lable at ScienceDirect

Biomaterials 54 (2015) 97e105

Contents lists avai

Biomaterials

journal homepage: www.elsevier .com/locate/biomater ia ls

Self healing hydrogels composed of amyloid nano fibrils for cellculture and stem cell differentiation

Reeba S. Jacob a, 1, Dhiman Ghosh a, 1, Pradeep K. Singh a, Santanu K. Basu a,Narendra Nath Jha a, Subhadeep Das a, e, Pradip K. Sukul b, Sachin Patil c,Sadhana Sathaye c, Ashutosh Kumar a, Arindam Chowdhury d, Sudip Malik b,Shamik Sen a, Samir K. Maji a, *

a Department of Biosciences and Bioengineering, Indian Institute of Technology Bombay, Mumbai, Maharashtra 400076, Indiab Polymer Science Unit, Indian Association for the Cultivation of Science, Jadavpur, Kolkata 700032, Indiac Institute of Chemical Technology, Matunga, Mumbai, Maharashtra 400031, Indiad Department of Chemistry, Indian Institute of Technology Bombay, Mumbai, Maharashtra 400076, Indiae IITB Monash Research Academy, IIT Bombay, Powai, Mumbai 400076, India

a r t i c l e i n f o

Article history:Received 3 March 2015Accepted 4 March 2015Available online

Keywords:Amyloid hydrogelNanofibrilsSelf-assemblyStem cellTissue engineering

* Corresponding author.E-mail address: samirmaji@iitb.ac.in (S.K. Maji).

1 Equally contributed authors.

http://dx.doi.org/10.1016/j.biomaterials.2015.03.0020142-9612/© 2015 Elsevier Ltd. All rights reserved.

a b s t r a c t

Amyloids are highly ordered protein/peptide aggregates associated with human diseases as well asvarious native biological functions. Given the diverse range of physiochemical properties of amyloids, wehypothesized that higher order amyloid self-assembly could be used for fabricating novel hydrogels forbiomaterial applications. For proof of concept, we designed a series of peptides based on the high ag-gregation prone C-terminus of Ab42, which is associated with Alzheimer's disease. These Fmoc protectedpeptides self assemble to b sheet rich nanofibrils, forming hydrogels that are thermoreversible, non-toxicand thixotropic. Mechanistic studies indicate that while hydrophobic, pep interactions and hydrogenbonding drive amyloid network formation to form supramolecular gel structure, the exposed hydro-phobic surface of amyloid fibrils may render thixotropicity to these gels. We have demonstrated theutility of these hydrogels in supporting cell attachment and spreading across a diverse range of cell types.Finally, by tuning the stiffness of these gels through modulation of peptide concentration and saltconcentration these hydrogels could be used as scaffolds that can drive differentiation of mesenchymalstem cells. Taken together, our results indicate that small size, ease of custom synthesis, thixotropicnature makes these amyloid-based hydrogels ideally suited for biomaterial/nanotechnology applications.

© 2015 Elsevier Ltd. All rights reserved.

1. Introduction

Self-assembly of biomolecules plays an important role in livingorganisms. For example, protein polymerization by actin andtubulin directly controls cell shape and other vital cellular functions[1,2]. The extracellular matrix, a composite mixture of various self-assembled biomaterials such as collagen, elastin, laminin, fibro-nectin and proteoglycans [3e5], can directly control cell adhesion,proliferation and tissue organization [6e8]. Guided by protein/peptide self-assembly in nature, artificial self-assembling systemshave been designed, which can mimic important biological

functions [9e11]. In this context, protein/peptide based hydro-gelators have found wide applications in tissue engineering (2D/3Dcell culture) and drug delivery [12e17]. Contrary to protein gels,where synthesis, manipulation and control of the self-assembly fordesired supramolecular formulation is challenging, small peptidegels allow ease of tunability of their functional properties as perdesired applications [17]. Such smartly designed small peptide gelswere found to have significant applications both in nanotechnologyand tissue engineering [11,18e20]. For instance, self-assembly ofpeptides with alternative charge and hydrophobic residue wasshown to form materials that can serve as a substrate for neuriteoutgrowth and synapse formation for the application in tissuerepair and tissue engineering [21,22].

Amyloids are self assembled protein/peptide aggregates thathave been implicated in the pathogenesis of many human diseasesincluding Alzheimer's and Parkinson's disease [23]. Intriguingly,

R.S. Jacob et al. / Biomaterials 54 (2015) 97e10598

recent studies have suggested that amyloid fibrils are less toxiccompared to their corresponding oligomers [24e26] and was pro-posed that amyloid formation could be defense mechanism of thecellular machinery against oligomer cytotoxicity [27,28]. Structur-ally, amyloids are composed of highly repetitive cross-b-sheetstructure, possess high stability and tensile strength; their physi-cochemical properties can be tuned by environmental conditionsand/or by amino acid substitutions [29,30]. These beneficial fea-tures of amyloids have been used in nature to perform variousnative functions for the host organism [23,31]. Examples of suchfunctional amyloids include component proteins in bacterial bio-films like curli [32], adhesins [33] and prion amyloids [34,35] inyeast. Moreover, recent reports have also demonstrated that pep-tide/protein hormones form amyloids like structures for its storagein pituitary glands [36]. Collectively, these examples of functionalamyloids, together with structural plasticity and reversible natureof amyloid under certain conditions [36] highlight the potential ofamyloids in development of nanostructured biomaterials.

Furthermore, the self-assembly of protein/peptides into amy-loid fibrils have been found to be useful as smart materials forvarious biomedical and technological applications [30,37e39]including development of nanowires/opto electronics [40e46],biosensors [47e49], drug delivery depot [50e52] and developingscaffold for tissue engineering [53e58]. In our current study, wepresent a class of short peptides, which forms hydrogels via self-assembling into nanoscale amyloid fibrils, which support cellculture and stem cell differentiation. Previous studies with pro-tein/peptide based amyloid hydrogels have shown their applica-bility in developing scaffolds that can support cell adhesion andspreading [53,54,56e59]. However, the applicability of amyloidbased hydrogels as scaffolds for stem cell differentiation intoparticular lineage (in our case neuronal) has not been exploredmuch. Here, we studied themechanism of amyloid based hydrogelformation and used their self-healing property in developingscaffolds for 2D, 3D cell culture and stem cell differentiation.

2. Materials and methods

2.1. Preparation of hydrogel

The peptide solutions were prepared by suspending the peptide powders(P1eP8) into a clean and dry glass vial at a concentration of 6 mg/ml in 20 mMsodium phosphate buffer, pH 7.4. This suspension was heated over a spirit lamp tillthe entire peptide was dissolved completely; the solution thenwas left to cool downat RT. The solution was allowed to stand for ~15 min without disturbance and theglass vial was inverted to test the gel formation (tube inversion test). Similarly, forpreparing gels with varying stiffness, different concentrations of P5 peptide (6, 12,48 and 60 mg/ml) were dissolved in 20 mM sodium phosphate buffer, pH 7.4 aspreviously described. For salts dependent study, 6 mg/ml of P5 peptides were dis-solved in presence and absence of 100, 200, 300, 400 and 500 mMNaCl and the gelswere prepared as described previously.

2.2. MTT assay

For cytotoxicity study, MTT assay was performed as described previously [60]with slight modification. Briefly, neuroblastoma cell line SH-SY5Y were culturedand maintained in DMEM media with 10% FBS and with 100-units/ml of penicillinand 100-mg/ml of streptomycin at 37 �C in a 5% CO2 incubator. For testing cyto-compatibility of amyloid gels, 20 ml of each peptide gel (6 mg/ml) was casted inwellsof 96 well plate and SH-SY5Y cells of density 4 � 104 cells were seeded in each welland were incubated for 24 h. After the required incubation with the gels, the MTTreagent at a concentration of 0.5 mg/ml per well was added to cells and incubatedagain for 4 h. Culture mediumwas then removed and the reduced MTT reagent wasdissolved by adding 100 ml solubilization solution containing 50% DMF and 20% SDS,pH 4.7 and incubated overnight. After incubation absorbance was taken at 560 nmusing a microplate reader (Thermofisher). All experiments were done in triplicates.Absorbance at 690 nmwas taken for background correction. Only gels andMTTwereused as control. Ab 40 fibrils were used as negative control of cell viability.

2.3. 2D and 3D cell culture and stem cell differentiation

For two dimensional (2D) cell culture, mouse fibroblast cell line L929 andneuroblastoma cell line SH-SY5Ywere used for studying the adhesion and spreading

of cells on different peptide gels. SH-SY5Yand L929 cells were maintained in DMEM(HIMEDIA, India) with 10% heat inactivated FBS and 100 units/ml penicillin and100 mg/ml streptomycin. The cell attachment was analyzed after 24 h of cell culture.The spreading area and circularity analysis of cells on each peptide gels werequantified using image J software (NIH, Version 1.47). Atleast 50 cells were analyzedfor each condition.

The 3D cell culture was performed with L929 and SH-SY5Y cell lines. Briefly, theP5 peptide gel at a concentration of 12 mg/ml was prepared and was vortexed toform the sol. 50 ml of sol was then mixed with 50 ml of DMEM with cells (density1 � 104) and was placed onto treated coverslips (with aminosilane and 0.5% glu-tearaldehyde). On gelation, additional growth media was provided. The final gelconcentration was maintained to 6 mg/ml. The cultures were maintained in a 37 �Cincubator with a humidified atmosphere of 5% CO2 and analyzed for the viabilityusing calcein AM staining. To do so, after 48 h of incubation of the cells within thegel, media was removed and cells were washed with PBS buffer. Then, 1 mg/mlcalcein (Invitrogen, USA) solution prepared in PBS buffer was added to the cellswithin the gel and incubated for 5 min. Excess calcein solution was then removedand equal volume of PBS was added and cells were imaged using confocal micro-scopy (IX 81, Olympus).

For stem cell attachment and its differentiation on the gels, human mesen-chymal stem cells (hMSCs) that were isolated from bone marrow (Stempeutics,India) were maintained in KnockOut DMEM (GIBCO) with 10% FBS and 2 mM Glu-tamax (GIBCO) and 0.25% penicillin and streptomycin. All cells were cultured andmaintained at 37 �C in a 5% CO2 incubator. The cell monolayers were then trypsi-nized and pelleted by centrifugation at 1500 rpm for 3 min. These cells (density of1 � 104) were then resuspended in the growth media and added directly on thesurface of peptide hydrogels layered on the treated (with aminosilane and 0.5%glutaraldehyde) coverslips. The cell attachment was analyzed after 24 h of cellculture. The spreading area and circularity analysis of cells on each peptide gels wereperformed using image J software (NIH, Version 1.47). Atleast 50 cells were analyzedfor each condition.

For studying stiffness dependent hMSC differentiation on peptide gels, gelswith different elastic modulus were prepared by varying peptide and salt con-centrations as previously described. hMSCs of cell density 1 � 104 were seeded onthese gels that was casted on treated coverslips. The morphological changes ofthese cells were monitored daily and imaged on day 1 and day 7. Morphologicalanalysis of cells in each condition was done with image J (NIH, Version 1.47)wherein at least 50 cells were analyzed for each condition. For quantifying bIIItubulin staining, tubulin stained cell images were analyzed for at least 50 cellsusing image J (NIH, Version 1.47).

2.4. Quantitative real time PCR (qPCR)

qPCR is routinely used for identification and quantification of gene expressionduring cell differentiation. hMSCswere cultured at a density of 104 per well on P5 gelin a 24 well plate as described for 2D cell culture. After seven days of culturing,hMSCs cultured on glass and P5 gel were pooled from 3 different wells for eachdifferent condition by trypsinisation. Further TRIzol (Invitrogen) were added andmixed with the cells. The whole mixture was then passed through a 24G sterileneedle 8e10 times for complete cell lysis. The cell lysate was then treated withchloroform to remove proteins, followed by RNA precipitation by isopropanol andnucleic acid carrier glycogen. The extracted RNA was quantified spectrophotomet-rically using ratio at 260 nm and 280 nm (Nanodrop spectrophotometer, ThermoScientific). The total RNA was then reverse transcribed to cDNA with ProtoScript®

First Strand cDNA Synthesis Kit (NEB, USA) using random hexamers and Oligo(dT)20primers according to the manufacturer's protocol. The cDNA obtained was used forqPCR analysis on Illumina Eco qPCR system using SYBR®Green mastermix (Ambion,USA). The SYBR green primers (SigmaeAldrich, USA) used for qPCR study were pre-designed and validated.

2.5. Statistical analysis

The statistical significance was determined by one-way ANOVA followed byNewmaneKeuls Multiple Comparison post hoc test; *P < 0.01; **P < 0.001.

3. Results and discussion

3.1. Peptides based on amyloid b-protein sequences form nontoxichydrogels

To design amyloid-based hydrogels, we chose the high aggre-gation and b-sheet prone C-terminal amyloid b-protein (Ab42)sequence (associated with Alzheimer's disease) [61] (Fig. 1AeC).The C-terminal of Ab, Ab(37e42) has previously been shown toform b-sheet zipper structure of amyloid by X-ray diffractionstudies [62]. Moreover, the two C-terminal residues (IleeAla) havebeen reported to increase the aggregation of Ab42 over Ab40

Fig. 1. Design and characterization of amyloidogenic peptide hydrogels. A) The peptide sequences of Ab40 and Ab42. The shaded portion showing the plausible b-sheet regions ofAb42/Ab40 fibrils. B) Fibril structure of Ab42 showing C-terminus in green color (PDB code used 2BEG). Crystal structure of amyloid microcrystal derived from Ab(37e42) showingb-sheet zipper formations (in blue). The structure is shown according to ZipperDB [62]. C) Designed peptides based on C-terminal of Ab42. D) Gel formation of the designed peptidesusing gel inversion test showing gel formation by Fmoc peptides. E) Rheological characterization of the peptide gel P5. Both storage modulus (G׳) and loss modulus (G ׳׳ ) were foundto increase with frequency. F) Morphology of P5 gel imaged using AFM. G) MTT assay of cells grown on different amyloid scaffolds showing cell viability of >80% on the differentpeptide hydrogels. Ab40 fibrils were used as negative control in cell viability assay.

R.S. Jacob et al. / Biomaterials 54 (2015) 97e105 99

significantly [63,64]. We propose that rapid intermolecularhydrogen bond formation by designed amyloidogenic peptide se-quences (designated as P1eP8 (Fig. 1C)) give rise to amyloid fibrilnetwork, which subsequently form hydrogel via entrapment ofwater. We found that most of the Fmoc peptides formed gels in20 mM phosphate buffer, pH 7.4 (Fig. 1D and Fig. S1). However,none of the unprotected peptides showed any gelation. The visco-elastic nature of these amyloid hydrogels was assessed using smalldeformation oscillatory rheology. Rheological characterizationrevealed that at lower frequency, the storage modulus G0 is higherthan that of the loss modulus G00 (Fig. 1E) by approximately oneorder of magnitude indicating the formation of hydrogels. Howeverat higher frequency the loss modulus increases and coincides withthe storage modulus indicating a phase transition to sol state(Fig. 1E). These hydrogels have low storage modulus below 10 Hz(at a concentration of 6 mg/ml) (<0.5 kPa) (Fig. 1E and Fig. S2)illustrating their high compliance. AFM and FEG-SEM of these gelsrevealed formation of dense nano-fibrils network (Fig. 1F andFigs. S3, S4), which are responsible for gel formation. The datasuggest that in Fmoc protected peptides, the intermolecular pepinteraction of the Fmoc group facilitates peptides to come closer

resulting in successful self-assembly and amyloid fibril formation.The cytocompatibility of these gels were tested using cells andin vivo (in rats using P5 gel). Ab40 fibrils were used as a control incell viability assay, which is known to be cytotoxic [65,66]. None ofthe gels showed any potential toxicity to SH-SY5Y cells; also, P5 geldid not cause any toxicity in the rat subjects (Fig. 1G and Fig. S5).

3.2. Amyloid nanofibril formation leads to hydrogelation

Since the designed peptides were based on the high aggregationprone C-terminal sequence of Ab42 (associated with Alzheimer'sdisease), we hypothesize that the fibril network responsible forgelation possessed amyloid-like character. To probe the amyloido-genicity of the fibril networks, CR and ThT binding studies wereperformed. Both CR and ThT are known to bind the amyloid form ofthe protein/peptide aggregates and not to their monomeric coun-terparts, and were hence used for amyloid detection in vitro andin vivo [36]. ThT generally exhibited an elevated fluorescence signalat 480 nmwhen bound to amyloid fibrils upon excitation at 450 nm[67]. All gels produced high ThT fluorescence at 480 nm (Fig. 2A andFig. S6). However, when peptides were solubilized in buffer by

R.S. Jacob et al. / Biomaterials 54 (2015) 97e105100

heating and subsequently diluted (without gel formation), signifi-cantly less ThT fluorescence was observed, suggesting that the fibrilnetwork formed for gelation possessed amyloid nature in contrastto soluble peptides (Fig. 2A, Fig. S6).

To further confirm the amyloidogenicity, CR dye binding wasperformed by CR UV absorption [68] and CR birefringence assay[69].When peptide solutions (6mg/ml) containing 360 mMCRwereheated upto 80 �C (sol state) and gradually cooled for gel formation,an increase in CR absorption and red shift of lmax occurred below50 �C for all gels suggesting that self-assembly and gelation areresults of higher order assembly of amyloid fibrils (Fig. 2B, Fig. S7).Consistent with these findings, an apple green birefringence wasalso observed for CR bound gels under cross-polarized light (Fig. 2Cand Fig. S8) further confirming the amyloid nature of the fibrilnetworks [69]. After establishing the amyloidogenic nature of thegel network, we used FTIR spectroscopy to check whether thepeptides in gel state were composed of b-sheet-rich structure [70].The FTIR spectra of the powder form of all Fmoc peptide showedFTIR peaks in the range of 1640 cm�1 to 1650 cm�1 (amide-I band)corresponding to random coil conformation (Fig. S9A). However,the FTIR spectra of all gels showed peaks in the range of 1639 and1619 cm�1 indicating b-sheet formation (Fig. 2D and S9B).

Since the Fmoc unprotected peptides were unable to form gels,we hypothesize that intermolecular interaction mediated by Fmoccould be one of the driving forces for gelation. To directly probethis, temperature dependent gelation was studied for all gels using1H NMR spectroscopy. At 80 �C, all peptides showed 1H chemicalshift of the Fmoc group in the region of (7.8e8.4) p.p.m. Withdecreasing temperature, NMR peak intensity of the 1H of Fmocgroup gradually decreased with the change of chemical shift to-wards the more shielded region (Fig. 2E and Fig. S10). For tem-peratures less than 60 �C, multiplicity of the individual peaks ofFmoc group were merged and all Fmoc peaks disappeared attemperatures below 40 �C, indicative of association of Fmoc groups

Fig. 2. Gelation of amyloidogenic peptides. A) ThT binding of P5 gel showing higher ThTtemperatures showing higher CR below 50 �C, suggesting self-assembly and gelation is a resuCR bound to the dried P5 gel under bright field (left) that showed greenish-yellow birefringeof P5 gels showing peaks at 1639 and 1619 cm�1 indicative of b-sheet structure. E) Temperatof Fmoc groups were found to merge at temperatures below 60 �C, which subsequently disagelation. F) Temperature-dependent fluorescence of Fmoc peptide showing gradual decreasstacking during gelation. (For interpretation of the references to color in this figure legend

during gelation (Fig. 2E and Fig. S10). To further probe the aromaticp-stacking interaction between Fmoc groups, temperature depen-dent fluorescence study of P5 gel was performed during gelation.The intensity of intrinsic Fmoc fluorescence decreased duringgelation (Fig. 2F), suggesting the involvement of FmoceFmocinteraction [71].

3.3. The thixotropicity of amyloid hydrogels

One of the most interesting property of these hydrogels wastheir thixotropic (i.e., self healing) nature [72e74], i.e., reversiblebreaking of the gel under mechanical stress and re-forming understatic conditions [72,74]. Self-healing property of amyloid basedhydrogels was also reported previously by Mezzenga and co-workers [75]. In our current study, all Fmoc peptide gels convertedto solution state (sol) after vortexing for 5 min and transformedback to their gel state upon keeping them undisturbed for~10e15 min at room temperature (Fig. 3A). The thixotropic natureof the gels was experimentally confirmed with stress-strainrheology experiments, wherein peptide gels were found toreadily transform into sol after applying high shear stress(g ¼ 100%; u ¼ 10 rad s�1), yielding a drop in the storage modulus(G0) to a value below the loss modulus (G00). However, when thestrain amplitude was decreased (g ¼ 0.5%) at the same frequency,the G0 value was instantly increased to its initial value and trans-formed into a semi-solid (gel) state (Fig. 3B and Fig. S11). We pro-posed that interactions mediated by sticky hydrophobic surfaces ofamyloid fibrils [60,76] may contribute to the quick recovery of thefibril network formation required for gel formation. If this hy-pothesis holds true, then molecules that bind to the exposed hy-drophobic surfaces of the fibrils would inhibit the fibril networkformation and gelation. To probe this, gel recovery time of P5 solwas examined in absence and presence of varying concentration offluorescent dye Nile Red (NR), which is known to bind to the

binding compared to solution (sol) state. B) Congo red (CR) binding of P5 at differentlt of higher order assembly of amyloid fibrils. C) Congo red (CR) birefringence of P5 gel.

nce under cross-polarized light (right) indicating amyloid state of P5 gel. D) FTIR spectraure-dependent NMR spectra showing 1H chemical shifts of Fmoc group. 1H NMR peaksppeared at temperature <40 �C, suggesting aromatic p stacking of Fmoc groups duringe in Fmoc fluorescence with decrease in temperature suggesting inter molecular pep, the reader is referred to the web version of this article.)

Fig. 3. Mechanism of thixotropic behavior and gel formation. A) Effect of nile red (NR) on thixotropicity of P5 hydrogels. Gel recovery time after vortexing (sol state) of P5 hydrogelin presence of NR was longer compared to P5 alone sample. B) Stress-strain rheology of P5 gel. Increase in shear stress (g ¼ 100%; u ¼ 10 rad s�1) led to the transition from gel to solsuggested by the decrease in storage modulus (G

0) lower than loss modulus (G00). Upon decreasing the shear stress to 0.5%, G0 value instantaneously increased indicating sol to gel

transition. C) The NR binding of P5 solution made by vortexing the P5 gel showing blue shift and an increase in the NR fluorescence intensity. D) ThT binding of P5 solution (made byvortexing P5 gel) in presence and absence of NR showing no significant effect of NR on the ThT binding to P5 solution. The data suggest that amyloid fibrils of P5 gels are still intacteven after binding to NR. E) Epifluorescence microscopy image of NR binding to individual P5 filaments. Inset showing non-uniform NR intensity along the long axis of selectedfilaments. Scale bar is 5 mm. F) Gel recovery time of P5 gels after vortexing in presence of various NR concentrations showing increase in gel recovery time in presence of higher NRconcentration. G) Proposed mechanism of gel formation and its thixotropicity. In presence of NR amyloid fibrils were formed but lacks required network for gelation. (Forinterpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

R.S. Jacob et al. / Biomaterials 54 (2015) 97e105 101

exposed hydrophobic surfaces of protein/peptide aggregates[60,77]. The NR was able to bind P5 sol (Fig. 3C) and delayed the gelformation (Fig. 3A) without interfering the amyloid fibril formation(similar extend of ThT binding as shown in Fig. 3D). Further, the NRimaging of spin casted P5 gels using epifluorescence microscopy(Fig. 3E) showed filaments of various lengths. Interestingly, the NRintensity was non-uniform along the fibrils axis, which also sug-gests the presence of non-uniform exposed hydrophobic surfacesacross the long axis of fibrils (Fig. 3E inset). With increasing NRconcentrations from 25 mM to 200 mM, the gelation time alsoincreased from 60 min up to 800 min as compared to the ~15 minrecovery time in untreated gels (Fig. 3F), suggesting that NR couldinterfere with gelation of amyloid hydrogels by blocking theexposed hydrophobic sites of the amyloid fibrils and making themless sticky (Fig. 3G). It is also possible that the binding of NR onamyloid fibrils could change its linear charge density and therebyaffecting the gel formation. To probe this, we measured the zetapotential of the diluted P5 gel in presence and absence of NR. Thezeta potential of 20 mM amyloid fibrils alone showed �30 mV,indicating the stable nature of these fibril solutions [78,79].

However, addition of increasing concentration of NR did not changethe zeta potential values significantly (Fig. S12). For instance, inpresence of 100 mM of NR, 20 mM P5 amyloid fibrils showed zetapotential of �28 mV. This indicates that the addition of NR, aneutral dye, to amyloid fibrils does not affect the linear chargedensity of amyloid fibrils. This data further support the hypothesisthat, NRmediated lag in gel recovery could be due to the blocking ofexposed hydrophobic sites of the amyloid fibrils.

We also studied the thixotropicity and amyloidogenicity of twocommercially available gels (Fmoc-YA and Fmoc-FF) as controls.Fmoc-FF hydrogel showed CR binding (Fig. S13A) without anythixotropicity and Fmoc-YA gels showed neither amyloid charac-teristics (lack of CR binding and absence of b sheet characteristicpeak in FTIR) (Fig. S13BeC) nor thixotropic behavior. The lack ofthixotropicity by Fmoc-FF hydrogel might be attributed to itshigher mechanical strength (>2 kPa) (Fig. S13D) due to extensive pstacking of Fmoc and aromatic Phenylalanine (F) amino acid[18e20]. Therefore, the delicate balance between p-stacking andamyloidogenicity of the fibrils is very likely governs the thixo-tropicity of these gels.

R.S. Jacob et al. / Biomaterials 54 (2015) 97e105102

3.4. Amyloid gels promote cell attachment

To assess the suitability of the peptide gels as biomaterials,initial experiments of cell attachment and spreading on the surfaceof the gels (i.e., 2D culture) were conducted with SH-SY5Y neuronaland L929 fibroblasts cells. Though both cell types were able toattach and spread on different gels (Fig. 4A and Fig. S14), subtledifferences in spreading area and cell shape (quantified by circu-larity) were observed (Fig. 4B). Compared to glass, L929 cells spreadsignificantly higher across all the gel surfaces (except P4) and weremore spindle-shaped (Fig 4B). In contrast, SH-SY5Y cells spreadmaximally on glass substrates. However, cells were more elongatedon P2 and P4 gels (Fig. 4B). SH-SY5Y cells grown on all these am-yloid gels expressed bIII- tubulin (neuronal marker) suggesting that

Fig. 4. 2D cell culture using amyloid hydrogels. A) Phase contrast images of attached cells ofall images are 100 mm. B) Spreading area and circularity of both L929 and SH-SY5Y cells calcgels surface. “**” indicates statistical significance (P < 0.001). C) SEM image depicting cell aincubation respectively. Scale bars are 1 mm (top) and 10 mm (bottom) respectively. D) Spreaspreading of cells on different peptide gel surfaces. “*” indicates statistical significance (P <and P5 gel surfaces. The actin filaments of the cells were stained with phalloidin (red) and nreferences to color in this figure legend, the reader is referred to the web version of this a

these hydrogels support cells of neuronal lineage (Fig. S15). TheSEM study showed that SH-SY5Y cells successfully attached on thenanofibrils of hydrogel after 1 h of incubation upon the P5 gel andafter 24 h, cells exhibited a well spread morphology (Fig. 4C). Thestudies imply that the nanofibrous structure of amyloid hydrogelssupport cell adhesion and cell spreading.

After establishing 2D, we further tested the suitability of presentclass of gels for culturing stem cells. We found that all gels are ableto support the attachment and spreading of human mesenchymalstem cells (hMSCs) (Fig. 4DeE and S16). After 24 h in culture, hMSCsgrown on P2 and P4 gels exhibited maximal spreading followed bycells on P7 and P5 (Fig. 4D). Further, low circularity of hMSCs on P5gels indicated that these gels induced anisotropic cell spreadingand cell elongation (Fig. 4D). In contrast, hMSCs on control gels

SH-SY5Y cells and L929 cells grown for 24 h on glass substrates and P5 gel. Scale bar forulated using Image J software showing different spreading of cells on different peptidedhesion and spreading of SH-SY5Y on P5 hydrogel after 1 h (top) and 24 h (bottom) ofding area and circularity of hMSC calculated using Image J software showing different0.01). E) Confocal microscopy images showing morphology of hMSCs cultured on glassucleus with DAPI (Blue). Scale bar for both images are 50 mm. (For interpretation of therticle.)

R.S. Jacob et al. / Biomaterials 54 (2015) 97e105 103

(Fmoc-FF and Fmoc-YA) exhibited morphology similar to that onglass substrates with increased spreading area and maximumcircularity (Fig. S17). Taken together, these results of cell attach-ment as well as adhesion with amyloid hydrogel indicate the uti-lization of this hydrogel as prospective biomaterials.

3.5. Mesenchymal stem cell differentiation using amyloid hydrogels

Since these hydrogels are soft gels (�0.5 KPa), it could be used asscaffold for stem cell differentiation into neuronal lineage. The celladhesion of SH-SY5Y on P5 hydrogel, where the cells were able tomaintain the neuronal morphology further supports this assump-tion. Further, several studies have demonstrated the profound in-fluence of microenvironments and their stiffness on stem cells fate,with soft brain-mimetic substrates shown to induce neuronal dif-ferentiation [80].

The stiffness of these amyloid gels was modulated for stem celldifferentiation by independently varying the peptide concentrationand the salt concentration. The stiffness of the P5 gels increasedexponentially upon increasing the peptide concentration (Fig. 5A).The storage modulus of each concentration was then extrapolatedto zero frequency and the G0 values obtained were plotted againstthe peptide concentration [81] (Fig. S18). Consistent with previousreports, the storage modulus of our Fmoc based peptide gels alsoobeyed a power law (G ~ Cn) function against peptide concentra-tion. However, the exponent “n” of the power law equation for thepresent class of peptide hydrogels was found to be less (n ¼ 1.7)(Fig. 5A) compared to previously reported protein fibrillar gels(n ¼ 2.5) [81]. Although similar power law was also reported forother Fmoc derived peptide hydrogels [82,83]. The low power lawfor concentration dependent rheology for Fmoc peptide gelcompared to protein gel could be that, in protein amyloid hydro-gels, gel formation is mediated by numerous H-bonding, charge

Fig. 5. Stiffness of gels and stem cell differentiation. A) Mechanical strength of P5 gels madeextrapolating G׳ at u ¼ 0) increased exponentially with peptide concentrations. B) Phase con1 day and 7 days of culture. Scale bar ¼ 100 mm. C) bIII tubulin (green) staining of hMSCs grstaining for nucleus are shown in blue. Scale bar is 50 mm. D) Gene expression profile of hMSTubulin (bIIIT) were upregulated and astrocyte marker GFAP was down regulated.

interaction and other non covalent interactions. Therefore thesephysical interactions are significantly increased with higher proteinconcentration resulting in enhanced storage modulus. However inFmoc di/tripeptides, due to the less number of residues present, thenumber of physical interactions (H-bonding, hydrophobic andionic) are less. Therefore, upon increasing the concentration, thestorage modulus of such gels would not increase in a similarmanner to protein gels.

The hMSCs cultures in selected P5 gel showed that the softestgels (6 mg/ml) led to maximal spreading and elongation of hMSCs(Fig. 5B and Fig. S19) with higher bIII- tubulin expression afterseven days in culture, indicating differentiation of hMSCs to aneuronal lineage (Fig. 5C and Fig. S20). The neuronal differentia-tion of stem cells on P5 gels was further confirmed by RT-PCR,where cells exhibited enhanced expression of various neuronspecific markers (Fig. 5D). In contrast to concentration dependentmodulation, the stiffness of the P5 gel only showed slight increaseby increasing the salt concentration (Fig. S21A) due to the chargeneutralization of C-terminal COO� group, which minimize theinter-peptide repulsion. These gels (fabricated via salt modula-tion), however, exhibited similar cell morphology (Fig. S21B-D),which suggests drastic changes in stiffness are necessary tochange the lineages for stem cell differentiation. Alternatively, P5gels when in contact with growth media during long termculturing could lead to dilution of the salt content from the gelcausing a decrease in stiffness of the gel to <1 KPa inducing dif-ferentiation of hMSC to neuronal lineage.

3.6. 3 Dimensional (3D) cell culture using thixotropic amyloidhydrogels

We showed that the present class of amyloid hydrogels isthixotropic and upon vortexing can convert to solution state. This

of various peptide concentrations (6e60 mg/ml). Storage modulus of gels (calculated bytrast images of human mesenchymal stem cells (hMSCs) cultured on different gels afterown on various P5 gels. Cells grown for 1st day and 7 days in culture are shown. DAPICs after 7 days of culturing on gels and glass. On P5 gels, neuronal markers ENO and bIII

Fig. 6. 3D cell culture using amyloid hydrogels. A) Schematic depicting entrapment ofcells inside thixotropic peptide gels. B) 3D cell culture using P5 gel showing cellviability of both SH-SY5Yand L929 cells indicated by calcein AM staining (green) insidethe 3D gel matrix. Scale bars are 50 mm. (For interpretation of the references to color inthis figure legend, the reader is referred to the web version of this article.)

R.S. Jacob et al. / Biomaterials 54 (2015) 97e105104

solution (sol) can pass through filters with a pore size of 0.22 mmand hence can be easily sterilized for in vivo applications. Thisthixotropic nature of amyloid gels can be successfully utilized forconstructing 3D cell culture system. For 3D cell culture, both L929and SH-SY5Y cell suspensions were mixed with the P5 solution(made by vortexing the corresponding gel) and were allowed toform gel (Fig. 6A). The viability of entrapped cells was evaluatedafter 48 h by calcein staining. Confocal imaging of 3D gels after 48 hdemonstrated that cells entrapped within the gels were viable(Fig. 6B). However in comparison to the 2D culture, cells entrappedin the hydrogel were found to spread less. Nevertheless the highlycompliant nature of amyloid hydrogel allows the cells to exertforces on the scaffold walls to assume spreadmorphology similar tocells in natural ECM matrix.

4. Conclusions

In summary, we have developed amyloid nano-fibril basedhydrogels for 2D/3D cell culture and stem cell differentiation. Wehave demonstrated that these hydrogels are composed of amyloidnano-fibrils, which are nontoxic and form a 3Dmeshwork. Detailedmechanistic study revealed that amyloid fibril formation, alongwith the pep interactions of the Fmoc group plays a crucial role inhydrogel formation. We further showed that these hydrogels arethixotropic in nature, which might arise from the sticky hydro-phobic surfaces of the amyloid fibrils. Thixotropicity is a favorabletrait for biomaterials that can be used for 3D cell culture systemsand also in the design of in situ implants. Thus, in contrast to theexisting protein and peptide hydrogels, amyloid based hydrogelsare advantageous not only due to their thixotropicity, but also dueto tunability of the gel properties by altering the amino acidsequence and/or environmental conditions. Detailed studies of oneof the representative hydrogels (P5), demonstrates their utility indriving neuronal differentiation, possibly by providing fibril-mediated contact guidance. Our study thus demonstrates theversatility of amyloid hydrogels as a novel class of biomaterials withwidespread possible applications including 2D and 3D cell culture,

scaffolds for stem cell differentiation, drug delivery systems andartificial implants.

Acknowledgments

Authors wish to acknowledge IRCC (IIT Bombay), CSIR(37(1404)/10/EMR-II) and DST (SR/FT/LS-032/2009), Governmentof India for financial support and cell biology lab (BSBE, IIT Bombay)for fluorescence imaging, CRNTS/IRCC central facilities for FEG SEM,FTIR, AFM and TIFR Mumbai for NMR facility. R.S.J and D.Gacknowledge UGC (Govt. of India) for their fellowship, SrivastavRanganathan and Saikat Ghosh for schematics and Dr. Shimul Salotand Roshni Ramachandran for giving critical comments duringmanuscript preparation. Authors also acknowledge Prof. D. Teplow,UCLA, Prof. R. Riek, ETH Zurich, Prof Rein Ulijn, University ofStrathclyde Glasgow for their valuable suggestions.

Appendix A. Supplementary data

Supplementary data related to this article can be found online athttp://dx.doi.org/10.1016/j.biomaterials.2015.03.002.

References

[1] Aida T, Ew M, Stupp SI. Functional supramolecular polymers. Science2012;335:813e7.

[2] Stupp SI, Zha RH, Palmer LC, Cui H, Bitton R. Self-assembly of biomolecular softmatter. Faraday Discuss 2013;166:9e30.

[3] Frantz C, Stewart KM, Weaver VM. The extracellular matrix at a glance. J CellSci 2010;123:4195e200.

[4] Iozzo RV. Matrix proteoglycans: from molecular design to cellular function.Annu Rev Biochem 1998;67:609e52.

[5] Schaefer L, Schaefer RM. Proteoglycans: from structural compounds tosignaling molecules. Cell Tissue Res 2010;339:237e46.

[6] Langer R, Tirrell DA. Designing materials for biology and medicine. Nature2004;428:487e92.

[7] Lutolf MP, Hubbell JA. Synthetic biomaterials as instructive extracellular mi-croenvironments for morphogenesis in tissue engineering. Nat Biotechnol2005;23:47e55.

[8] Patterson J, Martino MM, Hubbell JA. Biomimetic materials in tissue engi-neering. Mater Today 2010;13:14e22.

[9] Zhang S, Marini DM, Hwang W, Santoso S. Design of nanostructured biologicalmaterials through self-assembly of peptides and proteins. Curr Opin ChemBiol 2002;6:865e71.

[10] Kopecek J, Yang J. Peptide-directed self-assembly of hydrogels. Acta Biomater2009;5:805e16.

[11] Fichman G, Gazit E. Self-assembly of short peptides to form hydrogels: designof building blocks, physical properties and technological applications. ActaBiomater 2014;10:1671e82.

[12] Kretsinger JK, Haines LA, Ozbas B, Pochan DJ, Schneider JP. Cytocompatibilityof self-assembled beta-hairpin peptide hydrogel surfaces. Biomaterials2005;26:5177e86.

[13] Wang J, Zhang J, Zhang X, Zhou H. A protein-based hydrogel for in vitroexpansion of mesenchymal stem cells. PLoS One 2013;8:e75727.

[14] Petka WA, Harden JL, McGrath KP, Wirtz D, Tirrell DA. Reversible hydrogelsfrom self-assembling artificial proteins. Science 1998;281:389e92.

[15] Zhang S, Gelain F, Zhao X. Designer self-assembling peptide nanofiber scaf-folds for 3D tissue cell cultures. Semin Cancer Biol 2005;15:413e20.

[16] Ulijn RV, Smith AM. Designing peptide based nanomaterials. Chem Soc Rev2008;37:664e75.

[17] Stephanopoulos N, Ortony JH, Stupp SI. Self-assembly for the synthesis offunctional biomaterials. Acta Mater 2013;61:912e30.

[18] Mahler A, Reches M, Rechter M, Cohen S, Gazit E. Rigid, self-assembledhydrogel composed of a modified aromatic dipeptide. Adv Mater 2006;18:1365e70.

[19] Orbach R, Adler-Abramovich L, Zigerson S, Mironi-Harpaz I, Seliktar D, Gazit E.Self-assembled Fmoc-peptides as a platform for the formation of nano-structures and hydrogels. Biomacromolecules 2009;10:2646e51.

[20] Smith AM, Williams RJ, Tang C, Coppo P, Collins RF, Turner ML, et al. Fmoc-diphenylalanine self assembles to a hydrogel via a novel architecture based onpep interlocked b-sheets. Adv Mater 2008;20:37e41.

[21] Holmes TC, de Lacalle S, Su X, Liu G, Rich A, Zhang S. Extensive neuriteoutgrowth and active synapse formation on self-assembling peptide scaffolds.Proc Natl Acad Sci U S A 2000;97:6728e33.

[22] Ellis-Behnke RG, Liang YX, You SW, Tay DK, Zhang S, So KF, et al. Nano neuroknitting: peptide nanofiber scaffold for brain repair and axon regenerationwith functional return of vision. Proc Natl Acad Sci U S A 2006;103:5054e9.

R.S. Jacob et al. / Biomaterials 54 (2015) 97e105 105

[23] Chiti F, Dobson CM. Protein misfolding, functional amyloid, and human dis-ease. Annu Rev Biochem 2006;75:333e66.

[24] Hardy J, Selkoe DJ. Medicine - the amyloid hypothesis of Alzheimer's disease:progress and problems on the road to therapeutics. Science 2002;297:353e6.

[25] Lashuel HA, Hartley D, Petre BM, Walz T, Lansbury PT. Neurodegenerativedisease - amyloid pores from pathogenic mutations. Nature 2002;418:291.

[26] Winner B, Jappelli R, Maji SK, Desplats PA, Boyer L, Aigner S, et al. In vivodemonstration that alpha-synuclein oligomers are toxic. Proc Natl Acad Sci US A 2011;108:4194e9.

[27] Kopito RR. Aggresomes, inclusion bodies and protein aggregation. Trends CellBiol 2000;10(12):524e30.

[28] Carrotta R, Manno M, Bulone D, Martorana V, San Biagio PL. Protofibril for-mation of amyloid beta-protein at low pH via a non-cooperative elongationmechanism. J Biol Chem 2005;280(34):30001e8.

[29] Sunde M, Serpell LC, Bartlam M, Fraser PE, Pepys MB, Blake CC. Common corestructure of amyloid fibrils by synchrotron X-ray diffraction. J Mol Biol1997;273:729e39.

[30] Mankar S, Anoop A, Sen S, Maji SK. Nanomaterials: amyloids reflect theirbrighter side. Nano Rev 2011;2.

[31] Fowler DM, Koulov AV, Balch WE, Kelly JW. Functional amyloidefrom bacteriato humans. Trends Biochem Sci 2007;32:217e24.

[32] Chapman MR, Robinson LS, Pinkner JS, Roth R, Heuser J, Hammar M, et al. Roleof Escherichia coli curli operons in directing amyloid fiber formation. Science2002;295:851e5.

[33] Ramsook CB, Tan C, Garcia MC, Fung R, Soybelman G, Henry R, et al. Yeast celladhesion molecules have functional amyloid-forming sequences. EukaryotCell 2010;9:393e404.

[34] Wickner RB. [URE3] as an altered URE2 protein: evidence for a prion analog inSaccharomyces cerevisiae. Science 1994;264:566e9.

[35] Sondheimer N, Lindquist S. Rnq1: an epigenetic modifier of protein functionin yeast. Mol Cell 2000;5:163e72.

[36] Maji SK, Perrin MH, Sawaya MR, Jessberger S, Vadodaria K, Rissman RA, et al.Functional amyloids as natural storage of peptide hormones in pituitarysecretory granules. Science 2009;325:328e32.

[37] Cherny I, Gazit E. Amyloids: not only pathological agents but also orderednanomaterials. Angew Chem Int Ed Engl 2008;47:4062e9.

[38] Knowles TP, Oppenheim TW, Buell AK, Chirgadze DY, Welland ME. Nano-structured films from hierarchical self-assembly of amyloidogenic proteins.Nat Nanotechnol 2010;5:204e7.

[39] Gras SL. Surface- and solution-based assembly of amyloid fibrils forbiomedical and nanotechnology applications. Adv Chem Eng 2009;35:161e209.

[40] Reches M, Gazit E. Casting metal nanowires within discrete self-assembledpeptide nanotubes. Science 2003;300:625e7.

[41] Scheibel T, Parthasarathy R, Sawicki G, Lin XM, Jaeger H, Lindquist SL. Con-ducting nanowires built by controlled self-assembly of amyloid fibers andselective metal deposition. Proc Natl Acad Sci U S A 2003;100:4527e32.

[42] Reches M, Gazit E. Controlled patterning of aligned self-assembled peptidenanotubes. Nat Nanotechnol 2006;1:195e200.

[43] Herland B, Bjork P, Nilsson KPR, Olsson JDM, Asberg P, Konradsson P, et al.Electroactive luminescent self-assembled bio-organic nanowires: integrationof semiconducting oligoelectrolytes within amyloidogenic proteins. AdvMater 2005;17:1466e71.

[44] Li C, Bolisetty S, Chaitanya K, Adamcik J, Mezzenga R. Tunable carbon nano-tube/protein core-shell nanoparticles with NIR- and enzymatic-responsivecytotoxicity. Adv Mater 2013;25:1010e5.

[45] Hamedi M, Herland A, Karlsson RH, Inganas O. Electrochemical devices madefrom conducting nanowire networks self-assembled from amyloid fibrils andalkoxysulfonate PEDOT. Nano Lett 2008;8:1736e40.

[46] Bolisetty S, Adamcik J, Heier J, Mezzenga R. Amyloid directed synthesis oftitanium dioxide nanowires and their applications in hybrid photovoltaicdevices. Adv Funct Mater 2012;22:3424e8.

[47] Yemini M, Reches M, Gazit E, Rishpon J. Peptide nanotube-modified electrodesfor enzyme-biosensor applications. Anal Chem 2005;77:5155e9.

[48] Li C, Adamcik J, Mezzenga R. Biodegradable nanocomposites of amyloid fibrilsand graphene with shape-memory and enzyme-sensing properties. NatNanotechnol 2012;7:421e7.

[49] Sasso L, Suei S, Domigan L, Healy J, Nock V, Williams MA, et al. Versatile multi-functionalization of protein nanofibrils for biosensor applications. Nanoscale2014;6:1629e34.

[50] Maji SK, Schubert D, Rivier C, Lee S, Rivier JE, Riek R. Amyloid as a depot forthe formulation of long-acting drugs. PloS Biol 2008;6:e17.

[51] Nagai Y, Unsworth LD, Koutsopoulos S, Zhang S. Slow release of molecules inself-assembling peptide nanofiber scaffold. J Control Release 2006;115:18e25.

[52] Bhak G, Lee S, Park JW, Cho S, Paik SR. Amyloid hydrogel derived from curlyprotein fibrils of a-synuclein. Biomaterials 2010;31:5986e95.

[53] Gras SL, Tickler AK, Squires AM, Devlin GL, Horton MA, Dobson CM, et al.Functionalised amyloid fibrils for roles in cell adhesion. Biomaterials 2008;29:1553e62.

[54] Bongiovanni MN, Scanlon DB, Gras SL. Functional fibrils derived from thepeptide TTR1-cycloRGDfK that target cell adhesion and spreading. Bio-materials 2011;32:6099e110.

[55] Ryan DM, Nilsson BL. Self-assembled amino acids and dipeptides as non-covalent hydrogels for tissue engineering. Polym Chem 2012;3:18e33.

[56] Reynolds NP, Charnley M, Mezzenga R, Hartley PG. Engineered lysozymeamyloid fibril networks support cellular growth and spreading. Bio-macromolecules 2014;15:599e608.

[57] Reynolds NP, Styan KE, Easton CD, Li Y, Waddington L, Lara C, et al. Nano-topographic surfaces with defined surface chemistries from amyloid fibrilnetworks can control cell attachment. Biomacromolecules 2013;14:2305e16.

[58] Li C, Born AK, Schweizer T, Zenobi-Wong M, Cerruti M, Mezzenga R. Amyloid-hydroxyapatite bone biomimetic composites. Adv Mater 2014;26:3207e12.

[59] Yan H, Nykanen A, Ruokolainen J, Farrar D, Gough JE, Saiani A, et al. Thermo-reversible protein fibrillar hydrogels as cell scaffolds. Faraday Discuss2008;139:71e84.

[60] Singh PK, Kotia V, Ghosh D, Mohite GM, Kumar A, Maji SK. Curcumin mod-ulates alpha-synuclein aggregation and toxicity. ACS Chem Neurosci 2013;4:393e407.

[61] Selkoe DJ. Cell biology of the amyloid b-protein precursor and the mechanismof Alzheimer's disease. Annu Rev Cell Biol 1994;10:373e403.

[62] Goldschmidt L, Teng PK, Riek R, Eisenberg D. Identifying the amylome, pro-teins capable of forming amyloid-like fibrils. Proc Natl Acad Sci U S A2010;107:3487e92.

[63] Jarrett JT, Berger EP, Lansbury Jr PT. The carboxy terminus of the b amyloidprotein is critical for the seeding of amyloid formation: Implications for thepathogenesis of Alzheimer's disease. Biochemistry 1993;32:4693e7.

[64] Teplow DB. Structural and kinetic features of amyloid b-protein fibrillo-genesis. Amyloid 1998;5:121e42.

[65] Walsh DM, Hartley DM, Condron MM, Selkoe DJ, Teplow DB. In vitro studies ofamyloid beta-protein fibril assembly and toxicity provide clues to the aeti-ology of Flemish variant (Ala692–>Gly) Alzheimer's disease. Biochem J2001;355:869e77.

[66] Singh PK, Maji SK. Amyloid-like fibril formation by tachykinin neuropeptidesand its relevance to amyloid beta-protein aggregation and toxicity. Cell Bio-chem Biophys 2012;64:29e44.

[67] LeVine 3rd H. Quantification of beta-sheet amyloid fibril structures with thi-oflavin T. Methods Enzymol 1999;309:274e84.

[68] Klunk WE, Jacob RF, Mason RP. Quantifying amyloid by Congo red spectralshift assay. Methods Enzymol 1999;309:285e305.

[69] Westermark GT, Johnson KH, Westermark P. Staining methods for identifi-cation of amyloid in tissue. Methods Enzymol 1999;309:3e25.

[70] Barth A. Infrared spectroscopy of proteins. Biochim Biophys Acta 2007;1767:1073e101.

[71] Tang C, Ulijn RV, Saiani A. Effect of glycine substitution on Fmoc-diphenylalanineself-assembly and gelation properties. Langmuir 2011;27:14438e49.

[72] Barnes HA. Thixotropyda review. J Newt Fluid Mech 1997;70:1e33.[73] Pek YS, Wan AC, Shekaran A, Zhuo L, Ying JY. A thixotropic nanocomposite gel

for three-dimensional cell culture. Nat Nanotechnol 2008;3:671e5.[74] Haines-Butterick L, Rajagopal K, Branco M, Salick D, Rughani R, Pilarz M, et al.

Controlling hydrogelation kinetics by peptide design for three-dimensionalencapsulation and injectable delivery of cells. Proc Natl Acad Sci U S A2007;104:7791e6.

[75] Bolisetty S, Harnau L, Jung JM, Mezzenga R. Gelation, phase behavior, anddynamics of beta-lactoglobulin amyloid fibrils at varying concentrations andionic strengths. Biomacromolecules 2012;13:3241e52.

[76] Bolognesi B, Kumita JR, Barros TP, Esbjorner EK, Luheshi LM, Crowther DC,et al. ANS binding reveals common features of cytotoxic amyloid species. ACSChem Biol;5:735e40.

[77] Krishnan R, Goodman JL, Mukhopadhyay S, Pacheco CD, Lemke EA, Deniz AA,et al. Conserved features of intermediates in amyloid assembly determinetheir benign or toxic states. Proc Natl Acad Sci U S A 2012;109:11172e7.

[78] Fukuhara S, Nishigaki T, Miyata K, Tsuchiya N, Waku T, Tanaka N. Mechanismof the chaperone-like and antichaperone activities of amyloid fibrils of pep-tides from alphaA-crystallin. Biochemistry 2012;51:5394e401.

[79] Fast JL, Cordes AA, Carpenter JF, Randolph TW. Physical instability of a ther-apeutic Fc fusion protein: domain contributions to conformational andcolloidal stability. Biochemistry 2009;48:11724e36.

[80] Engler AJ, Sen S, Sweeney HL, Discher DE. Matrix elasticity directs stem celllineage specification. Cell 2006;126:677e89.

[81] Owczarz M, Bolisetty S, Mezzenga R, Arosio P. Sol-gel transition of chargedfibrils composed of a model amphiphilic peptide. J Colloid Interface Sci2015;437:244e51.

[82] Adams DJ, Mullen LM, Berta M, Chen L, Frith WJ. Relationship between mo-lecular structure, gelation behaviour and gel properties of Fmoc-dipeptides.Soft Matter 2010;6:1971e80.

[83] Raeburn J, Mendoza-Cuenca CBN, Cattoz, Little MA, Terry AE, ZamithCardoso AG, et al. The effect of solvent choice on the gelation and finalhydrogel properties of Fmoc-diphenylalanine. Soft Matter 2015;11:927e35.