This Article Abstract Full Text (PDF) Other Versions of this Article: MCB.02300-06v1 27/6/2324    most recent Alert me when this article is cited Alert me if a correction is posted Services Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Reprints and Permissions Copyright Information Books from ASM Press MicrobeWorld Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Solomon, S. Articles by Schrader, J. W. Search for Related Content PubMed PubMed Citation Articles by Solomon, S. Articles by Schrader, J. W.

Distinct Structural Features ofCaprin-1 Mediate Its Interaction with G3BP-1 and Its Induction of Phosphorylation of Eukaryotic Translation Initiation Factor 2, Entry to Cytoplasmic Stress Granules, and Selective Interaction with a Subset of mRNAs

The Biomedical Research Centre, University of British Columbia, Vancouver, Canada,¹ Eskitis Institute for Cell and Molecular Therapies, Nathan Campus, Queensland, Australia²

Received 8 December 2006/ Accepted 13 December 2006

	ABSTRACT

Top Abstract Introduction Materials and Methods Results Discussion References

 
Caprin-1 is a ubiquitously expressed, well-conserved cytoplasmic phosphoprotein that is needed for normal progression through the G₁-S phase of the cell cycle and occurs in postsynaptic granules in dendrites of neurons. We demonstrate that Caprin-1 colocalizes with RasGAP SH3 domain binding protein-1 (G3BP-1) in cytoplasmic RNA granules associated with microtubules and concentrated in the leading and trailing edge of migrating cells. Caprin-1 exhibits a highly conserved motif, F(M/I/L)Q(D/E)Sx(I/L)D that binds to the NTF-2-like domain of G3BP-1. The carboxy-terminal region of Caprin-1 selectively bound mRNA for c-Myc or cyclin D2, this binding being diminished by mutation of the three RGG motifs and abolished by deletion of the RGG-rich region. Overexpression of Caprin-1 induced phosphorylation of eukaryotic translation initiation factor 2 (eIF-2) through a mechanism that depended on its ability to bind mRNA, resulting in global inhibition of protein synthesis. However, cells lacking Caprin-1 exhibited no changes in global rates of protein synthesis, suggesting that physiologically, the effects of Caprin-1 on translation were limited to restricted subsets of mRNAs. Overexpression of Caprin-1 induced the formation of cytoplasmic stress granules (SG). Its ability to bind RNA was required to induce SG formation but not necessarily its ability to enter SG. The ability of Caprin-1 or G3BP-1 to induce SG formation or enter them did not depend on their association with each other. The Caprin-1/G3BP-1 complex is likely to regulate the transport and translation of mRNAs of proteins involved with synaptic plasticity in neurons and cellular proliferation and migration in multiple cell types.

	INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

 
Caprin-1 is a ubiquitously expressed, well-conserved cytoplasmic phosphoprotein (13). Its levels increase when resting cells enter the cell cycle and decrease when proliferation ceases and cells differentiate (13). In most tissues, levels of Caprin-1 correlate with the frequency of proliferating cells, although it also occurs at high levels in the adult brain (13, 40). Gene-targeting experiments showed that cells lacking Caprin-1 exhibit delays in transition from the G₁ to the S phase of the cell cycle (53). Although Caprin-1 was initially mischaracterized as a p137 glycosylphosphatidylinositol-anchored membrane protein (10), it does not occur on the plasma membrane and the current names for the locus, M11S1 (for membrane component, chromosome 11, surface marker 1) or Gpiap1 (glycosylphosphatidylinositol-anchored membrane protein 1), are misleading. Well-conserved orthologs of vertebrate Caprin-1 are present in the urochordate Ciona intestinalis (13) and the enchinoderm Strongylocentrotus purpurata (unpublished data). Caprin-1 shares two novel protein domains, homologous region-1 (HR-1) and HR-2, with a highly conserved paralog termed Caprin-2. Caprin-2 is present only in vertebrates and is distinguished from Caprin-1 by an additional carboxy-terminal domain homologous to C1q (13). The HR-2 domains in both Caprin-1 and Caprin-2 exhibit RGG motifs that are typical of RNA-binding proteins, with the carboxy terminus of Caprin-1 having three.

In the brain, Caprin-1 occurs in messenger ribonucleoprotein particles (mRNPs) that also contain RNA binding proteins (RBP) like hnRNPK, PABP-1 and Staufen, - and ß-tubulin, and the motor protein dynein (1, 2). Caprin-1 is present in both mRNPs associated with polysomes and in mRNPs with translationally silent mRNAs (1). In the course of our studies, Shiina et al. reported the characterization of the Xenopus ortholog of Caprin-1, which they termed XRNG-105 (40). Theyshowed that Caprin-1 is expressed in postsynaptic granules in dendrites in the hippocampus and neocortex and that immunoprecipitates of Caprin-1 from lysates of brain cells contained a selected subset of mRNAs that encoded proteins involved in synaptic plasticity (40).

To address the functional relationships of Caprin-1 in actively proliferating cells, we adopted a proteomic strategy and identified a series of binding partners in immunoprecipitates of Caprin-1 (M. D. David, P. Schubert, V. Lam, S. Solomon, J. Kast, and J. W. Schrader, unpublished). The most prominent binding partner was RasGAP SH3-domain binding protein-1 (G3BP-1) (16), an RBP originally characterized as a target of the SH3 domain of p120 RasGAP, the negative regulator and effector of p21 Ras (34). At the amino terminus of G3BP-1 is a nuclear transport factor-2 (NTF-2)-like domain homologous to NTF-2, followed by acidic and proline-rich regions, and an RNA-binding domain with an RNA-recognition motif and multiple RGG motifs. Given our evidence that Caprin-1 is involved in cellular proliferation, we were intrigued that multiple lines of evidence linked G3BP-1 with cell proliferation. Thus, G3BP-1 had been shown to bind to the 3' untranslated region (UTR) of c-Myc mRNA in quiescent cells and degrade it through its endonuclease activity, with this effect abrograted following localization of p120 RasGAP to the plasma membrane by activation of the p21 Ras pathway (11, 50). Expression of G3BP-1 in cells promotes entry to S phase (13a), and its levels increase in certain cancers (10a, 13a). Mice lacking functional G3BP-1 alleles die at birth and show defects in fetal growth and increased apoptosis in their central nervous system, with evidence that G3BP-1 is critical for the regulation of multiple imprinted growth regulatory transcripts (55).

G3BP-1 is also a marker for RNA granules called cytoplasmic "stress granules" (SG), which form in stressed cells (20, 21), and when overexpressed, G3BP-1 induces their formation (49). SG contain mRNA, certain translation initiation factors like eukaryotic translation initiation factor 3 (eIF-3), eIF-4E, eIF-4G, RBPs such as PABP-1, G3BP-1, and TIA-1 (20, 49), and 40S ribosomal components. SG are dynamic structures and are thought to function by triaging mRNA for salvage and subsequent transfer to polysomes for translation or destruction by transfer to associated processing bodies for degradation (20, 21). The formation of SG is induced by stalled preinitiation complexes that accumulate through two mechanisms. One involves phosphorylation of eIF-2, which results in a deficiency of the ternary complex eIF-2-GTP-tRNA (Met) and the accumulation of stalled preinitiation translation complexes (20). Phosphorylation of eIF-2 can be induced by stresses such as heat, arsenite, and the unfolded protein response (14, 20) or by stimulation of thymus-derived lymphocytes by antigen (38). Alternatively, SG can be induced by toxins that target the eIF-4F complex (8) or infection by poliovirus that cleaves eIF-GI and eIF-GII (30). By whatever means they are formed, the stalled preinitiation complexes then bind T-cell intracellular antigen-1 (TIA-1) and are recruited into SG by a process that depends on the prion-like domain of TIA-1 (12) and microtubules (48). We report here that Caprin-1 and G3BP-1 heteromerize in a tight complex that colocalizes in cytoplasmic RNA granules on microtubules. Moreover, Caprin-1 resembles G3BP-1 in entering SG induced by arsenite stress and, when overexpressed, induces SG formation through a mechanism that involves RNA binding and induction of phosphorylation of eIF-2.

We hypothesized that the prolongation of G₁-S transition in cycling cells lacking Caprin-1 reflected interaction of the Caprin-1/G3BP-1 complex with mRNAs for proteins involved in G₁/S transition in the cell cycle. Cellular proliferation is tightly regulated by multiple mechanisms that include control of mRNA translation. For example, AU-rich sequence elements (39) occur in the 3' UTR of mRNA for many proteins involved in cell cycling, such as c-Myc and c-Fos, and are bound by RBPs, e.g., of the Hu family (35, 45, 46). RBPs control not only the stability and translation of the mRNA which they bind but also their subcellular localization (25, 32) and interaction with microRNA and associated proteins (5, 6, 17). We asked whether Caprin-1 would selectively bind to the mRNA for two proteins that promote G₁/S transition, c-Myc, which plays a central role in G₁/S transition (18, 28, 33), and cyclin D2, which functions as a regulatory subunit of the CDK4 or CDK6 kinases, whose activity is required for G₁/S transition (7, 33). We show here that Caprin-1 directly and selectively binds mRNA for c-Myc and cyclin D2.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

 
Cells lines and transfections. Human 293T, HeLa, and NIH 3T3 cells were obtained from the American Type Culture Collection. Cells were maintained in Dulbecco's minimal essential medium (Invitrogen Life Technologies) containing 10% fetal bovine serum (FBS) at 7.0% CO₂. NIH 3T3 mouse fibroblasts that stably expressed full-length human Caprin-1 with carboxy-terminal hemagglutinin (HA) tags were described previously (13). Parental chicken DT40 cells and a clone of DT40 cells, R-Caprin-1^–/–, that lacked functional endogenous Caprin-1 genes but expressed human Caprin-1 under the control of a doxycycline-suppressible promoter (53), were maintained in log-phase proliferation at densities between 10⁴ to 10⁶ cells/ml in RPMI 1640 (Invitrogen Life Technologies) supplemented with 10% FBS, 1% chicken serum, and 50 µM 2-mercaptoethanol. Transfection was performed with Lipofectamine 2000 (Invitrogen Life Technologies) according to the manufacturer's instructions. To transfect DT40 cells, 10⁷ cells were electroporated with 50 µg of plasmid at 550 V, 50 µF, using a Gene Pulser (Bio-Rad).

Antibodies and reagents. The monoclonal anti-G3BP-1 antibody was obtained from BD Transduction Laboratories, and a chicken polyclonal anti-G3BP antiserum was obtained from Prosci, Inc. Anti-Caprin-1 serum was generated by immunizing rabbits with bacterially expressed, purified glutathione S-transferase (GST)-human Caprin-1 (hCaprin-1), kindly provided by Marees Harris-Brandts and David Rose. Polyclonal antibodies to TIA-1 (goat), actin (rabbit), and green fluorescent protein (GFP) (rabbit) were from Santa Cruz Biotechnology, those to eIF-2 (rabbit) and phospho-specific eIF-2 (Ser51) (rabbit) were from Cell Signaling Technology, and those to hemagglutinin (HA) (rabbit) and Flag (rabbit) were from Sigma-Aldrich. Mouse monoclonal antibodies (MAb) to Flag and ß-tubulin were from Sigma-Aldrich. Secondary antibodies for immunohistochemistry, Alexa 594-coupled goat antibodies against rabbit or mouse immunoglobulin G (IgG), Alexa 488-coupled goat antibodies against mouse IgG, and Alexa 594-coupled donkey antibodies against goat IgG, were obtained from Molecular Probes. Fluorescein isothiocyanate-coupled goat antibodies against rabbit IgG were from BD Transduction Laboratories. Horseradish peroxidase-conjugated goat secondary antibodies for blotting against mouse or rabbit IgG were from Dako Cytomation. Anti-Flag M2 mouse MAb-coupled affinity beads and anti-HA mouse MAb-coupled beads were from Sigma-Aldrich. Peptides with the sequences FIQDSMLDFE ("Core motif") and QDLMAQMQGPYNFIQDSMLDFE("Extended motif"), corresponding, respectively, to amino acids 372 to 381 and 360 to 381 of Caprin-1, were synthesized by solid-phase chemistry (Phil Owen, BRC, Vancouver, Canada). [³H]leucine for radiolabeling was from Amersham Biosciences. Arsenite, nocodazole, and cycloheximide (CHX) were from Sigma-Aldrich.

Plasmids and site-directed mutagenesis. The hCaprin-1 plasmids encoding human Caprin-1, pEGFP-C1-Caprin-1, pEGFP-C1-HR2 (352 to 709), pIRES-2-Caprin-1-HA, and pCMV-FLAG-Caprin-1 were described previously (13). pCMV-FLAG-Caprin-1 (47 to 380), pCMV-FLAG-Caprin-1 (47 to 327), pCMV-FLAG-Caprin-1 (381 to 709), and pCMV-FLAG-G3BP-1 (142 to 466) were generated by PCR amplification of DNA fragments encoding amino acid residues 47 to 380, 47 to 327, or 381 to 709 for Caprin-1 and 142 to 466 for G3BP-1 and cloning into the pCMV-Tag2a plasmid (Stratagene). The hG3BP-1 plasmids, pCMV-FLAG-G3BP-1, pCMV-FLAG-G3BP-1 (1 to 340), pGEX-4T3-GST-G3BP-1 (1 to 309), and GST-G3BP-1 (229 to 466) were described previously (22). pEGFP-C1-G3BP-1 was a kind gift from J. Tazi. The Stratagene QuikChange mutagenesis kit was used per the manufacturer's protocol to introduce stop codons at codon 328 or 607 in pEGFP-C1-Caprin-1 to generate plasmids pEGFP-C1-Caprin-1 (1 to 327) and pEGFP-C1-Caprin-1 (1 to 606) or at codon 381 or 607 in pEGFP-C1-HR2 (352 to 709) to generate pEFGP-C1-Caprin-1 (352 to 380) and pEFGP-C1-Caprin-1 (352 to 606). Likewise, it was used on pEGFP-C1-G3BP-1 to insert a stop codon at codon 142 to generate pEGFP-C1-G3BP-1 (1 to 141) and on pCMV-FLAG-Caprin-1 (381 to 709) to insert a stop codon at codon 606 to generate pCMV-FLAG-Caprin-1 (381 to 605). pCMV-FLAG-Caprin-1 (381 to 709)-AGGX3 was generated using site-directed mutagenesis to replace the arginines with alanines in the RGG motifs at codon positions 612, 633, and 690 of Caprin-1.

Immunoprecipitation and immunoblotting. 293T cells were transiently transfected with plasmids expressing tagged proteins. After 48 h, cells were lysed using lysis buffer with protease inhibitors on ice for 15 min, and the lysates were centrifuged and assayed for the protein concentration. For each immunoprecipitation (IP), 300 µg of lysate diluted in lysis buffer was agitated overnight at 4°C with 40 µl of agarose beads conjugated with anti-Flag or anti-HA antibodies. For peptide competition experiments, lysates of cells that were transiently expressing Flag-G3BP-1 and GFP-Caprin-1 were agitated overnight at 4°C with anti-Flag-coupled beads. The beads were washed and then agitated for 90 min at 4°C in 1 ml lysis buffer containing 50 µM, 200 µM, or 380 µM peptide or in buffer alone as a control. To evaluate the dependence of coprecipitation on RNA, antibody-coated beads were used to precipitate Flag-G3BP-1 and associated GFP-Caprin-1, and after washing, the beads were agitated in 1 ml buffer containing 100 µg of RNase A for 60 min at room temperature. In all of the above experiments, the beads were washed again and the bound protein was eluted by boiling with sample buffer, subjected to sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE), and transferred to nitrocellulose membranes (51). The membranes were blocked in Tris-buffered saline-Tween buffer with 5% milk powder, and proteins were detected by primary antibodies diluted in blocking buffer, followed by an appropriate secondary antibody conjugated to horseradish peroxidase, wit its binding detected using an enhanced chemiluminescence reagent, as recommended by the manufacturer (Amersham Biosciences).

RNP-IP, RNA extraction, and detection. 293T cells (7 x 10⁶) or 293T cells expressing Flag-tagged Caprin-1 or G3BP-1 or vector alone for 48 h were lysed in polysome-lysis buffer (100 mM KCl, 5 mM MgCl₂, 10 mM HEPES, pH 7.4, and 0.5% Triton X-100, 100 U/ml RNase inhibitor, 1 mM dithiothreitol, 1 mM phenylmethylsulfonyl fluoride, and 100 µg/ml mixture of pepstatin A, bastatin, and leupeptin). The supernatants from the polysome lysate were precleared for 30 min by agitation with 10 µl of normal rabbit sera (for rabbit serum RNP-IP) or 15 µg mouse IgG1 MAb (for mouse MAb RNP-IP) adsorbed on 50 µl protein A- or G-coupled Sepharose beads for rabbit sera and mouse IgG1, respectively. The precleared polysome lysates (3 mg) were diluted in 1 ml polysome lysis buffer and agitated for 2 h at 4°C with either 50 µl of protein-A-coupled Sepharose beads to which had previously been adsorbed IgG from rabbit anti-Caprin-1 serum (20 µl) or control rabbit serum (20 µl) for RNA-IP of endogenous Caprin-1 or, for the endogenous G3BP-1 RNP-IP, 50 µl of protein G-coupled Sepharose beads to which had previously been adsorbed mouse anti-G3BP-1 MAb (30 µg) or a control mouse IgG1 MAb (30 µg). For the Flag-tagged protein RNA-IP, the precleared polysome lysates were incubated with 50 µl of M2 anti-Flag MAb on Sepharose beads (Sigma Aldrich) for 2 h at 4°C. For all of the above RNA-IPs, the beads were washed five times with NT2 buffer (50 mM Tris, pH 7.4, 150 mM NaCl, 1 mM MgCl₂, 0.05% Triton X-100, 100 U/ml RNase inhibitor, and 10 µg/ml of protease inhibitors). The proteins on the beads were digested by resuspension in 100 µl NT2 buffer supplemented with 0.1% SDS and 30 µg RNase-free proteinase K and incubation at 55°C for 30 min. The immunoprecipitated RNA was extracted by phenol-chloroform-isoamyl alcohol and ethanol precipitation. The extracted RNA was treated with 10 U RNase-free DNase I for 15 min at room temperature, followed by reverse transcription (RT)-PCR with primer pairs for cyclin D2 (5'-GATGATCGCAACTGGAAGTG-3' and 5'-AGAGACCAGATTATGGACGC-3'), c-Myc (5'-CCAGAGGAGGAACGAGCTAA-3'and 5'-AGCCAAGGTTGTGAGGTTGC-3'), and glyceraldehyde-3-phosphate dehydrogenase (GAPDH) (5'-TTTGGCTACAGCAACAGGGT-3' and 5'-GGTTGAGCACAGG GTACTTT-3').

Immunohistochemistry. Adherent cells grown directly on glass coverslips or suspension cells centrifuged onto glass slides were fixed with 4% paraformaldehyde in phosphate-buffered saline for 15 min on ice and permeabilized for 10 min with ice-cold methanol. The fixed cells were treated with blocking solution (2% fetal bovine serum in phosphate-buffered saline) and stained with relevant primary antibodies in blocking solution for 1 h at room temperature or at 4°C overnight. The washed cells were stained with the relevant secondary antibody for 1 h at room temperature or 4°C overnight. DNA was stained with 0.2 µg/ml 4',6'-diaminidino-2-phenylindole (DAPI). RNA staining with ethidium bromide was performed as described previously (44). The cells were mounted with Fluoromount-G (Southern Biotech), and digital images were captured by a Qimaging charge-coupled device camera mounted on a Carl Zeiss Axioplan2 microscope with Plan-neofluar 5x/0.25, 10x/0.30, 20x/0.50, and 40x/0.75, Plan-apochromat 63x/1.40 oil, and Plan-neofluar 100x/1.30 oil objective lenses. The microscope was operated by Openlab software 4.0.4 (Improvision imaging software). The color images were finally processed with Adobe Photoshop software.

Arsenite-induced stress, treatment with cycloheximide, and disruption of the microtubules with nocodazole. Cells cultured on glass coverslips in Dulbecco's minimal essential mediumsupplemented with 10% fetal calf serum, 2 mM glutamine, and antibiotics (50 U/ml penicillin and 50 µg/ml streptomycin) at 37°C in 5% CO₂ were treated by the addition of sodium arsenite (0.5 mM for 1 h) to induce SG. To test whether cytoplasmic granules were affected by treatment with cycloheximide, HeLa cells, 48 h after transient transfection with plasmids expressing GFP-Caprin-1 or GFP-G3BP-1, were treated with 100 µg/ml cycloheximide for 1 h and processed for microscopy. To disrupt microtubules, HeLa cells in culture were treated with nocodazole at a final concentration of 33 µM for 90 min and then processed for staining.

Radioactive labeling and protein synthesis. HeLa cells transiently transfected with plasmids expressing GFP-Caprin-1 or GFP alone 24 h earlier were sorted by the FACSVantage (BD sciences) cell sorter for GFP fluorescence. Sorted fractions were incubated for 1 h at 37°C in leucine-free medium, containing 10 µCi/ml [³H]leucine and 5% dialyzed FBS. As a positive control for inhibition of protein synthesis, nontransfected HeLa cells were treated for 10 min with 10 µg/ml CHX prior to addition of [³H]leucine. For assessing global protein synthesis rates in Caprin-1 null cells, R-Caprin-1^–/– DT40 cells that had been grown for 3 days either in the presence or absence of doxycycline (0.5 µg/ml) were cultured with leucine-free medium supplemented with [³H]leucine as described above. In all cases, incorporation of [³H]leucine was halted by addition of excess unlabeled leucine to a final concentration of 0.8 µg/ml. Cells were harvested by centrifugation (13,200 x g, 10 min, 4°C), and washed cells were lysed in 0.1% NP-40 buffer. Ten microliters of the cell lysate was mixed with an equal volume of 10 mg/ml bovine serum albumin, spotted on a microfiber glass filter, washed with 10% trichloroacetic acid and 100% ethanol, and then after drying, immersed in a scintillant fluid. Radioactivity in each sample was determined using a PackardTri-Carb 2200CA liquid scintillation counter. The average incorporation efficiencies were calculated from triplicate samples.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

 
Caprin-1 associates with G3BP-1. During the course of an affinity-directed proteomic approach to elucidating the functions of Caprin-1, we observed that a Flag-tagged fragment of Caprin-1 that encompassed residues 47 to 380 coprecipitated with a series of proteins that either bound mRNA or were associated with its translation. The most prominent of these was G3BP-1, which was of particular interest because it provided a potential mechanistic link between Caprin-1 and cellular proliferation. To further investigate the association of G3BP-1 with Caprin-1, we transiently expressed Flag-tagged Caprin-1 in 293T cells and immunoprecipitated it from cell lysates. Precipitation of exogenous overexpressed Caprin-1 resulted in quantitative coprecipitation of endogenous G3BP-1 (Fig. 1A). The interaction between Caprin-1 and G3BP-1 was confirmed in reciprocal experiments in which endogenous Caprin-1 coprecipitated with overexpressed G3BP-1 (Fig. 1B). These data indicated that Caprin-1 and G3BP-1 were directly or indirectly associated in a stable complex. This was consistent with the report during the course of these studies that the host factor needed for in vitro transcription of vaccinia virus intermediate-stage genes was a heterodimer of G3BP-1 and Caprin-1 (19a).

View larger version (44K): [in this window] [in a new window]   FIG. 1. Caprin-1 and G3BP-1 associate and colocalize in RNA-rich cytoplasmic granules. (A, B) G3BP-1 coprecipitates with Caprin-1. In panel A, 293T cells were transfected with vector alone or Flag-Caprin-1, and total lysates were subjected to anti-Flag IP. The eluate of the immunoprecipitate (IP) and a 1/10 aliquot of the supernatant remaining after IP (S/N) together with a 1/10 aliquot of the whole-cell lysate (WCL) were run in parallel on SDS-PAGE and blotted for endogenous G3BP-1 with a mouse MAb to G3BP-1. Note that the efficient precipitation of the Flag-Caprin-1 was accompanied by a clearing of G3BP-1 from the post-IP supernatant. Panel B shows a reciprocal experiment in which coprecipitation of endogenous Caprin-1 was detected with anti-Caprin-1 rabbit serum. (C) Colocalization of Caprin-1 and G3BP-1 in cytoplasmic RNA granules. Actively proliferating HeLa cells were stained for endogenous Caprin-1 (green) and G3BP-1 (red). Nuclei were stained with DAPI (blue). The arrows in the enlarged area shown in the inset show colocalization of Caprin-1 and G3BP-1 in cytoplasmic granules. (D) Colocalization of Caprin-1 and RNA in cytoplasmic granules. NIH 3T3 cells were fixed and stained for total cellular RNA using 1 µM ethidium bromide (red) according to the method of Tang et al. (44) and with rabbit anti-Caprin-1 serum (green). As a control, some slides were pretreated with RNase before staining with ethidium bromide and anti-Caprin-1 serum. Note the lack of ethidium bromide staining in the cytoplasmic granules and the nucleoli in the nucleus after RNase treatment.

View larger version (53K): [in this window] [in a new window]   FIG. 2. Association of Caprin-1-containing granules with microtubules and with cellular processes. (A) Caprin-1-containing granules occur on a filamentous network resembling the microtubular network and are enriched in cellular processes. 3T3 cells that stably expressed low amounts of Caprin-1-HA were stained for HA (red). Arrows point to the Caprin-1-positive granules arrayed on a filamentous network and enriched in cellular processes. (B, C) Caprin-1 localizes at sites of adhesion and at the leading and trailing edges of migrating 3T3 fibroblasts and HeLa cells. In panel B, 3T3 cells that stably expressed low amounts of Caprin-1-HA were stained for HA (red). In panel C, HeLa cells were stained for endogenous Caprin-1 (green) and G3BP-1 (red). Nuclei were stained with DAPI (blue). The asterisks indicate concentrations of Caprin-1-HA in panel B and Caprin-1 and G3BP-1 in panel C at the sites of cell adhesion and at the leading and trailing edges of cells. (D) Caprin-1-containing cytoplasmic granules are associated with microtubules. Actively growing HeLa cells were treated with nocodazole (33 µM) for 90 min to disrupt microtubules and stained with rabbit anti-Caprin-1 serum (red) and anti-ß-tubulin MAb (green). Note that nocodazole treatment resulted in the loss of the filamentous distribution of Caprin-1 in the cytoplasm and the coredistribution of Caprin-1 and ß-tubulin into blebs. (E) Caprin-1-containing cytoplasmic granules are transport RNPs. Actively growing HeLa cells were stained with anti-Caprin-1 serum (green) and the stress granule marker TIA-1 (red). Note the absence of TIA-1 in Caprin-1-positive cytoplasmic granules.

Caprin-1 interacts with G3BP-1 through an evolutionarily conserved peptide motif. Caprin-1 and G3BP-1 remained associated in the presence of RNase, indicating that their interaction was not dependent upon RNA (Fig. 3C iii). We performed coprecipitation studies on fragments of GFP-Caprin-1 (Fig. 3A) and Flag-G3BP-1 (Fig. 4B) to define the minimal regions of each that are necessary or sufficient for their interaction. A GFP-Caprin-1 fragment that included HR-1, Caprin-1 (1 to 327), did not coprecipitate with Flag-G3BP-1 (Fig. 3Ai). However GFP-Caprin-1 fragments that lacked only the carboxy-terminal RGG-containing domain, Caprin-1 (1 to 606), or lacked both HR-1 and the carboxy-terminal RGG-containing domain, Caprin-1 (352 to 606), were efficiently coprecipitated with Flag-G3BP-1 (Fig. 3Aii). These results indicated that neither the amino-terminal HR-1 region nor the carboxy-terminal RGG-containing domain of Caprin-1 is involved in binding to G3BP-1. Endogenous G3BP-1 coprecipitated with the Flag-Caprin-1 (47 to 380) fragment but not the Flag-Caprin-1 (47 to 327) fragment, thus localizing the region needed for binding G3BP-1 to the 53 amino acids at the carboxy terminus of Flag-Caprin-1 (47 to 380) (Fig. 3Aiii). The fact that two fragments of Caprin-1 that corresponded, respectively, to residues 47 to 380 (Fig. 3Aiii) and 352 to 606 (Fig. 3Aii), both bound to G3BP-1, restricted the minimal region of Caprin-1 required to bind G3BP-1 to the overlapping 29 amino acids that correspond to residues 352 to 380 of Caprin-1. To confirm this conclusion, we fused these 29 residues at the beginning of HR-2 to GFP and performed coprecipitation studies. In contrast to GFP alone, GFP that was fused to these 29 residues of Caprin-1 was coprecipitated with Flag-G3BP-1 (Fig. 3Aiv). The amount of GFP-Caprin-1 (352 to 380) that coprecipitated with G3BP-1 was lower than that precipitated with GFP fused to a larger fragment of Caprin-1 (352 to 606), perhaps due to steric hindrance caused by interaction of the 29 amino acids with the GFP. Nevertheless, this result demonstrated that the sequence of 29 amino acids that corresponded to residues 352 to 380 of Caprin-1 was sufficient for recognition by G3BP-1.

View larger version (84K): [in this window] [in a new window]   FIG. 3. Caprin-1 interacts with G3BP-1 through an evolutionarily conserved peptide motif. (A) G3BP1 binds to a conserved peptide motif in Caprin-1. 293T cells were cotransfected with plasmids expressing Flag-G3BP-1 and various fragments of GFP-Caprin-1 comprising amino acids 1 to 327, 1 to 606, 352 to 606, or 352 to 380, as indicated. Cell lysates were subjected to anti-Flag IP, and precipitated proteins were eluted from the beads and subjected to SDS-PAGE and immunoblotting with anti-GFP, anti-Flag, or anti-G3BP-1 antibodies. (B) (i) Conserved features of Caprin-1- and insect HR-1-containing proteins. The amino-terminal MPSA motifs, the central conserved motif, and the RGG motifs are highlighted. (ii) Also shown is an alignment of the central conserved G3BP-1 binding motif in three insect HR-1-containing proteins and vertebrate Caprin-1 and Caprin-2 together with the core consensus. (C) Peptides containing the G3BP-1-binding Caprin-1 motif compete with Caprin-1 for binding to G3BP-1. (i) Sequence of the core consensus peptide and an extended consensus peptide from the G3BP-1 binding motif from human Caprin-1. (ii) 293T cells were transfected with plasmids expressing Flag-G3BP-1 and GFP-Caprin-1 and lysates were subjected to anti-Flag IP. Washed beads, with bound GFP-Caprin-1 and Flag-G3BP-1 were agitated in 1 ml buffer containing the indicated peptides at 50 µM or 200 µM or buffer alone for 90 min at 4°C. The proteins on the beads were eluted and immunoblotted with anti-GFP and anti-Flag antibodies. (iii) 293T cells were transfected with plasmids expressing Flag-G3BP-1 and GFP-Caprin-1, and lysates were subjected to anti-Flag IP. Washed beads, with bound GFP-Caprin-1 and Flag-G3BP-1 were agitated in 1 ml buffer containing the indicated peptide at 380 µM or buffer alone for 90 min at 4°C. Also shown is a coprecipitation of GFP-Caprin and Flag-G3BP-1, incubated with 100 µg of RNase A in 1 ml of buffer for 60 min. The proteins on the beads were eluted and subjected to SDS-PAGE and immunoblotting with anti-GFP and anti-Flag antibodies.

View larger version (48K): [in this window] [in a new window]   FIG. 4. G3BP-1 binds Caprin-1 through the NTF-2-like domain. (A) Conservation of the NTF-2-like domain and RNA-binding domain of G3BP-1 in human, Xenopus, and Drosophila cells. (B) G3BP-1 mutants used. (C) The RNA-binding domain of G3BP-1 is not necessary for its interaction with Caprin-1. 293T cells were transfected with plasmids expressing Caprin-1-HA and Flag-G3BP-1 or fragment of G3BP-1 (1 to 340), and the cell lysates were subjected to anti-Flag IP. The proteins on the beads were eluted and subjected to SDS-PAGE and immunoblotting with anti-GFP and anti-Flag antibodies. (D) G3BP-1 binds Caprin-1 through its amino terminus. Cell lysates from 293T cells expressing Flag-Caprin-1 were mixed with bacterially expressed GST-G3BP fragments (1 to 309 and 229 to 466) and subjected to anti-Flag IP. The proteins on the beads were eluted and subjected to SDS-PAGE and immunoblotting with anti-GFP and anti-Flag antibodies. (E) G3BP-1 recognizes Caprin-1 through its NTF-2-like domain. 293T cells were transfected with plasmids expressing Flag-Caprin-1 or Flag-Caprin-1 (47 to 380) and GFP-G3BP-1 (1 to 141), and the cell lysates were subjected to anti-Flag IP. The precipitates were subjected to SDS-PAGE and immunoblotting with anti-GFP and anti-Flag antibody.

Structural features of G3BP-1 that binds to Caprin-1. We next defined the region on G3BP-1 that was required for its binding to Caprin-1. Alignment of human G3BP-1 with its Xenopus and Drosophila orthologs showed that the NTF-2-like domain and the RNA-binding domain are well conserved, but the intervening acidic and proline-rich domains are not (Fig. 4A). We thus reasoned that the acidic and proline-rich domains of G3BP-1 were unlikely to mediate binding to the well-conserved peptide in Caprin-1. In contrast, the NTF-2-like domain and the region encompassing the RNA-binding domains were both well conserved and thus plausible candidates. To test these notions, we compared the ability of full-length G3BP-1 and a fragment of G3BP-1 that lacked the RNA-binding domains, G3BP-1 (1 to 340), to coprecipitate with Caprin-1 (Fig. 4B). The G3BP-1 fragment (1 to 340) that lacked the RNA-binding region still coprecipitated with Caprin-1 (Fig. 4C), indicating that the RNA-binding domain of G3BP-1 (residues 341 to 466) is not necessary for binding to Caprin-1.

We then incubated bacterially expressed GST-G3BP-1 fragments corresponding to amino acid residues 1 to 309 or 229 to 466 with lysates of 293T cells transiently expressing Flag-Caprin-1 and tested their ability to be coprecipitated with the Flag-Caprin-1. As seen in Fig. 4D, the amino-terminal fragment of G3BP-1 (1 to 309) efficiently coprecipitated with Caprin-1, while the carboxy-terminal fragment (229 to 466) did not. These observations suggested that the region spanning amino acid residues 1 to 229, which included the NTF-2-like domain, was sufficient for binding Caprin-1. We next tested the ability of a GFP-G3BP-1 fragment that corresponded to residues 1 to 141 and thus encompasses the NTF-2-like domain to bind to Caprin-1. As seen in Fig. 4E, the NTF-2-like domain of G3BP-1 was efficiently coprecipitated by Flag-Caprin-1 as well as by the Flag-Caprin-1 (47 to 380) fragment. We concluded that the interaction between G3BP-1 with Caprin-1 is mediated through the well-conserved NTF-2-like domain of G3BP-1.

Caprin-1 enters arsenite-induced cytoplasmic SG, and its overexpression induces SG assembly. G3BP-1 is a component of SG, and when overexpressed, induces their formation (49). We observed that Caprin-1, along with G3BP-1 and TIA-1, was recruited into cytoplasmic SG that formed in HeLa cells after treatment with arsenite (Fig. 5A). Likewise, arsenite treatment of 3T3 mouse fibroblasts that stably expressed low amounts of HA-tagged Caprin-1 resulted in recruitment of the HA-tagged Caprin-1 into SG (Fig. 5B). As was the case with G3BP-1, overexpression of GFP-Caprin-1 induced the formation of large cytoplasmic SG that contained GFP-Caprin-1 and proteins typical of SG like TIA-1 and G3BP-1 (Fig. 5C). SG can be distinguished from aggregates of unfolded proteins by their sensitivity to treatment with CHX, which stabilizes polysomes and stops the accumulation of stalled translation initiation complexes, resulting in the dissolution of SG (20). SG induced by overexpression of GFP-Caprin-1 disappeared after treatment with CHX (100 µg/ml) for 1 h (Fig. 5D), confirming that the Caprin-1-induced granules were SG and not cytoplasmic aggregates of misfolded Caprin-1.

View larger version (28K): [in this window] [in a new window]   FIG. 5. Caprin-1 enters cytoplasmic stress granules and its overexpression induces them. (A, B) Caprin-1 is recruited into SG induced with arsenite. HeLa cells (A) or 3T3 cells that stably expressed low amounts of Caprin-1 HA (B) were stressed by treatment with arsenite (0.5 mM for 1 h). The cells, before or after stress, were fixed and costained for endogenous Caprin-1 (red) and G3BP-1 (green) or with Caprin-1 (green) and TIA-1 (red) for panel A and for HA (red) for panel B. Nuclei were stained with DAPI (blue). Note the recruitment of Caprin-1 and G3BP-1 into cytoplasmic granules which were also positive for TIA-1. Arrows indicate SG in the cytoplasm. (C) Overexpressed Caprin-1 induces cytoplasmic SG containing G3BP-1 and the SG marker TIA-1. HeLa cells transfected with plasmid expressing GFP-Caprin-1 at 48 h were fixed and stained for G3BP-1 (red) or TIA-1 (red). Nuclei were stained with DAPI (blue). (D) Sensitivity of Caprin-1 induced granules to dissolution by treatment with cycloheximide. HeLa cells were transiently transfected with plasmids expressing GFP-Caprin-1 or with GFP-G3BP-1 and at 48 h, and aliquots were treated with 100 µg/ml cycloheximide for 1 h. Note in both cases the disappearance of SG upon treatment with cycloheximide.

View larger version (53K): [in this window] [in a new window]   FIG. 6. The carboxy-terminal RNA-binding domain of Caprin-1 is necessary for its ability to induce SG when overexpressed. HeLa cells were transfected with plasmid expressing GFP-fusion fragments of Caprin-1, GFP-Caprin-1 (1 to 327), GFP-Caprin-1 (352 to 709), or GFP-Caprin-1 (1 to 606). After 48 h, the cells were fixed and stained for G3BP-1 (red) or TIA-1 (red). Nuclei were stained with DAPI (blue). Note the absence of SG formation by GFP-Caprin-1 (1 to 327) and GFP-Caprin-1 (1 to 606).

View larger version (40K): [in this window] [in a new window]   FIG. 7. The RNA-binding domain of a single interacting partner of the Caprin-1-G3BP-1 complex is necessary and sufficient for the entry of the complex to SG. (A) HeLa cells were cotransfected with plasmids encoding Flag-G3BP-1 and those expressing GFP fusions of fragments of Caprin-1, namely GFP-Caprin-1 (1 to 606), GFP-Caprin-1 (352 to 606), or GFP-Caprin-1 (1 to 327). After 48 h, the cells were stained for Flag (red). Nuclei were stained with DAPI. None of these Caprin-1 fragments induce SG formation when expressed alone (data not shown). However, when coexpressed with Flag-G3BP-1, the Caprin-1 fragments 1 to 606 and 352 to 606 colocalized with G3BP-1-induced SG, but the Caprin-1 fragment, 1 to 327, that lacks the motif for binding G3BP-1 did not. (B) 293T cells were transfected with plasmid expressing GFP fused to a 29-amino-acid peptide from Caprin-1, GFP-Caprin-1 (352 to 380) with or without Flag-G3BP-1. After 48 h, the cells were fixed and stained for Flag (red). Nuclei were stained with DAPI. Note that the GFP-Caprin-1 (352 to 380) entered SG formed in cells coexpressing G3BP-1. (C) The NTF-2-like domain of G3BP-1 enters SG only when coexpressed with Caprin-1. HeLa cells were transfected with the GFP fusion of the fragment of G3BP-1 corresponding to the NTF-2 like domain, GFP-G3BP-1 (1 to 141), with and without Flag-Caprin-1. After 48 h, the cells were fixed and stained for Flag (red). Nuclei were stained with DAPI. Note that in cells overexpressing Flag-Caprin-1, the GFP-G3BP-1 (1 to 141) relocated from the nucleus into the cytoplasmic SG that were induced by overexpression of Flag-Caprin-1.

Caprin-1 and G3BP-1 can independently enter SG and induce their formation. To determine whether Caprin-1 is essential for the formation of SG, we made use of a clone of the chicken B-lymphocyte cell line R-Caprin^–/– DT40 cells in which the endogenous Caprin-1 genes had been ablated by gene targeting. In these cells, the expression of conditionally expressed human Caprin-1 that complements the lack of endogenous Caprin-1 can be completely suppressed by treatment with doxycycline for 3 days (53). We observed that overexpression of GFP-G3BP-1 in cells that lacked both human and endogenous Caprin-1 still resulted in formation of SG that contained GFP-G3BP-1 (Fig. 8A). Thus, the presence of Caprin-1 was not required for the induction of SG formation in response to overexpression of G3BP-1 or for the entry of G3BP-1 into these SG.

View larger version (20K): [in this window] [in a new window]   FIG. 8. Caprin-1 and G3BP-1 can independently induce the formation of SG and enter them. (A) Caprin-1 is not needed for the formation of SG by G3BP-1 or for the entry of G3BP-1 to SG. Avian R-Caprin-1–/– DT40 cells that express no endogenous Caprin-1 but expressed human Caprin-1 under the control of a DOX-suppressible promoter were grown for 3 days with (+) or without (–) DOX, which suppresses the expression of human Caprin-1 to undetectable levels. They were transfected with GFP-G3BP-1 and grown in the continued presence or absence of 0.5 µg/ml of DOX for 16 h. Note that expression of the GFP-G3BP-1 induced SG (arrows) in both the control cells and those that lacked Caprin-1. (B) Interaction with endogenous G3BP-1 is not needed for Caprin-1 for the formation of SG or for its entry to SG. HeLa cells were transfected with a Flag-tagged fragment of Caprin-1 (381 to 709) that does not interact with G3BP-1, and after 48 h, the cells were fixed and stained. Note the induction of SG containing Flag-Caprin-1 (381 to 709) in the cytoplasm.

Caprin-1 induces eIF-2 phosphorylation through a novel mechanism dependent on its interaction with RNA. Recently, it has been shown that SG formation can be induced by two mechanisms, only one of which depends on induction of phosphorylation of eIF-2 (8, 30). To explore the mechanism through which overexpression of Caprin-1 induced SG, we overexpressed Flag-Caprin-1 and investigated the phosphorylation of eIF-2 by immunoblotting. We observed that overexpression of Caprin-1 resulted in phosphorylation of eIF-2 (Fig. 9A).

View larger version (30K): [in this window] [in a new window]   FIG. 9. Overexpression of Caprin-1 induces of phosphorylation of eIF-2 through a mechanism that depends on RNA binding. (A) Overexpression of Caprin-1 induced phosphorylation of eIF-2. 293T cells were transfected with vector alone or Flag-tagged Caprin-1. After 48 h, the total cell lysates were separated on an SDS-PAGE gel and immunoblotted for phosphorylated eIF-2, for total eIF-2, and for ß-actin as a loading control. (B) Caprin-1 induces eIF-2 phosphorylation through an RNA-dependent mechanism. As in panel A, 293T cells were transfected with Flag-tagged Caprin-1 or Flag-tagged Caprin-1 mutants and, for controls, with Flag-smgGDS or empty vector and incubated for 48 h prior to lysis. As a positive control for eIF-2 phosphorylation, 293T cells were stressed with 0.5 mM sodium arsenite for 1 h. Total cell lysates were separated on an SDS-PAGE gel and immunoblotted for phosphorylated eIF-2 and total eIF-2. The expression of the Flag-tagged proteins was confirmed by immunoblotting with anti-Flag antibodies.

View larger version (36K): [in this window] [in a new window]   FIG. 10. The carboxy terminus of Caprin-1 selectively binds mRNAs associated with cellular proliferation through a mechanism dependent on the RGG motifs. (A) mRNA for c-Myc and cyclin D2 coprecipitate with endogenous Caprin-1 and G3BP-1. 293T cells were lysed in polysome lysis buffer. The supernatants of these lysates were precleared as described using beads coated with normal rabbit serum or mouse IgG1 for the anti-Caprin-1 and anti-G3BP-1 precipitations, respectively. Precleared polysome lysates were then incubated with protein A-Sepharose beads, which had been conjugated with anti-Caprin-1 rabbit serum, control rabbit serum, or protein G-Sepharose beads conjugated with the anti-G3BP-1 mouse MAb or an isotype-matched control mouse IgG1 MAb. RNA was extracted from the proteins bound to the beads after IP, as described above. RT-PCR was performed as described to detect the presence of c-Myc, cyclin D2, and GAPDH transcripts. (B) A carboxy-terminal fragment of Caprin-1 (381 to 709) that fails to bind G3BP-1 selectively binds mRNA for c-Myc and cyclin D2. As before, 293T cells were transfected with the indicated Flag-tagged fragments comprising the carboxy termini of Caprin-1 (381 to 709) or G3BP-1 (142 to 466) or the amino-terminal HR1 region of Caprin-1, Caprin-1 (47 to 380). The fragments were immunoprecipitated with anti-Flag as described above, and associated mRNA were assayed as before. Analysis of IP by SDS-PAGE and immunoblotting with antibody to Caprin-1 and G3BP-1 confirmed that fragments did not interact with endogenous protein (data not shown). (C) The RGG motifs in Caprin-1 are essential for selective binding of c-Myc and cyclin D2 mRNA. As before, 293T cells were transfected with a Flag-tagged carboxy-terminal fragment of Caprin-1 (381 to 709), a fragment in which each of the RGG motif had been mutated to AGG (AGGX3), or a Caprin-1 fragment that was truncated at residue 606 so it lacked the RGG motifs entirely, Caprin-1 (381to 605). The fragments were immunoprecipitated with anti-Flag and associated mRNAs were assayed as before.

Given that Caprin-1 and G3BP-1 interact, these experiments did not determine whether the c-Myc and cyclin D2 mRNA were binding to the Caprin-1 or to the G3BP-1 or whether binding was cooperative and required both partners in the heterodimer.

To test whether individual subunits of the Caprin-1/G3BP-1 complex could directly interact selectively with the mRNA and to define the structural requirement for RNA binding involved, we expressed truncation mutants of Flag-tagged Caprin-1 and G3BP-1 that did not bind to their endogenous binding partner. Thus, we used a carboxy-terminal fragment of Caprin-1, Flag-Caprin-1 (381 to 709), that lacked the first 29 amino acids of HR-2 that were necessary for recognition by G3BP-1, as well a carboxy-terminal fragment of G3BP-1, Flag-G3BP-1 (142 to 499), that lacked the NTF-2 domain needed for interacting with Caprin-1. We confirmed that these fragments did not coprecipitate endogenous G3BP-1 and Caprin-1, respectively, by overexpressing them in 293T cells (data not shown). We observed that the Caprin-1 (381 to 709) fragment associated selectively with c-Myc and cyclin D2 mRNA. In contrast, the carboxy-terminal fragment of G3BP-1 (142 to 499) bound only barely detectable amounts of c-Myc and cyclin D2 mRNA (Fig. 10B). This suggested that the carboxy terminus of Caprin-1 interacted directly with c-Myc and cyclin D2 mRNA and did not require interaction with G3BP-1 to selectively bind these mRNAs (Fig. 10B). We then tested the Caprin-1 (47 to 380) fragment, which included the HR-1 domain of Caprin-1, the intervening region, and the first 20 amino acids of HR-2 containing the G3BP-1 binding motif. We observed that this fragment did not selectively associate with c-Myc or cyclin D2 mRNA (Fig. 10B). Next we determined whether the RGG motifs in the carboxy-terminal domain of Caprin-1 were important for binding to c-Myc and cyclin D2 mRNA. We mutated all three RGG motifs to AGG motifs [Caprin-1 (382 to 709X3AGG)] or completely removed the RGG- and RG-rich domain by introducing a stop codon at amino acid position 606 [Caprin-1 (381 to 605)]. We observed that mutation of the three RGG motifs to AGG motifs resulted in a major reduction of binding of mRNA for c-Myc or cyclin D2 (Fig. 10C). Moreover, deletion of the glycine-rich region of the carboxy terminus with the RG and RGG motifs completely abolished binding of the Caprin-1 fragment to these mRNAs (Fig. 10C). These data indicated that Caprin-1 directly and selectively bound mRNAs for c-Myc and cyclin D2 through its carboxy-terminal RGG-rich region.

Overexpression of Caprin-1 results in global suppression of protein synthesis. The induction of phosphorylation of eIF-2 induced by expression of Caprin-1 would be predicted to induce a global block in protein synthesis. To investigate this possibility, we transfected HeLa cells with GFP-Caprin-1 or GFP alone and purified cells expressing GFP by fluorescence-activated cell sorting. We investigated the global rates of protein synthesis in these two populations of GFP-positive cells by quantifying their incorporation of [³H]leucine over a short incubation. We observed that cells expressing GFP-Caprin-1 exhibited a significant inhibition of global protein synthesis compared with cells expressing approximately equimolar levels of GFP (Fig. 11A). These findings provide an explanation for the inhibitory effects of GFP-Caprin-1 overexpression upon proliferation that was not seen with equimolar expression of GFP (13), suggesting that they were secondary to the induction of phosphorylation of eIF-2 and consequent inhibition of protein synthesis. Given that, in these experiments, Caprin-1 was expressed at nonphysiological levels, these data do not necessarily imply that Caprin-1 acts as a translational repressor.

View larger version (9K): [in this window] [in a new window]   FIG. 11. Global protein synthesis is inhibited by overexpression of Caprin-1 but is not affected by the absence of Caprin-1. (A) Overexpression of Caprin-1 inhibits protein synthesis. HeLa cells transiently expressing GFP alone or GFP-Caprin-1 were purified using a fluorescence-activated cell sorter and assessed for the rates of protein synthesis by incorporation of [3H]leucine as described above. As a positive control for inhibition of protein synthesis, nontransfected cells were treated with 10 µg/ml CHX. Note the significant decrease in global rates of protein synthesis in cells expressing GFP-Caprin-1 compared with those expressing equimolar amounts of GFP alone. (B) Cells lacking Caprin-1 exhibit no global changes in protein synthesis. Avian R-Caprin-1–/– DT40 cells were grown in the presence of DOX for 3 days to suppress the expression of human Caprin-1 to undetectable levels. Together with control R-Caprin-1–/– cells cultured in the absence of DOX, they were assessed for rates of protein synthesis by incorporation of radioactive leucine label as described. Parental DT40 cells were treated with CHX (10 µg/ml) as a positive control. Note that the R-Caprin–/– DT40 exhibited similar rates of protein synthesis in the presence (+) or absence (–) of DOX. Results are expressed as means ± standard errors of the means.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

 
These results show that in HeLa epithelial cells or 3T3 fibroblasts, Caprin-1 and G3BP-1 form a complex and that this complex is localized in cytoplasmic granules which contain a major part of the cytoplasmic RNA (Fig. 1C and 1D). Most of these granules were associated with microtubules (Fig. 2A), and disruption of microtubules resulted in the relocalization of Caprin-1 to tubulin aggregation regions adjacent to the plasma membrane (Fig. 2D). The Caprin-1- and G3BP-1-containing granules accumulated at the leading and trailing edges of migrating cells (Fig. 2B and C), raising the possibility that, at these sites, they were associated with the actin cytoskeleton. These observations are consistent with the notion that Caprin-1/G3BP-1 RNA granules are transported in a microtubule-dependent manner to sites such as the leading or trailing edges of migrating cells that are rich in F-actin and RNA (3). They are also consistent with reports that immunoprecipitates of Paxillin from focal adhesions contain Caprin-1 (9) and that G3BP-1 localized close to the membrane at areas of cell polarization and was concentrated in filopodia and sites where integrins had been cross-linked (31, 37). Our observations on epithelial cells and fibroblasts extend those of Shiina et al. (40), who reported that Caprin-1 was localized in presynaptic granules in the dendrites of neurons, by showing that Caprin-1-containing granules are also present in nonneuronal cells, where they are abundant and contain a significant fraction of the cytoplasmic RNA.

Our demonstration that Caprin-1 selectively binds to particular mRNAs (Fig. 10) is also in general agreement with the observations of Shiina et al. (40), who showed that Caprin-1 coprecipitated from lysates of brain cells with a selected subset of mRNAs for proteins involved in synaptic plasticity, viz., CaMKII, BDNF, TrkB, and MAP2, but not those for NMDAR or importin ß (40). However, there are important differences between their conclusions and ours with regard to the structural basis of the binding of mRNA to Caprin-1. Thus, based on their demonstration that in vitro mRNAs bound directly to the amino-terminal region of the recombinant Caprin-1, they concluded that the important region of Caprin-1 for mRNA-binding was the amino-terminal region and not the carboxy terminus. In contrast, we failed to precipitate mRNA for c-Myc or cyclin D2 from cell lysates with a transiently expressed amino-terminal fragment of Caprin-1 (Fig. 10B). However, the binding observed by Shiina et al. in their in vitro experiments was not sequence specific and may have been due to the abundance of basic residues in the amino-terminal region in HR-1. We postulate that in vivo these basic charges may be masked by interactions of HR-1 with other proteins and that the non-sequence-selective in vitro binding of recombinant HR-1 to mRNA observed by Shiina et al. may be artifactual. Certainly, our experiments indicate that the selective binding of c-Myc and cyclin D2 mRNA is a property of the carboxy terminus of Caprin-1 and, to a large extent, is mediated by the RGG motifs.

We were somewhat surprised that the truncation mutant of G3BP-1 that retained the RNA-binding domain but lacked the ability to interact with endogenous Caprin-1 bound very weakly to c-Myc mRNA (Fig. 10B), as G3BP-1 had previously been reported to bind selectively to the mRNA for c-Myc (11, 50), cdk7 and cdk9 (27), and tau (2a). However, given that Caprin-1 and G3BP-1 form a tight complex, it is possible that all of these mRNAs bind directly to Caprin-1 or, more likely, are bound cooperatively by the RNA-binding domains of both Caprin-1 and G3BP-1. The regulation of the association of a particular mRNA with Caprin-1/G3BP-1 may be highly regulated and depend on posttranslational modification and interactions with other proteins. For example, the interaction of G3BP-1 and cdk7 and cdk9 mRNA is dependent upon the interaction of G3BP-1 with RasGAP and filamin (27).

Caprin-1 joins a small group of proteins that enter SG and, when overexpressed, induce their formation. The mechanism of SG formation by overexpression of Caprin-1 involved phosphorylation of eIF-2, and only those mutants of Caprin-1 that induced phosphorylation of eIF-2 induced SG formation (Fig. 6 and 9B). Our data show that this induction of phosphorylation of eIF-2 was not due to the induction of an unfolded-protein response but instead correlated closely with the ability of fragments of Caprin-1 to bind selectively to mRNA (Fig. 9B). Indeed, the fact that the carboxy-terminal fragment of Caprin-1 with the intact RGG motifs that induced phosphorylation of eIF-2 also selectively bound particular mRNAs (Fig. 10) suggests that the induction of phosphorylation of eIF-2 by Caprin-1 fragments requires them to be well folded and in their native conformation. We conclude that the induction of phosphorylation of eIF-2 that occurs when Caprin-1 (and probably other RBPs) is overexpressed is not due to an unfolded-protein response but reflects an intrinsic property of the complex of Caprin-1 and mRNA. This is consistent with observations that SG formation is induced by overexpression of a series of structurally diverse RBPs that include G3BP-1 (49), TIA-1 (12), Fragile X mental retardation protein (FMRP) (29), survival of motor neurons protein (15), TTP (42), and Roquin (52). In the case of overexpression of FMRP, the formation of the cytoplasmic granules depended on the presence of its RGG-rich domain (29). Our results demonstrate that while RNA-binding proteins typically enter SG in stressed cells and, when overexpressed, induce SG formation, there are different structural requirements required for the induction of SG formation and for entry to SG. Although the ability to bind mRNA is required for RBPs to induce formation of SG and to enter them (Fig. 6), in the case of SG entry, this requirement can be replaced by an interaction with another protein that does enter SGs (Fig. 7). Thus, a non-RNA-binding protein such as TRAF-2 enters SG due to interaction with eIF-4GI (24).

The notion that the complex of Caprin-1 and mRNA has an intrinsic ability to trigger phosphorylation of eIF-2 raises the question of the mechanism involved. One possibility is that, when overexpressed, Caprin-1 stabilizes bound mRNA and presents it in a conformation that directly activates an eIF-2 kinase. The eIF-2 kinase protein kinase R (PKR) is characteristically activated by double-stranded RNA of viral origin but can be activated by endogenous mRNA (36), and cells lacking PKR activity exhibit increased expression of exogenous proteins (19, 47). Moreover, there is evidence that a pseudo-knot in the 5' UTR of gamma interferon mRNA can activate PKR and result in local phosphorylation of eIF-2 (4). It will be important to determine whether the formation of SG in response to overexpression of Caprin-1 and other RNA-binding proteins is dependent on PKR. It is conceivable that the intrinsic, RNA-dependent propensity of Caprin-1 to induce the phosphorylation of eIF-2 when overexpressed reflects an exaggeration of a physiological mechanism through which translation of Caprin-1-bound mRNA is suppressed through local phosphorylation of eIF-2. This would parallel the local PKR-dependent induction of phosphorylation of eIF-2 that results in the local suppression of translation of a mRNA for gamma interferon (4). PKR is well situated for such a role, being localized to the 40S ribosome, which associates with eIF-2 (56).

It remains to be determined whether the Caprin-1/G3BP-1 complex promotes or represses translation of mRNA to which it binds. Certainly the fact that Caprin-1 occurs in both polysome-associated and untranslated mRNPs (1) suggests that posttranslational modifications or the presence of other RBPs or microRNAs may determine whether the translation of a particular mRNA bound to Caprin-1 is repressed or promoted. For example, G3BP-1 associates with mRNAs for both Cdk7 and Cdk9 but increases levels of Cdk7 protein while decreasing levels of Cdk9 (27). Likewise, levels of many mRNAs increase and many decrease in cells lacking G3BP-1 (55), suggesting that Caprin-1/G3BP-1 may affect the translation and stability of mRNAs either positively or negatively. In demonstrating that overexpression of Caprin-1 induces phosphorylation of eIF-2 through its binding to mRNA, we have raised a caveat to the interpretation of those experiments in the literature in which overexpression of RNA-binding proteins has been used to probe their effects on translation. In light of our results, it is possible that overexpression of many RNA-binding proteins may readily induce eIF-2 phosphorylation and, thus, global inhibition of protein synthesis. This would certainly be predicted to be the case for those RBP that induce SG formation when overexpressed, such as FMRP (29), G3BP-1 (49), TIA-1 (12), Roquin (52) and SMN (15). For example, expression of exogenous FMRP resulted in the repression of translation of reporter genes, with the authors noting that it was associated with the appearance of granules that resembled SG (29). It should be noted that the levels of exogenously expressed Caprin-1 needed to induce global inhibition of protein synthesis and the resultant inhibition of proliferation (13) exceed even high physiological levels of Caprin-1 (13). That physiological levels of Caprin-1 do not repress global protein synthesis is demonstrated by the fact that cells that lacked Caprin-1 exhibited normal and not increased rates of global protein synthesis (Fig. 11B). It is likely that the effects of Caprin-1 on translation are highly regulated by posttranslational modification (13, 43) and interactions with other proteins, and, as appears to be the case with G3BP-1 (11, 27, 50), will be positive or negative depending on the mRNA and associated proteins. The mechanisms through which the Caprin-1/G3BP-1 complex regulates the translation of the mRNA it binds are likely to be complex.

	ACKNOWLEDGMENTS

 
We thank Jamal Tazi, Institute of Molecular Genetics, Montpellier, France, for the G3BP-1 plasmids and Marees Harris-Brandts and David Rose of the University of Toronto, Toronto, Canada, for GST-Caprin-1. We also thank Andrew Johnson for his excellent technical assistance with flow cytometry and the colleagues at the BRC for the helpful discussions and critical reading of the manuscript.

This study has been supported by grants from Canadian Institute for Health Research (CHIR) and a Fellowship to S.S. from the Canadian Arthritis Network.

	FOOTNOTES

 
* Corresponding author. Mailing address: The Biomedical Research Centre, University of British Columbia, 2222 Health Sciences Mall, Vancouver, B.C. V6T 1Z3, Canada. Phone: (604) 822-7822. Fax: (604) 822-7815. E-mail: john{at}brc.ubc.ca.

Published ahead of print on 8 January 2007.

	REFERENCES

Top Abstract Introduction Materials and Methods Results Discussion References

 
1. Angenstein, F., A. M. Evans, S. C. Ling, R. E. Settlage, S. Ficarro, F. A. Carrero-Martinez, J. Shabanowitz, D. F. Hunt, and W. T. Greenough.2005 . Proteomic characterization of messenger ribonucleoprotein complexes bound to nontranslated or translated poly(A) mRNAs in the rat cerebral cortex. J. Biol. Chem. 280:6496-6503.[Abstract/Free Full Text]

2. Angenstein, F., A. M. Evans, R. E. Settlage, S. T. Moran, S. C. Ling, A. Y. Klintsova, J. Shabanowitz, D. F. Hunt, and W. T. Greenough.2002 . A receptor for activated C kinase is part of m