HGF Protects Corneal Epithelial Cells from Apoptosis by the PI-3K/Akt-1/Bad- but Not the ERK1/2-Mediated Signaling Pathway

Abstract

purpose. Cell survival is critical during corneal epithelial regeneration after injury, and growth factors could be fundamental in cytoprotection. The goal of this study was to investigate the involvement of the paracrine hepatocyte growth factor (HGF) in the prevention of corneal epithelial cell apoptosis and to identify signal transducers in this process.

methods. Apoptosis in human and rabbit corneal epithelial (HCE and RCE) cells was induced with a nutrient-deprived exhausted medium (ExM) or by treatment with staurosporine (20–100 ng/mL) or the calcium ionophore A23187 (0.5 μM). Apoptotic cells were identified by DNA fragmentation in agarose gels and by Hoechst staining. Active Akt-1 overexpression (Akt-1 pUSEamp cDNA) and small interfering RNA (siRNA) specific for Akt mRNA were used. Immunofluorescence, Western immunoblot analysis, and Akt kinase assays were also used.

results. Staurosporine, ExM, and A23187 induced DNA fragmentation in HCE and RCE cells. HGF (20 ng/mL) in combination with the apoptotic agents greatly reduced DNA breakdown and the number of Hoechst-positive cells. In the presence of phosphatidylinositol-3 kinase (PI-3K) inhibitors (wortmannin and LY294002), HGF did not overcome apoptosis. However, PD98059, the ERK1/2 cascade pathway inhibitor, was ineffective in preventing HGF protection. HGF induced a sustained activation of Akt-1, and overexpression of active Akt-1 reduced apoptosis. HGF stimulated the downstream targets of Akt, glycogen synthase kinase (GSK-3), and Bad, a proapoptotic member of the Bcl-2 family, an effect that was blocked by PI-3K inhibitors but not by ERK1/2 inhibition. Suppressing the expression of Akt by Akt siRNA led to a decrease in the phosphorylation of Bad and GSK-3. Translocation of Bad to the mitochondria, a critical stage in apoptosis, was prevented by HGF when apoptosis was induced. Moreover, in epithelial cells overexpressing active Akt-1, Bad translocation was also prevented.

conclusions. HGF modulates multiple signaling cascades in corneal epithelial cells. The results demonstrated that HGF, in a paracrine fashion, protects cells from apoptosis through a PI-3K/Akt/Bad pathway but not through an ERK1/2 pathway. It was also demonstrated that GSK-3 is a target of PI-3K/Akt-1.

Hepatocyte growth factor (HGF), an important mediator in corneal epithelial–stromal interactions, is expressed in stromal cells and in the lacrimal gland, whereas its receptors are localized in the epithelial cells. ^{1 2} The ligation of HGF to its receptor, c-Met, produces activation and recruitment of multiple signaling molecules that are probably involved in diverse responses. After corneal injury, the expression of HGF is upregulated, ^{2 3 4} and this cytokine increases the rate of corneal epithelial wound healing ^{5 6} by mechanisms involving signal-transduction pathways that engage in cross-talk. We recently reported that, whereas the mitogen-activated kinase (MAPK) member p38 is involved in the migration of epithelial cells that cover the wound when stimulated by HGF, the ERK1/2 system is activated in those epithelial cells that proliferate. ⁶ HGF also stimulates phosphatidylinositol-3 kinase (PI-3K), ⁵ an enzyme that increases in expression during corneal wound healing. ⁷ PI-3K is recruited to the c-Met receptor on the inner surface of the plasma membrane to catalyze the transfer of phosphate from adenosine triphosphate (ATP) to the D-3 position of the inositol ring of membrane-localized phosphoinositides, generating 3′-phosphorylated phosphoinositides. These lipids then function as signaling intermediates that regulate downstream signal-transduction cascades. A downstream target of PI-3K activation in corneal epithelium is the 70-kDa S6 ribosomal protein kinase (p70S6K). ⁵ Because of the importance of this kinase in proliferation and cell-cycle progression, we suggested that the PI-3K/p70S6K pathway is linked to the proliferative phase of the wound-healing response elicited by HGF. ⁵  

Another downstream target of PI-3K is Akt/protein kinase B. In mammals, three closely related isoforms of Akt have been found: Akt-1, -2, and -3. These serine/threonine protein kinases are major inhibitors of apoptosis (for review see Ref. ⁸ ). 

Unlike corneal stroma keratocytes, epithelial cells are resistant to apoptosis, and previous studies have suggested that only a small percentage of the cells lost from the surface of the cornea (shedding) enter apoptosis. ⁹ In fact, we found that the inflammatory mediator platelet-activating factor (PAF) induces apoptosis in corneal keratocytes but not in epithelial cells. ¹⁰ Furthermore, PAF enhances apoptosis in epithelial cells only after the cornea is exposed to a previous stressor, such as UV radiation. ¹¹ Therefore, it is possible that growth factors confer cytoprotection to epithelial cells. Cell proliferation and survival are important processes that must be triggered so that re-epithelialization occurs after injury. Upregulation of HGF expression after corneal damage may not only increase epithelial migration and proliferation, ^{5 6} but may also enhance cell survival by activating antiapoptosis-signaling cascades that facilitate wound healing. 

This hypothesis prompted us to investigate the ability of HGF to suppress apoptosis in corneal epithelial cells. We found that HGF inhibited apoptosis induced by three different agents in corneal epithelial cells, and activated a PI-3K/Akt-1/Bad signaling pathway that promoted cell survival. Furthermore, overexpression and silencing mRNA experiments conducted in this study established for the first time the central role of HGF-mediated Akt activation in the antiapoptotic process in corneal epithelium. 

Methods

Materials

Human recombinant double-chain HGF was a gift from Genentech (San Francisco, CA). The PI-3K inhibitors wortmannin and LY294002 and the MEK (MAPK-K) inhibitor PD98059 were from Calbiochem (San Diego, CA); staurosporine and calcium ionophore A23187 from Sigma-Aldrich (St. Louis, MO). The mouse monoclonal antibodies Akt-1 and IκB, the phosphorylated form of IκB at Ser-32 (p-IκB), the rabbit polyclonal antibodies against Bad, the phosphorylated form of Bad at Ser-136 (p-Bad), the goat polyclonal antibody against voltage-dependent anion-selective channel-1 (VDAC1), and the secondary anti-goat horseradish peroxidase (HRP) antibody were purchased from Santa Cruz Biotechnology (Santa Cruz, CA). The rabbit polyclonal Ser-473 phosphorylated form of Akt (p-Akt), and the mouse monoclonal c-Myc antibodies, the Akt-1 cDNA expression (activated) in pUSEamp, and its empty expression vector were purchased from Upstate (Charlottesville, VA). The transfection reagent (FuGENE 6) was from Roche Diagnostics Corp. (Indianapolis, IN). The phosphorylated form of GSK-3α and -β at Ser-21/9, the Akt siRNA kit, the rabbit polyclonal p-Akt (Ser-437) antibody, and the p-Akt blocking peptide for immunocytochemistry were from Cell Signaling Technology (Beverly, MA). Anti-mouse and anti-rabbit HRP secondary antibodies were from BD-PharMingen (San Diego, CA). The secondary antibody goat anti-rabbit IgG FITC conjugate was from Jackson ImmunoResearch Laboratories (West Grove, PA). The Genomic DNA-purification kit was purchased from Promega (Madison, WI). The Western blot system was obtained from Bio-Rad (Hercules, CA). 

Cell Culture

Rabbit and human corneal epithelial (RCE and HCE, respectively) cells were used in this study. Rabbit eyes were obtained from Pel-Freeze Biologicals (Rogers, AR) and RCE cells were prepared as previously described. ¹² First-passage cells from primary cultures were used in the experiments. The HCE cell line was kindly provided by Roger Beuerman (LSU Eye Center, New Orleans, LA) and maintained as previously described. ⁶ Cells at passages between 35 and 50 were used for the experiments. When the cells reached 90% to 95% confluence, RCE cells were starved overnight in DMEM/F12 and HCE cells were maintained in keratinocyte basal medium (KBM; Clonetics, San Diego, CA). Cells were then stimulated with 20 ng/mL HGF. In some experiments, the inhibitor wortmannin (100 nM), LY294002 (10 μM), or PD98059 (10 μM) was added before HGF. 

Induction of Apoptosis

RCE or HCE cells were incubated overnight in DMEM/F12 or KBM, respectively, and treated with staurosporine (20–100 ng/mL) or the calcium ionophore A23187 (0.5 μM) for 7 hours (RCE) or 20 hours (HCE). RCE cells were also incubated for 24 hours with a nutrient-deprived exhausted medium (ExM) obtained from confluent RCE cultures incubated in the same medium for 7 days. ¹³ HGF, wortmannin, LY294002, and PD98059 were added according to the experimental design. 

Hoechst Staining

After treatment, cell cultures were washed twice with phosphate-buffered saline (PBS) and incubated with 2 μM Hoechst 33258 (Molecular Probes, Eugene, OR) for 1 hour at 37°C in the dark. After three washes with PBS, the cells were viewed with a fluorescence microscope (Nikon, Tokyo, Japan) equipped with a UV filter. The images were recorded on computer with a digital camera (DXM 1200; Nikon) attached to the microscope, and the images were processed by computer (MetaVue; Universal Imaging Co., Downingtown, PA). The Hoechst reagent is taken up by the nuclei of the cells, and apoptotic cells exhibit a bright blue fluorescence. Cells with nuclear condensation versus normal-appearing cells were counted in six different fields. At least 300 cells were counted in each treatment group. The number of apoptotic cells is expressed as a percentage of total cells counted. 

DNA Laddering

After the indicated treatments, RCE or HCE adherent and floating cells were harvested and washed twice in PBS. DNA was extracted with a genomic DNA-purification kit according to the manufacturer’s protocol. DNA (4 μg) was separated by electrophoresis on a 1.8% agarose gel containing ethidium bromide in TBE buffer (89 mM Tris-borate, 89 mM boric acid, 2 mM EDTA, pH 8.3). The fragmentation pattern was visualized under UV light. A 100 base-pair DNA ladder was used as a marker. 

Western Blot Analysis

Cells or mitochondrial fractions (isolated as described later) were homogenized in 50 mM Tris-HCl (pH 7.5), 1 mM EDTA, 1 mM EGTA, 0.5 mM sodium orthovanadate, 1 mM dithiothreitol (DTT), 1% Triton X-100, 5 mM sodium fluoride, 1 mM sodium pyrophosphate, 150 mM NaCl, 10 mM sodium β-glycerophosphate, 1 mM phenylmethylsulfonyl fluoride (PMSF), 1 μg/mL aprotinin and leupeptin, and 1 μM microcystin (lysis buffer). The homogenate was centrifuged at 12,000 rpm for 15 minutes, and total protein was determined in the supernatant by the Bradford method with a protein assay reagent (Bio-Rad). All procedures were performed at 4°C. Samples were resolved by SDS-polyacrylamide gel electrophoresis (9%–12% gel) and transferred to polyvinylidene difluoride (PVDF) membranes (Amersham Pharmacia Biotech, Piscataway, NJ). Biotinylated protein molecular weight standards were applied in one lane of each gel. The nonspecific proteins were blocked with 5% nonfat milk in Tris-buffered saline (TBS, 20 mM Tris-HCl, 150 mM NaCl [pH 7.6]) plus 0.1% Tween-20 for 1 hour and then probed with various primary antibodies, as described in the experiments for 2 hours at room temperature or overnight at 4°C. The membranes were washed six times with TBS plus 0.1% Tween-20 and further incubated with HRP-conjugated secondary antibodies. Protein bands were visualized using chemiluminescence detection reagents (ECL Plus; Amersham) and exposed to autoradiograph film. 

Isolation of Mitochondrial Fractions

Transfected (as explained later) or nontransfected HCE cells were starved overnight and then treated with 100 ng/mL staurosporine for 6 hours. The cells were resuspended in 20 mM HEPES buffer [pH 7.4] containing 10 mM KCl; 1.5 mM MgCl₂; 1 mM each of EDTA, EGTA, DTT, and PMSF; 25 μg/mL leupeptin; and 2 μg/mL aprotinin. Nuclei and broken cells were separated by centrifugation at 1500g for 10 minutes. The supernatant was further centrifuged at 12,000g for 30 minutes to collect the pellet. The pellet was washed with HEPES buffer and resuspended in the lysis buffer: this constituted the mitochondrial fraction. Cytosolic fractions were obtained after the mitochondrial supernatant was centrifuged at 100,000g for 30 minutes. 

Akt-1 Activity Assay

The assay was performed with a kinase assay kit from Upstate. RCE cells were homogenized in lysis buffer and 500 μg of protein was immunoprecipitated by incubating with the complex Akt-1 polyclonal antibody-protein G agarose for 2 hours at 4°C. The immunocomplex was washed three times with lysis buffer and then incubated with 100 μM of the substrate peptide (RPRAATF), 10 μM PKA inhibitor peptide, 18.75 mM MgCl₂, 125 μM ATP, and 10 μCi [γ-³³P]ATP (Amersham) for 10 minutes at 30°C in a total volume of 40 μL. The phosphorylated substrate was spotted on phosphocellulose paper (P81; Upstate), washed three times with 0.75% phosphoric acid, and quantified by liquid scintillation counting. ¹⁴  

Akt cDNA Transfection

The Akt-1 cDNA in a pUSEamp construct contained an src myristoylation signal sequence (Myr-Akt-1) added to its amino terminus that allowed membrane localization and caused activation without the need for elevated PI-3K activity. The construct contained a C-terminal c-Myc-His epitope tag. ¹⁵ The Akt construct was subcloned in HB101 competent cells (Invitrogen, Carlsbad, CA). The crude plasmid was purified by centrifugation in a CsCl-ethidium bromide gradient. The DNA was cut with the restriction enzyme XbaI, which generated the expected fragments of 1.39 and 5.5 kbp, and was probed by agarose gel electrophoresis. 

RCE or HCE cells were seeded into six-well culture plates (2 ×10⁵ cells) and allowed to grow to 60% to 70% confluence in 2 mL culture medium. Transfections were performed (FuGENE 6 reagent; Roche), and Akt-1 pUSEamp cDNA (microliters per microgram cDNA) was added to the cultures in a ratio of 3:1 and incubated overnight. The DNA-containing medium was removed and the cells were incubated in the same medium without DNA and the transfection reagent for 24 hours, to allow the expression of the activated form of Akt-1. Transient transfections were determined by immunoblot analysis with c-Myc and p-Akt antibodies. 

Immunofluorescence Staining

HCE cells (3 × 10⁴ cells) were seeded in each well of four-well glass slides. The cells were transfected with Akt-1 cDNA, as just described or maintained in KBM for 48 hours and stimulated with 20 ng/mL HGF for 2, 5, or 30 minutes. The cells were fixed with 4% paraformaldehyde for 30 minutes, then blocked with 10% normal goat serum and 1% BSA in PBS, and incubated with anti-p-Akt antibody overnight at 4°C. To evaluate the specificity of the immunostaining, the antibody was incubated with p-Akt blocking peptide for 2 hours at 4°C before addition to the slides. The cells were incubated with goat anti-rabbit IgG FITC-conjugated secondary antibody for 1 hour at room temperature. After each step, the slides were washed three times with PBS. Hoechst staining was performed to localize the nuclei after the immunostaining. Cells were examined with a fluorescence microscope, and images were recorded with a digital camera. 

Akt siRNA Transfection

HCE or RCE cells (2 × 10⁵) were seeded in six-well plates and allowed to grow to 70% confluence. Transient transfections were performed with Akt siRNA according to the manufacturer’s instructions. Briefly, 4 μL transfection reagent designed specifically for highly efficient siRNA delivery (Mirus, Madison, WI) was diluted in 200 μL of KBM or DMEM/F12 and incubated for 5 minutes at room temperature. Six microliters of the Akt siRNA (final concentration, 50 nM) was added to the mix and incubated for an additional 5 minutes. For transfection, the complex was added to the cells in KGM (HCE cells) or complete medium (RCE cells) and cultured for 24 hours. Afterward, the medium was changed to KGM or complete medium, and the cells were incubated for another 24 hours. The siRNA-transfected cells were starved overnight and then stimulated with HGF for the times specified in the experiments. Fluorescein-conjugated nonsilencing siRNA was used to define the transfection efficiency (monitored by fluorescence microscopy) and as a negative control. 

Results

HGF Protection of Corneal Epithelial Cells from Apoptosis

Apoptosis was produced in RCE/HCE cells by using three different inducers. Treatment with 20 ng/mL staurosporine or 0.5 μM A23187 for 7 hours in RCE cells produced DNA laddering and intense Hoechst-positive staining for condensed nuclei (Figs. 1A 1B 1C) , indicative of apoptosis. Cells treated with ExM overnight also showed DNA fragmentation and Hoechst staining (Figs. 1D 2B) . Significant decreases in DNA laddering (Figs. 1 2) and Hoechst staining were observed when cells were pretreated with 20 ng/mL HGF and the three apoptosis inducers. In earlier studies, this physiologic concentration of the growth factor produced activation of PI-3K signaling, as well as increases in proliferation and migration of corneal epithelial cells. ^{5 6} A higher concentration of 100 ng/mL staurosporine and a longer exposure time (20 hours) were needed to obtain DNA laddering in HCE cells. Under these conditions, cells treated with staurosporine in the presence of HGF were also protected (Fig. 1A) . The results indicate that HGF protected cells from apoptosis induced by multiple agents. 

Signal Dependency of HGF Protection of Corneal Epithelial Cells from Apoptosis

Previous studies have shown that HGF stimulates PI-3K as well as the p42/44 MAP-kinase (ERK1/2) in RCE cells. ^{5 6 16} To investigate the mechanisms involved in the protective effect of HGF against apoptosis, we incubated the cells in the presence of different kinase inhibitors. Pretreatment with the PI-3K inhibitors LY294002 or wortmannin abolished the protective effect of HGF in cells exposed to staurosporine or ExM (Fig. 2) . Moreover, LY294002 at a concentration of 10 μM enhanced the DNA laddering induced by staurosporine (Fig. 2A) , possibly by inhibition of basal PI-3K activity. ⁷ However, PD98059 in concentrations that block the activation of ERK1/2 ⁶ did not interfere with the protective effect of HGF. Similar results were obtained with HCE cells (data not shown). Analysis of the percentage of apoptotic cells by Hoechst staining showed that ExM induced apoptosis in approximately 50% of the cells, a percentage that was reduced to ∼15% in the presence of HGF. Both PI-3K inhibitors completely blocked the HGF protective effect, whereas ERK1/2 inhibition had no effect (Fig. 2C) . 

Activation and Inhibition of Akt-1

One of the main targets of PI-3K is the serine-threonine kinase Akt, ^{17 18} which has been reported to mediate cell survival in a wide range of cell types (for review see Refs. ⁸ , ¹⁹ ). We investigated whether HGF activates Akt-1 in corneal epithelial cells. An Akt kinase assay (Fig. 3A) showed that there was stimulation of Akt activity in RCE cells after 2 minutes of HGF treatment. By 5 minutes there was a 25-fold stimulation, and the activity continuously increased (to 38-fold) after 30 minutes of HGF treatment. This potent stimulation suggests the central role of Akt-1 in survival signaling by HGF in corneal epithelial cells. Activation of Akt depends on the phosphorylation of the molecule at Thr-308 in the catalytic domain and Ser-473 in the C terminus. ¹⁹ We determined the activation of Akt phosphorylated at Ser-473. HGF induced a significant phosphorylation of Akt that persisted for at least 1 hour, whereas total Akt-1 did not change (Fig. 3B) . Immunostaining of HCE cells with a polyclonal anti-p-Akt-1 produced a punctate green fluorescence localized in the cytosol surrounding the nuclei, which possibly included the Golgi apparatus and/or mitochondria ²⁰ in cells stimulated with HGF for 2 to 30 minutes (Fig. 3C) . To confirm the specificity of the immunostaining, we inhibited the antibody with a blocking peptide and observed no immunofluorescence (data not shown). Pretreatment with LY294002 markedly reduced the phosphorylation of Akt-1 in response to HGF, whereas inhibition of ERK1/2 with PD98059 had no effect (Fig. 3D) , suggesting that PI-3K but not ERK1/2 signaling is critical for HGF-stimulated Akt-1 phosphorylation. 

Involvement of Akt-1 in the Survival Effect of HGF

To demonstrate the role of Akt in the protective action of HGF against apoptosis in corneal epithelial cells, two different approaches were used. First, the activity of Akt-1 was assayed in RCE cells treated with ExM in the presence or absence of HGF (Fig. 4A) . Akt-1 activity decreased 50% when cells were treated with ExM, and the presence of HGF induced a fivefold increase in Akt-1 activity. Furthermore, the Akt-1 activity with HGF was 2.9 times higher than in control cells. Hoechst staining confirmed the apoptosis induced by ExM and the protective effect of HGF (data not shown, similar to Fig. 2B ). Second, the relevance of Akt-1 to the protection of RCE cells against apoptosis was investigated by transiently transfecting RCE cells with Akt-1 cDNA lacking the pleckstrin homology domain (PH) but containing an src myristoylation signal sequence fused to the c-Akt-coding sequence (Myr Akt) that targets the protein to the membrane and maintains the enzyme in its active form ¹⁵ (Fig. 4B) . In addition, Western-blot analysis with monoclonal antibodies against the Myc tag (or p-Akt-1, not shown) confirmed the presence of activated Akt-1 in the transfected cells (Fig. 4C) . Transfection with an empty vector did not produce expression of c-Myc. The transfection efficiency of the cells (50%–60%) was assessed with a cDNA-green fluorescent protein (GFP) control vector (pcMV-EGFP, kindly provided by Cindy Morris, Tulane University, New Orleans, LA) by expression of GFP. Overexpression of the activated form of Akt-1 decreased DNA laddering in RCE cells treated with staurosporine (Fig. 4D) in a manner similar to the reduction caused by HGF. 

Anti-apoptotic Targets of PI-3K/Akt Signaling

Several proteins have been reported to serve as substrates for Akt, which indicates the critical role of this kinase in cell signaling. ¹⁹ Activation of PI-3K/Akt may trigger the expression of survival genes such as nuclear factor κB (NFκB). This factor exists in the cytoplasm as an inactive complex bound to IκB proteins. Its activation involves the phosphorylation of IκB by an IκB kinase (IKK), which releases the complex and permits the translocation of NFκB to the nucleus, where it exerts its action. Akt induces IKK activation. ^{21 22} To determine whether this pathway is stimulated by HGF in corneal epithelial cells, RCE and HCE cells were treated with 20 ng/mL HGF for 5 to 60 minutes and phosphorylation of IκB at Ser-32 was analyzed. Under these conditions, we did not detect increased phosphorylation of IκB in corneal epithelial cells (data not shown). 

The first substrate identified for Akt was glycogen synthase kinase (GSK)-3. ²³ This enzyme was named for its ability to phosphorylate, then inactivate, glycogen synthase, a key regulatory enzyme in the synthesis of glycogen. More recently, it has been found to play major roles in processes related to development, gene expression, the cell cycle, and apoptosis. ^{24 25} GSK-3 negatively regulates downstream signaling mechanisms, and its inactivation by phosphorylation prevents its action. HGF induced a strong phosphorylation of two of the isoenzymes, GSK-3α at Ser-21 and GSK-3β at Ser-9, in RCE cells as early as 5 minutes after stimulation, which was sustained for up to 60 minutes (Fig. 5A) . Pretreatment with the PI-3K inhibitors LY294002 and wortmannin partially blocked the HGF-induced phosphorylation of both GSK-3α and -β. In contrast, the ERK1/2 inhibitor PD98059 had no effect (Fig. 5B) . 

Another downstream target of Akt is the proapoptotic member of the Bcl-2 family, Bad. This protein in the nonphosphorylated form complexes in the mitochondria with the antiapoptotic family members Bcl-2 or Bcl-XL, inhibiting their function and inducing apoptosis. ^{26 27} Phosphorylation of Bad at Ser-136 facilitates its dissociation from Bcl-XL or Bcl-2 and allows these proteins to exert a protective effect. For these experiments, HCE cells were used because they react with the Bad antibodies. When the cells were stimulated with HGF and analyzed by Western blot with antibody against p-Bad at Ser-136, there was increased phosphorylation within 5 minutes after stimulation that was maintained for up to 60 minutes (Fig. 5C) . No changes in total Bad were found under these conditions. The phosphorylation was inhibited by preincubation with LY294002 but not with PD98059 (Fig. 5D) . 

To prove further that Bad is an important target of Akt in the survival effect of HGF, epithelial cells were incubated in the presence or absence of HGF before induction of apoptosis with staurosporine. Some cells were also transfected with active Akt-1, then treated with staurosporine. Mitochondrial and cytosolic fractions were isolated, as described in the Methods section, and translocation of Bad was analyzed. Staurosporine produced translocation of Bad protein from the cytosol to the mitochondria (Fig. 6) . The presence of HGF effectively decreased the amount of Bad translocated to the mitochondria. A similar effect was obtained in Akt-1-transfected epithelial cells. Levels of VDAC, used as a mitochondrial marker, were similar in all the conditions, and no VDAC was detected in the cytosolic fraction. 

To demonstrate the direct effect of Akt on Bad and GSK-3 in corneal epithelial cells, Akt was downregulated by transfecting RCE and HCE cells with siRNA. The efficiency of the transfection was monitored by fluorescence microscopy and was between 80% and 90%. As shown in Figure 7A , transfection with an Akt-specific siRNA, but not with the nonsilencing control siRNA, resulted in an approximate 50% decrease in Akt-1 expression in HCE cells, as determined by Western blot analysis and quantification by densitometry of the bands. ⁷ A similar inhibition was found in RCE cells. When RCE cells were transfected with Akt siRNA and stimulated with HGF for 15 minutes, there was an approximate 43% inhibition in the phosphorylation of GSK-3β compared with HGF-stimulated nontransfected cells (Fig. 7B) . The gels were reprobed with p-Akt-1 and showed a correlation with its decreased activation in the transfected cells stimulated with the growth factor. These experiments demonstrate the involvement of Akt in GSK-3α and -β activation by HGF. Downregulation of Akt-1 also induced an approximate 44% inhibition of Bad phosphorylation stimulated by HGF (Fig. 7C) . 

Discussion

A healthy corneal epithelium is fundamental to the maintenance of the integrity and clarity of the cornea. With the increasing number of surgical procedures that alter corneal curvature to correct refractive errors, there is a growing recognition of the importance of minimizing disruptions of the epithelium. After epithelial injury, synthesis of HGF is upregulated in stromal cells and in the lacrimal gland. ^{2 3 4} In the present study, we demonstrated that, in addition to stimulating proliferation and migration, ⁶ HGF is essential for the protection of cells from apoptosis. This cytoprotection occurred with different inducers of apoptosis in both primary cultures of RCE cells and in an HCE cell line. This could be interpreted as a mechanism stimulated by HGF to prevent further corneal damage and to contribute to the rapid regeneration response by the epithelium after injury. 

We identified the PI-3K/Akt-1 signaling pathway as the one responsible for protecting corneal epithelial cells from apoptosis. Two PI-3K inhibitors, LY294002 and wortmannin, abolished the antiapoptotic effect of HGF. The role of Akt-1 in regulating apoptosis in corneal epithelial cells was demonstrated by inhibition of apoptosis by HGF, which correlated with increased Akt-1 activity, and by transfection of cells with an active Akt-1, which protected the cells from apoptosis in a manner similar to stimulation with HGF. 

The high and sustained level of activation of Akt-1 in the cells resulting from HGF stimulation suggests a delay in the downregulation of the c-Met receptor. In fact, a very recent study of a tumor cell line showed that the c-Met receptor remains on the cell surface much longer than does the epidermal growth factor (EGF) receptor, allowing HGF to sustain activation of Akt and exert antiapoptotic effects. ²⁸ Corneal epithelial cells express receptors for several growth factors ^{1 29 30 31} and it will be important in the future to determine whether differential downregulation of these receptors correlates with different functions. 

We also found that blockage of the HGF-stimulated ERK1/2 pathway by PD98059 had no effect on corneal epithelial survival. This is in contrast with a previous study using a human myosarcoma cell line, in which HGF protected those cells from UV-induced apoptosis by way of both the PI-3K/Akt and ERK1/2 pathways ³² ; however, it is in agreement with a more recent study in primary cultures of human keratinocytes ³³ and indicates that the initiating events in apoptosis are diverse and may involve cell-specific activity. 

Two downstream targets of Akt-1 were identified. One of them, Bad, was phosphorylated at Ser-136 in response to HGF. This allowed retention of the protein in the cytosol and prevented its binding to mitochondrial proteins. In fact, staurosporine caused dephosphorylation of Bad in the cytosolic fraction (data not shown). The involvement of PI-3K signaling was demonstrated by the inhibition of HGF-induced Bad phosphorylation with PI-3K inhibitors, and the role of Akt-1 was demonstrated by the decrease in Bad phosphorylation when synthesis of Akt was decreased by RNA interference. 

Bad was also translocated from the cytosol to the mitochondria in the presence of the apoptosis inducer, staurosporine, which binds and inhibits the actions of antiapoptotic proteins such as Bcl-2 and Bcl-XL. ^{26 27} Expression of these two proteins has been reported in corneal epithelial cells. ^{34 35} Translocation of Bad may produce opening of the mitochondrial permeability transition pore and decrease membrane potential and swelling of the membrane. ³⁶ Stimulation by HGF decreases the translocation and helps in the maintenance of mitochondrial membrane integrity. This protection is critical in preventing the release of cytochrome c, which results in the activation of a caspase cascade that concludes with cell death. ¹¹ The role of Akt-1 in the translocation of Bad was demonstrated by the decrease of Bad in the mitochondrial fraction of corneal epithelial cells transfected with active Akt-1. 

Phosphorylation of GSK-3 may represent another mechanism by which HGF decreases apoptosis in corneal epithelial cells. We demonstrated that HGF phosphorylates Ser-9 in GSK-3β and Ser-21 in GSK-3α by PI-3K/Akt signaling. These phosphorylations blocked the action of GSK-3. ³⁷ GSK-3β promotes apoptosis in neural and vascular smooth muscle cells. ³⁸ Overexpression of GSK-3 induces apoptosis in fibroblasts, and inhibitors of GSK protect neurons from apoptosis, ³⁷ although the mechanism by which inhibition of GSK-3 through PI-3K/Akt suppresses apoptosis is not known. It also phosphorylates a variety of transcription factors that cause decreases in DNA binding, and as a consequence, decreases in nuclear transcription (for review, see Ref. ²⁵ ). In addition, the activities of transcription factors that are critical promoters of cell survival, such as heat-shock factor (HSF-1) and cyclic AMP response element-binding protein (CREB), are inhibited by GSK-3. ^{39 40} Our results from experiments with two PI-3K inhibitors demonstrated a partial blockade of the phosphorylation of GSK-3α and -β, suggesting that phosphorylation and inhibition of GSK-3 in corneal epithelial cells is partially mediated by PI-3K/Akt. It is possible that inactivation of this enzyme by HGF is involved in other physiologic roles of the growth factor that require additional signaling mechanisms. One attractive possibility is activation of glycolysis. Corneal epithelium contains large deposits of glycogen that can be depleted by contact lenses wear or mild trauma, ⁴¹ but the mechanisms of this depletion have not been clarified. Inactivation of GSK-3α and -β by HGF phosphorylation could be one of the mechanisms that increase glycolysis. More work remains to decipher the precise biochemical targets and the importance of Akt/GSK-3 in cell survival. 

In summary, in our studies, HGF elicited multiple, diverse biological actions in corneal epithelial cells. In addition to its involvement in migration and proliferation ⁶ we clearly demonstrated that HGF has an antiapoptotic action. These diverse functions are mediated through different postreceptor signal-transduction pathways. The antiapoptotic action of HGF in corneal epithelial cells is exerted by a PI-3K/Akt pathway at the mitochondrial level through Bad phosphorylation in the cytosol and decreased translocation and probably through GSK-3 at the transcriptional level. Because many clinical conditions that affect the corneal surface are manifested by recurrent erosions and persistent epithelial defects, a better knowledge of the complex effects of HGF on corneal epithelial cells will allow the development of new therapeutic strategies to maintain a healthy corneal epithelium. 

 Supported by Grant R01EY06635 from the National Eye Institute.

 Submitted for publication April 2, 2004; revised May 7, 2004; accepted May 17, 2004.

 Disclosure: A. Kakazu, None; G. Chandrasekher, None; H.E.P. Bazan, Genentech (F)

 The publication costs of this article were defrayed in part by page charge payment. This article must therefore be marked “advertisement” in accordance with 18 U.S.C. §1734 solely to indicate this fact.

 

Figure 1.

 

Effect of HGF on corneal epithelial cell survival. RCE and HCE cells were incubated overnight in DMEM/F12 and KBM, respectively, with or without 20 ng/mL HGF. The medium was changed to a similar medium containing the apoptosis inducers (A, B) staurosporine at 20 ng/mL for RCE cells and 100 ng/mL for HCE cells or (C) the Ca²⁺ ionophore A23187 at 0.5 μM. The cells were further incubated for 7 (RCE) or 20 (HCE) hours. (D) ExM was kept on the cells for 24 hours. DNA was extracted for DNA ladder analysis. The first lane on the left of the gels represents a 100-bp DNA ladder used as marker. Cells were also stained with Hoechst reagent. (B) Normal nuclei showed faint staining, whereas increased brightness indicated apoptosis. HGF decreased the staining. Similar results were obtained in three independent experiments.

Figure 2.

 

Involvement of the PI-3K but not ERK1/2 pathway in the protective effect of HGF. RCE cells were treated with (A) staurosporine or (B) ExM as in Figure 1 . Wortmannin (100 nM), LY294002 (10 μM), or PD98059 (10 μM) was added 30 minutes before HGF. (C) Apoptotic cells were counted in six different fields and the results expressed as a percentage of total cells counted. *Significant difference compared with ExM; **significant difference compared with HGF (P < 0.05).

Figure 3.

 

HGF activation of Akt-1 through PI-3K activation. (A) RCE cells were starved overnight and then stimulated with 20 ng/mL HGF for the times shown. Akt-1 activity was measured in immunocomplexes incubated with a specific peptide substrate. Data are the average ± SD of results in three independent samples. (B) Immunoblot analysis with p-Akt for the detection of active Akt-1 and with total Akt-1 to demonstrate that gel loading was similar in all samples. (C) p-Akt staining of HCE cells stimulated with HGF for different times. The nuclei were stained with Hoechst reagent. Immunofluorescence micrographs show merged double stain. (D) Cells were preincubated for 30 minutes with LY294002 or PD98059 before addition of HGF for 15 minutes. Similar results were obtained in three independent experiments.

Figure 4.

 

Activation of Akt-1 protects RCE cells from apoptosis. (A) Cells were starved overnight and then treated with ExM for 24 hours and the activity of Akt-1 was assayed. Data are the average ± SD of results in three separate samples. (B) Immunostaining of Akt-1 cDNA-transfected cells shows p-Akt mostly in the plasma membrane. Hoechst reagent was used to label the nuclei of the cells. (C) Transfection of RCE cells with Akt-1 cDNA and the empty expression vector or transfecting reagent was assessed by immunoblot with c-Myc. (D) Transfected cells at a 3:1 ratio were treated with 20 ng/mL staurosporine for 7 hours. DNA was extracted and separated by agarose gel electrophoresis. The experiment was repeated twice with similar results.

Figure 5.

 

Activation of GSK-3 and Bad by HGF. (A) RCE cells were starved overnight and stimulated with HGF for the times shown. (B) LY294002 or PD98059 was added 30 minutes before stimulation with HGF for 15 minutes. Cell extracts were prepared and analyzed by Western blot with antibodies against p-GSK-3. (C) HCE cells were starved 24 hours, stimulated with HGF for the time shown, and probed with p-Bad and Bad antibodies. (D) LY294002 or PD98059 was added 30 minutes before HGF stimulation for 15 minutes. The experiments were repeated three times with similar results.

Figure 6.

 

HGF inhibited staurosporine-induced translocation of Bad to the mitochondria. HCE cells were transfected with active Akt-cDNA. Apoptosis was induced in transfected or nontransfected cells by incubation with 100 ng/mL staurosporine for 6 hours. HGF (20 ng/mL) was present in nontransfected cells, as indicated. Mitochondrial and cytosolic fractions were obtained, subjected to SDS-PAGE (12% gel), and probed with Bad antibody. VDAC1 was used as a mitochondrial marker. The experiment was repeated twice with similar results.

Figure 7.

 

Effect of Akt knockdown by siRNA on GSK-3α and -β and Bad. RNA interference by siRNA transfection was performed in HCE and RCE cells. Forty-eight hours after the transfections, cells were starved overnight and then stimulated with HGF for 15 minutes. The total cell lysate was analyzed by immunoblot with specific antibodies. (A) Transfected HCE cells probed with Akt-1. The first lane is a negative control with nonsilencing siRNA (fluorescein-conjugated, Fl-siRNA). (B) siRNA-transfected RCE cells stimulated with HGF for 15 minutes were probed with p-GSK-3α and -β and p-Akt antibodies. (C) HCE cells were treated the same as RCE cells and probed with p-Bad antibody.