P2Y Receptor-Mediated Stimulation of Müller Glial Cell DNA Synthesis: Dependence on EGF and PDGF Receptor Transactivation

Abstract

purpose. To determine whether P2Y receptor-evoked proliferation of Müller glial cells depends on transactivation of receptor tyrosine kinases.

methods. Primary cultures of Müller cells of the guinea pig were treated with test substances for 16 hours. The DNA synthesis rate was assessed by a bromodeoxyuridine (BrdU) immunoassay, and the phosphorylation states of the extracellular signal-regulated kinase (ERK1/2) and the p38 mitogen-activated protein kinase (p38 MAPK) were determined by Western blot analysis.

results. In Müller cells, the mitogenic effect of P2Y receptor activation by extracellular adenosine triphosphate (ATP) depended on transactivation of both the platelet-derived growth factor (PDGF) and the epidermal growth factor (EGF) receptor tyrosine kinases, as suggested by the blocking effects of the tyrphostins AG1296 and AG1478 on the ATP-induced proliferation and phosphorylation of ERK1/2. Moreover, the PDGF-induced proliferation may depend on transactivation of the EGF receptor kinase. Antibodies against heparin-binding EGF (HB-EGF) or PDGF, as well as inhibition of matrix metalloproteinases (MMPs) blocked ATP-evoked proliferation. At least one metalloproteinase (MMP-9), was implicated in the signal transfer from P2Y to EGF receptors. In contrast, the mitogenic effect of fetal calf serum was independent of growth factor receptor activity. P2Y receptor activation stimulated Müller cell proliferation by activating the ERK1/2 and the phosphatidylinositol 3 (PI3) kinase signaling pathways, whereas the p38 MAPK pathway was not involved in mitogenic signaling.

conclusions. The present data suggest that P2Y-receptor-induced mitogenic signaling in Müller cells is mediated by transactivation of the PDGF and EGF receptor tyrosine kinases. The transactivation may be mediated by release of PDGF and MMP-dependent shedding of HB-EGF from the Müller cell matrix, respectively. The transactivation of the receptor tyrosine kinases may result in activation of ERK1/2 and PI3 kinase and an increase in the proliferation rate.

Proliferation of Müller glial cells is a common feature of several different diseases of the sensory retina. ¹ During proliferative vitreoretinopathy (PVR), Müller cells proliferate continuously, migrate onto the retinal surfaces, and participate in the formation of periretinal cellular membranes. ^{2 3 4} Gliotic Müller cells are characterized by altered expression of various enzymes, ion channels, and receptors. During PVR of the human retina, Müller cells display an enhanced expression of a distinct type of purinergic P2 receptors. ⁵ In an animal model of PVR, an upregulation of uridine 5′-triphosphate (UTP)-evoked Müller cell responses was observed that is mediated by activation of P2Y receptors. ⁶ A possible role of extracellular adenosine 5′-triphosphate (ATP) in the induction or maintenance of reactive astrogliosis has been suggested. ⁷ Among other effects, extracellular ATP stimulates the proliferation of cultured astrocytes ⁷ and Müller cells. ^{5 8}  

The activation of several growth factor- and G-protein-coupled receptors has been shown to induce proliferation of cultured Müller glial cells. ⁹ It is noteworthy that stimulation of the epidermal growth factor (EGF) receptor or the platelet-derived growth factor (PDGF) receptor has been found to enhance the rate of DNA synthesis in Müller cells. ^{10 11 12} Cultured Müller cells of the guinea pig express G-protein-coupled P2Y receptors but apparently no ionotropic P2X or P1 receptors. ⁸ Stimulation of P2Y receptors by ATP results in a dose-dependent increase of the DNA synthesis rate. ⁸ The mitogenic effects of EGF ¹³ or of ATP ⁸ in Müller cells depend on an influx of calcium ions from the extracellular space, and the proliferation rate has been found to be positively correlated with the duration of ATP-induced calcium transients. ⁸  

Previous studies of several different cell types have demonstrated that growth factor receptors and G-protein-coupled receptors may activate similar signal transduction molecules and may use the same intracellular signaling cascades. ¹⁴ Particularly, G-protein-coupled receptors can transactivate growth factor receptors, including EGF or PDGF receptors. ^{15 16} The activation of the tyrosine kinases of growth factor receptors may be a crucial step in the activation of mitogen-activated protein kinases (MAPKs) by G-protein-coupled receptors. ¹⁷ However, with respect to P2 receptors, different cell systems seem to display different responses. Purinergic agonists may either reduce ¹⁸ or enhance ¹⁷ the tyrosine kinase activity of the EGF receptor. Moreover, activation of P2Y receptors may stimulate MAPKs without cross-activation of receptor tyrosine kinases, in a way that is dependent on integrin-mediated cell anchorage, intracellular calcium elevation, and activation of protein kinase C. ¹⁹ The goal of the present study was to determine whether the mitogenic effect of ATP on Müller cells is mediated by stimulation of the tyrosine kinase activities of growth factor receptors. In certain cell types, the EGF receptor has been described to be transactivated by cytokines and by other growth factors such as PDGF. ^{20 21} Thus, we asked whether activation of the EGF receptor or the PDGF receptor is involved in the mitogenic effect of ATP. The present results suggest that transactivation of both receptor tyrosine kinases is necessary for extracellular ATP to cause a stimulation of DNA synthesis in Müller cells and, partially, for the activation of extracellular signal-regulated kinases (ERK1/2), members of the MAPK protein family. 

Methods

Chemicals and Reagents

EGF (human recombinant), PD158870, PD98059, LY294002, AG1296, AG1478, and A1 (AG9) were obtained from Calbiochem (Bad Soden, Germany). U0126 and SB203358 were from Tocris (Bristol, UK); nagarse (subtilisin, EC 3.4.21.14) and ATP (sodium salt) were from Serva (Heidelberg, Germany). Hoechst 33258 and fura 2-acetoxymethylester (AM) were from Molecular Probes (Eugene, OR). Fetal calf serum was from Biochrom (Berlin, Germany). All other substances were obtained from Sigma (Diesenhofen, Germany). 

Antibodies

The following antibodies were used: monoclonal anti-bromodeoxyuridine (BrdU; clone BU33, 1:2000; Sigma), a neutralizing rabbit anti-human EGF antibody (polyclonal, 1:250; Chemicon International, Temecula, CA), a polyclonal neutralizing antibody directed against natural human PDGF (20 μg/mL; Calbiochem), a polyclonal antibody directed against the human and rat PDGF-B chain (10 μg/mL; Calbiochem), a neutralizing anti-human heparin binding (HB)-EGF (polyclonal, 5 μg/mL; Calbiochem), a monoclonal neutralizing anti-MMP9 (10 μg/mL, Calbiochem), a polyclonal rabbit anti-rat p44/p42 MAPK (New England Biolabs, Frankfurt-am-Main, Germany; 1:1000), a polyclonal rabbit anti-phosphorylated MAPK (Promega, Madison, WI; 1:5000), a polyclonal rabbit anti-human EGF receptor (Santa Cruz Biotechnology, Heidelberg, Germany; 1:200), a polyclonal rabbit anti-human-phosphorylated EGF receptor (Tyr1045) antibody (1:1000; New England Biolabs), a polyclonal rabbit anti-human p38 MAPK (1:1000; New England Biolabs), a polyclonal rabbit anti-human-phosphorylated p38 MAPK (1:1000; New England Biolabs), Cy3-conjugated goat anti-mouse secondary antibody (1:400; Jackson ImmunoResearch Laboratory, West Grove, PA), a secondary biotinylated anti-rabbit IgG (Vector Laboratories, Burlingame, CA), and a secondary anti-rabbit IgG conjugated with alkaline phosphatase (Sigma; 1:5000). 

Cell Culture

Primary cultures of Müller glial cells were obtained from guinea pigs (250–400 g) by a method previously described. ⁸ Animals were deeply anesthetized with urethane (2.0 g/kg, intraperitoneally) before decapitation and enucleation of the eyes. The excised retinas were dispersed in Ca²⁺, Mg²⁺-free phosphate buffer supplemented with nagarse (1 mg/mL) for 30 minutes at 37°C. After a wash in phosphate buffer containing DNase I (200 U/mL), dissociated cells were seeded on uncoated coverslips (diameter 15 mm; Glaswarenfabrik Hecht, Sontheim/Rhön, Germany). Retinal cells from two eyes were distributed on 54 coverslips, at 100 μL cell suspension per slip. Cells were cultured in minimal essential medium supplemented with 10% fetal calf serum at 37°C in 5% CO₂. The medium was exchanged twice a week. Just before cells achieved confluence after 8 days in culture, the test substances were added to the culture medium 16 hours before the cultures were fixed. During this latter period, substances were tested in serum-free medium. Approximately 96% of the cultured cells expressed immunoreactivity for vimentin, and approximately 99% were positively stained for glial fibrillary acidic protein. ¹³ Because the guinea pig retina does not contain astrocytes, most of the cultured cells were considered to represent Müller cells. Lipophilic substances were dissolved in dimethyl sulfoxide. Vehicle alone had no effect on the DNA synthesis rate. 

DNA Synthesis Rate

The DNA synthesis rate was determined by measuring incorporation of BrdU. BrdU (10 μM) was added 16 hours before fixation with 4% paraformaldehyde. Incorporation of BrdU into the nuclei of mitotically active cells was revealed by an anti-BrdU antibody and a Cy3-tagged secondary antibody. Counterlabeling of all cell nuclei was performed with acridine orange or Hoechst 33258. In the peripheral (i.e., nonconfluent) regions of the cultures, six distinct areas of each coverslip (each approximately 60,000 μm², resulting in a total area of 0.42 mm² per coverslip) were studied by means of a semiautomatic image-analysis system (SIS; Soft-Imaging Systems, Münster, Germany). The results from three coverslips per culture were summarized. Every experiment involved at least three independent cultures. The ratio of BrdU-immunoreactive versus total cell nuclei was taken as a marker for the DNA synthesis rate. 

Calcium Imaging

For fluorescence measurements, cells were cultured for 8 days in medium containing 10% serum and then for 16 hours in serum-free medium in the absence or presence of AG1478 (300 nM). The cells were loaded with 10 μM fura 2-AM for 30 minutes at 37°C. Measurements were performed at room temperature in a bath solution that contained (mM) 129 NaCl, 3 KCl, 1 CaCl₂, 0.2 MgCl₂, 20 glucose, and 10 HEPES (pH 7.4 adjusted with NaOH). Fluorescence was excited at 340 (F₃₄₀) and 380 nm (F₃₈₀), and images were recorded every 6 seconds using a measurement system (Fucal 5.12B; Till-Photonics, Munich, Germany). 

Western Immunoblot Analysis

After 8 days in serum-containing medium, the cultures were grown in serum-free medium for 24 hours and then treated with test substances for 10 minutes before cell harvest. The media were removed, the cells were washed twice with cold phosphate-buffered saline (pH 7.4; Biochrom), and the monolayer was scraped into 200 μL lysis buffer (Mammalian Cell Lysis-1 Kit; Sigma). The total cell lysates were centrifuged at 10,000g for 10 minutes, and the supernatant was analyzed by immunoblot. Equal amounts of protein (30 μg) were separated by 12% SDS-polyacrylamide gel electrophoresis. Immunoblots were probed with primary and secondary antibodies, and immunoreactive bands were visualized with diaminobenzidine (peroxidase substrate kit; Vector Laboratories) or 5-bromo-4-chloro-3-indolyl phosphate/nitro blue tetrazolium (Sigma). 

Data Presentation

Statistical analysis (paired Mann-Whitney test, two-tailed) was performed on computer (Prism; Graphpad Software Inc., San Diego, CA). The fluorescence ratio F₃₄₀ to F₃₈₀ is presented to describe relative changes of the intracellular calcium concentration. An increase in the ratio indicates an increase in intracellular free calcium. Data are expressed as the mean ± SEM (BrdU incorporation levels) or as the mean ± SD (calcium imaging). Statistical significance was accepted at P < 0.05. 

Results

Comparison of the Mitogenic Effects of ATP, EGF, HB-EGF, PDGF, and Serum

As previously described, extracellular ATP (Fig. 1A) or EGF (Fig. 1B) increased the DNA synthesis rate of cultured guinea pig Müller cells. ^{8 13} Addition of ATP (1 mM) or EGF (100 ng/mL) to the culture medium resulted in a similar increase of in incorporation of BrdU. Simultaneous application of both substances did not further stimulate DNA synthesis rate. Similarly, simultaneous exposure of the cultures to ATP (500 μM) and native PDGF (a mixture of 70% PDGF-AB and of 30% PDGF-BB, 25 ng/mL) did not additively stimulate the DNA synthesis rate (Fig. 1C) . Another natural ligand of the EGF receptor, HB-EGF (100 ng/mL), was found to stimulate the DNA synthesis rate similar to PDGF-AB/BB (25 ng/mL; Fig. 1D ) as well as ATP (500 μM; Fig. 1E ). The nonadditive mitogenic effects of the growth factors and of ATP may indicate that they share, at least partially, common signal transduction-pathways in Müller cells. In contrast, the effects of fetal calf serum (5%) and ATP (500 μM) were additive (Fig. 2D) , suggesting that ATP and serum stimulate different intracellular signaling pathways. 

Receptor Tyrosine Kinase Dependence of ATP-Induced DNA Synthesis

To investigate whether stimulation of P2Y receptors activates the tyrosine kinase of the EGF receptor, the effect of AG1478 on ATP-evoked DNA synthesis was determined. AG1478 is a tyrphostin that selectively blocks the tyrosine kinase of the EGF receptor by interaction with its ATP-binding site and by inducing the formation of inactive EGF receptor dimers. ²² Although AG1478 (300 nM) per se had no effect on the incorporation of BrdU, it significantly decreased the DNA synthesis stimulated by ATP (500 μM) or EGF (100 ng/mL; Fig. 2A ). Similar results were obtained with another blocker of the EGF receptor tyrosine kinase, PD158780 (500 nM; Fig. 2B ). These observations may suggest that stimulation of P2Y receptors by ATP in Müller cells leads to transactivation of the EGF receptor. Exposure of AG1478 (300 nM) did not affect the ATP (500 μM)-induced intracellular calcium response measured by fura-2 fluorometry (Fig. 2C) , indicating that the tyrosine kinase of the EGF receptor had no effect on ATP binding to P2Y receptors or on subsequent activation of G protein. The mitogenic effect of serum (5%) was not influenced by coapplication of AG1478 (300 nM; Fig. 2D ), further supporting the conclusion that the effects of ATP and serum on proliferation are mediated by different intracellular-signaling pathways. When ATP (500 μM), serum (5%), and AG1478 (300 nM) were simultaneously applied, the tyrphostin blocked the effect of ATP but not the effect of serum on DNA synthesis (Fig. 2D) . 

To explore whether extracellular ATP also transactivates the PDGF receptor, the effect of tyrphostin AG1296 was investigated. AG1296 selectively inhibits the tyrosine kinases of PDGFα- and -β receptors. ²³ Addition of AG1296 (10 μM) to the culture medium suppressed the stimulating effect of extracellular ATP on DNA synthesis but failed to suppress the mitogenic effect of EGF (Fig. 2E) . To disclose possible nonspecific effects of the tyrphostins used, we tested the inactive tyrphostin A1 (10 μM) as a negative control. As shown in Figure 2F , tyrphostin A1 inhibited neither ATP- nor the EGF-induced DNA synthesis. PDGF itself had a stimulating effect on DNA synthesis. The effects of PDGF-AA (100 ng/mL; Fig. 2G ) or of PDGF-AB/BB (25 ng/mL; Fig. 2H ) were blocked by simultaneous application of AG1478 (300 nM) or AG1296 (10 μM). To confirm the assumption that stimulation of the P2Y or PDGF receptor may lead to transactivation of the EGF receptor in Müller cells, Western blot analysis was performed with an antibody directed against the phosphorylated Tyr1045 site of the EGF receptor. Application of ATP (500 μM) or PDGF-AB/BB (25 ng/mL) led to elevated phosphorylation at this site (Fig. 2 , inset). The results suggest that the EGF receptor is an integrator of mitogenic signals from G-protein-coupled receptors and from tyrosine kinase receptors in Müller cells and that the PDGF receptor may be involved in transmitting mitogenic signals from the P2Y receptor to the EGF receptor. 

Mechanism of the ATP-Induced Transactivation of the EGF Receptor

To determine how P2Y receptor activation may result in a transactivation of the EGF receptor, antibodies against EGF receptor ligands were tested. Simultaneous application of an EGF-neutralizing antibody inhibited EGF (100 ng/mL)-induced but not ATP (500 μM)-induced DNA synthesis (Fig. 3A) . However, a neutralizing antibody directed against HB-EGF inhibited the stimulating effect of ATP on the DNA synthesis (Fig. 3B) . HB-EGF (100 ng/mL) increased the DNA synthesis rate, and the effect of HB-EGF was reversed by simultaneous addition of the anti-HB-EGF antibody or of the tyrphostin AG1478 (300 nM; Fig. 3C ). The results suggest that P2Y receptor activation by ATP leads to release of HB-EGF from the Müller cell matrix, which then mediates the transactivation of the EGF receptor. However, we also found a partial block of the mitogenic effect of HB-EGF when the PDGF receptor tyrosine kinase was inhibited by tyrphostin AG1296 (10 μM; Fig. 3C ). The antibody against HB-EGF also blocked the effect of PDGF-AB/BB (25 ng/mL) but not that of EGF (100 ng/mL) on DNA synthesis (Fig. 3D) . It is suggested that activation of the PDGF receptor in Müller cells causes a release of HB-EGF from the extracellular matrix. 

To determine whether the effect of ATP is mediated by activation of matrix metalloproteinases (MMPs), which may cleave membrane-bound pro-HB-EGF, ²⁴ the broad-spectrum MMP inhibitor 1,10-phenanthroline (10 μM) was tested. This inhibitor decreased significantly the ATP (500 μM)- and PDGF-AB/BB (25 ng/mL)-evoked effects on DNA synthesis, but not the effect of HB-EGF (100 ng/mL; Fig. 4A ). To find out whether MMP-9 may be involved in the mitogenic effect of ATP, a neutralizing antibody was tested. The antibody against MMP-9 reversed the mitogenic effect of ATP (Fig. 4B) , suggesting that MMP-9 is involved in the mitogenic signaling cascade of P2Y receptors. The anti-MMP-9 antibody also blocked the mitogenic effect of HB-EGF and partially reversed the effect of PDGF-AB/BB but had no effect on the EGF- and serum-mediated proliferation (Fig. 4B) . 

PDGF Receptor Transactivation by P2Y Receptors

We then determined whether the transactivation of the PDGF receptor by the P2Y receptor is mediated by a release of PDGF from Müller cells. Different antibodies directed against PDGF were tested. Both PDGF-AB/BB (25 ng/mL)- and ATP (500 μM)-induced DNA synthesis were inhibited by simultaneous application of a neutralizing antibody directed against native PDGF (Fig. 5A) . A similar effect was observed with an antibody directed against the PDGF-B chain (Fig. 5B) . In contrast, the antibody did not block the mitogenic effects of HB-EGF and of EGF. The data suggest that the P2Y receptor-induced transactivation of the PDGF receptor is mediated by a release of PDGF from Müller cells. 

Depletion of Cholesterol

In other cell systems, EGF ²⁵ and PDGF receptors ²⁶ are concentrated in caveolae of the cell surface. The confinement of the two receptors to this narrow cellular substructure may facilitate their interaction. Stimulation with PDGF of PDGF receptors in caveolae induces tyrosine phosphorylation of neighboring EGF receptors. ²¹ The dependence of the mitogenic effect of PDGF on EGF receptor transactivation in Müller cells may support the assumption that the two receptors are expressed in close spatial proximity in the Müller cell membrane, probably within caveolae or lipid rafts. Lipid rafts are dispersed by cholesterol depletion. ²⁵ Lovastatin (10 μM) and mevalonate (250 μM), blockers of the 3-hydroxy-3-methylglutaryl-coenzyme A (HMG CoA) reductase, cause a reduction in the intracellular concentration of cholesterol. Simultaneous application of both blockers abolished the mitogenic effect of ATP (500 μM) but not the effect of serum (5%) in Müller cells (Fig. 6A) . The results once more support the assumption that ATP and serum exert their mitogenic effects through different intracellular signaling mechanisms. Both blockers of cholesterol synthesis also inhibited the mitogenic effect of PDGF-AB/BB (25 ng/mL) but not of EGF (100 ng/mL; Fig. 6B ). 

Because HMG CoA reductase inhibitors may block mitogenic signaling through several different mechanisms other than cholesterol deficiency, ^{27 28 29} we tested another way to deplete the cells of cholesterol. Lipid rafts may be destroyed by extraction of cholesterol from the plasma membrane by incubating cells with the membrane-impermeable cholesterol-binding drug methyl-β-cyclodextrin (MCD). Simultaneous application of MCD (1 mM) blocked the mitogenic effects of ATP (500 μM) and PDGF-AB/BB (25 ng/mL) but not of serum (5%; Fig. 6C ). MDC treatment also resulted in a partial reversal of the mitogenic effects of HB-EGF and EGF, with a significantly stronger effect in the case of HB-EGF (Fig. 6C) . The results support the view that the PDGF and the EGF receptors (and, perhaps, also the P2Y receptors) are expressed in proximity in the Müller cell membrane and that this spatial relationship may be necessary for the mitogenic effect of PDGF receptor activation (and, therefore, of P2Y receptor activation as well). 

MAPK Dependence of Proliferation

To investigate the involvement of MAPKs in the ATP-induced Müller cell proliferation, different blockers were tested. Inhibition of ERK1/2 by using the MEK inhibitors PD98059 (20 μM) and U0126 (20 μM) resulted in a complete block of the proliferation evoked by ATP (500 μM), PDGF-AB/BB (25 ng/mL), HB-EGF (100 ng/mL), or EGF (100 ng/mL; Fig. 7A ). In contrast, the blocker of the p38 MAPK, SB203358 (10 μM), did not decrease the DNA synthesis evoked by any of the agonists tested (Fig. 7B) . 

PI3 Kinase Dependence of Proliferation

The PI3 kinase inhibitors LY294002 (25 μM) or wortmannin (100 nM) partially decreased the mitogenic effects of ATP (500 μM), PDGF-AB/BB (25 ng/mL), HB-EGF (100 ng/mL), and EGF (100 ng/mL; Fig. 7C ). A similar effect on ATP (500 μM)-evoked Müller cell proliferation was observed in the presence of rapamycin (100 nM), an inhibitor of mammalian target of rapamycin (mTOR) kinase which is indispensable for the activation of p70S6 kinase, a downstream target of PI3 kinase (Fig. 7D) . It is suggested that the mitogenic effect of ATP in Müller cells is mediated by activation of ERK1/2, but not of p38 MAPK, and that the ATP-induced proliferation is also partially dependent on the PI3 kinase pathway. 

Phosphorylation of MAPK

Western blot analysis was performed to determine whether P2Y receptor activation causes phosphorylation of ERK1/2 and p38 MAPK in Müller cells. Although ATP (500 μM), EGF (100 ng/mL), and PDGF-AB/BB (25 ng/mL) did not alter the amount of ERK1/2 protein expressed by Müller cells (Fig. 8B , top), they significantly increased the degree of phosphorylation of these kinases (Fig. 8A) . The effect of ATP was partially reversed by simultaneous exposure to AG1478 (300 nM) or to AG1296 (10 μM; Fig. 8A ). The effect of EGF on phosphorylation ERK1/2 was partially blocked by simultaneous exposure to AG1478 but remained largely unaffected in the presence of AG1296 (Fig. 8A) . Both blockers had no effect on the total amount of ERK1/2 protein expressed by the cells (not shown). The effects of ATP or EGF on the ERK1/2 phosphorylation were fully blocked by the MEK inhibitor U0126 (20 μM; Fig. 8B ). Moreover, neutralizing anti-HB-EGF and anti-MMP-9 antibodies decreased significantly the effect of ATP and left the effect of EGF unchanged. An MMP-9-neutralizing antibody also caused a decrease in the stimulating effects of PDGF-AB/BB (25 ng/mL) and HB-EGF (100 ng/mL; Fig. 8B ). Furthermore, the effect of PDGF on the phosphorylation degree of ERK1/2 was significantly decreased by AG1478 (300 nM). In contrast, ATP (500 μM) did not increase the phosphorylation level of p38 MAPK (Fig. 8C) . Similarly, PDGF-AB/BB (25 ng/mL), HB-EGF (100 ng/mL), and EGF (100 ng/mL) displayed no effects on the amount and the phosphorylation state of p38 MAPK (data not shown). 

Discussion

In cultured Müller cells of the guinea pig, extracellular ATP increases the DNA synthesis rate through activation of G-protein-coupled P2Y receptors. ⁸ In the current study, P2Y receptor stimulation led to activation of ERK1/2 (Fig. 8A) , and Müller cell proliferation depended on activated ERK1/2 (Fig. 7A) and —although to a lesser extent—on the activity of PI3 kinase (Fig. 8C) . In contrast, p38 MAPK was not involved in the ATP- and growth-factor-induced Müller cell proliferation (Fig. 7B) , in agreement with the absence of agonist-induced phosphorylation of the p38 MAPK protein (Fig. 8D) . 

The tyrosine kinase of the EGF receptor is essential to transmission of mitogenic signals from G-protein-coupled receptors to the Ras-Raf-MAPK pathway in various cell types. ¹⁶ The results presented in our study suggest that transactivation of more than one receptor tyrosine kinase is involved in the mitogenic signaling in Müller cells after P2Y receptor stimulation. The transactivation may be mediated by release PDGF and MMP-dependent shedding of HB-EGF from the Müller cell matrix, respectively. The transactivation of the PDGF-α and EGF receptor tyrosine kinases result in activation of ERK1/2 and PI3 kinase and increase of the proliferation rate (Fig. 9) . To our knowledge, this is the first description of a case in which the proliferation-stimulating effect of P2Y receptor activation has been shown to depend on transactivation of two different receptor tyrosine kinases. The results support the view that the EGF receptor in Müller cells receives mitogenic signals from both tyrosine kinase receptors and G-protein-coupled receptors. The cholesterol-depletion experiments may suggest that the PDGF and EGF receptors (and, perhaps, the P2Y receptors) are expressed in spatial proximity in Müller cell membranes, perhaps in caveolae or in lipid rafts, which may facilitate their interaction. Cholesterol depletion of the cell membrane disperses lipid rafts ³⁰ and, therefore, may destroy this close spatial relationship of PDGF and EGF receptors on the Müller cell membrane. This may cause an interruption of ATP-induced mitogenic signaling in Müller cells (Fig. 6) . 

Although the effect of ATP on Müller cell proliferation was fully inhibited by the blockers of receptor tyrosine kinases tested (AG1478, PD158780, and AG1296; Fig. 2 ), the effect of ATP on ERK1/2 phosphorylation was only partially reversed (Fig. 8B) . Therefore, it cannot not be fully ruled out that stimulation of P2Y receptors in Müller cells causes ERK1/2 phosphorylation through multiple intracellular signaling cascades: one involving PDGF-α and EGF receptor transactivation and the other involving independent mechanisms of cross-activation of receptor tyrosine kinases, as recently described in other cell systems. ¹⁹ The presence of multiple ERK1/2 activation pathways in cultured Müller cells has been suggested in regard to basic fibroblast growth factor signaling. ³¹ In other cell systems, P2Y receptor-mediated stimulation of MAPKs without cross-activation of receptor tyrosine kinases has been described which was mediated by activation of phospholipase C and protein kinase C. ¹⁹ In cultured astrocytes, signaling from P2Y receptors to ERKs involved stimulation of phospholipase D and of a calcium-independent protein kinase C isoform. ³²  

In addition to the transactivation of tyrosine kinase, activation of calcium-dependent intracellular signaling has been suggested to be necessary for the mitogenic effect of ATP in Müller cells. ⁸ Whether the influx of calcium ions from the extracellular space and the activation of protein kinase C are necessary for the transactivation of the receptor tyrosine kinases remains to be determined in future experiments. In PC12 cells, EGF receptor transactivation by bradykinin or by membrane depolarization is critically dependent on the presence of extracellular calcium, ³³ and calcium influx through voltage-gated channels is sufficient to cause tyrosine phosphorylation of the EGF receptor. ³⁴ In contrast, EGF receptor transactivation by m1-muscarinergic acetylcholine receptors is calcium independent. ³⁵ Protein kinase C has been shown to be either a positive mediator ³⁵ or an inhibitor of EGF receptor transactivation ³⁶ by different G-protein-coupled receptors. Phorbol ester, calcium entry from the extracellular space, or activation of calcium-independent protein kinase C, all can induce proteolytic cleavage of membrane-bound proHB-EGF. ^{37 38 39}  

The two ligands of the EGF receptor tested, HB-EGF and EGF, showed slightly different effects in our experiments. Whereas the effect of EGF on proliferation was apparently independent of activation of the PDGF receptor (Fig. 2E) , the effect of HB-EGF was partially inhibited after blocking the PDGF receptor tyrosine kinase (Fig. 3C) . The effect of EGF was not blocked by an anti-MMP-9 antibody, whereas the effect of HB-EGF was strongly influenced (Fig. 4B) . Similarly, the effect of EGF was significantly less influenced by cholesterol depletion of the plasma membrane than was the effect of HB-EGF (Fig. 6C) . These results may suggest that these two ligands of the EGF receptor initiate proliferation somewhat differently, with a dependence on PDGF receptor activation in the case of HB-EGF and with no dependence on PDGF receptor activation in the case of EGF. The reason for this difference is unclear. A direct physical association of EGF and PDGF receptors and an EGF-induced transactivation of the PDGF receptor has been described in other cell systems. ⁴⁰ Moreover, different interactions with the extracellular matrix (to heparan sulfate proteoglycans in the cases of PDGF ^{41 42} and of HB-EGF ⁴³ but not in the case of EGF) may cause different ligand effects. The binding of PDGF and of HB-EGF to extracellular matrix components may amplify the binding to and activation of their receptors and may augment MAPK phosphorylation. ^{42 44} We assume that a dispersion of lipid rafts or of caveolae alters the spatial relationship between lipid rafts and the extracellular matrix. Because HB-EGF activity is dependent on binding to heparan sulfate proteoglycans, ⁴⁴ dispersion of caveolae would influence the mitogenic effect of HB-EGF but not that of EGF. The weak effect of cholesterol depletion on the mitogenic activity of EGF suggests that the coupling of the EGF receptor to downstream signaling molecules is not significantly affected. The anti-MMP-9 antibody may sterically prevent PDGF and HB-EGF, but not EGF, from interacting with their receptors when MMP-9, the PDGF and EGF receptors, and heparan sulfate proteoglycans are expressed in proximity on the Müller cell membrane. 

There are several results suggesting that in Müller cells, serum and ATP mediate their mitogenic effects through different intracellular signaling pathways. In contrast to the nonadditive effects of EGF and ATP on DNA synthesis, the effects of serum and ATP were additive. The mitogenic effects of ATP or of EGF were blocked in the presence of AG1478 (Fig. 2A) , whereas the effect of serum was not (Fig. 2D) . Moreover, the mitogenic signaling of ATP was interrupted by cholesterol depletion of the Müller cell membrane, but Müller cell proliferation remained unaffected by serum (Figs. 6A 6C) . A similar difference between the mitogenic effects of serum and EGF has been reported associated with the dependence on the activity of calcium-activated potassium channels in cultured Müller cells. Although the mitogenic effects of ATP or of EGF were dependent on the activity of this type of channel, serum-induced DNA synthesis was not. ^{8 13} The data suggest also that the stimulating effects of serum are probably not caused by transforming growth factor-α or by lysophosphatidic acid, both of which are native constituents of serum. Lysophosphatidic acid, which acts through a G-protein-coupled receptor, has been shown to induce ligand-independent tyrosine autophosphorylation of the EGF and PDGF receptors. ⁴⁵  

The present results imply the involvement of Müller cells in proliferative retinal diseases in several ways. Human Müller cells from patients with PVR show an upregulation of a distinct type of P2X receptors, ⁵ whereas rabbit Müller cells from PVR eyes display increased responses to activation of uridine triphosphate (UTP)-sensitive P2Y receptors. ⁶ Upregulation of PDGF ^{46 47} and of its receptors ⁴⁶ has been described in eyes with proliferative retinal diseases. In a rabbit model of PVR, the PDGF-α receptor has been implicated in the generation of PVR, ⁴⁸ and inhibition of the PDGF-α receptor has been shown to attenuate experimental PVR. ⁴⁹ In the current study, application of native PDGF (PDGF-AB/BB) or of PDGF-AA increased the proliferation rate of cultured Müller cells, which suggests that the cells express functional PDGFα receptors and that AG1296, the tyrphostin that specifically inhibits the tyrosine kinase of PDGF receptors, ²³ blocks the effects of these agonists. In the mitogenic signaling pathways of P2Y receptors, an activation of MMPs is likely to be involved (Fig. 4A) , and at least one type of MMP (MMP-9) was suggested to be active (Fig. 4B) . MMP-2 and, to a lesser degree, MMP-9 have been implicated in the development of postoperative PVR. ^{50 51} Because different proliferation-enhancing molecules act through transactivation of the EGF receptor, it can be speculated that inhibiting the tyrosine kinase activity of the EGF receptor is a useful tool to suppress uncontrolled intraocular proliferation. However, when the proliferation is additionally stimulated by serum after a destruction of the blood-retinal barrier, other strategies are needed to inhibit intraocular proliferation. 

 Supported by a grant from the Federal Ministry for Education and Research (Interdisciplinary Center for Clinical Research at the University of Leipzig, 01KS9504, Project C5) and by Grant Re 849/8-1 from the German Research Foundation.

 Submitted for publication March 13, 2002; revised August 28, 2002; accepted October 17, 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.

 

Comparison of the mitogenic effects of ATP, EGF, HB-EGF, and PDGF. (A) Extracellular ATP (10 and 500 μM) increased the DNA synthesis rate in cultured Müller cells. (B) EGF (100 ng/mL) and ATP (1 mM) increased the DNA synthesis rate to a similar degree. Simultaneous application of EGF and ATP had no additive effect. (C) ATP (500 μM) and native PDGF (PDGF-AB/BB, 25 ng/mL) had similar mitogenic effects. Simultaneous application of both agonists did not further stimulate DNA synthesis. Native PDGF is a mixture of 70% PDGF-AB and 30% PDGF-BB. (D) HB-EGF (100 ng/mL) and PDGF-AB/BB (25 ng/mL) increased the DNA synthesis rate similarly. (E) Simultaneous application of ATP (500 μM) and HB-EGF (100 ng/mL) did not result in additive stimulation of DNA synthesis. Data are the mean ± SEM of results in three to six independent experiments. (•) Significant difference compared with the control (P < 0.05).

Figure 2.

 

The mitogenic effects of ATP were abolished in the presence of blockers of receptor tyrosine kinases. (A) AG1478 (300 nM), the tyrphostin selective for the tyrosine kinase of the EGF receptor, reversed the increases of DNA synthesis induced by ATP (500 μM) and EGF (100 ng/mL). (B) Another blocker of the EGF receptor tyrosine kinase, PD158780 (500 nM), also inhibited the mitogenic effects of ATP (500 μM) and EGF (100 ng/mL). (C) AG1478 did not block the ATP-induced increase in intracellular calcium, as indicated by fura-2 fluorometry. Extracellular ATP (500 μM) induced a calcium transient in control cells (left) that was similar in other cells that were cultured for 16 hours in the presence of AG1478 (300 nM; middle). Preincubation of a third group of cells with AG1478 (300 nM) for 3 minutes before application of ATP (500 μM) did not significantly alter the ATP-induced increase in calcium (right). AG1478 was also applied simultaneously with ATP. Data are the mean ± SD curves of measurements in 29, 77, and 37 cells (left to right), respectively, from sister cultures. (D) Addition of ATP (500 μM) or fetal calf serum (5%) to the culture medium resulted in increases in DNA synthesis. Simultaneous application of both agonists resulted in additive stimulation of DNA synthesis. The mitogenic effect of serum was not inhibited in the presence of AG1478 (300 nM), whereas the mitogenic effect of ATP was blocked by this tyrphostin when ATP, serum, and AG1478 were simultaneously applied. (E) AG1296 (10 μM), the tyrphostin selective for PDGF receptor tyrosine kinases, inhibited the mitogenic effect of ATP (500 μM) but did not alter the effect of EGF (100 ng/mL). (F) Tyrphostin A1 (10 μM), a negative control for other tyrphostins, had no inhibitory effects on ATP (500 μM)- or on EGF (100 ng/mL)-induced DNA synthesis. (G) The stimulating effect of PDGF-AA (100 ng/mL) on DNA synthesis was reversed by the tyrphostins AG1478 (300 nM) and AG1296 (10 μM). (H) Both tyrphostins also inhibited the mitogenic effect of native PDGF (PDGF-AB/BB, 25 ng/mL). Data are the mean ± SEM of three to six independent cultures. (H, inset) Immunoblots using an antibody against the Tyr1045 site of the EGF receptor showed an elevated phosphorylation level after stimulation with ATP (500 μM) and PDGF-AB/BB (25 ng/mL). The total amount of EGF receptor protein remained unaltered during agonist stimulation (not shown). (•) Significant difference versus the control (P < 0.05); (○) significant effect of the blocker (P < 0.05).

Figure 3.

 

The mitogenic effect of ATP is likely to be mediated by release of heparin-binding EGF-like growth factor (HB-EGF) from Müller cells. (A) An EGF-neutralizing antibody had no effect on the basal incorporation of BrdU or on the mitogenic effect of ATP (500 μM). The effect of ATP was inhibited by application of AG1478 (300 nM). The anti-EGF antibody inhibited the DNA synthesis that was induced by EGF (100 ng/mL). (B) A neutralizing antibody directed against human HB-EGF inhibited the effect of ATP on DNA synthesis. (C) The stimulating effect of HB-EGF (100 ng/mL) on DNA synthesis was inhibited by the anti-HB-EGF antibody and by the tyrphostin AG1478 (300 nM). The tyrphostin AG1296 (10 μM) partially blocked the effect of HB-EGF. (D) The effect of PDGF-AB/BB (25 ng/mL) on DNA synthesis was inhibited by the anti-HB-EGF antibody, whereas EGF (100 ng/mL) did not affect it. Data are the mean ± SEM of three to five independent experiments. (•) Significant differences versus the control (P < 0.05); (○) significant effects of the antibodies and blockers (P < 0.05).

Figure 4.

 

Activation of an MMP was involved in the mitogenic effect of ATP. (A) 1,10-Phenanthroline (10 μM), a broad-spectrum metalloproteinase inhibitor, significantly decreased the stimulating effects of ATP (500 μM) and PDGF-AB/BB (25 ng/mL) on DNA synthesis but had no effect on the mitogenic action of HB-EGF (100 ng/mL). (B) The effect of ATP (500 μM) is suggested to be at least partially mediated by MMP-9, as revealed by the blocking effect of a neutralizing anti-MMP-9 antibody. The antibody depressed PDGF-AB/BB (25 ng/mL)- and the HB-EGF (100 ng/mL)-induced proliferation but had no effect on the actions of EGF (100 ng/mL) and fetal calf serum (5%). Data are the mean ± SEM of results in four to six independent experiments. (•) Significant differences versus the control (P < 0.05); (○) significant effects of the antibody and the blocker (P < 0.05).

Figure 5.

 

Extracellular ATP may cause release of PDGF from Müller cells. (A) A neutralizing antibody directed against native PDGF inhibited the mitogenic effects of native PDGF (PDGF-AB/BB, 25 ng/mL) and ATP (500 μM). (B) An antibody directed against the PDGF-B chain showed the same effects, whereas it did not inhibit the effects of HB-EGF (100 ng/mL) and EGF (100 ng/mL). Data are the mean ± SEM of results in three to six independent experiments. (•) Significant differences versus the control (P < 0.05); (○) significant effects of the antibodies (P < 0.05).

Figure 6.

 

The mitogenic effects of ATP and PDGF were inhibited by cholesterol depletion. (A) Simultaneous application of lovastatin (10 μM) and mevalonate (250 μM) reversed the stimulating effect of ATP (500 μM) on DNA synthesis but not the effect of serum (5%). (B) Lovastatin (10 μM) and mevalonate (250 μM) blocked the mitogenic effect of PDGF-AB/BB (25 ng/mL) but not the effect of EGF (100 ng/mL). (C) MCD (1 mM) inhibited the mitogenic effects of ATP (500 μM) and of PDGF-AB/BB (25 ng/mL), but did not inhibit the effect of serum (5%). The effects of EGF (100 ng/mL) and HB-EGF (100 ng/mL) were partially reversed. Means ± SEM of 4 to six independent experiments. (•) Significant differences versus the control (P < 0.05); (○) significant effects of the blockers and of MCD (P < 0.05). *P < 0.05.

Figure 7.

 

The mitogenic effect of ATP was mediated by ERK1/2 and PI3 kinase but not by p38 MAPK. (A) Inhibition of ERK1/2 activation by the MEK inhibitors PD98059 (20 μM) or U0126 (20 μM) resulted in complete block of the DNA synthesis that was evoked by ATP, PDGF-AB/BB, HB-EGF, or EGF. (B) Inhibition of p38 MAPK by SB203358 (10 μM) did not decrease the DNA synthesis that was evoked by the agonists in (A). (C) The PI3 kinase inhibitor LY294002 (25 μM) or wortmannin (100 nM) partially decreased the mitogenic effects of the agonists. (D) Rapamycin (100 nM) significantly decreased the mitogenic effect of ATP. The agonists were tested at the following concentrations: ATP, 500 μM; PDGF-AB/BB, 25 ng/mL; HB-EGF, 100 ng/mL; and EGF, 100 ng/mL. Data are the mean ± SEM of results in three to five independent cultures. (•) Significant difference versus the control (P < 0.05); (○) significant effect of the blocker (P < 0.05).

Figure 8.

 

Extracellular ATP increased the phosphorylation of ERK1/2 but not of p38 MAPK in Müller cells. (A) A 10-minute exposure of cells to ATP (500 μM) or EGF (100 ng/mL) increased the amount of phosphorylated ERK1/2. Simultaneous exposure to AG1478 (300 nM) or AG1296 (10 μM) partially blocked the effect of ATP, whereas the effect of EGF was decreased only by AG1478. (B) Effects of different blockers (U0126, 20 μM; AG1478, 300 nM) and neutralizing antibodies, respectively, on the increase of the amount of phosphorylated ERK1/2 that was evoked by ATP (500 μM), EGF (100 ng/mL), PDGF-AB/BB (25 ng/mL), or HB-EGF (100 ng/mL; bottom). The substances and antibodies tested displayed no effect on the total amount of ERK1/2 protein expressed by the cells (top). (C) ATP (500 μM) did not increase the phosphorylation level of the p38 MAPK (bottom) and did not alter the total amount of p38 protein present in Müller cells (above). AG1478, 300 nM; AG1296, 10 μM. The cultures were preincubated with the blockers for 15 minutes and then stimulated with the agonists for 10 minutes. The data are representative of results in two to four independent experiments.

Figure 9.

 

Proposed mechanism of receptor tyrosine kinase-dependent stimulation of Müller cell proliferation by extracellular ATP. Activation of P2Y receptors by ATP leads to an increase in intracellular calcium concentration and may cause a release of PDGF from the cells. Released PDGF mediates the transactivation of the PDGF-α receptor. Activation of the PDGF-α receptor may result in a release of HB-EGF from the extracellular matrix through shedding of membrane-bound pro-HB-EGF by an MMP. The released HB-EGF subsequently transactivates the EGF receptor tyrosine kinase. The activated EGF and PDGF receptors induce proliferative activity in Müller cells, through the Ras-Raf-MEK-ERK and the PI3 kinase pathways. However, there may be also other signaling pathways that contribute to the P2Y receptor-mediated mitogenic signaling in Müller cells (indicated by broken lines). PLC, phospholipase C.

The authors thank Jana Krenzlin for excellent technical assistance.