RasGAP Promotes Autophagy and Thereby Suppresses Platelet-Derived Growth Factor Receptor-Mediated Signaling Events, Cellular Responses, and Pathology

Abstract

Platelet-derived growth factors (PDGFs) and their receptors (PDGFRs) make profound contributions to both physiology and pathology. While it is widely believed that direct (PDGF-mediated) activation is the primary mode of activating PDGFRs, the discovery that they can also be activated indirectly begs the question of the relevance of the indirect mode of activating PDGFRs. In the context of a blinding eye disease, indirect activation of PDGFRα results in persistent signaling, which suppresses the level of p53 and thereby promotes viability of cells that drive pathogenesis. Under the same conditions, PDGFRβ fails to undergo indirect activation. In this paper, we report that RasGAP (GTPase-activating protein of Ras) prevented indirect activation of PDGFRβ. RasGAP, which associates with PDGFRβ but not PDGFRα, reduced the level of mitochondrion-derived reactive oxygen species, which are required for enduring activation of PDGFRs. Furthermore, preventing PDGFRβ from associating with RasGAP allowed it to signal enduringly and drive pathogenesis of a blinding eye disease. These results indicate a previously unappreciated role of RasGAP in antagonizing indirect activation of PDGFRβ, define the underlying mechanism, and raise the possibility that PDGFRβ-mediated diseases involve indirect activation of PDGFRβ.

INTRODUCTION

The receptors for platelet-derived growth factor (PDGF) are essential for mouse development and are implicated in a variety of human diseases (1, 2). Furthermore, these observations are the basis for the consensus that, while there may be overlap in what the two PDGF receptors (PDGFRs), PDGFRα and PDGFRβ, are capable of, they also have nonredundant functions in physiology and pathology.

Because the two PDGFRs engage nonidentical signaling events in acutely stimulated cultured cells (3), a plausible reason for the distinct phenotype of mice lacking pdgfra and/or pdgfrb (4, 5) relates to signaling. Characterization of mice that express chimeric receptors in which the cytoplasmic domains were interchanged indicated that PDGFRβ was more capable than PDGFRα. PDGFRα/β chimeric mice had no phenotype, whereas PDGFRβ/α chimeric mice showed some of the defects seen in mice in which PDGFRβ lacked a major portion of the cytoplasmic domain (6, 7). Thus, in the context of mouse embryogenesis, the two PDGFRs do not appear to trigger the same signaling events, and more specifically, PDGFRβ does something that PDGFRα cannot.

The disparity in signaling events between the two PDGFRs that is germane to this report involves RasGAP (GTPase-activating protein of Ras), which is recruited by PDGFRβ but not PDGFRα (8–10). RasGAP promotes the inactivation of Ras (11–13). RasGAP is an SH2 domain-containing protein, and its association with PDGFRβ is dependent on tyrosine phosphorylation of PDGFRβ within a context that is preferred by the SH2 domains of RasGAP (14–19). PDGFRα does not interact with RasGAP because none of its phosphorylation sites are within such an amino acid motif (9, 10, 20).

Consistent with the known function of RasGAP, PDGF stimulates a substantially larger accumulation of active Ras in early-passage fibroblasts isolated from Rasa1-null mouse embryos than in cells isolated from control mice (21). While PDGF-driven entry into S phase is unaffected by either the absence of RasGAP expression or mutating PDGFRβ so that it could not associate with RasGAP, there is an increase in the number of PDGFRβ-dependent cells in mice expressing the RasGAP binding mutant instead of wild-type (WT) PDGFRβ (7, 8, 21). Thus, RasGAP limits PDGF-stimulated Ras activation, and this appears to govern the expansion of PDGFRβ-dependent populations of cells during development.

In addition to the direct, PDGF-mediated mode of activating PDGFRs, recent studies indicate that there is also an indirect mode (also called transactivation). The quintessential features of direct activation include the facts that it is driven by PDGF, commences within seconds, is enhanced by transient inactivation of phosphotyrosine phosphatases (PTPs), and is extinguished by ligand-induced receptor degradation (22, 23).

The indirect mode of activation of PDGFR is unique to PDGFRα and proceeds by an exclusively intracellular route; while PDGFRs without an extracellular domain fail to be activated by PDGF, they remain fully responsive to indirect activation (24). The key elements of this intracellular route include non-PDGF (growth factors outside the PDGF family)-mediated elevation of reactive oxygen species (ROS) and activation of Src family kinases, which phosphorylate and activate monomeric PDGFRα (25, 26). Unlike directly activated PDGFRα, which is rapidly extinguished (its half-life is less than 5 min [27, 28]), indirectly activated PDGFRα signals persistently (28). This is because it has not been dimerized by PDGF, which greatly shortens its half-life. Furthermore, indirectly activated PDGFRα engages signaling events that suppress autophagy, and thereby elevate mitochondrial ROS, and perpetuates activation (26). Thus, indirect activation of PDGFRα involves monomeric receptors, which are activated persistently because of a mitochondrial-ROS-driven feed-forward loop.

The realization that PDGFRα can be activated indirectly begs the questions of when this occurs and its relevance. Its discovery was within the context of a blinding ocular condition called proliferative vitreoretinopathy (PVR), which afflicts up to 5,500 individuals in the United States annually (29). Indirect activation of PDGFRα promotes the survival of cells displaced into the vitreous (which contains many non-PDGFs) and thereby contributes to the pathogenesis of the disease (23, 30). Given that promoting the viability of cells may also contribute to a variety of physiological settings, it is possible that the current association of indirect activation of PDGFRα with pathology is an underestimate of the relevance of this mode of activating PDGFRα.

In contrast to PDGFRα, PDGFRβ cannot be indirectly activated and is largely unable to promote pathogenesis of PVR (31, 32). In this report, we considered if RasGAP was responsible. Indeed, our results indicate that recruitment of RasGAP to PDGFRs antagonized the feed-forward loop that is necessary to persistently activate PDGFR. Furthermore, preventing RasGAP from associating with PDGFRβ enabled the receptor to undergo indirect activation and substantially boosted its ability to drive pathology. These results indicate a previously unappreciated role of RasGAP in regulating activation of PDGFRβ and suggest that this is one mechanism that restrains PDGFRβ-mediated disease.

MATERIALS AND METHODS

Major reagents.

Antibodies against autophagy 5 (Atg5), Atg7, Akt, phospho-Akt (p-Akt) (S473), mammalian target of rapamycin (mTOR), p-mTOR (S2448), p70 S6 kinase (S6K), p-S6K (T389), beclin 1, LC3B, p62, parkin, and Pink1 were purchased from Cell Signaling Technology (Danvers, MA). Antibodies against PDGFRα (27P), PDGFRβ (30A), p-PDGFRα (Y742), p-PDGFRβ (Y751), and RasGAP were produced and characterized as previously described (8, 33–35). The anti-Hsp90α antibody was from Affinity BioReagents (Golden, CO). The horseradish peroxidase (HRP)-conjugated goat anti-rabbit IgG secondary antibody was purchased from Santa Cruz Biotechnology (Santa Cruz, CA). Enhanced chemiluminescent substrate for detection of HRP was from Pierce Protein Research Products (Rockford, IL). Thirty percent H₂O₂, 2′,7′-dichlorofluorescein diacetate (DCFH-DA), dithiothreitol (DTT), bafilomycin a1, puromycin, tetracycline, G418, and Na₃VO₄ were purchased from Sigma (St. Louis, MO).

Cell culture.

F cells are mouse embryo fibroblasts derived from mice null for both the PDGFRα and PDGFRβ genes and then immortalized with simian virus 40 (SV40) large T antigen (31). Fα cells are F cells that reexpressed human PDGFRα. F_DR/MA cells are F cells that expressed human PDGFRα that harbored the following two mutations: D763M and R764A. Fβ cells are F cells that reexpressed human PDGFRβ, and F₇₇₁ cells are F cells that expressed human PDGFRβ harboring the following mutation: Y771F.

GPG293 cells were cultured in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% fetal bovine serum (FBS), 2 mM l-glutamine, 1 μg/ml tetracycline, 2 μg/ml puromycin, 0.3 mg/ml G418, and 16.7 mM HEPES (Life Technologies, Chicago, IL). The medium used during virus collection was DMEM supplemented with 10% FBS, 2 mM l-glutamine, and 16.7 mM HEPES. 293T cells were cultured in DMEM with 10% FBS. All the above-mentioned cells were cultured at 37°C in a humidified 5% CO₂ atmosphere.

Expressing mutants of PDGFRα and -β in F cells.

Two mutations (D763M and R764A) were generated in PDGFRα by following instructions provided with the QuikChange XL site-directed mutagenesis kit (Stratagene, La Jolla, CA) using the following set of mutagenic primers: 5′-GACATCCAGAGATCACTCTATATGGCTCCAGCCTCATATAAGAAG-3′ and its complementary oligonucleotide. The primers were synthesized by the DNA core facility of Massachusetts General Hospital (MGH) (Boston, MA). The mutated PDGFRα was named PDGFRαDR/MA and was constructed in the pLHDCX³ retroviral vector (36) and confirmed by DNA sequencing at the MGH DNA core facility. The construct was transfected into 293GPG cells, and the resulting virus was used to infect F cells. The successfully infected cells were selected for the ability to proliferate in medium containing histidinol (5 mM). The selected cells were called F_DR/MA. Expression of PDGFRαDR/MA in F cells was determined by Western blotting using an anti-PDGFRα antibody. The mutant PDGFRβ Y771F was created as previously described (17), and F cells expressing PDGFRβ Y771F were called F₇₇₁.

Immunoprecipitation and Western blotting.

F, Fα, F_DR/MA, Fβ, and F₇₇₁ cells were cultured to 80 to 90% confluence and then incubated for 24 h in DMEM. The cells were exposed to the desired agents and then washed twice with ice-cold phosphate-buffered saline (PBS). The cells were lysed in extraction buffer (10 mM Tris-HCl, pH 7.4, 5 mM EDTA, 50 mM NaCl, 50 mM NaF, 1% Triton X-100, 20 μg/ml aprotinin, 2 mM Na₃VO₄, 1 mM phenylmethylsulfonyl fluoride), and the insoluble material was removed by centrifugation for 15 min at 13,000 × g and 4°C. PDGFRα and -β were immunoprecipitated from clarified lysates using antibodies against PDGFRα or -β, respectively. The immunoprecipitated proteins were separated by SDS-10% PAGE, transferred to polyvinylidene difluoride (PVDF) membranes, and then subjected to a Western blot analysis using the indicated antibodies. At least three independent experiments were performed. Signal intensity was determined by densitometry using NIH Image J.

Active Ras pulldown and detection.

F, Fα, F_DR/MA, Fβ, and F₇₇₁ cells were cultured to 80 to 90% confluence, serum starved for 24 h, and then stimulated with rabbit vitreous (RV) for 2 h. Active Ras was pulled down by glutathione S-transferase (GST)–Raf 1 from clarified lysates, and the recovered material was subjected to Western blotting by following the instructions in an active Ras pulldown and detection kit from Thermo Scientific (Rockford, IL).

Suppression of RasGAP, Atg5, and Atg7 expression by short hairpin RNA (shRNA).

An oligonucleotide (CAGATTGTTGAAGGCTATTAT) corresponding to Mus musculus RAS p21 protein activator 1 (RasGAP) cDNA (1840 to 1860) (NM_145452.3; Open Biosystems clone identification no. TRCN0000322372), an oligonucleotide (CCAGCTCTGAACTCAATAATA) corresponding to the mouse Atg7 3′ untranslated region (UTR) (2483 to 2504) (NM_028835.1; TRCN0000092163), an oligonucleotide (AGCCTCCTCTTCTCGTGAAAT) corresponding to the mouse Atg5 3′ UTR (1315 to 1336) (NM_053069.5; TRCN0000375754), a control oligonucleotide (ACAACAGCCACAACGTCTATA) corresponding to green fluorescent protein (GFP) 437 to 457 (TRCN0000072181), the hairpin-pLKO.1 retroviral vector, the packaging plasmid (pCMV-dR8.91), the envelope plasmid (VSV-G/pMD2.G), and 293T packaging cells were from the Dana-Farber Cancer Institute/Harvard Medical School (Boston, MA).

To prepare GFP, RasGAP, Atg7, and Atg5 shRNA lentivirus, a mixture of packaging plasmid (0.9 μg), envelope plasmid (0.1 μg), hairpin-pLKO.1 vector (1 μg) (or a hairpin-pLKO.1 vector containing GFP, RasGAP, Atg7, or Atg5 shRNA oligonucleotide), and TransIt-LT1 were mixed and incubated at room temperature for 30 min. The transfection mixture was transferred to 293T cells that were approximately 70% confluent. After 18 h, the medium was replaced with growth medium modified to contain 30% FBS, and virus was harvested at 24 h after the medium switch. The viral harvest was repeated at 24-h intervals 3 times. The virus-containing media were pooled and centrifuged at 800 × g for 5 min, and the supernatant was used to infect F, Fα, F_DR/MA, Fβ, and F₇₇₁ cells. Successfully infected cells were selected on the basis of their ability to proliferate in media containing puromycin (6 μg/ml). The resulting cells were characterized by Western blotting using antibodies against RasGAP, Atg5, or Atg7, respectively.

Dichlorofluorescein assay.

The level of intracellular H₂O₂ was determined by measuring the fluorescence of cells stained with DCFH-DA. Briefly, cells were rinsed twice with Krebs-Ringer solution (118.1 mM NaCl, 3.4 mM KCl, 2.5 mM CaCl₂, 0.8 mM MgSO₄, 1.2 mM KH₂PO₄, 25.0 mM NaHCO₃, 11.1 mM glucose) and incubated in Krebs-Ringer solution containing DCFH-DA (5 μM) for 5 min. DCFH-DA is nonpolar and readily diffuses into cells, where it is hydrolyzed to the nonfluorescent polar derivative DCFH, which cannot cross the cell membrane. In the presence of H₂O₂, DCFH is oxidized to the highly fluorescent 2′,7′-dichlorofluorescein (DCF). The culture dishes were sealed with paraffin film and placed in a CO₂ incubator at 37°C for 5 min, after which they were rinsed three times with PBS, and then the fluorescence was read with a Bio-Tek fluorescence plate reader at excitation and emission wavelengths of 485 and 528 nm, respectively.

Mito-RoGFP.

The mitochondrion-localized redox-sensitive GFP mutant (Mito-RoGFP) has been described previously (37). It contains two engineered cysteine thiols that were generated by introducing the following 4 point mutations in enhanced green fluorescent protein (EGFP): C48S, Q80R, S147C, and Q204C. Localization to the mitochondria was achieved by appending a 48-bp region encoding the mitochondrial targeting sequence from cytochrome oxidase subunit IV at the 5′ end of the coding sequence. To stably express Mito-RoGFP in F, Fα, F_DR/MA, Fβ, and F₇₇₁ cells, the Mito-RoGFP cDNA in pLNCX³ was transfected into 293GPG cells to obtain retrovirus, which was used to infect F, Fα, F_DR/MA, Fβ, and F₇₇₁ cells.

The F, Fα, F_DR/MA, Fβ, and F₇₇₁ cells that expressed Mito-RoGFP were selected in G418 (4 mg/ml) and then treated with DTT (10 mM) or H₂O₂ (1 mM) for 1 h or RV for 16 h. The treated cells were observed under a Leica TCS SP5 confocal microscope and photographed at excitation wavelengths of 408 and 490 nm and an emission wavelength of 525 nm. A total of 5 photographs were collected for each treatment, and the experiment was repeated three times. The pixel density of each photograph was quantitated using the Confocal LAS AF software.

Cell proliferation assay.

Cells were seeded into 24-well plates at a density of 30,000 cells/well in DMEM plus 10% FBS. After the cells had attached (approximately 8 h), the medium was aspirated, and the cells were rinsed twice with PBS and cultured in serum-free DMEM or DMEM plus RV (1:2). The cells were counted in a hemocytometer on day 3; at least three independent experiments were performed.

Apoptosis assay.

Cells were seeded into 6-cm dishes at a density of 1 × 10⁵ cells per dish in DMEM plus 10% FBS. After the cells had attached, they were treated as described for the proliferation assay. On day 3, the cells were harvested and stained with fluorescein isothiocyanate (FITC)-conjugated annexin V and propidium iodide according to the instructions provided with the apoptosis kit (BD Biosciences, Palo Alto, CA). The cells were analyzed by flow cytometry in a Coulter Beckman XL instrument. At least three independent experiments were performed.

p53 reporter assay.

Oligonucleotides containing the p53 binding element (reporter) (38) 5′-CACGTTTGCCTTGCCTGGACTTGCCTGGCCTTGCCTTGGACATGCCCGGGCTGTCAGATCTGGGTATATAATGGA-3′ were synthesized by the DNA core facility of MGH and subcloned into the pGL3-basic vector encoding firefly luciferase (Promega, Madison, WI) by restriction endonucleases KpnI/HindIII. The resultant vector, pGL3-p53, was confirmed by DNA sequencing. F, Fα, F_DR/MA, Fβ, and F₇₇₁ cells were cultured in 96-well plates to 80 to 90% confluence and transfected with pGL3-p53 and pRL-RK encoding Renilla luciferase following the manufacturer's instructions (Dual-Luciferase reporter assay system; Promega). At 24 h after transfection, the cells were treated with RV (1:2 dilution in PBS). Two hours later, the cells were washed in PBS and lysed in passive lysis buffer (Promega) by gentle rocking for 15 min. The lysates were mixed with luciferase assay reagent II (LAR II) (Promega) in a luminometer tube, and firefly luciferase activity was recorded in a Turner BioSystems luminometer model TD-20/20 (Promega). Subsequently, Stop & Glo reagent (Promega) was added and Renilla luciferase activity was determined.

Collagen gel contraction assay.

The collagen gel contraction assay was performed as previously described (39). Briefly, cells were suspended in 1.5 mg/ml of neutralized collagen I (Inamed, Fremont, CA; pH 7.2) at a density of 10⁶ cells/ml and transferred into a 24-well plate (Falcon, Franklin Lakes, NJ) that had been preincubated with PBS plus 5 mg/ml bovine serum albumin (BSA) overnight. The gel was solidified by incubating at 37°C for 90 min and then overlaid with 0.5 ml DMEM or 1:2 DMEM plus RV. The gel diameter was measured on days 1, 2, 3, and 4. The area was calculated using the formula πr². Each experimental condition was assayed in duplicate.

Experimental PVR.

Pigmented rabbits were purchased from Covance (Denver, PA). PVR was induced in the right eye as previously described (31). Briefly, a gas-induced posterior vitreous detachment was performed by injecting 0.1 ml of 100% perfluoropropane (C₃F₈) (Alcon, Fort Worth, TX) using a 30-gauge needle on a 1-ml tuberculin syringe into the vitreous cavity 4 mm posterior to the limbus. One week later, all the rabbits received two injections: (i) 0.1 ml of platelet-rich plasma (PRP) and (ii) 0.1 ml DMEM containing the desired cells (2 × 10⁵). The retinal status was evaluated with an indirect ophthalmoscope fitted with a +30 diopter fundus lens at days 1, 3, 5, 7, 14, 21, and 28 after injection of the cells. PVR was graded according to the classification of Fastenberg et al. from 0 through 5 (40). On day 28, the animals were sacrificed, and the eyes were enucleated and frozen at −80°C. All procedures were performed under aseptic conditions and pursuant to the Association for Research in Vision and Ophthalmology (ARVO) Statement for the Use of Animals in Ophthalmic and Vision Research. The protocol for the use of animals was approved by the Schepens Animal Care and Use Committee.

To prepare the rabbit vitreous, the vitreous was dissected from the eyeball while it was still frozen, permitted to thaw, and then centrifuged at 4°C for 2 min at 13,000 × g. The resulting supernatant was used for all analyses.

On day 28, immediately prior to sacrifice, electroretinogram (ERG) analysis was performed as previously described (41). Following euthanasia, the eyes were enucleated, fixed in 10% formalin, and embedded in paraffin. The eyes were sectioned, and the resulting sections were stained with hematoxylin and eosin. Photographs of representative sections viewed with a microscope are presented.

Statistics.

The results from the rabbit studies were subjected to Mann-Whitney analysis. All other data involved a pairwise comparison between receptors that did and did not associate with RasGAP (F_DR/MA and Fα or Fβ and F₇₇₁) and were analyzed using the unpaired t test. P values of less than 0.05 were considered statistically significant.

RESULTS

Design and characterization of PDGFRs that do (WT PDGFRβ and DR/MA PDGFRα) and do not (WT PDGFRα and F771 PDGFRβ) associate with RasGAP.

Direct (PDGF-mediated) activation of PDGFRβ results in autophosphorylation of Tyr 771, which lies in an amino acid context that is favorable for association of the SH2 domains of RasGAP (14, 15–19). Replacing Tyr 771 with Phe profoundly reduces the interaction of RasGAP with activated PDGFRβ, even though the F771 PDGFRβ mutant is phosphorylated at numerous additional tyrosine residues, which enable interaction with other SH2 domain-containing proteins. While Tyr 762 in PDGFRα roughly corresponds to Tyr 771 in PDGFRβ, the sequence downstream of Tyr 762 (DRP) is not preferred by the SH2 domains of RasGAP. In an attempt to enable PDGFRα to associate with RasGAP, we replaced Asp 763 and Arg 764 with Met and Ala, respectively, which corresponds to the sequence in PDGFRβ (Fig. 1A). The resulting mutant (DR/MA), along with the other 3 PDGFRs (WT PDGFRα, WT PDGFRβ, and F771 PDGFRβ), was expressed in F cells, which are immortalized mouse embryo fibroblasts derived from mice that are null for both pdgfr genes (31). The resulting panel of cells was used to characterize the ability of the PDGFRs to associate with RasGAP and promote activation of Ras (Fig. 1B and C).

FIG 1.

Characterization of PDGFR mutants. (A) Amino acid sequence of a portion of the kinase insert of the PDGFRs. For each pair, the wild-type sequence is on top, and the residues that were changed are in boldface and boxed. (B) F, Fα, F_DR/MA, Fβ, and F₇₇₁ cells were cultured to 80 to 90% confluence and then incubated for 24 h in serum-free DMEM. The cells were exposed to RV for 2 h and lysed, and PDGFRα or PDGFRβ was immunoprecipitated using antibodies directed against PDGFRα or -β, respectively. The lysates from the F cells were immunoprecipitated with anti-PDGFRα antibodies. The immunoprecipitated proteins were subjected to Western blot analysis using the indicated antibodies. (C) Cells were treated as described for panel B and lysed, and activated Ras was isolated from the clarified lysates using a GST-Raf 1 fusion protein. The recovered proteins were subjected to Western blot analysis using an anti-Ras antibody. The blots shown in panels B and C are representative of three independent experiments. The values below the blot were calculated as follows. The band intensity was quantified and divided by the untreated (−) value. The bar graph shows the means ± standard deviations (SD) of Ras activation from three independent experiments; the asterisks indicate statistically significant differences between the indicated groups.

Because the focus of these experiments was the indirect mode of activating PDGFRs, RV was used as the stimulus. RV contains a plethora of growth factors outside the PDGF family (non-PDGFs) and no PDGFs (42, 43). Exposing cells to RV directly activates receptor tyrosine kinases, such as ErbB1, ErbB2, IGF-1R, and insulin receptor, and indirectly activates PDGFRα (26). RV promoted association of RasGAP with PDGFRβ but not PDGFRα or F771 PDGFRβ (Fig. 1B). Furthermore, the DR/MA PDGFRα mutant was able to associate with RasGAP (Fig. 1B). All 4 PDGFRs associated with the p85 subunit of phosphatidylinositol 3-kinase (PI3K), which was expected, since all of them had intact PI3K binding sites. Consistent with the ability of RasGAP to antagonize activation of Ras, the amplitude of Ras activity was lower in those cells that expressed PDGFRs that associated with RasGAP. The extent of Ras activation in such cells was comparable to the level observed in cells that expressed no PDGFRs (F cells). We conclude that, when activated indirectly, PDGFRs associate with RasGAP provided that they harbor a canonical RasGAP binding site and that the consequence of this interaction is suppression of Ras activity.

RasGAP suppressed PDGFR activation and downstream signaling events.

In light of the reduced ability of PDGFRs that associated with RasGAP to activate Ras, we speculated that the Ras-driven downstream signaling events would also be suppressed. Indeed, the amplitude of Akt and Erk activation was lower in cells expressing PDGFRs that associated with RasGAP than in cells expressing PDGFRs that did not (Fig. 2). This phenomenon was most pronounced at the latest time point (16 h). As shown in Fig. 1C, the magnitude of activation of signaling events in cells expressing PDGFRs that associated with RasGAP was comparable to the level seen in cells that expressed no PDGFRs. As mentioned above, these signaling events are likely to result from direct activation of receptor tyrosine kinases by non-PDGFs that are present in RV. When one subtracts the level of activation seen in F cells from that in PDGFR-expressing cells, it becomes apparent that association with RasGAP essentially silenced the output of the PDGFR, especially at the latest (16-h) time point.

FIG 2.

RasGAP restricts the duration of Akt and Erk activation. (A and B) F, Fα, F_DR/MA, Fβ, and F₇₇₁ cells were cultured to 80 to 90% confluence, deprived of serum for 24 h, and then left unstimulated or treated with RV for 10 min, 2 h, or 16 h. The cells were lysed, and the clarified total cell lysates were subjected to Western blot analysis using the indicated antibodies. The values below the blots were calculated as follows. The band intensity was quantified, normalized for loading, and then divided by the untreated (−) value. (C and D) The intensities of signals from Western blots were quantified for three independent experiments and expressed as a fold change in the phospho/total signal. The means ± SD are plotted. The a asterisks indicate statistically significant differences between the indicated groups.

To begin to investigate the mechanism by which RasGAP suppressed the output of PDGFRs, we considered if association with RasGAP impacted activation of the receptor itself. While association with RasGAP did not influence acute activation (at the 10-min time point), persistent activation was affected; PDGFRs that did not associate with RasGAP remained activated at the 16-h time point (Fig. 3). Activation of PDGFR was monitored with a phospho-PDGFR antibody that recognized the PI3K binding sites, which were functional regardless of RasGAP binding (Fig. 1B). We conclude that binding of RasGAP limited activation of PDGFR, which is a plausible explanation for downstream signaling events being suppressed in cells expressing PDGFRs that associate with RasGAP.

FIG 3.

RasGAP restricts the duration of PDGFR activation. Same as Fig. 2, except that total cell lysates were subjected to Western blot analysis using antibodies directed to PDGFRs. Antibodies raised against peptides, including pY742 or pY751, were used to detect the phosphorylated PDGFRα or PDGFRβ, respectively. The pan-PDGFR antibodies were specific for each of the receptors. The values below the blots were calculated as described in the legend to Fig. 2. The asterisks indicate statistically significant differences between the indicated groups.

In contrast to the comparable extents of phosphorylation of PDGFRs that do and do not bind RasGAP at the 10-min time point (Fig. 3), downstream signaling events were lower in cells expressing PDGFRs that associated with RasGAP at this time point (Fig. 2). This is likely because downstream signaling events are dependent on Ras activity, which is antagonized by RasGAP (21), whereas acute phosphorylation of PDGFRs is not (28). As discussed below, RasGAP promotes autophagy and thereby reduces mitochondrial ROS, which is required for phosphorylation of PDGFRs at the later time points (26).

The RasGAP-mediated suppression of PDGFR's output was associated with a reduced ability to elevate ROS.

We considered two mechanisms by which association with RasGAP suppressed activation of PDGFRs. The first involved a PTP, which might associate with RasGAP and/or be activated when RasGAP bound to PDGFR. If this were true, then PTP inhibitors would rescue the output of PDGFRs that associated with RasGAP. As expected, pretreating cells with a PTP inhibitor (Na₃VO₄) increased both basal and RV-stimulated phosphorylation of PDGFRs. However, the difference in the extents of PDGFR phosphorylation between PDGFRs that did and did not associate with RasGAP persisted (Fig. 4A and B). These results did not support the possibility that RasGAP reduced PDGFR activation via PTPs.

FIG 4.

RasGAP-mediated suppression of PDGFR's output was associated with a reduced ability to elevate ROS. (A and B) F, Fα, F_DR/MA, Fβ, and F₇₇₁ cells were cultured to 80 to 90% confluence, deprived of serum for 24 h, treated with 1 mM Na₃VO₄ for 30 min, and stimulated with RV for 16 h. The cells were lysed, and clarified total cell lysates were subjected to Western blot analysis using the indicated antibodies. The values below the blots were calculated as described in the legend to Fig. 2. The blots shown are representative of the results of three independent experiments. (C) F, Fα, F_DR/MA, Fβ, and F₇₇₁ cells were cultured to 80 to 90% confluence, deprived of serum for 24 h, and stimulated with RV for 16 h. The cells were rinsed with Krebs-Ringer solution and incubated with 5 μM DCFH-DA for 5 min in a CO₂ incubator at 37°C. The fluorescence was read with a Bio-Tek fluorescence plate reader at excitation and emission wavelengths of 485 and 528 nm, respectively. The results from three independent experiments were pooled, and the means ± SD are shown; the asterisks indicate statistically significant differences between the indicated groups.

The second mechanism we investigated involved ROS, because a rise in ROS is essential for RV-mediated activation of PDGFRα (25, 26). More specifically, we considered if binding RasGAP attenuated the PDGFR's ability to elevate ROS. Figure 4C shows that this was true. PDGFRs that associated with RasGAP were unable to increase the total cellular level of ROS. We concluded that the RasGAP-mediated suppression of PDGFR's output was associated with a reduced ability to elevate ROS.

RasGAP limited the output of indirectly activated PDGFR by preventing a decline in autophagy.

Our recent observations (26), together with the data presented in Fig. 1 to 4, suggest that RasGAP limited the output of PDGFR by preventing a decline in autophagy. These findings included resolution of the mechanism by which the indirect mode of activating PDGFRα resulted in its enduring activation. Namely, activating PDGFRα suppresses autophagy and thereby elevates mitochondrial ROS, which drives a feed-forward loop that is responsible for persistently activating PDGFRα. Because of these previous observations, we considered the possibility that RasGAP prevented enduring activation of PDGFRs by preventing the PDGFR-dependent decline in autophagy following exposure to RV. Autophagy (monitored by the levels of beclin 1, LC3, and p62) and mitophagy (monitored by the levels of parkin and Pink 1) declined (relative to F cells, which express no PDGFRs) in cells expressing PDGFRs that were unable to associate with RasGAP (Fig. 5A). In contrast, this reduction was not observed in cells expressing PDGFRs that associated with RasGAP (Fig. 5A). Thus, there was an association between suppression of autophagy, failure to associate with RasGAP, and enduring activation of PDGFRα.

FIG 5.

RasGAP binding to PDGFRs prevents a decline in autophagy/mitophagy. (A) F, Fα, F_DR/MA, Fβ, and F₇₇₁ cells were cultured to 80 to 90% confluence, deprived of serum for 24 h, and stimulated with RV for 16 h. The cells were lysed, and the clarified total cell lysates were subjected to Western blot analysis using the indicated antibodies. Hsp90α served as a control for loading. The values below the blots were calculated as described in the legend to Fig. 2. (B) Parental F cells were treated for 3 h with vehicle (ethanol) or bafilomycin a1 (BAF) (100 nM), and then lysed. The lysates were also prepared from unstimulated cells which had been infected with a lentivirus harboring shRNAs directed against gfp, atg5, or atg7. The resulting cell lysates were clarified and then subjected to Western blot analysis using the indicated antibodies. The values below the blots were calculated as follows. The band intensity was quantified, normalized for loading, and then divided by the vehicle value. The blots are representative of the results of three independent experiments.

If preserving autophagy was essential for RasGAP to subdue activation of PDGFRs, then blocking autophagy should overcome the repressive affect of RasGAP. Silencing expression of Atg5 or Atg7 suppressed autophagy/mitophagy (Fig. 5B). In such cells, the repressive effect of RasGAP was attenuated; PDGFRs that associated with RasGAP were activated to nearly the same extent as PDGFRs that did not associate with RasGAP (Fig. 6 and data not shown). Similarly, when autophagy/mitophagy was compromised, activation of signaling events downstream of PDGFR (Akt, mTOR, and S6K) were no longer suppressed in cells expressing PDGFRs that associated with RasGAP (Fig. 6 and data not shown). Furthermore, the difference in kinetics of Erk activation between cells expressing PDGFRs that did and did not associate with RasGAP (Fig. 3) was greatly reduced because Erk activation by PDGFRs that associated with RasGAP improved when autophagy/mitophagy was compromised (data not shown). These data indicate that RasGAP suppressed the output of PDGFRs by preventing the decline in autophagy that occurred in cells expressing PDGFRs unable to associate with RasGAP.

FIG 6.

Antagonizing autophagy improves the output of PDGFRs that associate with RasGAP. (A) F, Fα, F_DR/MA, Fβ, and F₇₇₁ that had been infected with lentiviruses containing shRNAs directed toward gfp or atg5 were lysed and subjected to Western blot analysis using the indicated antibodies. The band intensities were quantified and normalized for loading, and the values in the shATG5 cells were divided by the value from the corresponding shGFP cell. Expression of Atg5 was reduced by 90% in shAtg5 cells. (B and C) The cells from panel A were cultured to 80 to 90% confluence, deprived of serum for 24 h, and stimulated with RV for 16 h. The cells were lysed, and the clarified total cell lysates were subjected to Western blot analysis using the indicated antibodies. The values below the blots were calculated as described in the legend to Fig. 2. The blots shown are representative of the results of three independent experiments.

Bafilomycin, which causes autophagosomes to accumulate by preventing their fusion with the lysosome (e.g., the increase in LC3 II in Fig. 5B), was not useful in this series of experiments. This was because cells could not tolerate exposure to bafilomycin for 16 h, which was the duration of exposure to RV (data not shown).

As in PDGFR-expressing cells, we consistently observed that RV-stimulated signaling events (Akt, mTOR, and S6K) were enhanced in F cells, which expressed no PDGFRs. These findings indicate that RasGAP-mediated suppression of PDGFRs was not the only way in which autophagy antagonized signaling events, such as Akt, mTOR, and S6K.

The finding that RasGAP preferentially reduced the extent of PDGFR phosphorylation at the prolonged time points (Fig. 3) suggested that it was governing the mitochondrial source of ROS. This is because ROS generated by the mitochondria are responsible for phosphorylation of PDGFRα at the late time points, whereas acute phosphorylation of PDGFRα is dependent on ROS generated by NADPH oxidases (26). These previous findings, as well as the observation that the effect of RasGAP was dependent on autophagy (which regulates the mitochondrial source of ROS), led us to posit that there will be less mitochondrial ROS generated in response to RV stimulation of cells expressing PDGFRs that associate with RasGAP than of cells expressing PDGFRs that are unable to associate with RasGAP. To test this idea, we generated a panel of PDGFR-expressing cells that stably expressed a mitochondrion-localized redox-sensitive GFP mutant (37), which previously allowed us to ascertain that mitochondrial ROS was required for persistent activation of PDGFRα (26). Control experiments demonstrated that the emission ratio measured at 484 and 400 nm decreased when ROS increased (Fig. 7B). Representative cells are shown in Fig. 7A; Fig. 7C summarizes the quantified results from three independent experiments. The results show that cells expressing PDGFRs that did not associate with RasGAP caused a greater decline (generated more mitochondrial ROS) than PDGFRs that did associate with RasGAP.

FIG 7.

RasGAP restricts mitochondrial ROS. (A and B) F, Fα, F_DR/MA, Fβ, and F₇₇₁ cells stably expressing the mitochondrial ROS sensor (mitochondrion-localized redox-sensitive GFP mutant [37]) were cultured to near (80 to 90%) confluence, deprived of serum for 24 h, and left resting or stimulated with 1 mM H₂O₂ for 30 min, with 1 mM DTT, for 1 h or with RV for 16 h. The images in panel A are of cells viewed using a confocal microscope; the excitation wavelengths were 400 and 484 nm, and the emission wavelength was 525 nm. For each experimental condition, five different areas containing approximately 30 cells were quantified. (C) The results from three independent experiments were pooled, and the means ± SD are presented; the asterisks indicate statistically significant differences between the indicated groups. (D) Diagram summarizing the results in Fig. 5 and 6 and this figure. PDGFRs that do not bind RasGAP engage a signaling pathway (shown in the box on the right), which suppresses autophagy and thereby drives a mitochondrial-ROS-driven feed-forward loop (shown in the box on the left) by which the PDGFR stays active. Association with RasGAP antagonizes this signaling pathway and thereby attenuates the decline in autophagy that is required to drive the feed-forward loop.

The results shown in Fig. 5 to 7 are summarized in Fig. 7D. PDGFRs that do not bind RasGAP engage a signaling pathway (shown in the box on the right) that suppresses autophagy and thereby drives a mitochondrial ROS-driven feed-forward loop (shown in the box on the left) by which the PDGFR stays active. Association with RasGAP antagonizes this signaling pathway and thereby attenuates the decline in autophagy that is required to drive the feed-forward loop.

RasGAP was the PDGFR binding protein responsible for suppressing its output.

More than a single SH2 domain-containing protein may bind to a given docking site. For instance, phosphorylation of two tyrosine residues with the carboxy terminus of the hepatocyte growth factor receptor enables association with four different SH2 domain-containing proteins (44). Consequently, while RasGAP is the only SH2 domain-containing protein that is known to bind with high affinity to the YMAP motif of PDGFRβ, it is possible that other SH2 domain-containing proteins bind to this site and contribute to inhibiting activation of PDGFRs. One way to test this possibility is to silence expression of RasGAP. If RasGAP is essential for suppression of PDGFR activation, then YMAP motif-containing PDGFRs will be activated, as well as PDGFRs that lack the motif, in cells that have reduced expression of RasGAP. This was what we observed (Fig. 8). Similarly, activation of Akt in cells expressing YMAP motif-containing receptors improved when RasGAP was silenced (Fig. 8). These results, together with those shown in Fig. 1B, indicate that RasGAP binds to PDGFRs within the YMAP motif and is essential for inhibiting the activation and output of PDGFRs.

FIG 8.

RasGAP was essential for suppressing the activation of PDGFRs. F, Fα, F_DR/MA, Fβ, and F₇₇₁ cells that had been infected with lentiviruses containing shRNAs directed toward gfp or Rasa1 were lysed and subjected to Western blot analysis using antibodies directed against RasGAP or Hsp90α. Expression of RasGAP was reduced by 90% in shRasGAP cells. The cells were cultured to 80 to 90% confluence, deprived of serum for 24 h, and stimulated with RV for 16 h. The cells were lysed, and the clarified total cell lysates were subjected to Western blot analysis using the indicated antibodies. The values below the blots were calculated as described in the legend to Fig. 2. The blots shown are representative of the results of three independent experiments.

RasGAP suppressed signaling events and cellular responses that are associated with PVR.

Indirect activation of PDGFRα drives a series of signaling events and cellular responses that are associated with PVR and is essential for development of the disease in experimental animals (23). The realization that RasGAP inhibits indirect activation of PDGFR provides a plausible explanation for the inability of PDGFRβ to induce PVR in the way that PDGFRα can (31). We performed the following series of experiments to directly test the hypothesis that RasGAP prevents PDGFRs from driving signaling events and cellular responses associated with PVR.

Indirectly activated PDGFRα triggers sustained activation of Akt and a decline in p53 reporter activity (32). Both of these signaling events were attenuated in cells expressing PDGFRs that associated with RasGAP (Fig. 2 and 9A). Furthermore, cellular responses that are associated with PVR (protection from apoptosis, proliferation, and contraction of collagen gels) were all attenuated in cells expressing PDGFRs that associated with RasGAP (Fig. 9B and C and 10). These results indicate that RasGAP antagonizes signaling events and cellular responses that are associated with PVR.

FIG 9.

Association of RasGAP with PDGFRs increases p53 activation and apoptosis. (A) F, Fα, F_DR/MA, Fβ, and F₇₇₁ cells were cotransfected with pGL3-p53 (a p53 activity reporter) and pRL-TK (to control for the efficiency of transfection) plasmids. After 12 h, the cells were stimulated with RV for 16 h, and the activities of the reporter constructs were monitored using a dual-luciferase reporter assay system. (B) The cells from panel A were cultured to 80 to 90% confluence, deprived of serum for 24 h, and then treated or not with RV for 16 h. The percentage of apoptotic cells was determined by fluorescence-activated cell sorter (FACS) analysis of cells stained with FITC-conjugated annexin V and propidium iodide (PI). Cells in early apoptosis were defined as the population that was positive for annexin V but negative for propidium iodide. For both panels A and B, the results from at least three independent experiments were pooled, and the means ± SD are presented; the asterisks indicate statistically significant differences between the indicated groups. (C) Representative set of raw FACS data.

FIG 10.

Association of RasGAP with PDGFRs decreases RV-induced cell proliferation and collagen gel contraction. (A) F, Fα, F_DR/MA, Fβ, and F₇₇₁ cells (3 × 10⁴) were plated in individual wells of a 24-well dish in serum-free medium that was or was not supplemented with RV. After 3 days, the cells were counted. (B) F, Fα, F_DR/MA, Fβ, and F₇₇₁ cells were embedded in a collagen matrix at a density of 10⁶ cells/ml and cultured in serum-free medium that was supplemented or not with RV. The diameter of the collagen gel was measured on day 3. For both panels A and B, the results from at least three independent experiments were pooled, and the means ± SD are presented; the asterisks indicate statistically significant differences between the indicated groups.

Preventing PDGFRβ from associating with RasGAP improved its PVR potential.

The results presented thus far predict that preventing PDGFRβ from associating with RasGAP will improve its ability to induce PVR. To test the PVR potential of the F771 PDGFRβ mutant, we intravitreally injected rabbits with F cells expressing similar levels of the mutant or WT PDGFRβ (Fig. 3B). Consistent with previous results (31), cells expressing WT PDGFRβ failed to effectively induce PVR; the vast majority of these rabbits did not develop retinal detachment (stage 3 or higher) (Fig. 11A), the retina retained electrical functionality (Fig. 11B), and the histology of the retina was unremarkable (Fig. 11C). In contrast, the majority of the rabbits injected with cells expressing the F771 mutant developed retinal detachment, the retina became nonfunctional, and an epiretinal membrane was apparent (Fig. 11). As indicated in Fig. 11A, differences between the two experimental groups became apparent on day 5 and persisted until week 3. These results strongly support the idea that PDGFRβ is fully capable of inducing PVR but fails to do so because it associates with RasGAP, which prevents it from signaling persistently and thereby facilitating pathogenesis.

FIG 11.

RasGAP binding suppresses the PVR potential of PDGFRβ. PVR was induced in the right eyes of pigmented rabbits as described in Materials and Methods. All rabbits received two injections: (i) 0.1 ml of PRP and (ii) 0.1 ml DMEM containing 2.0 × 10⁵ Fβ or F₇₇₁ cells. The retinal status was evaluated on days 1, 3, 5, 7, 14, 21, and 28. PVR was graded according to the Fastenberg classification from 0 through 5. (A) Clinical status of each rabbit at the indicated time points. The asterisks indicate statistically significant differences between the two experimental groups. (B) Results of ERG analysis that was performed on day 28 on rabbits that were injected with Fβ cells and had a clinical score of 0 (left), that were injected with F₇₇₁ cells and had a clinical score of 0 (middle), or that were injected with F₇₇₁ cells and had a clinical score of 5 (right). (C) Following euthanasia, the eyes were enucleated, fixed, embedded in paraffin, sectioned, stained with hematoxylin and eosin, and photographed. Representative images are presented; the scale bar is 100 μm. The cells that were injected and the clinical score for each eye are indicated at the top left.

It is noteworthy that a subset of rabbits injected with cells expressing the F771 PDGFRβ mutant remained disease free (Fig. 11A). While the response within experimental groups is typically heterogeneous, it is rare to have a complete lack of response. The retinas in these animals remained functional, and the histology was overtly normal (Fig. 11B and C). The existence of this subgroup suggests that recruitment of RasGAP may not be the only mechanism to suppress activation and/or signaling downstream of PDGFRβ.

DISCUSSION

We report that RasGAP antagonizes indirect activation of PDGFRβ and describe the underlying mechanism. RasGAP prevents the mitochondrial-ROS-driven feed-forward loop by which the PDGFR stays activated (26). Finally, PDGFRβ that is no longer subject to regulation by RasGAP displays a marked increase in its ability to promote pathology.

The negative impact of RasGAP on activation of PDGFRs is more apparent at the prolonged time points (2 or 16 h). This is because enduring activation of PDGFRs is dependent on mitochondrially produced ROS (26), which is the step in the activation process that RasGAP antagonizes (Fig. 7D). In contrast, the magnitude of PDGFR phosphorylation at the early time points is largely unaffected by association of RasGAP. Thus, RasGAP appears to function as a determinant of how long PDGFRs stay active. Duration is a key variable in signal transduction; the duration of Erk activity can determine if activating Erk results in altered expression of genes and, hence, the nature of the ensuing cellular response (45).

While ROS contributes to both the direct and indirect modes of activating PDGFRs, the position within the activation process and the effectors via which ROS acts are distinct. In the context of direct activation, ROS acts downstream of the activated PDGFRs to transiently inhibit PTPs and thereby increase the amplitude and/or duration of PDGFR phosphorylation (22). In the indirect scenario, ROS functions upstream of PDGFRs by activating SFKs, which phosphorylate the PDGFR and stimulate its kinase activity (26). In addition, ROS promotes a feed-forward loop that is essential for persistent activation of indirectly activated PDGFRα (26). Thus, there are fundamental differences regarding the contributions of ROS to the two modes of activating PDGFRs.

It is unlikely that autophagy/mitophagy negatively impacts PDGF-mediated activation of PDGFRs. This is because PDGF activates PDGFRs transiently and in a ROS-independent manner (25, 26, 28). Hence, PDGF-mediated receptor activation and downstream signaling events are suppressed by the time that autophagy becomes relevant. Indeed, PDGF-mediated activation of PDGFRβ is unaffected by binding of RasGAP (46, 47).

RasGAP's ability to antagonize Ras provides a plausible explanation for its suppressive influence on PDGFR-stimulated signaling described in this report. However, since RasGAP contains a variety of functional domains and complexes with several signaling proteins (48), our intentional focus on Ras antagonism may incompletely describe the mechanism of RasGAP's influence on the output of PDGFRs. In support of this idea are results from studies with PDGFRβ mutants. Binding of RasGAP to PDGFRβ reduces the amount of PI3K- and phospholipase C-γ (PLC-γ)-generated products that accumulate in PDGF-stimulated cells (47, 49). While the antagonism of Ras maybe the reason that PI3K was inhibited (Ras promotes PI3K activity [49, 50]), how RasGAP attenuates tyrosine phosphorylation of PLC-γ is not obvious (47). Additional studies are required to determine if RasGAP-mediated events, other than reducing the level of active Ras, contribute to blocking indirect activation of PDGFRβ.

Characterization of mice expressing PDGFR chimeras indicates that PDGFRβ is more capable than PDGFRα (6, 7). This is unlikely to be related to indirect activation for the following two reasons. First, PDGFRβ does not undergo indirect activation, at least under standard tissue culture conditions. Second, the phenotypes of mice lacking ligands and those lacking receptors are quite similar (2). Therefore, direct activation of PDGFRs is likely to be the predominant mode during development. If indirect activation was contributing, then mice lacking the receptors (and hence both direct and indirect activation) would have a more severe phenotype than ligand-null mice, in which PDGFRs could still be activated indirectly. The nature of the additional capabilities of PDGFRβ relative to PDGFRα remains an intriguing and unanswered question.

The realization that PDGFRβ is fully capable of undergoing indirect activation begs the question of whether it ever does, and if so, what the relevance is. RasGAP is the only impediment, and there are at least two ways to overcome it. Perturbing the motif within PDGFRβ that is necessary for its association with RasGAP is one approach. Within the single-nucleotide polymorphism database (dbSNP), there are entries that are predicted to do this; however, they are not associated with disease, and hence, this clue remains very much undeveloped. A second way to enable PDGFRβ to undergo indirect activation is by reducing expression of RasGAP. MicroRNAs (such as miR-132) with this capacity have been identified (51). Whether changes that enable PDGFRβ to undergo indirect activation occur and contribute to PDGFRβ-related pathology (e.g., cardiovascular disease, cancer, and fibrotic diseases [2]) becomes an open question in light of the findings presented here.