lable at ScienceDirect

Biomaterials 29 (2008) 4268–4274

Contents lists avai

Biomaterials

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

The behavior of aged regenerated Bombyx mori silk fibroin solutionsstudied by 1H NMR and rheology

Zainuddin a,b,c,*, Tri T. Le d, Yoosup Park e,f, Traian V. Chirila a,b,e,g, Peter J. Halley e,f,Andrew K. Whittaker c,e

a Queensland Eye Institute, Brisbane, Qld, Australiab School of Medicine, Faculty of Health Sciences, University of Queensland, Brisbane, Qld, Australiac Centre for Magnetic Resonance, University of Queensland, Brisbane, Qld, Australiad Department of Chemistry, University of Queensland, Brisbane, Qld, Australiae Australian Institute for Bioengineering and Nanotechnology, University of Queensland, Brisbane, Qld, Australiaf Department of Chemical Engineering, University of Queensland, Brisbane, Qld, Australiag School of Physical and Chemical Sciences, Queensland University of Technology, Brisbane, Qld, Australia

a r t i c l e i n f o

Article history:Received 12 June 2008Accepted 28 July 2008Available online 19 August 2008

Keywords:Silk fibroinAgeing1H NMRRheologyChain conformationPseudo-plastic flow

* Corresponding author Queensland Eye InstituteBrisbane, Qld 4101, Australia. Tel.: þ61 7 3010 3381;

E-mail address: z.zainuddin@qei.org.au (Zainuddin

0142-9612/$ – see front matter � 2008 Elsevier Ltd.doi:10.1016/j.biomaterials.2008.07.041

a b s t r a c t

As part of a project to utilize the regenerated silk fibroin (RSF) membranes as a supporting matrix forthe attachment and growth of corneal stem/progenitor cells in the development of tissue engineeredconstructs for the surgical restoration of the ocular surface, the behavior of the aged RSF solutions hasbeen investigated. The solutions were produced from domesticated silkworm (Bombyx mori) cocoonsaccording to a protocol involving successive dissolution steps, filtration and dialysis. The solutions werekept at 4 �C in a refrigerator for a certain period of time until near the gelation time. The changes inmolecular conformation were studied by solution-state 1H NMR, while the flow of the solutions wascharacterized by rheological method. Upon ageing turbidity developed in solutions and the viscositycontinuously decreased prior to a drastic increased near the gelation time. The 1H resonances of agedsolutions showed a consistent downfield shift as compared to the 1H resonances of the fresh solution.Shear thinning with anomalous short recovery within a certain range of low shear rates occurred inboth fresh and aged solutions. While the solutions behave as pseudo-plastic materials, the chainconformation in aged solutions adopted all secondary configurations with b-strand being predominant.

� 2008 Elsevier Ltd. All rights reserved.

1. Introduction

It has been known that the ocular surface disorders (OSDs)caused by various chronic conditions such as thermal or chemicalburns, Stevens–Johnson syndrome, cicatricial pemphigoid andchronic contact lens wear can damage the progenitor cells, leadingto deficiency of limbal epithelial stem cells. A severe implication ofprolonged limbal stem cell deficiency is blindness due to progres-sive ingrowth of fibrous tissue and opacification of the cornea [1].Clinical studies have shown that the surgical treatment to restoreOSDs either by autografts or allografts has many limitations.Autografting is restricted by the amount of tissue that can beremoved from the other eye and may cause complications to thehealthy eye, whilst allografting is limited by the availability of

, 41 Annerley Road, Southfax: þ61 7 3010 3390.).

All rights reserved.

donor tissue, possible rejection and biosafety concerns. In a rela-tively new approach, limbal stem cells are isolated through biopsy,expanded in vitro on a suitable matrix and then transplanted toinjured/diseased corneal surface [2–5]. Providing that the sup-porting material is capable of promoting cell attachment, growthand spreading, the latter technique offers a viable alternative forrestoring the ocular surface.

Currently, denuded human amniotic membrane (AM) is the mostwidely used substrate for ocular surface repair [2,6–10]. However, asthe AM is a human-derived tissue, it is a potential vector for trans-ferring infectious disease [11]. Other drawbacks of using this bio-logical material have also been reported [12–14]. Therefore, thesearch for alternative materials is increasingly important. To date,several alternative matrices have been evaluated including collagenand its derivatives, fibrin gel, Matrigel� and some synthetic poly-mers [4,5], but only few of them have reached animal experimen-tation or clinical trials and the results were unsatisfactory.

Hence, in our previous reports [15,16], we have proposed andevaluated the feasibility of the regenerated silk fibroin (RSF)

Zainuddin et al. / Biomaterials 29 (2008) 4268–4274 4269

membranes as a supporting matrix for cultivation of humanlimbal epithelial (HLE) cells. We demonstrated that the RSFmembranes were able to support the growth of HLE cells at a levelcomparable to that on tissue culture plastic. A comprehensive studyon cell behavior and its long-term differentiation is currentlyundergoing in our laboratories and the results will be published indue course.

It is known that the cell-adhesive properties of the RSFmembranes are also influenced by the protocol for preparation ofthe membranes [17–22], a phenomenon which probably relates tothe conformational change in solution that occurs during pro-cessing and storage [23–25] and by solvent treatment of themembranes [17–19,26–28], leading to the formation of, and aninterplay between crystallizable forms known as silks I, II and III,where b-forms and a-helix structures predominate, and amor-phous forms containing mainly random coil structures [29–36]. Itwas Coleman and Howitt who, in their seminal paper on fibroin[37], showed probably for the first time that different conforma-tions of the silk fibroin can be reciprocally converted. They noticedthat if water-soluble fibroin (‘‘denatured’’), which we know now assilk II, was dissolved in aqueous solution of copper ethylenediamineand then dialyzed against water, the resulting water-soluble fibroin(renatured) does not differ essentially from native fibroin (as foundin the silk gland), which we call now silk I. However, at that time,they could not refer to silk I or silk II as these terms were to becoined a few years later by Kratky and coworkers [38,39], who werealso the first to give a correct interpretation to this observation byshowing that silk II can be converted into silk I after dissolution ofthe former in an appropriate solvent followed by dialysis againstwater [40,41].

In the last decades, there have been considerable efforts tounderstand the transformation of the viscous silk I in the middlesection of domesticated silkworm’s glands (about 25% of concen-tration) to crystalline fibrous silk II spun through the spinneret[42–47]. As a result, many factors have been identified to affect thetransition of silk I to silk II, including pH, removal of calcium ionsand water molecules from the ducts and the presence of externalforces during the spinning process [28,42–47]. Based on theexperiments developed to mimic the natural spinning process ofsilkworm silk it was revealed that the external force played animportant role in the transition of silk I to silk II [42–47]. Within thiscontext, the rheometer has also been utilized to generate a similarstructural transformation by applying shear force to the silk fibroinsolutions [42–55]. Although limited studies on the rheology of theRSF aqueous solutions have been reported [47–50,53,55], itappeared that shear thinning dominated the process and theformation of aggregates was occasionally observed at considerablyhigh shear rate [47,53,55].

Another factor which also contributes to the transformation ofsilk I to silk II is the structural nature of the silk fibroin. It is wellknown that the silk fibroin is constituted by highly repeatedhydrophobic and crystallisable molecules with the primarysequences of amino acid residues of Gly–Ala–Gly–Ala–Gly–Ser,[Gly–Ala]n–Gly–Tyr and [Gly–Val]n–Gly–Ala (n ¼ 1–8) separated by11 amorphous region of mainly Gly–Ala–Gly–Ser and Gly–Ala–Gly–Ala–Gly–Ser sequences [51]. Therefore, upon storage, the RSFsolution tends to undergo self-transformation to adopt a morestable conformation by forming intra- and/or intermolecularhydrogen bonds [23,56–58].

In this study, to shed more light on the conformational changeduring storage and to determine the useful life time of the RSFsolutions prior to gelation, the behavior of RSF solutions wasmonitored by using rheological and proton NMR methods. Thisknowledge is essential, particularly when the RSF membranesare intended to be used as substrates for tissue engineeringapplications.

2. Materials and methods

2.1. Materials

The Bombyx mori cocoons were purchased from Tajima Shoji Co. Ltd. (Yokohama,Japan). Sodium carbonate (Na2CO3) and lithium bromide (LiBr) were obtained fromSigma-Aldrich. Methanol and deionized water (18.2 MUcm) were used as solvents.

2.2. Preparation and ageing of RSF aqueous solutions

The Bombyx mori cocoons were cut in smaller pieces, vacuum dried, weighed,and placed in 1 L boiling solution of Na2CO3 (0.02 M) for 1 h to remove sericins.Subsequently the silk fibers were rinsed three times in 1 L hot water (w70 �C) for20 min each and then let to dry in a fume hood overnight. After sericin-free silkfibers were dried under vacuum for few hours, they were dissolved in an aqueouslithium bromide solution (9.3 M) at 60 �C for 4 h to obtain a silk concentration ofabout 10%. The solution was pre-filtered through a syringe filter with a pore size of0.8 mm (Minisart�-GF, Sartorius) and followed by a 0.20-mm pore size filter (Minis-art�High-Flow, Sartorius). About 10 mL of the filtrate was injected into a 3–12 mLdialysis cassette with a molecular weight cut-off of 3500 (Slide-A-Lyzer�, Pierce) tobe dialysed against water. Following six changes of water within 3 days of dialysis,the resulting solution was collected and filtered through a 0.20-mm pore size filter(Minisart�High-Flow). The resulting aqueous solution with a concentration in silkfibroin of about 3.8% was kept at 4 �C for 1 to 4 months.

2.3. 1H NMR measurements

The samples were diluted to four times of the initial volume of the RSF aqueoussolutions with deuterated water (D2O). 1H NMR spectra were obtained on a 500 MHzAvance Bruker spectrometer operating at 500.13 MHz with a 5 mm probe. Water(HDO) resonance was used as an internal standard for determination the 1Hchemical shifts. The water signal was suppressed by a thousand fold or more usingwatergate pulse sequence with gradient double echo. Phasing and baselinecorrections were completed manually. The software used for these procedures wasTOPSPIN version 1.3.

2.4. Rheological measurements

The viscosity of the RSF aqueous solutions was measured using an AR-G2Rheometers (TA Instrument Ltd., USA). The shear viscosity was acquired by linearlyincreasing the shear rate without oscillation from 0.05 to 500 s�1. Meanwhile, themeasurement of storage (G0) and loss (G00) modulus as a function of frequency (u)was performed within the linear viscoelastic region using 40-mm diameter coneplate with 57 mm gap. All rheological measurements were carried out at roomtemperature (25 � 0.5 �C). A solvent trap was used throughout the experiment toavoid the possible loss of water.

3. Results and discussion

3.1. 1H NMR analysis

Though the proton (1H) resonances of many proteins andpeptides in the solution-state have been well documented, littledata are available on the solutions of silk fibrous proteins, includingthe regenerated Bombyx mori silk fibroin [59]. The fact that the silkfibroin does not readily dissolve in most organic solvents and it isstructurally unstable in aqueous environment due to conforma-tional change. Our observation on the effect of ageing on thebehavior of the aqueous silk fibroin solution, however, hasprompted us to carry out the 1H NMR measurement in aqueoussystem. Under this condition we are aware that the interference ofthe water 1H resonance is imminent. However, because all of thesamples contain the same amount of water and this study was tocompare the 1H resonances of the aged and fresh silk fibroinsolutions, the same magnitude of interference would apply to allsamples, thus the changes in the molecular structure due to ageingeffect would directly correspond to the changes in chemical shiftand line shape. Fig. 1 shows the 1H NMR spectra of the fresh and4 month old silk fibroin solutions with and without water peaksuppression. As expected, the major peaks were mainly contributedby the primary structure of the silk fibroin (i.e. alanine, glycine,serine, tyrosine and valine). Close inspection of the spectra (seeFig. 2a–f) reveals that while the global chemical shift of the agedsolutions consistently moved downfield (relative to the fresh silk

4

4

0

0

HDO

Gly αSer β

Tyr βTyr 4HAla, β

Val, γ

9 8 7 6 5 4 3 2 1 0

A

B

(ppm)

Fig. 1. 1H NMR spectra of fresh and aged RSF aqueous solutions without (A) and with(B) water peak suppression measured at 298 K. 0 denotes fresh solution and 4 denotes4 months old solution.

Zainuddin et al. / Biomaterials 29 (2008) 4268–42744270

solution) the shape of the lines remains unchanged. This behaviorsuggests that both fresh and aged solutions contained all of thethree secondary structures of the silk fibroin proteins (random coil,a-helix and b-strand) with b-strands content increased withincreasing ageing time. The presence of b-strands in the fresh RSFsolutions is not unexpected, particularly in the solutions that arenot clarified after dialysis against water [24]. Additionally, thetransformation of silk fibroin from random coil/a-helix to b-strandsand/or b-sheets (adjacent b-strands link together via hydrogenbonds) during storage has been reported elsewhere [23–25].Accordingly, the appearance (even if it is weak and only observed inthe 4 month old sample) of new 1H resonances at 5.60 ppm andpossibly at around 5.30 ppm (Fig. 1) which are characteristic of theb forms [60–62] was anticipated. It seems that even though b formsis structurally more packed than the random coil the molecularchain remains flexible, as reflected by small downfield shift (about0.06 ppm). Hence, our hypothesis is that ageing the RSF solution upto near the gelation time would result in a transformation ofrandom coil to mainly b-strand, while formation of b-sheetprimarily occurred at the onset of gelation.

Therefore, the 1H resonances of the RSF aqueous solution priorto gelation have been assigned by closely examining the expanded1H NMR spectra (Fig. 2a–f) and matching their values with thevalues of all amino acid residues available in the literature [60–66],as summarized in Table 1. Taking into account that both fresh andaged silk solution strongly demonstrated the presence of mixedconformations, so the 1H resonances were determined as anaverage of all resonances emerged from random coil, a-helix andb-strand. The peaks appearing at 0.6–0.9 ppm were attributed to Ile(Hg,d) and Leu (Hd), with a small contribution of Hg of Val (Fig. 2a).The multi-peaks at 1.1–1.5 ppm were originated mainly from Hg ofIle overlapped with Hb of Ala, Hg of Lys and Hg of Thr residues(Fig. 2a,b). As shown in Fig. 2b, at least eight amino acid residueswere detected under the peaks which stretch from 1.5 to 2.5 ppm,namely Leu (Hb,g), Arg (Hb,g), Lys (Hb,d), Met (Hb,g,3), Ile (Hb), Pro(Hb,g), Val (Hb) and Glu (Hb,g). The main peak appearing at2.65–3.05 ppm was contributed by Hb of Tyr, Phe, Asp, Cys and Hb,3

of Lys residues (Fig. 2c). A significant Ha resonances emerge at3.5–4.5 ppm was mainly due to Ha of Gly as major component ofthe silk fibroin, overlapping with Hb of Ser and Hd of Pro residues(Fig. 2d). Other Ha resonances which should appear at 4.4–5.7 ppmwere not observable (n/a) due to water peak suppression. Whilst

the aromatic side-chain 1H resonances of Tyr and Phe can be seen at6.50–7.50 ppm (Fig. 2e), the amide protons of all amino acid resi-dues expand at 7.85–8.45 ppm (Fig. 2f). The appearance of few verynarrow peaks in all spectra (1.76/1.81, 2.12, 3.25 and 8.29/8.34 ppm)suggests that both fresh and aged silk solutions contained smallportion of low molecular weight peptides which were free fromhydrogen bonding [67].

3.2. Rheological study

Izuka [48] reported that the viscosity of the RSF aqueous solu-tions measured at 14 �C was extremely low, even at a fibroinconcentration of about 10% (x0.09 Pa s). The authors suggestedthat the silk fibroin solution might be considered as liquid crys-talline. To further elaborate on this behavior the shear viscosity ofthe freshly prepared and aged silk solutions has been measured andreported in this study. Fig. 3 shows the dependence of viscosity onthe shear rate for fresh and aged silk fibroin solutions. It can be seenthat all solutions demonstrated a similar trend of viscositybehavior. Initially, the viscosity is almost independent of shear rate,approaching Newtonian flow, particularly for silk solutions of1–3 month old and then follows plastic flow behavior withcontinuous shear thinning. More interestingly, there was a sensitiveregion in which a drastic drop of viscosity followed by a shortrecovery occurred before a continuous shear thinning. Thisbehavior suggests that under shear forces the silk fibroin under-went structural transformations. Having the viscosity continuouslydecreasing with increasing shear rate the model of non-Newtonianpseudoplastic flow can be applied. Typically, the profile of thismodel shows three zones, i.e. a low shear rate plateau followed bya near power-law decrease that ends finally at high shear rateplateau [68]. There numerous mathematical models have beenproposed to explain this behavior [69], however, the one proposedby Cross [70] was found to be more practical in the analyses of ourexperimental results. The model is expressed in the followingequation.

h ¼ hN þ ðh0 � hNÞ=�1þ l�ym� (1)

Parameters h0 and hN are the limiting viscosity values at shearrate �y / 0 and �y / N, respectively. l is a constant with units oftime and m > 0 is a dimensionless constant that measure theseverity of shear thinning, typically ranges from 2/3 to 1. As shownin Fig. 3, irrespective of anomalous (a drastic drop and shortrecovery) region, the congruency of fitting curves and experimentaldata is sufficient to estimate the zero-shear viscosity for each of thesolutions, and the results are presented in Fig. 4. It can be seen thatthe viscosity of the solution gradually decreased with extension ofageing time up to 3 months, and then it increased significantly inthe following month. This behavior is consistent with our earlyhypothesis, i.e. following b-strand transformation which led tolower viscosity, b-sheet transformation occurred, causing anincrease in the viscosity. Noting that the solution gelled just a fewdays after 4 months timepoint, the viscosity of the solution musthave sharply increased and reached an infinite value at the onset ofgelation. Clearly, the macroscopic effect of the conformationalchanges can be visually observed from the development ofturbidity in the solutions (see inserted images in Fig. 4). The moreb-strand and/or b-sheet was formed the more light would bediffracted, thus increasing the turbidity.

Based on the above observation, it is unambiguous that theviscosity behavior shown in Fig. 3 was dictated by both shear rateand the propensity of the silk fibroin to undergo self-trans-formation. Under shear, the molecules aligned into more uniformorientation and promoted the formation of either folded orunfolded chain, depending on the shear rate. At low shear rate

4

3

2

1

0

4

3

2

1

0

4

3

2

1

0

4

3

2

1

0

4

3

2

1

0

4

3

2

1

0

AlaThrIIe

IIe

IIeIIeVal Val

LeuGlu Glu

Glu

Glu

Pro

Pro

Pro

Met

Met

Met Met

Arg

ArgArg

Lys

Lys Lys Lys

LeuLeu

Leu

1.5 1.4 1.3 1.2 1.1 1.0 0.9 0.8 0.7 0.6 0.5 2.5 2.4 2.3 2.2 2.1 2.0 1.9 1.8 1.7 1.6 1.5 1.4 1.3

Pro HisHis

HisTyr TyrPhe

Phe

Arg

Cys

Cys

Asp

Asp

Lys SerThrPro

Pro

LeuArgLysMet

GlyAlaValGluThrIIe Ser

SerGly

3.6 3.5 3.4 3.3 3.2 3.1 3.0 2.9 2.8 2.7 2.6 2.5 2.4 4.5 4.4 4.3 4.2 4.1 4.0 3.9 3.8 3.7 3.6 3.5

PhePheArg

Lys

Tyr

Tyr

TyrTyr

His

Arg

NH ofall residues Phe

LysHis

7.6 7.5 7.4 7.3 7.2 7.1 7.0 6.9 6.8 6.7 6.6 8.66.5 8.5 8.4 8.3 8.2 8.1 8.0 7.9 7.8 7.7 7.6(ppm)

a

c

e

b

d

f

Fig. 2. Expanded 1H NMR spectra for fresh and aged RSF aqueous solutions with water peak suppression. 0 denotes fresh solution and 1–4 denote aged solution to 1, 2, 3 and4 months, respectively.

Zainuddin et al. / Biomaterials 29 (2008) 4268–4274 4271

Table 11H NMR chemical shifts of the RSF aqueous solutions measured at 298 K

Residue (%) Fresh solution Aged solution

Averaged chemical shift values for all conformations Averaged chemical shift values for all conformations

Ha Hb NH Others Ha Hb NH Others

Gly (46.2) 4.15, 3.65 – 8.35 – 4.20, 3.70 – 8.39 –Ala (29.7) 4.18 1.24 8.10 – 4.24 1.28 8.16 –Ser (10.8) 4.31 3.95, 3.81 8.29 – 4.37 4.00, 3.86 8.34 –Tyr (4.9) n/a 3.08, 2.76 8.26 d-CH (7.15, 7.04) n/a 3.12, 2.81 8.30 d-CH (7.21, 7.10)

3-CH (6.91, 6.64) 3-CH (6.97, 6.69)Val (2.1) 4.10 2.05 8.20 g-CH3 (n/a, 0.72) 4.16 2.11 8.25 g-CH3 (n/a, 0.78)Asp (1.5) n/a 2.89, 2.56 8.36 – n/a 2.94, 2.62 8.40 –Glu (0.9) 4.22 2.10, 1.98 8.37 g-CH2 (2.36, 2.28) 4.26 2.11, 1.99 8.35 g-CH2 (2.39, 2.32)Thr (0.8) n/a 4.20 8.16 g-CH3 (1.19) n/a 4.24 8.20 g-CH3 (1.24)Phe (0.6) n/a 3.12, 2.84 7.99 3-CH (7.38, 7.26) n/a 3.16, 2.90 8.02 3-CH (7.44, 7.30)Ile (0.5) 4.12 1.77 8.05 g-CH2 (1.23, 1.04) 4.16 1.81 8.10 g-CH2 (1.26, 1.10)

g-CH3 (0.76) g-CH3 (0.82)d-CH3 (0.71) d-CH3 (0.77)

Pro (0.4) 4.30 2.18, 1.84 – g-CH2 (1.89, 1.74) 4.36 2.22, 1.90 – g-CH2 (1.94, 1.78)d-CH2 (3.67, 3.48) d-CH2 (3.69, 3.54)

Leu (0.6) 4.32 1.74, 1.57 8.23 g-CH (1.53) 4.38 1.80, 1.61 8.27 g-CH (1.57)d-CH3 (0.78, 0.67) d-CH3 (0.84, 0.72)

Arg (0.4) 4.33 1.86, 1.71 8.27 g-CH2 (1.60, 1.52) 4.37 1.92, 1.77 8.32 g-CH2 (1.65, 1.56)d-CH2 (3.12, 3.08) d-CH2 (3.16, 3.13)NH (7.36, 6.62) NH (7.41, 6.67)

Lys (0.3) 4.28 1.90, 1.73 8.29 g-CH2 (1.40, 1.36) 4.34 1.97, 1.78 8.35 g-CH2 (1.45, 1.41)d-CH2 (1.61, 1.57) d-CH2 (1.66, 1.63)3-CH3 (2.96, 2.93) 3-CH3 (3.02, 2.99)3-NH3 (7.44) 3-NH3 (7.49)

His (0.2) n/a 3.28, 2.86 8.28 d-CH (7.70, 7.14) n/a 3.34, 2.90 8.33 d-CH (7.76, 7.20)Met (0.1) 4.29 1.84, 1.58 8.22 g-CH3 (2.20, 1.87) 4.35 1.92, 1.62 8.32 g-CH3 (2.26, 1.98)

3-CH3 (1.50) 3-CH3 (1.55)Cys (0.1) n/a 3.17, 2.80 8.17 – n/a 3.21, 2.84 8.23 –

Zainuddin et al. / Biomaterials 29 (2008) 4268–42744272

(<0.15 s�1), the viscosity of the fresh solution was slightly reducedwith increasing the shear rate, suggesting that the alignment ofmolecules which are consisted of mainly random coil improved themolecular flow and promoted the formation of b-strand. On theother hand, in aged solutions, the viscosity was independent ofshear rate up to 3 months and then slightly reduced at 4 monthstime. This behavior might be attributed to the b-strand conforma-tion formed by self-transformation during storage. As b-strand isa folded conformation, it is more packed and more easily tumbling,thus alignment of the molecules by shear would not significantlyalter the molecular flow and so as the viscosity. In the 4 months oldsolution, there was an indication that the b-sheet has also beenformed, as reflected by an increase in the zero-shear viscosity (seeFig. 4) and further shifting of the chemical shift to downfield

1

1 10 100 1000

0.1

0.1

0.01

0.010.001

Fresh, m=0.971 month, m=0.802 months, m=0.773 months, m=0.704 months, m=0.93Curve fitting

Visco

sity (P

a.s)

Shear rate (s-1

)

Fig. 3. Viscosity behavior of the fresh and aged RSF aqueous solutions (3.8% w/v) undershear force.

(Fig. 2a–f). However, it seemed that this infant b-sheet structurewas easily disrupted, even by low shear rate, resulting in a gradualdecrease of the viscosity.

At shear rates of 0.2–1.0 s�1, the viscosity of all solutionsappeared to be highly sensitive with a drastic drop of viscosityfollowed by a short recovery. We suggest that the shear force at thissensitive region was able to induce a massive breakage of transientcrosslinked points (H-bonds), leading to the unfolding of the foldedchains. Clearly, after a short recovery, the viscosity continuouslydecreased and eventually reached plateau at end shear rate of500 s�1. This suggests that once all of the unfolded chains werealigned and straighten by the flow field the adjacent moleculesstarted to aggregate. Episodically, the macroscopic effect of thisaggregation was observed as white material surrounded by a clear

1.0

0.8

0.6

0.4

0.2

0.0

0 1 2 3 4 5

Solution

Infinity

Gel

Zero

-sh

ear visco

sity (P

a.s)

Time (month)

Fig. 4. Effect of ageing time on zero-shear viscosity of the RSF aqueous solutions(3.8% w/v). Turbidity (inserted images) developed as the ageing time increased.

1000

100

10

1

0.1 1 10 100 10000.01

0.1

0.01(rad s-1)

tan

(G

′′/G′)

Fig. 6. Tan delta (d) of fresh and aged RSF aqueous solutions (3.8% w/v).

Zainuddin et al. / Biomaterials 29 (2008) 4268–4274 4273

liquid at the end of shear experiments, a phenomenon which hasbeen previously noticed [52–54,71].

More evidences for the hybrid behavior of the RSF solutionswere obtained from the storage (G0) and loss (G00) modulusmeasurements. As shown in Fig. 5, within the oscillatory frequency(u) range tested (0.1–100 rad s�1), initially, both G0 and G00 werealmost independent of u. However, after reaching certain u values,the solutions seemed to follow the typical plastic flow behaviorwith G0 and G00 increased with increasing u. The value of G0 for thefresh solution was higher than the G00, in agreement with the trendreported by Ochi et al. [51] for the natural silk solution of similarconcentration. For the aged solutions, the above trend fluctuatedagainst the ageing time. At 1 and 4 months, it followed the freshsolution – indicating a network-like structure, but at 2 and3 months the trend reversed – suggesting a liquid crystal-likestructure. Whilst the G0 at 4 month was higher than the G0 of freshsolution, the G0 at 1–3 months was lower. This variation is inter-preted as the result of conformational change during storage. Asearly mentioned that upon ageing the silk fibroin underwentstructural transformation from random coil/a-helix to b-strand,then b-sheet and eventually formed three dimensional networksthrough self-assembly. It seemed that in early transition (at1 month) the formed b-strand was still not yet aligned and prob-ably entangled to some extent; consequently, the elastic modulusG0 dominated the loss modulus G00 and the G0 value became slightlylower than the G0 of the fresh solution. In contrast, at 2 and3 months the b-strand has relatively well aligned and separated;accordingly, the solutions might be considered as a liquid crystal. Inthis condition, the modulus values, particularly the G0, wouldappreciably drop below the modulus values of the fresh solutionand the G00 dominated the G0. While at 4 months, the adjacentb-strands started to link together through formation of hydrogenbonds between carbonyl and amine groups of the amino acidresidues. As expected, the latter association would shift bothmodulus values to a higher level with G0 being predominant.Another phenomenon which also reflects the complexity of the RSFsolutions is an appreciably high increase of G0 and G00 at u above10 rad s�1. This increase implies that at higher frequency the silkfibroin tends to form aggregates, regardless of the ageing time.Moreover, since there was no crossover point between G0 and G00

observed in all samples the solutions behave entirely like a liquid(no gelation). By plotting tan d (¼G00/G0), it is immediately seen thatthe values of tan d for aged solutions were all above the freshsolution (Fig. 6). As seen, the tan d at 1 and 4 months was justslightly higher, while the tan d at 2 and 3 months it was

100

10

1

0.1

0.01

0.01 0.1 1 10 100 1000

0.001

0.0001

G′ a

nd G

′′ (Pa

)

G′ G′′Fresh1 month2 months3 months4 months

(rad s-1)

Fig. 5. Storage (G0) and loss (G00) modulus of fresh and aged RSF aqueous solutions(3.8% w/v)

significantly higher than the fresh solution. This again confirms thatat 1 and 4 months the solutions behave like a network-structure,while at 2 and 3 months they behave like liquid crystals. Also, thetendency of the tan d to approach unity at u above 10 rad s�1

suggests that the aggregation of the b-strands is likely to occur athigh frequency.

4. Conclusions

The tendency of 1H NMR chemical shifts and the shape ofspectral line suggested that all three molecular conformations, i.e.random coil, a-helix and b-strand were present in both fresh andaged RSF aqueous solutions. As anticipated, the transformationfrom random coil/a-helix to b-strand increased upon ageing andthis transformation eventually formed b-sheet structure at theonset of gelation. The flow of RSF aqueous solutions showed char-acteristic of non-ideal pseudoplastic behavior with the viscositygradually decreased from approximately 0.30 Pa s to as low as0.012 Pa s, which is nearly the same as the viscosity of water(z0.01 Pa s) at 3 months time. Our observation revealed that theuseful life-time of RSF aqueous solution (3.8% w/v) in which theresultant membranes were still easy to handle with forceps ina similar fashion to pieces of cellophane was 3 months. Beyond thisperiod, the membranes were very brittle.

Acknowledgements

This work was sponsored by an unrestricted grant from PreventBlindness Foundation, Brisbane, Australia, through Viertels Vision.We would like to thank Professor Kaplan and Dr Matsumoto, bothat Tufts University, Medford, MA, USA, for expert advice. We alsoacknowledge the support from the group of Professor JustinCooper-White, allowing us to use the rheometer at the AustralianInstitute for Bioengineering and Nanotechnology, University ofQueensland, Brisbane, Australia.

References

[1] Tseng SC. Regulation and clinical implications of corneal epithelial stem cells.Mol Biol Rep 1996;23:47–58.

[2] Nishida K. Tissue engineering of the cornea. Cornea 2003;22:S28–34.[3] Pellegrini G, Traverso CE, Franzi AT, Zingirian M, Cancedda R, De Luca M. Long-

term restoration of damaged corneal surfaces with autologous cultivatedcorneal epithelium. Lancet 1997;349:990–3.

[4] Selvam S, Thomas PB, Yiu SC. Tissue engineering: current and futureapproaches to ocular surface reconstruction. Ocul Surf 2006;4:120–36.

[5] Yang J, Yamato M, Nishida K, Ohki T, Kanzaki M, Sekine H, et al. Cell delivery inregenerative medicine: the cell sheet engineering approach. J ControlledRelease 2006;116:193–203.

Zainuddin et al. / Biomaterials 29 (2008) 4268–42744274

[6] Schwab IR. Cultured corneal epithelia for ocular surface disease. Trans AmOphthalmol Soc 1999;97:891–986.

[7] Koizumi N, Inatomi T, Quantock AJ, Fullwood NJ, Dota A, Kinoshita S. Amnioticmembrane as substrate for cultivating limbal corneal epithelial cells forautologous transplantation in rabbits. Cornea 2000;19:65–71.

[8] Tseng SCG. Amniotic membrane transplantation for ocular surface recon-struction. Biosci Rep 2001;21:481–9.

[9] Bouchard CS, Thomas J. Amniotic membrane transplantation in the manage-ment of severe ocular surface disease: indications and outcomes. Ocul Surf2004;2:201–11.

[10] Dua HS, Gomes JAP, King AJ, Maharajan VS. The amniotic membrane inophthalmology. Surv Ophthalmol 2004;49:51–77.

[11] Schwab IR, Johnson NT, Harkin DG. Inherent risks associated with manufactureof bioengineered ocular surface tissue. Arch Ophthalmol 2006;124:1734–40.

[12] Chuck RS, Graff JM, Bryant MR, Sweet PM. Biomechanical characterization ofhuman amniotic membrane preparations for ocular surface reconstruction.Ophthalmic Res 2004;36:341–8.

[13] Tseng SCG. Evolution of amniotic membrane transplantation. Clin ExperOphthalmol 2007;35:109–10.

[14] Maharajan VS, Shanmuganathan V, Currie A, Hopkinson A, Powell-Richards A,Dua HS. Amniotic membrane transplantation for ocular surface reconstruc-tion: indications and outcomes. Clin Exper Ophthalmol 2007;35:140–7.

[15] Chirila TV, Barnard Z, Zainuddin, Harkin DG. Silk as substratum for cellattachment and proliferation. Mater Sci Forum 2007;561-565:1549–52.

[16] Chirila TV, Barnard Z, Zainuddin, Harkin DG, Schwab IR, Hirst LW. Bombyx morisilk fibroin membranes as potential substrata for epithelial constructs used inthe management of ocular surface disorders. Tissue Eng 2008;14:1203–11.

[17] Motta A, Migliaresi C, Faccioni F, Torricelli P, Fini M, Giardino R. Fibroinhydrogels for biomedical applications: preparation, characterization and invitro cell culture studies. J Biomater Sci Polym Edn 2004;15:851–64.

[18] Min BM, Jeong L, Nam YS, Kim JM, Kim JY, Park WH. Formation of silk fibroinmatrices with different texture and its cellular response to normal humankeratinocytes. Int J Biol Macromol 2004;34:223–30.

[19] Jin HJ, Park J, Karageorgiou V, Kim UJ, Valluzzi R, Cebe P, et al. Water-stable silkfilms with reduced b-sheet content. Adv Funct Mater 2005;15:1241–7.

[20] Kim HJ, Kim UJ, Vunjak-Novakovic G, Min BH, Kaplan DL. Influence of mac-roporous protein scaffolds on bone tissue engineering from bone marrowstem cells. Biomaterials 2005;26:4442–52.

[21] Servoli E, Maniglio D, Motta A, Predazzer R, Migliaresi C. Surface properties ofsilk fibroin films and their interaction with fibroblasts. Macromol Biosci2005;5:1175–83.

[22] Hakimi O, Grahn MF, Knight DP, Vadgama P. Interaction of myofibroblasts withsilk scaffolds (Abstract). Eur Cells Mater 2005;10(Suppl 2):46.

[23] Kim UJ, Park J, Li C, Jin HJ, Valuzzi R, Kaplan DL. Structure and properties of silkhydrogels. Biomacromolecules 2004;5:786–92.

[24] Matsumoto A, Chen J, Collette AL, Kim UJ, Altman GH, Cebe P, et al. Mecha-nisms of silk fibroin sol-gel transitions. J Phys Chem 2006;B:21630–8.

[25] Xie F, Shao H, Hu X. Effect of storage time and concentration on structure ofregenerated silk fibroin solution. Int J Modern Phys 2006;B:3878–83.

[26] Magoshi J, Maghosi Y, Nakamura S. Physical properties and structure of silk.VII. Crystallization of amorphous silk fibroin induced by immersion inmethanol. J Polym Sci Polym Phys Edn 1981;19:185–6.

[27] Tsukada M, Gotoh Y, Nagura M, Minoura N, Kasai N, Freddi G. Structuralchanges of silk fibroin membranes induced by immersion in methanolaqueous solutions. J Polym Sci Part B Polym Phys 1994;32:961–8.

[28] Chen X, Shao Z, Knight DP, Vollrath F. Conformation transition kinetics ofBombyx mori silk protein. Protein Struct Funct Bioinfo 2007;68:223–31.

[29] Asakura T, Kuzuhara A, Tabeta R, Saito H. Conformation characterization ofBombyx mori silk fibroin in the solid state by high-frequency 13C crosspolarization-magic angle spinning NMR, X-ray diffraction, and infrared spec-troscopy. Macromolecules 1985;18:1841–5.

[30] He SJ, Valluzzi R, Gido SP. Silk I structure in Bombyx mori silk foams. Int J BiolMacromol 1999;24:187–95.

[31] Valluzzi R, Gido SP, Muller W, Kaplan DL. Orientation of silk III at the air–waterinterface. Int J Biol Macromol 1999;24:237–42.

[32] Wilson D, Valluzzi R, Kaplan DL. Conformational transitions in model silkpeptides. Biophys J 2000;78:2690–701.

[33] Asakura T, Yao J, Yamane T, Umemura K, Ulrich AS. Heterogeneous structure ofsilk fibers from Bombyx mori resolved by 13C solid-state NMR spectroscopy.J Am Chem Soc 2002;124:8794–5.

[34] Asakura T, Ohgo K, Ishida T, Taddei P, Monti P, Kishore R. Possible implicationsof serine and tyrosine residues and intermolecular interactions on theappearance of silk I structure of Bombyx mori silk fibroin-derived syntheticpeptides: high-resolution 13C cross-polarization/magic-angle spinning NMRstudy. Biomacromolecules 2005;6:468–74.

[35] Kawahara Y, Furukawa K, Yamamoto T. Self-expansion behaviour of silk fibroinfilm. Macromol Mater Eng 2006;291:458–62.

[36] Sashina ES, Bochek AM, Novoselov NP, Kirichenko DA. Structure and solubilityof natural silk fibroin. Russian J Appl Chem 2006;79:869–76.

[37] Coleman D, Howitt FO. Studies on silk proteins. I. The properties and consti-tution of fibroin. The conversion of fibroin into a water-soluble form and itsbearing on the phenomenon of denaturation. Proc Roy Soc A 1947;190:145–69.

[38] Kratky O, Schauestein E, Sekora A. An instable lattice in silk fibroin. Nature1950;165:319–20.

[39] Kratky O, Schauestein E. X-Ray and UV spectroscopic investigations of fibrousand globular modifications silk fibroin. Disc Faraday Soc 1951;11:171–8 (disc207-208).

[40] Kratky O, Schauestein E, Sekora A. X-Ray and ultra-violet absorption spectrumof ‘‘renatured’’ silk fibroin. Nature 1950;166:1031–2.

[41] Kratky O. An X-ray investigation of silk fibroin. Trans Faraday Soc1956;52:558–70.

[42] Magoshi J, Magoshi Y, Nakamura S. Liquid crystal and fibre formation of silk.J Appl Polym Sci Appl Polym Symp 1985;41:187–204.

[43] Trabbic KA, Yager P. Comparative structural characterization of naturally- andsynthetically-spun fibers of Bombyx mori fibroin. Macromolecules1998;31:462–71.

[44] Li G, Zhou P, Shao Z, Xie X, Chen X, Wang H, et al. The natural silk spinningprocess. A nucleation-dependent aggregation mechanism? Eur J Biochem2001;268:6600–6.

[45] Jin HJ, Kaplan DL. Mechanism of silk processing in insects and spiders. Nature2003;424:1057–61.

[46] Dicko C, Kenney JM, Vollrath F. b-silks: enhancing and controlling aggregation.Adv Protein Chem 2006;73:17–53.

[47] Yamaura K, Okumura Y, Matsuzawa S. Mechanical denaturation of highpolymers in solution. XXXVI. Flow-induced crystallization of Bombyx mori Lsilk fibroin from solution under steady-state flow. J Macromol Sci Phys1982;21:49–69.

[48] Iizuka E. Silk thread: mechanism of spinning and its mechanical properties.J Appl Polym Sci Appl Polym Symp 1985;41:173–85.

[49] Iizuka E. Mechanism of fiber formation by silkworm, Bombyx mori L. Bio-rheology 1966;3:141–52.

[50] Chen X, Knight DP, Zhao Z, Vollrath F. Regenerated Bombyx mori silk solutionsstudied with rheometry and FTIR. Polymer 2001;42:9969–74.

[51] Ochi A, Hossain KS, Magoshi J, Nemoto N. Rheology and dynamic light scat-tering of silk fibroin solution extracted from middle division of Bombyx morisilkworm. Biomacromolecules 2002;3:1187–96.

[52] Terry AE, Knight DP, Porter D, Vollrath F. pH induced changes in rheology ofsilk fibroin solution from middle division of Bombyx mori silkworm. Bio-macromolecules 2004;5:768–72.

[53] Rossle M, Panine P, Urban VS, Riekel C. Structural evolution of regenerated silkfibroin under shear: combined wide- and small-angle x-ray scatteringexperiments using synchroctron radiation. Biopolymers 2004;74:316–27.

[54] Holland C, Terry AE, Porter D, Vollrath F. Comparing the rheology of nativespider and silkworm spinning dope. Nat Mater 2006;5:870–4.

[55] Holland C, Terry AE, Porter D, Vollrath F. Natural and unnatural silks. Polymer2007;48:3388–92.

[56] Asakura T, Kashiba H, Yoshimizu H. NMR of silk fibroin. 8. 13C NMR analysis ofthe conformation and the conformational transition of philosomia cynthiaricini silk fibroin protein on the basis of Bixon–Scheraga–Lifson theory.Macromolecules 1988;21:644–8.

[57] Ayub ZH, Arai M, Hirabayashi K. Mechanism of the gelation of fibroin solution.Biosci Biotech Biochem 1993;57:1910–2.

[58] Hanawa T, Watanabe A, Tsuchiya T, Ikoma R, Hidaka M, Sugihara M. New oraldosage form for elderly patients: preparation and characterization of silkfibroin gel. Chem Pharm Bull 1995;43:284–8.

[59] Ohgo K, Bagusat F, Asakura T, Scheler U. Investigation of structural transitionof regenerated fibroin aqueous solution by Rheo-NMR spectroscopy. J AmChem Soc 2008;130:4182–6.

[60] Wishart DS, Sykes BD, Richards FM. Simple techniques for the quantificationof protein secondary structure by 1H NMR spectroscopy. FEBS Lett1991;293:72–80.

[61] Wishart DS, Sykes BD, Richards FM. Relationship between nuclear magneticresonance chemical shift and protein secondary structure. J Mol Biol1991;222:311–33.

[62] Wishart DS, Sykes BD. Chemical shifts as a tool for structural determination.Methods Enzym 1994;239:363–92.

[63] Griffiths DV, Feeney J, Roberts GCK, Burgen ASV. Preparation of selectivelydeuterated aromatic amino acids for use in 1H NMR studies of proteins. Bio-chem Biophys Acta 1976;446:479–85.

[64] Mendz GL, Moore WJ, Carnegie PR. N.M.R. studies of myelin basic protein. IV.Proton spectra in aqueous solutions of proteins from mammalian and avianspecies. Aust J Chem 1982;35:1979–2006.

[65] Mendz GL, Moore WJ, Smith SE, Linthicum DS. Proton-n.m.r. study of inter-action of myelin basic protein with monoclonal antibody. Biochem J1985;228:61–8.

[66] Sims CJ, Fujito DT, Burholt DR, Dadok J, Giles HR, Wilkinson A. Quantificationof human amniotic fluid constituents by high resolution proton nuclearmagnetic resonance (NMR) spectroscopy. Prenat Diagnos 1993;13:473–80.

[67] Nagura M, Ishikawa H. Molecular motion in poly(amino acid). III. Proton broadline NMR of amorphous silk fibroin. Polym J 1979;11:159–61.

[68] Berli CLA, Deiber JA, Anon MC. Connection between rheological parametersand colloidal interactions of a soy protein suspension. Food Hydrocolloids1999;13:507–15.

[69] Braun DB, Rosen MR. Rheology modifiers handbook: practical use and appli-cations. New York: William Andrew Publisher; 2000. p. 27-48.

[70] Cross MM. Rheology of non-Newtonian fluids: a new flow equation forpseudoplastic systems. J Colloid Sci 1965;20:417–37.

[71] Iizuka E. Silk: an overview. J Appl Polym Sci Appl Polym Symp 1985;41:163–71.