Normalization of Wound Healing and Diabetic Markers in Organ Cultured Human Diabetic Corneas by Adenoviral Delivery of c-Met Gene

Abstract

Purpose.: Diabetic corneas display altered basement membrane and integrin markers, increased expression of proteinases, decreased hepatocyte growth factor (HGF) receptor, c-met proto-oncogene, and impaired wound healing. Recombinant adenovirus (rAV)–driven c-met overexpression in human organ-cultured corneas was tested for correction of diabetic abnormalities.

Methods.: Forty-six human corneas obtained postmortem from 23 donors with long-term diabetes (5 with diabetic retinopathy) were organ cultured and transduced with rAV-expressing c-met gene (rAV-cmet) under the cytomegalovirus promoter at approximately 10⁸ plaque-forming units per cornea for 48 hours. Each control fellow cornea received control rAV (rAV expressing the β-galactosidase gene or vector alone). After an additional 4 to 5 days of incubation, 5-mm epithelial wounds were created with n-heptanol, and healing was monitored. The corneas were analyzed afterward by immunohistochemistry and Western blot analysis. Signaling molecule expression and role was examined by immunostaining, phosphokinase antibody arrays, Western blot analysis, and inhibitor analysis.

Results.: rAV-cmet transduction led to increased epithelial staining for c-met (total, extracellular, and phosphorylated) and normalization of the patterns of select diabetic markers compared with rAV-vector–transduced control fellow corneas. Epithelial wound healing time in c-met–transduced diabetic corneas decreased twofold compared with rAV-vector–transduced corneas and became similar to normal. c-Met action apparently involved increased activation of p38 mitogen-activated protein kinase. c-Met transduction did not change tight junction protein patterns, suggesting unaltered epithelial barrier function.

Conclusions.: rAV-driven c-met transduction into diabetic corneas appears to restore HGF signaling, normalize diabetic marker patterns, and accelerate wound healing. c-Met gene therapy could be useful for correcting human diabetic corneal abnormalities.

In pathologic conditions, such as diabetes mellitus, the cornea is seriously affected, which can cause visual impairment. ^1,2 The most recognized corneal complication caused by both type I (insulin dependent, IDDM) and type II (non–insulin-dependent, NIDDM) diabetes is referred to as diabetic keratopathy. Epithelial basement membrane (BM), epithelial wound healing, epithelial–stromal interactions, and endothelial and nerve function are impaired in the corneas of diabetic patients. ^3–10 In human diabetic corneas, we have found alterations of specific BM proteins (laminin-8, laminin-10, and nidogen-1/entactin) and an α₃β₁ laminin-binding integrin. ^5,11 In addition, such corneas, intact or organ cultured, display abnormal overexpression of insulin-like growth factor (IGF)-I and the proteinases MMP-10 and cathepsin F, as well as downregulation of fibroblast growth factor (FGF)-3 and its receptor FGFR3, thymosin β₄, and the hepatocyte growth factor (HGF) receptor c-met proto-oncogene. ^12–14 These results suggest that diabetic keratopathy is a result of decreased migratory growth factors and elevated proteinase levels that could lead to BM degradation, lessened epithelial adhesion, and clinically observed ¹⁵ delayed wound healing compared with normal corneas. 

The HGF/c-met system has been recognized as essential in cell migration and wound healing in different systems including the cornea. ^16–22 Overexpression of c-met and/or its constitutive activation by truncation contributes to increased invasive, angiogenic, and metastatic properties of various tumors. ^23–25 Using gene microarray analysis validated by quantitative RT-PCR and immunohistochemistry, we found an increased expression of HGF (possibly because of diabetes-elevated hypoxia ²⁶ ) in the diabetic cornea, whereas we observed diminished c-met expression. ¹⁴ Therefore, the signaling and functional effects of HGF in diabetic corneas could have been impaired because of decreased c-met expression. This finding prompted us to test c-met as a target for viral-mediated gene therapy. The data reported herein show that c-met overexpression in organ cultured human diabetic corneas driven by a recombinant adenoviral (rAV) vector led to amelioration of the BM and integrin marker proteins and a c-met-specific normalization of epithelial wound healing, possibly by the activation of p38 mitogen-activated protein kinase (MAPK). 

Materials and Methods

Corneal Organ Culture

Postmortem diabetic human eyes or corneas were purchased from the National Disease Research Interchange (NDRI, Philadelphia, PA). NDRI has a human tissue collection protocol approved by a managerial committee and subject to National Institutes of Health oversight, and the donor corneas were managed by us in accordance with the guidelines of the Declaration of Helsinki for research involving human tissue. A total of 46 corneas (Table 1) from 23 patients with IDDM (n = 9) and NIDDM (n = 14) were used (15 men, 8 women; mean age, 69.3 ± 14.0 years). The corneas were organ cultured over agar-collagen gel, as described. ¹¹ The concavity of the corneas with a scleral rim was filled with a warm agar-collagen mixture that quickly solidified. The corneas were cultured with epithelium facing upward, at a liquid–air interface, in serum-free medium with insulin-transferrin-selenite, antibiotics, and antimycotic (Invitrogen, Carlsbad, CA) covering the limbus. Medium (100 μL) was added one to two times a day, to moisten the epithelium. 

Table 1.

 

Donor Characteristics

Case Number	Diabetes Type	Age, Sex	Disease Duration (y)	Cause of Death
07-27	IDDM	81, M	>10	Acute renal failure
07-32	NIDDM	84, M	20	Stroke
07-34	IDDM, DR	37, M	22	Intracranial hemorrhage
08-35	IDDM, DR	88, M	38	Cardiac arrest
08-36	NIDDM	82, M	15	Stroke
08-38	IDDM	73 F	>10	Diabetic ketoacidosis
08-40	NIDDM	59, M	20	Cardiac arrest
08-44	IDDM, DR	71, M	15	Cardiopulmonary arrest
08-49	NIDDM	59, F	28	Intracranial hemorrhage
08-51	NIDDM	68, F	7	End-stage renal disease
08-52	NIDDM, then IDDM	58, F	10, 5 (15 total)	End-stage renal disease
08-54	NIDDM	57, M	10	Myocardial infarction
08-57	IDDM	78, F	15	Respiratory failure
08-59	IDDM	64, M	30	Sepsis
09-5	IDDM	77, M	8	Congestive heart failure
09-6	NIDDM	85, M	25	Congestive heart failure
09-8	NIDDM	40, M	10	Anoxic brain injury
09-12	NIDDM, DR	61, F	Unknown	Myocardial infarction
09-14	NIDDM	85, M	25	Alzheimer's disease
09-16	NIDDM	61, F	>10	Congestive heart failure
09-21	NIDDM	79, M	5	Myocardial infarction
09-24	NIDDM	80, M	23	Coronary artery disease
09-26	IDDM, DR	67, F	30	Pneumonia

Culture Treatments

Some cultures were treated with small molecule specific inhibitors of c-met kinase (PHA 665752; Tocris Bioscience, Ellisville, MO) and of p38 MAPK signaling (SB 202190; EMD Biosciences, San Diego, CA) during wound healing. Inhibitors dissolved in dimethylsulfoxide (DMSO; Sigma-Aldrich, St. Louis, MO) were added to the medium right after the wound was made and were kept with the cells until complete healing. Control fellow corneas received the same amount of DMSO for PHA 665752 or of an inactive inhibitor SB 202474 (EMD Biosciences) for SB 202190. The concentration of PHA 665752 was 150 nM ^27,28 with 0.15% DMSO in the medium or 10 μM for SB 202190 and SB 202474 ^29,30 with 0.33% DMSO in the medium. Medium with inhibitors was changed every 2 to 3 days. It was important to dissolve the stock solutions of SB 202190 (and its analogue) in prewarmed medium because of its reported concentration/solubility fluctuations in aqueous solutions. ³¹  

rAV transduction efficiency was increased by routinely adding 75 μg/mL sterile sildenafil citrate (Viagra; Pfizer Corp., New York, NY) to the culture, together with the virus for 3 hours, after which another portion was added, because of the short half-life of sildenafil in aqueous solutions. ³²  

Epithelial Wound Healing

Epithelial debridement was performed by placing a 5-mm paper disc soaked in n-heptanol (1-heptanol; Sigma-Aldrich) on the central corneal surface for 1 minute. ¹¹ Contrary to mechanical debridement, n-heptanol does not disrupt fragile diabetic epithelial BM. ^33,34 Both healing corneas—target gene and control treated—were photographed every 24 hours until the epithelial defect was completely healed and the healing time recorded. Photography of live corneas was performed with a microscope (BX40; Olympus USA, Melville, NY), equipped with a high-sensitivity, 2-megapixel color digital camera (MicroFire; Optronics, Goleta, CA), using ×4 (for large field) and ×10 (for cell visualization) plan apochromatic objectives. After they healed, the corneas were cut in half, with one half embedded in OCT compound (Ted Pella, Inc., Redding, CA), and processed for indirect immunofluorescence on cryostat sections. ¹¹ The other half, or its separated epithelium and stroma, were frozen in liquid nitrogen for Western blot analysis or quantitative RT-PCR. 

Adenoviruses

The recombinant adenoviruses rAV-gal (with the β-galactosidase gene), rAV-vector (no gene inserted), and rAV-cmet (with the c-met gene) were E1/E3-deleted type 5 rAV expressing genes under the control of the major immediate early cytomegalovirus promoter. All the viruses were generated based on recombination technology (Gateway; Invitrogen), as per the manufacturer's instructions. Briefly, the human c-met full-length open reading frame (ORF) clone (Ultimate; Invitrogen) provided in an entry vector (Gateway pENTR221; Invitrogen) was transferred into the rAV vector pAd/CMV/V5-DEST, by using a kit and enzyme mix (ViraPower Adenoviral Gateway Expression Kit and LR Clonase II enzyme mix; Invitrogen). pAd/CMV/V5/lacZ for β-galactosidase expression is included with each kit for use as a positive control in the adenoviral expression system. The reaction mixtures were transformed into DH5 chemically competent Escherichia coli to select expressing clones. The selected expression clones were sequenced at the UCLA Core Facility to confirm the correct orientation of the ORF. The Pac I-digested vectors were used to transfect 293A cells to produce rAV stocks. The rAVs were amplified by infecting the 293A producer cells with the crude viral lysates and purified (Vivapure AdenoPACK kit; Sartorius-Stedim, Concord, CA). The titers of rAV stocks were determined twice in HEK293 cells, as described. ³⁵  

Cell lines used for viral production and initial transduction included human embryonic kidney (HEK293), human glioma U87MG, and Chinese hamster ovary (CHO) cell lines from the American Type Culture Collection (Manassas, VA). The 293A cell line for rAV stock production was from Invitrogen. The cells were cultured in Dulbecco's modified Eagle's medium (DMEM; Invitrogen) or a mixture of DMEM-Ham's F-12 (1:1) with 10% fetal calf serum (FCS), l-glutamine, antibiotics, and sodium pyruvate, in a humidified atmosphere at 37°C with 5% CO₂. 

Viral Transduction

The cell lines U87MG and CHO were initially used for rAV transduction to confirm target gene overexpression. The diabetic organ-cultured corneas were transduced with rAV-cmet. The other cornea of each pair (control cornea) was treated with control rAV-gal or rAV-vector. The rAV particles were given at 0.8 to 1.25 × 10⁸ plaque-forming units (pfu) per cornea in culture medium for 48 hours. During incubation with rAV, the corneas were kept under the medium surface. After the additional 4-day standard incubation at the liquid–air interface, corneas were tested for epithelial wound healing (see above) or processed for immunohistochemistry, Western blot or quantitative real-time RT-PCR. 

Human Phospho-MAPK Array

The human phospho-MAPK array kit (Proteome Profiler Array, ARY002; R&D Systems, Minneapolis, MN) was used to determine the relative level of phosphorylation of 19 MAP kinases. The nitrocellulose membrane array contains the capture and control antibodies spotted in duplicate. It was used according to the manufacturer's instructions with total lysates of trephined corneas. Total lysates of rAV-vector and rAV-cmet trephined fellow corneas were diluted to 150 to 200 μg/mL protein in a detergent-, urea-, and phosphatase inhibitor-containing solubilizing buffer (R&D Systems) and incubated with the arrays overnight at 4°C. After washing away the unbound material, membranes were incubated with a cocktail of phospho-site–specific, biotinylated antibodies, to detect phosphorylated kinases with streptavidin-horseradish peroxidase and a chemiluminescent substrate kit (Chemi Glow; Alpha Innotech, San Leandro, CA). The signals were revealed on ECL film (Amersham Hyperfilm; GE Healthcare, Little Chalfont, UK). Independent experiments were performed on six pairs of corneas. 

Immunohistochemistry

Immunohistochemistry was performed as described before. ^14,32,36 The antibodies and their reactivities in different assays are presented in Table 2. For each marker of fellow corneas, the same exposure time was used when photographing stained sections of a vector-transduced one and a c-met–transduced one. Routine controls with omission of primary antibodies were negative. 

Table 2.

 

Antibodies

Antigen	Antibody and Cat. No.	Positive In	Source/Reference
c-Met total	Rabbit pAb sc-161	IHC, WB	Santa Cruz Biotechnology
c-Met total	Rabbit pAb sc-10	IHC, WB	Santa Cruz Biotechnology
c-Met total	Rabbit pAb sc-8307	WB*	Santa Cruz Biotechnology
c-Met total	Rabbit pAb 71-8000		Zymed
c-Met extracellular	Mouse mAb MAB358	WB	R&D Systems
c-Met extracellular	Mouse mAb 05-238	IHC	Millipore
p-c-Met (Tyr1003)	Rabbit pAb ab61024	IHC	Abcam
p-c-Met (Tyr1234/Tyr1235)	Rabbit pAb 44887G	WB*	Invitrogen
p-c-Met (Tyr1230/Tyr1234/Tyr1235)	Rabbit pAb 44888G		Invitrogen
HGF	Rabbit pAb sc-7949	IHC	Santa Cruz Biotechnology
Nidogen-1	Mouse mAb A9	IHC	Kabosova et al. 37
Nidogen-1	Mouse mAb MAB2570	IHC	R&D Systems
Nidogen-2	Rabbit pAb 1080	IHC	Kabosova et al. 37
Laminin γ1 chain	Rat mAb A5	IHC	Kabosova et al. 37
Integrin α3β1	Mouse mAb MAB1992	IHC	Millipore
Akt	Mouse mAb 610860	WB	BD Transduction Labs
p-Akt (Ser472/Ser473)	Mouse mAb 550747		BD Pharmingen
p-Akt (Ser473)	Rabbit pAb sc-7985-R	IHC, WB	Santa Cruz Biotechnology
p-Akt (Ser473)	Rabbit pAb 9271	IHC, WB	Cell Signaling
p-Akt (Thr308)	Rabbit mAb 2965		Cell Signaling
ERK1/2	Rabbit mAb 4695	IHC, WB	Cell Signaling
p-ERK1/2 (Thr202/Tyr204)	Rabbit pAb sc-101760		Santa Cruz Biotechnology
p-ERK1/2 (Tyr 204)	Mouse mAb sc-7383		Santa Cruz Biotechnology
p-ERK1/2 (Thr202/Tyr204)	Rabbit mAb 4370	WB	Cell Signaling
p-ERK1/2 (Thr185/Thr202)	Rabbit pAb ab4819	IHC	Abcam
p-ERK1/2 (Thr183/Tyr185)	Mouse mAb ab50011		Abcam
p38 MAPK	Rabbit pAb 9212	WB	Cell Signaling
p-p38 (Thr180/Tyr182)	Rabbit pAb AB3828	WB*	Millipore
p-p38 (Thr180/Tyr182)	Rabbit mAb 9215	WB	Cell Signaling
p-p38 (Thr180/Tyr182)	Mouse mAb 612280		BD Transduction Labs
p-p38 (Thr180/Tyr182)	Mouse mAb ab50012	IHC	Abcam
p85α PI3 kinase	Mouse mAb sc-56939		Santa Cruz Biotechnology
p-p85α PI3 kinase (Tyr508)	Goat pAb sc-12929		Santa Cruz Biotechnology
p-EGFR (Tyr845)	Rabbit pAb 44-784G	IHC	Invitrogen
p-EGFR (Tyr845)	Rabbit mAb 2342-1		Epitomics
ZO-1	Rabbit pAb 40-2300	IHC†	Invitrogen
Claudin-1	Rabbit pAb 51-9000	IHC†	Invitrogen
Ki-67	Mouse mAb sc-101861	IHC‡	Santa Cruz Biotechnology
Activated caspase-3	Rabbit pAb G7481	IHC	Promega
β-Actin	Mouse mAb A5316	WB	Sigma-Aldrich

 Santa Cruz Biotechnology, Santa Cruz, CA; Zymed, South San Francisco, CA; R&D Systems, Minneapolis, MN; Millipore, Billerica, MA; Abcam, Cambridge, MA; Invitrogen, Carlsbad, CA; BD Transduction Laboratories, Lexington, KY; BD Pharmingen, San Diego, CA; Cell Signaling, Beverly, MA; Epitomics, Burlingame, CA; Promega, Madison, WI; Sigma-Aldrich, St. Louis, MO. mAb, monoclonal antibody; pAb, polyclonal antibody; IHC, immunohistochemistry; WB, Western blot; p-, phosphorylated.

*  Weak reactivity.

†  95% ethanol followed by acetone fixation needed.

‡  Acetone fixation needed.

Western Blot Analysis

Transduced corneas were trephined with a 8.25-mm stainless-steel trephine, and a cut out part for protein analysis was snap-frozen in liquid nitrogen and stored there until further use. Frozen corneas were wrapped in aluminum foil, smashed with a hammer to a powder, and dissolved in Laemmli's buffer with 1% SDS, 5% β-mercaptoethanol (Invitrogen), and proteinase inhibitors (Sigma-Aldrich). For Western blot analysis, 8% to 16% gradient Tris-glycine SDS polyacrylamide gels were used (Invitrogen). Gel loading was normalized by β-actin content. ³² After transfer to nitrocellulose membranes, blots were blocked in 5% defatted milk and incubated with primary antibodies. Alkaline phosphatase–conjugated secondary antibodies (Jackson ImmunoResearch Laboratories, West Grove, PA) were used with standard substrates or with chemiluminescent substrate (Lumi-Phos WB; Thermo Scientific, Rockford, IL) as per the manufacturer's protocol. 

RNA Extraction and cDNA Generation and Amplification

Total RNA was extracted from corneal epithelium and stroma (TRIzol; Invitrogen), treated with DNase I to eliminate possible genomic DNA contamination, and purified (RNeasy minicolumns; Qiagen, Valencia, CA). The RNA yield was determined by spectroscopy (model ND-1000; NanoDrop Technologies, Wilmington, DE). Total RNA (0.5 μg) was reverse transcribed into first-strand cDNA (High Capacity cDNA Reverse Transcription Kit; Applied Biosystems, Inc. [ABI], Foster City, CA) and random primers. Reactions were incubated at 25°C for 10 minutes, 37°C for 120 minutes and inactivated at 85°C for 5 seconds on a thermal cycler (PTC-200 Peltier; MJ Research, Waltham, MA). 

Quantitative Real-Time RT-PCR

Quantitative RT-PCR was performed with a sequence-detection system (Prism 7900HT; ABI) in 384-well plates, to evaluate the expression levels of the c-met gene along with the expression of the endogenous control gene, β₂-microglobulin (TaqMan Gene Expression Assays; ABI). The kit contains custom optimized primer probe sets consisting of two unlabeled PCR primers and the FAM dye-labeled MGB probe (TaqMan; ABI) formulated into a single mixture. All probes are designed to extend an intron–exon junction, to avoid signal from any possible genomic DNA contamination. The following gene-specific assay kits were used as per ABI: c-met proto-oncogene assay ID Hs00179845_m1 (GenBank RefSeq NM_001127500.1, exon boundary 10-11; assay location, 2605; amplicon size, 81 bp) and β₂-microglobulin assay ID Hs99999907_m1 (GenBank RefSeq NM_004048.2, exon boundary 2-3; assay location, 413; amplicon size, 75 bp) (www.ncbi.nlm.nih.gov/locuslink/refseq/ RefSeq is provided in the public domain by the National Center for Biotechnology Information, Bethesda, MD). 

PCR reactions were performed (Taqman Universal PCR Master Mix; ABI) in a final volume of 20 μL. Briefly, all mixtures contained 40 ng of cDNA template, 1× primer probe sets (Taqman Gene Expression Assay; ABI), and 1× PCR master mix (TaqMan Universal; ABI). The reactions were cycled as follows: 2 minutes at 50°C to allow uracil N-glycosylase (UNG) activity (AmpErase; ABI), 10 minutes at 95°C, and 40 cycles at 95°C for 15 seconds and 60°C for 1 minute. The samples were simultaneously run in triplicate for target and endogenous β₂-microglobulin housekeeping gene control. Signals from each sample were normalized to values obtained for β₂-microglobulin. For analysis, the comparative threshold cycle (Ct) method (ΔΔCt) was used, which corrects for any difference in the expression of the internal normalization gene. 

Statistical Analysis

Times to complete wound healing after chemical debridement of the epithelium in pairs of vector and c-met–transduced corneas were analyzed by the two-tailed Wilcoxon matched-pairs test; paired two-tailed t-test gave similar results (all statistical analyses by InStat 3 software program; GraphPad Software, San Diego, CA). Comparison of times to complete wound healing with an additional group of normal corneas ¹¹ was performed with ordinary ANOVA with the Bonferroni multiple-comparisons post test; the same results were obtained with the Tukey-Kramer post test. Densities of Western blot bands relative to β-actin (or internal negative controls on phosphoproteomic arrays) were determined by using the histogram tool (Photoshop CS software; Adobe Systems, San Jose, CA) as described ³⁸ ; a paired t-test was used for statistical analysis. For all tests P < 0.05 was considered significant. Data are presented as the mean ± SE. 

Results

Validation of Viral Gene Transfer

When U87MG or CHO cells were transduced with pretitered rAV-cmet, an elevated expression of c-met protein was shown by Western blot analysis (Fig. 1A, right, for U87MG; mostly the processed 140- to 145-kDa heavy chain was increased) compared with the control rAV-vector. This validated the efficacy of the transduction. Similar experiments were then performed in organ cultured diabetic corneas. Corneas transduced with rAV-cmet had increased expression on Western blot (Fig. 1A, left) of the 170-kDa band (1.37 ± 0.07 times; n = 4, P < 0.02 vs. vector), apparently corresponding to the c-met precursor and of the 55- to 60-kDa band (1.54 ± 0.2 times; n = 7, P < 0.04 vs. vector), apparently corresponding to the cell-associated c-met without extracellular domain. ^39,40 The observation that the smaller band was strong in the rAV-vector–transduced corneas but rather faint in the U87MG cells suggests that corneal cells process c-met in a different, cell-specific way. The same processing pattern was observed in normal corneas (data not shown here). Immunostaining for c-met compared with rAV-gal– or rAV-vector–transduced fellow corneas showed predominant localization to epithelial cells, ⁴¹ and it was increased on rAV-cmet transduction. This staining pattern concerned total, phosphorylated, and extracellular c-met (Fig. 2). Stromal or endothelial staining was similar to that in the control corneas. These data are in good agreement with quantitative RT-PCR results. On rAV-cmet transduction, c-met mRNA showed on average a three-fold increase in the epithelium compared with rAV-vector, whereas stromal c-met mRNA did not change on rAV-cmet transduction (Fig. 1B). 

Figure 1.

 

rAV-cmet transduction into organ cultured human diabetic corneas and U87MG cells led to c-met overexpression. (A) Immunoblot analysis of total protein from corneas transduced with rAV-cmet and rAV-vector probed with rabbit pAb sc-161 (left), and from U87MG cell lysates (right). Arrows: 170-kDa full-length c-met, 140- to 145–kDa processed chain, and 55- to 60-kDa cell-associated fragment. Note that, in glioma cells, transduced c-met gets processed and mainly a 145-kDa chain is detected, whereas in the corneas, a full-length protein is present together with a prominent smaller band, apparently corresponding to a cell-associated kinase fragment. β-Actin was used to normalize protein loading. M, markers in kilodaltons. (B) Quantitative RT-PCR. On c-met gene transduction, its mRNA level increased approximately threefold in the epithelium, but stromal level did not change. Ratios to vector control are presented. β₂-microglobulin was used as the housekeeping gene.

Figure 2.

 

Transduction of c-met to organ cultured human diabetic corneas caused its increased expression and phosphorylation. Left: rAV-vector transduction; right: rAV-cmet transduction. In corneas transduced with rAV-cmet, there was increased epithelial staining for total c-met (revealed with a C-terminal rabbit pAb sc-10), tyrosine-phosphorylated (PY) c-met, and extracellular (EX) c-met. Indirect immunofluorescence of corneal sections. Sections of wounded and healed corneas are shown; the same exposure time was used for each pair of fellow corneas. e, epithelium; s, stroma.

Distribution of Diabetic Markers in rAV-cmet–Transduced Organ Cultured Corneas

Using immunostaining, we determined that the markers altered in the in vivo diabetic corneas (integrin α₃β₁, laminin isoforms, nidogen-1, and nidogen-2) all returned to a nearly normal pattern after rAV-cmet transduction of diabetic organ cultured corneas compared with the control rAV-vector transduction. Integrin staining became more uniform, whereas BM staining became markedly stronger and more continuous (Fig. 3). The staining for HGF did not change with c-met overexpression (data not shown). 

Figure 3.

 

c-Met transduction to organ cultured human diabetic corneas normalized expression patterns of select diabetic markers. Staining for integrin α₃β₁ became more homogeneous, and its intensity increased (top row). Staining for the BM component laminin-γ1 chain appeared after c-met transduction (second row). Staining for BM nidogen-1 and -2 (double labeling is shown in the two bottom rows) remained interrupted on vector transduction but became homogeneous after c-met transduction, similar to normal corneas. Indirect immunofluorescence of corneal sections. e, epithelium; s, stroma.

Tight Junction Protein Patterns after c-Met Overexpression

Human diabetic corneas, by some accounts, have a compromised epithelial barrier function. This function is largely controlled by epithelial tight junctions. Because the HGF/c-met system promotes cell motility, which may loosen these junctions, ⁴² the patterns of the major tight junction proteins ZO-1 and claudin-1 were studied after c-met overexpression. As shown in Figure 4, c-met overexpression did not result in appreciable changes of ZO-1 and claudin-1 patterns. These data suggest that c-met did not have an adverse influence on epithelial barrier function, in accordance with previous data. ^43,44  

Figure 4.

 

c-Met transduction did not change tight junction protein patterns. Top row: immunostaining for claudin-1 with typical membrane staining of all cell layers; bottom row: staining for ZO-1 mostly expressed in upper epithelial layers. None of these tight junction proteins appeared to change after c-met transduction. Indirect immunofluorescence of corneal sections. e, epithelium; s, stroma.

Effect of rAV-cmet on Corneal Epithelial Wound Healing

To test whether c-met gene transfer would correct abnormal wound healing in the diabetic organ cultured corneas, corneas transduced with rAV-cmet and rAV-vector were wounded, and the healing process was monitored every 24 hours. It took much less time for diabetic corneas to heal after c-met transduction compared with the control (Fig. 5). Wounded diabetic corneas transduced with rAV-cmet (n = 7) completely closed n-heptanol–induced epithelial defects in 3.1 ± 0.4 days on average (Fig. 6). This healing time was slightly higher than average for normal corneas in our previous study (2.3 ± 0.2 days; n = 13), ¹¹ but the values were not significantly different (P > 0.05, Fig. 6). The other cornea of each diabetic pair (n = 7) transduced with control rAV-gal or rAV-vector healed significantly slower than the rAV-cmet–transduced one (Figs. 5, 6), in 6.1 ± 0.7 days on average (P < 0.001 vs. rAV-cmet–treated or normal). Staining for proliferating cells with Ki-67 or for apoptotic cells with activated caspase-3 antibodies did not reveal any apparent change in both parameters in rAV-cmet– versus rAV-vector–transduced corneas (data not shown here). 

Figure 5.

 

Dynamics of wound healing. A typical course of healing is presented for a c-met–transduced and vector-transduced cornea. rAV-vector–transduced diabetic cornea (top row) healed in 7 days, whereas rAV-met–transduced fellow cornea (bottom row) healed in 3 days. Pictures of live healing corneas are shown. Dashed line: shows the boundaries of the nonhealed wound region. W, wound zone; E, migrating epithelium.

Figure 6.

 

c-Met overexpression led to a significant decrease in corneal epithelial wound healing time. The average healing time of rAV-vector–transduced corneas was 6.1 days, whereas rAV-cmet–transduced corneas healed in 3.1 days on average. For comparison, previous data on normal corneas ¹¹ are also shown. They healed in 2.3 days on average. Note that vector-treated diabetic corneas healed significantly slower than normal. c-Met transduction led to a significant decrease in healing time, bringing it close to normal. Statistical analysis of the time to complete healing (n = 7 for vector and c-met, and n = 13 for normal). Significance was determined using ANOVA with the Bonferroni post test.

Acceleration of corneal epithelial wound healing on c-met transduction was most likely due to c-met activity, as a highly specific small molecule inhibitor of c-met kinase, PHA 665752, completely abrogated this effect (data not shown here). In the presence of this compound, the corneal epithelium healed even slower than in the control, possibly because the inhibitor blocked both the rAV-transduced and endogenous c-met. 

Effect of rAV-cmet on the Expression of Major Signaling Intermediates

It has been established that the most important downstream mediators of c-met signaling for cell proliferation, migration, and survival included kinases Akt, p38 MAPK, and ERK (extracellular signal-regulated kinase). ^21,45–48 Immunofluorescent staining with specific antibodies did not reveal consistent changes in the patterns of phosphorylated (activated) Akt and ERK on c-met gene transduction, both during wound healing and after it was complete (Fig. 7, middle and bottom rows). However, staining for phosphorylated p38 MAPK (Thr180/Tyr182) was noticeably increased in 8 of 12 cases after c-met overexpression (Fig. 7, top row). P38 MAPK antibody predominantly stained the epithelial cell nuclei, consistent with previous results in rabbits. ⁴⁶ Data on increased p38 activation after c-met transduction were corroborated by phosphoproteomic MAPK arrays (Fig. 8A; 2.1 ± 0.4 times increase in p-p38α expression, n = 6, P < 0.05 vs. vector), which were additionally validated by Western blot analysis with a specific antibody to phosphorylated p38 MAPK (Fig. 8B). 

Figure 7.

 

c-Met overexpression leads to increased phosphorylation of p38 MAP kinase. Top row: increased epithelial staining for the phosphorylated form of p38 (p-p38). Staining for phosphorylated forms of Akt (p-Akt, middle row) or ERK (p-ERK, bottom row) kinases did not show noticeable change on c-met transduction. Indirect immunofluorescence of corneal sections. e, epithelium; s, stroma.

Figure 8.

 

Phosphoproteomic array analysis of MAPKs. (A) On c-met transduction, phosphorylated p38α (p-p38α) was significantly (2.1-fold, n = 6, P < 0.05) increased compared with vector transduction. Array membranes were probed with total corneal protein, and signal was detected with chemiluminescence. (B) Validation of array results by Western blot analysis. Two separate cases are shown; in both of them p-p38 was increased on c-met transduction compared with vector alone. Mean increase in c-met versus vector was 1.7 ± 0.2 (n = 5, P < 0.04). Gel loading was normalized by β-actin. M, markers in kilodaltons.

Activation of p38 MAPK was reported to stimulate corneal epithelial migration. ^29,30,47,49 To test whether it may also be involved in c-met effects on diabetic corneal wound healing, we performed p38 MAPK inhibitor studies. In the first setting, corneal wound healing on rAV-cmet transduction proceeded in the presence of specific p38 MAPK inhibitor SB 202190 (mostly active against p38α and p38β ⁵⁰ ), whereas rAV-vector transduced fellow cornea healed in the presence of inactive analogue SB 202474 (control). As shown in Figure 9 (top rows), p38 MAPK inhibitor resulted in markedly slower wound healing than in the control cornea. Similar results were obtained when both fellow corneas were transduced with rAV-cmet. In the second setting, one cornea received the inhibitor during wound healing, and the fellow one received the inactive analogue (control). The inhibitor again slowed down the healing, so that the control cornea healed faster (Fig. 9, bottom rows). Overall, analogue-treated corneas healed on average at 3.2 ± 0.4 days, whereas p38 inhibitor-treated ones, at 5.7 ± 0.9 days (P < 0.05). These data suggest that p38 MAPK activation may mediate c-met-dependent acceleration of corneal epithelial wound healing. 

Figure 9.

 

Dynamics of wound healing on treatment with p38 inhibitor. (A, top row) rAV-vector–transduced diabetic cornea completely healed in 4 days in the presence of 10 μM of the inactive p38 inhibitor analogue SB 202474. Bottom row: rAV-met transduced fellow cornea healed in 6 days in the presence of 10 μM of the specific p38 inhibitor SB 202190. (B, top row) rAV-cmet–transduced diabetic cornea healed in 3 days in the presence of 10 μM SB 202474. Bottom row: rAV-met–transduced fellow cornea healed in 7 days in the presence of 10 μM SB 202190. Pictures of live corneas are shown. Analogue-treated corneas healed on average at 3.2 ± 0.4 days, compared with 5.7 ± 0.9 days for inhibitor-treated corneas (P < 0.05). p38 inhibitor thus abrogates c-met–induced acceleration of epithelial wound healing. Dashed line: shows the boundaries of the nonhealed wound region. W, wound zone; E, migrating epithelium.

Discussion

Diabetic corneal disease is a frequent ocular manifestation of diabetes mellitus, ^1,2 but its pathophysiology is not fully understood. The hallmarks of this disorder include epitheliopathy (keratopathy) and neuropathy. ^1,51 Despite wide occurrence, the diabetic corneal disease treatment remains symptomatic. ^52,53 Experimental therapies, including aldose reductase inhibitors, thymosin β₄, insulin, and the opioid growth factor antagonist naltrexone, have been tested only in animals or early-phase clinical trials. ^54–56 The need for a specific and effective therapy for diabetic corneal alterations may be filled with gene therapy. Because of its transparency, ready availability for topical treatment, and immune privilege status, the cornea is an ideal tissue for viral and nonviral gene therapy. ^57,58  

To implement the gene therapy approach, one must target specific disease markers. Our previous studies have identified several BM, growth factors, and proteinase markers in human diabetic corneas. ^5,12–14 The HGF/c-met system is of particular interest among these markers because of its role in cell survival, cell migration, and wound healing. In the cornea, its mode of action is probably paracrine, with the epithelium expressing c-met, and the keratocytes, HGF. ^40,59 However, during epithelial wound healing (physiological or surgery-induced), anterior stromal keratocytes die, ⁶⁰ and HGF could be supplied by tears. ^39,61–63 It can also act in an autocrine manner, ⁶³ which is in line with immunostaining revealing not only c-met but also HGF in corneal epithelium. ¹⁴ In the diabetic corneas, both ex vivo and organ cultured, HGF expression is increased, but c-met expression is depressed. ¹⁴ This effect suggests decreased signaling and activity of this system, as well as a possibility of restoring it by upregulating c-met. We further hypothesized that, because HGF/c-met system has been implicated in cell migration and wound healing, its alteration in the diabetic cornea could contribute to poor epithelial healing, which is characteristic of these corneas. ^9,11,15 We also looked at the expression of previously established BM and integrin markers after overexpression of c-met driven by an adenoviral vector able to transduce corneal epithelium. ⁶⁴  

As expected, rAV-cmet distinctly increased HGF receptor expression and phosphorylation/activity in the corneal epithelium of diabetic organ cultured human corneas. As a result, we also observed a substantial effect of c-met expression in restoring patterns of specific markers (laminin, nidogens, and integrin α₃β₁). This effect could be due to either HGF-mediated amelioration of cell survival ⁴⁷ or to decreased expression of proteinases upregulated in diabetic corneas. ^12,14 In fact, at least in some rAV-cmet–treated corneas, decreased expression of MMP-10 and cathepsin F was noted (data not shown). 

Another important effect of c-met overexpression in diabetic corneas was restoration of nearly normal epithelial wound healing times (Figs. 5, 6). Because the HGF/c-met system is a potent mediator of cell migration that is necessary for wound healing, the return of normal wound healing is another indication that HGF signaling was restored in rAV-cmet–transduced corneas. The effect appeared to be specific for c-met because an inhibitor of this kinase abrogated wound healing amelioration. Overexpression of one protein, c-met, was thus able to markedly affect a complex process such as wound healing. This effect may be explained by the interactions of HGF signaling with pathways triggered by other growth factors in a cross-talk mechanism. Such a mechanism has been described for HGF and epidermal growth factor (EGF) and its receptor (EGFR). HGF has been shown to transactivate EGFR for the induction of full motility in different cell cultures. ^65,66 It should be noted, however, that insulin can also induce EGFR phosphorylation, thereby promoting corneal epithelial cell migration. ⁶⁷ At the same time, diabetic patients are often treated with insulin and still have delayed wound healing. In our organ cultures also maintained with insulin c-met caused significant wound healing improvement. These results suggest that cross-talk between EGFR and c-met may not play a major role in the observed effects. On the other hand, c-met could cooperate with receptors of other growth factors, such as IGF-I, ⁶⁸ which is also elevated in diabetic corneas, ¹³ vascular endothelial growth factor, ⁶⁹ and keratinocyte growth factor. ^46,70 Alternatively, a strong action of c-met on corneal wound healing could be explained by a shift in the balance of phosphatases positively and negatively regulating its activity, such as SHP-2 and PTP1B. ⁷¹ These possibilities deserve further investigation. 

Phosphokinase arrays, immunostaining, Western blot analysis, and inhibitor studies suggest that activation of p38 MAPK could mediate the effects of overexpressed c-met on wound healing in diabetic corneas. The phosphorylated form of this kinase was increased in the epithelium on rAV-cmet transduction and on phosphokinase arrays. By phosphokinase arrays, this increase concerned p38α, the most abundant isoform in many cell types. ⁷² An increase in other isoforms (β, γ, and δ) cannot be ruled out, but their relative levels were probably too low for their changes to be reliably detected by the array analysis. A specific inhibitor of p38 MAPK (mostly active against the α and β isoforms) abrogated c-met–mediated acceleration of epithelial wound healing in two different settings, similar to previous results. ^29,46,49 It should be mentioned that other signaling intermediates have been implicated in HGF-induced c-met–mediated cell migration including Akt, protein kinase C (PKC), and ERK. ^21,71,73,74 However, all these studies were performed in the presence of exogenous HGF. Moreover, activation of Akt in corneal wound healing may be attributable to the action of the EGF/EGFR system, ⁷⁵ which may cross-talk with c-met. Overall, the presented data do not exclude the possibility of a role of Akt, PKC, and ERK in the c-met–mediated effect on wound healing in diabetic cornea. At the same time, our results suggest a major role of p38 MAPK in this process. This conclusion is in good agreement with previous data identifying p38 MAPK as the main cell migration and wound healing mediator in rodent corneal epithelial cultures and in mouse and rabbit corneal wounds. ^29,46,49  

Recombinant adenoviral vectors may have a potential for successful gene therapy in the intact cornea with overexpression of target proteins in the epithelial layer. It should be noted that the transient nature of transgene expression when using rAV may be a deterrent to be used in therapy for inherited gene defects. However, the main problem in the diabetic corneas remains delayed and incomplete wound healing. Therefore, our system may be enough to positively influence wound healing, and then the transgene expression could be turned off after a month because the wound would have healed. In this respect, the rAV-driven c-met transduction may be useful for clinical purposes as well. 

In summary, this study confirms the role of the HGF receptor c-met in normalizing the expression of diabetic corneal markers and epithelial wound healing, which are compromised in the corneas in diabetes. We have thus established an efficient procedure for transferring a target gene into the cornea and have demonstrated the therapeutic efficacy of this approach using the human corneal organ culture model. Future studies might focus on a similar gene therapy, specifically for corneal epithelial stem cells, for an increased duration of the effect. The ability to genetically transduce corneal epithelial cells provides a useful strategy for understanding the molecular defects in diabetic corneas. It also offers a potential approach for the development of novel therapeutic strategies for corneal diabetic disease with a perspective for future clinical implementation. 

Footnotes

 Presented at the annual meetings of the Association for Research in Vision and Ophthalmology, Fort Lauderdale, Florida, April 2008 and May 2009; at the 68th Scientific Sessions of the American Diabetes Association, San Francisco, California, June 2008; and at the XVIIIth International Congress for Eye Research, Beijing, China, September 2008.

Footnotes

 Supported by National Eye Institute (NEI) Grants R01 EY13431 (MS, AVL, AAK) and R01 EY10869 (FXY) and National Institutes of Health Research Resources Grant M01 RR00425; a Winnick Family Foundation Clinical Research Scholar award (AVL); an Alcon Travel Grant (AVL); the Eye Defects Research Foundation (AAK, AVL); an NEI Travel Award (MS); National Institute of Neurological Disorders and Stroke Grants U01 NS052465, R01 NS057711, and R21 NS054143; and the Cedars-Sinai Board of Governors (MGC). MGC holds the Medallions Group Endowed Chair in Gene Therapeutics Research.

Footnotes

 Disclosure: M. Saghizadeh, None; A.A. Kramerov, None; F.X. Yu, None; M.G. Castro, None; A.V. Ljubimov, None

The authors are indebted to the late Rupert Timpl (The Max Planck Institute of Biochemistry, Martinsried, Germany) for his kind donation of antibodies to nidogen-2.