Abstract
purpose. Ciliary neurotrophic factor (CNTF) is undergoing testing in human clinical trials to rescue degenerating retina, whereas studies show that the CNTF-binding α-subunit of the CNTF receptor (CNTFRα) is released from injured tissues. Here, the recombinant human (rh) CNTFRα was shown to restore functional CNTFRα in human corneal endothelial (CE) cells that lost endogenous CNTFRα during corneal storage
methods. In CE cells of stored human donor corneas, endogenous CNTFRα levels were quantified (by Western blot analysis), CNTF stimulation leading to the upregulation of connexin-43 was demonstrated, and the effectiveness of rhCNTFRα (8.3 nM) in augmenting the CNTF (0.83 nM) effect was tested. Paired human donor corneas were used as vehicle versus CNTF-treated or CNTF- versus (rhCNTFRα+CNTF)-treated (24 hours, 37°C), followed by analysis of CE cell connexin-43 mRNA and protein by semiquantitative RT-PCR and Western blot analysis, respectively. After 90-minute incubation with stored human corneas, rhCNTFRα incorporation into the CE membrane fraction was demonstrated by Western blot analysis
results. CE cell CNTFRα levels decreased as corneal storage time increased. CE cell connexin-43 mRNA levels in CNTF-treated and (rhCNTFRα+CNTF)-treated paired corneas averaged (mean ± SEM) 0.26 ± 0.08 and 0.58 ± 0.21, respectively (P = 0.029; eight pairs; storage time ≥25 days). rhCNTFRα augmentation was confirmed at the protein level. In corneas with short storage times (≤9 days) that retained abundant endogenous CNTFRα, rhCNTFRα decreased the effectiveness of CNTF. rhCNTFRα was incorporated into CE membranes
conclusions. rhCNTFRα acted as a surrogate to the lost endogenous membrane-bound CNTFRα in CNTF signaling, suggesting the potential of an adjuvant rhCNTFRα therapy in CNTF-therapy.
Ciliary neurotrophic factor (CNTF) was discovered in an extract of eye tissues consisting of ciliary body, iris, and choroid and was characterized as a survival factor for chick ciliary ganglion neurons.
1 2 It has since been shown to exert neurotrophic activities in various neuronal injury models, including axotomy-induced motor neuron degeneration and retinal ganglion cell apoptosis in vivo.
3 4 Recently, because the beneficial effects of CNTF treatment have been observed in a variety of animal models of photoreceptor cell degeneration,
5 a phase 1 (safety) human clinical trial was conducted in which encapsulated cells engineered to secrete CNTF were implanted into the vitreous cavity of the eyes of patients with retinitis pigmentosa.
6 Although the precise CNTF targets have not been identified, this therapy demonstrated a trend of beneficial effects on the visual outcome of the participants. Because CNTFRα, the CNTF-binding subunit of the CNTF receptor complex, has been localized to retinal pigment epithelial cells, rods and cones, inner nuclear cells, and retinal ganglion cells,
7 these cells are potential targets of CNTF therapy. However, to the best of our knowledge, the status of CNTFRα in the CNTF therapy-targeted tissues has never been addressed. It is a distinct possibility that CNTFRα is lost during the progression of degenerative eye diseases, which CNTF therapy is designed to treat, and the inclusion of CNTFRα in CNTF therapy may further improve the outcome. This possibility must be addressed given that phase 2 and 3 clinical trials are under way or have been conducted to test the efficacy of CNTF in the treatment of macular degeneration, retinitis pigmentosa, Usher type 2 and 3, and choroideremia.
8 9 10 11
Although the mechanism has not been established, studies describe the circumstances in which CNTFRα is lost from the tissues under stress. CNTFRα is lost from the fibers of the optic nerve after optic nerve injury, which may explain why CNTF is not effective in promoting regeneration of the injured optic nerve.
12 Although CNTF does not have a signal peptide sequence for secretion and has been postulated as an injury factor released only after injury
13 and CNTFRα lacks a transmembrane domain and is anchored to the cell membrane through a glycosylphosphatidylinositol anchor,
14 we have previously demonstrated that CNTF is released in a complex with CNTFRα by sublethal oxidative stress-injured corneal endothelial (CE) cells in corneal organ cultures.
15 In addition, CNTFRα is released by the skeletal muscle in response to peripheral nerve injury.
16 Furthermore, CNTFRα has been detected in cerebrospinal fluid samples of patients with degenerative diseases of the central nervous system,
16 and the level detected in the urine of patients with amyotrophic lateral sclerosis is 4.4-fold that of healthy persons.
17
Recombinant human (rh) CNTFRα has been shown to bind with CNTF at a 1:1 stoichiometry and transduces signal as the membrane-anchored CNTFRα.
18 CNTF in combination with CNTFRα (but not alone) transduces signals upregulating connexin-43 expression in glioma and astrocytes, cells that do not express CNTFRα.
19 20 The present study was designed to demonstrate that rhCNTFRα can restore in cells with diminished endogenous CNTFRα functional levels of CNTFRα, augmenting the effect of CNTF on these cells.
The corneal endothelium, a neural crest-derived tissue,
21 22 consists of a single cell layer of CE cells. CE cells in fresh human donor corneas express CNTFRα.
23 As a common clinical practice, human donor corneas for transplantation are stored in the preservation medium in the eye bank (for only a few days) before they are transplanted into recipient eyes. We have found in CE cells in stored human donor corneas a functional CNTF/CNTFRα signaling pathway that induces the expression of vasoactive intestinal peptide,
23 which is a trophic
24 and differentiation-maintaining factor of CE cells.
25 We have also found that the level of CE cell CNTFRα gradually diminishes during human donor cornea storage. The diminishing level of CNTFRα in stored human donor corneas in which the CE cells remain viable provides a unique opportunity to examine whether functional CNTFRα can be restored by rhCNTFRα.
Whereas CNTF/CNTFRα signaling leads to the upregulation of connexin-43 expression in astrocytes and in C6 glioma cells,
18 19 connexin-43 is expressed in the corneal endothelium of rat,
26 27 rabbit,
28 and human.
29 Recently, connexin-43 has been shown to play an important role in wound repair of the corneal endothelium.
30 In the present study we demonstrated that stimulation of the CNTF/CNTFRα signaling pathway in the corneal endothelium also upregulated connexin-43 expression and that, in the presence of rhCNTFRα, CE cells that lost endogenous CNTFRα during human donor cornea storage were more responsive to stimulation by CNTF.
Media used were as follows: medium A—Dulbecco’s modified Eagle’s medium (DMEM) supplemented with 100 U/mL penicillin, 100 μg/mL streptomycin sulfate, and 20 mM HEPES; medium B—DMEM supplemented with 100 U/mL penicillin, 100 μg/mL streptomycin sulfate, and 0.292 mg/mL l-glutamine; complete medium B—medium B plus 5% fetal calf serum and fungizone (250 ng/mL amphotericin; Invitrogen, Grand Island, NY).
Viable human corneoscleral explants (human donor corneas) stored in medium (Optisol-GS; Bausch & Lomb Surgical, Irvine, CA) at 4°C were obtained from the Lions Eye Institute for Transplant and Research (Tampa, FL). These corneas were deemed not suitable for transplantation because of their less than optimal CE cell densities and the advanced ages of the donors. In addition, using the same procedure as that used by the eye bank, fresh human donor corneoscleral explants were retrieved from cadavers (within 30 hours of death) in the Anatomy Board of the State of Maryland (Baltimore, MD). Cadavers were deidentified and were not considered human subjects by the Human Research Protection Office of the University of Maryland School of Medicine.
Paired human donor corneas were removed from the storage medium (Optisol-GS; Bausch & Lomb Surgical)-containing vials and incubated in 3.5 mL complete medium B in 35-mm Petri dishes for 24 hours at 37°C in 5% CO2-95% air. For testing the effect of CNTF on upregulating connexin-43 expression, complete medium B containing either no (right eye) or 0.83 nM rhCNTF (catalog no. 257-NT; R&D Systems, Minneapolis, MN; left eye) was used to treat the paired human donor corneas. For showing CNTFRα augmentation of the CNTF effect, complete medium B containing 0.83 nM CNTF (right eye) or 0.83 nM CNTF plus 8.3 nM rhCNTFRα (catalog no. 303-CR; R&D Systems; left eye) was used to treat paired human donor corneas.
Total RNA was isolated from CE cells scraped from human corneas (RNA-Bee; Tel-Test, Friendswood, TX) and subjected to reverse transcription (RT) using a kit (RETROscript; catalog no. 1710; Ambion, Austin, TX). RT products were subjected to PCR using the primer set corresponding to the gene sequences of human connexin-43 (5′-CCTTCTTGCTGATCCAGTGGTAC-3′ [forward] and 5′-ACCAAGGACACCACCAGCAT-3′ [reverse]).
31 Two negative controls were conducted: a reagent control in which the PCR reaction was conducted in the absence of the RT product and a PCR control in which RNA was not subjected to RT. PCR used RT products derived from 10% of isolated human RNA. Between an initial 94°C (2-minute) and a final 72°C (5-minute) treatment, 25 to 26 thermocycles (20 seconds at 94°C, 25 seconds at 54°C, and 40 seconds at 72°C) were conducted. The identity of the specific band of appropriate size (154 bp) revealed after agarose gel electrophoresis of the PCR product was proven by sequencing (Biopolymer Laboratory, University of Maryland) and was demonstrated to be identical with the sequence for human connexin-43. For quantifying mRNA levels, semiquantitative PCR was conducted in the presence of the18S primers and the 18S competomers (Universal 18S internal standards; catalog no. 1718; Ambion) in the ratio of either 2:8 or 3:7. RT-PCR products were electrophoresed (2% agarose gel), and the optical densities of bands were measured using a densitometer (NucleoVision; NucleoTech, San Carlos, CA).
Corneal endothelium was scraped from the corneas using a razor blade and homogenized in a glass homogenizer in the solubilization buffer containing 50 mM HEPES, pH 7.2, 10% glycerol, 1.25% triton X-100, 300 mM NaCl, 0.2 mM EGTA, 0.2 mM MgSO4, 100 μM sodium orthovanadate, and protease inhibitor cocktail (one tablet of Complete Mini [Roche Diagnostics, Mannheim, Germany] per 10 mL buffer).
CE cells scraped from the corneal endothelium with a razor blade were homogenized in a glass homogenizer in a buffer (10 mM Tris-HCl, pH 7.3 containing 5 mM MgCl2) at 4°C. The homogenate was centrifuged at 4°C for 10 minutes at 12,000g to remove unbroken cells and nuclei. Crude membrane fractions were prepared by centrifugation of the supernatants at 100,000g for 60 minutes. Final pellets were dispersed in 2× sample buffer for Western blot analysis.
Samples of CE cell extracts and membrane preparation were electrophoresed under reducing conditions using preformed Tris/glycine polyacrylamide gradient gels (NuPage; Novex, San Diego, CA) and electrophoretically transferred to nitrocellulose membranes. For the detection of CNTFRα, nitrocellulose membranes were immunostained with an affinity-purified goat anti-human CNTFRα primary antibody (catalog no. AF-303-NA; R&D Systems) and an anti-goat IgG-alkaline phosphatase conjugate secondary antibody (catalog no. 401512; Calbiochem, La Jolla, CA). CNTFRα on nitrocellulose membranes was detected by a chromogenic method using an alkaline phosphatase substrate solution made from tablets (Fast Red TR/Naphthol AS-MX; Sigma, St Louis, MO). For the detection of connexin-43 and actin (internal standard), chemiluminescence was used with affinity-purified rabbit anti-connexin-43 (catalog no. AB19012; Chemicon, Temecula CA) and mouse monoclonal anti-actin (catalog no. CP01; Ab-1; Calbiochem, La Jolla, CA) primary antibodies, horseradish peroxidase-linked anti-mouse and anti-rabbit IgG secondary antibody, and horseradish peroxidase substrate (ECL kit; Amersham Pharmacia, Piscataway, NJ).
We previously demonstrated the presence of CNTFRα (53 kDa) in CE cells from each of the 12 fresh human corneas from 12 donors with postmortem times of less than 24 hours.
23 In the present study, the status of CE cell CNTFRα in human donor corneas that had been in storage medium (Optisol-GS; Bausch & Lomb Surgical) for varying lengths of time was examined, with Western blot analysis and actin as the internal standard. Although persistently expressed in CE cells of stored corneas, CNTFRα levels in CE cells decreased as the time of storage increased
(Fig. 1A) .
Figure 1Bshows the normalized CNTFRα (against the actin internal standard) levels of nine corneas of nine human donors. There was a general correlation between the level of CNTFRα and the storage time of the corneas
(Fig. 1B) . However, there was no correlation between CNTFRα level and the age of the donor (data not shown). Furthermore, there was disintegration of the cell-cell adhesion in the corneal endothelial cell layer in corneas under long-term storage, as revealed by light microscopic examination of Alizarin red S (a dye that binds to the calcium-enriched cell-cell adhesion)-stained, flat-mounted corneas (data not shown).
The presence of CE cell connexin-43 mRNA was established and was followed by the demonstration of CNTF upregulation of connexin-43 mRNA expression in CE cells with endogenous CNTFRα. Total RNA isolated from pooled CE cells from six corneas of three donors was subjected to RT-PCR using the connexin-43 primer set. Gel electrophoresis of RT-PCR products demonstrated a specific band of the appropriate size (154 bp) in the positive and not in the two negative controls
(Fig. 2A) . Sequencing of the 154-bp band revealed it to be identical with the sequence for human connexin-43 (nucleotides 668–821; accession no. M65188). CNTF upregulation of connexin-43 mRNA expression was demonstrated by semiquantitative RT-PCR. As shown in
Figure 2B , one pair of human corneas that had been stored in storage medium (Optisol-GS; Bausch & Lomb Surgical) for 11 days and used as a control (right eye) compared with CNTF (0.83 nM)-treated (left eye) demonstrated the effect of CNTF in increasing the connexin-43 mRNA level.
rhCNTFRα Augmentation of CNTF-Upregulated Connexin-43 mRNA Expression in CE Cells of Human Donor Corneas Stored in Storage Medium
We have hypothesized that the loss of CNTFRα from CE cells during extended storage of the human donor corneas in storage medium (Optisol-GS; Bausch & Lomb Surgical;
Fig. 1 ) can be compensated by rhCNTFRα. Thus, rhCNTFRα was expected to augment the effect of CNTF in upregulating connexin-43 expression in CE cells in human donor corneas in long-term storage. As shown in
Figure 3A , semiquantitative RT-PCR of connexin-43 mRNA demonstrated that CE cells in (CNTFRα+CNTF)-treated human donor corneas (storage time, 33 days) expressed higher connexin-43 mRNA levels than those in CNTF-treated paired corneas. In eight pairs of human donor corneas (storage time, 25 days and longer), the normalized connexin-43 (against the 18S internal standard) mRNA levels in the CE cells in CNTF-treated corneas and those in the (CNTFRα+CNTF)-treated paired corneas averaged (mean ± SEM) 0.26 ± 0.08 and 0.58 ± 0.21, respectively (
P = 0.029;
Fig. 3B ).
CNTFRα augmentation of CNTF upregulated connexin-43 expression in CE cells, as confirmed at the protein level by Western blot analysis with the use of actin as the internal standard.
Figures 4A and 4Bdemonstrated that in the CNTF (0.8 nM) treatment of one pair of human donor corneas in storage for 23 days, the presence of CNTFRα (8.3 nM) increased CNTF effectiveness to 168% that observed in the absence of CNTFRα.
The possibility that rhCNTFRα competed with endogenous CNTFRα for binding with CNTF in CE cells that retained abundant endogenous CNTFRα was examined. In two pairs of human donor corneas in short-term storage for 9 and 3 days (therefore retaining endogenous CNTFRα), rhCNTFRα (8.3 nM) decreased the effectiveness of CNTF (0.83 nM) in upregulating connexin-43 protein levels to 65% and 68% those observed in paired corneas treated with CNTF alone, respectively
(Figs. 4C 4D and 4E 4F) .
Increased CE Cell Membrane-Bound CNTFRα Found in Long-term Stored Human Donor Corneas after Short-term Incubation with rhCNTFRα
Increased CE cell membrane CNTFRα levels resulting from short-term (90-minute) incubation with rhNTFRα was demonstrated in human donor corneas in storage medium (Optisol-GS; Bausch & Lomb Surgical) for 40 days and longer. As shown in
Figure 5 , the membrane fractions isolated from paired human donor corneas in storage medium (Optisol-GS; Bausch & Lomb Surgical) for 40 (two pairs), 41 (three pairs), and 42 (two pairs) days before treatment with CNTF (right eye) versus (CNTFRα+CNTF) (left eye) were subjected to Western blot analysis for CNTFRα. Results demonstrated a higher CNTFRα level in CE cell membrane fractions from (CNTFRα+CNTF)-treated corneas than that from CNTF-treated paired corneas
(Figs. 5A 5B) . The normalized CNTFRα (against a 71-kDa molecule revealed by Ponceau stain as the most abundant protein molecule in the membrane fraction) levels in the CE cells of all 14 corneas demonstrated that the presence of CNTFRα in the CNTF-incubation increased the membrane-bound CNTFRα levels from (mean ± SEM) 1.80 ± 0.12 to 5.75 ± 1.00 (
P = 0.001;
Fig. 5B ).
On binding of CNTF to the endogenous membrane-bound CNTFRα, the two β subunits (gp130 and LIFRβ) of the receptor dimerize, forming a trimeric receptor complex that activated the Janus family of tyrosine kinases Jak1/Jak2, leading to tyrosine phosphorylation, dimerization, and nuclear translocation of signal transducer and activator of transcription STAT3, which then bound to specific responsive elements in the promoter regions of the CNTF-responsive genes.
32 33 CNTF-responsive genes included genes of vasoactive intestinal peptide
34 35 and connexin-43.
18 19 We have recently demonstrated CNTF induction of vasoactive intestinal peptide in CE cells of human donor corneas preserved in storage medium (Optisol-GS; Bausch & Lomb Surgical).
23
In mature astrocytes, which do not express CNTFRα, CNTF and the recombinant CNTFRα in combination induce connexin-43 expression in a JAK/STAT pathway-dependent manner.
20 Nevertheless, the mechanism of CNTF in combination with the recombinant CNTFRα-transducing signals through the JAK/STAT pathway has not been established. In the present study, CNTF and the rhCNTFRα might have formed a complex before modulating the two β subunits (gp130 and LIFRβ), similar to that of IL-6 and its soluble receptor (sIL-6R), forming a complex to stimulate cells that lack the endogenous membrane IL-6 receptor.
36 In contrast, rhCNTFRα might have functioned as a surrogate to the membrane-bound CNTFRα and transduced the CNTF signal because (in the absence of CNTF) rhCNTFRα can bind to LIFRβ and to gp130 in vitro.
37 38 Regardless of the mechanism, the insertion of rhCNTFRα in the membrane was involved because rhNTFRα became membrane bound after 90-minute incubation with the endogenous CNTFRα-depleted CE cells
(Fig. 5) .
According to the specification and use manual provided by the manufacturer, the rhCNTFRα (324 amino acids) used in the present study was expressed in Sf21 insect cells and was glycosylated, resulting in a protein with an apparent molecular size of 52 kDa, which was similar to that of the endogenous CNTFRα (53 kDa) found in CE cells.
23 The glycosylphosphatidyl inositol anchor of the endogenous CNTFRα
14 is also the result of posttranslational modification adding the anchor to the protein.
39 It is, therefore, likely that glycosylphosphatidyl inositol posttranslation modification also took place in the insect cell-produced rhCNTFRα, allowing it to be inserted in the cell membrane. This notion was in agreement with our finding that rhCNTFRα became enriched in the membrane fraction after only 90 minutes of incubation with endogenous CNTFRα-depleted CE cells
(Fig. 5) . Although future studies will be necessary to determine whether insertion of the rhCNTFRα in the membrane requires the presence of CNTF, studies of other glycosylphosphatidyl inositol-anchored proteins indicate that these proteins by themselves can be exogenously incorporated onto the cell membrane.
39 Transfer of glycosylphosphatidylinositol-anchored proteins (CD55 and CD59) contained in vesicles shed by normal erythrocytes in blood stored for transfusion to erythrocytes deficient in these proteins in patients have been demonstrated in vivo.
40 It has also been demonstrated that CD59 in proteasomes, when transferred to CD59-deficient erythrocytes, retains its function and protects erythrocytes against complement-mediated hemolysis.
41
In human donor corneas that had been stored for a short period (and that retained endogenous CNTFRα), the rhCNTFRα, while present in large excess over CNTF (10:1), actually decreased the effectiveness of CNTF
(Figs. 4C 4D 4E 4F) , suggesting that rhCNTFRα competed with the endogenous CNTFRα for CNTF and that the complex of rhCNTFRα/CNTF was not as effective as the endogenous CNTFRα/CNTF complex in signal transduction. As a result, the endogenous CNTFRα level was inversely related to the effectiveness of rhCNTFRα in further increasing the CNTF-upregulated connexin-43 level.
Finally, the mechanism of CNTFRα loss during human donor corneal storage is unknown. In blood stored for transfusion, it has been demonstrated that levels of CD59 and another glycosylphosphatidylinositol-anchored, membrane-associated complement regulatory protein, the decay-accelerating factor (DAF), gradually decline during storage.
42 Loss of glycosylphosphatidylinositol-anchored proteins is mediated by enrichment of these proteins in vesicles and the release of these vesicles from erythrocytes in stored blood.
43 44
In conclusion, the present study demonstrated that rhCNTFRα can act as a surrogate to the lost endogenous membrane-bound CNTFRα in transducing the CNTF signal, suggesting an adjuvant role for the rhCNTFRα in CNTF therapy.
The authors thank Dante Gloria for excellent technical support.