Ciliary Neurotrophic Factor Released by Corneal Endothelium Surviving Oxidative Stress Ex Vivo

Abstract

purpose. To demonstrate that corneal endothelial (CE) cells that survive oxidative stress in corneas release ciliary neurotrophic factor (CNTF), which does not possess the secretion signal sequence.

methods. CNTF and CNTF receptor α subunit (CNTFRα) in CE cells and cell-conditioned medium (H₂O₂ and PBS placed in bovine corneal cups for 30 minutes at 37°C) in explant cultures were demonstrated by Western blot (WB) and immunoprecipitation (IP). The number of dead CE cells was determined microscopically with a viability kit, by an observer uninformed of the explants’ identities. CNTF and CNTFRα synthesis and release by CE cells in ³⁵S-methionine–labeled (0.1 mCi/mL for 8 hours at 37°C) corneal cups were shown by autoradiography and WB.

results. CE cells in fresh bovine eyes expressed a 25-kDa CNTF that was recognized by three different antibodies. CE cells expressed a 61-kDa CNTF-immunoreactive molecule (IM), which disappeared from the CE cells in H₂O₂-conditioned corneal cups, concomitant with the appearance of the 25-kDa CNTF in the conditioned medium. Corneal cups containing 0, 0.006, 0.012, 0.023, 0.045, 0.09, 0.18, and 0.35 mM H₂O₂ demonstrated relative levels of CE cell 61-kDa CNTF-IM of 100%, 84%, 77%, 61%, 52%, 39%, 35%, and 35%, respectively, whereas levels of 25-kDa CNTF in the conditioned medium were 23%, 32%, 39%, 63%, 80%, 90%, 100%, and 63%, respectively. CE cells expressed a 53-kDa CNTFRα that, along with trace amounts of a 61-kDa CNTFRα-IM, appeared concomitantly with the 25-kDa CNTF in the conditioned medium. H₂O₂ (0–0.56 mM) did not affect the viability of CE cells (15 dead cells per 600 cells). CE cells in ³⁵S-methionine–labeled corneal cups synthesized and released a ³⁵S 61-kDa molecule that was both CNTF- and CNTFRα-immunoreactive in an H₂O₂-dependent manner, whereas 25-kDa CNTF was detected in the ³⁵S-methionine labeling medium.

conclusions. CE cells release autocrine CNTF under sublethal oxidative stress by a mechanism that involves CNTFRα and the formation of a 61-kDa CNTF/CNTFRα-IM.

The ciliary neurotrophic factor (CNTF), the second neurotrophic factor to be identified, was first characterized as a survival factor for the chick ciliary ganglion neurons and was detected in an extract of eye tissues consisting of ciliary body, iris, and choroid. ^{1 2 3} It has since been shown to exert neurotrophic effects on various neuronal cells, including purified retinal ganglion cells in culture and motor neurons in vivo. ^{4 5 6 7} CNTF is a protein of 200 amino acids ⁸ that shares sequence homology with interleukin-6, leukemia inhibitory factor (LIF), oncostatin M, and interleukin-11. ^{9 10} The receptor for CNTF consists of a ligand-binding α subunit, CNTFRα, and two β subunits, gp130 and LIFRβ. Ligand binding to CNTFRα induces formation of a dimer by the two β subunits gp130/LIFRβ, leading to activation of the associated protein tyrosine kinase JAK, which in turn phosphorylates tyrosine residues on gp130/LIFRβ. ^{11 12} Phosphotyrosine residues on gp130/LIFRβ become the docking sites for the SH2 domain, containing downstream target molecules of JAK. ^{11 12}  

CNTFRα is widely expressed in the central and peripheral nervous systems. ^{11 12} In the rat retina, CNTFRα is expressed in horizontal cells and subpopulations of amacrine and ganglion cells. ¹³ Many studies have demonstrated the beneficial effects of applying exogenous CNTF to protect against naturally occurring or induced neuronal degeneration. For example, exogenous CNTF has been shown to protect the retinas of normal albino rats from the damaging effect of constant light ¹⁴ and to slow the progression of retinal degeneration in mice with genes that cause inherited retinal degeneration and in transgenic mice with a mutated rhodopsin gene. ^{15 16} Furthermore, CNTF has been shown to promote the axonal genesis of dissociated retinal ganglion cells, ¹⁷ and, when applied intravitreously, CNTF promotes axonal regeneration of axotomized retinal ganglion cells in adult hamsters. ¹⁸ The expression of CNTFRα is not limited to neural tissues, because it also has been identified in skeletal muscle of the rat and human. ^{19 20} CNTF exerts myotrophic effects on experimentally denervated rat skeletal muscle and increases expression of the mRNA of the three subunits of CNTF receptor. ¹⁹ Moreover, biopsy and autopsy samples of denervated human skeletal muscle have shown increased expression of CNTFRα mRNA. ²⁰ CNTF is also a promoter of neuronal differentiation. It stimulates the appearance of cholinergic markers in cultures of retinas and sympathetic neurons. ^{13 21 22} In cultured sympathetic neurons, CNTF induces the differentiation and synthesis of vasoactive intestinal peptide (VIP), ²³ a neuropeptide with trophic properties, ²⁴ by modulating the cytokine responsive element (CyRE) in the VIP gene. ^{25 26 27} Recently, CNTF has been shown to enhance stem cell self-renewal in the adult forebrain in vivo and to prevent neural stem cells from becoming glial progenitor cells, resulting in enhanced expansion of the number of stem cells in vitro. ²⁸  

CNTF does not have a classic secretory signal sequence ⁸ and is thought to be a lesion factor that is released only after injury, ²⁹ and there has been some indirect evidence to support this notion. It has been demonstrated that, after axotomy, significant CNTF accumulates extracellularly at the lesion site. ²⁹ In the retina, Müller cells are immunocytochemically identified as the site of production of CNTF. ¹³ Expression of CNTF is upregulated in mechanically injured normal mouse retina, ³⁰ whereas CNTF plays a role in the focal mechanical-injury–induced slowing down of photoreceptor degeneration in inherited retinal dystrophy in rats. ³¹ In peripheral nerve, CNTF is expressed in the Schwann cells. ³² In pmn mutant mice that normally exhibit motor neuron degeneration, facial nerve transection leads to an increased number of surviving motor neurons, a beneficial effect not observed in pmn mutant mice, in which CNTF is not expressed. ³³ Nevertheless, the mechanism of release of CNTF is completely unknown. Adler et al. ¹ have reported that, on the basis of trophic units per milligram protein, the cornea contains a slightly higher level of CNTF than the retina. By an organ culture system of bovine corneas, the present study demonstrates, for the first time, the release of CNTF in response to H₂O₂ presented at sublethal levels. 

The corneal endothelium, a monocellular layer that functions as a barrier to movement of fluid into the cornea and actively pumps fluid out of the cornea, maintains the transparency of the cornea. ³⁴ The endothelium’s developmental origin is that of the neural crest ^{35 36 37 38} and the corneal endothelial (CE) cells express neuron-specific enolase. ^{39 40} Recently, we have reported that the mRNA and protein of VIP are expressed in CE cells in fresh bovine and donor human eyes, and CE cell survival in cornea organ cultures subjected to lethal oxidative stress is promoted by exogenous VIP. ⁴¹ The present study demonstrated the presence, biosynthesis, and sublethal H₂O₂-induced release of CNTF in CE cells in fresh bovine eyes and in corneoscleral explant cultures. 

Materials and Methods

Corneoscleral Explants

Fresh bovine eyes were obtained from the local abattoir within 6 hours after death. The procedures for dissecting corneoscleral explants were the same as those previously reported, ^{41 42} except a scleral incision 1 mm (instead of 2 mm) posterior to the limbus was used, so that the prepared corneoscleral explants were free of the trabecular meshwork. For those to be used for ³⁵S-methionine labeling of the corneal cups, corneoscleral explants were dissected under sterile conditions after brief immersion of eyeballs in a 0.5% iodine solution (1:4 dilution in PBS of povidone iodine [Baxter Health Care Co., Deerfield, IL]) and rinse with a stream of PBS. 

H₂O₂ Conditioning of Corneal Cups in Cultured Corneoscleral Explants

The corneoscleral explants were placed in PBS on ice until all the eyes were dissected. With the corneas’ concave side up, the explants were then incubated for 75 minutes in Eagle’s MEM supplemented with 20 mM HEPES (pH 7.2; EMEM-HEPES) in 35-mm culture dishes at 37°C in 5% CO₂-95% air before incubation with H₂O₂. The corneoscleral explants (concave side up) were then placed on caps of 50-mL conical centrifuge tubes (Costar, Cambridge, MA), and a conditioning medium (0.5 mL) of H₂O₂ (0–0.35 or 0–0.56 mM) in PBS was placed in each of the corneal cups. The corneoscleral explants were placed in humidified chambers and incubated for 30 minutes at 37°C in 5% CO₂-95% air. The CE cell extract and the concentrated conditioned medium were prepared for Western blot analysis and immunoprecipitation, as described below. 

CNTF Synthesis and Release by CE Cells in ³⁵S-Methionine–Labeled Corneal Cups in Corneoscleral Explant Culture

The corneoscleral explants were dissected under sterile conditions, with the procedures described above and rinsed once with 12 to 14 mL PBS in 60 × 15-mm culture dishes. With the corneas’ concave side up, the explants were then incubated for 30 minutes at room temperature in 12 to 14 mL medium A (DMEM supplemented with 100 U/mL penicillin, 100 μg/mL streptomycin sulfate, 0.25 μg/mL amphotericin B, and 20 mM HEPES) in 60 × 15-mm culture dishes. At the end of incubation in medium A, the explants were transferred to a 5% fetal bovine serum-containing medium B (DMEM supplemented with 100 U/mL penicillin, 100 μg/mL streptomycin sulfate, and 0.292 mg/mL l-glutamine) and incubated at 37°C under 5% CO₂-95% air for 4 hours before they were transferred to 35-mm culture dishes containing 3.5 mL methionine-free EMEM-HEPES and incubated for an additional 20 hours. The explants (concave side up) were then placed on caps of 50 mL-conical centrifuge tubes (Costar). Each of the corneal cups was filled with a labeling medium (∼0.5 mL; methionine-free EMEM-HEPES containing 0.1 mCi/mL ³⁵S-methionine) and the corneoscleral explants were placed in humidified chambers and incubated for 6 to 8 hours at 37°C. For studies of CNTF synthesis, the CE cells were scraped off the corneas and homogenized in the lysis buffer to obtain the CE cell extract, as described below. For studies of the release of CNTF, the ³⁵S-methionine–labeled corneal cups in corneoscleral explants were incubated with H₂O₂, as described above. The CE cell extract, the conditioned medium, and the ³⁵S-methionine–containing labeling medium were subjected to immunoprecipitation with a polyclonal goat anti-human CNTF antibody (AF-257-NA; R&D Systems, Minneapolis, MN) and a polyclonal goat anti-human CNTFRα antibody (AF-303-NA; R&D Systems) for autoradiography and Western blot analysis. 

CE Cell Extract

CE cells were scraped off the corneas in fresh and cultured corneoscleral explants and homogenized in either the RIPA buffer (25 mM Tris [pH 7.2] 0.1% SDS, 1% Triton X-100, 1% sodium deoxycholate, 150 mM NaCl, 1 mM EDTA, and one tablet of protease inhibitor cocktail [complete, mini; Roche Diagnostics, Mannheim, Germany]) or a lysis buffer (20 mM Tris [pH 7.4], 1% NP40, 137 mM NaCl, 50 μM EDTA, and one tablet of protease inhibitor cocktail/10 mL [complete, mini; Roche Diagnostics, Mannheim, Germany]). The homogenates were centrifuged at 12,000g for 10 minutes to remove the insoluble materials to obtain the CE cell extracts. 

Concentrated CE Cell–Conditioned Medium

The CE cell–conditioned medium was collected, cleared of cell debris by centrifugation (12,000g, 10 minutes), and frozen at −70°C before being concentrated in centrifugal filter units with a molecular mass cutoff of 5000 daltons (Biomax 5; Millipore, Bedford MA). One fourth of the concentrated conditioned medium collected from each of the corneal cups was used for Western blot analysis. For immunoprecipitation, the concentrated conditioned medium was diluted with an equal volume of 2× lysis buffer (1× contains 20 mM Tris [pH 7.4] 1% NP40, 137 mM NaCl, 50 μM EDTA, and one tablet of protease inhibitor cocktail/10 mL [complete, mini; Roche Diagnostics]). 

Western Blot Analysis

Samples of CE cell extracts in the RIPA buffer and the concentrated conditioned medium were prepared in a sample buffer containing 2.5% β-mercaptoethanol for SDS-polyacrylamide gel electrophoresis (PAGE) on preformed Tris-glycine 8% to 16% polyacrylamide gels (Novex, San Diego, CA). Samples of CE cell extract and conditioned medium subjected to antibody immunoprecipitation were prepared in a nonreducing sample buffer for SDS-PAGE on preformed tricine polyacrylamide gels (10%–20%; Novex). The electrophoresed proteins were electrophoretically transferred to nitrocellulose membranes for Western blot analysis. Mouse monoclonal and goat polyclonal anti-human CNTF antibodies (MAB257 and AF-257-NA; R&D Systems), a chick polyclonal anti-rat CNTF antibody (G1631; Promega, Madison, WI), and a goat polyclonal anti-human CNTFRα antibody (AF-303-NA; R&D Systems) were used at either 1:1250 or 1:625 dilution in 1% BSA-PBS. For detecting the immunoreactive molecules using an enhanced chemiluminescence (ECL) method, a kit was used (Amersham Pharmacia Biotech, Piscataway, NJ), with either its horseradish peroxidase (HRP)–linked anti-mouse IgG secondary antibody or with a HRP-linked anti-chick IgY secondary antibody (Promega, Madison, WI). For detecting immunoreactive molecules using a color method, Fast Red TR/Naphthol AS-MX I tablets (Sigma, St. Louis, MO) were used in conjunction with an alkaline phosphatase-linked anti-goat IgG secondary antibody (ICN, Costa Mesa, CA). The optical densities of bands were measured by a densitometer (NucleoVision; NucleoTech, San Carlos, CA). After ECL detection of CNTF, some of the nitrocellulose membranes were stripped (Restore Western Blot Stripping Buffer, no. CA46067; Pierce, Rockford, IL) and reprobed for actin, with a monoclonal antibody that recognized α-, β-, and γ-actins (Oncogene, Cambridge, MA). 

Immunoprecipitation

For immunoprecipitation, the CE cell extracts in lysis buffer were incubated with a goat polyclonal anti-human CNTF antibody (2.0 μg IgG/mg protein, AF-257-NA; R&D Systems), whereas the concentrated medium in lysis buffer (above) was incubated with the goat polyclonal anti-human CNTF antibody and a goat polyclonal anti-human CNTFRα antibody (AF-257-NA and AF-303-NA; R&D Systems) at the designated concentrations for 22 hours at 4°C. An equal volume of packed protein A Sepharose beads (Sigma) was added, and the incubation continued for 1 hour, followed by centrifugation to obtain the immune complex. After washing the immune complex with the lysis buffer (three times) and PBS (one time), a nonreducing sample buffer was added to the complex, which was then boiled for 5 minutes for SDS-PAGE on preformed tricine polyacrylamide gels (10%–20%; Novex). The molecules on the gels were electrophoretically transferred to nitrocellulose membranes, which were then either apposed to x-ray films for autoradiography or immunoblotted as described in Western blot analysis. 

Viability of CE Cells in H₂O₂-Conditioned Corneal Cups in Cultured Corneoscleral Explants

Corneoscleral explants were pretreated and corneal cups were conditioned with H₂O₂ (0–0.56 mM) as described above. Corneoscleral explants with corneal cups conditioned with identical concentrations of H₂O₂ were divided into two groups. One group of corneoscleral explants were incubated with reagents from a cell viability kit (Live/Dead Viability kit; Molecular Probes, Eugene, OR) and placed on coverslips. ⁴¹ Nuclei of the CE cells with compromised plasma membranes, which allowed entrance of the DNA-binding fluorescent ethidium homodimer in the reagent mixture from the kit, appeared red under a fluorescence microscope. With identities of the corneoscleral explants withheld from the examiner and under an inverted microscope (×200 magnification) equipped with an epifluorescence attachment (Diaphot-TMD; Nikon, Tokyo, Japan), the number of dead cells shown by red nuclei in the field (0.33 mm²; 600 bovine CE cells) ⁴¹ defined by the photograph mask that was placed in the optical path of the microscope, were counted. Fourteen fields were counted in each of the corneoscleral explants. In the other group of corneoscleral explants, CE cells were scraped off the corneas with a razor blade and homogenized in the RIPA buffer for CE cell extract preparation and Western blot analysis for CNTF immunoreactivity, with the mouse monoclonal anti-human CNTF antibody (MAB257, R&D Systems) described above. 

Statistical Analysis

Data were analyzed by analysis of variance (ANOVA) and the Dunnett post hoc test or Student’s t-test. 

Results

CNTF in CE Cells

Western blot analysis of CE cell extracts from fresh bovine corneas showed a 25-kDa molecule immunoreactive to all three anti-CNTF antibodies tested (Fig 1) . Using the monoclonal anti-human CNTF antibody, Western blot analysis of the polyclonal anti-human CNTF antibody–immunoprecipitated CE cell extracts also demonstrated that CE cells in fresh bovine eyes expressed the 25-kDa CNTF (Fig. 1B) . Autoradiography was then used to demonstrate that CE cells synthesized the CNTF-immunoreactive molecule. After metabolically labeling CE cells in the corneal cups in corneoscleral explant cultures with ³⁵S-methionine, followed by immunoprecipitation of the CE cell extract with the polyclonal anti-human CNTF antibody and SDS-PAGE, a major 61-kDa ³⁵S-labeled molecule and not the 25-kDa molecule was detected by autoradiography (Fig. 2A) . Two separate experiments for CNTF biosynthesis were conducted with identical results. The possibility that the CNTF-immunoreactive molecule in the CE cells in corneoscleral explant culture existed as a 61-kDa molecule was further investigated. Western blot analysis of CE cell extracts from corneoscleral explants that have been placed in organ culture showed a major CNTF-immunoreactive molecule with a molecular mass of 61-kDa and the level of the 25-kDa CNTF that was detected in CE cells from fresh corneas was greatly diminished (Fig. 2B) . To investigate the possibility that the 61-kDa, CNTF-immunoreactive molecule was an adduct of the 25-kDa CNTF and another molecule through the formation of disulfide bonds, CE cell extracts were treated with dithiothreitol (up to 100 mM, at 4°C for 4 hours). The result showed that dithiothreitol had no effect on the 61-kDa molecule (data not shown). 

Disappearance of the 61-kDa CNTF-Immunoreactive Molecule from CE Cells in H₂O₂-Conditioned Corneal Cups in Cultured Corneoscleral Explants

Incubation of the corneal cups with H₂O₂-PBS resulted in H₂O₂-concentration–dependent decreases in the level of the 61-kDa CNTF-immunoreactive molecule in the CE cells (Fig. 3A) . Reprobing of the Western blot analysis that showed H₂O₂-concentration–dependent decreases in the level of the 61-kDa CNTF-immunoreactive molecule in the CE cells (Fig. 3A) with an antibody against actin indicated that the levels of this cytoskeletal protein remained unaffected by the H₂O₂ (Fig. 3B) . Densitometry of blots from seven separate experiments showed that levels of the 61-kDa CNTF-immunoreactive molecule in the CE cells in corneal cups conditioned with 0.006, 0.012, 0.023, 0.045, 0.09, 0.18, and 0.35 mM H₂O₂ were (mean ± SEM) 84% ± 13%, 77% ± 11%, 61% ± 14%, 53% ± 14%, 39% ± 13%, 35% ± 12%, and 35% ± 12% of that found in the control cups (0 mM H₂O₂), respectively (Fig. 3C) . Statistical analysis (ANOVA) indicated that there was a significant difference (P = 0.0007) among various groups, and the Dunnett post hoc test indicated that levels found in the CE cells conditioned in 0.045, 0.09, 0.18, and 0.35 mM H₂O₂ were significantly lower (P < 0.05) than that found in the control (0 mM H₂O₂). 

Appearance of 25-kDa CNTF in the CE Cell–Conditioned Medium in an H₂O₂-Concentration–Dependent Manner

Western blot analysis demonstrated that the level of 25-kDa CNTF in the CE cell–conditioned medium in corneal cups increased in an H₂O₂ concentration-dependent manner and the highest level was observed in medium containing 0.18 mM H₂O₂ (Fig. 4A) . Densitometry of the blots from five separate experiments showed that the levels of the 25-kDa CNTF-immunoreactive molecule in the conditioned medium containing 0, 0.006, 0.012, 0.023, 0.045, 0.09, and 0.35 mM H₂O₂ relative to that containing 0.18 mM H₂O₂ (100%) were (mean ± SEM) 24% ± 6%, 32% ± 9%, 39% ± 9%, 63% ± 14%, 80% ± 8%, 89% ± 10%, and 61% ± 9%, respectively (Fig. 4B) . Statistical analysis (ANOVA) indicated that there was a significant difference (P = 0.0001) among various groups, and the Dunnett post hoc test indicated that levels found in the conditioned medium containing 0.023, 0.045, 0.09, 0.18, and 0.35 mM H₂O₂ were significantly higher (P < 0.05) than that in the control (0 mM H₂O₂; Fig. 4B ). 

Viability of CE Cells in H₂O₂-Conditioned Corneal Cups

To show that the appearance of the 25-kDa CNTF in CE cell–conditioned medium and the disappearance of the 61-kDa CNTF-immunoreactive molecule from the CE cells did not result from leakage of these molecules through compromised plasma membranes of the dead CE cells, the viability of the CE cells was examined. Nuclei of dead CE cells with compromised plasma membranes appeared red under a fluorescence microscope. The number of dead CE cells in corneas was low and similar in all corneas (P = 0.7, ANOVA): 11 ± 2, 15 ± 3, 19 ± 4, and 16 ± 7 (mean ± SEM, n = 4) per field of 600 CE cells in corneal cups conditioned in 0, 0.14, 0.28, and 0.56 mM H₂O₂, respectively (Fig. 5A) . In none of the H₂O₂-treated corneas was the number of dead CE cells statistically different from the number in the control corneas (Fig. 5A) , whereas the endogenous 61-kDa CNTF-immunoreactivity in CE cells decreased in an H₂O₂-dependent manner (P = 1 × 10⁻⁶, ANOVA, Fig. 5B ) with those in 0.14-, 0.28-, and 0.56-mM H₂O₂-treated corneal cups, showing significantly lower levels (P < 0.05, Dunnett post hoc test) than that in the control (0 mM H₂O₂; Fig. 5B ). 

Effect of H₂O₂ on CNTFRα in CE Cell– Conditioned Medium

Western blot analysis with an antibody against the human CNTFRα demonstrated an H₂O₂-dependent increase in the levels of a major (53-kDa) and a minor (61-kDa) CNTFRα-immunoreactive molecule in the CE cell–conditioned medium of the corneal cups (Fig. 6A) . Densitometry of the blots from five separate experiments showed that the level of 53-kDa CNTFRα-immunoreactive molecule in the conditioned medium containing 0.006, 0.012, 0.023, 0.045, 0.09, 0.18, and 0.35 mM H₂O₂ increased to 3.4 ± 1.2, 2.6 ± 0.5, 3.0 ± 0.9, 3.1 ± 0.9, 3.5 ± 0.9, 4.8 ± 0.3, and 3.7 ± 0.8 times that in the control (0 mM H₂O₂), respectively (Fig. 6B) . Whereas the level of the 53-kDa CNTFRα-immunoreactive molecule in the H₂O₂-containing conditioned medium, except that conditioned with 0.006 mM H₂O₂, was significantly higher (P < 0.05, Student’s t-test) than that in the control (0 mM H₂O₂), no statistically significant difference (P = 0.6, ANOVA) was found among these H₂O₂-elevated levels. Although the 53-kDa molecule was the CNTFRα in CE cells in fresh bovine eyes (Fig. 6C) , the 53-kDa CNTFRα-immunoreactive molecule was also present in the CE cells in organ cultures, but its level was not significantly affected by the presence of H₂O₂ in the conditioned medium of the corneal cups (data not shown). 

CNTF and CNTFRα Released from ³⁵S-Methionine–Labeled CE Cells in Corneal Cups

Autoradiography demonstrated the presence of a ³⁵S 61-kDa molecule in the medium that had conditioned the CE cells in ³⁵S-methionine–labeled corneal cups for 30 minutes. As shown in Figure 7A , the ³⁵S 61-kDa molecule appeared in both anti-human CNTF and anti-human CNTFRα immunoprecipitates obtained from the medium. The level of this ³⁵S 61-kDa molecule was increased by H₂O₂ in the conditioned medium (Fig. 7A) . Whereas Figure 2A showed that an anti-CNTF–immunoprecipitable ³⁵S 61-kDa molecule was synthesized by CE cells, it was shown in this experiment (Fig. 7A , lanes 1 and 2) that the ³⁵S 61-kDa molecule was released in CE cell–conditioned medium in an H₂O₂-dependent manner in organ cultures. To learn whether this anti-CNTF–immunoprecipitable ³⁵S 61-kDa molecule was the same as the 61-kDa CNTF-immunoreactive molecule that disappeared from the CE cells in an H₂O₂ concentration–dependent manner (Fig. 3) , the immunoreactivity of the anti-CNTF–immunoprecipitated ³⁵S 61-kDa molecule was examined. With the use of the polyclonal anti-human CNTF antibody for immunoprecipitation followed by immunoblot analysis, it was demonstrated that the anti-CNTF–immunoprecipitated ³⁵S 61-kDa molecule in the CE cells and that in the CE cell–conditioned medium (with H₂O₂) were CNTF immunoreactive (Fig. 7B , lanes 1 and 2), indicating that it was the same molecule as the 61-kDa CNTF-immunoreactive molecule that disappeared from the CE cells in an H₂O₂ concentration–dependent manner in corneoscleral organ cultures (Fig. 3) . Furthermore, studies were also undertaken to determine whether the anti-CNTFRα–immunoprecipitable ³⁵S 61-kDa molecule (Fig. 7A , lanes 3 and 4) in the conditioned medium was the same as the 61-kDa CNTFRα-immunoreactive molecule that increased in an H₂O₂ concentration–dependent manner in the CE cell–conditioned medium in corneoscleral organ cultures (Fig. 6A) . With the use of the polyclonal anti-human CNTFRα antibody for immunoprecipitation followed by immunoblot analysis, the ³⁵S 61-kDa molecule immunoprecipitated by the anti-CNTFRα antibody from the H₂O₂-containing conditioned medium was shown to be anti-CNTFRα immunoreactive (Fig. 7B , lane 3). Thus, the CNTFRα-immunoprecipitable ³⁵S 61-kDa molecule in the CE-conditioned medium was the same as the 61-kDa CNTFRα-immunoreactive molecule (Fig. 6A) that was detected in the CE-conditioned medium in an H₂O₂-dependent manner. The possibility that the 61-kDa molecule had dual immunoreactivity and was immunoreactive to both anti-CNTF and anti-CNTFRα antibodies was then examined. While the polyclonal anti-human CNTF antibody precipitated a 61-kDa CNTF-immunoreactive molecule from the ³⁵S-labled CE cells (Fig. 7B , lane 1), this molecule was recognized by the anti-CNTFRα antibody in the immunoblot (Fig. 7C) . 

In addition to the 61-kDa molecule, CE cells in organ cultures synthesized and released in the conditioned medium a series of molecules that demonstrated a molecular mass higher than and immunoreactivity the same as the 61-kDa molecule. As demonstrated in autoradiographs (Figs. 2A 7A) , a series of ³⁵S-labeled molecules with molecular mass higher than 61-kDa were coimmunoprecipitated with the ³⁵S 61-kDa CNTF/CNTFRα-immunoreactive molecule from the CE cells and its conditioned medium. As demonstrated by immunoblots, this series of molecules had molecular masses of 100, 150, 175, and 210 kDa (Figs. 7A) and were recognized by the anti-CNTF (Fig. 7B , lanes 1 and 2) and the anti-CNTFRα (Figs. 7B , lane 3, 7C ) antibodies. 

Although the 25-kDa CNTF-immunoreactive molecule was released by CE cells in explant cultures in an H₂O₂-dependent manner (Fig. 4) , the level of ³⁵S 25-kDa molecule released by CE cells in explant cultures that have been prepared for and subjected to metabolic ³⁵S-methionine labeling, a process that took more than 24 hours, appeared to be too low to be detected (data not shown). Nevertheless, from the ³⁵S-methionine labeling medium that had been placed in the corneal cups for 8 hours, a ³⁵S 25-kDa molecule was coimmunoprecipitated with the ³⁵S 61-kDa molecule by either anti-human CNTF (Fig. 8 , lane 1) or anti-human CNTFRα antibodies (Fig. 8 , lane 2). Immunoblots showed that the ³⁵S 25-kDa molecule was recognized by the anti-human CNTF antibody (Fig. 8 , lanes 3 and 4). In addition to the ³⁵S 25-kDa and the ³⁵S 61-kDa molecules, a series of high-molecular-mass ³⁵S-labeled molecules (100, 150, 175, 210, and 260 kDa) and a detectable level of a ³⁵S 41-kDa molecule were also immunoprecipitated by either anti-human CNTF (Fig. 8 , lane 1) or anti-human CNTFRα antibody (Fig. 8 , lane 2) from the labeling medium. Although the ³⁵S 25- and ³⁵S 61-kDa molecules and the series of high-molecular-mass ³⁵S-labeled molecules were recognized by the anti-human CNTF antibody, the ³⁵S 41-kDa was not (Fig. 8 , lanes 3 and 4). 

Discussion

The present study provided direct evidence for the first time that CNTF can be released from cells subjected to sublethal oxidative stress of H₂O₂. In vivo, the released CNTF would be destined for the aqueous humor, which fills the anterior chamber and constantly bathes the corneal endothelium. The potential targets of CNTF are cells in tissues that border the anterior and posterior chambers and are constantly bathed in the aqueous humor, including the corneal endothelium, trabecular meshwork, iris, ciliary body, and lens. The presence of the CNTFRα in the corneal endothelium (Fig. 6C) , trabecular meshwork, ⁴³ iris, and ciliary body (unpublished results) has been demonstrated. Thus, the release of CNTF by the CE cells that survived the oxidative stress may play a role in protecting these CNTFRα-expressing cells in tissues bordering the anterior and posterior chambers. 

H₂O₂ is a normal constituent of the aqueous humor. ⁴⁴ The physiological concentration of H₂O₂ in the aqueous humor that has appeared in various reports in the literature (25–60 μM) has been controversial, because the high concentration of ascorbic acid present in the aqueous humor may have interfered with the measurements. ⁴⁵ Nevertheless, Spector et al. ⁴⁶ have demonstrated that although the aqueous humor contains both H₂O₂-generating and -degrading substances, a steady state presence of 1 μM H₂O₂ is maintained in bovine aqueous humor. Regardless of what the true physiological concentration of H₂O₂ in the aqueous humor may be, its elevation after ocular surgery is likely, due to the presence of infiltrating macrophages and other immune cells ⁴⁷ that produce H₂O₂. ⁴⁸ Thus, under certain conditions, H₂O₂ may be present in the aqueous humor at concentrations high enough to release CNTF from the CE cells. Although the present study is the first to demonstrate the effect of sublethal concentrations of H₂O₂ on the release of a trophic factor (CNTF) that has no signal sequence for secretion, previous reports have demonstrated that on the secretion of vascular endothelial growth factor (VEGF) ⁴⁹ and transforming growth factor-β1 (TGF-β1). ⁵⁰ The mechanism of H₂O₂-induced release of CNTF likely involves several signaling pathways. H₂O₂ (0.05–0.2 mM)-induced oxidative stress causes the activation of the serine/threonine kinase Akt (protein kinase B) in vascular smooth muscle cells. ⁵¹ Low levels of oxidative stress induced by H₂O₂ (0.02 mM) has been shown to activate p38 MAP kinase in human lymphoid cells. ⁵² Phospholipase D2 in PC12 cells is activated by H₂O₂ in a concentration-dependent manner, with a maximal effect observed at 0.5 mM H₂O₂. ^{53 54} p38 MAP kinase and ERK 1/2 MAP kinase, ⁵³ as well as protein kinase C, ⁵⁴ have been shown to mediate the activation of H₂O₂-induced phospholipase D2. Very recently, H₂O₂ at concentrations as low as 0.2 mM has been shown to cause the activation of a phosphatidylinositol-specific phospholipase, phospholipase Cγ1, which plays a crucial role in promoting cell survival. ⁵⁵ In cultured nonpigmented ciliary epithelium, a low concentration of H₂O₂ (0.2 mM) stimulates ouabain-sensitive active sodium-potassium transport. ⁵⁶  

Whereas the molecular mass of the CNTF-immunoreactive molecule in CE cells from fresh bovine eyes was that expected of CNTF (25 kDa; Fig. 1 ), ³³ that from the medium in corneal cups in explant cultures had a molecular mass of 61 kDa (Fig. 2) , which disappeared from the H₂O₂-PBS–conditioned CE cells in corneal cups in an H₂O₂ concentration–dependent manner (Fig. 3) . In contrast, the CNTF-immunoreactive molecule, which appeared in the conditioned medium in a concentration-dependent manner, had the expected molecular mass of 25 kDa (Fig. 4) , not 61 kDa. However, a ³⁵S 61-kDa molecule was immunoprecipitated with either anti-CNTF or anti-CNTFRα antibodies from the 30-minute–conditioned medium of ³⁵S-methionine–labeled CE cells in an H₂O₂-dependent manner (Fig. 7A) . Whereas Western blot analysis demonstrated that the anti-CNTF antibody did not cross-react with the CNTFRα (molecular mass, 53 kDa, Fig 2B ), the anti-CNTFRα antibody did not recognize the CNTF (25 kDa, Fig. 6C ). The dual CNTF/CNTFRα immunoreactivities of the 61-kDa molecule suggest that the molecule is an adduct of CNTF and CNTFRα. From a series of immunoprecipitation and immunoblot experiments (Figs. 7B 7C) , it was tentatively concluded that the 61-kDa molecule in the CE-conditioned medium (Figs. 6 7) and the 61-kDa molecule in CE cells (Figs. 2 3) in corneoscleral explant cultures were the same molecule, that the 61-kDa molecule was both anti-CNTF- and anti-CNTFRα immunoreactive (Figs. 2 3 6 7B 7C) , and that the 61-kDa molecule was synthesized and released by the CE cells under sublethal oxidative stress (Figs. 2A 3 6A 7A) . Although only the ³⁵S 61-kDa molecule was immunoprecipitated with either anti-CNTF or anti-CNTFRα antibodies in 30-minute–conditioned medium of ³⁵S-methionine–labeled CE cells in an H₂O₂-dependent manner (Fig. 7A) , both ³⁵S 25-kDa and ³⁵S 61-kDa molecules were immunoprecipitated from the ³⁵S-methionine–containing labeling medium after 8 hours’ incubation in corneal cups in explant cultures (Fig. 8 , lanes 1, 2). These data suggest that CNTF-immunoreactive molecule was released as a 61-kDa molecule, which was then cleaved to form the 25-kDa CNTF. Although cleavage of the 25-kDa CNTF-immunoreactive molecule occurred rapidly after the 61-kDa CNTF/CNTFRα was released by the CE cells, resulting in accumulation of the 25-kDa CNTF-immunoreactive molecule in the CE-conditioned medium (Fig. 4) , in those corneoscleral explant cultures that were prepared for and subjected to metabolic ³⁵S-methionine labeling (a process that took >24 hours), the cleavage did not occur rapidly and allowed the 61-kDa CNTF/CNTFRα molecule to accumulate in the CE-conditioned medium (Fig. 7A) . 

At the present time, the relationship between the 25-kDa CNTF and the 61-kDa CNTF/CNTFRα-immunoreactive molecules is not known. It is possible that before its release from CE cells in corneoscleral explant cultures the 25-kDa CNTF formed a complex (molecular mass, 61 kDa) with its putative binding molecule. The identity of the putative CNTF-binding molecule in the 61-kDa CNTF/CNTFRα-immunoreactive molecule remains to be established, but CNTFRα is a logical candidate. Whereas CNTFRα is anchored to the cell membrane through a glycosylphosphatidyl (GPI) linkage, it can be released from the membranes by exogenous phosphatidylinositol-specific phospholipase C. ⁵⁷ Although H₂O₂ at concentrations as low as 0.2 mM has been shown to cause the activation of the phosphatidylinositol-specific phospholipase, phospholipase Cγ1, ⁵⁵ how the 53-kDa CNTFRα may have been cleaved from its membrane anchor is unknown. In any event, because both the 25-kDa CNTF-immunoreactive molecule (Fig. 4) and the 53-kDa CNTFRα-immunoreactive molecule (Fig. 6) appeared in the CE-conditioned medium, it is tempting to speculate that the 61-kDa molecule was an adduct formed by one of each of these two molecules and released in the conditioned medium. 

However, there are two lines of evidence arguing against the involvement of the 53-kDa CNTFRα. First, the H₂O₂ concentration-dependent curve of the level of 53-kDa-CNTFRα in the conditioned medium (Fig. 6B) appeared to be very different from that of the 25-kDa CNTF-immunoreactive molecule (Fig. 4B) . Second, the combined molecular mass of 53 and 25 kDa is much larger than 61 kDa. Whereas the CNTFRα gene encodes a protein of 372 amino acids (molecular mass, 41 kDa) with four potential glycosylation sites, ^{57 58} the 53-kDa species appeared to be the only CNTFRα-immunoreactive molecule detected in CE cells from either fresh bovine eyes (Fig. 6C) or corneal cups in explant cultures (data not shown). It is possible that the 61-kDa CNTF/CNTFRα-immunoreactive molecule synthesized by CE cells (as demonstrated by autoradiography in Figs. 2A and 7A ) and that detected in CE cells (Figs. 2B 3 7B 7C) in corneoscleral explant cultures was an adduct of CNTF (molecular mass, 25 kDa) and the unglycosylated CNTFRα. There is an apparent discrepancy between the added molecular masses (of CNTF [25 kDa] and the unglycosylated CNTFRα [41 kDa]) and 61 kDa. However, as has been described in Plun-Favreau et al. ⁵⁹ and references therein, dramatic conformation changes take place on binding of CNTF to CNTFRα, which may have caused the CNTF and CNTFRα complex to assume an apparent molecular mass that was 5 kDa smaller than the two corresponding molecular masses combined. A minor 41-kDa band that appeared in autoradiography (Fig. 8 , lanes 1 and 2) of the immunoprecipitated and electrophoresed ³⁵S-methionine labeling medium may be the unglycosylated CNTFRα. The minute amount of the 41-kDa CNTF-immunoreactive molecules compared with that of the 25-kDa and 61-kDa molecules in the 8-hour conditioned medium (Fig. 8) may have resulted from the recycling of the 41-kDa unglycosylated CNTFRα. Although acting as a carrier of the 25-kDa CNTF, the 41-kDa unglycosylated CNTFRα may reenter the cell after the 25-kDa CNTF is cleaved off, to form a new 61-kDa adduct in the cell, which is then released in the conditioned medium in the next cycle. Recently, CNTFRα has been shown to be essential for the release of cardiotrophin-like cytokine (CLC), which shows a CNTFRα-binding characteristic similar to that of CNTF, in cells transfected with cDNAs of the cytokine and cytokine receptor. ^{59 60 61}  

The present study is the first to demonstrate that naturally expressed CNTF and CNTFRα can be released simultaneously by cells under sublethal oxidative stress. Whether the release of the 53-kDa CNTFRα was a prerequisite for the release of the 61-kDa CNTF/CNTFRα is not known at present. Nevertheless, Davis et al. ⁶² have hypothesized that the released CNTFRα and exogenous CNTF can form a complex that acts as a ligand to confer CNTF responsiveness on cells that have no the CNTFRα but express the other two components, gp130 and LIFRβ, of the receptor complex for CNTF. The fate of CNTF released from the CE cells is not yet known. The 25-kDa molecule, as expected, was immunoprecipitated from ³⁵S-methionine–containing labeling medium by the anti-CNTF antibody (Fig. 8) , indicating the presence of a basal level of CNTF release by CE cells in response to injury (enucleation and corneoscleral explant dissection). The 25-kDa molecule was also present in the immune complex of anti-CNTFRα, which likely resulted from binding of 25-kDa CNTF to CNTFRα in the anti-CNTFRα immune complex. Whether the released 25-kDa CNTF can bind to the CNTFRα on the cell membranes or it forms a complex with the released CNTFRα and acts on cells that have no CNTFRα but express gp130 and LIFRβ in tissues bathed in the aqueous humor remains to be investigated. 

The present study also demonstrated that CNTF and CNTFRα can form a series of complexes with molecular mass of 100, 150, 175, 210, and 260 kDa (Figs. 2A 7) . It has been reported that both CNTF and CNTFRα can form dimers and that CNTF assembles a hexameric complex of two CNTFs, two CNTFRαs, one gp130, and one LIFRβ in vitro. ⁶³ The nature of the series of CNTF/CNTFRα complexes observed in the present study remains to be investigated. 

In conclusion, CE cells express CNTF, which may be an autocrine trophic factor that protects CE cells from H₂O₂ and other oxidative insults, because CE cells that have survived H₂O₂ stress can release CNTF and CNTFRα. H₂O₂ produced by the immune cells that have infiltrated the aqueous humor may serve as the signal for release of CNTF. 

 Supported in part by National Eye Institute Grant RO1EY11607, the Pangborn award from the University of Maryland at Baltimore, and a developmental grant to the Department of Ophthalmology from Research to Prevent Blindness.

 Submitted for publication November 1, 2001; revised April 3, 2002; accepted April 10, 2002.

 Commercial relationships policy: N.

 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.

 

The 25-kDa CNTF in CE cells from fresh bovine corneas. (A) Western blot analysis (WB) of CE cell extract, with a mouse monoclonal anti-human CNTF antibody (lane 1), a chick polyclonal anti rat-CNTF antibody (lane 2), and a goat polyclonal anti-human CNTF antibody (lane 3) in RIPA buffer. Each lane contained 15 μg protein from CE cell extract. (B) Western blot analysis (using the mouse monoclonal anti-human CNTF) of the immune complex derived from goat polyclonal anti-human CNTF antibody immunoprecipitation (IP) of CE cell extract in lysis buffer (650 μg protein; lane 1) and that of lysis buffer alone (lane 2). Except for experiments with the chick antibody (A, lane 2), all experiments were repeated at least three times with identical results.

Figure 2.

 

The 61-kDa CNTF-immunoreactive molecule synthesized and expressed by CE cells in corneal cups in corneoscleral explant cultures. (A) Autoradiography of electrophoresed immune complex derived from the goat polyclonal anti-human CNTF antibody–immunoprecipitated (IP+Autorad) ³⁵S-methionine–labeled CE cell extract in lysis buffer (450 μg protein). (B) Western blot analysis (WB) of the CE cell extract isolated from corneoscleral explant cultures, performed with the mouse monoclonal anti-human CNTF antibody. The lane contained 11 μg protein of CE cells extracted in RIPA buffer. The experiment was repeated three (A) and at least seven (B) times with identical results. M.W. Std., molecular mass standard.

Figure 3.

 

H₂O₂ concentration dependency of the decrease in the level of a 61-kDa CNTF-immunoreactive molecule in CE cells in H₂O₂-conditioned corneal cups. (A) Western blot (top) and optical densities (bottom) of the 61-kDa CNTF-immunoreactive bands depicted in the Western blot. (B) Reprobing of the Western blot in (A) with anti-α, β-, and γ-actin antibodies. H₂O₂ concentrations (in mM) in both (A) and (B): 0 (lane 1), 0.006 (lane 2), 0.012 (lane 3), 0.023 (lane 4), 0.045 (lane 5), 0.09 (lane 6), 0.18 (lane 7), and 0.35 (lane 8). Each lane contained 32 μg of protein of CE cells extracted in RIPA buffer. The experiment was repeated at least seven (A) and three (B) times with similar results. (C) Optical densities of the 61-kDa bands same as those depicted in (A) from seven separate experiments were averaged and plotted against the H₂O₂ concentrations. The difference was significant (P = 0.0007, ANOVA) among various groups. *Significant difference between the control and the H₂O₂-treated corneas (P < 0.05).

Figure 4.

 

Release of a 25-kDa CNTF-immunoreactive molecule by CE cells in H₂O₂-conditioned corneal cups increased in an H₂O₂-concentration–dependent manner. (A) Western blot of the conditioned medium (top) and optical densities of the 25-kDa CNTF-immunoreactive molecule (bottom). Corneal cups were conditioned with H₂O₂ at the concentrations described in Figure 3 . There was a significant difference (P = 0.0001, ANOVA) among various groups. Each lane contained one fourth of the volume of the concentrated conditioned medium collected from one corneal cup. The experiment was repeated at least five times with similar results. (B) Optical densities of the same 25-kDa bands as those depicted in (A) from five separate experiments were averaged and plotted against the H₂O₂ concentrations. *Significant difference between the H₂O₂-treated and control corneas (P < 0.05).

Figure 5.

 

Viable CE cells demonstrated H₂O₂-concentration–dependent decreases in the level of 61-kDa CNTF-immunoreactive molecule in corneal cups. (A) The number of red nuclei of the dead cells in a field of 600 CE cells is shown. The number of dead CE cells, averaged from 14 fields counted per cornea, was similar in corneas conditioned in 0, 0.14, 0.28, and 0.56 mM H₂O₂: (mean ± SEM, n = 4 corneas) 11 ± 2, 15 ± 3, 19 ± 4, and 16 ± 7, respectively (P = 0.7, ANOVA). The data represent the average of results in two separate experiments, each with two corneal cups examined for each designated H₂O₂ concentration. (B) Optical densities of the 61-kDa CNTF-immunoreactive bands depicted in the Western blot analysis (data not shown) decreased from the control (100%) to 51% ± 11%, 30% ± 3%, and 33% ± 9% in CE cells isolated from corneal cups conditioned in 0.14, 0.28, and 0.56 mM H₂O₂, respectively. There were significant differences (P = 0.000001, ANOVA) among various groups. Each lane contained 20 μg CE cell protein extract. *Significant difference between the control and the H₂O₂-treated corneas (P < 0.05).

Figure 6.

 

CNTFRα released from CE cells in H₂O₂-conditioned corneal cups in corneal explant cultures and present in CE cells from fresh bovine eyes. (A) The level of 53-kDa CNTFRα-immunoreactive molecule in the CE cell–conditioned medium increased in an H₂O₂-dependent manner. Each lane contained one fourth of the volume of the concentrated conditioned medium collected from one corneal cup. (B) Levels of the 53-kDa anti-CNTFRα immunoreactive molecules depicted in (A) from five separate experiments. The optical densities of the bands were normalized to those of the bands derived from the control corneal cups in each of the five experiments. *Significant difference between the control and the H₂O₂-treated corneas (P < 0.05; t-test). (C) The CNTFRα-immunoreactive molecule in the CE cell extract (15 μg protein) from fresh bovine eyes had a molecular mass of 53 kDa. The experiment was repeated at least five times with similar (A) and identical (C) results.

Figure 7.

 

³⁵S 61-kDa molecules in ³⁵S-methionine–labeled corneal cups in corneoscleral explant cultures. (A) Autoradiograph showing ³⁵S 61-kDa molecule immunoprecipitated in the 30-minute–conditioned medium in the corneal cups, with polyclonal anti-human CNTF (lanes 1 and 2) and polyclonal anti-human CNTFRα (lanes 3 and 4) antibodies. H₂O₂ in the conditioned medium (in mM): 0 (lanes 1 and 3) and 0.27 (lanes 2 and 4). (B) Blots showing immunoreactivity of the ³⁵S 61-kDa molecule. Using identical antibodies in paired immunoprecipitation and Western blot analysis, CNTF-immunoreactive ³⁵S 61-kDa molecule in CE cell extract (lane 1), the conditioned medium (lane 2), and CNTFRα-immunoreactivity of that in the conditioned medium (lane 3) were demonstrated. (C) Blot showing CNTFRα immunoreactivity of anti-CNTF–immunoprecipitated ³⁵S 61-kDa molecule in the CE cell extract. (A, B, C) Each lane contained immunoprecipitate from 225 μg CE cell protein extract or concentrated conditioned medium collected from two corneal cups and 1.5 μg IgG analyzed by nonreducing SDS-PAGE. Data shown were from one of two separate experiments.

Figure 8.

 

³⁵S 25-kDa CNTF-immunoreactive molecule detected in ³⁵S-methionine labeling medium placed in corneal cups for 8 hours. Autoradiographs showing coprecipitation of the 25- and 61-kDa molecules in the labeling medium by anti-CNTF (lane 1) and anti-CNTFRα (lane 2) antibodies and immunoblot analysis with the monoclonal anti-human CNTF antibody showing CNTF immunoreactivity of the 25-kDa molecule (lanes 3 and 4). Each lane contained immunoprecipitate from concentrated labeling medium collected from five corneal cups and 1.5 μg IgG.

The author thanks Timothy J. Coll for technical assistance.