Attenuation of Experimental Proliferative Vitreoretinopathy by Inhibiting the Platelet-Derived Growth Factor Receptor

Abstract

purpose. The work from numerous laboratories has led to the idea that the growth factors such as platelet-derived growth factor (PDGF) contribute to proliferative vitreoretinopathy (PVR) in experimental models of the disease, as well as in humans. In support of this idea, the authors have previously reported that cells unable to respond to PDGF had a greatly reduced PVR potential, compared with PDGF-responsive versions of the same cells. The goal of this study was to test the effect of blocking the output of the PDGF receptor in an experimental model of PVR.

methods. Polymerase chain reaction–based site-directed mutagenesis was used to generate point mutations in the human PDGF α receptor (αPDGFR) cDNA, which resulted in single amino acid substitutions. These changes were based on naturally occurring point mutations in the c-kit receptor tyrosine kinase, which suppresses the function of wild-type c-kit. A truncated αPDGFR was also made, in which the receptor ended just after the juxtamembrane domain. As with the point mutants, truncated receptors have been shown to block the action of wild-type receptors. All the αPDGFR mutants were introduced into cells that naturally express the wild-type receptor, and the PDGF-dependent output of the resultant cell lines was determined. In addition, the PVR potential of cell lines expressing the mutant receptors was tested in a PVR rabbit model.

results. Although the mutants differed in their ability to suppress PDGF-dependent signaling of the wild-type receptor, each mutant effectively blocked cell cycle progression. When expressed in rabbit conjunctival fibroblasts, a cell line that effectively induces PVR, the mutant receptors blocked PVR to various degrees. The most effective receptor was the truncated mutant.

conclusions. These data suggest that the αPDGFR plays an important role in PVR. In addition, these mutant receptors appear to have therapeutic potential for prevention of this blinding disease.

Proliferative vitreoretinopathy (PVR) is the major cause for failed retinal detachment surgery. Some of the events thought to contribute to pathogenesis include migration of the retinal pigment epithelial (RPE) cells and retinal glial cells (Müller cells) and synthesis of extracellular matrix molecules such as collagen. ¹ The fibrous material often contracts, and this tractional force causes retinal detachment with or without a rhegmatogenous component. It has been reported that 20% to 40% of the patients with PVR fail to achieve the final anatomic success, ² and this occurs in 5% to 10% of the patients who undergo retinal reattachment surgery. Complicated retinal detachment associated with trauma or uveitis has a higher risk for this disease. 

Growth factors such as transforming growth factor (TGF)-β ^{3 4} and platelet-derived growth factor (PDGF) ^{5 6 7 8 9} are believed to play an important role in promoting the events that contribute to PVR. Other growth factors such as hepatocyte growth factor, ¹⁰ basic fibroblast growth factor, or interleukin-6 ^{4 8} have also been implicated. We have recently reported that cells unable to respond to PDGF induce PVR poorly and that the PVR potential increases substantially when they are made responsive to PDGF by expression of the PDGF receptor. ¹¹ This finding strongly suggests that PDGF is an important growth factor in at least an experimental model of PVR. 

PDGF is a potent mitogen for fibroblasts and induces DNA synthesis and chemotaxis and sometimes serves as a survival factor. Two PDGF genes have been identified, and they encode the PDGF-A and PDGF-B chains. Biologically active PDGF is either a homo- or heterodimer; therefore, there are three kinds of combinations, PDGF-AA, -AB, and -BB. The receptor for PDGF is a homo- or heterodimer of the α and β subunits. The receptor subunits differ in their affinity for ligand, and, hence, the composition of receptor subunits is in part dependent on the isoform of PDGF. For instance, PDGF-AA only binds to αα homodimer, -AB to αα homo- or αβ heterodimer, and –BB to any subunit combination. In the studies described herein, we focus on the PDGF α receptor (αPDGFR), which is a homodimer of the α subunits and can be assembled by any of the three PDGF isoforms. PDGF dimerizes the αPDGFR, leading to activation of the receptor’s tyrosine activity, which is encoded in the intracellular domain of the receptor. Activation of the receptor’s kinase is a prerequisite for subsequent signal relay and biological responses. 

The c-kit receptor belongs to the same family of tyrosine receptor kinases as the αPDGFR; and like the αPDGFR it has an extracellular domain, transmembrane domain, juxtamembrane domain, and tyrosine kinase that is interrupted by a kinase insert. ^{12 13 14} Single point mutations in c-kit are responsible for the deficits of W³⁷, W^v, W⁴², and W⁴¹ strains of mice. The abnormalities include white spotting on the skin, infertility, stem cell deficiency, and anemia. W³⁷ has a substitution of Glu to Lys at position 582 (juxtamembrane domain); W^v has Met instead of Thr at 660 (first half of the kinase domain); Asp replaces Asn at 790 (second half of the kinase domain) in W⁴²; and W⁴¹ has a Val to Met substitution at 831 (second half of the kinase domain). ^{15 16} The affected mice were heterozygous for the mutations, suggesting that the mutant form of c-kit was dominant to the normal copy of c-kit, which was also expressed. In fact, all the c-kit receptor mutants have been shown to function as dominant negatives in a mast cell proliferation assay. ¹⁵ Mice do not survive when both c-kit alleles harbor either the W³⁷ or W⁴² point mutant. ¹⁷  

The point mutants in c-kit lie in regions of the intracellular domain that are highly conserved within this class of receptor tyrosine kinases. When the single amino acid substitution corresponding to W³⁷ was introduced into Xenopusα PDGFR, the resultant mutant blocked αPDGFR-dependent events during development. Furthermore, this mutant prevented PDGF-dependent tyrosine phosphorylation of the wild-type (WT) αPDGFR when the mutant and WT receptor were coexpressed. ¹⁸ These findings suggest that making the W mutations in the αPDGFR would yield a panel of mutants able to block activation of the WT αPDGFR. 

A widely used approach to make dominant negative receptor tyrosine kinase is to truncate the kinase domain. Such a truncated receptor heterodimerizes with a WT receptor and prevents activation of the WT receptor. An important distinction between the truncated receptors and point mutants in the W series is the level of expression needed to effectively block signaling of the WT receptor. For the PDGFR, the WT receptor was silenced by a 90-fold overexpression of the truncated receptor. ¹⁹ In contrast, a 4-fold overexpression of theα PDGFR point mutant corresponding to W³⁷ was sufficient to block activation of the WT receptor. ¹⁸ In the W mice, even a single copy of the mutant was sufficient to have an effect, because the W phenotype was seen in heterozygotes. Hence, the W panel of the receptor mutants may be more potent than truncated mutants. 

Here we generated a panel of αPDGFR mutants that either had a single amino acid substitution, which corresponded to the W³⁷, W^v, W⁴², or W⁴¹ c-kit mutants, or was truncated to eliminate the kinase domain and carboxyl terminus. Each of the mutants was expressed in cells that naturally express theα PDGFR and proved to block PDGF-dependent entry into S phase. We also tested the effect of expressing these mutants in rabbit conjunctival fibroblasts (RCFs), which can induce PVR in rabbit eyes when coinjected with platelet-rich plasma (PRP). ²⁰ Importantly, the cells with these mutants were impaired to induce PVR. 

Methods

Cells

NIH 3T3 cells were cultured and maintained in Dulbecco’s modified Eagle’s medium (DMEM; GIBCO–BRL, Grand Island, NY) with 10% fetal bovine serum (FBS; Gemini Bio Products, Calabasas, CA) supplemented with antibiotics. Primary RCFs were isolated from rabbit conjunctiva and maintained as previously described, ²⁰ except that instead of amphotericin B and gentamicin, we used 500 U/ml of penicillin G and 500 μg/ml of streptomycin as the antibiotics supplement. 

Polymerase Chain Reaction Mutagenesis

The truncated receptor was generated as follows. The 4.3-kb NotI/XbaI cDNA fragment containing the α/β chimeric receptor ²¹ was subcloned into pBlueScript SK (Stratagene, La Jolla, CA). This construct was cut with SacII and XbaI, and the liberated cDNA fragment (which contains all of the βPDGFR sequence) was discarded. The remaining fragment was treated with Klenow to blunt-end the DNA and then religated. The 2.0-kb NotI/SalI αPDGFR fragment was subcloned into pLXSN^{2 21} that had also been cut with this enzyme pair. The protein encoded by this portion of theα PDGFR cDNA includes all of the extracellular, transmembrane, and juxtamembrane domains and has a stop codon at nucleotide 1972, with no change in the predicted amino acid sequence. The last amino acid of the truncated receptor is proline 589, in the sequence “DSRWEFP,” and is near or at the juxtamembrane/kinase domain junction. 

Because the protein sequence surrounding the mutated amino acid in the W³⁷, W^v, W⁴², and W⁴¹ in c-kit receptors was highly conserved, we made the corresponding substitution in the αPDGFR. More specifically, Glu to Lys at the position 587 for W³⁷, Thr to Met at 665 for W^v, Asp to Asn at 818 for W⁴², Val to Met at 858 for W⁴¹ (Fig. 1A ). The mutants were generated using a polymerase chain reaction (PCR)–based strategy and the template was 18G generated from 18F. The 18F was made by subcloning the 3.5-kb wild-type human αPDGFR cDNA into pBlueScript II SK+ (Stratagene) using NotI/BamHI site. ²² We generated 18G from 18F by introducing SacII site at 1975, which is a unique site for this construct. ²¹ The PCR-generated mutants were subcloned into 18G as an NcoI/SacII fragment (E587K), as a SacII/StuI fragment (T665M), or as a StuI/SphI fragment (D818N and V858M). The sequence of the point mutants and the truncated receptor was confirmed by sequencing DNA. 

Retrovirus Expression System

The mutated αPDGFR cDNAs were subcloned into pLHDCX² retroviral vector using the NotI/SalI site. This vector has a modified multiple cloning site containing NotI-BglII-SalI-HindIII driven by cytomegalovirus promoter. The vector also encodes a histidinol-resistant gene, which is driven by the long terminal repeat (LTR) promoter. Purified DNA (25 μg) was transfected into 293 GPG replication-incompetent retrovirus-producing cells ²³ using lipofectamine (GIBCO–BRL) according to the manufacturer’s instruction, and virus in the supernatant was collected from days 3 through 8. The virus was concentrated by centrifugation at 30,000g at 4°C for 90 minutes and resuspended in TNE solution (50 mM Tris–HCl, pH 7.8, 130 mM NaCl, 1 mM EDTA), and it was stored at −70°C until use. To express the mutant receptors, NIH 3T3 cells and RCFs were incubated overnight with the virus harboring mutated αPDGFR or empty vector in the presence of 4 μg/ml of polybrene in DMEM with 10% FBS. The infected cells were passaged into new dishes and cultured in DMEM with 10% FBS supplemented with 5 mM histidinol (Sigma, St. Louis, MO). Mass population of drug-resistant cells was used in the experiments. Fluorescent based cell sorting (FACS) analysis of the E587K, T665M, D818N, and V858M mutant receptor–expressing RCFs indicated a single population of receptor-expressing cells. Two populations of truncated receptor–expressing cells were detected by FACS, and the higher expressing population was sorted and collected, and these cells were used for the PVR studies shown in Figure 6 . 

Immunoprecipitation and Western Blot Analyses

NIH 3T3 cells and RCFs were grown to 80% confluence and then incubated for 20 hours in DMEM containing 0.1% FBS and 0.4 mg/ml bovine serum albumin (BSA). Cells were exposed at 37°C for 5 minutes to either 50 ng/ml of PDGF-AA or left unstimulated with buffer. Cells were washed with H/S (20 mM Hepes, pH 7.4, 150 mM NaCl) twice and then lysed in EB (10 mM Tris–HCl, pH 7.4, 5 mM EDTA, 50 mM NaCl, 50 mM NaF, 1% Triton X-100, 0.1% BSA, 20 μg/ml aprotinin, 2 mM Na₃VO₄, 1 mM phenylmethylsulfonyl fluoride) or RIPA buffer (150 mM NaCl, 10 mM NaPO₄, 2 mM EDTA, 1% sodium deoxycholate, 1% Nonidet P-40 [NP-40], 0.1% sodium dodecyl sulfate, 20 μg/ml aprotinin). ²⁴ Lysates were centrifuged for 15 minutes at 13,000g, the pellet was discarded, and the soluble fraction was used as the total cell lysate. The protein concentration was measured using protein assay kit (Pierce, Rockfield, IL) following the manufacturer’s instructions. 

Receptors were immunoprecipitated from the soluble fraction with the 27P or 292 antibody. Immune complex was bound to formalin-fixed membranes of Staphylococcus aureus, spun through an EB sucrose gradient, and washed twice with EB, then with PAN (10 mM Pipes, pH 7.0, 100 mM NaCl, 1% aprotinin) + 0.5% NP-40, and, finally, with PAN. The samples were resuspended in PAN before using for kinase assay or Western blot analysis. 

Total cell lysates containing 20 μg of protein or receptor immunoprecipitates from 1.0 × 10⁶ cells were resolved in 7.5% SDS—polyacrylamide gel electrophoresis (PAGE) gel under reducing conditions. Proteins were transferred onto Immobilon (Millipore, Bedford, MA). Membranes were blocked using Block (10 mg/ml BSA, 10 mg/ml ovalbumin, 0.05% Tween-20, dissolved in Western Rinse; 8 mM Tris-HCl, 2 mM Tris-base, pH 7.5, 150 mM NaCl) for anti-phosphotyrosine blotting. The membranes were blocked in Blotto (10 mg/ml nonfat dry milk, 0.05% Tween-20 in Western Rinse) for other blotting. Membranes were incubated with primary antibodies for 1 hour at room temperature and washed 5 times with Western Rinse. Consequently, they were incubated with secondary antibody for 1 hour at room temperature and washed 5 times in Western Rinse as well. Finally, all blots were visualized using ECL (Amersham Pharmacia Biotech, Piscataway, NJ). 

Antibodies

The 27P antibody is a crude polyclonal rabbit antibody raised against a glutathione S-transferase (GST) fusion protein, including the human αPDGFR carboxyl terminus (amino acids 951–1089). The 80.8 antibody was raised against a GST-fusion protein, including a portion of the first immunoglobulin domain (amino acids 52–94) of humanα PDGFR. These antibodies recognize the human and mouse αPDGFR. The 292 antibody is a mouse monoclonal antibody that specifically recognizes primate αPDGFR; therefore, this antibody does not recognize the endogenous receptor in NIH 3T3 cells. The Ras GTP–activating protein (RasGAP) antibody is a crude rabbit antisera against the SH2-SH3-SH2 of the human RasGAP (69.3). 4G10 and PY20 are mouse monoclonal anti-phosphotyrosine antibodies, purchased from Upstate Biotechnology (Lake Placid, NY) or Transduction Laboratories (San Diego, CA). The phospho-extracellular signal related kinase (Erk) rabbit polyclonal antibody was purchased from New England Biolabs (Beverly, MA). For Western blot analysis the following dilutions were used for each primary antibody: αPDGFR, 27P:80.8 (1:1), 1:1000; anti-phosphotyrosine, 4G10:PY20 (1:1), 1:5000; 69.3, 1:4000; anti–phospho-Erk, 1:1000. Secondary antibodies were horseradish peroxidase–conjugated goat anti-rabbit or anti-mouse antibodies (Amersham Pharmacia Biotech) diluted 1:5000. 

In Vitro Kinase Assay

Mutant αPDGFRs were selectively immunoprecipitated with the 292 antibody, and samples representing 2 × 10⁵ cells were subjected to an in vitro kinase assay. Immunoprecipitates were preincubated with 2 μg of GST protein for 10 minutes at 0°C, then 2 μg of GST–phospholipase C (PLC)-γ, 10 μCi ofγ -[³²P] ATP (DuPont–NEN Research Products, Boston, MA), and universal kinase buffer (20 mM Pipes, pH 7.0, 10 mM MnCl₂, 20 μg/ml aprotinin) were added. The samples were incubated at 30°C for 5 minutes, the proteins were separated by 7.5% SDS–PAGE, the gel was dried, and the radiolabeled protein was detected by autoradiography. 

DNA Synthesis Assay

NIH 3T3 cells were trypsinized, resuspended in DMEM with 10% FBS, and plated at a density of 5 × 10⁵ cells/well in a 24-well tissue culture plate and cultured overnight. The cells were rinsed twice with phosphate-buffered saline (PBS), 0.5 ml of DMEM with 0.1% FBS and 0.4 mg/ml of BSA (Sigma) was added, and the cells were incubated for 48 hours. The cells were then exposed to 50 ng/ml of PDGF-AA, 10% FBS (vol/vol), or buffer for 22 hours, after which time the cells were pulsed for 4 hours in DMEM with 10% FBS containing 0.8 μCi/ml of [³H]-thymidine (DuPont–NEN Research Products). Finally, the cells were washed twice with PBS, washed once with 5% trichloroacetic acid, and lysed in 250μ l of 0.25 N NaOH. The lysates were transferred into scintillation tubes containing 50 μl of 6 N HCl, and then 3 ml of scintillation fluid (ICN Biochemical) was added. The incorporated radioactivity was quantitated in a scintillation counter (Packard, Meriden, CT). The data were expressed as fold induction, which was calculated by dividing stimulated samples by the buffer control. Each condition was assayed in triplicate, and the mean ± SD was obtained. 

Rabbit Model for PVR

PVR was induced in the rabbit eyes as previously described. ²⁰ Briefly, gas vitrectomy was performed by injecting 0.4 ml of perfluoropropane (C₃F₈) into the vitreous cavity 4 mm posterior to the corneal limbus under anesthetic conditions. Three days later, the rabbits were anesthetized and the pupils were dilated. Then 0.1 ml DMEM containing 1 × 10⁵ of RCFs expressing empty vector or αPDGFR mutant was injected into the vitreous cavity together with 0.1 ml of PRP using a 30-gauge needle. Ten rabbits underwent surgery for each group, with the cells expressing the empty vector or the truncated, E587K, T665M, D818N, V858M αPDGFR mutant. One rabbit in the D818N group died just after the surgery; therefore, the total number was 9 for this group. The retinal status was evaluated with an indirect ophthalmoscope fitted with a +30 D fundus lens at days 1, 4, 7, 14, 21, and 28 after the surgery. The PVR was graded according to the Fastenberg score from 0 through 5. ²⁵ All surgeries were performed under aseptic conditions and pursuant to the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. Only the left eye of each rabbit was used for the experiments. 

Statistical Analysis

To determine whether the differences among groups of rabbits were statistically significant, we performed the Mann–Whitney U test for nonparametric ordinal data. The response of rabbits injected with empty vector–expressing cells were compared with the response of those injected with mutant receptor–expressing cells. In all cases, P < 0.05 was considered significant. 

Results

Expression of αPDGFR Mutants

The first goal was to make and characterize a panel of αPDGFR mutants capable of blocking the output of the WT αPDGFR. To this end, we made the five mutants shown in Figure 1A . All five were expressed in NIH 3T3 cells using the replication-incompetent retroviral approach described in the Methods section. NIH 3T3 cells were chosen as the cell line for the characterization studies because it is a well-characterized cell line that responds mitogenically to PDGF-AA (see Fig. 4 below). A Western blot analysis of total cell lysates using a combination of αPDGFR antibodies that recognizes both endogenous and introduced receptors indicated that each of the mutants were expressed two- to fourfold above the level of the endogenous WTα PDGFR (Fig. 1B) . The smaller size of the truncated receptor is expected, because it lacks the kinase domain and carboxyl terminus. Because only one of the antibodies recognizes the truncated receptor (it lacks the 27P epitope), the level of expression of this receptor may be underestimated. We conclude that all five of the mutant receptors were successfully constructed and stably expressed. 

PDGF-Dependent Signaling in Cells Expressing the Mutant Receptors

The next step was to test PDGF-dependent tyrosine phosphorylation of the αPDGFR, which is an early required event in the signaling cascade. Cells expressing an empty vector or the αPDGFR mutants were arrested by serum deprivation and then left resting or stimulated with PDGF-AA for 5 minutes. The cells were lysed and the receptors were immunoprecipitated using 27P, an antibody that recognizes both the introduced mutant and endogenous WT receptors, except for the truncated receptor which is lacking the epitope seen by this antibody. Anti-phosphotyrosine Western blot analysis of these samples indicated that PDGF triggered the expected increase in the phosphorylation content of the WT receptor (Fig. 2A , lane “EMP”). In contrast, the WT receptor was poorly phosphorylated in cells expressing the truncated receptor, even though there were comparable amounts of WT receptor recovered in the“ TRUNC” and “EMP” samples (Fig. 2A) . The phosphotyrosine signal of the immunoprecipitated receptor was not inhibited in any of the other cells. This is probably at least in part because some of these mutants retain kinase activity (see Fig. 3 ) and were immunoprecipitated with the 27P antibody. 

We also investigated PDGF-dependent activation of Erk in our panel of cell lines. Erk is a member of the mitogen-activated protein (MAP) kinase pathway that is activated by the αPDGFR, as well as many other receptors. Indeed, stimulation of the control cells (EMP) resulted in enhanced phosphorylation of Erk (Fig. 2B) , which is a commonly used indicator of activation. Similar to the effect on tyrosine phosphorylation of the receptor, expression of the truncated receptor greatly diminished PDGF-dependent activation of Erk (Fig. 2B) . Cells expressing the D818N mutant also failed to fully activate Erk in response to PDGF, whereas Erk activation in cells expressing the other point mutants was unaffected or even enhanced in some experiments (Fig. 2B and data not shown). We conclude that the truncated receptor efficiently blocks PDGF-dependent signaling events, whereas the point mutants either had no effect or effects were only partially inhibited. 

To better characterize the kinase activity of the point mutants we immunoprecipitated them using the 292 monoclonal antibody, which selectively recognizes an extracellular epitope of the introduced receptor and recognizes all the mutants used in this study. Although PDGF-stimulation is expected to dimerize mutant and WT receptors, lysing cells in RIPA buffer breaks receptor dimers. ²⁶ Consequently, 292 immunoprecipitates are not expected to contain a coimmunoprecipitating WT receptor. NIH 3T3 cells expressing the empty vector or receptor mutants were left resting or stimulated with 50 ng/ml of PDGF-AA for 5 minutes, the cells were lysed in RIPA buffer, and the resultant samples were subjected to anti-phosphotyrosine and anti-αPDGFR Western blot analyses. In this series of experiments, we included the previously described Fα cell line, which is an NIH 3T3–like cell line that expresses the introduced humanα PDGFR. ¹¹ This cell line was included as a positive control, because the WT receptor in NIH 3T3 cells is mouse, and not recognized by the 292 antibody. As expected, the αPDGFR was immunoprecipitated from the Fα cells, but not the NIH 3T3 cells expressing the empty vector, and PDGF stimulation increased the phosphotyrosine content of the receptor (Fig. 3A) . PDGF promoted tyrosine phosphorylation of three of the four point mutants. There was no detectable basal or PDGF-stimulated tyrosine phosphorylation of the truncated and D818N αPDGFR receptor mutant. 

We also tested the ability of immunoprecipitated receptors to phosphorylate an exogenous substrate. The WT αPDGFR phosphorylated the exogenous substrate and there was a modest enhancement of this activity when the receptor was immunoprecipitated from PDGF-stimulated cells (Fig. 3B) . In contrast, the substrate was not phosphorylated by the truncated or D818N receptor. The kinase activity of the E587K mutant was comparable to the WT receptor, whereas the T665M and V858M mutants were more active than the WT receptor. We conclude that the truncated and D818N mutants appear to be kinase dead and that the E587K is comparable to WT, whereas T665M and V858M are activated as kinases. Note that the behavior of these αPDGFR mutants is similar, although not identical, to that of the analogous c-kit receptor mutants, in which kinase activity was lowest in W⁴² (corresponds to D818N) and W³⁷ (E587K), W^v (T665M) was intermediate, and W⁴¹ (V858M) was the best, although still below the WT levels. ¹⁵  

PDGFR Mutants Inhibit PDGF-AA–Dependent DNA Synthesis

The next question we addressed is whether these mutants inhibit PDGF-dependent biological responses. We focused on cell cycle progression, which can readily be monitored in NIH 3T3 cells in a DNA synthesis assay. Cells were plated in 24-well dishes, arrested by serum deprivation, then tested for their ability to enter the S phase after exposure to PDGF. As shown in Figure 4 , cells expressing an empty vector responded to PDGF-AA, and the magnitude of the response was typically at least 50% of the response seen when cells were stimulated with 10% FBS. In all the other cell lines PDGF-AA failed to induce a robust response, whereas each of the cell lines did respond normally to serum. The somewhat elevated response to serum in the V858M cells was not routinely observed. We conclude that expression of the mutant receptors selectively blocks PDGF-dependent cell cycle progression, whereas it has little effect on the response of these cells to serum. 

PDGFR Mutants Attenuated Experimental PVR in Rabbits

Data presented thus far indicated that the mutant receptors were capable of blocking PDGF-dependent cellular responses such as cell cycle progression. Consequently, we wanted to test whether they could prevent PVR, which we have recently found is dependent on PDGF in a rabbit model of the disease. ¹¹ To avoid species variables, we switched from the mouse NIH 3T3 cell line to primary RCF, which has previously been used extensively with this PVR model. 

An empty vector or each of the αPDGFR mutants was introduced into fourth passage RCF, and mass populations of drug-resistant cells were obtained. Western blot analysis of the resultant cell lines indicated that the cells do indeed express endogenous αPDGFR, and the introduced receptor was expressed at a 6- to 10-fold higher level (Fig. 5) . The truncated receptor was expressed at least 30-fold over the endogenous receptor. Because mass populations of the cells were used, we also tested the heterogenicity of the population with respect to receptor expression. FACS analysis, using the 292 monoclonal antibody, indicated that there was a single population of receptor-expressing cells for all the samples, except the truncated receptor. In this case, there were two populations, and we sorted the cells to obtain a single population of high expressors, which were used in the analysis shown in Figure 5 . These experiments show that the αPDGFR mutants were expressed in RCF and that the cells used in our experiments were homogenous with respect to receptor expression. 

To compare the PVR potential of RCFs expressing an empty vector to the cells expressing αPDGFR mutants we performed the following experiments. For each of the six experimental groups, 9 to 10 rabbits were first subjected to gas compression vitrectomy and then 1 × 10⁵ cells were coinjected with 0.1 ml of PRP. The PVR grade was evaluated on days 1, 4, 7, 14, 21, and 28. Under these experimental conditions PVR is induced rapidly, such that 30% of the rabbits injected with cells expressing the empty vector underwent total retinal detachment (stage 5) by day 4. By day 7, 100% of the rabbits in the control group had reached the most severe form of the disease. In contrast, PVR was less severe in all the experimental groups (Fig. 6 , top panel) at day 7. At this time point the truncated receptor appeared to be the best in preventing PVR. The experiment was extended for 3 more weeks, during which time PVR worsened in all experimental groups. However, in all cases there was a statistically significant difference in severity of the disease between the animals injected with cells expressing αPDGFR mutants versus cells of the control group. As with the earlier time point, the truncated receptor was the best in preventing PVR. None of the rabbits in this group achieved stage 5, and 80% remained at stage 1 or below. We conclude that the mutantα PDGFRs are capable of significantly attenuating experimental PVR. 

Discussion

We have constructed and characterized a series of αPDGFR mutants. Although the mutants differed in their intrinsic kinase activity and potential to prevent PDGF-dependent signaling, they were all effective in blocking PDGF-stimulated cell cycle progression. Furthermore, these receptor mutants were able to prevent PVR, and the truncated receptor was the most effective. 

One reason why the truncated receptor was more effective in blocking PVR might be because it was expressed to a higher level than the other receptors. To investigate this possibility we compared PDGF-dependent signaling in cells expressing high or low levels of the point mutants. Unlike the truncated receptor, increasing the expression level of the point mutants did not further block PDGF-dependent responses (authors’ unpublished observations, July 1999). These findings suggest that the higher level of expression of the truncated receptor may not be the reason it was the most effective in blocking PVR. The key difference may relate to the step in signaling cascade at which the mutant receptors inhibit. The truncated receptor seems to block at a very early step, whereas the point mutants appear to have an effect further downstream in the signaling cascade. 

The mechanism by which the truncated αPDGFR inhibits DNA synthesis is thought to be by competitive inhibition of ligand binding to the WT receptor. ²⁷ To be effective, this type of dominant negative receptor typically needs to be expressed at levels that greatly exceed the level of the WT receptor. In contrast, mice heterozygous for the W mutants of c-kit display a phenotype, ^{15 16 17} indicating that these mutants are effective when expressed at levels comparable to those of the WT receptor. This suggests that the point mutants block receptor function by a mechanism that is different from that of the truncated receptor. The idea that the truncated and point mutants used in this study function by different mechanism is further supported by the finding that some of the point mutants are kinase active, whereas the truncated receptor is not (Fig. 3) .The mechanism by which the point mutants block PDGF-dependent cellular responses, as well as the possibility that there may be more than one mechanism of action within the group of 4 point mutants, is actively under investigation. 

Given that a key step in activation of the WT receptor is engaging its kinase activity, it is a bit puzzling how a kinase active receptor could have a negative influence on the overall response of a cell to the growth factor. However, other investigators have also shown that kinase active receptor tyrosine kinase mutants are dominant negative. The Y845F mutant of epidermal growth factor receptor (EGFR) has full kinase activation but inhibits EGF- and serum-dependent mitogenesis. ²⁸ Another example is the thanatophoric dysplasia (TD) II mutant of fibroblast growth factor receptor (FGFR). The TDII mutant is constitutively active, and it causes cell cycle arrest by activating Stat 1 and consequent upregulation of cell cycle inhibitor p21^waf1/cip1. ²⁹ The T665M and V858M αPDGFR mutants, which have elevated kinase activity (Fig. 3) , may be mimicking these ways to inhibit the endogenousα PDGFR-dependent cell cycle progression. 

Our data that inhibition of αPDGFR can reduce the PVR score strongly suggest that αPDGFR is a critical contributor in this experimental model of PVR. This is somewhat surprising granted that the RCFs are likely to have receptors for many growth factors. Furthermore, the PRP that is coinjected with the RCFs is a rich source of serum growth factors. The idea that the αPDGFR is important for experimental PVR is consistent with our previous findings, in which a different approach was used to address this question. ¹¹ In this study we found that the PVR potential of cells that lack PDGFR is low and that expressing the αPDGFR significantly enhances the ability of such cells to induce PVR. Taken together, our findings make a strong case for the idea that the αPDGFR is contributing to experimental PVR. 

One possible mechanism by which the αPDGFR promotes the events culminating in PVR is by driving cell proliferation. However, the observation that the panel of mutants were equipotent in blocking PDGF-dependent DNA synthesis (Fig. 4) , but differed in their ability to prevent PVR (Fig. 6) , suggests that there are processes in addition to cell proliferation that are required for PVR. In fact, other investigators have found that even irradiated cells, which fail to proliferate, can induce PVR in animal eyes, provided that a sufficient number of cells is injected. ²⁵ Our studies support those previous findings that PVR requires more than just cell proliferation. Further characterization of the panel of αPDGFR mutants may help in identifying the other cellular processes required for PVR. 

Other than cell proliferation, what are other cellular processes that are likely to be contributing to PVR? In humans, it seems that cell migration is an important component in PVR, because the cells of the epiretinal membranes have arrived from distant locations. In the commonly used PVR rabbit that was used in this series of experiments, cell migration may not be an important issue because the cells are injected into the final anatomic location. An additional contributing factor to PVR is likely to be secretion of extracellular matrix (ECM), which constitutes a major portion of the membrane in experimental and clinical PVR. Whether the αPDGFR is particularly capable of triggering ECM production is an area of active investigation. In addition, the αPDGFR may be part of a more elaborate network of growth factors. For instance, TGF-β is able to promote PVR, ^{3 4} and TGF-β upregulates PDGF-AA secretion andα PDGFR activation. ^{30 31} PDGF-AA also has a synergistic effect on TGF-β–associated collagen synthesis. ^{32 33} It may be interesting to investigate the relationship between αPDGFR and TGF-β, especially in terms of collagen and/or ECM synthesis. Finally, contraction of the epiretinal membrane or vitreous makes a major contribution to retinal detachment in later stages of PVR, and PDGF enhances the contraction of fibroblasts, ³⁴ and RPE cells. ^{35 36 37}  

The establishment of experimental PVR that is dependent on the αPDGFR and the availability of αPDGFR signaling mutants provide an opportunity to begin to define the signaling enzymes that are involved with PVR. For instance, an αPDGFR mutant that is unable to bind or activate the Src family of kinases is better than the WT receptor in the early phase of experimental PVR. ³⁸ In the present study, we found a positive correlation between activation of Erk signals (Fig. 2) and PVR scores at 7 days (Fig. 6 ; P = 0.035, by Spearman rank correlation test). Thus, in addition to identifying important cellular responses that contribute to PVR, we are also actively investigating the signaling enzymes that participate in these cellular responses and in PVR. Finally, PVR may be a simple model for other fibrotic diseases such as atherosclerosis and kidney fibrosis, ^{39 40} and so our PVR findings may shed light on the cellular process that contribute to other diseases as well. 

The present study points to the αPDGFR as a target for the prevention of PVR. Several approaches have been developed to block PDGFR activation. These include a variety of ways to prevent the ligand from interacting with the receptor: antibodies to the ligand ⁴¹ or receptor ⁴² and peptides that compete with the ligand for binding to the receptor. ^{43 44} Others have focused on ways to inhibit activation of the receptor’s kinase activity, and developed antibodies that block receptor dimerization, ⁴⁵ or drugs that inhibit the receptor’s enzymatic activity (AG1296). ⁴⁶ Elucidation of the signaling enzymes that theα PDGFR uses to promote PVR will further expand the targets and available drugs to treat PVR. Finally, we believe that gene therapy is also viable approach to prevent PVR, and our success with the ex vivo approach described in this article encourages us to try to develop an in vivo, gene therapy–based treatment for PVR. 

 Supported in part by the SERI Ocular Gene Therapy Program.

 Submitted for publication January 26, 2000; revised April 24, 2000; accepted May 22, 2000.

 Commercial relationships policy: N.

 

Figure 1.

 

(A) Schematic of αPDGFR mutants. The schematic of theα PDGFR indicates the last amino acid of the truncated receptor and the position of each point mutation. A stop codon terminates the truncated receptor such that the cytoplasmic domain encodes only the juxtamembrane domain. For the other mutations the amino acid sequence of the WT receptor is indicated in the top line, and the amino acid change present in the mutant receptor is shown in the bottom line. The numbers indicate the position of each of the mutations. TM, transmembrane domain; TRUNC, truncated receptor; E587K, glutamic acid at position 587 was changed to lysine; T665M, threonine at position 665 was changed to methionine; D818N, aspartic acid at position 818 was changed to asparagine; and V858M, valine of position 858 was changed to methionine. (B) Expression of the mutants in NIH 3T3 cells. NIH 3T3 cells infected with a replication-incompetent retrovirus harboring the αPDGFR mutant indicated in (A) or empty vector (EMP) were tested for expression of the introduced and endogenous receptors. Serum-starved cells were lysed, and 20 μg of the protein was subjected to Western blot analysis for αPDGFR (top panel) or Ras GTP-activating protein (RasGAP; lysate control, bottom panel). Two distinct sizes of αPDGFR were detected at around 190 kDa (full-length receptor, top arrow) and approximately 120 kDa (truncated receptor, bottom arrow). The point mutants migrated slightly faster than endogenous WT receptor. The data shown are representative of three independent experiments.

Figure 2.

 

PDGF-dependent signaling is most effectively inhibited by the truncated receptor. (A) Tyrosine phosphorylation of the WT and mutant receptors. NIH 3T3 cells expressing a mutant receptor or the empty expression vector (EMP) were serum-starved overnight and exposed to either buffer (−) or 50 ng/ml of PDGF-AA (+) for 5 minutes. The cells were lysed and immunoprecipitated with the 27P antibody, which recognizes both the endogenous mouse and introduced human αPDGFR mutants. Immunoprecipitates representing approximately 1 × 10⁶ cells were subjected to an anti-phosphotyrosine (P-Y) Western blot analysis (top panel). After stripping off the antibodies used in the first blot experiments, Western blot analysis with an anti-αPDGFR antibody was performed (bottom panel). In both panels, the arrow indicates the mature αPDGFR, whereas the arrowhead points to the immature form of the receptor. Three independent experiments showed similar results. IP, immunoprecipitation; IB, immunoblot (Western); the abbreviations for the receptor mutants are detailed in the legend of Figure 1 . (B) PDGF-dependent Erk activation. Cells were stimulated as described in (A), 20 μg of total cell lysate was subjected to Western blot analysis using anti–phospho-Erk (P-Erk, top panel) or anti–Ras GTP–activating protein (RasGAP; lysate control, bottom panel) antibody. Arrowheads point to the p44 and p42 forms of Erk, both of which are activated by growth factors such as PDGF. Similar results were obtained in three independent experiments.

Figure 3.

 

Kinase activity of the mutant receptors. (A) Tyrosine phosphorylation of the mutant PDGFRs. NIH 3T3 cells expressing an empty vector (EMP) or the indicated mutant receptors were serum-starved and exposed to either buffer (−) or 50 ng/ml of PDGF-AA (+) for 5 minutes. F cells expressing the human αPDGFR (Fα) ¹¹ were processed in parallel and served as a positive control. Cells were lysed and immunoprecipitated with the 292 antibody, which is primate-specific, and recognizes an extracellular epitope. This antibody selectively recovers the mutant receptor. Immunoprecipitates representing approximately 1 × 10⁶ cells were first subjected to anti-phosphotyrosine (P-Y) Western blot analysis (top panel), the blots were stripped and then reprobed with an anti-αPDGFR antibody (bottom panel). The data presented are representative of three independent experiments. IP, immunoprecipitation; IB, immunoblot (Western); the abbreviations for the receptor mutants are detailed in the legend of Figure 1 . (B) Kinase activity of the mutant receptors. The mutant receptor, immunoprecipitated with the 292 antibody, was incubated with 2 μg of GST–PLCγ, an exogenous substrate, and[ ³²P]-γ ATP in an in vitro kinase assay, as described in the Methods section. The proteins were resolved by 7.5% SDS–PAGE, and the gel was dried and exposed to film. A portion of the resultant autoradiogram is shown, and the arrowhead indicates the position of GST–PLCγ. In at least three independent experiments, the T665M and V858M mutants showed substantially elevated kinase activity, compared with either the other mutants, or the WT receptor.

Figure 4.

 

All mutants block PDGF-dependent DNA synthesis. NIH 3T3 cells expressing either empty vector (EMP) or the indicated mutant were serum-starved for 48 hours and then PDGF-AA (50 ng/ml), FBS (10%), or buffer was added for 22 hours. Finally, [³H]-thymidine was added for 4 hours, the cells were harvested, and the amount of[ ³H]-thymidine incorporated was measured by scintillation counting. Data are expressed as a ratio of stimulated to unstimulated samples, and the mean ± SD of triplicates are presented. In at least three experiments, we consistently found that the PDGF-dependent response was greatly reduced or completely eliminated in cells expressing any of the mutant receptors. The abbreviations for the receptor mutants are detailed in the legend of Figure 1 .

Figure 5.

 

Expression of the mutants in RCFs. The empty expression vector (EMP) or the indicated receptor mutant was introduced into primary RCFs. The resultant cells were serum-starved overnight and lysed and 20 μg of protein was subjected to an anti-αPDGFR (upper panel) or anti–Ras GTP–activating protein (RasGAP; lower panel) Western blot analysis. Arrowheads point to the mature and the truncated receptors. IB, immunoblot (Western); the abbreviations for the receptor mutants are detailed in the legend of Figure 1 .

Figure 6.

 

All the mutants attenuated experimental PVR. The panel of cell lines shown in Figure 5 were compared for their ability to induce PVR in a rabbit model of the disease. After gas compression vitrectomy, 1 × 10⁵ cells were coinjected with 0.1 ml of PRP, and disease progression was monitored with an indirect ophthalmoscope on days 1, 4, 7, 14, 21, and 28. The Fastenberg classification was used to score the disease, and the results on days 7 and 28 are shown. Each circle represents an individual (there were 9 or 10 in each group), and the horizontal bar indicates the mean of the group. Statistical analysis was performed using the nonparametric Mann–Whitney U test, and the probability values are indicated. One hundred percent of the rabbits injected with the control group developed stage 5 PVR by day 7, whereas disease progression was slower in all the groups injected with cells expressing the mutant receptors. The severity of PVR was increased in all the experimental groups during weeks 1 through 4; however, a statistically significant difference in PVR was observed between control and experimental groups even at the end of the experiment. The probability values between the control (EMP) and experimental group are indicated. The abbreviations for the receptor mutants are detailed in the legend of Figure 1 .

The authors appreciate the constructive discussions from members of the Kazlauskas laboratory.