lable at ScienceDirect

Biomaterials 32 (2011) 2445e2458

Contents lists avai

Biomaterials

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

Leading Opinion

Silk fibroin in ocular tissue reconstruction

Damien G. Harkin a,b,c,*, Karina A. George a,b,c, Peter W. Madden c,d, Ivan R. Schwab c,e,Dietmar W. Hutmacher b, Traian V. Chirila c,d,f,g

aDiscipline of Medical Sciences, Faculty of Science and Technology, Queensland University of Technology, 2 George Street, Queensland 4000, Brisbane, Australiab Institute of Health and Biomedical Innovation, Queensland University of Technology, Kelvin Grove, AustraliacQueensland Eye Institute, South Brisbane, Queensland, Australiad School of Medicine, University of Queensland, Herston, Queensland, AustraliaeUC Davis Health Service Eye Center, University of California at Davis, Sacramento, CA, USAfAustralian Institute for Bioengineering and Nanotechnology, University of Queensland, St Lucia, Queensland, AustraliagDiscipline of Chemistry, Faculty of Science and Technology, Queensland University of Technology, Brisbane, Australia

a r t i c l e i n f o

Article history:Received 13 December 2010Accepted 27 December 2010Available online 19 January 2011

Keywords:SilkFibroinCorneaRetinaTransplantation

* Corresponding author. Discipline of Medical ScTechnology, Queensland University of Technology,Queensland 4000, Australia. Tel.: þ61 7 3138 2552; fa

E-mail address: d.harkin@qut.edu.au (D.G. Harkin)

0142-9612/$ e see front matter � 2011 Elsevier Ltd.doi:10.1016/j.biomaterials.2010.12.041

a b s t r a c t

The silk structural protein fibroin displays potential for use in tissue engineering. We present here ouropinion of its value as a biomaterial for reconstructing tissues of clinical significancewithin the human eye.We review the strengths and weaknesses of using fibroin in those parts of the eye that we believe aremostamenable to cellular reconstruction, namely the corneoscleral limbus, corneal stroma, corneal endotheliumand outer blood-retinal barrier (Ruysch’s complex). In these areas we find that by employing the range ofmanufacturing products afforded by fibroin, relevant structural assemblies can bemade for cells expandedex vivo. Significant questions now need to be answered concerning the effect of this biomaterial on thephenotype of key cell types and the biocompatibility of fibroin within the eye. We conclude that fibroin’sstrength, structural versatility and potential for modification, combined with the relative simplicity ofassociated manufacturing processes, make fibroin a worthy candidate for further exploration.

� 2011 Elsevier Ltd. All rights reserved.

1. Introduction

Fibroin is a group of fibrous proteins produced by somemembers of the classes Insecta and Arachnida, generally to makewebs or cocoons. The secondary structure of fibroin proteins isresponsible for the mechanical properties of silk fibres and has leadto these proteins being considered as biomaterials for tissue engi-neering and other biomedical applications [1]. The commercialavailability of silk cocoons from the domesticated silkworm Bombyxmori has focusedmost studies on this source. Whilst the majority ofapplications have been considered with respect to reconstructingmusculoskeletal and vascular tissues [2e8], a small number ofrecent studies, including those by our own group, have begunexploring the potential use of fibroin for engineering more func-tionally complex tissues such as those found within the eye [9e12].The results from these preliminary studies are encouraging, butthere is still much to be learned about optimising fibroin-basedmaterials for biomedical applications. Moreover, strategies for

iences, Faculty of Science &2 George Street, Brisbane,x: þ61 7 3138 6030..

All rights reserved.

using fibroin within the eye have yet to be fully explored withina clinical context. The purpose of this article is therefore to clarifywhich tissues within the eye and which diseases stand to benefitmost from use of bioengineered tissue. From these analyses wepropose clinically appropriate strategies for treatment with fibroin-based materials. While doing so, we highlight current deficienciesin the general understanding of cellefibroin interactions which inour opinion will need to be addressed to properly exploit theseprocesses for biomedical applications.

2. What types of bioengineered ocular tissue are needed?

The eye like many organs is prone to a variety of pathologicconditions throughout life which can lead to loss of normal tissuestructure and function. These losses result in reduced vision,sometimes blindness. In some cases vision can be restored throughremoval of the diseased tissue and implantation of amedical device.For example, the development of intraocular lenses (IOLs) usingvarious acrylic polymers has transformed the treatment of cataractsto the pointwhere their use can nowbe considered routine [13]. Theartificial cornea or keratoprosthesis synthesized from poly(2-hydroxyethyl methacrylates) (PHEMA) has also proved to be usefulas a last resort in cases when conventional surgical options have

D.G. Harkin et al. / Biomaterials 32 (2011) 2445e24582446

failed [14]. The purpose of this article, however, is to consider thoseeye conditions for which transplants of bioengineered tissue arebeing developed as a first-line therapy. These conditions relate tofour specific regions within the eye: the corneoscleral limbus, thecorneal stroma, the corneal endothelium, and the outer blood-retinal barrier also known as Ruysch’s complex. The location andhistological structure for each region is shown in Fig. 1.

2.1. Overview of the cornea

The cornea of the eye is a relatively simple structure measuringapproximately half a millimetre in thickness and consisting of threedistinct tissues: the corneal epithelium consisting of 3e5 layers ofstratified squamous epithelial cells residing anterior to an acellularzone known as Bowman’s layer (BL), an avascular corneal stromasparsely populated by a mesenchymal cell type known as the ker-atocyte, and an innermost single layer of corneal endothelial cellswhich resides posterior to an acellular zone known as Descemet’smembrane (DM) (Fig. 1B). While the smoothness of the epitheliumcombined with the optical properties of the stroma constitute theprimary refractive element in the visual pathway, the endotheliallayer is also critical for vision as this single layer of cells regulatesthe degree of corneal hydration and in doing so maintains theprecise spacing of extracellular matrix (ECM) components neces-sary for a clear cornea.

The corneoscleral limbus is a narrow transitional zone of tissuewhich separates the clear cornea from the surrounding “white”

Fig. 1. Regions within the human eye where there is greatest need for bioengineered tissue.progenitor cells for maintaining the corneal epithelium. The absence of Bowman’s layer (BL;and arrows) may well contribute to maintenance of this stem cell niche. (B) Corneal stroma (within the stroma provides a clear refractive element essential for vision. The correct spendothelium which regulates hydration of the CS. Descemet’s membrane (DM) can also bemembrane (BrM) and choriocapillaris (Ch). Together these structures nourish and support

sclera (Fig. 1A). Across this transitional zone, the surface of the eyechanges from corneal epithelium covering the cornea, to conjunc-tival epitheliumwhich covers the scleral surface.While the stratifiedsquamous epithelial cells of the cornea maintain a smooth ocularsurface, the conjunctival epithelium contains mucus-secretinggoblet cellswhichhelp toprotect and lubricate the surfaceof theeye.Both tissues are maintained throughout adult life via the actions ofepithelial progenitor cells. Progenitor cells for the corneal epithe-lium in humans reside predominantly within the basal layer of thelimbal epithelium[15],which in someareas canbe seen tobulge intothe adjacent stroma resulting in the formation of epithelial crypts[16]. The limbal region differs from the cornea by the absence of BL,a more densely populated stromal layer, and the presence ofa network of terminating blood vessels. Together these histologicalfeatures contribute to formation of an epithelial stem cell niche [17].

2.2. Tissue for repairing the corneoscleral limbus

Injuries and diseases affecting the limbus have marked effects onthe stability of the ocular surface (Fig. 2B and Fig. 3D). A typicalexample occurs in the case of chemical burns whereby after limbaldamage, subsequenthealingof theocular surface canoftenonlyoccurvia ectopic expansionof theperipheral conjunctival epithelium. Sincethe conjunctival epithelium is relatively opaque and less able towithstand frictional forces associatedwith normal eyelidmovement,vision is significantly impaired and the unstable surface leads tochronic inflammation and scarring within the underlying stroma.

(A) Corneoscleral limbus: this transitional zone between the cornea and sclera containsapproximate end point marked with asterisk) and presence of stromal blood vessels (BVCS) and corneal endothelium (CE): precise orientation and spacing of ECM componentsacing of ECM components required for a clear cornea is maintained by the cornealseen. (C) Ruysch’s complex: consists of the retinal pigment epithelium (RPE), Bruch’sattachment of the retina to the underlying choroid.

Fig. 2. Clinical presentation of common and/or severe eye disorders for which bioengineered tissue transplants are being developed. (A) Example of normal cornea. (B)Encroachment of conjunctival tissue (Conj. with associated blood vessels, BV) across the corneal surface following loss of limbal epithelial progenitor cells. (C) Side view of eyedisplaying ectasia of the cornea associated with keratoconus. (D) Comparison of corneal thickness in H&E stained sections of normal cornea (top) and cornea removed from patientwith keratoconus (bottom). (E) Eye displaying typical symptoms of endothelial dysfunction with cornea becoming cloudy. (F) Fundus image of left eye displaying a healthy retina.The approximate location of the macular is encircled (M). Normal healthy retinal blood vessels (BV) can be seen emerging from the optic disc (OD). (G) Fundus image of left eyedisplaying advanced or “wet” form of age-related macular degeneration (AMD). Punctate deposits of drusen (d and arrows) can be seen across the field. Discolouration of the macula(M) is largely due to bleeding associated with aberrant choroidal blood vessels.

D.G. Harkin et al. / Biomaterials 32 (2011) 2445e2458 2447

Themost obvious choice of therapy is to transplant limbal tissuefrom a deceased donor but unfortunately this procedure is associ-ated with a high rejection rate [18], presumably owing to vascu-larity of the limbus and associated presence of immune cells.Current strategies for treatment of limbal dysfunction are thereforebased upon transplantation of epithelial progenitor cells followingex vivo expansion in the laboratory (Fig. 3 AeE). If only one eye isaffected, the most common application of this technique is to takea small limbal biopsy measuring only a fewmillimetres in diameterfrom the patient’s non-injured eye, expanding the number ofepithelial cells by cultivation on donor amniotic membrane, thengrafting the resulting sheet of cultured cells back onto the ocularsurface [19,20]. Alternatively, dissociated biopsies have been sub-jected to ex vivo expansion in the presence of irradiated fibroblast

feeder cells, then transferred to either amniotic membrane [21] ora substrate prepared from fibrin glue prior to transplantation [22].While it remains difficult to assess the relative efficacy of thevarious techniques, the long-term outcomes indicate that there aresubstantial benefits for select cohorts [23]. A limitation of thecurrent techniques, however, is that they contain cells for onlyrepairing the epithelial surface. While residual scarring of thecentral cornea can be successfully treated using a conventionaldonor corneal transplant [24], there are currently no reliabletechniques for repairing the limbal stroma and as such long-termfunction of the limbal stem cell niche is likely to remain impaired. Itis our opinion therefore that the next generation of the culturedlimbal transplants should incorporate stromal as well as epithelialtissue components.

Fig. 3. Surgical options for corneal tissue reconstruction. (A) Primary culture of limbal epithelial cells following ex vivo expansion in presence of murine 3T3 feeder cells. (B) Grossview of human amniotic membrane (supported by nitrocellulose backing paper) carrying culture of limbal epithelial cells. (C) Application of limbal epithelial cells on amnioticmembrane to the ocular surface of a patient with limbal stem cell deficiency. (D) Pre-operative appearance of patient with limbal stem cell deficiency resulting from burn injury.(E) Post-operative appearance of same patient eye as shown in D, 5 months following application of limbal epithelial cells cultured on donor amniotic membrane. (F) Keratoconiceye displaying scarring to central cornea. (G) Appearance of same patient eye displayed in F following receipt of a full thickness donor corneal transplant (penetrating keratoplasty).

D.G. Harkin et al. / Biomaterials 32 (2011) 2445e24582448

Biomaterials for limbal reconstruction should be designed tosupport the attachment and growth of both epithelial as well asstromal cell types including those of fibroblastic and vascularlineage (Table 1). Physical separation of the epithelial and stromalcells to retain normal tissue boundaries could be achievedthrough use of a composite material such as a prosthetic base-ment membrane on which to grow the epithelium, overlying a 3Dscaffold which would be more biomimetic for stromal cells. Theprosthetic basement membrane should also be designed tosupport sufficient exchange of nutrients and regulatory mole-cules necessary for maintenance of the stem cell niche. Trans-parency is not a primary requirement given that the limbusresides at the outer margin of the visual pathway, however, it willbecome an issue if the transplant is required to repair the centralcornea as well.

2.3. Tissue for repairing the corneal stroma

Primary causes of stromal dysfunction include scarring, subse-quent to either trauma or infection, and keratoconus, a degenera-tive disorder which is characterised by thinning and ectasia of thecorneal stroma (Fig. 2C and D). In each case, the conventionaltreatment is to transplant tissue from a deceased donor (Fig. 3 F andG). Owing to the absence of blood vessels and immune cells withinthe cornea, such transplants have proven to be highly successfulwith many lasting over 10 years. Traditionally, most transplants ofcorneal stroma have utilised a procedure known as penetratingkeratoplasty (PK) which involves replacing the entire thickness ofthe cornea including potentially an otherwise healthy endotheliallayer. Recent approaches aimed at preserving a healthy endothe-lium have therefore favoured lamellar grafts whereby only the

Table 1Ideal properties of biomaterials for bioengineering ocular tissues.

Ocular Tissue Functions Biomaterial properties

Thickness Permeability Transparency Cell types requiredto interact

Degradationproperties

Potential design

Corneosclerallimbus

Maintains cornealepithelium viaresidentpopulation ofepithelialprogenitor cells.

w250 mm Essential to supportstem cell niche.

Non-essential. Limbal epithelialcells.Potentially requireadditional celltypes to fullyrecreate the limbalstem cell niche (e.g.limbal fibroblastsand vascular celltypes).

Should not elicit animmunologicalresponse.Should notinterfere withfunctioningof the stem cellniche.

Thin porousmembraneto support limbalepithelium,combined withporousscaffold to supportstromal layer.

Cornealstroma

Refractive elementin the visualpathway.

w500 mm Essential tofacilitatemovement ofnutrientsand waste productsthroughstroma.

Essential forvision.

Keratocytes.Considerationshould also begiven toincorporatingcorneal epithelialand endothelialcells.

Should not elicit animmunologicalresponse.Should remaintransparent.

Poroustransparentscaffold.

Cornealendothelium

Maintains clarity ofcornea by acting asa barrier andpumping of water.

w2e10 mm Essential to supportmovement of waterand nutrients.

Essential forvision.

Corneal endothelialcells.

Should not elicit animmunologicalresponse.Should remaintransparent.

Ultrathin,permeable,transparentmembrane.

Ruysch’scomplex

Maintains retinalattachment andforms outer blood-retinal barrier.

w2e10 mm Essential to supportepithelial transportmechanisms of RPE.

Non-essential. Retinal pigmentepithelial cells.Potentialincorporation ofmicrovascularendothelial cells torecreatechoriocapillaris.

Should not elicit animmunologicalresponse.Degradationproducts shouldnot adverselyaffect functioningof RPE and retinalcells.

Ultrathin porousmembranePotential tocombinewith porousscaffoldto supportmicrovascularendothelial cells.

D.G. Harkin et al. / Biomaterials 32 (2011) 2445e2458 2449

stroma and attached epithelium are replaced. Despite the relativesuccess of both of these types of donor corneal transplants, therecan be restrictions in regions affected by low donation rates.Furthermore, the use of tissue collected at short notice fromnumerous deceased donors runs the risk of some enabling trans-mission of infectious agents. For these reasons, techniques forcreating bioengineered constructs of corneal stroma have becomea popular area of research. The resulting methods can be dividedinto two main categories: (1) those which utilise collagen-basedhydrogels [25e27], and (2) tissue self-assembled from culturedstromal cells [28,29]. Some collagen hydrogels are designed to beimplanted without pre-seeding of stromal cells, relying instead onrecruitment of these cells following implantation [30]. Thisapproach is perhaps influenced by the fact that standard techniquesfor growing cells from the corneal stroma utilise serum-supple-mented growth medium which promotes differentiation of kera-tocytes into fibroblasts and myofibroblasts [31]. This isa disadvantage since keratocytes normally only differentiate intofibroblasts and myofibroblasts during wound healing and retentionof myofibroblasts is associated with scarring which reduces vision.Careful attention should therefore be paid to the effects of bioma-terials on stromal cell phenotype.

Biomaterials for reconstructing the corneal stroma (Table 1)should ultimately support growth of an epithelial layer as well asproviding a vehicle for transplantation of keratocytes. An addedcomplication however is that, unlike tissue for reconstructing thelimbus, materials for reconstructing the corneal stroma shouldpreferably maintain a similar level of transparency to that seenwithin the native tissue. Significant progress in this area has beenachieved in the case of collagen hydrogels through a number oftechniques including incorporation of chondroitin sulphate [32] and

via arranging collagen fibres into orthogonal sheets using magneticalignment techniques [27]. Ultimately, the stromal cells that residewithin the reconstituted tissue, either through prior seeding in vitroor subsequent recruitment in vivo, should ideally maintaina phenotype similar to that displayed by native keratocytes.

2.4. Tissue for repairing the corneal endothelium

Approximately one third of all donor corneal transplants per-formed in developed countries are due to endothelial dysfunctionarising through trauma (including that associated with surgicalprocedures) or various endothelial dystrophies [18] (Fig. 2E). Thetraditional approach once again has been to replace all layers of thecornea (through PK) with the emphasis being on restoring a suffi-cient density of viable endothelial cells to maintain a clear corneavia the tissue’s ability to pump water. Given this emphasis, recentdevelopments in surgical technique have focussed on replacingonly the endothelial layer with or without a portion of posteriorstroma thus giving rise to the concept of an endothelial transplant[33]. The principal advantage to this technique is that the patient’sstroma essentially remains intact and thus avoids distortions instructure associated with full thickness grafts that can produceastigmatism. Due to difficulties encountered in integrating host anddonor stroma, it would be better to transplant donor endotheliumattached to DM alone. Furthermore, general limitations associatedwith use of donor tissue (as described above) have prompted effortsto manufacture corneal endothelial tissue equivalents in the labo-ratory from biomaterials and cultured cells. Unfortunately, adulthuman corneal endothelial cells display little evidence of mitosis insitu and are notoriously difficult to grow in vitro. Laboratoriesdedicated to the task have nevertheless produced some promising

D.G. Harkin et al. / Biomaterials 32 (2011) 2445e24582450

results [34] and there also remains the potential to produce one daycell substitutes via manipulation of embryonic stem (ES) cells orinduced pluripotent stem (iPS) cell lines. While a variety of mate-rials including collagen [35] and polylactides [36] have been shownto support the growth of corneal endothelial in vitro, a prostheticDM must ultimately support endothelial cell function and more-over must be amenable to handling and insertion into the anteriorchamber of the eye by the surgeon.

The paramount characteristics for a prosthetic DM are sum-marised in Table 1. Firstly, the material used should naturallysupport the attachment and growth of endothelial cells toa minimum density (2000 cells/mm2) necessary to maintaina normal level of corneal hydration (and thus clarity) in vivo.Secondly, the material should be sufficiently permeable to allow foradequate movement of water and solutes including gases via thebasal surface of the cells. Thirdly, the optical properties of thematerial itself and any eventual degradation products should notinterfere with the visual pathway. Finally, the transplanted cellsshould become permanently integrated into the posterior surfaceof the cornea despite any potential degradation of the carriermaterial. In addition to these considerations, the carrier materialitself and any potential degradation products should not incite aninflammatory or immunological response.

2.5. Overview of the retina

The retina is a vastly more complex structure compared to thecornea, containing a highly ordered arrangement of neurons andsupporting glial cells which together are responsible for lightsensation. While photoreceptor neurons (rods and cones) are theprimary sensory element in the visual pathway, in order for light toreach these cells it must first pass through several distinguishablelayers of nervous tissue which together with the photoreceptorlayer constitute the sensory retina. The outermost layer of theretina immediately adjacent to the photoreceptor cells is formed bya single layer of pigmented epithelial cells known as the retinalpigment epithelium (RPE).

The RPE displays numerous activities that are essential to visionincluding: daily phagocytosis of photoreceptor outer segments(POS) containing toxic photo-oxidation products, transport of waterand nutrients to the photoreceptor cells, and growth factor secre-tion [37]. These functions are highly dependent upon maintenanceof a polarised epithelial phenotype which includes extension ofmicrovilli from the apical surface, a polarised distribution oftransport molecules, and differences in-growth factor secretionbetween the apical (pigment epithelium derived factor; PEDF) andbasal surface of the RPE (vascular endothelial growth factor; VEGF)[38]. The highest density of RPE cells is observed within a 6-mmdiameter circular zone located at the optical centre of the retinaknown as the macula which contains the highest concentration ofcone photoreceptors and is essential for observing fine detail.

A thin layer of connective tissue, known as Bruch’s membrane(BrM), resides beyond the RPE. It lies immediately adjacent toa network of fenestrated capillaries known as the choriocapillaris.The choriocapillaris constitutes the innermost part of the vasculartunic of the eye known as the choroid. BrM is a pentalaminarstructure consisting of: (1) the RPE basement membrane, (2) aninner fibrous layer, (3) a central elastic layer, (4) an outer fibrouslayer, and (5) the basement membrane of blood vessels within thechoriocapillaris. The net result is a dense layer of ECM which actslike a sieve to materials travelling between the retina and choroid.Together, the RPE, BrM and choriocapillaris form an importantfunctional unit which has been historically referred to as Ruysch’scomplex [39]. The primary functions of Ruysch’s complex are toanchor the retina to the back of the eye and to regulate the

transport of waste products and essential nutrients between thephotoreceptors and the choroid (Fig. 1C).

2.6. Tissue for repairing Ruysch’s complex

While diseases of the retina are relatively rare in childhood, theRPE and BrM display cumulative damage with age. These changesare especially noticeable within the macula and are presumablydue to the increased workload associated with a higher number ofphotoreceptor cells within this region. In time these changes oftenlead to a condition referred to as age-related macular degeneration(AMD) which in developed countries is the leading cause ofblindness in people over 50 years of age [39]. While the exactcauses of AMD have yet to be determined there is a strong associ-ation between thickening of BrM, reduced numbers of RPE cells,and eventually death of the overlying photoreceptor cells [39].Thickening of BrM is initially due to normal ageing processesincluding lipid deposition, calcification (of central elastic layer) andalterations in ECM deposition, but is subsequently exacerbated bythe deposition of extracellular deposits beneath the RPE andwithinBrM known as drusen [40] (Fig. 2G). The origins of drusen arecomplex (including products from photoreceptor cells, RPE, choroidand serum), and its presence is associated with early or “dry” formsof AMD. Advanced or “wet” forms of the disease are additionallyassociated with aberrant growth of leaky blood vessels from thechoroid into the retina, entering via breaks that develop withinRuysch’s complex. For the most part, little can be done to treat theinitial stages of the disease, but some success in treating progres-sion of advanced AMD has recently been achieved through use ofanti-VEGF drugs [41,42].

The apparent links between AMD and reduced numbers of RPEcells has lead to the view that the condition might benefit fromtransplants of RPE cells. Encouraged by promising results in animalmodels a number of surgeons have attempted RPE transplants usingeither autologous flaps (transferred to the macula from moreperipheral locations) or allogenic transplants of foetal RPE [43].Despite the impressive surgical skills required for this work, theclinical results to date have overall been disappointing. A recentreview of the field has led to the conclusion that efforts arehamperedby twomajor factors [44], (1) the requirement for a robustsupply of functioning RPE cells which will integrate and avoidrejection, and (2) the requirement to remove thediseasedBrMalongwith drusen deposits and replace this with a prosthetic BrM.Unfortunately, adult RPE cells, like corneal endothelial cells, aredifficult to cultivate but recent breakthroughs in manipulating thephenotypeof iPS cells [45] aswell as ES cell lines [46] indicate that analternative supply of human RPE is possible. While a variety ofmaterials havebeen shown to support the attachment andgrowthofRPE cells invitro,onlya fewof these have thus far been trialled in vivoand as such a prosthetic BrM has yet to be clearly identified [47].

A prosthetic BrM should support the attachment, growth anddevelopment of a normal RPE phenotype including those featuresassociated with apical-basal polarity in these cells. Given thetransport functions of the RPE, permeability of the biomaterial isnaturally again a key consideration. Recent in vitro studies of BrMpermeability provide valuable data for comparison [48]. Opticaltransparency is not essential since the material would not beplaced in the visual pathway. Crucially, however, the materialshould be designed to integrate with the choriocapillaris. Thebasal surface of the biomaterial might therefore need to beadjoined to a thin 3D scaffold to encourage in-growth of capil-laries from the choroid. Alternatively, the basal surface of theprosthetic BrM could be seeded with cultured microvascularendothelial cells to reconstruct components of the choriocapillarisin vitro prior to implantation. Given the close proximity of

Fig. 4. Examples of materials fabricated from silk fibroin in dry state. (A, D, G) Thin transparent membrane cast from aqueous solutions of purified fibroin. Images demonstrateappearance of 16-mm disc when either held with forceps, positioned over printed text, or when viewed by scanning electron microscopy respectively. (B, E, H) Analogous images ofan ultrathin porous membrane created using a 1:0.03 mixture of aqueous fibroin with PEO (Mv 900 kg/Mol). (C, F, I) Fibrous mat prepared from degummed and partially solubilisedsilk fibres. (J) Fibroin membrane prepared in similar manner to that shown in A, but patterned by casting on the grooved surface from a compact disc. (K) Porous sponge created byfreeze drying aqueous solution of fibroin. (L) Fibrous mat created by electrospinning an aqueous solution of purified fibroin.

D.G. Harkin et al. / Biomaterials 32 (2011) 2445e2458 2451

transplanted material to highly sensitive photoreceptor cells, itwill also be important to consider the potential toxicity ofbreakdown products including those that are likely to producelocalised changes in osmolarity or pH.

3. Is there a role for fibroin?

Given that a variety of biomaterials are already being investi-gated for ocular tissue engineering and that some of these have

D.G. Harkin et al. / Biomaterials 32 (2011) 2445e24582452

already produced promising results in clinical trials [23,30], therecent addition of fibroin to this list therefore requires carefulconsideration as to what novel combination of properties might beafforded by this material. Since the basic properties of fibroin havepreviously been reviewed in detail, we refer the reader to theseexisting resources [49,50] and limit our comments here to thoseproperties which are most relevant to potential applications inocular tissues.

3.1. What types of materials can be produced from fibroin?

The majority of prior studies are based upon fibroin extractedfrom cocoons produced by the domesticated silkworm B. mori. Thefirst step in this process is to remove the glue-like protein sericinwhich is associated with adverse immunological reactions to silksutures [49]. The resulting degummed fibres can then be solubilisedusing highly concentrated solutions of lithium bromide, and dia-lysed to produce aqueous solutions of fibroin which can then befabricated into a variety of different materials (Fig. 4). Perhaps thesimplest of these fibroin-based materials are the transparent filmscreated by drying aqueous solutions of fibroin on a hydrophobicsurface. Ultrathin films measuring just a few microns across canroutinely be achieved using these methods and commercial-scalemanufacturing is possible using conventional tape-casting equip-ment. The architecture of fibroin films can be varied via soft litho-graphic techniques such as casting onto patterned surfaces [51]. Thesecondary structure of the films can also be manipulated bychanging the rate and temperature at which the fibroin solution isdried and by the method of annealing with either alcohol or watervapour [52,53]. These processing variables have all been shown toinfluence the secondary structure which in turn affects both themorphology and properties of the films (i.e. surface hydrophilicity,mechanical properties and degradation). Moreover, the perme-ability of films can be improved by introducing pores by castingfibroin solutions in the presence of more hydrophilic polymers suchas poly(ethylene oxide) (PEO) (Fig. 4 panels B, E and H). Lowmolecular weight PEO (300 g/Mol) results in films with sub-nano-meter diameter pores [54], while higher molecular weight PEO(>900 kg/Mol) can result in micron sized pores [55]. Increasedporosity howeverhas detrimental effects on transparencyespeciallyin thicker fibroin films. Other techniques enable the production of3D structures from fibroin. The simplest of these approaches is toallow partially solubilised degummed fibroin fibres to dry intoa fibrous mat [56]. More commonly, however, 3D porous scaffoldshave been produced from regenerated aqueous or hexafluoroaceticacid solutions of fibroin. Sponges can be prepared by dehydration offrozenfibroin solutions under vacuum (freeze drying) [57], or by theincorporation of porogens such as salt particles into the fibroinsolution [58]. Electrospinning of concentrated (8e21 wt %) aqueousfibroin solutions can also be used to create 3D fibrous mats [59]. Asthe mats, films or scaffolds can all be manufactured solely fromaqueous solutions, thepossibility that residual toxic solvent couldbepresent in the final material is eliminated. Fibroin is therefore quiteversatile when it comes to fabricating different structures forpotential tissue engineering applications in the eye. Nevertheless,this versatility is not unique to fibroin and so other factors includingbiocompatibility and handling properties will ultimately decide ifthis biomaterial can be translated into the clinic.

3.2. The advantages and disadvantages of using fibroin for oculartissue reconstruction

A number of features in addition to structural versatility suggestpotential advantages of fibroin over some other natural andsynthetic biomaterials. To begin, the more complex structure of

fibroin compared with synthetic materials such as polylactidesmeans that there are a number of functional groups already presentin the material that can be reacted with molecules of interest (e.g.RGD sequences). Fibroin is usually functionalised through eitherthe lysine residue (contains amine group) which accountsfor w 0.3% of all amino acid residues in fibroin or through tyrosineresidues (contains phenol group) which accounts for w 5% of allamino acids in fibroin (NCBI Refs. AAN28165.1 and AAC32606.1).Furthermore, advances in genetic engineering are enablingproduction of chimeric molecules in which fibroin sequences arecombined with those found in other ECMmolecules [50]. Of coursesimilar strategies can be used with ECM proteins such as collagenswhich unlike fibroin have the added benefit of being found natu-rally within the eye and thus perhaps represent a more appropriatechoice. Nevertheless, while materials produced from collagen havebeen shown to support growth of a variety of ocular cell types, theylack mechanical strength unless modified through cross-linking orcombined with other materials [60]. In comparison, however,fibroin-based materials still display strength at least one orderabove that observed for collagen-based materials [49] and as suchthis property may prove to be a critical advantage for recon-structing some ocular tissues.

The biodegradation properties of fibroin may also be exploitedto some advantage within the eye. Fibroin’s amino acid sequencefavours the formation of b-sheet structures which are exceptionallystable and hydrophobic in nature. As a result, fibroin-based mate-rials degrade more slowly than other commonly used materialsincluding collagen, fibrin and polylactides. By varying the propor-tion of b-sheet structure, however, the rate of degradation can becontrolled. A faster rate of degradation may be helpful in avoidingchronic foreign body reactions that promote a fibrotic response.Conversely, a slower rate of degradation may be useful to allowmore time for tissue integration especially in the case of trans-planting single layers of cells (e.g. corneal endothelial cells) whichcould be easily lost if the supporting material breaks down tooquickly following transplantation. The more significant advantagecompared with synthetic materials, however, may well reside inthe nature of the degradation products (peptides) which are lesslikely to cause local changes in tissue osmolarity and pH than canbe expected with degradation of aliphatic polyesters. For thisreason, there may therefore be clear advantages in using fibroin asa prosthetic BrM for RPE cells within the retina given the proximityto highly sensitive photoreceptor cells.

3.3. Prior studies of ocular cell responses to fibroin

The potential use of fibroin for tissue engineering applications inthe eye was first highlighted by our own group by demonstratingthat growth of primary limbal epithelial cells on fibroin films (over72 h) is similar to that observed on tissue culture plastic [9,10]. Inmore recent studies we have demonstrated that attachment (over4 h) of a human corneal epithelial cell line (HCE-T) to fibroin issignificantly improved in the presence of serum proteins (Fig. 5).Furthermore, attachment of HCE-T cells to fibroin in the presence ofserum is comparable to that seen on denuded amniotic membrane,a popular choice of substrate for transplanting cultured epithelialcells to the ocular surface. Evidence of basement membraneformation can also be seen via staining using the periodic acidSchiff (PAS) method. While these results are encouraging, itremains unclear whether primary cultures of epithelial cells grownon fibroin display evidence of immature progenitor-like cells whichwill be necessary for clinical efficacy. For example, primaryepithelial cultures when grown on tissue culture plastic rapidly loseprogenitor cell characteristics unless grown in the presence ofa fibroblast feeder layer. Traditionally, 3T3 fibroblast cell lines

Fig. 5. Visual comparison of the human corneal epithelial cell line, HCE-T, growing on either (A/D) tissue culture plastic, (B/E) human donor amniotic membrane, or (C/F) films castfrom aqueous solutions of silk fibroin. Cells were seeded at a density of 5000 cells per square centimetre and cultured in the presence of 10% foetal bovine serum for 4 h beforestaining with the vital dye fluorescein diacetate. (A, B & C) Phase contrast images. (D, E & F) Corresponding fluorescence images captured using fluorescein filter set. (G) Quantitativecomparison of HCE-T cell attachment to tissue culture plastic, donor amniotic membrane and films cast from silk fibroin, when seeded in the presence or absence of 10% foetalbovine serum. Bars represent the mean � SEM for the total number of viable cells (judged by staining with fluorescein diacetate) per five standardised low power (10� objective)fields. Asterisk indicates significant difference (Student’s T two-tailed test, p < 0.05) for cell attachment to fibroin in the presence of serum (H) Histological section of confluent HCE-T culture after two weeks growth on donor amniotic membrane at the aireliquid interface. Staining with Schiff’s reagent following oxidation with periodic acid (PAS method)reveals evidence of glycoprotein deposition (arrow) at the cultureemembrane interface. (I) Comparative histological section of HCE-T after growth on silk fibroin membrane.Glycoprotein deposition can again be seen at the cultureemembrane interface (arrow).

D.G. Harkin et al. / Biomaterials 32 (2011) 2445e2458 2453

D.G. Harkin et al. / Biomaterials 32 (2011) 2445e24582454

established from mouse embryos are used for this purpose, buta number of recent studies including those by our own group haveshown that primary cultures of human limbal fibroblasts can alsobe used [31]. At time of submitting this paper, a detailed study byHiga et al. has just appeared in press [61] characterising thephenotype of rabbit primary corneolimbal epithelial cells whengrown on fibroin films mounted above a 3T3 feeder layer by usinga porous cell culture insert. Two particularly important observa-tions are made in this study. Firstly, poor growth was observedunless the fibroin films were rendered porous by casting in thepresence of low molecular weight PEO (300 g/Mol). Secondly, thesuccessful cultures developed a normal corneal phenotype andretained good evidence of progenitor cell activity. Logically, theseresults suggest that growth of the rabbit cells on fibroin wasdependent upon factors secreted by the 3T3 feeder layer below.Since growth was not examined on porous films in the absence of3T3 cells, however, it remains possible that porosity alonemay haveimproved growth. Further studies are therefore required to clarifythis point. Moreover, these studies need to be repeated using

Fig. 6. (A) Proposed flow diagram of manufacturing processes required to produce fibroin-either alone or as multiple compressed layers has been established through prior studies by Ccontain both epithelial as well as stromal tissue components are an important area for futuquality of fibroin/cell constructs for ocular tissue reconstruction.

human cells since rabbit corneas contain a broader distribution ofepithelial progenitor cells [62].

With respect to the corneal stroma, Lawrence et al. [12] havedemonstrated that porousultrathinfibroinfilms support the growthof primary rabbit corneal fibroblasts (activated keratocytes) as wellas a transformed cell line established from human corneal fibro-blasts. Fibroblast growth on fibroin was found to be less than thatobserved on tissue culture plastic, but the cells retained productionof ECM molecules associated with a normal corneal phenotype.Moreover, alignment of fibroblasts, which is considered tocontribute to corneal clarity, can be achieved through patterning thefibroin surface via simple lithographic techniques. In further studiesfrom this group, fibroblast alignment has been improved throughvarying the depth and width of grooves imprinted in the fibroinsurface [11], andfibroblast growth has been improved bychemicallybinding RGD peptide to the membrane surface [63]. Assuming thatfibroin can also be used as substrate for keratocytes, incorporation ofcorneal epithelial cell and corneal endothelial may well lead to theproduction of a functioning corneal tissue equivalent.

based materials for ocular tissue reconstruction. Feasibility of using transparent filmshirila et al [9,10] and Lawrence et al. [12] respectively. Composite materials designed tore studies. (B) Examples of criteria which are likely to be important for assessing the

D.G. Harkin et al. / Biomaterials 32 (2011) 2445e2458 2455

3.4. Questions that require further investigation

Like many other biomaterials of natural origin, fibroin presentsits own list of challenges. For the most part, these challenges arequestions which remain unanswered concerning how fibroinshould best be optimised for use with ocular cells and indeed cellsin general. To begin, the mechanism of cell attachment to fibroinremains an unanswered question especially in the case of the morecommonly used B. mori fibroin since this protein lacks structuralmotifs known to be associated with cell attachment (NCBI Refs.NP_001037488.1 and NP_001106733.1). Fibroin isolated fromthe non-domesticated silk worms Antheraea mylitta (NCBIRef. AAN28165.1) and Antheraea pernyi (NCBI Ref. AAC32606.1)contain the cell-adhesion peptide RGD and so it might be expectedthat these sources of fibroin are perhaps more suitable than thatobtained from B. mori. Nevertheless, wild silkworm cocoons areproblematic to obtain and regenerated fibroin solutions are difficultto prepare. Consequently, fibroin is usually extracted directly fromthe silk glands of the final instar larva. In the absence of recognis-able cell-binding motifs, the mechanism of cell attachment toB. mori fibroin most likely involves a combination of non-specific(e.g. electrostatic) as well as specific processes as seen when cellsare seeded onto tissue culture plastic, with the latter being medi-ated through ECM proteins such as those found in serum (e.g.fibronectin and vitronectin) which become attached to the culturesurface via simple hydrophobic interactions. Novel cell-adhesionmotifs have been proposed to exist near the N-terminus of fibroinbut await confirmation [64]. Similarly, a detailed examination ofthe effects of different processing techniques including annealingconditions (alcohol versus water vapour) is required to determinehow this can provide the optimal substrate for encouraging theattachment and growth of various ocular cell types. Whatever themechanism of cell attachment, it remains possible that currentformulations of materials produced from fibroin may preferentiallyselect for only certain subsets of cells and thus it must be estab-lished whether those that are selected are likely to support clinicalefficacy (e.g. progenitor cells in case of limbal epithelium). Studiesof cell phenotype following attachment of ECM proteins or cell-adhesion peptides will be required to address this problem.

Assuming that fibroin formulations can be tuned to selectivelyencourage attachment of the correct cell types, it will then benecessary to ensure that these cells retain or develop a normal

Table 2Potential strategies for use of fibroin in ocular tissue reconstruction.

Ocular Tissue Biomaterial(s) require

Corneosclerallimbus

Thin porous membraneepithelium.Combined with porous sstromal cell types includvascular cell types.

Corneal stroma Porous transparent scaff

Corneal endothelium Thin, permeable, transpa

Ruysch’s complex Ultrathin porous membrPotential to combine wito support microvascula

phenotype required for tissue function. In the case of limbalepithelial progenitor cells, these cells should ideally retain theability to divide and differentiate into normal corneal epitheliumboth in vitro and following delivery to the ocular surface. A varietyof markers are available to enable identification of immature limbalepithelial cells including ABCG2 [65], C/EBPd [66], Bmi-1 [67] andΔNp63a [68]. Moreover, the degree of differentiation can bedetermined morphologically and by increasing expression of thekeratin pair K3/K12 which is specific for the corneal epithelium[15]. A range of markers are also available to assess stromal cellphenotype. In situ, corneal keratocytes express CD34 [69], butduring cultivation in the presence of serum this marker is quicklylost and replaced by markers for wound healing fibroblasts (CD90)and myofibroblasts (a-smooth muscle actin) [70]. The differenti-ated corneal endothelium is generally characterised by a tightlycobblestoned morphology, but perhaps a more important studywill be to assess the degree of cell density and pump functionswhich are required to maintain a clear corneal stroma. Likewise,RPE cells are generally characterised by expression of ZO-1, cyto-keratin 18 and RPE-65, but their function following transplantationwill be dependent upon maintenance of a polarised morphologyand various associated activities including phagocytosis andgrowth factor secretion.

A final issue of prime importance is that there is presently littledata available on what happens when fibroin is implanted into theeye. While there is a long and mixed history of silk sutures beingused in ophthalmic surgery, these results are often complicated bythe presence of sericin in many formulations [49]. Moreover, it ispossible for fibroin alone to elicit a damaging immunologicalresponse in select patients. Results from our own preliminarystudies using a rabbit model indicate that 2-mm discs cut fromfibroin films are well tolerated when implanted within the centralcorneal stroma, with little evidence of leukocyte recruitment orfibroblast proliferation being observed over 60 days [71]. A similarresult has recently been reported by Higa et al. [61] using 5-mmfibroin film discs. Nevertheless, both results may well have beeninfluenced by the avascular nature of the central cornea. Quitedifferent results may therefore be encountered for similar formu-lations of fibroin when implanted in or near the limbus, within thesub-retinal space, or when larger sheets of fibroin are used. Studiesof biocompatibility and biodegradation should therefore beengiven high priority. We envisage that an important component to

Potential strategies supported through useof fibroin

to support limbal

caffold to supporting fibroblasts and

Upper porous membrane can be createdusing film casting techniques incorporatingPEO to create pore through phaseseparation [12].Lower scaffold could be produced fromaqueous fibroin solutions via a variety oftechniques including freeze drying [57], saltleaching [58] or electrospinning [59].Degummed and partially solubilised silkfibres also offer a potential solution.

old. Compressed layers of porous ultrathinfibroin films as described by Lawrence et al.[12] and more recently by Gil et al [11,63].

rent membrane. As described above for upper layer of limbalcomposite.

ane for RPE.th porous scaffoldr endothelial cells.

As described above for upper layer of limbalcomposite.As described above for limbal stromalscaffold but with significantly reducedthickness.

D.G. Harkin et al. / Biomaterials 32 (2011) 2445e24582456

these studies will be to develop an improved understanding ofpermeability requirements given that thicker materials can lead totissue necrosis and subsequent inflammation by simply impairingthe movement of nutrients and waste products. Logically, porousfibroin materials described in this paper should provide a solutionto any inadequate permeability. Unfortunately, conventionalporous scaffolds (Fig. 4 panels I, K & L) lack transparency andultrathin microporous films (Fig. 4 panel H) may yet prove toofragile and difficult for surgeons to handle. Therefore a solutionmay therefore be to produce ultrathin fibroin membranes withporosity in the sub-micron range via use of track etching tech-niques as routinely used in manufacture of cell cultureemembraneinserts (Nuclepore�, Whatman). It also remains possible thatsufficient permeability for some ocular tissues may be provided bysimply using nonporous membranes if they are produced thinenough. Thus the key issue is one of fibroin permeability ratherthan fibroin porosity per se and this will be especially importantwhen developing substrata for corneal endothelial cells and RPEcells given the transport functions of both cell types.

4. Conclusion and forward vision

Despite the existence of significant knowledge gaps, weconclude that fibroin displays potential value as a biomaterial forocular tissue reconstruction and propose that this be exploredthrough the following strategies (Fig. 6). First and foremost, there isan urgent requirement for studies investigating the effects ofvarious fibroin-based materials when implanted into the eye. Keysurgical sites should include the limbal stroma, corneal stroma andsub-retinal space. The materials should be selected based upontheir structural relevance to tissues being repaired (Table 1) andmanufactured according to standardised protocols includinga range of appropriate quality control tests (Fig. 6. and Table 2).

In the case of the corneoscleral limbus, we are currentlyexploring the effects of fibroin on the phenotype of primary humanlimbal epithelial cells and are developing a composite material toenable incorporation of limbal stromal cells. We propose to culti-vate the limbal epithelial cells on an ultrathin porous film con-structed from an aqueous fibroin solution, adhered to the surface ofa 3D scaffold seeded with limbal stromal cells [72]. The stromalscaffold could be manufactured by either using degummed,partially solubilised silk fibres or fibroin mats produced by elec-trospinning. Given the role of the limbus as a stem cell niche weanticipate the requirement to incorporate additional componentsas found in situ including ECM molecules and potentially growthfactors as well. Key molecules of interest include tenascin C andselective isoforms of laminin (containing g3 chain) owing to theircolocalisation with limbal epithelial progenitor cell markers inhistological specimens [73,74].

For the corneal stroma, the achievements of Lawrence et al. [12]using compressed layers of transparent porous ultrathin fibroinfilms, encourage further work in this area to carefully evaluate thephenotype of primary keratocyte cultures established on andwithin this 3D material. Moreover, the effects of epithelial cellswhen applied to the upper (limbal epithelial cells) and lowersurfaces (corneal endothelial cells) of this material should also bestudied. Given the promising clinical results achieved through useof modified collagen gels [30] there are also significant reasons toinvestigate fibroin-collagen composites including potential fibroin-collagen chimeric proteins.

Fibroin-based materials for supporting reconstruction of thecorneal endothelium may well be incorporated into those forrepairing the stroma, but it is important tofirst understandwhetherfibroin is a suitable substrate for corneal endothelial cells. Trans-parent, permeable, ultrathin membranes of fibroin would seem to

provide the best place to begin investigations. Once again, cellattachment and correct cell phenotype will be important consid-erations with the emphasis being on establishing a suitable densityand pump function tomaintain a clear cornea post-transplantation.

Reconstruction of the blood-retinal barrier (Ruysch’s complex)should initially commencewith developing a suitable substitute forBrM which supports the attachment, growth and polarity requiredto support the numerous activities of RPE cells. Permeability of thematerial should be a key consideration in order to facilitateadequate movement of nutrients and waste products between thesub-retinal space and the choriocapillaris. Given these priorities,we are presently evaluating porous ultrathin fibroin films asa substrate for RPE cells [75] and there is potential to incorporatecells for reconstructing the choriocapillaris by seeding of micro-vascular endothelial cells to the alternate surface.

Acknowledgements

This work was supported by funding received from the NationalHealth & Medical Research Council of Australia (Project Grant ID:553038) awarded to authors DGH, TVC, DWH and IRS. The authorsacknowledge Professor David Kaplan (Tufts University, Boston,USA) for his advice and encouragement. Thanks are also extendedto Dr Sunil Kale (presently of Sahajanand Medical Technologies Pvt.Ltd, Gujarat, India) for assistance with creating the mat of electro-spun fibres of fibroin (presented in Fig. 4 panel L) during his time asa visiting postgraduate student in our laboratory. Clinical images ofcornea (Fig. 2 panels A, B, C and E; Fig. 3 panels F and G) are courtesyof Professor Lawrence Hirst (Queensland Eye Institute, Brisbane,Australia). Fundus images (Fig. 2 panels F and G) were provided byAssociate Professor Tony Kwan (Queensland Eye Institute, Brisbane,Australia). Images of procedure and outcomes for transplantation ofcultured limbal epithelial cells (Fig. 3 panels B to E) were producedby DGH in collaboration with Dr Andrew Apel (Queensland EyeHospital, Brisbane, Australia). Images presented in Fig. 6. wereproduced with assistance from Ms Cassie Rayner (Institute forHealth and Biomedical Innovation, Queensland University ofTechnology, Brisbane, Australia). We further acknowledge that allstudies involving animals or humans referred to in this manuscriptwere performed with appropriate institutional research ethicscommittee approvals.

References

[1] Vepari C, Kaplan DL. Silk as a biomaterial. Prog Polym Sci 2007;32:991e1007.[2] Bondar B, Fuchs S, Motta A, Migliaresi C, Kirkpatrick CJ. Functionality of

endothelial cells on silk fibroin nets: comparative study of micro- and nano-metric fibre size. Biomaterials 2008;29:561e72.

[3] Hofmann S, Knecht S, Langer R, Kaplan DL, Vunjak-Novakovic G, Merkle HP,et al. Cartilage-like tissue engineering using silk scaffolds and mesenchymalstem cells. Tissue Eng 2006;12:2729e38.

[4] Kim HJ, Kim UJ, Kim HS, Li C, Wada M, Leisk GG, et al. Bone tissue engineeringwith premineralized silk scaffolds. Bone 2008;42:1226e34.

[5] Lovett M, Cannizzaro C, Daheron L, Messmer B, Vunjak-Novakovic G,Kaplan DL. Silk fibroin microtubes for blood vessel engineering. Biomaterials2007;28:5271e9.

[6] Park SY, Ki CS, Park YH, Jung HM, Woo KM, Kim HJ. Electrospun silk fibroinscaffolds with macropores for bone regeneration: an in vitro and in vivo study.Tissue Eng Part A 2010;16:1271e9.

[7] Unger RE, Sartoris A, Peters K, Motta A, Migliaresi C, Kunkel M, et al. Tissue-like self-assembly in cocultures of endothelial cells and osteoblasts and theformation of microcapillary-like structures on three-dimensional porousbiomaterials. Biomaterials 2007;28:3965e76.

[8] Zhang X, Baughman CB, Kaplan DL. In vitro evaluation of electrospun silkfibroin scaffolds for vascular cell growth. Biomaterials 2008;29:2217e27.

[9] Chirila T, Barnard Z, Zainuddin, Harkin D. Silk as a substratum for cellattachment and proliferation. Mater Sci Forum 2007;561-565:1549e52.

[10] Chirila TV, Barnard Z, Zainuddin, Harkin DG, Schwab IR, Hirst LW. Bombyxmori silk fibroin membranes as potential substrata for epithelial constructsused in the management of ocular surface disorders. Tissue Eng Part A2008;14:1203e11.

D.G. Harkin et al. / Biomaterials 32 (2011) 2445e2458 2457

[11] Gil ES, Park SH, Marchant J, Omenetto F, Kaplan DL. Response of humancorneal fibroblasts on silk film surface patterns. Macromol Biosci 2010;10:664e73.

[12] Lawrence BD, Marchant JK, Pindrus MA, Omenetto FG, Kaplan DL. Silk filmbiomaterials for cornea tissue engineering. Biomaterials 2009;30:1299e308.

[13] Ashwin PT, Shah S, Wolffsohn JS. Advances in cataract surgery. Clin ExpOptom 2009;92:333e42.

[14] Chirila TV. An overview of the development of artificial corneas with porousskirts and the use of PHEMA for such an application. Biomaterials 2001;22:3311e7.

[15] Schermer A, Galvin S, Sun TT. Differentiation-related expression of a major64K corneal keratin in vivo and in culture suggests limbal location of cornealepithelial stem cells. J Cell Biol 1986;103:49e62.

[16] Dua HS, Shanmuganathan VA, Powell-Richards AO, Tighe PJ, Joseph A. Limbalepithelial crypts: a novel anatomical structure and a putative limbal stem cellniche. Br J Ophthalmol 2005;89:529e32.

[17] Stepp MA, Zieske JD. The corneal epithelial stem cell niche. Ocul Surf2005;3:15e26.

[18] The Australian corneal graft Registry. 1990 to 1992 report. Aust N Z J Oph-thalmol 1993;21:1e48.

[19] Tsai RJ, Li LM, Chen JK. Reconstruction of damaged corneas by transplantationof autologous limbal epithelial cells. N Engl J Med 2000;343:86e93.

[20] Zakaria N, Koppen C, Van Tendeloo V, Berneman Z, Hopkinson A,Tassignon MJ. Standardized limbal epithelial stem cell graft generation andtransplantation. Tissue Eng Part C Methods 2010;16:921e7.

[21] Schwab IR. Cultured corneal epithelia for ocular surface disease. Trans AmOphthalmol Soc 1999;97:891e986.

[22] Rama P, Bonini S, Lambiase A, Golisano O, Paterna P, De Luca M, et al.Autologous fibrin-cultured limbal stem cells permanently restore the cornealsurface of patients with total limbal stem cell deficiency. Transplantation2001;72:1478e85.

[23] Rama P, Matuska S, Paganoni G, Spinelli A, De Luca M, Pellegrini G. Limbalstem-cell therapy and long-term corneal regeneration. N Engl J Med2010;363:147e55.

[24] Harkin DG, Barnard Z, Gillies P, Ainscough SL, Apel AJ. Analysis of p63 andcytokeratin expression in a cultivated limbal autograft used in the treatmentof limbal stem cell deficiency. Br J Ophthalmol 2004;88:1154e8.

[25] Duan X, McLaughlin C, Griffith M, Sheardown H. Biofunctionalization ofcollagen for improved biological response: scaffolds for corneal tissue engi-neering. Biomaterials 2007;28:78e88.

[26] Liu W, Merrett K, Griffith M, Fagerholm P, Dravida S, Heyne B, et al.Recombinant human collagen for tissue engineered corneal substitutes.Biomaterials 2008;29:1147e58.

[27] Torbet J, Malbouyres M, Builles N, Justin V, Roulet M, Damour O, et al.Orthogonal scaffold of magnetically aligned collagen lamellae for cornealstroma reconstruction. Biomaterials 2007;28:4268e76.

[28] Carrier P, Deschambeault A, Talbot M, Giasson CJ, Auger FA, Guerin SL, et al.Characterization of wound reepithelialization using a new human tissue-engineered corneal wound healing model. Invest Ophthalmol Vis Sci2008;49:1376e85.

[29] Guo X, Hutcheon AE, Melotti SA, Zieske JD, Trinkaus-Randall V, Ruberti JW.Morphologic characterization of organized extracellular matrix deposition byascorbic acid-stimulated human corneal fibroblasts. Invest Ophthalmol Vis Sci2007;48:4050e60.

[30] Fagerholm P, Lagali NS, Carlsson DJ, Merrett K, Griffith M. Corneal regenera-tion following implantation of a biomimetic tissue-engineered substitute. ClinTransl Sci 2009;2:162e4.

[31] Ainscough SL, Linn M, Barnard Z, Schwab IR, Harkin DG. Effects of fibroblastorigin and phenotype on the proliferative potential of limbal epithelialprogenitor cells. Exp Eye Res 2011;92:10e9.

[32] Orwin EJ, Borene ML, Hubel A. Biomechanical and optical characteristics ofa corneal stromal equivalent. J Biomech Eng 2003;125:439e44.

[33] Price MO, Price Jr FW. Endothelial keratoplasty - a review. Clin Exp Oph-thalmol 2010;38:128e40.

[34] Joyce NC, Zhu CC. Human corneal endothelial cell proliferation: potential foruse in regenerative medicine. Cornea 2004;23:S8e19.

[35] Mimura T, Yamagami S, Yokoo S, Usui T, Tanaka K, Hattori S, et al.Cultured human corneal endothelial cell transplantation witha collagen sheet in a rabbit model. Invest Ophthalmol Vis Sci 2004;45:2992e7.

[36] Hadlock T, Singh S, Vacanti JP, McLaughlin BJ. Ocular cell monolayers culturedon biodegradable substrates. Tissue Eng 1999;5:187e96.

[37] Strauss O. The retinal pigment epithelium in visual function. Physiol Rev2005;85:845e81.

[38] Sonoda S, Sreekumar PG, Kase S, Spee C, Ryan SJ, Kannan R, et al. Attain-ment of polarity promotes growth factor secretion by retinal pigmentepithelial cells: relevance to age-related macular degeneration. Aging 2010;2:28e42.

[39] de Jong PT. Age-related macular degeneration. N Engl J Med 2006;355:1474e85.

[40] Booij JC, Baas DC, Beisekeeva J, Gorgels TG, Bergen AA. The dynamic nature ofBruch’s membrane. Prog Retin Eye Res 2010;29:1e18.

[41] Bashshur ZF, Bazarbachi A, Schakal A, Haddad ZA, El Haibi CP, Noureddin BN.Intravitreal bevacizumab for the management of choroidal neovascularizationin age-related macular degeneration. Am J Ophthalmol 2006;142:1e9.

[42] Heier JS, Antoszyk AN, Pavan PR, Leff SR, Rosenfeld PJ, Ciulla TA, et al. Rani-bizumab for treatment of neovascular age-related macular degeneration:a phase I/II multicenter, controlled, multidose study. Ophthalmology2006;113:641e4.

[43] Steindl KSB. Retinal degeneration processes and transplantation of retinalpigment epithelial cells: past present and future trends. Spektrum Augen-heilkd 2008;22:357e61.

[44] Binder S, Stanzel BV, Krebs I, Glittenberg C. Transplantation of the RPE inAMD. Prog Retin Eye Res 2007;26:516e54.

[45] Buchholz DE, Hikita ST, Rowland TJ, Friedrich AM, Hinman CR, Johnson LV,et al. Derivation of functional retinal pigmented epithelium from inducedpluripotent stem cells. Stem Cells 2009;27:2427e34.

[46] Osakada F, Ikeda H, Sasai Y, Takahashi M. Stepwise differentiation of plurip-otent stem cells into retinal cells. Nat Protoc 2009;4:811e24.

[47] Hynes SR, Lavik EB. A tissue-engineered approach towards retinal repair:scaffolds for cell transplantation to the subretinal space. Graefes Arch Clin ExpOphthalmol 2010;248:763e78.

[48] Hussain AA, Starita C, Hodgetts A, Marshall J. Macromolecular diffusioncharacteristics of ageing human Bruch’s membrane: implicationsfor age-related macular degeneration (AMD). Exp Eye Res 2010;90:703e10.

[49] Altman GH, Diaz F, Jakuba C, Calabro T, Horan RL, Chen J, et al. Silk-basedbiomaterials. Biomaterials 2003;24:401e16.

[50] Wang Y, Kim HJ, Vunjak-Novakovic G, Kaplan DL. Stem cell-based tissueengineering with silk biomaterials. Biomaterials 2006;27:6064e82.

[51] Perry H, Gopinath A, Kaplan DL, Dal Negro L, Omenetto FG. Nano- andmicropatterning of optically transparent, mechanically robust, biocompatiblesilk fibroin films. Adv Mater 2008;20:3070e2.

[52] Lu Q, Hu X, Wang X, Kluge JA, Lu S, Cebe P, et al. Water-insoluble silk filmswith silk I structure. Acta Biomater 2010;6:1380e7.

[53] Tretinnikov ON, Tamada Y. Influence of casting temperature on the near-surface structure and Wettability of cast silk fibroin films. Langmuir2001;17:7406e13.

[54] Demura M, Asakura T. Porous membrane of Bombyx mori silk fibroin: struc-ture characterization, physical properties and application to glucose oxidaseimmobilization. J Membr Sci 1991;59:39e52.

[55] Lawrence BD, Omenetto F, Chui K, Kaplan DL. Processing methods to controlsilk fibroin film biomaterial features. J Mater Sci 2008;43:6967e85.

[56] Dal Pra I, Freddi G, Minic J, Chiarini A, Armato U. De novo engineering ofreticular connective tissue in vivo by silk fibroin nonwoven materials.Biomaterials 2005;26:1987e99.

[57] Tamada Y. New process to form a silk fibroin porous 3-D structure. Bio-macromolecules 2005;6:3100e6.

[58] Kim HJ, Kim HS, Matsumoto A, Chin I-J, Jin H-J, Kaplan DL. ProcessingWindows for forming silk fibroin biomaterials into a 3D porous matrix. Aust JChem 2005;58:716e20.

[59] Jin H-J, Fridrikh SV, Rutledge GC, Kaplan DL. Electrospinning Bombyx mori silkwith poly(ethylene oxide). Biomacromolecules 2002;3:1233e9.

[60] Deng C, Li F, Hackett JM, Chaudhry SH, Toll FN, Toye B, et al. Collagen andglycopolymer based hydrogel for potential corneal application. Acta Biomater2010;6:187e94.

[61] Higa K, Takeshima N, Moro F, Kawakita T, Kawashima M, Demura M, et al.Porous silk fibroin film as a transparent carrier for cultivated corneal epithelialsheets. J Biomater Sci Polym Ed, in press.

[62] Majo F, Rochat A, Nicolas M, Jaoude GA, Barrandon Y. Oligopotent stem cellsare distributed throughout the mammalian ocular surface. Nature 2008;456:250e4.

[63] Gil ES, Mandal BB, Park SH, Marchant JK, Omenetto FG, Kaplan DL. Helicoidalmulti-lamellar features of RGD-functionalized silk biomaterials for cornealtissue engineering. Biomaterials 2010;31:8953e63.

[64] Yamada H, Igarashi Y, Takasu Y, Saito H, Tsubouchi K. Identification of fibroin-derived peptides enhancing the proliferation of cultured human skin fibro-blasts. Biomaterials 2004;25:467e72.

[65] Watanabe K, Nishida K, Yamato M, Umemoto T, Sumide T,Yamamoto K, et al. Human limbal epithelium contains side populationcells expressing the ATP-binding cassette transporter ABCG2. FEBS Lett2004;565:6e10.

[66] Barbaro V, Testa A, Di Iorio E, Mavilio F, Pellegrini G, De Luca M. C/EBPdeltaregulates cell cycle and self-renewal of human limbal stem cells. J Cell Biol2007;177:1037e49.

[67] Umemoto T, Yamato M, Nishida K, Yang J, Tano Y, Okano T. Limbal epithelialside-population cells have stem cell-like properties, including quiescent state.Stem Cells 2006;24:86e94.

[68] Di Iorio E, Barbaro V, Ruzza A, Ponzin D, Pellegrini G, De Luca M. Isoforms ofDeltaNp63 and the migration of ocular limbal cells in human corneal regen-eration. Proc Natl Acad Sci U S A 2005;102:9523e8.

[69] Toti P, Tosi GM, Traversi C, Schurfeld K, Cardone C, Caporossi A. CD-34 stromalexpression pattern in normal and altered human corneas. Ophthalmology2002;109:1167e71.

[70] Pei Y, Sherry DM, McDermott AM. Thy-1 distinguishes human cornealfibroblasts and myofibroblasts from keratocytes. Exp Eye Res 2004;79:705e12.

[71] Vieira A, Forwood K, Yu A, Kim C, Harocopos G, Schwab IR. In vivo silk fibroinbiocompatibility. Annual Meeting of the association for research in vision andOphthalmology. Fort Lauderdale; 2010. Abstract No. 4872.

D.G. Harkin et al. / Biomaterials 32 (2011) 2445e24582458

[72] Sinfield LJ, George K, Zainuddin Z, Chirila TV, Hutmacher DW, Schwab IR, et al.Fibroin-based materials support cultivation of limbal stromal cells. AnnualMeeting of the Association for Research in Vision and Ophthalmology. FortLauderdale; 2010. Abstract No. 6211.

[73] Schlotzer-Schrehardt U, Dietrich T, Saito K, Sorokin L, Sasaki T, Paulsson M,et al. Characterization of extracellular matrix components in the limbalepithelial stem cell compartment. Exp Eye Res 2007;85:845e60.

[74] Yeung AM, Schlotzer-Schrehardt U, Kulkarni B, Tint NL, Hopkinson A, Dua HS.Limbal epithelial crypt: a model for corneal epithelial maintenance and novellimbal regional variations. Arch Ophthalmol 2008;126:665e9.

[75] Harkin DG, George KA, Shadforth AMA, Cheng S, Kwan AS, Chirila TV.Development of an ultra-thin fibroin membrane for RPE cell transplantation.Annual Meeting of the Association for Research in Vision and Ophthalmology.Fort Lauderdale; 2010. Abstract No. 5248.