± SD, n = 3).

PEG type Density of amino groups (nmol/cm2)

PU-PEG-Lys(P)

PEG300 3.46 ± 0.12PEG1000 4.28 ± 0.23

* Difference PU-PEG-Lys � PU-PEG-Lys(P).

out that the interaction of proteins with the PU-PEG-NHS surface may involve covalent binding, as suggestedpreviously [3] based on resistance to elution by SDS. Aninteresting observation is that the effect of PEG on proteinadsorption is nonspecific even when the interactions arecovalent, as indicated by the result showing that PU-PEG300-NHS adsorbs much more protein than PU-PEG1000-NHS. This nonspecificity could also affect theability of distally conjugated ligands to capture their target(as discussed in the next section).

3.3. Plasminogen adsorption from plasma

Data on plasminogen adsorption from plasma are summa-rized in Fig. 2. The PU, PU-PEG and PU-PEG-Lys(P) sur-faces do not possess affinity binding sites for plasminogen,and thus showed very low adsorption (below 35 ng cm�2);in particular, the PU-PEG surfaces showed the lowest adsorp-tion due to the repellent effect of PEG. Both of the e-NH2

exposed lysine surfaces showed significantly higher adsorp-tion than the precursor PEG surfaces, indicating specific bind-ing of plasminogen to e-NH2 free lysine residues. PU-PEG300-Lys adsorbed more plasminogen than PU-PEG1000-Lys, possibly due to the lower resistance ofPEG300.

As mentioned previously, little attention has been paidto the effects of PEG protein resistance on the binding oftarget proteins to surface-immobilized ligands when PEGis used as a spacer. In this work, the effect of PEG chainlength on plasminogen binding to distally attached lysineswas investigated. It seems likely that plasminogen will beresisted as it approaches the PU-PEG-Lys surfaces, andthat the longer PEG chains will have a greater inhibitoryeffect. Furthermore, the motions of the chains could also

Lysine density* (nmol/cm2)

PU-PEG-Lys

3.97 ± 0.10 0.515.04 ± 0.23 0.76

Fig. 2. Plasminogen adsorption from plasma on modified and unmodifiedsurfaces (3 h exposure; mean ± SD, n = 3).

Fig. 4. Clot formation in plasma expressed as absorbance at 405 nm vs.time for polyurethane and lysine-derivatized surfaces.

1868 D. Li et al. / Acta Biomaterialia 5 (2009) 1864–1871

slow plasminogen attachment and the longer PEG chainswould have more extensive motions than the shorter ones.Previous work indicated that these effects could be reducedby prolonging incubation time. Plasminogen adsorption onthe PU-PEG and PU-PEG-Lys surfaces was thereforeinvestigated as a function of incubation time from 3 to24 h.

As shown in Fig. 3, adsorption was more rapid initially(3–7 h) on the 300 Lys than on the 1000 Lys surface, presum-ably due to the different protein resistance of PEG300 andPEG1000. The levels were the same at longer times with aplateau of about 260 ng cm�2, suggesting the same lysinedensity on both surfaces and plasminogen saturation at thisdensity. It should be noted that both PU-PEG surfacesshowed very low adsorption independent of adsorptiontime. These data support the strategy of incorporating bio-active ligands via a PEG spacer to produce a surface withdual ‘‘bioinert” and ‘‘bioactive” functions.

3.4. Plasma clot lysis

Typical data on clot lysis are shown in Fig. 4. The onsetof coagulation is indicated by a steep rise in the absorbance

Fig. 3. Plasminogen adsorption from plasma as a function of adsorptiontime (mean ± SD, n = 3).

vs. time curve following recalcification of the plasma. ThePU surface showed a typical clot formation curve: a pla-teau in absorbance was reached and maintained indicatinga fully formed, stable clot. In contrast, for both PU-PEG-Lys surfaces the absorbance returned to baseline, indicat-ing that the clot formed and then was lysed by the actionof surface localized plasmin. It is evident that the clotwas lysed more rapidly on the PU-PEG300-Lys surface(�20 min) than on the PU-PEG1000-Lys surface(�40 min). This result reflects the plasminogen adsorptiondata, and indicates that the surface with the greater plas-minogen binding capacity showed more rapid clot lysis.

3.5. Western blots of adsorbed proteins

Immunoblots of proteins eluted from the surfacesfollowing a 3 h exposure to plasma were probed with anti-bodies directed against 21 different plasma proteins and areshown in Fig. 5. It should be emphasized that the bandscannot be compared quantitatively among proteins withina given blot since the intensities depend on the sensitivitiesof the antibodies and the staining reactions. However, theintensity of a given protein on one surface should be com-parable to the intensity of the same protein on anothersurface.

From an overall perspective, the unmodified PU, con-sidered as a control surface, appeared to adsorb the great-est amounts of the proteins. Also it is evident that the PU-PEG surface adsorbed the smallest amounts, again indicat-ing the protein-resistant properties of PEG. PU-PEG-Lys(P) showed relatively more and stronger bands thanits precursor (PU-PEG), presumably due to the increasedhydrophobicity of this surface.

The contact phase coagulation proteins (factor XI, fac-tor XII, prekallikrein, HMWK) were either absent or unde-tectable. The complement proteins investigated include C3,factor B, factor H and factor I. Only C3 was detected, andshowed two bands at �70 and �45 kDa on all surfaces.The band at �40 kDa indicates activation of the comple-ment system [20]. It is well-known that surfaces containingnucleophiles such as hydroxyl and amino groups tend tobind C3b covalently and promote formation of the C3

Fig. 5. Immunoblots of plasma proteins eluted from (a) PU, (b) PU-PEG300, (c) PU-PEG300-Lys(P) and (d) PU-PEG300-Lys after exposure to 100%plasma for 3 h. The molecular weight scale is in kilodaltons.

D. Li et al. / Acta Biomaterialia 5 (2009) 1864–1871 1869

and C5 convertases [21]; thus it is possible that, at least inpart, C3 is adsorbed to the PU-PEG-Lys surface and acti-vated via the amino groups. For albumin, all four surfacesshowed bands at �66 kDa representative of the intact mol-ecule. For vitronectin, although two bands, at �70 kDa(intact molecule) and �60 kDa, were observed on the blotsfor all the surfaces, they appeared to be much less intenseon PU-PEG relative to the others. Apolipoprotein A-I gavea strong response at �27 kDa on all surfaces. Corneliuset al. [22] showed that this protein is an important compo-nent of the blood-derived protein layer on biomaterials,although its role in blood–biomaterial interactions isunclear. The PU-PEG surface did not resist apolipoproteinA-I, possibly due to the small size of this protein allowing itto penetrate to the underlying substrate.

Plasminogen is of the greatest importance in this studybecause of its central role in fibrinolysis. Only the PU-PEG-Lys blot showed clear evidence of plasminogenadsorption, with a strong band at �94 kDa, correspondingto the intact form of plasminogen, again indicating the spe-cific affinity of e-NH2-free lysine for plasminogen.

Fibrinogen is also important due to its role in bloodcoagulation and platelet adhesion. Strong bands were

detected at about 68, 56 and 48 kDa on the PU blot, corre-sponding to the a, b and c chains, respectively. Similar tothe trend observed in the radiolabeling data, fibrinogenwas completely resisted in plasma by the PEG-grafted sur-face and only weak responses were observed in the blots forPU-PEG-Lys(P) and PU-PEG-Lys. The trace amounts offibrinogen adsorbed on the latter two surfaces are appar-ently not sufficient to support significant platelet adhesion,as shown in the next section. All other proteins were unde-tectable on any surface except for IgG, which showed weakresponses on PU and PU-PEG-Lys.

As shown in the blots, most of the plasma proteins gavevery weak responses on the PU-PEG surface. Thesebecame stronger after further modification despite the pres-ence of PEG. However, compared with the PU blot, pro-tein responses on the blots of PU-PEG-Lys(P) and PU-PEG-Lys were diminished.

3.6. Platelet adhesion and activation

Platelet interactions are also important for blood com-patibility and it is generally believed that platelet adhesionon artificial surfaces is mediated by adsorbed proteins,

Fig. 6. Fibrinogen adsorption and platelet adhesion from whole blood topolyurethane control and modified surfaces under flow conditions: 300 s�1

shear rate, 15 min exposure (mean ± SD, n = 3).

1870 D. Li et al. / Acta Biomaterialia 5 (2009) 1864–1871

especially fibrinogen. Experiments measuring simulta-neously fibrinogen adsorption and platelet adhesion fromwhole blood were conducted using a cone-and-plate appa-ratus which allows for well-defined and easily variable flowconditions. Typical data are shown in Fig. 6. It is evidentthat fibrinogen adsorption and platelet adhesion show par-allel trends, confirming a strong correlation between theseresponses. It appears that fibrinogen is the most importantadhesive protein in this regard [23]. All of the modified sur-faces showed much less fibrinogen adsorption and plateletadhesion than the unmodified PU and the levels on lysine-derivatized PU were similar to those on PEG-grafted PU.It should be noted that platelet adhesion was relativelylow on the unmodified PU surface compared with valuesreported by others [24], possibly due to the particular bloodsamples used. However, the effects of the surface modifica-tions are clear.

Fig. 7. SEM images of platelet adhesion after 15 min of exposure to whole bloPU-PEG300-Lys. Scale bars = 10 lm.

SEM studies were undertaken to assess the morphologyof the platelets adherent on the PU and modified surfaces.As shown in Fig. 7, the surface distribution of platelets onthe unmodified PU surface was non-random, perhapsreflecting chemical heterogeneity. The adherent plateletswere generally activated and aggregated to some extent(inset, Fig. 7a). Some platelets appeared to be in the earlystages of spreading and pseudopod extension, and somewere in the fully spread state.

In agreement with the data from radiolabeling, plateletadhesion was very low on all of the modified surfaces, sug-gesting the relative absence of adhesive proteins such asfibrinogen, vitronectin and fibronectin. It is likely that theplatelet resistant properties shown by the lysine-derivatizedsurfaces may be attributed to the presence of PEG, whichprevents adsorption of adhesive proteins. It can also beconcluded that neither surface-bound lysine nor adsorbedplasminogen promotes platelet adhesion.

4. Summary and conclusions

Fibrinolytic polyurethane surfaces were prepared byconjugating e-amino-free lysine to the distal terminus ofsurface-grafted PEG. PEGs of different molecular weight,PEG300 and PEG1000, were used to investigate theeffects of PEG chain length on both the resistance to non-specific protein adsorption and the specific binding ofplasminogen. Protein adsorption results showed that bothPU-PEG300-Lys and PU-PEG1000-Lys appeared to befibrinogen resistant while the former adsorbed plasmino-gen from plasma more rapidly than the latter. In addition,PU-PEG300-Lys dissolved clots formed on the surfacemore effectively than PU-PEG1000-Lys. Immunoblots ofproteins eluted from the surfaces after plasma contact

od at 300 s�1 on (a) PU, (b) PU-PEG300, (c) PU-PEG300-Lys(P) and (d)

D. Li et al. / Acta Biomaterialia 5 (2009) 1864–1871 1871

showed that only the PU-PEG-Lys surface adsorbed plas-minogen; most of the other plasma proteins were resistedby this surface. The PEG spacer was also shown to assistin the prevention of platelet adhesion and activation inflowing whole blood. Neither surface-bound lysine noradsorbed plasminogen promoted platelet adhesion. Theresults of this study suggest that, when used both as aspacer for a bioactive moiety and a protein-repellingagent, PEG with a relatively short chain may be optimal.More specifically, the PEG–lysine surface system withdual ‘‘bioinert” and ‘‘bioactive” functions appears to bea promising approach to surfaces of improvedhemocompatibility.

Acknowledgements

This work was supported by the Canadian Institutes ofHealth Research (CIHR), the National Natural ScienceFoundation of China (20634030, 20574055), the Ministryof Science and Technology of China (2008CB617510) andProgram for New Century Excellent Talents in University(NCET0606055).

References

[1] Chen H, Yuan L, Song W, Wu Z, Li D. Biocompatible polymermaterials: role of protein–surface interactions. Prog Polym Sci2008;33:1059–87.

[2] Chen H, Chen Y, Sheardown H, Brook MA. Immobilization ofheparin on a silicone surface through a heterobifunctional PEGspacer. Biomaterials 2005;26:7418–24.

[3] Chen H, Zhang Y, Li D, Hu X, Wang L, McClung WG et al. Surfaceshaving dual fibrinolytic and protein resistant properties by immobi-lization of lysine on polyurethane through a PEG spacer. J BiomedMater Res doi: 10.1002/jbm.a.32152.

[4] Kang IK, Kwon OH, Kim MK, Lee YM, Sung YK. In vitro bloodcompatibility of functional group-grafted and heparin-immobilizedpolyurethanes prepared by plasma glow discharge. Biomaterials1997;18:1099–107.

[5] Li XL, Ji J, Pu M, Shen JC. Promote endothelialization of PETsurface by the grafting of PEG terminated by RGD. Tissue Eng2006;12(4):985–6.

[6] Park KD, Okano T, Nojiri C, Kim SW. Heparin immobilization ontosegmented polyurethaneurea surfaces – effect of hydrophilic spacers. JBiomed Mater Res 1988;22(11):977–92.

[7] Angles-Cano E. Overview on fibrinolysis: plasminogen activationpathways on fibrin and cell surfaces. Chem Phys Lipids 1994;67–68:353–62.

[8] Woodhouse KA, Brash JL. Plasminogen adsorption to sulfonatedand lysine derivatized model silica glass materials. J Colloid InterfaceSci 1994;164:40–7.

[9] Woodhouse KA, Weitz JI, Brash JL. Lysis of surface-localized fibrinclots by adsorbed plasminogen in the presence of tissue plasminogenactivator. Biomaterials 1996;17(1):75–7.

[10] McClung WG, Clapper DL, Hu SP, Brash JL. Adsorption ofplasminogen from human plasma to lysine-containing surfaces. JBiomed Mater Res 2000;49(3):409–14.

[11] McClung WG, Clapper DL, Anderson AB, Babcock DE, Brash JL.Interactions of fibrinolytic system proteins with lysine-containingsurfaces. J Biomed Mater Res 2003;66A:795–801.

[12] Gombotz WR, Guanghui W, Hoffman AS. Immobilization ofpoly(ethylene oxide) on poly(ethylene terephthalate) using a plasmapolymerization process. J Appl Polym Sci 1989;37:91–107.

[13] Gombotz WR, Wang GH, Horbett TA, Hoffman AS. Proteinadsorption to poly(ethylene oxide) surfaces. J Biomed Mater Res1991;25(12):1547–62.

[14] Moon JH, Kim JH, Kim KJ, Kang TH, Kim B, Kim CH, et al.Absolute surface density of the amine group of the aminosilylatedthin layers: ultraviolet–visible spectroscopy, second harmonic gener-ation, and synchrotron-radiation photoelectron spectroscopy study.Langmuir 1997;13(16):4305–10.

[15] Cornelius RM, Brash JL. Identification of proteins absorbed tohemodialyser membranes from heparinized plasma. J Biomater SciPolym Ed 1993;4(3):291–304.

[16] Mustard JF, Perry DW, Ardlie NG, Packham MA. Preparation ofsuspensions of washed platelets from humans. Br J Haematol1972;22(2):193–204.

[17] Dosne AM, Drouet L, Dassin E. Usefulness of 51 chromium-plateletlabelling for the measurement of platelet deposition on subendothe-lium. Microvasc Res 1976;11:111–4.

[18] Skarja GA, Kinlough-Rathbone RL, Perry DW, Rubens FD, BrashJL. A cone-and-plate device for the investigation of platelet bioma-terial interactions. J Biomed Mater Res 1997;34(4):427–38.

[19] Unsworth LD, Sheardown H, Brash JL. Protein resistance of surfacesprepared by sorption of end-thiolated poly(ethylene glycol) to gold:effect of surface chain density. Langmuir 2005;21(3):1036–41.

[20] Mulzer SR, Brash JL. Identification of plasma proteins adsorbed tohemodialyzers during clinical use. J Biomed Mater Res1989;23(12):1483–504.

[21] Gorbet MB, Sefton MV. Biomaterial-associated thrombosis: roles ofcoagulation factors, complement, platelets and leukocytes. Biomate-rials 2004;25:5681–703.

[22] Cornelius RM, Archambault J, Brash JL. Identification of apolipo-protein A-I as a major adsorbate on biomaterial surfaces after bloodor plasma contact. Biomaterials 2002;23(17):3583–7.

[23] Tsai WB, Grunkemeier JM, McFarland CD, Horbett TA. Plateletadhesion to polystyrene-based surfaces preadsorbed with plasmasselectively depleted in fibrinogen, fibronectin, vitronectin, or vonWillebrand’s factor. J Biomed Mater Res 2002;60(3):348–59.

[24] Tan J. PEO-containing block copolymers as surface modificationadditives in polyurethanes for protein and cell resistance. Ph.D.Thesis, McMaster University, Hamilton, Ontario, Canada; 2004.