Laminin-511 and -521-based matrices for efficient ex vivo-expansion of human limbal epithelial progenitor cells

Optimization of culture conditions for human limbal epithelial stem/progenitor cells (LEPC) that incorporate the in vivo cell-matrix interactions are essential to enhance LEPC ex vivo-expansion and transplantation efficiency. Here, we investigate the efficacy of laminin (LN) isoforms preferentially expressed in the limbal niche as culture matrices for epithelial tissue engineering. Analyses of expression patterns of LN chains in the human limbal niche provided evidence for enrichment of LN-α2, -α3, -α5, -β1, -β2, -β3, -γ1, -γ2 and -γ3 chains in the limbal basement membrane, with LN-α5 representing a signature component specifically produced by epithelial progenitor cells. Recombinant human LN-521 and LN-511 significantly enhanced in vitro LEPC adhesion, migration and proliferation compared to other isoforms, and maintained phenotype stability. The bioactive LN-511-E8 fragment carrying only C-terminal domains showed similar efficacy as full-length LN-511. Functional blocking of α3β1 and α6β1 integrins suppressed adhesion of LEPC to LN-511/521-coated surfaces. Cultivation of LEPC on fibrin-based hydrogels incorporating LN-511-E8 resulted in firm integrin-mediated adhesion to the scaffold and well-stratified epithelial constructs, with maintenance of a progenitor cell phenotype in their (supra)basal layers. Thus, the incorporation of chemically defined LN-511-E8 into biosynthetic scaffolds represents a promising approach for xeno-free corneal epithelial tissue engineering for ocular surface reconstruction.

in vitro culturing of LEPC is that they readily differentiate, hampering their use for therapeutic applications 19,20 . These limitations underscore the need for developing novel standardized LEPC culture techniques that ensure preservation of the stem/progenitor cell phenotype and function during cultivation and after transplantation.
In vivo, LEPC reside in a highly specialized and complex microenvironment that is known as the limbal niche, where stem cell quiescence, proliferation and differentiation are maintained in balance 21 . Besides several types of supporting niche cells that include melanocytes and mesenchymal stromal cells, the limbal niche comprises a specific extracellular matrix (ECM) composition 22 . Indeed, the ECM has been recognized as a crucial part of many stem cell niches mediating biophysical, mechanical and biochemical signals that regulate stem cell survival, proliferation and differentiation [23][24][25][26] . Laminins (LNs) are the best-described ECM constituents present in basement membranes (BM) of adult stem cell niches, where they influence cell behaviors, such as cell adhesion, differentiation, and phenotype stability. To date, five alpha (α1-5), four beta (β1-4), and three gamma (γ1-3) chains have been identified that can combine to form up to 16 heterotrimeric isoforms in mammalian tissues, showing tissue-specific distribution patterns [27][28][29][30] . We have previously shown that LN chains α1, α2, α5, β1, β2, γ1 and γ3 are preferentially localized to the BM of the limbal epithelium compared to that of the corneal epithelium, where LN chains α3, β3 and γ2 predominate 31 . Our work has further suggested that LEPC are anchored to LN-α2 and -α5 in their niche by expression of LN receptors α3β1 and α6β4 integrins 32 . The heterogeneity of LN isoform expression in BMs of ocular surface epithelia, which has been also reported by other groups [33][34][35] , suggests a functional role for LN-α2 and -α5 containing isoforms in the limbal stem cell niche.
Stem cell-based tissue engineering aims to mimic the native stem cell niche and to present the appropriate microenvironmental cues, including ECM components, in a controlled and reproducible fashion in order to maintain stem cell function within the graft 36,37 . We have previously reported that the LN-332 isoform can serve as a suitable substrate to induce transdifferentiation of hair follicle stem cells into corneal epithelial-like cells in the presence of limbal fibroblast conditioned medium 38 . It has also been suggested that proteolytic fragments of LN-332 γ2 chain potentiate the outgrowth of limbal epithelial cells on intact amniotic membrane 39 , and that vitronectin adsorbed to therapeutic contact lenses support LEPC expansion 40 . Otherwise, little attention has been paid to the role of the ECM in LEPC ex vivo-expansion, and the effect of limbus-specific LNs on LEPC phenotype and function in vitro has not been investigated. We hypothesize that the LN isoforms that are specifically expressed in the limbal stem cell niche may be used as exogenous cues to promote ex vivo-expansion and maintenance of LEPC. In this study, we systematically analyze expression patterns of LN chains in LEPC clusters isolated by laser capture microdissection (LCM), in tissue sections and in cultured LEPC by quantitative RT-PCR (qPCR) and confocal microscopy, and test the effects of niche-specific recombinant LN isoforms on LEPC adhesion, migration, proliferation and differentiation in vitro. We also examine the efficacy of a recombinant human LN-511-E8 fragment 41,42 for epithelial tissue engineering using fibrin-based hydrogels as carrier. Our findings demonstrate that LN-α5 constitutes a signature BM component of the limbal niche and that LN-511 or -521 promote ex vivo-expansion and maintenance of LEPC. The C-terminal fragment of LN-511 (LN-511-E8 fragment) shows comparable efficacy as full-length LN-511 and supports the formation of well stratified epithelial constructs with maintenance of a progenitor cell phenotype. Thus, LN-511-E8 represents a chemically defined, xeno-free substrate for improved corneal epithelial tissue engineering and future clinical application.
Concurrent with the qPCR data, the immunohistochemical findings suggest that the human limbal niche is specifically enriched for LN-α2 and -α5 chains.These were found to co-localize with the progenitor cell markers cytokeratin (CK)15, N-cadherin, and p63α as well as with integrins α3, α6 and ß1 expressed by LEPC clusters in the basal epithelium at the limbus (data for LN-α5 is shown in Fig. 1C).
Together, these expression data obtained in vivo and in vitro suggest that LN-α5 constitutes a signature BM component of the limbal niche, which is endogenously produced by LEPC and becomes strongly enriched in the BM of the limbal niche.
LN-511/521 has been reported to bind to integrin α3β1 and α6β1, while LN-332 binds to integrin α6β4 on the cell membrane 44 . Because integrins α3β1and α6β4 have been previously reported to mediate anchorage of LEPC to their niche 32 , we analyzed the expression of these integrin subunits on the cell surface of cultured LEPC (n = 3) and evaluated the effect of integrin blocking on LEPC adhesion to LN-521, -511, -511-E8 and -332 (n = 4). As demonstrated by flow cytometry, almost 100% of LEPC expressed α3, α6, β1 and β4 integrins on their surface ( To assess a potential effect of LN isoforms on the differentiation state of LEPC, mRNA expression levels of KRT3, a cornea-specific differentiation marker, and KRT15, an established progenitor cell marker, were analyzed by qRT-PCR in LEPC passages 1 to 3 (n = 5). Although there was a trend towards suppression of KRT3 expression in LEPC cultivated on LN-521, -511, -511-E8, and -421 compared to control (Fig. 4D), differences were not statistically significant. Expression levels of KRT15 were also not significantly different from controls, except for LN-332, which supported maintenance of high KRT15 expression levels for up to 3 passages (p = 0.02).
These observations suggest that LN-332 supports an undifferentiated phenotype but inhibits proliferation, whereas LN-521, -511, and 511-E8 support cell proliferation without affecting differentiation, rendering LN-α5 containing isoforms as suitable candidates for tissue engineering purposes.
Tissue engineering of corneal epithelial constructs. Since recombinant LN-E8 fragments have been shown to promote efficient and scalable culture of human stem cells under defined xeno-free conditions 41, 45 , we tested the applicability of LN-511-E8 for tissue engineering of corneal epithelial constructs (n = 5). To generate multilayered epithelial cell sheets on 3D-scaffolds suitable for clinical application, LEPC were seeded and cultivated for 12 to 14 days on fibrin gels prepared with or without incorporation of LN-511-E8 (10 µg/ml). LEPC expansion was markedly increased on LN-containing gels compared with untreated gels resulting in a confluent monolayer on LN-511-E8 by 5 days after seeding (Fig. 5A). Light microscopic analyses of tissue constructs showed a stratified epithelial cell sheet consisting of a cuboidal basal layer and 5 to 6 layers of flattened suprabasal cells on LN-511-E8 containing gels, but only 3 to 4 cell layers on LN-free gels (Fig. 5B). Transmission electron microscopy confirmed formation of well-organized multi-layered epithelial cell sheets consisting of a basal layer of cuboid cells covered by several layers of flattened elongated cells (Fig. 5C,i). Suprabasal cells showed typical ultrastructural signs of epithelial differentiation, such as desmosomes, cytoplasmic filaments, and apical microvilli. Basal cells appeared to be firmly attached to the LN-511-E8 containing fibrin gels by formation of hemidesmosomes and newly produced BM, whereas they appeared to only loosely adhere to LN-free fibrin gels where no BM deposition was evident (Fig. 5C,ii).
Immunofluorescence analysis of epithelial constructs showed expression of corneal epithelial differentiation marker CK3 in superficial layers and expression of progenitor cell markers CK15 and p63α in (supra) basal layers (Fig. 5D). The cell-matrix receptors α3, α6 and ß1 integrin could be mainly immunolocalized to the basal aspects of epithelial sheets, which also provided evidence of endogenous expression of LN-α5 and -α3. Protein expression was found to be generally more pronounced in epithelial constructs established on  LN-511-E8 containing gels than on pure fibrin gels. These data suggest that LN-511-E8-coated fibrin gels promote ex vivo-expansion of LEPC and the generation of a stratified, firmly adherent, corneal epithelial-like cell sheet containing both differentiated and undifferentiated cells, which appears amenable for transplantation onto the corneal surface.

Discussion
Using qPCR on microdissected LEPC clusters and confocal microscopy on tissue sections, we show that the BM of the limbal niche is enriched for LN chains α2, α3, α5, β1, β2, β3, γ1, and γ2, co-localizing with integrins α3, α6 and ß1 on the LEPC surface. Of note, mRNA expression of LN-γ3 could not be detected in LEPC, although LN-γ3 protein was clearly identified in the limbal zone by immunohistochemistry in this and previous studies 31,46,47 . In order to determine the role of LEPC in LN deposition within the limbal niche, we performed differential mRNA expression analyses in cultivated LEPC and supporting LMSC populations, both derived from LEPC clusters within the niche 32 . Our results clearly show the predominant expression of LN-α3 and -α5 by LEPC together with LN-β1, -β3, and -γ2 chains, suggesting the probable secretion of LN-332 and LN-511 by epithelial cells. By contrast, the LMSC preferentially expressed LN-α2 and -α4 chains together with -β2, -γ1 and -γ3 chains, and thus probably secrete and deposit LN-211/221 and LN-411/421/423 into the limbal BM. This differential expression pattern of laminin chains by epithelial and stromal fibroblast-like cells is consistent with findings in the epidermal BM, containing both epidermally derived LN-α3 and -α5 and mesenchymally derived LN-α2 and -α4, and suggests that, as in the skin, interaction between epithelial and mesenchymal cells is required to establish the limbal niche 48,49 . The differential expression of LN-α5, but not LN-α3, by LEPC compared to BCEC further suggests that the LN-α5 chain may be an endogenous signature component of progenitor cells and may also present an appropriate exogenous cue for ex vivo-expansion of LEPC.
As a major component of BMs, LN isoforms are also enriched in other adult stem cell niches, where they regulate stem/progenitor cell function through interaction with cell surface receptors transducing molecular signals to the intracellular compartment [23][24][25][26]50 . Consistent with the data presented here, interactions are predominantly mediated by one of the four LN-binding integrins, α3β1, α6β1, α6β4 43,44 , and in the case of myogenic cells also α7β1, which activate specific signaling networks to modulate diverse cellular functions, such as adhesion, migration, proliferation, differentiation, stability of phenotype, and resistance towards apoptosis. The epidermal stem cell niche, for instance, contains LN-332 and LN-511, which interact with α6β4 and α3β1 integrins, respectively, and are hypothesized to maintain stem cell homeostasis through a precise ratio of LN-332 to LN-511 [51][52][53] . Deposition of LN-332 and LN-511 gradients within the epidermal BM is regulated by integrin-linked kinase (ILK), which plays a crucial role in linking the ECM to the actin cytoskeleton of skin and hair follicle stem cells 52 . Detachment from LN-511/521-containing BMs has been considered a key factor in promoting epithelial differentiation. LN-332 and LN-511/521 have been also reported to be synthesized and secreted by human corneal epithelial cells in vivo, and to play a crucial role in the adhesion and migration of corneal epithelial in vitro 54,55 . Together with the data presented here, there is convincing evidence that both LN-332 and LN-511/521 through interaction with integrins α6β4 and α3β1/α6β1, respectively, form part of the limbal niche. However, whether the ratio between the two LN isoforms and their gradient of expression in the limbal BM determine LEPC function, requires further investigation. It also remains to be determined, whether ILK, which has been reported to transfer β1 integrin-mediated signals from the ECM to the Wnt/ß-catenin pathway in LEPC cultivated on amniotic membrane 56 , is a key regulator of the limbal niche ECM microenvironment.
Here, we further show that the different LN isoforms have different functions on LEPC in vitro. Whereas LN-511/521 supported integrin α3β1/α6β1-mediated LEPC adhesion, migration and proliferation without major effects on cellular differentiation, LN-332 promoted LEPC adhesion and migration, but suppressed proliferation, and maintained an undifferentiated phenotype, as reflected by high expression of the progenitor marker KRT15. All other LN isoforms tested, i.e. LN-421, -411, -211 and -111, showed no superior effects on adhesion and proliferation of LEPC compared to tissue culture control wells. These observations are different to the skin, where LN-α3 has been shown to promote and LN-α5 to inhibit keratinocyte proliferation 48 . However, in accordance with our observations, LN-511 and -521, which are also expressed in Descemet's membrane and corneal endothelium, strongly supported in vitro adhesion and proliferation of human corneal endothelial cells 42 . Thus, the particular characteristics of LN-511/521, i.e. support of adhesion, proliferation and phenotype stability of corneal cells, render these isoforms best qualified for tissue engineering of LEPC-derived epithelial cell sheets without abandoning the progenitor phenotype. Indeed, recombinant LN-511 and -521 have been frequently used as substrates for efficient stem/progenitor cell expansion in vitro and have been shown to protect against apoptosis, to stimulate proliferation, and to maintain pluripotency and long-term self-renewal of both pluripotent and adult stem cells including human keratinocyte stem cells [57][58][59][60] .
Short recombinant E8 fragments derived from the long arm of LN have been reported as a viable and safe alternative to intact LN proteins in tissue engineering applications, because they do not involve the risk of contamination with xenogenic pathogens and immunogens 41 .These fragments have been shown to represent the minimum fraction conferring β1 integrin-binding activity and supporting efficient adhesion and proliferation of pluripotent stem cells in defined xeno-free media 45,61,62 . E8 fragments of LN-511 have also been used as a substrate for cultivation of human corneal endothelial cells with a similar efficacy to that obtained with full-length LN 42 . This could be confirmed by the present study, showing that LN-511-E8 fragments had similar potency in stimulating LEPC adhesion, migration, proliferation, and phenotype stability as intact LN-511. We further show that recombinant LN-511-E8 fragments incorporated into and on top of fibrin-based hydrogels improve ex vivo-expansion of LEPC and generation of a stratified epithelial cell sheet. The epithelial constructs generated on LN-511-E8 pre-functionalized scaffolds proved to be superior to those on bare scaffolds, displaying a well-organized stratified structure and firm attachment by α3β1 and α6β1 integrin-mediated anchorage to LN-α5 as well as preservation of a progenitor cell phenotype in (supra)basal layers and a differentiated phenotype in superficial layers.
In conclusion, in vitro presentation of appropriate ECM cues influencing stem cell function in vivo may represent a useful strategy to improve LEPC expansion and to maintain stem/progenitor cell function within the graft. The incorporation of chemically defined, bioactive, recombinant human E8 fragments of the limbal niche constituent LN-511 into clinically approved biosynthetic scaffolds like fibrin gels represents a promising avenue for safe, xeno-free corneal epithelial tissue engineering for ocular surface reconstruction. This defined culture technique may provide a significant improvement over current protocols for ex vivo-expansion of LEPC on undefined matrices, such as amniotic membrane or murine feeder cells, suffering from an intrinsic biologic variability, but this proposition has to be evaluated in preclinical in vivo studies. Future studies are also required to elucidate the nature of signaling pathways that are activated by LN-integrin interactions to determine LEPC function in vivo and in vitro.

Methods
Human tissues and study approval. Human donor corneas (n = 5 for LCM, n = 10 for immunohistochemistry) not suitable for transplantation with appropriate research consent were procured by the Erlangen Cornea Bank. In addition, organ-cultured corneoscleral tissue (n = 42) with appropriate research consent was provided by the Erlangen Cornea Bank after corneal endothelial transplantation. Informed consent to corneal tissue donation was obtained from the donors or their relatives. Experiments using human tissue samples were approved by the Institutional Review Board of the Medical Faculty of the University of Erlangen-Nürnberg (No. 4218-CH) and adhered to the tenets of the Declaration of Helsinki.
Laser capture microdissection (LCM) and amplification of RNA. LCM and amplification of RNA was performed as previously described 32 . Briefly, corneal specimens destined for LCM were obtained from five donors (mean age, 69.6 ± 10.4 years) within 15 hours after death. After labeling of the superior, inferior, nasal, and temporal quadrants of donor globes, tissue sectors were embedded in optimal cutting temperature (OCT) compound (Tissue-Tek, Sakura Finetek Europe) and snap frozen in liquid nitrogen. Serial cryosections of 12 μm thickness were obtained under RNAse-free conditions from the superior or inferior quadrants, placed onto UV-irradiated (3000 mJ/cm 2 ) PEN (polyethylene naphtalate) Membrane Slides (Carl Zeiss Microscopy, Göttingen, Germany), and stained with 0.01% cresyl violet. The PALM MicroBeam IV system (Carl Zeiss Microscopy) was used to isolate clusters of basal limbal epithelial cells (LEPC) and basal epithelial cells from central cornea (BCEC). From each donor eye, 100 cryosections were dissected (50 from the superior and 50 from the inferior quadrants, respectively) yielding a total of 500 cryosections used for sample collection.
RNA isolation from these specimens was achieved using the RNeasy Micro Kit (Qiagen, Hilden, Germany) including an on-column DNase digestion step according to the manufacturer's instructions. Quality control was performed on a 2100 Agilent Bioanalyzer using the RNA 6000 Pico Kit (Agilent Technologies, Santa Clara, CA). Samples with an RNA concentration of 650-2,000 pg/µl and a RIN (RNA integrity number) of ≥7.0 were used for further analysis. Following RNA-amplification using the MessageAmp II aRNA Amplification Kit (Life Technologies GmbH, Darmstadt, Germany) according to the manufacturer's protocol, aRNA (amplified RNA) concentration was measured on a Nanodrop ND1000 spectrophotometer (Thermo Scientific, Wilmington, DE) and quality control was again performed using Agilent technology.

Real time RT-PCR.
First-strand cDNA synthesis was performed using 5 µg of aRNA from tissue samples or RNA from cultured cells and Superscript II reverse transcriptase (Invitrogen, Karlsruhe, Germany) as previously described 32 . PCR reactions were run in triplicate in 1× TaqMan Probe Mastermix (Roche Diagnostics, Mannheim, Germany) or 1× SsoFast EvaGreen Supermix (Bio-Rad Laboratories, München, Germany), respectively, according to the manufacturers' recommendations. Primer sequences (Eurofins, Anzing, Germany) are given in Table 2. For normalization of gene expression levels, ratios relative to the housekeeping gene GAPDH were calculated by the comparative C T method (ΔΔC T ). Genes were considered as differentially expressed when their expression levels exceeded a two-fold difference in all specimens analyzed.
Immunohisto-and immunocytochemistry. Corneoscleral tissue samples obtained from 10 normal human donor eyes (mean age, 72.3 ± 11.6 years) and 3D-cultures were embedded in optimal cutting temperature (OCT) compound and frozen in isopentane-cooled liquid nitrogen. Cryosections of 4 μm thickness were cut from the superior or inferior quadrants, fixed in cold acetone for 10 minutes, blocked with 10% normal goat serum, and incubated in primary antibodies (Table 1) diluted in PBS overnight at 4 °C. Antibody binding was detected by Alexa 488-conjugated secondary antibodies (Molecular Probes, Eugene, OR) and nuclear counterstaining was performed with DAPI (Sigma-Aldrich, St. Louis, Missouri). Immunolabelled cryosections and cultured LEPC were examined with a fluorescence microscope (Olympus BX51; Olympus, Hamburg, Germany) or a laser scanning confocal microscope (LSM 780; Carl Zeiss Microscopy). In negative control experiments, the primary antibodies were replaced by PBS or equimolar concentrations of an irrelevant isotypic primary antibody.
Cell culture. LEPC clusters were isolated from 42 corneoscleral buttons (mean age, 69.2 ± 7.1 years) as previously described 32 . Briefly, organ-cultured corneoscleral tissue with appropriate research consent was provided by the Erlangen Cornea Bank after corneal endothelial transplantation. After rinsing in Hanks' balanced salt solution, the tissues were cut into 12 one-clock-hour sectors, from which limbal segments were obtained by incisions made at 1 mm before and beyond the anatomical limbus. Each limbal segment was enzymatically digested with 2 mg/mL collagenase A at 37 °C for 18 hours to generate epithelial-mesenchymal stem/progenitor cell clusters. Cell clusters were isolated from single cells by using reversible cell strainers with a pore size of 37 µm (Stem Cell Technologies, Köln, Germany). Isolated cell clusters were further dissociated into single cells by digestion with 0.25% trypsin and 0.02% EDTA (Pan Biotech, Aidenbach, Germany) at 37 °C for 10-15 min. Single cell suspensions were seeded into T75 flasks (Corning, Tewksbury, MA) in Keratinocyte serum free medium (KSFM) supplemented with bovine pituitary extract, epidermal growth factor (Life Technologies) and 1× penicillin-streptomycin-amphotericin B mix (Pan Biotech) to enrich epithelial cell population (LEPC) and the flasks were incubated at 37 °C under 5% CO 2 and 95% humidity. To enrich the associated mesenchymal stromal cells (LMSC), single cell suspensions were seeded into T75 flasks in Mesencult media (Stem Cell Technologies) and incubated at 37 °C under 5% CO 2 and 95% humidity. The media was changed every second day. Cultivated cells were used for real time PCR (n = 10), flow cytometry (n = 3), functional assays (adhesion: n = 4, integrin blocking: n = 4, migration: n = 3, proliferation: n = 13) and tissue engineering (n = 5).   Table 2. Primers used in qRT-PCR primer assays. Primers were used for Probe based (Universal Probe Library) or SYBR Green based qRT-PCR assays with an annealing temperature of 60 °C. Japan) as per manufacturers' recommendations. Coating solutions were prepared in 1x DPBS (Dulbecco's phosphate-buffered saline) with calcium and magnesium and incubated in culture wells either for two hours at 37 °C (fast coating) or overnight at 4 °C (slow coating) as per manufacturer's instructions. Cells (P1-P2) were seeded at a density of 50,000 cells/cm 2 and left to adhere for 30 and 60 min at 37 °C. Standard tissue culture treated plates were used as control. After incubation, plates were washed with DPBS using a Static Cell Adhesion Wash Chamber (Glycotec, Maryland, USA) to remove non-adherent cells, and the adherent cells were fixed with 4% paraformaldehyde/PBS for 15 min and stained with 0.1% crystal violet for 20 min. After three washes with water, stained cells were extracted with 1% sodium dodecyl sulfate and quantified by measuring optical density (OD) at 570 nm using a spectrophotometer (Multiskan Spectrum; Thermo Scientific, Waltham, USA). All experiments (n = 4) were performed in quadruplicates. The fold change values were calculated as OD of the LN/OD of control. The effect of an interaction between cellular integrins and extracellular LNs on cell adhesion was evaluated by seeding LEPC (50,000 cells/cm 2 ) in 96-well plates coated with recombinant LN-332, -511, -521, and LN-511-E8 in the presence or absence of integrin-neutralizing antibodies, i.e., anti-α3 integrin (20 µg/ml) (Merck Millipore, Darmstadt, Germany), anti-α6 integrin (20 µg/ml) (Merck Millipore) and anti-β1 integrin (2.5 µg/ml) (R&D Systems Inc., Minneapolis, MN, USA). 60 minutes after seeding, the numbers of adherent cells were determined as described above. LEPC used for integrin blocking assays were isolated from 4 organ-cultured corneoscleral buttons (mean age, 67.7 ± 6.1 years) and all experiments were performed in quadruplicates.

Cell migration assay.
To exactly measure the change in the cell-covered area over time, 2 well-culture inserts with a defined cell-free gap were used (ibidi GmbH, Planegg, Germany). Briefly, the wells were coated with LN isoforms as described above and seeded with 70 µl of a LEPC suspension containing 5 × 10 5 cells/ml. LEPC were isolated from 3 organ-cultured corneoscleral buttons (mean age, 71.3 ± 6.1 years). After formation of a cellular monolayer (24 hours), the silicone inserts were removed and the culture medium was supplemented with 2.5 µg/ ml of soluble LNs. Images of each well were acquired immediately following insert removal (0 hours) and after 3 and 6 hours by using an inverted microscope (CKX41, Olympus). Image analysis software Cell F (Olympus) was used to measure areas that were free of migrating cells. All experiments (n = 3) were performed in triplicates.
Cell proliferation assays. LEPC used for cell proliferation assays were isolated from a total of 13 corneoscleral buttons after corneal endothelial transplantation (mean age, 70.0 ± 8.0 years).The effect of LNs on LEPC proliferation was quantified using the Cell Proliferation ELISA BrdU Colorimetric Assay Kit (Roche Diagnostics). Cells (P1-P2) were seeded into 96-well plates, coated with LN isoforms as described above, at a density of 5000 cells/well, cultured for 48 and 72 hours, and labeled with BrdU according to the manufacturer's instructions. Absorbance was measured at 450 nm using a spectrophotometer (Multiskan Spectrum), and fold change values were calculated as described above. Experiments (n = 5) were performed in quadruplicates.
For immunocytochemical analysis of cell proliferation, LEPC (n = 3) were seeded at a density of 10,000 cells/ well into 4 well-glass chamber slides (LabTek; Nunc, Wiesbaden, Germany), cultured for 72 hours, and stained with antibodies against Ki-67 (Abcam; Cambridge, UK). Ki-67 positive cells were counted in the 4 wells using Cell F image analysis software (Olympus). Ki-67 expression was also analyzed by real time PCR as described above (n = 5).
Tissue engineering of epithelial constructs. Scaffolds for tissue engineering and 3D-cell culture were prepared from fibrin as previously described 38 . Briefly, the gels were prepared by dissolving fibrinogen and thrombin stock solutions (Tisseel; Baxter Deutschland GmbH, Unterschleißheim, Germany) in 1.1% NaCl and 1 mM CaCl 2 to a final concentration of 10 mg/ml fibrinogen and 3 IU/ml thrombin. Recombinant LN-511-E8 (10 µg/ml) was incorporated into the gels, which were placed into 24 well-culture inserts and allowed to polymerize at 37 °C. After washing with PBS, gels were additionally coated with LN-511-E8 (5 µg/ml) overnight. LN-511-E8 free gels served as controls. LEPC (P1) were isolated from 5 organ-cultured corneoscleral buttons (mean age, 65.0 ± 7.5 years) and seeded onto coated and uncoated control gels at a density of 1 × 10 5 cells/cm 2 and cultivated in KSFM media under low calcium concentrations (0.09 mM Ca 2+ ) for 5 to 6 days. Culture conditions were switched to DMEM/Ham's F12 (Hyclone; GE Health Care Life Sciences, Freiburg, Germany) supplemented with Human Corneal Growth Supplement (Life Technologies), 10% FCS (GE Health Care Life Sciences), and high calcium concentrations (1.2 mM Ca 2+ ) in order to promote cell differentiation. After another 6 to 8 days of cultivation, gels were fixed for immunohistochemistry as described above and for light and electron microscopy.
For light and electron microscopy, gels were fixed in 2.5% glutaraldehyde in 0.1 M phosphate buffer, dehydrated, and embedded in paraffin or epoxy resin, respectively, according to standard protocols. Paraffin sections were stained with hematoxylin and eosin, and ultrathin sections were stained with uranyl acetate-lead citrate and examined with an electron microscope (EM 906E; Carl Zeiss Microscopy).
Statistical analysis. Statistical analyses were performed using the GraphPad InStat statistical package for Windows (Version 5.04; Graphpad Software Inc., La Jolla, CA). Data are expressed as mean ± standard error of the mean from individual experiments. The Mann-Whitney U test was performed to assess statistical significance. A p value of <0.05 was considered statistically significant. Data availability. Any additional data beyond those included in the main text that support the findings of this study are also available from the corresponding author upon request.