Vitronectin: A Matrix Support Factor for Human Limbal Epithelial Progenitor Cells

Abstract

Purpose.: The authors recently developed a therapeutic technique for patients with limbal stem cell deficiency by harvesting ocular surface stem cells (SCs), expanding them on therapeutic contact lenses (CLs), and applying them to diseased corneas. The present study determined the proteins that bind to CLs and whether such factors, along with transplanted cells, are critical determinants for corneal rehabilitation using this method.

Methods.: Therapeutic CLs were exposed to human serum, and adherent proteins were analyzed by proteomics. The distribution of vitronectin (VN) on the ocular surface was determined with specific antibodies. Cadaveric human corneas were chemically wounded, and cell transfer by CLs was assessed in organ culture.

Results.: VN was identified as a serum factor that binds and desorbs from CLs. VN localized to the limbal and basement membranes (BM) of other SC-harboring organs. Clonogenic assays demonstrated higher colony-forming efficiency on VN compared with uncoated surfaces. Cell transfer from CLs was achieved through in vitro models and was abrogated by RGD peptides and inhibitory antibodies to VN and its receptor.

Conclusions.: Identification of VN within the limbal BM, its effect on limbal SC activity, and the discovery of this factor on serum-exposed CL polymers implies a role in supporting progenitor cells and facilitating corneal regeneration.

The human ocular surface spans the cornea to the conjunctiva and plays an integral role in vision. The cornea, the most anterior structure of the eye, contributes significantly to light focusing through its intrinsic convexity and transparency. The corneal epithelium requires continuous maintenance and renewal, ¹ a function served by unipotent stem cells (SCs) located within the limbus, a narrow transition zone circumscribing the peripheral cornea. ^{2 –4} On activation, SCs divide asymmetrically with one daughter cell retained within the SC niche while the other detaches from its basement membrane (BM) and migrates centripetally and superficially to replenish aged, damaged, or diseased corneal epithelial cells. ⁵  

Excessive damage to the limbal SC (LSC) niche, through chemical or thermal burns, multiple surgical procedures, genetic anomalies, or manifestations that arise through mucous membrane disorders, can induce a condition known as limbal stem cell deficiency (LSCD). ^{6 –9} This is characterized by persistent, recurrent, and painful corneal epithelial defects. Conjunctivalization is a process commonly observed in LSCD and involves the invasion of an inflamed and vascularized pannus of conjunctival tissue ¹⁰ over the cornea that may ultimately result in blindness. ⁸  

Current treatment strategies for LSCD use ex vivo expanded LSCs from a tissue biopsy, ^{11 –14} a technique first developed by Rheinwald and Green ¹⁵ for managing severe cutaneous burns. Many substrates for the expansion of LSCs have been identified, including tissue culture plastic, ¹¹ human amniotic membrane (HAM), ¹² and fibrin. ¹³ Although techniques using these substrates have proven efficacious, ¹⁶ most incorporate foreign biomaterials such as murine 3T3 fibroblast feeder cells, HAM, and fetal bovine serum (FBS), thereby increasing the risk for xenogeneic disease transmission and the potential for graft rejection if foreign antigens persist. ^17,18  

Our recently developed technique was the first to use a (contact lens) CL scaffold and autologous serum-supplemented media to deliver syngeneic SCs to patients with LSCD. ¹⁴ Silicone-hydrogel CLs provide a suitable substrate for corneal epithelial cells ¹⁹ and have been used as cell delivery vehicles in animal models of LSCD. ^20,21 However, the factors that support cells in culture and on the ocular surface before and after transplantation are undefined. 

Limbal niche factors are thought to maintain SC characteristics. ²² Espana et al. ²³ demonstrated the impact of the stromal niche on the differentiation state of corneal epithelial cells, observing that central corneal epithelial cells lost expression of cytokeratin 3 (a differentiation marker) when placed on a limbal stroma while limbal epithelial cells gained expression of this protein when placed on a corneal stroma. Similarly, other studies have shown evidence that fibronectin (FN) and collagen type IV, factors found within the limbal stroma, ²² promote retention of SC properties ²⁴ and dramatically increase the lifespan of human limbal epithelial cells. ²⁵ Interestingly, vitronectin (VN) is also concentrated in the limbal stroma, ²² supporting a role for this factor in SC maintenance. VN has been found to support corneal epithelial cell migration and proliferation, ²⁶ and FN has been found to be a prominent component of the provisional matrix deposited after corneal wounding, ²⁷ with increased levels correlating with epithelial migration. ²⁸ In addition, VN and FN are proteins abundant in blood and are thought to contribute to the successful outcomes associated with the use of serum eyedrops in the treatment of persistent corneal epithelial defects. ^29,30  

The aim of this study was to define the factors that preserve SC activity on a CL scaffold and on the human ocular surface, speculating the involvement of a complex interplay between extracellular matrix (ECM) proteins and their integrin receptors, which facilitate cell attachment, detachment, spreading, migration, and wound healing. ³¹ Herein we report the binding and release of VN from CL polymers, the localization of this factor on the limbal BM, and its ability to support SC activity in vitro. 

Materials and Methods

Human Tissue and Serum Specimens

Human biological material was obtained in accordance with the tenets of the Declaration of Helsinki and was approved by the University of New South Wales Human Research Ethics Committee (approval numbers 06219, 06290, 10166). A total of 26 human cadaveric eyes (14 men, 5 women; age range, 36–96 years) not suitable for transplantation, with <72 hours postmortem delay (Lions NSW Eye Bank, Sydney, Australia), were used in this study. In addition, pterygium, normal skin, and small intestine were included from our tissue bank. Serum was obtained from seven patients (4 men, 3 women; age range, 29–71 years) with LSCD who consented to participate in our pilot clinical trial. ¹⁴  

Mass Spectrometry

Night & Day (Lotrafilcon A; CIBA Vision, Duluth, MN) and PureVision (Balafilcon A; Bausch & Lomb, Rochester, NY) therapeutic CLs were submerged in 600 μL Eagle's minimal essential medium (EMEM; Thermo Electron, Bremen, Germany) containing 10% human serum in 24-well culture plates (CellStar; Greiner Bio-One, Frickenhausen, Germany) and incubated at 37°C in a 5% CO₂ humidified incubator for 11 days, replacing 100 μL solution on days 4 and 7. These conditions simulated our cultivation procedure for autologous cells on a CL for clinical application ¹⁴ ; however, in these experiments, cells were omitted. CLs were rinsed in sterile phosphate-buffered saline (PBS), and adsorbed proteins were trypsin-digested (Promega, Madison, WI) and analyzed using a mass spectrometer (LTQ Velos Orbitrap; Thermo Electron, Bremen, Germany). Results were entered into the Mascot database v2.2 (Matrix Science, Boston, MA), and a protein profile was generated. 

Gel Electrophoresis and Western Blot Analysis

CLs were placed in nonreducing buffer (62.5 mM Tris-HCl, pH 6.8, 2% [wt/vol] SDS, 10% glycerol in 0.01% [wt/vol] bromophenol blue) to strip proteins that were bound to the lens polymer, and samples (20 μL) analyzed by SDS-PAGE with molecular weight markers (Precision Plus Protein Dual Color Standards; Bio-Rad, Hercules, CA) placed in adjacent lanes. Gels were stained with Coomassie blue or silver nitrate and then imaged (Gel Doc XR System and Quantity One software; Bio-Rad). 

Some CLs were incubated in either 10% human serum/EMEM (as described) or 5 μg/mL recombinant human VN (rhVN; R&D Systems, Minneapolis, MN), submerged in PBS for 7 days, and placed in 1.5-mL microcentrifuge tubes (Eppendorf, Hamburg, Germany) containing 100 μL nonreducing sample buffer or were reduced in the presence of 25 mM dithiothreitol in 2× Laemmli sample buffer (Bio-Rad). Tubes were gently agitated for 3 hours at 4°C, and the solution was collected and stored at −20°C until use in biochemical assays. Proteins were transferred to polyvinylidene fluoride membranes (Perkin Elmer Life Sciences, Boston, MA) that were washed in Tris-buffered saline containing 0.01% Tween-20 (MP Biochemicals, Solon, OH) (TBST, pH 7.6) before they were blocked in 5% (wt/vol) skim milk powder (Fonterra Brands, Mount Waverly, Australia) for 1 hour. Membranes were exposed to a mouse monoclonal anti-VN antibody (10 μg/mL; clone HV-2, gift from Meg Evans and Penny Bean [Australia's Commonwealth Scientific and Industrial Research Organisation, Sydney, Australia]). A secondary biotinylated polyclonal goat anti-mouse immunoglobulin antibody (Dako, Glostrup, Denmark) and horseradish peroxidase (HRP)-conjugated streptavidin (Dako) were sequentially added with ample washing between reagents. Membranes were placed in a chemiluminescence reagent (Western Lightning Plus-ECL; PerkinElmer, Waltham, MA) to reveal immunoreactive bands that were detected with x-ray film (GE Healthcare Bio-Sciences, Uppsala, Sweden) and were analyzed (LAS-3000 Imager; Fujifilm, Kanagawa, Japan). 

Immunofluorescence and Immunohistochemistry

To localize VN, CLs were cut into wedges with a scalpel blade, and immunofluorescence was conducted after they were incubated in human serum for 11 days. CLs were washed in PBS, fixed in 10% neutral buffered formalin (Fronine Laboratory Supplies, Taren Point, Australia), rehydrated, transferred to wells of a 24-well plate (CellStar; Greiner Bio-One), and blocked in 20% goat serum (Sigma, St. Louis, MO) made in 2% bovine serum albumin (BSA; Sigma) containing TBS (2% BSA/TBS). An anti-human VN antibody (clone HV-2) was diluted (2 μg/mL) in 2% BSA/TBS and incubated with each CL segment overnight at 4°C. Lens segments were extensively washed in TBS, and an Alexa Fluor 488-conjugated goat anti-mouse secondary antibody (Invitrogen, Eugene, OR) was added to develop VN reactivity. CL segments were mounted in anti-fade mountant (Vectashield; Vector Laboratories, Burlingame, CA), coverslipped, and photographed using a camera (DP70; Olympus, Center Valley, PA) mounted on a microscope (BX51; Olympus). Images were processed with graphics editing software (Photoshop, version 8; Adobe, San Jose, CA). 

To determine the spatial distribution of VN in vivo, formalin-fixed, paraffin-embedded sections of normal human cadaveric ocular tissue (n = 13), pterygium (used as a positive control), small intestine, and skin (organs known to harbor epithelial SCs) were assessed using a procedure similar to that described. Briefly, corneal specimens were bisected vertically and horizontally and embedded on the cut surface. Tissue was dewaxed, equilibrated in TBS, and blocked for endogenous peroxidase activity with 3% H₂O₂/methanol for 5 minutes. Sections were washed in TBS, blocked (20% goat serum in 2% BSA/TBS), and processed as described for immunofluorescence. Immunoreactivity for VN using the HV-2 clone was compared with a commercial antibody (clone VIT-2; Abcam, Cambridge, UK); both antibodies displayed a similar pattern of immunoreactivity on pterygium specimens (data not shown). Negative control reactions included omitting the primary antibody or adding an IgG₁ (R&D Systems, Minneapolis, MN) or IgG_2b isotype (gift from Meg Evans and Penny Bean, CSIRO, Sydney, Australia). Secondary (biotin-conjugated, goat anti-mouse antibody) and tertiary reagents (HRP-linked streptavidin) were added to each section, as previously described. ³² The substrate 3-amino-9-ethyl-carbazole (Sigma) was added, slides were lightly counterstained in hematoxylin (Dako), staining was preserved in aqueous mounting medium (Crystal Mount; ProSciTech, Kirwan, Australia), and results were imaged. A semiquantitative grading scale was used ³² to estimate VN levels in different regions of the ocular surface (no staining, 1; mild, 2; moderate, 3; intense, 4). Each specimen was reviewed and scored by three independent masked observers. 

Primary and Cell Line Cultures of Human Corneal Epithelial Cells

Limbal biopsy specimens were excised from donor corneal rims (Lions NSW Eye Bank) and placed in six-well plates or on Lotrafilcon A CLs (CIBA Vision). CnT-50 medium (CELLnTEC; Advanced Cell Systems, Bern, Switzerland) supplemented with 100 U/mL penicillin and streptomycin (Invitrogen, Eugene, OR) was added, and tissue explants were incubated at 37°C in a humidified 5% CO₂ environment. When substantial outgrowth was observed, explants were removed, and adherent cells were enzymatically detached (TrypLE; Invitrogen), subcultivated, and phenotyped. ^14,19 Some limbal epithelial cells or limbal tissue explants were placed on the concave surface of Lotrafilcon A CLs. Culture medium containing either human or bovine serum or CnT-50 was added such that lenses were submerged and, on reaching confluence, were rested over corneas that had epithelial wounds. Some lenses were coated with SV40-transformed human corneal epithelial cells (HCECs) and were cultured in Dulbecco's modified Eagle medium (DMEM/F12; Invitrogen) supplemented with 10 ng/mL epidermal growth factor (EGF; Invitrogen) and 10% FBS (HyClone, South Logan, UT). 

Corneal Wound Model

Fresh cadaveric human corneas (n = 13) were subjected to chemical injury by applying a 6-mm filter paper disc soaked in 1 M sodium hydroxide over the central cornea for 30 seconds. The defect was observed macroscopically using fluorescein sodium 2.0% (Minims; Chauvin Pharmaceuticals Ltd., Surrey, UK) or microscopically after performing standard histology. After chemical debridement, some corneas were placed in organ culture with an uncoated CL under air-lifting conditions for up to 10 days to determine their regenerative capacity. Other corneas were wounded, incubated with primary human limbal or HCEC-laden CLs, and histologically assessed. 

Colony-Forming Efficiency

Six-well plates (Corning Inc., Corning, NY) were coated with recombinant human VN (rhVN) for 24 hours at 4°C. Passage 1 primary human limbal epithelial cells were seeded on coated surfaces at a density of 200 cells/well. Colonies, the progeny of a single cell, ¹⁵ were defined as dense clusters that were counted 12 days after inoculation. The number of cells required per well was determined empirically to avoid colonies merging within the assay period. Colony-forming efficiency (CFE) was calculated as the percentage of cells able to generate colonies (i.e., number of colonies divided by the number of cells seeded). Colonies were fixed and stained in 1% rhodamine B in 3.7% formaldehyde (Sigma) for 1 hour at room temperature. Images were captured by digital photography. Colonies (holoclone-like, 10–30 mm²; meroclone-like, 5–9 mm²; paraclone-like, <5 mm²) were determined as previously described. ³³ Holoclone-like colonies were not serially propagated in these experiments. 

Cell Adhesion

Twenty-four-well tissue culture plates were coated with rhVN (R&D Systems) at 10 μg/mL for 24 hours at 4°C. Wells were washed in PBS and incubated in 0.5% BSA/PBS for 1 hour at 37°C to block nonspecific binding sites. Lotrafilcon A CLs containing confluent early-passage (P1-P2) primary human limbal epithelial cells were cut into four equal segments, placed epithelium-side down onto VN-coated or uncoated tissue culture plastic, and incubated for 72 hours in the presence of 1 μg/mL each linear or cyclic RGD peptide (Merck, Darmstadt, Germany), 1 μg/mL function-blocking antibodies to αvβ3 and αvβ5 (clones 23C6 and P5H9, respectively; R&D Systems), or 1 μg/mL each of anti-VN (clone VIT-2; Abcam, Cambridge, MA) or isotype control antibody (IgM; Abcam). Before removing the lens, a marker pen was used to trace the area it covered, and only cells within that zone were counted. Plates were washed in PBS, fixed in 100% methanol, and stained with hematoxylin, and cells in five-high powered fields (×40 objective) were counted per CL segment, in triplicate. 

Statistical Analysis

Statistical analyses were performed (Prism, version 5.0b; GraphPad Software Inc., La Jolla, CA). Immunoblot band intensities were calculated using ImageJ software (developed by Wayne Rasband, National Institutes of Health, Bethesda, MD; available at http://rsb.info.nih.gov/ij/index.html), and one-tailed paired t-tests or one-way analysis of variance (ANOVA) with Bonferroni's posttest were used to compare means between groups. Immunohistochemical scores for VN staining in different ocular surface regions were assessed by Kruskal-Wallis one-way ANOVA with Dunn's posttest, and the inhibition of cell adhesion was assessed with one-way ANOVA with Dunnett's posttest. Statistical analysis of the CFE was performed by one-way ANOVA with Bonferroni's test. P < 0.05 was considered significant. 

Results

Vitronectin Binds to CL Polymers

Serum proteins were visibly bound to Lotrafilcon A CLs after staining with Coomassie blue (Fig. 1A). Further analysis by SDS-PAGE (Fig. 1B) confirmed the presence of proteins, as denoted by faint bands (Fig. 1B; arrows). Mass spectrometry identified an array of serum-derived proteins that bound to the CL polymer, including the glycoprotein VN (Table 1). Because VN plays a key role in cell migration, ³⁴ cell attachment, ³⁵ SC maintenance, ³⁶ and corneal regeneration, ³⁷ it became a focus of our investigation. The affinity of native VN toward the CL polymer was initially assessed by Western blotting that revealed two immunoreactive bands, a major band corresponding to full-length VN (75 kDa) and a minor truncated species (65 kDa). The intensity of these bands increased with increasing serum concentration and was absent from lenses incubated in media alone (Fig. 1C). Similarly, CLs incubated with increasing serum displayed enhanced immunofluorescence (Figs. 1D–G). 

Figure 1.

 

Serum proteins bind to CL polymers. CLs were incubated in varying serum concentrations for 10 days and stained with Coomassie blue. Deposited proteins were denoted by the dark blue staining (A). After releasing proteins from CLs, SDS-PAGE was performed and was followed by Coomassie blue staining to display protein bands (B, black arrows). In other experiments, CLs were incubated in varying concentrations of serum for 10 days, proteins were extracted, and VN content was displayed by Western blot analysis (C). CLs were cut into segments after serum exposure, and VN deposits were confirmed by immunofluorescence (D–G, green staining). The white hatched lines represent the cut edge of the CL. Original magnification, ×100 (D–G).

Table 1.

 

Proteomic Analysis of Contact Lens–Bound Proteins

1% Serum	10% Serum
Vitronectin	Vitronectin
Apolipoproteins	Apolipoproteins
Serum amyloid proteins	Serum amyloid proteins
Complement (components 1, 4)	Complement (component 1)
Hemoglobin	ND
ND	Coagulation factors
ND	Prepromultimerin

 CLs were incubated for 10 days in 0%, 1%, or 10% human serum. CL-bound proteins were trypsin digested and analyzed by mass spectrometry, and a protein profile was generated. Proteins were not found when lenses were incubated without (0%) serum (not shown). ND, not detected.

Vitronectin Is Adsorbed and Released from CLs

Previously, we demonstrated that Lotrafilcon A (but not Balafilcon A) lenses support epithelial cell growth. ¹⁹ To determine whether these lenses adsorb different amounts of VN, they were incubated for 10 days in sera from four patients with LSCD. VN bound to each lens type was assessed by immunoblotting, and results revealed that levels of this protein on Lotrafilcon A were significantly higher (∼5.5-fold, *P < 0.05) than on Balafilcon A CLs (Fig. 2A). 

Figure 2.

 

Serum VN content on therapeutic CLs. CLs (Lotrafilcon A, Balafilcon A) were incubated for 10 days in 10% sera from patients (P1-P4) with LSCD. VN content was assessed by Western blot analysis. A positive control sample (C⁺), consisting of 2% serum from patient 1 (P1), was included in an adjacent lane, and differences in level of bound VN between the CL types were determined densitometrically (A). In other experiments, CLs were incubated for 10 days in 10% patient sera (B) or rhVN (C) before they were placed in PBS for a further 7 days. Protein that remained attached to lenses at day 0, 4, or 7 was assessed by Western blot analysis. Differences in the amount of native (B) and recombinant (C) VN remaining on lenses was displayed graphically. Histograms represent the mean band intensity ± SE mean. Appropriate statistical analyses were performed. *P < 0.05; ns, not significant.

Given that previous studies have shown the delivery of biologicals and therapeutic agents to the ocular surface by CLs, ^38,39 we sought to determine whether lenses could be coated with VN. Lotrafilcon A CLs were incubated in sera from patients with LSCD for 10 days before they were placed in PBS for an additional 7 days. Western blot analysis for CL-bound VN (after buffer exchange) revealed a progressive and significant loss of this protein with time (Fig. 2B; P < 0.05). Under identical experimental conditions, a similar trend was noted for rhVN, although this did not reach statistical significance (Fig. 2C; P = 0.08). 

Vitronectin Is Localized to the Human Limbus

Next, we explored whether VN was localized on the ocular surface. Immunohistochemical assessment of 13 cadaveric human corneas revealed intense and specific VN staining along the limbal BM (Figs. 3A, 3C), including within the palisades of Vogt and the limbal stroma (Fig. 3B). However, VN was absent along the central corneal BM directly adjacent to Bowman's layer, indicating a potential role for this factor in maintaining progenitor cells within their natural microenvironment. Interestingly, VN was also localized to BMs of other epithelia known to harbor SCs, including the outer root sheath of hair follicles (Figs. 3D–G) and crypts of the small intestine (Fig. 3H). 

Figure 3.

 

(A, B) VN expression in the normal human cornea. Immunofluorescence for VN was performed on human cadaveric corneas (green staining). (A, B, arrows) Intense immunoreactivity on the limbal BM. (A, arrowheads) Weak to absent staining in the peripheral corneal region. VN immunoreactivity was semiquantitatively assessed along the BM of the ocular surface (C) using the HV-2 antibody clone. An isotype control antibody was included in each experiment (inset, B). Nuclei were counterstained with DAPI (A, B). Photomicrographs were taken at ×400 (A) or under ×1000 oil immersion (B). Immunoreactivity scores (C) were collated from three independent masked observers. Sections of human skin (n = 3, D–G) and small intestine (n = 2, H) were stained for VN using the HV-2 antibody. Immunoreactivity (red) and hematoxylin (blue) established the nuclei. (F–H, arrows) Staining within BM regions. The small box in (D) (×100) is magnified in (F) (×1000). The small box in (E) (×200) in magnified in (G) (×1000).

Vitronectin Enhances LSC Clonogenic Activity

To further explore the role of VN in maintaining LSC, in vitro clonogenic assays were performed. VN dose dependently enhanced the colony-forming efficiency of early-generation primary limbal epithelial cells (Fig. 4A). In addition, the colonies that formed on VN were larger and denser, indicative of SC-containing holoclones. ³³ Smaller, irregularly shaped colonies, resembling abortive paraclones, ³³ were also evident; however, there were fewer of these. A significant (P < 0.05) linear relationship was noted for holoclone-like colonies on VN (Fig. 4B). 

Figure 4.

 

VN enhances colony formation. Early-passage (P1) human limbal epithelial cells were seeded in triplicate wells (a, b, c) on VN-coated tissue culture plastic. Dishes were stained with rhodamine B, and colonies were counted. Colony-forming efficiency (A) and colony type (B) were compared between treatment groups. **P < 0.01, *P < 0.05. These data are representative of experiments performed on limbal epithelial cells from three donors.

Transplanted Limbal Epithelial Cells Reconstitute Chemically Damaged Human Corneas

An obvious disadvantage of our recently described transplantation strategy for patients with LSCD is the lack of direct evidence showing how cells transfer from a CL carrier to the patient's cornea and how they are maintained on CLs and on the ocular surface during implantation and regeneration. ¹⁴ Using an ex vivo model whereby a central corneal epithelial defect was chemically induced (Fig. 5), ocular surface epithelial cells grown on CLs were successfully transferred and partially (Fig. 5I) or completely (Figs. 5J–L) resurfaced the corneal epithelial defect. CLs without cells were ineffective at resolving the epithelial defect (Figs. 5G, 5H). Interestingly, in experiments in which epithelial cells were grown on CLs in media supplemented with FBS, after implantation, an antibody directed against bovine VN (Ab-62; CSIRO) was able to detect its presence on the BM of regenerating cells (Fig. 5K), a zone normally void of VN (Figs. 3A, 3C). In contrast, bovine VN was not identified in cells that were expanded without FBS (Fig. 5L). These data imply that VN is either released from the CL and adsorbed onto the cornea or is derived from the serum-containing media. The latter explanation is less likely because corneas were air-lifted while the lenses were on the ocular surface. 

Figure 5.

 

VN expression after cell transfer and integration. An ex vivo model of severe corneal damage was induced using donor human corneas (A). After applying a sodium hydroxide-soaked filter paper disc to the central cornea (B), a defect developed (C) that was also evident on fluorescein staining (D). Corneas were histologically assessed before (E) and immediately after (F) chemical debridement. Macroscopically (G) and microscopically (H), central corneal defects remained unresolved even after application of an uncoated CL. Arrows in (E) and (F) identify an intact Bowman's layer. Corneal defects partially resolved after 10 days with a primary epithelial cell-laden CL (I) or completely resolved using a corneal epithelial cell line-coated lens (J). Arrowheads in (I) and (J) indicate corneal integration by transplanted cells. VN immunoreactivity (red) adjacent to transplanted cells was confirmed using an anti-bovine VN antibody (Ab-62) (K, arrows). In cells that were not cultured in FBS, bovine VN could not be detected after using the same antibody clone (L). Negative controls consisted of omitting the primary antibody or adding an isotype control (not shown). Sections in (E), (F), and (H–J) were stained with hematoxylin and eosin (H&E), whereas those in (K) and (L) were counterstained with hematoxylin alone. All images were acquired under oil immersion (×1000 magnification) except those in (I) (×400). Scale bar, 1000 μm (G).

Cell Transfer from CLs Involves Vitronectin

To determine the importance of VN in cell transfer, an in vitro model was developed whereby cell migration from CLs onto VN-coated tissue culture plastic was assessed (Fig. 6). We reasoned this model would provide more accurate and reliable proof-of-concept data compared with using cadaver human or animal corneas in which donor-to-donor variation, time of death, age, and condition of the ocular surface could influence our readout. The results show that VN increased cell transfer and adhesion from the lens polymer to tissue culture plastic (compare Fig. 6A to Fig. 6F), whereas RGD peptides (competing for the VN cell adhesion site), a VN receptor antibody (αvβ5 but not αvβ3 or an isotype control), and a VN antibody (VIT-2) significantly suppressed VN-mediated cell transfer and adhesion (Figs. 6A–G). 

Figure 6.

 

VN and VN receptor blockade suppresses cell adhesion. Human limbal epithelial cells were photographed (A–F) and counted (G) after migration from confluent CL segments onto VN (A–E, G, black bars) or uncoated (F, G, white bar) tissue culture plastic. Cells were treated with RGD peptides (B), anti-αvβ5 (C), anti-VIT-2 (D), or anti-IgM isotype control antibody (E). Cells were counterstained with hematoxylin and photographed (×400, final magnification). Bars represent mean cell counts derived from triplicate experiments. **P < 0.001; ***P < 0.0001.

Discussion

Recently, we developed a novel strategy to deliver autologous ocular surface epithelial progenitor cells, cultured in media supplemented with serum, to patients with LSCD using therapeutic CLs. ¹⁴ We now demonstrate the binding and exchange of important serum proteins, including VN, from CLs (Figs. 1, 2; Table 1). VN is a factor likely to support progenitor cells while they expand in vitro on a CL scaffold and to facilitate cell migration and reepithelialization during and after transplantation. In support of this hypothesis, VN was identified on the limbal BM, a zone known to harbor SCs, ^2,3 but was absent from the central corneal BM (Fig. 3), the location for differentiated epithelial cells. ⁴⁰ In addition, VN increased colony formation (Fig. 4), an indicator of SC activity. VN deposits were found in the central cornea adjacent to transplanted cells (Fig. 5), and selective blockade of VN or its integrin receptor suppressed cell transfer (Fig. 6). 

CLs support ocular surface epithelial cells, with therapeutic implications for corneal repair. ^{14,19 –21} In the context of our technique, ¹⁴ we sought to determine whether serum-derived wound healing and cell support factors adhere to silicone hydrogels and aid in ex vivo LSC maintenance before and during the transplantation procedure. Although other investigators have reported the adhesion of tear-derived cholesterol, ⁴¹ lysozyme, ⁴² and other proteins ⁴³ to CL polymers, we observed the ability of numerous serum proteins (including VN) to transiently absorb on similar surfaces. Moreover, our observation that rhVN elutes from CLs in a manner similar to that of its native counterpart has implications for the development of a slow-diffusion delivery system for ocular surface wound-healing factors. Application of recombinant proteins could be a standardized approach for treating ocular surface disorders that do not require a cell-based therapy, including persistent epithelial defects. ^{37,44 –47} Indeed, CLs have been used to deliver EGF to wounded corneas, ³⁹ and pharmacologic agents have been absorbed onto lenses for glaucoma management. ³⁸ Interestingly, though EGF alone induced rapid reepithelialization, the regenerating cells were loosely bound to the corneal BM, indicating the necessity for coapplication of adhesion-like molecules to facilitate stability. ³⁹ Our system has this added potential because VN is known to couple growth factors such as insulin-like growth factor-binding proteins (IGF-BPs), which partake in corneal ⁴⁸ and cutaneous ⁴⁹ wound repair. VN/IGF-BP complexes are also supportive factors for cultured SCs, ⁵⁰ and VN, FN, and nerve growth factor have each been found to promote the healing of persistent corneal epithelial defects. ^37,44,45  

We noted that the ability of limbal epithelial cells to adhere to and expand on CL surfaces was type specific. ¹⁹ The capacity of proteins to deposit on synthetic substrates such as hydrogels relies on polymer composition, water content, ionicity, hydrophilicity, and hydrophobicity, otherwise known as wettability. ^51,52 Here, we report that Lotrafilcon A CLs more readily adsorb VN than the alternative material, Balafilcon A. Although this seems contradictory to reports that indicate Balafilcon A CLs adsorb the highest amount of protein, ⁵³ the composition of Lotrafilcon A lenses may be more conducive to VN adsorption. The same investigators also showed that VN adsorption is greater on next-generation Lotrafilcon B than on Balafilcon A CLs. ⁵⁴ In conjunction with supporting more VN, Lotrafilcon A CLs have significantly higher oxygen permeability than Balafilcon A CLs ⁵⁵ and may be conducive to sustaining epithelial cell growth. ¹⁹  

ECM proteins, including VN, FN, collagen IV, tenascin-C, and laminin, have been identified along the BM or within the stroma directly underlying putative LSCs, ⁵⁶ and key roles for these proteins in SC maintenance have been identified. For example, cells that rapidly adhere to collagen IV have higher DNA label retention, ²⁴ a universal SC feature indicative of slow-cycling cells. ³ In addition, FN has been shown to delay the onset of terminal differentiation by triggering involucrin expression, ⁵⁷ and VN maintains SCs in culture. ^36,50 Similarly, cell surface receptors for ECM molecules (integrins) are involved in corneal wound healing, ^27,58 are often expressed in a polarized manner along BMs, ⁵⁹ potentiate limbal epithelial outgrowth from tissue explants placed on HAM, ⁶⁰ and support an undifferentiated phenotype on the same bioscaffold. ⁶¹ Interestingly, a VN/VN-receptor (αvβ5) axis has been implicated in promoting cell migration and spreading in explant outgrowth, ⁶² an approach analogous to ours. ¹⁴ Therefore, it is tempting to speculate that VN released from CLs and engagement with its receptor may be a mode by which cells transfer between CLs and the ocular surface. Consolidating this view was the observation of VN deposits directly beneath the donor-transplanted cells, along the recipient corneal BM (Fig. 5), a region normally void of this factor (Fig. 3). Although VN appears to be one of the major proteins involved in supporting the expansion of limbal epithelial cells on silicone hydrogel lenses and in the transfer of these cells in organotypic models, it is likely that the highly related molecule FN might also play a role. ⁶³ We have previously identified FN on cell-laden CLs before transplantation ¹⁴ and detected its presence in the CL proteome; however, this factor was not stably bound to the lens polymer (data not shown) and was, therefore, not pursued in our investigations. It is also possible that other ECM proteins were bound but were below the sensitivity of our proteomic approach. Notably, a host of ECM and BM proteins have been identified within the limbal zone ²² that may be of equal importance to or greater importance than VN in supporting LSC. 

Our study elucidated a potential role of VN in supporting LSCs in vitro and in vivo. Further clarification of the role of this protein and its integrin receptors in LSC biology is required. Future investigations will focus on manipulating ECM/integrin interactions at immunologic, pharmacologic, and molecular levels using animal models of LSCD and corneal wounding to dissect the pathways involved. Finally, CLs could potentially be used as slow-delivery devices for the application of wound-healing factors to facilitate rapid corneal wound closure and to minimize the risk for infection, scarring, and corneal melting in patients with persistent epithelial defects or after surgery. Understanding how epithelial progenitors adhere, detach, and transfer from CLs to the cornea and understanding their interactions with ECM proteins and their receptors will provide additional knowledge to refine our current clinical strategy and will provide insight for the treatment of more prevalent ocular surface disorders, including those that do not require a cell-based therapy. 

Footnotes

 Supported by National Health and Medial Research Council of Australia Career Development Award 455358 (ND), a University of NSW GoldStar Award, and an Australian Stem Cell Centre Strategic Development Award.

Footnotes

 Disclosure: T.J. Echevarria, None; S. Chow, None; S. Watson, None; D. Wakefield, None; N. Di Girolamo, None