Supramolecular design of self-assemblingnanofibers for cartilage regenerationRamille N. Shaha,b, Nirav A. Shahc, Marc M. Del Rosario Limd, Caleb Hsieha,b, Gordon Nubere, and Samuel I. Stuppa,b,f,g,1

aInstitute for BioNanotechnology in Medicine, Northwestern University, 303 E. Superior Street 11th floor, Chicago, IL 60611; bDepartment of MaterialsScience and Engineering, Northwestern University, 2220 Campus Drive, Evanston, IL 60208; cDepartment of Orthopaedic Surgery, Northwestern University,676 N. Saint Clair, Suite 1350, Chicago, IL 60611; dDepartment of Biomedical Engineering, Northwestern University, 2145 Sheridan Road, Evanston, IL60208; eNorthwestern Orthopaedic Institute, 680 N. Lakeshore Dr., Suite 1028, Chicago, IL 60611; fDepartment of Chemistry, Northwestern University,2145 Sheridan Road, Evanston, IL 60208; and gDepartment of Medicine, Northwestern University, 251 East Huron Street, Suite 3-150, Chicago, IL 60611

Edited by Robert Langer, Massachusetts Institute of Technology, Cambridge, MA, and approved December 23, 2009 (received for review June 12, 2009)

Molecular and supramolecular design of bioactive biomaterialscould have a significant impact on regenerative medicine. Idealregenerative therapies should be minimally invasive, and thusthe notion of self-assembling biomaterials programmed to trans-form from injectable liquids to solid bioactive structures in tissueis highly attractive for clinical translation.We report here on a coas-sembly system of peptide amphiphile (PA) molecules designed toform nanofibers for cartilage regeneration by displaying a highdensity of binding epitopes to transforming growth factor β-1(TGFβ-1). Growth factor release studies showed that passiverelease of TGFβ-1 was slower from PA gels containing the growthfactor binding sites. In vitro experiments indicate these materialssupport the survival and promote the chondrogenic differentiationof human mesenchymal stem cells. We also show that thesematerials can promote regeneration of articular cartilage in a fullthickness chondral defect treated with microfracture in a rabbitmodel with or even without the addition of exogenous growthfactor. These results demonstrate the potential of a completelysynthetic bioactive biomaterial as a therapy to promote cartilageregeneration.

self-assembling biomaterials ∣ chondral defects ∣ microfracture ∣peptide amphiphiles ∣ transforming growth factor

Damaged articular cartilage in our joints is an importantregenerative medicine target because adults lack the ability

to effectively form cartilage with the architecture and morphol-ogy of the native tissue. This can eventually lead to joint pain withloss of physical function (1–3), a serious health care issue in anaging and physically active global population. Full thickness focalchondral lesions may progress to osteoarthritis that has anestimated economic impact approaching $65 billion in the USalone when considering healthcare costs, loss of wages, and so-cietal impact costs (4). With the limited natural healing capabilityof articular cartilage, clinical intervention is necessary to preventfurther articular cartilage degradation and early progression ofdegenerative osteoarthritis.

Microfracture is a common clinical procedure used for the re-pair of cartilage defects (5). The benefits of microfracture includethe fact that it is a single-stage procedure, is relatively simple froma technical point of view, is cost-effective with low patient morbid-ity, and involves the patients’ ownmesenchymal stem cells (MSCs)as a cell source to stimulate cartilage repair. The reparative pro-cess in microfracture involves a clot of pluripotent MSCs thatadhere to the subchondral bone. Histological assessment ofmicrofracture in animal (6, 7) and clinical testing (5) have shownthat most lesions form fibrous cartilage with predominant type Icollagen and a limited amount of type II collagen present. Addi-tionally, there is a significant drop in clinical outcome scores after18 months as well as in patients older than 40 years (8). This sug-gests that there is deficient bioactivity, quantity, quality, and reten-tion of chondrocyte phenotype within the repair tissue. An idealregenerative medicine solution to augment this current clinicalprocedure would be a bioactive scaffold capable of being im-

planted throughminimally invasivemeans that localizes andmain-tains cells and growth factors within the defect site, promotes stemcell chondrogenic differentiation, and stimulates biosynthesis.

In prior work, we developed self-assembling biomaterialsbased on a broad class of molecules known as peptide amphi-philes (PAs) that self-assemble from aqueous media into supra-molecular nanofibers of high aspect ratio. These molecules,targeted to serve as the components of artificial extracellular ma-trices, consists of a peptide segment covalently bonded to a morehydrophobic segment such as an alkyl tail (9–12). PAs are nor-mally charged molecules so that screening ions in the biologicalenvironment can trigger self-assembly into cylindrical nanofibers,which form by hydrogen bonding among peptide segments intoβ-sheets and the hydrophobic collapse of their alkyl segments(13). Furthermore, PAs can have a terminal biosignaling peptidedomain, which upon self-assembly becomes exposed in very highdensities on the surfaces of the nanofibers. At specific pH valuesand PA concentrations, these amphiphilic molecules can assem-ble into self-supporting gels made up of an interconnected net-work of nanofibers (12). Over time, PA gels should biodegradeinto amino acids and lipids that can be safely cleared by the body.

We had previously discovered by phage display a peptide se-quence (HSNGLPL) with a binding affinity to transforminggrowth factor β1 (TGF-β1), which is known to be important inthe formation of connective tissues and for other biologicalfunctions (14). Prior work has demonstrated that TGF-β1 playsa significant role in the regulatory network of growth factors thatmaintains articular cartilage in the differentiated phenotype (15)and is a critical factor for inducing chondrogenesis in marrow-derived MSCs (16). Furthermore, in articular cartilage tissueengineering TGF-β1 has been shown to increase collagen andproteoglycan production and inhibit matrix breakdown (17).The effectiveness of TGF-β1, however, has also been demon-strated to be significantly dependent on the delivery kineticsand/or simultaneous delivery with other proteins (18, 19).

In this study, we designed and evaluated in vivo a unique self-assembling PA molecule, which includes a TGF-binding domain(Fig. 1A) for specific use in articular cartilage regeneration. Thisparticular element in artificial matrix design, that is, the bindingof growth factors to components of the extracellular space such ascollagens and heparin, have been shown to occur naturally inbiological systems (20, 21). The matrix studied here also con-tained a nonbioactive PA of smaller molecular dimensions(Fig. 1B) that coassembles with the TGF-binding molecules.

Author contributions: R.N.S., N.A.S., and G.N. designed research; R.N.S., N.A.S., M.M.D.R.L.,and C.H. performed research; R.N.S., N.A.S., M.M.D.R.L., and C.H. analyzed data; andR.N.S., N.A.S., and S.I.S. wrote the paper.

The authors declare no conflict of interest.

This article is a PNAS Direct Submission.1To whom correspondence should be addressed at: Northwestern University, Cook Hall,Room 1127, 2220 Campus Drive, Evanston, IL 60208. E-mail: s-stupp@northwestern.edu.

This article contains supporting information online at www.pnas.org/cgi/content/full/0906501107/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.0906501107 PNAS ∣ February 23, 2010 ∣ vol. 107 ∣ no. 8 ∣ 3293–3298

APP

LIED

BIOLO

GICAL

SCIENCE

SSP

ECIALFEAT

URE

Dow

nloa

ded

by g

uest

on

Mar

ch 2

4, 2

020

The supramolecular concept in designing bioactivity in thesematerials was to coassemble both molecules so that the bindingepitope could adequately capture and display the growth factorfor signaling (Fig. 1C). In vitro studies were performed to estab-lish the ability of these PA systems to support mesenchymal stemcell viability and chondrogenic differentiation. Furthermore, weutilized an in vivo chondral defect microfracture model in rabbitsto test the ability of these systems to promote hyaline cartilageregeneration. To our knowledge, this is the first study to investi-gate the potential use of designed (growth factor binding) self-assembling PA systems for the treatment of articular cartilagedefects in an in vivo model.

Results and DiscussionGrowth Factor Release Kinetics.Growth factor release studies wereperformed to determine if the presence of binding epitopes toTGFβ-1 on the PA nanofibers are able to slow the release ofthe growth factor from the gel. Gels (n ¼ 4) containing 10 mol% TGFBPA (mixed with the filler PA without the binding epi-tope) were compared to gels made up of 100% filler PA. PA gelswere loaded with 1 μg∕mL of TGFβ-1 and incubated in a buffersolution. The buffer was collected and replaced at 6, 24, 48, and72 h and the quantity of TGFβ-1 in the collected aliquots wasassayed by ELISA. Results revealed release of the growth factorfrom both the filler and TGFBPA gels, with a slower release ofgrowth factor from gels containing the TGF-binding epitopes(Fig. 1D). After 72 h, the cumulative percentage of TGFβ-1 re-leased from the filler PA gel was 3-fold greater (approximately60%) compared to the TGFBPA gel (approximately 20%).The slower release of TGFβ-1 from TGFBPA gels suggestssuccessful binding is occurring between the growth factor and epi-topes, which may help localize the growth factor and prolong itsrelease at defect sites for enhancing tissue regeneration.

In Vitro Viability and Differentiation of Human MSCs Cultured WithinPeptide Amphiphile Gels. In vitro studies were necessary to ensurethat PA gels did not cause cell death or inhibit MSC differentia-tion. MSCs were cultured in gels of nonbioactive filler PA or gelscontaining 5 or 10 mol% TGF-binding PA (TGFBPA) mixed withfiller PA. Gels alone or gels mixed with 100 ng∕mL of recombi-nant human (rh)TGF-β1 were also compared. A Live/Dead stain(Invitrogen) showed that MSCs remained viable within PA gelsthroughout the culture period (Fig. 2A). Scanning electron

microscopy (SEM) revealed cells extended their processes tointeract with the surrounding PA nanofiber matrix (Fig. 2B).When grown in chondrogenic media, most MSCs obtained amore rounded phenotype as expected, indicating that the PA gelssupported and did not inhibit MSC differentiation.

Gene expression analysis for cartilage markers (aggrecanand type II collagen) at 3 and 4 wks showed upregulation of thesegenes in both filler and TGFBPA that incorporated 100 ng∕mLrhTGF-β1 within the gels (Fig. 2C and Fig. S1). At 3 wks, glyco-saminoglycan (GAG) content in the scaffolds was also assessedand showed significantly higher (p < 0.05) GAG production inthe PA gels that contained supplemented growth factor (Fig. 2D),but there was no apparent difference between the filler andTGFBPA at this time point. It is expected that the presence ofsupplemented TGFβ-1 in both the filler and TGFBPA wouldresult in chondrogenic differentiation of MSCs as long as thetherapeutic amount of growth factor is present. At 4 wks, how-ever, there was a significantly higher aggrecan expression levelfor the TGFBPA (10 mol%) compared to the filler PA(p < 0.03). This most likely is the result of prolonged and morelocalized delivery of the growth factor from the TGFBPA com-

Fig. 1. Design of PAs for articular cartilage regeneration. Chemical structureof (A) TGF-binding PA and (B) nonbioactive filler PA. (C) Illustration of co-assembly of the TGF-binding PA and the filler PA showing binding epitopesexposed on the surface of the nanofiber. (D) ELISA results showing TGFβ-1release up to 72 hours from filler PA and 10 mol% TGF-binding PA (TGFBPA).100 ng∕mL of TGFβ-1 were loaded in all gels. Error bars equal standarddeviation, n ¼ 4.

Fig. 2. In vitro viability and differentiation of hMSCs cultured in PA scaf-folds. (A) Live/dead images of cells cultured in PA gels (green ¼ live;red ¼ dead). (B) SEM of hMSC on nanofiber gel surface. (C) Aggrecan geneexpression from hMSCs cultures in PA gels at 2, 3, and 4 wks.Filler PA ¼ filler; 100 ng∕mL of TGF ¼ 100TGF; 5 mol% TGFBPA ¼5%TGFBPA; 10 mol% TGFBPA ¼ 10%TGFBPA. (D) GAG per DNA quantifica-tion in digested PA gels at 3 wks for filler, fillerþ 100TGF, and 10%TGFBPA þ100TGF groups. Error bars equal standard deviation, n ¼ 3.

3294 ∣ www.pnas.org/cgi/doi/10.1073/pnas.0906501107 Shah et al.

Dow

nloa

ded

by g

uest

on

Mar

ch 2

4, 2

020

pared to the filler PA (as supported by the growth factor releaseexperiments), and it is logical that the biological effect of thisdifference is observed at later time points when therapeutic dosesof growth factor are still maintained within the TGFBPA.Collagen type II expression results revealed a significant effectof time (p < 0.0001) and rhTGF-β1 supplementation (p <0.0001), but there were no significant differences in type IIcollagen expression between the PA gels with and without theTGF-binding epitope at 4 wks (Fig. S1). It is speculated, however,that longer term in vitro culture may lead to differential type IIcollagen expression between the PA groups. Based on the geneexpression data, the filler PA and the 10 mol% TGF-bindingPA were chosen for in vivo evaluation.

In Vivo Evaluation of PAs in a Full Thickness Articular Cartilage DefectRabbit Model Treated with Microfracture. Full thickness chondraldefects treated with microfracture in the trochlea of adult rabbitswere used to evaluate the potential of PAs to promote cartilageregeneration in the presence of bone marrow-derived stem cells(Fig. 3). 10% pyrene-labeled PA was used in preliminary studiesto assess the retention and adherence of the PA within the defectsafter application. Fluorescence was detected under a UV lampand revealed that the PA is successfully contained within thedefect following application (Fig. 3D). A preliminary 4-weekstudy showed no obvious chronic inflammatory response inPA-treated defects—no apparent swelling, redness, or synovialhypertrophy of the joint, as well as negative staining of the tissuesections for CD4, an inflammatory cell marker specificallyfor T helper cells (Fig. S2).

A 12-week study was performed comparing the followinggroups: (i) control group treated with 10 μL of rhTGF-β1(100 ng∕mL) per defect (TGF); (ii) defects treated with thenonbioactive filler PA þ 100 ng∕mL rhTGF-β1 (filler/TGF); (iii)defects treated with 10 mol% TGFBPA mixed with the fillerPA þ 100 ng∕mL rhTGF-β1 (TGFBPA/TGF); and (iv) defectstreated with 10 mol% TGFBPA mixed with the filler PA withoutgrowth factor (TGFBPA). At the end of the study, all rabbits ineach group appeared to have full range of motion of their knees.One rabbit knee was noted to have a dislocation of the patellathat was not recognized until the day of sacrifice, and thereforethis sample was not included in the analysis. As in the 4-week

study, none of the rabbits in any of the groups developed grosslyapparent degeneration or synovial hypertrophy of the joint at12 wks. Macroscopic observation of defects after 12 wks revealedno obvious differences between defects treated with the growthfactor alone and ones treated with the nonbioactive filler PA withregard to tissue fill or appearance of repair tissue (Fig. 4A and B).In these groups, defects in the trochlea were still very obvious,with defined defect boundaries and noticeable color and texturedifferences compared to the surrounding cartilage tissue. In greatcontrast, defects treated with the TGF-binding PA with and with-out growth factor (TGFBPA/TGF and TGFBPA) showed nearlycomplete tissue fill in most defects, and the formed tissue wassimilar in color and texture to the surrounding cartilage (Fig. 4Cand D). In some of these defects, the defect boundaries couldbarely be distinguished, indicating excellent integration of theregenerated tissue to the surrounding cartilage.

Qualitative assessment of histological sections of the growthfactor alone treated and filler PA/TGF groups revealed incom-plete fill of the defects and repair tissue that was not integratedwith the surrounding cartilage and/or the subchondral bone(Fig. 5A and B). Tissue in defects treated with growth factor aloneresembled more fibrocartilage with cells having a fibroblast-likemorphology, abnormal cell density and organization, and little tono staining for glycosaminoglans (GAGs, Safranin-O stain)(Fig. 5A) and type II collagen (Fig. 5E). Defects treated withthe filler PA with growth factor also showed little staining forGAGs (Fig. 5B), but did show some positive staining for typeII collagen (Fig. 5F). In comparison to the TGF group, somecells in the repair tissue of the filler PA/TGF group had a morerounded chondrocyte-like morphology, cells located in lacunae,and some cell clustering. Cell density and organization within thisgroup, however, was still abnormal compared to the surroundingcartilage (Fig. 5B and F). These results indicate that the presenceof a nanofiber scaffold without any bioactive epitope can stillhave an enhanced effect on cell differentiation and synthesisof articular cartilage specific matrix molecules (in this case typeII collagen). This may be due to increased localization of thegrowth factor within the defect (mechanically trapped withinthe PA gel) compared to treatment with growth factor solu-tion alone.

Fig. 3. Full thickness articular cartilage defect microfracture rabbit model.Surgical procedure creating (A) full thickness articular cartilage defects inrabbit trochlea using a microcurette and (B) microfracture holes throughthe subchondral bone using a microawl to induce bleeding into the defect.(C) PA gel in defect after injection (Arrow). (D) Pyrene-labeled PA gel illus-trating containment of the gel within articular cartilage defects afterinjection.

Fig. 4. Observation of articular cartilage defects 12 wks after treatment.Macroscopic views of articular cartilage defects after 12 wks postop treatedwith (A) 100 ng∕mL TGF-β1 (100TGF), (B) filler PAþ 100TGF, (C) 10%TGFBPAþ100TGF, and (D) 10%TGFBPA alone.

Shah et al. PNAS ∣ February 23, 2010 ∣ vol. 107 ∣ no. 8 ∣ 3295

APP

LIED

BIOLO

GICAL

SCIENCE

SSP

ECIALFEAT

URE

Dow

nloa

ded

by g

uest

on

Mar

ch 2

4, 2

020

As in the macroscopic observations, there was a significant en-hanced effect on tissue regeneration when using the PA systemcontaining the TGF-binding epitope (with and without TGFβ-1), resulting in more hyaline-like tissue formation (Fig. 5C andD). In these groups, tissue formed within the defect space showednearly complete fill to the level of the undamaged surroundingcartilage tissue. Furthermore, voids between the regeneratedtissue and surrounding cartilage or subchondral bone were notapparent, indicating excellent integration. GAG (Fig. 5C andD) and collagen type II (Fig. 5G and H) staining in these samplesnearly matched the intensity and quantity of the surroundingundamaged cartilage. Additionally, cell morphology resembleda chondrocyte morphology (rounded and in lacunae), cells wereorganized in a columnar architecture, and cell density and clus-tering were similar to the undamaged native cartilage. In somesamples, the boundaries of the defects could not even be identi-fied due to the close similarity in architecture and morphology tothe surrounding cartilage.

Quantitative scoring of histological sections (stained withSafranin-O) was performed using a modified O’Driscoll (22)24-point scoring system (see Table S1) that is commonly usedfor evaluating the quality of new tissue formation in articular car-tilage defects in vivo. The major categories scored in this systeminclude assessment of cellular morphology, Safranin-O staining,surface regularity, structural integrity, thickness, bonding toadjacent cartilage, hypocellularity, chondrocyte clustering, andfreedom from degenerative changes in adjacent cartilage. Theaverage scores in each category and total scores are presentedin Table 1 and the spread of the scores in each group is also pre-sented in Fig. 6 (n ¼ 8–10 defects). There was no statistically sig-nificant difference in histological scoring between groups treatedwith growth factor alone (15.5� 4.7) or with filler PA and growthfactor (15.1� 3.7). Both groups treated with the TGF-binding PA

with (21.9� 1.2) and without (20.8� 2.1) growth factor, how-ever, had higher scores in each O’Driscoll category and about1.5-fold higher total histological scores compared to the othertwo groups (p < 0.0001). Interestingly, there was no significantdifference between the groups treated with the TGF-bindingPA with or without growth factor. The significant enhancementin hyaline cartilage formation in defects without the additionof exogenous TGF-β1 may indicate that binding events of endo-genous TGF-β1 (i.e., from the bleeding marrow through themicrofracture holes or from the surrounding synovial fluid) tothe epitopes displayed by the supramolecular PA nanofibersare occurring in vivo, and at a level that increases the localconcentration of the protein within the defect space to see a re-generative response. Based on our in vitro growth factor releasestudies, there is greater retention (slower release) of TGF-β1within PA gels containing TGF-β1 growth factor binding sitescompared to the filler PA alone. This may be the reason forthe significant enhanced regeneration in the TGFBPA treateddefects compared to the ones treated with the filler PA. It is plau-sible that as long as the TGFBPA is present within the defects,there is continuous localization of TGF-β1 that can encouragechondrogenic differentiation of mesenchymal stem cells fromthe bone marrow and subsequent production of cartilage matrixmolecules. Future studies are warranted to investigate if bindingof TGF-β1 to the PAs would be affected in the presence of othergrowth factors or matrix molecules in vivo.

Not only is it possible for the TGF-binding epitope to prolongthe localization of growth factor within the defect, but it may alsohelp preserve the integrity and activity of the growth factor byprotecting it from proteolytic degradation through the molecu-larly specific binding events. For example, it is known that growthfactors that bind to heparin in the extracellular space (e.g., angio-genic growth factors) through highly specific heparin binding

Fig. 5. Histological evaluation of sample sections 12 wks after treatment. Safranin-O staining for glycosaminoglycans (A–D) and type II collagen staining (E–H)in articular cartilage defects treated with (A, E) 100 ng∕mL TGF-β1 (100TGF), (B, F) filler PAþ 100TGF, (C, G) 10%TGFBPAþ 100TGF, and (D, H) 10%TGFBPAalone 12 wks postop.

3296 ∣ www.pnas.org/cgi/doi/10.1073/pnas.0906501107 Shah et al.

Dow

nloa

ded

by g

uest

on

Mar

ch 2

4, 2

020

domains are protected from enzymatic degradation (23). Oneadditional observation in our study is the narrow spread in his-tological scores for specimens treated with the TGF-bindingPA compared to the broader spread in scores for the growth fac-tor alone or filler plus growth factor groups (Fig. 6). This suggestsmore consistent tissue formation within defects treated withthe TGF-binding PA, potentially demonstrating the dependablebioactivity of this PA system.

These results show that the healing of chondral defects treatedwith microfracture can be accelerated and enhanced with de-signed PA systems. Results from the TGFβ1 growth factor alonegroup (without any gel) and the filler PA þ growth factor treatedgroups were similar, and demonstrated that the PA gels do notimpede healing. The addition of the TGFβ1 binding epitopeson the PA molecules, however, significantly promoted the forma-tion of cartilage matrix rich in GAG with similar histological ap-pearance to the native cartilage tissue. These promising resultswarrant future in vivo studies in larger defects as well as in largeranimal models to further assess the regenerative potential of thistechnology.

ConclusionsWe have observed extensive cartilage regeneration in the pre-sence of supramolecular nanofibers designed to bind the growthfactor TGFβ-1 relative to a nonbioactive system. Our in vitro stu-dies showed that self-assembling PA scaffolds can support hMSCviability and chondrogenic differentiation and that the presenceof the binding epitope at a specific concentration (10 mol%) canbetter retain TGFβ-1 within gels, which can lead to upregulated

gene expression of cartilage specific markers over prolongedtimes. We demonstrated in vivo that PAs synthesized with a pep-tide binding sequence to TGF-β1 significantly enhanced the re-generative potential of microfracture-treated chondral defects.Most importantly, we were able to induce a significant regenera-tive response in vivo in the presence of marrow-derived mesen-chymal cells, without the addition of exogenous growth factor.Our findings demonstrate the potential of molecularly designedsupramolecular biomaterials in promoting a specific biologicalresponse without the need for exogenous growth factors or trans-planted cells. The chemical versatility of these peptide based self-assembling systems and the ability to potentially apply thesetherapies through minimally invasive injection into the joint spacemakes them promising therapeutic candidates to improve currentclinical cartilage repair strategies.

Materials and MethodsPeptide Synthesis and Purification. The synthesis of linear peptide sequenceswas performed using standard solid phase methods in an AppliedBiosystems 433A automated peptide synthesizer. Filler PA with sequenceH3CðCH2Þ14CO-VVVAAAEEE, was grown on a preloaded glutamic acid Wangresin, using 3.8 M equivalents of HBTU, 6.0 equivalents of diisopropylethyla-mine (DIEA) and 4 equivalents of each amino acid. A palmitoyl alkyl tail wasadded to the N-terminus of the peptide manually, by adding 4 M equivalentsof palmitic acid and the same molar equivalents of HBTU and DIEA as above.The PA was cleaved from the resin and amino acid side groups weredeprotected in 95% trifluoroacetic acid (TFA), 2.5% triisopropylsilane,2.5% deionized water. TFA was removed in a rotary evaporator and peptideswere collected by precipitation in cold diethyl ether.

TGF-binding sequences were previously designed in the Stupp laboratoryusing phage display methods (24, 25). For the TGFBPA with sequenceHSNGLPLGGGSEEEAAAVVVðKÞ-COðCH2Þ10CH3, a lysine with a dodecylamineside chain was reacted with a Rink amide resin. The remainder of the peptidewas then synthesized as described above, but yielding a PA with reverse po-larity (9). The resulting products were purified using standard preparativeHPLC methods.

Growth Factor Release Studies. One weight percent solutions of PA wereused for growth factor release studies by dissolving PA in aqueous solutionscontaining 1 μg∕mL of TGFβ-1. Filler PA gels were compared with gels con-taining 10 mol% TGFBPA. PA gels were made by injecting 100 μL of the PAsolution in 200 μL PBS containing 25 mM calcium chloride and 50 μg∕mL ofBSA and incubating at 37 °C for 1 h. After 1 h, 200 μL of buffer (PBS contain-ing 50 μg∕mL BSA) was added to each gel. The buffer was collected andreplaced at 6, 24, 48, and 72 h. TGFβ-1 released in the buffer solutionswas quantified by an Enzyme-Linked Immunosorbent Assay (eBioscience).

In Vitro Cell Viability and Differentiation Studies. Human mesenchymal stemcells (hMSCs) from one donor (Lonza) were used at passage 5 for culture with-in PA scaffolds at 40,000 cells per 20 μL gel (n ¼ 3). For cell encapsulation, cellsuspensions were mixed with aqueous PA solutions (pH 7) and 20 μL aliquotsof the PA/cell suspension were injected into 200 μL of MSC growth media(Lonza) supplemented with 25 mM calcium chloride to induce gelation (ina 96-well U-bottom plate). For differentiation studies, media was changedto a serum-free chondrogenic media containing high glucose DMEM (highglucose 4.5% without L-glutamine), 0.1 mM nonessential amino acids,10 mM Hepes buffer, 100 U∕mL penicillin, 100 μg∕mL streptomycin gluta-

Table 1. Histological scores for TGF, filler PAþ TGF, 10% TGFBPAþ TGF, and 10% TGFBPA groups at 12 wks

TGF Filler PAþ TGF 10% TGFBPAþ TGF 10% TGFBPA

Category Mean SD Mean SD Mean SD Mean SD

Cellular morphology 2.56 1.43 2.67 1.38 3.93 0.22 3.80 0.45Safranin-O staining of the matrix 1.48 0.60 1.21 0.67 2.44 0.24 2.30 0.58Surface regularity 2.00 0.80 1.67 0.69 2.41 0.49 2.37 0.51Structural integrity 1.07 0.62 0.88 0.40 1.70 0.31 1.57 0.32Thickness 1.15 0.60 1.29 0.55 1.85 0.18 1.77 0.32Bonding to the adjacent cartilage 1.33 0.53 1.13 0.62 1.81 0.18 1.80 0.23Hypocellularity 1.96 0.72 2.08 0.53 2.89 0.17 2.63 0.40Chondrocyte clustering 1.26 0.40 1.46 0.25 1.93 0.15 1.73 0.31Freedom From Degenerative Changes in Adjacent Cartilage 2.70 0.35 2.75 0.24 2.89 0.33 2.83 0.24Total 15.52 4.74 15.13 3.66 21.85 1.19 20.80 2.06

Fig. 6. Histological scores of 12 wk in vivo samples showing significantlyhigher scores for the groups treated with the 10% TGF-binding PA withor without growth factor. Circles represent scores for individual specimensin each group (n ¼ 8–10). Of note is the narrow distribution of scores forthe defect groups treated with the 10% TGFBPA compared to the widerspread in scores for those treated with 100 ng∕mL rhTGF-β1 (100TGF) aloneor filler PAþ 100TGF.

Shah et al. PNAS ∣ February 23, 2010 ∣ vol. 107 ∣ no. 8 ∣ 3297

APP

LIED

BIOLO

GICAL

SCIENCE

SSP

ECIALFEAT

URE

Dow

nloa

ded

by g

uest

on

Mar

ch 2

4, 2

020

mate, ITSþ1 (100x, by Sigma Chemical), 0.1 mM ascorbic acid 2-phosphate,1.25 mg∕mL BSA, 10 ng∕mL of TGF-β1, and 100 nM dexamethasone. Mediawas completely changed every 2–3 d (200 μL of media per change) and gelswere cultured up to 4 wks.

Viability of cells was assessed by fluorescence microscopy using a Live/Dead stain (Invitrogen). At the end of the culture period, total RNA was ex-tracted from each sample with TRIzol and reverse-transcribed into cDNAusing the SuperScript® III First-Strand Synthesis kit (Invitrogen). Real-timePCRwas performedwith the BioRad iQ5 Real-Time PCR system and Promega’sPlexor® qPCR kit. Expression of Aggrecan (AGC1) and Collagen II Alpha I(COL2A1) was used to assess chondrogenic differentiation, and Glyceralde-hyde 3-phosphate dehydrogenase was as the housekeeping gene.

For GAG quantification after 3 wks in culture, the PA gels containingMSCswere first digested in a papain solution (100 μL) consisting of 0.01 ML-cysteine and 0.5% papain (25 mg∕mL) dissolved in phosphate bufferedEDTA (0.04 M Na2HPO4, 0.06 NaH2PO4· H2O, 0.01 M Na2EDTA · 2H2O) ad-justed to a pH of 6.5. Samples were allowed to digest for 24 h in a 60 °C waterbath. Following digestion, 20 μL from each sample was added to a 96 wellplate. 200 μL of a dimethylmethylene blue solution was added and mixedper well, and absorbance was read at 535 nm. GAG quantities were obtainedusing a chondroitin sulfate standard curve and normalized to DNA contentobtained from a Picogreen assay (Invitrogen).

In Vivo Full Thickness Articular Cartilage Rabbit Model Treated with Microfrac-ture. Ten adult male New Zealand White Rabbits (3–3.5 kg, approximately6 months old) were anesthetized by intramuscular injection of ketamineat 30–40 mg∕kg and xylazine 5–7 mg∕kg. Isoflourane (1–3%) and oxygenwere supplied by face mask for sedation and general anesthesia during theentire procedure. Under sterile aseptic technique, a midline 2 cm incision wasmade with the knee flexed at about 20 ° and subsequently a medial para-patellar capsulotomy was performed and the patella was translated laterallyto expose the articular surface of the trochlea. Two 2 mm diameter fullthickness chondral defects followed by microfracture were created in therabbit trochlea (proximal–medial and distal–lateral). The articular cartilage,including the calcified cartilage layer, was removed with a micro-curette(Fig. 3A). Care was taken not to disrupt the underlying subchondral plate,and sharp (perpendicular) defect edges were created with the aid of a dermalpunch. To allow bone marrow MSCs into the defect space, microfracture wasperformed within the defects by creating 3 holes spaced equally apart and2 mm deep into subchondral bone with a micro-awl (Fig. 3B). Marrow bloodwas observed emerging out of each microfracture hole. After application ofthe PA solution (1 wt%) within the defect, self-assembly of the supramo-lecular nanofibers into a gel network is ensured by adding an aliquot of

dilute calcium chloride solution. Calcium ions are used to screen the negativecharges on the PA molecules to induce self-assembly. Before the addition ofcalcium chloride, gelation of the PA solution was already triggered by elec-trolytes naturally present in marrow blood from the microfracture holes(Fig. 3C). Postoperatively each rabbit was given IM antibiotics (Baytril 72 hduration) and IM pain medicine (Buprenex 24 h duration). Signs of infectionand the ability of the rabbits to bear weight and move within their cageswere evaluated. A 12-week study was performed comparing the four condi-tions described previously using 10 rabbits. For each trochlea, 2 defects werecreated and treated the same, and in each rabbit two different conditionswere tested (i.e., one treatment in one knee and a different treatment inthe other knee). Overall, there were 10 defects per experimental condition.At each end point, rabbits were euthanized by injection of pentobarbitalintravenously and secondary measures of bilateral thoracotomy. The distalfemur was harvested and processed for histological analysis.

Histological Analysis. Extracted specimens were fixed in 10% neutral bufferedformalin, decalcified for 24 h, and tissue processed for paraffin embedding.Four-micrometer-thick sections were obtained from the center cross sectionsof the defects (1,000 μm from the defects interface) and histochemicallystained for hemotoxylin and eosin and Safranin-O/Fast Green, and immuno-histochemically stained for type II collagen. Immunohistochemical stainingfor CD4þ cells was also used in a preliminary 4-week study to assess immuneresponse to the PA gels (5 rabbits were used in this experiment). For the12-week study, scoring of histological sections was performed by 3 indepen-dent, blinded observers using a 24-point scale (Table S1).

Sample Size Determination and Statistical Analysis. Sample size was based onan a priori power analysis using alpha <0.05 (CI ≥ 95%) and 1 − β ≥ 0.8. Aminimal sample size of five (defects) was required per group. Significantdifferences were evaluated with ANOVA (F < 0.05) testing and a leastsignificant difference post hoc test (p < 0.05). Each defect was consideredan independent sample. Data are expressed as mean �SD.

ACKNOWLEDGMENTS. We thank A. Cheetham for PA synthesis, A. Mata forSEM, M. Seniw for assistance with illustrations, the Center for ComparativeMedicine (Northwestern University) for assistance with animal surgeries,Pathology Core (Robert H. Lurie Comprehensie Cancer Center, NorthwesternUniversity) for histological processing, and the Institute for BioNanotechnol-ogy in Medicine at Northwestern University for facilities support. This workwas supported by the National Institutes of Health Grant 5-R01-EB003806and Nanotope, Inc.

1. Buckwalter JA, Mankin HJ (1998) Articular cartilage: Degeneration and osteoarthritis,repair, regeneration, and transplantation. Instr Course Lect 47:487–504.

2. O’Driscoll SW (1998) Current concepts review—The healing and regeneration ofarticular cartilage. J Bone Joint Surg Am 80:1795–1812.

3. Hunziker EB (2002) Articular cartilage repair: Basic science and clinical progress. Areview of the current status and prospects. Osteoarthr Cartilage 10:432–463.

4. Jackson DW, Simon TM, Aberman HM (2001) Symptomatic articular cartilage degen-eration: The impact in the new millennium. Clin Orthop Relat Res 391 (Suppl):S14–25.

5. Williams RJ, Harnly HW (2007) Microfracture: Indications, technique, and results. InstrCourse Lect 56:419–428.

6. Frisbie DD, et al. (2003) Early events in cartilage repair after subchondral bone micro-fracture. Clin Orthop Relat Res 407:215–227.

7. Breinan HA, Martin SD, Hsu HP, Spector M (2000) Healing of canine articular cartilagedefects treated with microfracture, a type-II collagen matrix, or cultured autologouschondrocytes. J Orthop Res 18:781–789.

8. Kreuz PC, et al. (2006) Results after microfracture of full-thickness chondral defects indifferent compartments in the knee. Osteoarthr Cartilage 14:1119–1125.

9. Behanna HA, Donners JJ, Gordon AC, Stupp SI (2005) Coassembly of amphiphiles withopposite peptide polarities into nanofibers. J Am Chem Soc 127:1193–1200.

10. Claussen RC, Rabatic BM, Stupp SI (2003) Aqueous self-assembly of unsymmetricPeptide bolaamphiphiles into nanofibers with hydrophilic cores and surfaces. J AmChem Soc 125:12680–12681.

11. Hartgerink JD, Beniash E, Stupp SI (2001) Self-assembly and mineralization ofpeptide-amphiphile nanofibers. Science 294:1684–1688.

12. Hartgerink JD, Beniash E, Stupp SI (2002) Peptide-amphiphile nanofibers: A versatilescaffold for the preparation of self-assembling materials. Proc Natl Acad Sci USA99:5133–5138.

13. Jiang H, Guler MO, Stupp SI (2007) The internal structure of self-assembled peptideamphiphiles nanofibers. Soft Matter 3:454–462.

14. McLennan IS, Koishi K (2002) The transforming growth factor-betas: Multifacetedregulators of the development and maintenance of skeletal muscles, motoneuronsand Schwann cells. Int J Dev Biol 46:559–567.

15. Dounchis JS, et al. (1997) Chondrogenic phenotype of perichondrium-derived chon-droprogenitor cells is influenced by transforming growth factor-beta 1. J OrthopRes 15:803–807.

16. Johnstone B, Hering TM, Caplan AI, Goldberg VM, Yoo JU (1998) In vitro chondrogen-esis of bone marrow-derived mesenchymal progenitor cells. Exp Cell Res 238:265–272.

17. O’Connor WJ, Botti T, Khan SN, Lane JM (2000) The use of growth factors in cartilagerepair. Orthop Clin North Am 31:399–410.

18. Hunziker EB (2001) Growth-factor-induced healing of partial-thickness defects in adultarticular cartilage. Osteoarthr Cartilage 9:22–32.

19. StevensMM, et al. (2005) In vivo engineering of organs: The bone bioreactor. Proc NatlAcad Sci USA 102:11450–11455.

20. Paralkar VM, Vukicevic S, Reddi AH (1991) Transforming growth factor beta type 1binds to collagen IV of basement membrane matrix: Implications for development.Dev Biol 143:303–308.

21. RajangamK, et al. (2006) Heparin binding nanostructures to promote growth of bloodvessels. Nano Lett 6:2086–2090.

22. O’Driscoll SW, Keeley FW, Salter RB (1986) The chondrogenic potential of freeautogenous periosteal grafts for biological resurfacing of major full-thickness defectsin joint surfaces under the influence of continuous passive motion. An experimentalinvestigation in the rabbit. J Bone Joint Surg Am 68:1017–1035.

23. Klagsbrun M (1992) Mediators of angiogenesis: The biological significance ofbasic fibroblast growth factor (bFGF)-heparin and heparan sulfate interactions. SeminCancer Biol 3:81–87.

24. Smith GP, Petrenko VA (1997) Phage Display. Chem Rev 97:391–410.25. Hoess RH (2001) Protein design and phage display. Chem Rev 101:3205–3218.

3298 ∣ www.pnas.org/cgi/doi/10.1073/pnas.0906501107 Shah et al.

Dow

nloa

ded

by g

uest

on

Mar

ch 2

4, 2

020