A Novel Phosphorylation-Dependent RNase Activity of GAP-SH3 Binding Protein: a Potential Link between Signal Transduction and RNA Stability

Abstract

A potential p120 GTPase-activating protein (RasGAP) effector, G3BP (RasGAP Src homology 3 [SH3] binding protein), was previously identified based on its ability to bind the SH3 domain of RasGAP. Here we show that G3BP colocalizes and physically interacts with RasGAP at the plasma membrane of serum-stimulated but not quiescent Chinese hamster lung fibroblasts. In quiescent cells, G3BP was hyperphosphorylated on serine residues, and this modification was essential for its activity. Indeed, G3BP harbors a phosphorylation-dependent RNase activity which specifically cleaves the 3′-untranslated region of human c-myc mRNA. The endoribonuclease activity of G3BP can initiate mRNA degradation and therefore represents a link between a RasGAP-mediated signaling pathway and RNA turnover.

The Ras protein belongs to a family of low-molecular-weight GTPases which are essential components of multiple receptor-mediated signal transduction pathways controlling cell proliferation, differentiation, and cytoskeletal organization (23). Activated Ras is bound to GTP, while the GDP-bound form of Ras is inactive (27). Extracellular stimuli induce the exchange of GDP for GTP on Ras through a series of protein-protein interactions involving activated receptors, adaptor proteins (such as Grb2 or Shc), and Ras guanine nucleotide exchange factors (5, 9, 33, 38). Mutations in the Ras gene which lock Ras in the GTP-bound form lead to cell growth in the absence of mitogenic signals and are associated with an oncogenic phenotype (17). Physiological inactivation of Ras involves interaction with GTPase-activating proteins (GAPs) (40), such as p120 (RasGAP) (41, 43) or the product of the NF1 gene (neurofibromin) (26, 44), which accelerate the hydrolysis of Ras-associated GTP, thereby converting Ras from an active to an inactive form. Disruption of either the RasGAP or the NF1 gene in mice results in an embryonic lethal phenotype (3, 14), indicating that Ras inactivation is a key process in normal cell signaling and development.

In addition to being a negative regulator of Ras, RasGAP may also represent a downstream target of Ras (35). RasGAP is a widely expressed modular protein which comprises several structural features that likely enable it to function in the transduction cascade (29). While the carboxyl-terminal domain of RasGAP constitutes a catalytic domain (25), the N-terminal region is believed to mediate interactions with other signaling proteins (20). The N-terminal region is characterized by a Src homology 3 (SH3) domain flanked by two SH2 domains, as well as pleckstrin homology (PH) and calcium-dependent lipid binding domains (4, 34). Upon activation of many growth factor receptors, RasGAP becomes phosphorylated and associates with cytosolic proteins as well as with the autophosphorylated tyrosine kinase receptors (19). RasGAP has been shown to form a complex with G3BP (RasGAP SH3 binding protein) in a Ras-GTP-dependent manner (32). G3BP is composed of 466 amino acid and has a predicted molecular mass of 52 kDa; the carboxyl-terminal region contains a structural motif implicated in RNA binding, the RRM-type domain (6). However, the exact function of G3BP in RasGAP-dependent signaling remains to be defined.

In this study, we analyzed the RNA binding function of G3BP along with its ability to interact with RasGAP during cell proliferation. Using immunopurified G3BP from mammalian cells and highly purified recombinant G3BP, we demonstrated that G3BP harbors an intrinsic endonuclease activity that cleaves the 3′-untranslated region (3′-UTR) of human c-myc mRNA. The phosphorylation-dependent RNase activity of G3BP and its interaction with RasGAP suggest that the latter constitutes a link between the Ras signal transduction pathway and mRNA decay.

MATERIALS AND METHODS

Cell culture and antibodies.

Chinese hamster lung fibroblasts (CCL39 fibroblasts) were generously provided by J. Pouysségur (Université de Nice, Nice, France) and were maintained in Dulbecco’s modified Eagle’s medium (DMEM) supplemented with 10% fetal calf serum and antibiotics (50 U of penicillin and 50 μg of streptomycin per ml) at 37°C in 5% CO₂–95% air. They were rendered quiescent by serum starvation for 24 h. CCL39 fibroblasts transformed with Ha-ras were cultured as described previously (39).

Affinity-purified monoclonal antibody (MAb) against RasGAPs (15F8) was obtained in mice by immunization with the carboxyl-terminal GAP domain (residues 702 to 1044) (Rhône-Poulenc Rorer). A MAb against RasGAP SH3 (GP200) was the generous gift of J. Grassi (Gif-sur-Yvette, France). Affinity-purified polyclonal antibodies to G3BP were obtained as described previously (32). An anti-G3BP MAb (1F1) was directed against full-length recombinant G3BP (Rhône-Poulenc Rorer). The anti-p21^ras antibody was from Oncogene Science, the anti-Sos antibody was from AFFINITI, and the anti-epidermal growth factor receptor (EGFR) antibody was from TEBU. Polyclonal antibodies against glyceraldehyde-3-phosphate dehydrogenase (GAPDH) were kindly provided by J. M. Blanchard (Montpellier, France).

Subcellular fractionation.

Cell lysates were typically prepared from 10⁹ adherent cells, which were collected with a cell scraper and harvested by centrifugation in a Beckman J.S 7.5 rotor for 5 min at 1,200 rpm and 4°C. The pelleted cells (2 ml) were washed twice with 10 ml of cold phosphate-buffered saline (PBS) and centrifuged in a Beckman J.S7.5 rotor for 5 min at 1,200 rpm. The cells were subsequently washed in 10 ml of ice-cold buffer A, consisting of 10 mM triethanolamine (pH 7.9), 10 mM KCl, 1.5 mM MgCl₂, 0.5 mM dithiothreitol (DTT), 2 μg of leupeptin per ml, and 0.5% aprotinin, centrifuged in a Beckman J.S7.5 rotor for 5 min at 1,200 rpm and 4°C, and resuspended in 4 ml of buffer A. At this step, the cells were still intact but were inflated (twice their initial volume), as determined by phase-contrast light microscopy. These cells were lysed with 10 strokes of a type B pestle in a Dounce homogenizer. Cell lysates were centrifuged in a Beckman J.S7.5 rotor for 10 min at 3,000 rpm and 4°C. The supernatant was mixed with 0.11 volume of buffer B, consisting of 300 mM HEPES (pH 7.9), 1.4 M KCl, and 30 mM MgCl₂, and centrifuged in an SW60 rotor (Beckman) for 60 min at 100,000 × g. The supernatant (S100) and the pellet (P100) were separated, and the S100 fractions were adjusted to 5% glycerol and stored at −70°C. To prepare the nuclear extracts, the pelleted nuclei were centrifuged in a Beckman J.S13.1 rotor for 20 min at 10,000 rpm. The pellets were resuspended in 2 ml of buffer C, consisting of 20 mM triethanolamine (pH 7.9), 25% glycerol, 0.2 mM EDTA, 0.4 M NaCl, 1.5 mM DTT, 2 μg of leupeptin per ml, 0.5% aprotinin, and 0.5 mM phenylmethylsulfonyl fluoride (PMSF), and homogenized with 10 strokes of a type A pestle in a glass Dounce homogenizer. The lysates were incubated for 1 h at 4°C under mild agitation and then centrifuged in a Beckman SW60 rotor for 25 min at 15,500 rpm. The recovered supernatants were dialyzed for 3 h against buffer D, consisting of 20 mM triethanolamine (pH 7.9), 100 mM KCl, 10% glycerol, 0.5 mM DTT, 0.2 mM EDTA, and 0.5 mM PMSF, and stored at −70°C.

Total extracts were prepared as described previously (32). Protein concentrations were determined by the Bio-Rad protein assay with bovine serum albumin as a standard.

Immunoprecipitation and immunoblotting.

Immunoprecipitations were performed as described previously (32). Briefly, immune complexes were collected by centrifugation, washed three times with buffer F, consisting of 20 mM HEPES (pH 7.5), 150 mM NaCl, 0.1% Triton X-100, 1 mM EGTA, 10 mM pyrophosphate, 1 mM MgCl₂, 1 mM Na₃VO₄, 10 mM Na₄P₂O₇, 100 mM NaF, 1 μg of leupeptin per ml, 1 μg of trypsin inhibitor per ml, 1 μg of pepstatin A per ml 2 μg of aprotinin per ml, 10 μg of benzamidine per ml, 1 mM PMSF, 1 μg of antipain, and 1 μg of chymostatin per ml, and boiled for 10 min in 2× Laemmli sample buffer (50 μl). Proteins were resolved on sodium dodecyl sulfate (SDS)–10% polyacrylamide gels and electrophoretically transferred to nitrocellulose membranes in 10 mM 3-(cyclohexylamino)-1-propanesulfonic acid (pH 11.0) containing 10% methanol for 2 h. The nitrocellulose membranes were incubated overnight at 4°C in blocking solution (5% dried milk in PBS) and then incubated with the following dilutions of primary antibody in blocking solution for 1 h at room temperature: 1/200 for the anti-G3BP antibody (200 ng/ml) (32), 1/500 for the anti-RasGAP antibody (GP200) (32) in blocking solution supplemented with 0.05% Tween 20, 1/500 for the anti-p21^ras antibody, 1/400 for the anti-Sos antibody, 1/1,000 for the anti-EGFR antibody, and 1/300 for the anti-GAPDH antibody. The nitrocellulose filters were washed three times in PBS, and the bound antibodies were detected with an appropriate anti-immunoglobulin G–horseradish peroxidase conjugate followed by enhanced chemiluminescence (Amersham) according to the manufacturer’s protocols.

³²P labeling and phosphoamino acid and phosphopeptide mapping.

For metabolic labeling (see Fig. 3), 500 μCi of ³²P_i (Amersham) was used per 60-mm dish of proliferating or quiescent CCL39 fibroblasts. Labeling was carried out with 6 ml of phosphate-free DMEM, in either the presence or the absence of 10% phosphate-free fetal calf serum, for 4 or 8 h at 37°C in a CO₂ incubator. Cells were lysed as described above, and lysates were precleared for 1 h with 50 μl of protein G-Sepharose beads (Pharmacia). The lysates were clarified by centrifugation at 12,000 × g for 30 s at 4°C. An aliquot of the lysates was used to determine the protein concentration by the Bio-Rad protein assay with bovine serum albumin as a standard. Equal amounts of proteins were immunoprecipitated with an anti-G3BP antibody. Immunocomplexes were collected on protein G-Sepharose beads, washed three times with buffer F and then three times with PBS, solubilized in 2× Laemmli sample buffer, boiled for 5 min, and analyzed by SDS-polyacrylamide gel electrophoresis (PAGE) on a 10% gel. Labeled proteins were transferred to polyvinylidene difluoride membranes and detected by autoradiography. Membrane slices containing radioactive G3BP were excised, and ³²P counts were determined with a liquid scintillation counter. Membrane-bound G3BP was hydrolyzed with 6 N HCl, and the resulting amino acids were separated by two-dimensional electrophoresis (2). ³²P-labeled amino acids were detected by autoradiography and identified by comigration with internal cold standards for phosphoserine (Pser), phosphothreonine (Pthr), and phosphotyrosine (Ptyr).

FIG. 3.

G3BP becomes hyperphosphorylated on Ser and Thr residues following serum starvation of CCL39 fibroblasts. (A) Proliferating (+FCS [with fetal calf serum]) and resting (−FCS [without fetal calf serum]) CCL39 fibroblasts were metabolically labeled with ³²P_i for 4 or 8 h, and phosphorylated G3BP was affinity purified with an anti-G3BP antibody (α G3BP), separated by gel electrophoresis, transferred to a PVDF membrane, and detected by autoradiography. IP, immunoprecipitates. (B) Phosphoamino acid analysis of labeled G3BP. Equal amounts of G3BP, as estimated by Ponceau red staining, were excised, subjected to Cerenkov counting, and analyzed for phosphoamino acids. Numbers below the panels show total counts per minute incorporated into affinity-purified G3BP. (C) (Panel I) Comparison of G3BP phosphorylation levels between total extracts derived from dividing (lane 1) and quiescent (lane 2) CCL39 fibroblasts. Proteins were subjected to IEF gel analysis as described in Materials and Methods, and G3BP phosphorylation variants were revealed by immunoblotting with an anti-G3BP antibody. (Panels II and III) Total extracts (20 μg) derived from proliferating (panel II) or quiescent (panel III) CCL39 fibroblasts were incubated at 30°C for 1 h in the absence (lane 2) or the presence (lane 3) of 200 U of calf intestinal alkaline phosphatase (IAP) and analyzed as described for panel I. CTL, control fraction. (D) Phosphotryptic peptide mapping of immunopurified ³²P-labeled G3BP from dividing (panel I) and quiescent (panel II) cells. Panel I+II represents the analysis of the mixture of the samples in panels I and II. ori, origin.

For tryptic phosphopeptide mapping, quiescent and dividing fibroblasts were ³²P labeled for 8 h, and G3BP was immunopurified as described above. Labeled proteins from the immunocomplexes were analyzed by SDS-PAGE on a 10% gel and stained with Coomassie blue. After the gel was dried, the G3BP band was excised and digested (in the gel) as described previously (36). Phosphopeptide analysis was performed as described previously (2) with pH 1.9 buffer in the first dimension and phosphochromatography buffer in the second dimension.

Chemical cross-linking.

Proliferating or quiescent CCL39 fibroblasts (5 × 10⁷ cells) were collected with a cell scraper and washed twice with cold PBS. The cells were resuspended in 1 ml of buffer G, consisting of 100 mM morpholinepropanesulfonic acid (MOPS) (pH 7), 150 mM NaCl, 1 mM EGTA, 10 mM pyrophosphate, 100 mM NaF, 1.5 mM MgCl₂, 1 mM Na₃VO₄, 10 μg of aprotinin per ml, 10 μg of leupeptin per ml, and 100 μM PMSF, and lysed with 10 strokes of a type B pestle in a Dounce homogenizer. Total extracts were obtained after centrifugation of the lysates in a Beckman SW60 rotor at 13,000 rpm for 10 min at room temperature. Cross-linking was done with 400-μl reaction mixtures consisting of 200 μl of total extracts (2 mg of protein per ml) supplemented with 10 mM 1-ethyl-3(dimethylamino)-propyl carbodiimide (EDC) (Sigma) and 20 mM N-hydroxysuccinimide (Sigma). After incubation for 20 min at 25°C, half of each reaction mixture was mixed with 2× Laemmli loading buffer (200 μl), and aliquots (50 μl) were loaded onto SDS–10% polyacrylamide gels. Fractionated proteins were analyzed by immunoblotting with anti-G3BP or anti-RasGAP antibodies. For the other half of each sample, the reaction was terminated with 300 mM glycine, and the sample was subjected to immunoprecipitation with anti-G3BP or anti-RasGAP antibodies before immunoblot analysis.

Purification of recombinant G3BP.

Monolayer cultures of Sf9 cells were infected with recombinant baculovirus expressing G3BP (32) at a multiplicity of infection of 5 to 10 and harvested by centrifugation 4 days postinfection. Infected cells were lysed in 120 ml of HNTG buffer, consisting of 50 mM HEPES (pH 7.5), 150 mM NaCl, 1% Triton X-100, 1 mM EGTA, 10 mM sodium pyrophosphate, 1.5 mM MgCl₂, 1 mM Na₃VO₄, 100 mM NaF, 1 μg of leupeptin per ml, 1 μg of trypsin inhibitor per ml, 1 μg of pepstatin A per ml, 2 μg of aprotinin per ml, 10 μg of benzamidine per ml, 1 mM PMSF, 1 μg of antipain, and 1 μg of chymostatin per ml, and centrifuged at 15,000 × g for 15 min. The supernatant was diluted fivefold in HG wash buffer, consisting of 50 mM HEPES (pH 7.5), 1 mM EGTA, 10% glycerol, and phosphatase and protease inhibitors, and incubated for 12 to 14 h in a heparin-Sepharose Fast Flow gel (Pharmacia) equilibrated with HNTG buffer supplemented with 50 mM HEPES (pH 7.5), 30 mM NaCl, 1 mM MgCl₂, 1 mM EGTA, 0.2% Triton X-100, and 10% glycerol. Subsequently, the mixture, consisting of 15 mg of total protein per 1 ml of gel, was poured into a K26 column (5 cm² by 5.6 cm; Pharmacia) and washed with 10 volumes of equilibration buffer at a flow rate of 50 ml h⁻¹. Bound proteins were eluted stepwise with increasing salt concentrations (from 100 to 600 mM NaCl). Far-Western blotting and immunoblotting were performed (32) to identify the fractions containing G3BP, which was found to elute at 600 mM NaCl. The G3BP fractions were pooled, diluted 10-fold in HG wash buffer, and passed over an agarose-poly(U) type 6 column (19 cm² by 1 cm; Pharmacia) preequilibrated with supplemented HNTG buffer at a flow rate of 5 ml/cm² h⁻¹. In order to remove weakly bound proteins, the column was washed with HNTG buffer containing 60 mM NaCl and eluted with a 229-ml linear 60 to 320 mM NaCl gradient. Subsequently, G3BP-containing fractions were eluted at 700 mM NaCl. The G3BP fractions from several poly(U) columns were pooled, diluted 11.7-fold in phosphate buffer (50 mM phosphate [pH 7.5], protease inhibitors), and loaded onto a 1-ml MonoS HR5/5 column (Pharmacia) preequilibrated with supplemented HNTG buffer containing 60 mM NaCl at a flow rate of 30 ml h⁻¹. After the column was washed, the proteins were eluted at a flow rate of 1 ml min⁻¹ with a 40-ml linear 0.06 to 1 M NaCl gradient. G3BP was found to elute between 150 and 200 mM NaCl. These fractions, corresponding to apparently purified G3BP, were pooled, dialyzed against storage buffer, consisting of 50 mM HEPES (pH 7.5), 30 mM NaCl, and 10% glycerol, and stored at −80°C.

To perform further purification of G3BP under denaturing conditions, purified recombinant G3BP protein (2.6 mg) was denaturated in storage buffer containing 6 M urea and loaded into an S-Sepharose Fast Flow (SFF) column (0.2 cm² by 2.5 cm; HR 5/5; Pharmacia) equilibrated with the same denaturing buffer. The SFF column was washed with 10 times the bed volume at a flow rate of 30 ml h⁻¹ and eluted with a 15-ml linear 0 to 500 mM NaCl gradient. Denatured G3BP was found to elute in two peak fractions, at 280 mM (peak 1) and 395 mM (peak 2). The fractions corresponding to each peak were pooled, adjusted to 150 mM NaCl with buffer E (50 mM HEPES [pH 7.5], 6 M urea, 10% glycerol), and stirred for 5 min at 200 rpm. With continued stirring, buffer H (50 mM HEPES [pH 7.5], 150 mM NaCl, 10% glycerol) was added to each pool in a dropwise fashion at 4°C in order to renature G3BP by diluting the urea by 10-fold. After the mixture was stirred for 1 h, the aggregated proteins were pelleted by centrifugation at 10,000 × g for 10 min at 4°C. Each pool was adjusted to 100 mM NaCl and concentrated over a heparin-Sepharose Fast Flow gel by a batch procedure. G3BP was eluted from the heparin column at 1 M NaCl in 50 mM HEPES (pH 7.5)–10% glycerol, dialyzed against storage buffer, and stored at −80°C. The same procedure was applied to eluted fractions not containing G3BP, which were used as a control.

For metabolic labeling (see Fig. 6), 2.5 mCi of ³²P_i was used per 175-mm dish of monolayer cultures of Sf9 cells infected for 1 day with recombinant baculovirus expressing G3BP. Labeling was carried out with 25 ml of SF-900 serum-free insect cell culture media (GIBCO BRL) for 24 h at 27°C. Recombinant G3BP was immunopurified from infected cell lysates as described above.

FIG. 6.

Dephosphorylation of recombinant G3BP inhibits its ability to cleave the c-myc 3′-UTR RNA. (A) Comparative tryptic phosphopeptide maps of recombinant G3BP and G3BP in quiescent cells. ³²P-labeled recombinant G3BP (panel I) and G3BP from quiescent cells (panel II) were (in the gel), and the resulting phosphopeptides were separated in the first dimension by electrophoresis (pH 1.9) and in the second dimension chromatography on thin-layer plates and detected by autoradiography. Panel I+II represents a mixture of the samples in panels I and II. ori, origin. (B) RNase assays performed with 2 μl of baculovirus control fraction (Baculo CTL), 6 pmol of purified recombinant G3BP, or 20 U of sequencing-grade RNase T₁ incubated in the absence (lanes 1, 3, and 5, respectively) or the presence (lanes 2, 4, and 6, respectively) of calf intestinal alkaline phosphatase (CIAP) for 1 h at 37°C.

In vitro RNase assays.

The c-myc 3′-UTR RNA fragment labeled with [α-³²P]UTP was prepared by in vitro transcription of plasmid pkSGMW1 (42) linearized at BglII with T3 RNA polymerase (Promega) according to the manufacturer’s protocols. Full-length transcripts were purified on denaturing polyacrylamide gels prior to RNase assays. RNase assays were performed with 10 fmol (10,000 cpm) of radiolabeled RNA. The RNase reaction buffer contained 50 mM Tris-HCl (pH 6), 150 mM NaCl, and 10% glycerol. Purified recombinant G3BP (0.2 to 8 pmol) or dephosphorylated purified recombinant G3BP (8 pmol) (as described below) and 10 μg of yeast tRNA were added to 10 μl of reaction mixture prior to the labeled probe. The samples were incubated for 10 min at 25°C and extracted with phenol-chloroform; following ethanol precipitation, the cleavage products were resolved on an 8% polyacrylamide–8 M urea gel. For the experiment shown in Fig. 5B, the samples were directly analyzed on a 4% (acrylamide/bisacrylamide ratio, 80:1) polyacrylamide gel in 50 mM Tris (pH 8.8)–50 mM glycine (21). The gel was electrophoresed in the same buffer at 120 V for 3 h, dried, and autoradiographed.

FIG. 5.

Dose-dependent RNase activity associated with recombinant G3BP after purification under denaturing conditions. Lane 1, 2 μl of baculovirus control fraction (Baculo CTL). The labeled c-myc 3′-UTR was incubated with increasing amounts of purified recombinant G3BP (lane 2, 1.6 pmol; lane 3, 3.2 pmol; and lane 4, 6 pmol of recombinant G3BP). Aliquots of purified recombinant G3BP that were subjected to denaturation in the presence of 6 M urea, renatured, and concentrated over a heparin column were tested for RNase activity (lane 5, 1.6 pmol; lane 6, 3.2 pmol; and lane 7, 6 pmol of G3BP after purification). Denatured purified G3BP was separated over an SFF column into two peaks, and proteins from each peak were renatured, concentrated over a heparin column, and tested for RNase activity (peak 1: lane 8, 1.6 pmol; lane 9, 3.2 pmol; and lane 10, 6 pmol) (peak 2: lane 11, 1.6 pmol; lane 12, 3.2 pmol; and lane 13, 6 pmol). A control fraction (CTL) was obtained by pooling all other fractions not containing G3BP and assessing the RNase activity (lanes 14 to 16). nt, nucleotides; Baculo G3BP, control or recombinant baculovirus G3BP.

To assay for RNase activity in the immunoprecipitates, 50 μg of the anti-G3BP MAb 1F1 was coupled to 100 μl of a 50% suspension of protein G-Sepharose beads in buffer H by incubation for 2 h at 4°C. The beads were washed three times with buffer H and 10 times with PBS. Total extracts (500 μl) from quiescent, dividing fibroblasts or insect cells expressing recombinant G3BP were precleared by incubation under rotation with 50 μl of protein G-Sepharose beads for 1 h at 4°C. The unbound fractions were incubated with the corresponding protein G-Sepharose-coupled serum for 12 h at 4°C. The beads were washed three times with buffer H and 10 times with PBS. One fifth of the immune complexes were analyzed by Western blotting to estimate the G3BP concentration. RNase assays were performed with 5 and 25 μl of each immunoprecipitate in 50 μl of RNase reaction buffer containing 10 μg of yeast tRNA and 10 fmol of radiolabeled c-myc 3′-UTR. The samples were incubated for 10 min at 25°C and extracted with phenol-chloroform; following ethanol precipitation, the cleavage products were resolved on an 8% polyacrylamide–8 M urea gel and visualized by autoradiography. The same experiment was performed with extracts previously treated with alkaline phosphatase as described below.

Detection of isoelectric variants of G3BP.

In order to dephosphorylate G3BP, total extracts prepared as described previously (32) or recombinant G3BP was treated with 10 U of calf intestinal alkaline phosphatase (GIBCO BRL) per μg of protein for 1 h at 37°C. The protein samples (20 μg of total extracts or 5 μg of purified recombinant G3BP), incubated with or without phosphatase, were precipitated with 5 volumes of cold acetone, resuspended in 50 μl of loading buffer, consisting of 9.5 M urea, 5% Bio-lyte (pH 3 to 10), 4% 3-[(3-cholamidopropyl)-dimethylammonio]-1-propanesulfonate (CHAPS), 10% glycerol, and 50 mM DTT, and analyzed on vertical slab isoelectric focusing (IEF) gels essentially as described previously (37). Briefly, we used vertical slab gels of 1.5 mm (Bio-Rad) containing 9.2 M urea, 4.5% acrylamide (acrylamide/piperazine diacrylamide ratio, 28.38:1.68), 5% CHAPS, 1% Bio-lyte (pH 3 to 10), 2% Bio-lyte (pH 5 to 7), 2% Bio-lyte (pH 5 to 8), and 0.025% ammonium persulfate. Samples were loaded into dry wells and overlaid with overlay solution (8 M urea, 0.1% Nonidet P-40, 2.5% Bio-lyte [pH 3 to 10]). The gels were run at 10 mA per gel, with an 800-V limit, for 17 h with 50 mM NaOH at the cathode (top) and 10 mM orthophosphoric acid at the anode (bottom). After electrophoresis, the gels were soaked in transfer buffer (12.5 mM Tris, 26 mM glycine, 0.1% SDS, 25% methanol), and proteins were transferred to nitrocellulose membranes for at least 3 h and revealed by immunoblot analysis.

The pH gradient was determined by slicing a slot of IEF gel; 5-mm-long slices were soaked for 1 h in 500 μl of H₂O. The pH of the supernatant was measured at room temperature. G3BP was found to have a measured pI of approximately 6.5.

RESULTS

RasGAP-G3BP interaction occurs in dividing cells but not in quiescent cells.

We previously reported that immunoprecipitation of G3BP with a purified anti-GAP MAb (15F8) was critically dependent on cell growth conditions (32). To further characterize the RasGAP-G3BP interaction, we monitored the recovery of RasGAP by an anti-G3BP MAb at different times after stimulation of quiescent CCL39 fibroblasts. Probing of anti-G3BP immunoprecipitates by an anti-RasGAP MAb revealed that RasGAP recovery was maximal at 1 and 8 h of serum addition but was not observed between these times and persisted after 8 h for at least 16 h (Fig. 1A, upper panel). However, the same amounts of G3BP were immunoprecipitated at any time of serum stimulation (Fig. 1A, lower panel), indicating that the level of the RasGAP-G3BP complex oscillated in the cell cycle. The complex observed after 1 h of serum stimulation was short lived and may have coincided with the entry of most quiescent cells into the G₁ phase. In contrast, the complex observed after 8 h of stimulation was very stable and was formed when more than 40% of CCL39 fibroblasts had entered the S phase. Thus, G3BP appeared to be associated with RasGAP in exponentially growing cells but not in quiescent cells.

FIG. 1.

Time course of anti-G3BP antiserum recovery of RasGAP after serum stimulation of quiescent cells showing a physical interaction between RasGAP and G3BP. (A) Anti-RasGAP (upper panel) and anti-G3BP (lower panel) immunoblot analysis of G3BP immunoprecipitates (IP) from cells either unstimulated (lane 0) or stimulated with 10% fetal calf serum for the indicated times. (B) Total extracts obtained from quiescent (lane 2) or dividing (lane 3) cells were treated with EDC, and cross-linked proteins were analyzed by immunoblotting with either anti-G3BP (α G3BP) or anti-RasGAP (α RasGAP) antibody. Lane 1 represents total extracts not treated with EDC. −FCS, without fetal calf serum; +FCS, with fetal calf serum; CTL, control fraction.

Since the coimmunoprecipitation of RasGAP with an anti-G3BP antibody could be due to direct or indirect association with G3BP, we further analyzed complexes after treatment with the chemical cross-linker EDC. This reagent preferentially cross-links biomolecules which interact through salt bridges between carboxylate and ammonium groups and is well suited for studies of interactions between proteins that form electrostatically stabilized complexes (11). Cross-linked proteins were fractionated by SDS-PAGE, blotted, and probed with either anti-G3BP or anti-RasGAP antibodies (Fig. 1B). In addition to G3BP and RasGAP, which were revealed with the corresponding antibodies, both antibodies detected a 180,000-molecular-weight (180K) band in the presence of EDC (Fig. 1B, lanes 3) but not in the absence of EDC (lanes 1). Furthermore, the 180K band was not observed following EDC treatment of extracts derived from quiescent cells (Fig. 1B, lanes 2). Also, the 180K band was the only complex which was recognized by both antibodies, even though a crude total extract from exponentially growing cells was used. As a complex containing both RasGAP and G3BP would migrate at approximately 180K to 190K, it is likely that the 180K band represents this complex. Indeed, an anti-RasGAP antibody was able to immunoprecipitate the 180K species which was recognized by an anti-G3BP antibody and, conversely, an anti-G3BP antibody was able to immunoprecipitate a complex of the same size which was recognized by an anti-RasGAP antibody (data not shown). Thus, G3BP and RasGAP are physically associated in dividing cells so as to form a 180K EDC-cross-linked complex.

The RasGAP-G3BP complex is associated with the membrane fraction.

Immunolocalization experiments established that RasGAP and G3BP are cytosolic. However, the RasGAP-G3BP complex was not detected in the S100 fraction derived from exponentially growing cells (data not shown). Since the S100 fraction was obtained following ultracentrifugation of the cell lysate at 100,000 × g, it remained likely that the RasGAP-G3BP complex was associated with the particulate membrane fraction (P100) and was depleted from the S100 fraction. We therefore analyzed the distribution of RasGAP and G3BP between the S100 and P100 fractions by Western blotting (Fig. 2) and determined the percentage of total cellular RasGAP and G3BP found in each fraction. In exponentially growing CCL39 cells, 40% of total RasGAP and 30% of total G3BP were contained in the P100 fraction (Fig. 2, α G3BP and α RasGAP, compare lanes 3 and 4). In contrast, following serum deprivation, both proteins were detected only in the S100 fraction (Fig. 2, α G3BP and α RasGAP, compare lanes 1 and 2). These results suggested that upon serum stimulation, RasGAP and G3BP associated with the P100 fraction containing the plasma membrane.

FIG. 2.

G3BP associates with RasGAP exclusively in subcellular particulate fractions derived from dividing cells and not quiescent cells. Proteins contained in the S100 and the P100 fractions derived from quiescent cells (lanes 1 and 2, respectively) or dividing cells (lanes 3 and 4, respectively) were immunoblotted with anti-G3BP (α G3BP), anti-RasGAP (α RasGAP), anti-Sos (α Sos), anti-Grb2 (α Grb2), anti-p21^ras (α Ras), anti-EGFR (α EGFR), and anti-GAPDH (α GAPDH) antibodies. −FCS, without fetal calf serum; +FCS, with fetal calf serum.

Since previous studies showed that the association of RasGAP with the membrane is required to potentiate its negative regulatory activity on p21^ras (28, 30), we also assessed the subcellular distribution of two regulators of the Ras pathway; Sos, a factor required for changing Ras from the GDP-bound to the GTP-bound state, and Grb2, an adaptor protein that connects Sos to the receptor. As expected, Ras itself (24) and the EGFR were entirely associated with the membrane fraction (Fig. 2, α Ras and α EGFR, compare lanes 1 and 2), and this localization did not depend on cell proliferation (compare lanes 3 and 4). Significantly, both Sos and Grb2 demonstrated the same distribution as RasGAP and G3BP (Fig. 2, compare α Sos and α Grb2 to α G3BP and α RasGAP); they were associated with the P100 fraction in exponentially growing cells (lane 4) but not in quiescent cells (lane 2). In contrast, growth conditions did not influence the localization of GAPDH (Fig. 2, α GAPDH, compare lanes 1 and 2 to lanes 3 and 4), a known cytosolic protein. These results demonstrated that G3BP was recruited to the plasma membrane in proliferating cells, presumably through its association with RasGAP.

Hyperphosphorylation of G3BP on serine residues upon serum starvation.

Tyrosine phosphorylation of RasGAP and RasGAP-associated proteins p62 and p190 accompanies stimulation of cells with growth factors or transformation of cells with activated tyrosine kinases (10). We therefore assessed whether cell stimulation would modify the phosphorylation status of G3BP and/or affect its interactions with RasGAP. For this purpose, we examined the phosphorylation status of G3BP in quiescent and serum-stimulated CCL39 cells. Cells were metabolically labeled in vivo with ³²P_i before G3BP was isolated by immunoprecipitation. The bound proteins were separated by SDS-PAGE and visualized by autoradiography (Fig. 3A). Phosphoamino acid determination of the labeled amino acid residues demonstrated that G3BP was extensively phosphorylated on serine residues in quiescent and serum-stimulated cells, but trace amounts of Pthr were also detected (Fig. 3B). A comparison of amounts of ³²P-labeled Pser derived from equivalent amounts of immunopurified G3BP from quiescent and dividing cells revealed that G3BP labeling was threefold higher in quiescent cells than in dividing cells (Fig. 3B), indicating that the protein becomes more phosphorylated in quiescent cells.

To further confirm the hyperphosphorylation of G3BP in quiescent fibroblasts, total extracts were analyzed on a one-dimensional IEF gel, which can resolve G3BP derivatives that differ in their degree of phosphorylation. The electrophoretic patterns revealed that there were three isoelectric variants of G3BP in proliferating cells (Fig. 3C, panel I, lane 1) but only two in quiescent cells (Fig. 3C, panel I, lane 2). Fractionation of the isoelectric variants in the second dimension by SDS-PAGE confirmed that all variants had the mobility of G3BP (data not shown). Phosphorylation appeared to be the major modification responsible for G3BP electrophoretic behavior, as all the G3BP variants were sensitive to treatment with alkaline phosphatase (Fig. 3C, panels II and III, compare lanes 2 and 3); this treatment resulted in the occurrence of two prominent bands with decreased mobility (lane 3). The faster migrating of the two bands was likely to correspond to a phosphorylation variant of G3BP, since 20% of immunopurified labeled G3BP was resistant to dephosphorylation with a large excess of alkaline phosphatase (data not shown). A comparison of the G3BP IEF patterns in quiescent and dividing cells revealed that G3BP was more phosphorylated in the former (Fig. 3C, panel I, compare lanes 1 and 2).

Phosphotryptic peptide mapping analysis (2) was used to compare the fingerprints of labeled immunopurified G3BP derived from quiescent and dividing cells which were metabolically labeled for 8 h with ³²P_i. The results showed that there was no discernible difference in the patterns of phosphorylation of G3BP from quiescent and dividing cells (Fig. 3D), implying that the hyperphosphorylation of G3BP did not involve changes in the phosphorylation sites. However, phosphopeptides a and c were more extensively labeled in quiescent cells than in dividing cells, indicating that phosphorylation at those sites was responsible for G3BP hyperphosphorylation.

Recombinant G3BP exhibits endoribonuclease activity.

G3BP contains sequence motifs found in RNA binding proteins and binds RNA homopolymers (data not shown). To examine isolated effects of G3BP-RNA interactions, a baculovirus vector that expresses the recombinant G3BP protein at levels corresponding to 10% of the total soluble protein of infected insect cells was used to obtain large amounts of purified G3BP. G3BP was purified to apparent homogeneity (Fig. 4A) from infected cells by sequential chromatography on heparin-Sepharose, agarose-poly(rU), and ion-exchange columns. During purification, G3BP was tracked by monitoring RasGAP SH3-binding activity and by Western blotting with a G3BP-specific antibody. Reactivity with an anti-G3BP antibody and interaction with RasGAP SH3 were coincident, implying that recombinant G3BP could still bind RasGAP (data not shown).

FIG. 4.

Cleavage of the c-myc 3′-UTR by purified recombinant G3BP. (A) SDS-PAGE analysis of purified recombinant G3BP (R-G3BP) (2 μg) stained with Coomassie blue (lane G3BP). Lane M, molecular size markers (in thousands). (B) The homogeneously ³²P-labeled c-myc 3′-UTR was incubated for 10 min at 25°C in 10 μl of 50 mM Tris-HCl (pH 6)–150 mM NaCl–10% glycerol–10 mg of yeast tRNA per ml with 2 μl of buffer (lane 1), 2 μl of baculovirus control fraction (Baculo-CTL) (lane 2), 8 pmol of purified G3BP previously incubated at 100°C for 10 min (lane 3), 8 pmol of purified G3BP digested with proteinase K (PK) (lane 4), or 8 pmol of purified recombinant G3BP (lane 5). (Panel I) Aliquots (6 μl) from each reaction mixture were directly resolved on a nondenaturing polyacrylamide gel as described in Materials and Methods. Labeled RNA was visualized by autoradiography. The asterisk corresponds to abnormal migration of the labeled c-myc 3′-UTR in the native gel. (Panel II) Alternatively, the reaction mixtures were quenched by one phenol-chloroform extraction, and the RNA was electrophoresed on an 8% polyacrylamide–8 M urea gel. The asterisk indicates labeled c-myc.

Given the strong homology between G3BP and the hnRNP C RRM domains (32) and the preferential binding of G3BP to poly(rU), a G3BP target sequence was expected to be U rich (data not shown). Since G3BP is localized in the cytoplasm, we next examined the ability of a recombinant protein to bind RNA sequences known to regulate either mRNA turnover or mRNA translation or both. The 3′-UTR of the c-myc mRNA was used in initial studies because this region, previously shown to be involved in the regulation of c-myc mRNA turnover (18), contains many poly(U) tracts embedded in a U-rich sequence. A radiolabeled 145-nucleotide RNA probe containing the sequence spanning positions 2094 to 2195 of mouse c-myc mRNA was mixed with purified recombinant G3BP in the presence of an excess of competing tRNA, and the mixture was then loaded onto a native 4% polyacrylamide gel in order to resolve the G3BP-RNA complex from free RNA. The free RNA showed a complex migration profile, with two bands, suggesting the presence of several intramolecular isomers (Fig. 4B, panel I, lane 1). Interestingly, following a 10-min incubation with G3BP at 25°C, the majority of the RNA migrated as small fragments below the free RNA (Fig. 4B, panel I, lane 5). Analysis of the same samples on a denaturing polyacrylamide-urea gel confirmed that incubation of c-myc RNA with recombinant G3BP resulted in its specific degradation (Fig. 4B, panel II, lane 5). Cleavage was not observed when the incubation was performed at 0°C or in the absence of recombinant G3BP (Fig. 4B, lanes 1). Both heat denaturation (Fig. 4B, lanes 3) and proteinase K treatment (lanes 4) of recombinant G3BP prevented c-myc RNA degradation, demonstrating the protein origin of the RNase activity.

A more detailed characterization of the RNase activity in a standardized in vitro degradation assay, which will be presented elsewhere (11a), revealed that the enzyme generated discrete c-myc cleavage fragments in a dose-dependent manner (Fig. 5). A characteristic initial cleavage product was detected as a doublet of approximately 130 nucleotides at a low protein concentration (Fig. 5, lane 2), while higher concentrations of enzyme (lane 4) resulted in the appearance of cleavage fragments of 65, 45, and 12 nucleotides. These fragments were not cleaved further following the addition of recombinant G3BP, indicating that they were the likely end products of the nuclease reaction (data not shown). In contrast, under the same conditions, neither poly(rU), poly(rG), poly(rC), nor poly(rA) homopolymers were degraded by the RNase (data not shown), suggesting that a relatively specific sequence and/or structure was required for this RNase activity.

To test whether the protein responsible for this activity had the same purification profile as recombinant G3BP, we derived a control fraction from insect cells infected with a baculovirus vector carrying Grb2 and found that this fraction did not have detectable RNase activity (Fig. 5, lane 1). This finding indicated either that G3BP was itself an RNase or that the RNase was tightly associated with recombinant G3BP. To distinguish between these two possibilities, purified recombinant G3BP was subjected to additional chromatographic purification under denaturing conditions. Recombinant G3BP was denatured in 6 M urea in order to disrupt any hydrophobic interactions with contaminating proteins, loaded onto an SFF column, and eluted with a linear salt concentration in the presence of 6 M urea. This process resulted in the separation of G3BP into two peaks, which were concentrated on a heparin column. Renatured G3BP from these two peaks was tested for RNase activity (Fig. 5, lanes 8 to 13); additionally, other SFF fractions were pooled and tested in order to assess whether there was a contaminating RNase activity (lanes 14 to 16). Indeed, RNase activity was purified with both G3BP peaks, albeit at a higher activity with peak 2 (Fig. 5, lanes 11 to 13) than with peak 1 (lanes 8 to 10), and none was observed in the control fraction (lanes 14 to 16). Additional cleavage products were seen with either total renatured G3BP before fractionation with the SFF column (Fig. 5, lanes 5 to 7) or G3BP in peak 2 (lanes 11 to 13), whereas the pattern of c-myc RNA cleavage obtained with peak 1 was indistinguishable from that observed in the presence of G3BP before denaturation and renaturation (compare lanes 2 to 4 with lanes 8 to 10).

Although the reason for the separation of G3BP into two peaks is not yet clear, the differences in substrate preference indicate that the RNase activity could be modulated by changes in the conformation of G3BP. The results presented below suggest that these changes may occur normally due to the phosphorylation of the protein. Given that the specific activities before and after purification were identical (a comparison was made between total renatured G3BP and the sum of peak 1 and peak 2), these results clearly identified G3BP as an RNase. Under the conditions of the RNase assay, G3BP was twofold more active than RNase T₁, a well-known endonuclease (see Fig. 6B).

Regulation of G3BP RNase activity by phosphorylation.

As the activity of recombinant G3BP was higher in peak 2 than in peak 1, we next assessed the phosphorylation status of G3BP in the two fractions. Probing of Western blots of recombinant G3BP with anti-Pser, anti-Ptyr, and anti-Pthr antibodies revealed that recombinant G3BP in both peaks, like cellular G3BP, was phosphorylated extensively on serine residues (data not shown). We also compared the phosphorylation status of G3BP expressed in insect cells to that of G3BP in quiescent mammalian cells. Proteins were metabolically labeled with ³²P_i, immunopurified with an anti-G3BP antibody, and subjected to tryptic phosphopeptide mapping. Four labeled phosphopeptides characterized the trypsin digestion pattern of recombinant G3BP (Fig. 6A, panel I). Three of them were identical to those forming the trypsin digestion pattern of labeled immunopurified G3BP from quiescent cells (Fig. 6A), but one phosphopeptide (d) was different. An additional difference between the two patterns was the level of phosphopeptide c phosphorylation (Fig. 6A, compare panels I and II). This phosphopeptide was more phosphorylated in quiescent mammalian cells than in insect cells. These results could be easily explained if one or more kinase(s) responsible for the phosphorylation of G3BP in insect cells became limiting because of G3BP overexpression (see Discussion).

To assess the effects of phosphorylation on G3BP RNase activity, we evaluated this activity upon phosphatase treatment. As shown in Fig. 6B, degradation of c-myc RNA by alkaline phosphatase-treated recombinant G3BP was marginal (lane 4). Background cleavage of c-myc RNA was negligible compared with cleavage with recombinant G3BP that was not treated with alkaline phosphatase (Fig. 6B, compare lanes 3 and 4). The trivial explanation that the alkaline phosphatase preparation contained a component that inhibited RNase activity was eliminated, because heat-denatured phosphatase did not prevent G3BP-dependent degradation of c-myc RNA (data not shown). Inhibition also was not due to direct interaction of the phosphatase with c-myc RNA, which would make RNA sites on the myc sequence unavailable to nucleases, since cleavage of c-myc RNA with RNase T₁ was not altered in the presence of alkaline phosphatase (Fig. 6B, lane 6). Thus, phosphorylation appeared to have major effects on G3BP RNase activity.

Phosphorylation-dependent G3BP RNase activity in mammalian cells.

To investigate whether G3BP from mammalian cells had RNase activity, immunoprecipitates from quiescent cells, dividing cells, and insect cells expressing G3BP were tested for RNase activity. Aliquots of the immunoprecipitates were incubated with labeled c-myc 3′-UTR, and the products of the reaction were resolved on a polyacrylamide gel and visualized by autoradiography (Fig. 7A). The patterns of c-myc RNA cleavage obtained with immunoprecipitates from quiescent (Fig. 7A, lanes 12 and 13) and insect (lanes 8 and 9) cells were indistinguishable from that observed in the presence of purified recombinant G3BP (lanes 14 to 16); no cleavage occurred with the corresponding amount of immunoprecipitates of protein G without antibody which were previously incubated with the same extracts (lanes 2 to 4). However, a low level of RNase activity was associated with immunoprecipitates from dividing cells (Fig. 7A, lanes 10 and 11) compared to immunoprecipitates from quiescent cells (lanes 12 and 13).

FIG. 7.

RNase activity of G3BP in mammalian cells. (A) RNase assays were performed as described in Materials and Methods with 5 or 25 μl of G3BP immunoprecipitates (IP) of extracts obtained from insect cells infected with baculovirus expressing G3BP (recombinant G3BP [R-G3BP]) (lanes 8 and 9), from dividing cells (with fetal calf serum [+FCS]) (lanes 10 and 11), or from quiescent cells (without fetal calf serum [−FCS]) (lanes 12 and 13) or with 25 μl of G3BP immunoprecipitates of calf intestinal alkaline phosphatase (CIAP)-treated extracts obtained from insect cells infected with baculovirus expressing G3BP (lane 5), from dividing cells (lane 6), or from quiescent cells (lane 7). Lane 1, RNA alone (CTL, control fraction). As controls, RNase assays were performed with 25 μl of protein G-Sepharose (without antibody) incubated with extracts obtained from insect cells infected with baculovirus expressing G3BP (lane 2), from dividing cells (lane 3), or from quiescent cells (lane 4). α-G3BP, antibody to G3BP. (B) Immunoblot analysis of G3BP in immunoprecipitates (IP) of extracts obtained from insect cells infected with baculovirus expressing G3BP (Baculo CTL), from dividing cells (+FCS), or from quiescent cells (−FCS). The asterisk indicates labeled RNA. Lanes 14 to 16, RNase assays with recombinant G3BP.

Western blot analysis showed that these immunoprecipitates contained the same amounts of G3BP (Fig. 7B, compare lanes +FCS and −FCS), suggesting that the association with RasGAP or the hypophosphorylation of G3BP led to a decrease in the RNase activity. This analysis also revealed that immunoprecipitates from insect cells contained at least 10 times as much G3BP as immunoprecipitates from quiescent cells (Fig. 7B, compare lanes 1 and 2) and therefore had the lowest specific activity. Given that recombinant G3BP was not associated with RasGAP and that the SH3 domain of RasGAP did not affect the RNase activity in vitro (data not shown), it seems unlikely that RasGAP could be responsible for the low level of RNase activity of G3BP from dividing cells. However, the quantitative and qualitative differences in the tryptic phosphopeptide maps, together with differences in the RNase specific activities between immunoprecipitates from quiescent cells and immunoprecipitates from insect cells, strengthened the relationship between G3BP phosphorylation status and G3BP RNase activity. In agreement with this idea, immunoprecipitates from extracts that were pretreated with alkaline phosphatase had no RNase activity (Fig. 7B, lanes 5 to 7).

DISCUSSION

We have demonstrated that G3BP, a protein which is localized in the cytoplasm and which shares characteristic structural features with RNA binding proteins, is a phosphorylation-dependent RNase. Although it exhibits strong RNase activity, the primary amino acid sequence of G3BP does not show significant overall similarity to any known nuclease. Therefore, G3BP could represent a new class or subclass of RNases which are likely to play a major role in cell proliferation. It specifically interacts with RasGAP in dividing but not quiescent cells. The absence of an interaction in quiescent cells was further linked to the accumulation of a soluble hyperphosphorylated form of G3BP exhibiting enhanced RNase activity. Since it is vital to rapidly degrade specific mRNAs that encode critical regulatory proteins that function only briefly during cell proliferation, the phosphorylation-dependent RNase activity of G3BP could be involved in mRNA degradation. A central focus of many cell cycle-regulated mRNA turnover pathways is the initiation of degradation by endonucleolytic cleavage, usually in the 3′-UTR of target transcripts. G3BP cleaves the 3′-UTR of c-myc in vitro and therefore might initiate mRNA turnover.

G3BP localization and relationship to other signaling proteins.

Parker et al. (32) first reported that RasGAP associates with G3BP in cells that were mitogenitically stimulated or transformed by activated Ras. By using the covalent cross-linking reagent EDC, we showed that, in vivo, the RasGAP-G3BP association is a result of a direct interaction between the two proteins in dividing and not quiescent cells (Fig. 1B). The predominance of the 180K species as the major cross-linked complex containing both G3BP and RasGAP argued in favor of a stable interaction between the two proteins rather than the formation of complexes by random collisions of unstably interacting proteins. We found that the complex between RasGAP and G3BP was localized exclusively in subcellular particulate fractions of dividing cells, whereas free G3BP or RasGAP was primarily cytosolic. Still open to question is whether the RasGAP-G3BP complex is recruited at the plasma membrane through the interaction of RasGAP with a third partner or through the recruitment of RasGAP by one of its plasma membrane localization domains (i.e., PH domain or calcium-dependent lipid binding domain). A complex between RasGAP and tyrosine-phosphorylated p62 is believed to occur in subcellular particulate and cytosolic fractions (7, 30, 45). Since p62 has several phosphorylated tyrosine residues, it has been suggested that p62 acts as a docking protein for RasGAP recruitment at the membrane upon phosphotyrosine kinase activation (7, 45). The fact that we did not detect p62 from G3BP immunocomplexes with an anti-Ptyr antibody (data not shown) argued against a p62-G3BP-RasGAP ternary complex and hence against a role for p62 in recruiting G3BP.

Our results also indicate that serum growth factors might modulate the membrane anchoring of G3BP. Therefore, it is worth noting that RasGAP has a PH domain (residues 474 to 577), a sequence motif that is present in a large variety of proteins involved in intracellular signaling or cytoskeletal organization (22, 31). Recently, the PH domain of Sos was shown to participate in regulating the inducible association of Sos with the membrane (8); similarly, the PH domain of RasGAP might target G3BP to the plasma membrane following serum stimulation. The calcium-dependent lipid binding domain of RasGAP (residues 594 to 676) also appears to be a prevalent protein module that may affect the association of RasGAP with the particulate fraction of cells in response to elevated intracellular Ca²⁺ levels (12). This could be another way in which RasGAP changes the subcellular localization of the RasGAP-G3BP complex.

RasGAP and the regulation of G3BP phosphorylation.

The finding that the level of G3BP phosphorylation changed during the transition from resting to dividing cells suggested that RasGAP could play a role in changing the phosphorylation status of G3BP. The results did not, however, establish how these two events are mechanistically related: Is the dephosphorylation of G3BP required for its interaction with RasGAP, or is the association with RasGAP an essential prerequisite for the dephosphorylation of G3BP? In other words, is the deficiency of G3BP phosphorylation in dividing cells a cause or a consequence of RasGAP binding? Preliminary indications are that the interaction with RasGAP is not determined by the level of G3BP phosphorylation, since the SH3 domain of RasGAP is able to bind G3BP in vitro regardless of its phosphorylation status. Additionally, we did not detect differences in the levels of G3BP phosphorylation between cytosolic free G3BP and G3BP that interacted with RasGAP in the particulate fractions (data not shown). The results at this stage imply that the dephosphorylation of G3BP does not guarantee the formation of the RasGAP-G3BP complex, but cell proliferation ensures that G3BP undergoes dephosphorylation, probably through an interaction with RasGAP. Consistent with this view is the finding that fibroblasts transformed with an activated ras^Val-12 mutant (39), which could not be arrested upon serum starvation and which accumulated RasGAP-G3BP complexes in the presence and absence of serum, had a low level of G3BP phosphorylation (11a). The mechanism by which RasGAP allows the dephosphorylation of G3BP is presently unclear. The interaction with the SH3 domain of RasGAP either could inhibit a kinase that constitutively phosphorylates serine residues of G3BP or could recruit G3BP to a specific phosphatase. Elucidation of this mechanism will be an important step toward understanding the regulation of G3BP function by RasGAP.

Phosphorylation and possible roles of G3BP RNase.

G3BP is a stable protein whose expression does not vary during the cell cycle (Fig. 1A), but as demonstrated here, it is posttranscriptionally modified by phosphorylation. This modification concerns serine residues, which appear to carry the information required to modulate, at least in vitro, the RNase activity of G3BP. Given that the dephosphorylation of G3BP by alkaline phosphatase abolishes its RNase activity, it is possible that recombinant G3BP, which had the lowest RNase activity, lacked an important phosphorylation site(s). Consistent with this view, recombinant G3BP had a pattern of phosphorylation different from that observed for the highly phosphorylated G3BP isolated from quiescent CCL39 fibroblasts. Tryptic phosphopeptide mapping of recombinant G3BP revealed an additional phosphopeptide, designed d, which could be either a new site that was phosphorylated only in the insect cells or a phosphoisomer that was generated by multiple phosphorylations of a single tryptic peptide containing more than one phosphoacceptor site. Although further experiments will be required to distinguish between these two possibilities, the observation that phosphopeptides b and d lay on a diagonal line sloping toward the anode was consistent with these two phosphopeptides being phosphoisomers (2). In accordance with this hypothesis, phosphopeptide d had one phosphate less than phosphopeptide b, implying that recombinant G3BP lacked one phosphorylation site compared to G3BP in mammalian cells. Under phosphorylation of recombinant G3BP was also demonstrated by the finding that phosphopeptide c was less phosphorylated in insect cells than in mammalian cells.

The determination and localization of cellular RNA(s) with which G3BP interacted are elusive tasks due to the RNase activity of the protein. A comparison of G3BP with other RNA binding proteins revealed strong homology with the RRM domain of hnRNP C, which is valuable information concerning the RNA sequences that might interact with G3BP. Interestingly, hnRNP C is known to avidly bind to poly(U) stretches (13). The strategy of using the 3′-UTR of c-myc mRNA, which harbors a U-rich sequence, was therefore important for observing significant sequence specificity. The ability of G3BP to introduce cleavage at several sites in the c-myc 3′-UTR, which is known to mediate rapid mRNA decay (15), suggests that G3BP may be a component of an mRNA degradation system that controls normal cell growth and differentiation. Indeed, the c-myc proto-oncogene is rapidly and transiently activated by extracellular stimuli, and the rapid disappearance of this message is due not only to a shutoff of transcription but also to a short half-life (16).

The observation that phosphorylation is required to significantly influence the RNase activity of G3BP, together with the fact that the phosphorylation of G3BP is affected by extracellular stimuli, is compatible with the possibility that G3BP plays a role in the c-myc mRNA short half-life. It is striking that the transient increase in the steady-state levels of c-myc mRNA observed after serum stimulation of quiescent CCL39 fibroblasts (1) parallels the formation of a transient RasGAP-G3BP complex (Fig. 1A). It can be speculated that the formation of this complex leads to a decrease in RNase activity (through the dephosphorylation of G3BP), thereby allowing the accumulation of c-myc mRNA. Conversely, excess active RasGAP-free G3BP would lead to c-myc mRNA degradation. Supporting this view is our finding that the overexpression of G3BP reduced the steady-state levels of c-myc mRNA (data not shown).

Another possibility, which is not mutually exclusive with the latter, is that RNase activity is not very specific and should not be considered to degrade specific mRNAs such as c-myc but rather to attack almost any exposed U-rich motif. In this context, G3BP may act as a growth factor sensor. For instance, when growth factors are withdrawn from the cells, G3BP RNase is activated and degrades mRNAs whose expression is required for cell cycle progression. Conversely, the stimulation of quiescent cells with mitogen induces the expression of several short-lived mRNAs and allows the transient formation of a RasGAP-G3BP complex to transiently inhibit RNase activity. The finding that the RasGAP-G3BP complex was reformed after quiescent cells entered the S phase may also suggest that the RNase activity of G3BP plays a role during progression through G₁. The growing family of cyclin genes and their products which have been identified as important regulatory participants in the eukaryotic cell cycle could also be potential targets of G3BP RNase activity.