Selective Killing of RPE with a Vascular Endothelial Growth Factor Chimeric Toxin

Abstract

purpose. To determine the sensitivity of retinal pigment epithelial (RPE) cells to a vascular endothelial growth factor (VEGF) chimeric toxin.

methods. A targeted toxin was developed using recombinant methods to fuse VEGF₁₆₅ to the diphtheria toxin (DT) translocation and enzymatic domain (DT₃₉₀-VEGF₁₆₅). Human RPE cells, choroidal endothelial cells (CECs), and scleral fibroblasts were isolated, and a dose–response for DT₃₉₀-VEGF₁₆₅ was determined by measurement of cell proliferation and cell number. In parallel experiments, cultures were pretreated with transforming growth factor (TGF)-β₂. VEGF-receptor (VEGFR-1 and -2) expression was determined using reverse transcription–polymerase chain reaction and fluorescence-activated cell sorting, and affinity was measured using Scatchard analysis.

results. RPE cells and CECs were similarly prone to killing by the VEGF-toxin, but scleral fibroblasts were unaffected. Pretreatment with TGF-β₂ selectively increased the sensitivity of RPE cells to the VEGF-toxin. RPE cells expressed both VEGFR-1 and -2 in vitro; however, the expression of VEGFR-1 was very low. Pretreatment with TGF-β₂ (10 ng/ml) was associated with increased expression of the VEGFR-1 in RPE cells and increased receptor affinity for VEGF detected by Scatchard analysis.

conclusions. Dose-dependent killing of RPE cells by the DT₃₉₀-VEGF₁₆₅ conjugate is selectively enhanced by pretreatment with TGF-β₂. This study provides further strong support for the presence of functional VEGFRs on human RPE cells, and demonstrates for the first time the ability to target a normal nonendothelial cell type through VEGFR expression.

Vascular endothelial growth factor (VEGF) is an angiogenic, permeability-inducing factor that functions by binding two distinct high-affinity receptors, VEGF receptor (VEGFR)-1 (flt-1) and VEGF receptor (VEGFR)-2 (flk-1 or KDR), that are expressed predominantly on endothelial cells. Each of these receptors has tyrosine kinase activity, but each mediates its functions through a somewhat different signal transduction pathway. ^{1 2} The recent localization of VEGFR on cell types other than endothelial cells has suggested other functions for VEGF. ^{2 3} In the retina, VEGFR mRNAs have been identified on retinal pigment epithelial (RPE) cells in vivo and in vitro. ^{3 4 5} The coexistent secretion of VEGF by RPE cells in vitro has suggested the possibility that VEGF is an autocrine growth factor for activated RPE cells. ³ Further support for this contention has come from immunohistochemical studies that show increased expression of VEGF and of VEGFR-1 and -2 on RPE cells in nonvascular epiretinal membranes from patients with proliferative vitreoretinopathy (PVR). ⁴  

A number of strategies have been developed to inhibit growth factor–activated cells, including the use of soluble receptors, inhibitors of signal transduction, and antibody-mediated toxin delivery to target the expression of specific receptors. ^{2 6} Proteins conjugating diphtheria toxin (DT) to VEGF have recently been used to target the toxin specifically to cells expressing a high density of VEGFR. ^{7 8} We hypothesized that activated RPE cells could be targeted by such an approach. 

The purpose of this study was to determine whether RPE cells could be selectively killed by a DT-VEGF conjugate. Because both transforming growth factor-β₂ (TGF-β₂) and platelet-derived growth factor (PDGF)-BB are commonly found in PVR membranes, ⁴ the effect of these cytokines on VEGFR expression and the targeting of RPE cells by the DT-VEGF conjugate was also determined. 

Methods

Isolation and Culture of Cells

RPE cells were isolated from human fetal eyes (>22 weeks of gestation) obtained from the Anatomic Gift Foundation (Woodbine, GA). The RPE cells were dissociated using 2% dispase (Gibco, Grand Island, NY), nylon mesh, and gentle pipetting and were cultured in Dulbecco’s modified Eagle’s medium (DMEM; Irvine Scientific, Santa Ana, CA) with 10% fetal bovine serum (FBS) and 1% penicillin-streptomycin and glutamine at 37°C in a humidified incubator saturated with 5% CO₂ and 95% air. The second and third passages were used for all experiments. Epithelial origin was confirmed by immunohistochemical staining for cytokeratin using a pan-cytokeratin antibody (Sigma, St Louis, MO). The cells were tested and found free of contaminating macrophages (anti-CD11, Sigma) and endothelial cells (anti-von Willebrand factor, Sigma). 

The same eyes were used to isolate scleral fibroblasts. Fibroblasts were cultured in DMEM with 20% FBS and 1% penicillin-streptomycin and glutamine. Cells were passaged and expanded after reaching confluence. The medium was exchanged every 3 days. Second-passage human fibroblasts were used for all experiments. 

Human choroidal endothelial cells (CECs) were isolated as previously described. ⁹ CECs were chosen as the control endothelial cell population, because they are an ocular endothelial cell type that responds well to VEGF stimulation and can be isolated and grown from human fetal eye specimens in relative abundance. Cells were confirmed to be vascular endothelial cells by positive immunostaining for von Willebrand factor and by binding of dil-acetylated low-density lipoprotein (LDL). Epithelial contamination was excluded by staining for cytokeratin. 

VEGF-Toxin Production

VEGF₁₆₅ fusion protein containing 390 amino acids of DT made up of the enzymatic and translocation domains was produced as a tripartite fusion protein with glutathione-S-transferase, as previously described. ⁸ The sample was passed over a glutathione column to remove the glutathione-S-transferase domain, and purified proteins were analyzed on sodium dodecyl sulfate gels. Western blot analysis of the recombinant protein confirmed reactivity with specific antibodies to VEGF and DT. ⁸  

Cytotoxicity Studies

Subconfluent cells were seeded at a density of 1 × 10⁴ cells/well and treated with DT₃₉₀-VEGF₁₆₅ (0.1, 1, 10, 50, and 100 ng/ml) for 3 days. [³H]thymidine uptake and cell counting were used to determine effects on cell proliferation and survival. Viability was determined by exclusion of 0.4% trypan blue dye. In parallel experiments, cells were pretreated with 10 ng/ml TGF-β₂ (Genzyme, Cambridge, MA) for 1 to 5 days before treatment with the VEGF toxin. 

Reverse Transcription–Polymerase Chain Reaction

For reverse transcription–polymerase chain reaction (RT-PCR) RPE cells were cultured to subconfluence in DMEM + 10% FBS, grown overnight in DMEM + 1% FBS, and then stimulated with or without growth factor in low serum conditions for 48 hours. Total RNA was extracted from untreated RPE cells and from RPE cells pretreated with TGF-β₂ (1 or 10 ng/ml) and PDGF (10 ng/ml). RT-PCR was performed as previously described ¹⁰ but modified to use ready-to-go PCR beads with hot-start conditions (Pharmacia Biotech, Piscataway, NJ). Primers were synthesized from the coding region of the human VEGFR-1 and VEGFR-2 genes by the University of Southern California Norris Microchemical facility. ¹⁰ Amplification products specific for VEGFR-1 and VEGFR-2 mRNA measured 498 bp and 709 bp, respectively. Samples were amplified for 35 cycles and resolved on a 1% agarose gel. Loading was equalized by adding equal amounts of cDNA to each lane and by comparing amplification products for the housekeeping gene glyceraldehyde 3-phosphate dehydrogenase (results not shown). 

Flow Cytometry

Cell surface expression of VEGFR-1 and -2 was determined by labeling cells with polyclonal antibodies (Santa Cruz Biotechnology, Santa Cruz, CA), with or without 3 to 5 days of TGF-β₂ pretreatment. Nonspecific binding was identified using a polyclonal anti-glial fibrillary acidic protein antiserum (Sigma). Cells were detached using a 0.25% solution of EDTA and phosphate-buffered saline (PBS; Sigma). Nonspecific binding was blocked with 5% goat serum for 10 minutes (Fig. 4C) . The cells were then incubated with primary antibody for 1 hour (4°C), washed with PBS, and labeled with fluorescein-isothiocyanate–labeled anti-mouse IgG (Vector, Burlingame, CA) for 1 hour (Fig. 4C) . Cells were washed, fixed with 1% paraformaldehyde, and analyzed by fluorescence-activated cell sorting using flow cytometry (FACScan with Consort 30 software; Becton–Dickinson; Mountain View, CA). 

Scatchard Plot

Scatchard analysis was performed as previously described. ¹¹ RPE cells were seeded at a density of 1 × 10⁵ cells per well in 24-well plates and treated for 3 or 5 days with 10 ng/ml TGF-β₂. The cells were washed twice with Hanks’ balanced salt solution and incubated with the desired concentration of iodinated VEGF₁₆₅ (1–2 ng/ml; Amersham, Arlington Heights, IL). Nonspecific binding was determined in the presence of an excess of unlabeled VEGF. All binding studies were performed at 4°C. The radioactivity was measured in a gamma counter, and specific and nonspecific binding was calculated. 

Statistical Analysis

All experiments were performed at least in triplicate. Analysis of variance was performed on all data. All results with P < 0.05 were considered statistically significant. 

Results

Effect of DT-VEGF on RPE Cells

There was a significant inhibition of[ ³H]thymidine uptake by RPE cells (median effective dose [ID₅₀] 82 ng/ml) treated with DT₃₉₀-VEGF₁₆₅ in the range of 50 to 100 ng/ml. As predicted by their known expression pattern for VEGFR, CEC showed significant inhibition of[ ³H]thymidine uptake in a dose–response manner after treatment with DT₃₉₀-VEGF₁₆₅, (ID₅₀ 85 ng/ml), but scleral fibroblasts were unaffected (Fig. 1) . Unconjugated DT (DAB₃₉₀) had no effect on any of the cell types, which was consistent with its having no binding domain (results not shown). Cell-counting experiments confirmed that the inhibition of proliferation was a result of cytotoxicity. RPE cell counts were decreased by 44% and by 56%, after 3 days of treatment with 50 ng/ml and 100 ng/ml of conjugate, respectively. 

After pretreatment with TGF-β₂ (10 ng/ml) for 5 days, RPE cells showed a significantly increased sensitivity to the VEGF-toxin (Fig. 1 ; P < 0.05, ID₅₀ 22 ng/ml). There was no such increase in sensitivity for the CECs, and scleral fibroblasts remained unaffected. The RPE cells had became sensitive to the VEGF-toxin at previously ineffective concentrations (P < 0.05, 10 ng/ml). 

Expression of VEGFR

RPE cells express VEGFR-2 mRNA even when unstimulated by TGF-β₂ (Fig. 2 ; bottom; lane 1). VEGFR-1 mRNA expression was not seen using unstimulated RPE cells under these conditions (Fig. 2 ; top; lane 1) but was present at very low levels when identified by Northern blot analysis using an internal probe (results not shown). TGF-β₂ treatment induced VEGFR-1 mRNA expression that was weak at 1 ng/ml (Fig. 2 ; top; lane 2) and strong at 10 ng/ml (Fig. 2 ; top; lane 3). The intensity of the strong VEGFR-1 band was confirmed by varying the number of amplification cycles: a positive band was seen with as few as 20 cycles of amplification (results not shown). Treatment with TGF-β₂ did not affect the mRNA expression of VEGFR-2 (Fig. 2 ; lower panel). Pretreatment of RPE cells with PDGF-BB (10 ng/ml) did not induce VEGFR-1 mRNA expression and did not appear to alter VEGFR-2 mRNA expression significantly (Fig. 2 , lane 4). 

To determine whether increased mRNA expression for VEGFR-1 after TGF-β₂ pretreatment was associated with increased surface expression of the receptor, we examined the RPE cells by flow cytometry, using specific polyclonal antibodies against the VEGFR. Although no apparent change in VEGFR-2 surface expression was seen (results not shown), there was a prominent increase in both intensity and number of cells expressing VEGFR-1 (Fig. 3) . 

Scatchard Analysis of TGF-β₂–Treated and Untreated RPE Cells

The binding of iodinated recombinant VEGF₁₆₅ to RPE cells was compared with cells that had been pretreated with 10 ng/ml TGF-β₂ for 3 and 5 days. Untreated RPE cells demonstrated a monophasic VEGF binding curve with a dissociation constant of 4400 pM (Fig. 4A 4B ). Three-day TGF-β–treated RPE cells also showed a linear binding curve with a dissociation constant of 1000 pM, whereas 5 days of treatment led to a biphasic binding curve with a dissociation constant of 220 pM (Fig. 4C 4D) . Analysis of the biphasic Scatchard plots found in the 5-day TGF-β–treated samples suggested the expression of two distinct high-affinity receptor binding sites on RPE cells resulting in an overall higher affinity of the VEGFR for VEGF. 

Discussion

VEGF is the prototype member of an expanding family of angiogenic polypeptides that also includes placenta growth factor, VEGF-B, VEGF-C, and VEGF-D. ^{1 2} VEGF acts through at least two tyrosine kinase receptors that are widely regarded as restricted almost exclusively to endothelial cells ^{1 2} ; however, RPE cells have recently been shown to express VEGFR, both in vivo and in vitro, and to respond to VEGF stimulation. ^{3 4 5} The data presented herein further support the contention that RPE cells express VEGFR that actively binds VEGF and that this receptor expression can be modulated and used to selectively target RPE cells. 

Cytokines and antibodies conjugated with translocation and enzymatic domains of bacterial toxins have been studied to target various cell types and tumors. ^{7 8} Use of a truncated DT in which the native receptor-binding domain is chemically conjugated to a cytokine or is replaced by recombinant methods with a synthetic gene encoding for a cytokine can allow for purification of a chimeric toxin that binds only to cells bearing the corresponding receptor. ^{7 8} Internalization of the chimeric protein by receptor-mediated endocytosis occurs only in cells expressing high levels of receptor. Passage through an intracellular acidic compartment results in cytotoxicity by inhibition of adenosine diphosphate (ADP)-ribosylation of elongation factor 2. ^{7 8} A DT-VEGF conjugate has been developed to target the vasculature of tumors. ⁷ This conjugate was effective at selectively killing endothelial cells in vitro and blocking basic fibroblast growth factor (bFGF)–induced angiogenesis in vivo. ⁷ More recently, a DT-VEGF chimeric protein has been developed that demonstrates significantly higher activity against tumor angiogenesis than the chemical conjugate. ⁸ Our study clearly shows that a VEGF chimeric toxin can be used to target RPE cells as well as ocular endothelial cell populations. 

The differential functions mediated by each of the high-affinity VEGFR is uncertain in endothelial cells and even more unclear in RPE cells. ^{1 2 3} VEGFR-2 is thought to be involved in the mitogenic and chemotactic response of endothelial cells to VEGF, whereas VEGFR-1 may be primarily involved in chemotaxis and cellular differentiation. ^{1 2} VEGFR-1 also exists in a soluble form that can exert a dominant negative effect on VEGFR-2 signal transduction ¹ ; whether TGF-β increases secretion of soluble VEGFR-1 in RPE cells is unknown. The increased activity of the toxin conjugate on the TGF-β₂–treated RPE cells suggests an increase in affinity for the VEGFR. Consistent with this hypothesis, TGF-β treatment led to increased mRNA and surface protein expression of the higher affinity VEGFR-1, and Scatchard analysis confirmed the presence of two high-affinity binding sites with a higher overall receptor affinity. There is currently very little information reported about regulation of VEGFR expression in RPE cells. A recently reported semiquantitative analysis of cytokine and cytokine receptor mRNA expression in RPE cells by RT-PCR ⁵ showed much higher expression of VEGFR-2 than of VEGFR-1, a result that was similar to ours. Another study appeared to show higher VEGFR-1 expression in native RPE cells than in cultured cells ⁴ ; but whether this is due to confluence, culture conditions, or some other factor is unknown. In endothelial cells, VEGFR-2 expression is decreased by TGF-β ¹² ; however, our results did not show any significant alterations in VEGFR-2 mRNA or surface protein expression after treatment with TGF-β₂. Of particular interest is the significant increase in VEGFR-1 mRNA and surface protein expression in RPE cells. The time course of this expression suggests the involvement of an intermediary factor; but PDGF-BB did not produce a similar effect. The effect of TGF-β₂ is likely to be of clinical relevance, because it is abundantly expressed in the membranes and pathologic vitreous of patients with PVR. ⁴ Although there is only limited immunohistochemical data about localization of VEGFR on normal RPE cells in situ, studies of PVR membranes suggest that the migrating RPE cells show increased expression of VEGFR and therefore may be targets for such a conjugate. ⁴  

If such a conjugate were to be used in vivo to target RPE cells in nonvascular membranes, the possibility of a cytotoxic action on normal retinal endothelial cells (RECs) must be considered. Although RECs were not studied, previous reports have shown the presence of a similar range of high-affinity VEGFRs in endothelial cells of diverse origin, suggesting that RECs may behave similarly to CECs in these assays. ¹¹ Isolation of sufficient numbers of RECs from our human samples precluded complete study; however, preliminary experiments showed a very similar response. Localization of VEGFRs in normal primate eyes by in situ hybridization indicates expression in the vessels of the inner retina raising concern about the toxicity of such an approach. ¹³ A minimal number of receptors is necessary to mediate sufficient internalization of conjugate to result in cytotoxicity, suggesting that resting cells with low receptor number may be much less affected than the activated target cells. ^{7 8} Previous studies have shown that only proliferating cells are sensitive to the action of cytotoxic conjugates and that this differential susceptibility may be related to changes in the endocytic pathway. ⁷ The absence of endothelial cell proliferation in PVR membranes should protect these vascular cells from the effects of the conjugate. In particular, the increased sensitivity to VEGF-toxin induced by TGF-β₂ in RPE cells but not in CECs or fibroblasts suggests that in the TGF-β₂–rich environment of the pathologic vitreous, there should be even more specificity of action toward the RPE cells. 

RPE cells play a critical role in the cytokine network of the retina, both in its ability to synthesize and its ability to respond to these signaling polypeptides. The regulation of VEGFR mRNA and surface protein expression by TGF-β₂ demonstrates how the response of a specific cell is profoundly influenced by its environment. The killing of RPE cells using a VEGF-toxin provides further strong support for the presence of functional VEGFRs on RPE cells. This study demonstrates for the first time the ability to target a normal non–endothelial cell type through VEGFR expression and suggests a novel approach to the targeting of specific ocular cell populations in patients with proliferative ocular disorders. 

 Supported by Grant EY-01545 and by Core Grant EY-03040 from the National Institutes of Health. The Department of Ophthalmology is the recipient of an award from Research to Prevent Blindness and the Hoover Foundation.

 Submitted for publication September 3, 1999; revised January 20, 2000; accepted February 29, 2000.

 Commercial relationships policy: N.

 

Figure 1.

 

Human RPE cells, CECs, and scleral fibroblasts were examined by[ ³H]thymidine uptake after 3 days in culture with DT₃₉₀-VEGF₁₆₅ (0.1–100 ng/ml). (A, C, and E) Untreated cells; (B, D, and F) cells pretreated with TGF-β₂ for 5 days.[ ³H]thymidine uptake in cells without DT₃₉₀-VEGF₁₆₅ is normalized to a value of 100.

Figure 2.

 

RT-PCR analysis (35 cycles) of RPE cells for mRNA expression of VEGFR-1 and -2. Lane M: molecular weight ladder. The specific amplification band for VEGFR-1 measures 498 bp, whereas VEGFR-2 amplification product measures 709 bp (arrowheads). Nonspecific primer–dimer bands are located at the bottom of each lane and measure less than 100 bp. Lane 1: Unstimulated cells; lane 2: cells stimulated for 48 hours with 1 ng/ml TGF-β₂; lane 3: cells stimulated for 48 hours with 10 ng/ml TGF-β₂; lane 4: cells treated for 48 hours with PDGF-BB; lane 5: negative control cells without reverse transcriptase.

Figure 3.

 

Flow cytometric analysis of VEGFR-1 expression on RPE cells with (gray) and without (black) pretreatment with TGF-β₂. Control staining using a nonreactive primary polyclonal antibody and the same secondary antibody has been subtracted from these curves.

Figure 4.

 

VEGF binding to RPE cells. VEGF binding was examined in RPE cells (A, B) and compared with that in RPE cells pretreated with TGF-β for 5 days (C, D). Scatchard binding analysis was performed. The experiment was repeated three times. Results of a representative experiment are shown.