C3G dynamically associates with nuclear speckles and regulates mRNA splicing.

C3G (Crk SH3 domain binding guanine nucleotide releasing factor) (Rap guanine nucleotide exchange factor 1), essential for mammalian embryonic development, is ubiquitously expressed and undergoes regulated nucleocytoplasmic exchange. Here we show that C3G localizes to SC35-positive nuclear speckles and regulates splicing activity. Reversible association of C3G with speckles was seen on inhibition of transcription and splicing. C3G shows partial colocalization with SC35 and is recruited to a chromatin and RNase-sensitive fraction of speckles. Its presence in speckles is dependent on intact cellular actin cytoskeleton and is lost on expression of the kinase Clk1. Rap1, a substrate of C3G, is also present in nuclear speckles, and inactivation of Rap signaling by expression of GFP-Rap1GAP alters speckle morphology and number. Enhanced association of C3G with speckles is seen on glycogen synthase kinase 3 beta inhibition or differentiation of C2C12 cells to myotubes. CRISPR/Cas9-mediated knockdown of C3G resulted in altered splicing activity of an artificial gene as well as endogenous CD44. C3G knockout clones of C2C12 as well as MDA-MB-231 cells showed reduced protein levels of several splicing factors compared with control cells. Our results identify C3G and Rap1 as novel components of nuclear speckles and a role for C3G in regulating cellular RNA splicing activity.

INTRODUCTION

Many molecules function in signaling pathways through dynamic nucleocytoplasmic exchange to regulate nuclear functions like chromatin organization, gene expression, and RNA processing. Within the nucleus, proteins may be present in the nucleoplasm or associated with nuclear substructures such as chromatin, the nuclear matrix, nuclear membrane, nucleoli, Cajal bodies, or nuclear speckles (Handwerger and Gall, 2006). Their localization often provides insights into the functions they perform in the nucleus. Replication, transcription, and DNA repair take place in distinct nuclear regions and are generally defined by dynamics of chromatin remodeling and nuclear architecture (Stein ).

Nuclear speckles are distinct, irregularly shaped structures present in interchromatin regions (Spector and Lamond, 2011). They are complexes of a large number of proteins and snRNPs and are sites for regulation of transcription and splicing in cells (Spector, 1993; Zhou ; Galganski ). Precise excision of introns from the primary transcript is important to generate spliced translatable mRNAs in eukaryotes. RNA splicing is a regulated process mediated by a multiprotein complex called spliceosome that facilitates the chemical removal of introns (Misteli ). The serine arginine (SR) family of proteins localize dynamically to speckles along with several other regulatory molecules like kinases and phosphatases (Spector and Lamond, 2011). Some of these proteins have signature sequences that enable their targeting to speckles (Jagiello ; Eilbracht and Schmidt-Zachmann, 2001; Salichs ), and phosphorylation plays a major role in regulating localization to speckles (Mermoud ; Stamm, 2008). While some proteins are constitutive to speckles and play a role in functional mRNA biogenesis, others show restricted expression and tissue specific alternate splicing events (Castle ; Wang ). Speckle size and structure are dependent on transcription and splicing activity in cells, and active transcription takes place at the periphery of speckles (Misteli ; Lamond and Spector, 2003). RNA transcription and splicing are altered on transient heat stress, which also affects the localization of various factors to speckles (Spector ; Lallena and Correas, 1997; Biamonti and Caceres, 2009; Velichko ).

C3G is a ubiquitously expressed guanine nucleotide exchange factor (GEF) that regulates important cellular functions like cytoskeletal remodeling, adhesion, cell proliferation, and apoptosis (Radha ). It is a 140-kDa protein with a catalytic domain at the C-terminus and a central protein interaction domain containing multiple poly-proline tracts. The N-terminus has an E-cadherin–binding domain, and the noncatalytic domains serve to negatively regulate its catalytic activity (Ichiba ; Hogan ). In addition, C3G activity is regulated by tyrosine phosphorylation and membrane localization (Ichiba ). Functions of C3G are dependent on its catalytic and/or protein interaction properties. C3G targets GTPases Rap1, Rap2, R-Ras, and TC10 (Gotoh ; Mochizuki ) and interacts with signaling molecules like Hck, Src, Crk, c-Abl, TC-48, and β-catenin (Knudsen ; Tanaka ; Shivakrupa ; Radha ; Mitra ; Dayma ). C3G suppresses β-catenin activity, a determinant of cell fate decisions (Dayma ). C3G levels increase and are required for differentiation of muscle and neuronal cells (Radha ; Sasi Kumar ). C3G is essential for early embryonic development in mammals, as mice lacking C3G do not survive beyond 6–7.5 d postcoitum (Ohba ).

Indirect immunofluorescence experiments have shown that cellular, as well as overexpressed C3G localizes primarily to the cytoplasm (Hogan ; Radha ). The primary sequence of C3G has functional NLSs, and fractionation of a variety of cells has shown the presence of C3G in the nuclear compartment (Shakyawar ). Inhibition of chromosomal maintenance 1– (CRM1) dependent export, as well as glycogen synthase kinase 3 beta (GSK3β) activity, caused an increase in nuclear levels of C3G. In an attempt to understand functions of C3G in the nucleus, we examined its localization to distinct subnuclear structures. Our findings demonstrate that C3G and its target Rap1 show dynamic association with nuclear speckles dependent on cellular transcription activity. Differentiation of myocytes to myotubes enhances association of C3G with nuclear speckles and knockdown of C3G alters cellular splicing activity.

RESULTS

Inhibition of nuclear export causes enrichment of C3G in SC35-positive speckles

The dynamic exchange of C3G between nuclear and cytoplasmic compartments prompted a study of its subnuclear localization, to understand its functions in the nucleus. Since C3G localized predominantly to interchromatin regions on inhibition of nuclear export, we examined its localization to nuclear speckles by costaining for SC35 (SRSF2), a spliceosomal protein of SR family, and an established nuclear speckle marker. This antibody recognizes a phosphoepitope of SC35 (Fu and Tom, 1990). A commercial antibody raised against N-terminal region of C3G that specifically recognizes endogenous C3G was used (Figure 1A). This antibody recognizes endogenous C3G in methanol-fixed cells when examined by indirect immunofluorescence. Confocal microscopy showed prominent staining for C3G in nuclear foci positive for SC35 on treatment of MDA-MB-231 cells with nuclear export inhibitor leptomycin B (LMB) (Figure 1B). Untreated cells show cytoplasmic and weak nuclear staining. The efficacy of LMB treatment was validated by observing nuclear localization of NF-κB, a protein regulated by CRM1-dependent nuclear export (Supplemental Figure S1A). Localization of C3G to speckles in response to LMB was seen in other cell types such as MCF-7 (Supplemental Figure S2A). Compared to MDA-MB cells, C2C12 myoblast cells were more responsive to LMB, and the number of interphase cells showing C3G in nuclear foci peaked by 11 h of treatment (Supplemental Figure S2, B–D). Fluorescence intensity of C3G in SC35 speckles was significantly higher in LMB-treated cells compared with untreated cells (Supplemental Figure S2E). Colocalization of endogenous C3G was also seen with overexpressed GFP-SC35 on LMB treatment (Supplemental Figure S2F). All cells overexpressing GFP-SC35 showed prominent localization of C3G to splicing factor compartments, suggesting that enhanced expression of SC35 could induce localization of C3G to speckles when nuclear export is inhibited.

FIGURE 1:

C3G localizes to SC35 nuclear speckles. (A) MDA-MB-231, HeLa, and C2C12 cell lysates were subjected to immunoblotting and probed for C3G expression using C3G antibody. (B) Normally growing MDA-MB-231 cells (UT) on cover slips or those treated with LMB were fixed and immunostained for C3G and SC35. Panels show optical sections taken using a confocal microscope. (C) MDA-MB-231 cells were subjected to α-amanitin treatment and immunostained for C3G and SC35. BL indicates cells processed similarly without addition of primary antibodies. (D) The bar diagram represents the Pearson correlation coefficient of colocalization between C3G and SC35 analyzed from multiple images on indicated treatments. ***p < 0.001. (E) Optical sections (z-plane step, 0.30 µm) of nuclei from amanitin-treated cells were captured on Leica SP8 confocal microscope and were reconstructed to form a three-dimensional image. Three-dimensional visualization of speckle regions of cells dually labeled with antibodies against C3G (Red) and SC35 (Green) in XY or XZ plane are shown. Line scans showing local intensity distributions of C3G in red and SC35 in green in the ROI drawn across two speckles are shown to the right of the panels.

Inhibition of transcription results in enhanced localization of C3G to speckles

The shape, size, and number of speckles change, depending on cellular transcription levels (Melčák ). Inhibition of transcription causes formation of large intranuclear foci containing splicing factors in cells (Carmo-Fonseca ; Spector, 1993; Hall ). This is a characteristic feature of nuclear speckles (interchromatin granule clusters) that undergo reorganization and fuse to form large structures that are more rounded and fewer in number. To examine if C3G localization to nuclear speckles responded to inhibition of transcription, α-amanitin (AMA)-treated MDA-MB cells were dual stained with C3G and SC35 antibodies. We observed a redistribution of C3G into enlarged speckles in response to inhibition of transcription, similarly to that seen for SC35 (Figure 1C). No signal was seen in speckles when C3G primary antibody was omitted, and cells were exposed to only secondary antibodies. Quantitation of the extent of colocalization between C3G and SC35 showed a significant increase in LMB and amanitin-treated cells (Figure 1D). Inhibition of transcription by amanitin was confirmed by examining RNA pol II using the antibody H5, which recognizes the active Ser2 phosphorylated form of RNA Pol II (Supplemental Figure S1B) (Warren ). Treatment with amanitin resulted in all interphase cells showing C3G in speckles. Overexpressed C3G was also seen associated with SC35 speckles in amanitin-treated cells. A deletion construct lacking N- and C-terminal sequences, and containing only the Crk-binding region (CBR), which primarily localizes to the nucleus, did not show association with speckles in amanitin-treated cells, even on detergent extraction prior to fixation (Supplemental Figure S3). These results suggest that the proline-rich interaction domain of C3G is insufficient for nuclear speckle location.

Localization of C3G to speckles was confirmed using two alternate antibodies, C9 and C19, that target the central CBR and C-terminus region of C3G, respectively. In amanitin-treated cells, C3G showed weak localization to SC35 speckles in formaldehyde, as well as methanol-fixed cells when stained using C9 or C19 antibodies (Supplemental Figures S4, A and B). In both cases, prior extraction of cells with mild detergent resulted in better visualization of C3G localization to speckles. While SC35 staining looks similar between extracted and unextracted cells when fixed with methanol, formaldehyde-fixed cells showed diffuse nucleoplasmic staining after extraction, in addition to speckle localization. To rule out cross-reactivity of C3G antibodies with other speckle-associated proteins, C3G-GFP fusion protein was transiently expressed in MDA-MB cells and examined for colocalization of GFP with SC35. C3G-GFP expressing cells show predominant cytoplasmic localization, similarly to that of C3G. On amanitin treatment, distinct GFP fluorescence was seen at SC35-positive structures in methanol-fixed cells (Supplemental Figure S5A). Formaldehyde-fixed cells showed GFP at speckles only when subjected to detergent extraction prior to fixation (unpublished data). Cells transiently coexpressing Flag-tagged C3G with GFP-SC35 also showed colocalization of Flag signals with that of GFP (Supplemental Figure S5B). Localization of C3G to nuclear speckles in response to amanitin treatment was not specific to a particular cell type, as it was also observed in HeLa and C2C12 cells (Supplemental Figure S6A, B). Endogenous C3G also showed colocalization in nuclear speckles with other ribonucleoproteins like U1 snRNP70, U2 snRNP B”, and Sm snRNP (Y12) on inhibition of transcription (Supplemental Figure S7, A–D). These results confirmed that C3G localizes to nuclear speckles.

When examined at higher magnification, C3G was distributed irregularly in speckles and appeared to be present only partially colocalized with SC35. Individual speckles were examined by capturing optical z sections at 0.30 microns. Three-dimenionsal reconstruction showed distribution of C3G nonuniformly in the speckle, being prominently seen in the periphery enveloping the core region where SC35 is primarily located. This was also evident from XY and XZ projections (Figure 1E, left panel). Quantitative assessment of fluorophore localization using a region of interest (ROI) was carried out to assess the spatial distribution of C3G and SC35 within speckles, which showed that intensity peaks do not correspond though the signals overlapped (Figure 1E). These data indicated that C3G localizes to discrete regions that show only partial overlap with SC35.

The dynamics of C3G localization to speckles was examined using a reversible inhibitor of transcription, 5, 6-dichloro-1-β-d-ribofuranosylbenzimidazole (DRB) (Sehgal ). C3G showed reversible association, similar to SC35, as its prominence in speckles was reduced 1 h after removal of DRB (DRB Rec) (Figure 2A). Since C3G dynamically exchanges between nuclear and cytoplasmic compartments, we examined whether the prominent association with speckles in response to α-amanitin treatment was due to enhanced nuclear translocation or retention. Levels of C3G in nuclear and cytoplasmic fractions were compared in untreated (UT) and amanitin-treated cells. Unlike LMB treatment (Shakyawar ), amanitin does not increase nuclear C3G levels, indicating that enhanced association with speckles was due to redistribution of nuclear C3G (Figure 2B).

FIGURE 2:

C3G shows reversible association with nuclear speckles. (A) Untreated MDA-MB-231 cells (UT) plated on coverslips were treated with DRB and either fixed immediately or after recovery from drug treatment for 1 h (DRB Rec). Cells were subjected to indirect immunofluorescence to detect C3G and SC35 expression. (B) MDA-MB-231 cells grown in the presence or absence of amanitin were subjected to cell fractionation. Lysates were processed for immunoblotting and probed for expression of indicated proteins. Calnexin and Lamin B1 were used to indicate the purity of the cytoplasmic and nuclear fractions, respectively. H5 antibody (RNA pol II pS2) was used to show efficacy of α-amanitin treatment. Numbers indicate ratio of C3G levels in nuclear vs. cytoplasmic fraction determined, based on protein loading. (C, D) Effect of heat shock on C3G localization. MDA-MB-231 cells plated on coverslips were grown at 37°C (UT) or subjected to heat shock (HS). Panel shows confocal images of cells immunostained with antibodies against C3G and SC35 (C) or C3G and U1 snRNP70 (D).

SC35 and snRNPs respond differently to heat shock (HS), which is known to inhibit transcription (Spector ; Lallena and Correas, 1997). We examined the localization of C3G in response to heat shock and observed that ∼40–50% of cells showed the presence of C3G in alternate nuclear structures that were irregular and very few in number per cell. These structures did not colocalize with SC35-positive speckles (Figure 2C). Colocalization of C3G and U1 snRNP70 was examined under similar conditions, and it was found that, unlike SC35, U1 snRNP70 partially colocalized with C3G in these alternate nuclear structures (Figure 2D).

The rounding up of speckles and clustering of splicing factors on inhibition of transcription is caused by their accumulation at storage sites, due to reduction in pre-mRNA levels. Reduction in splicing activity also results in similar changes in speckle morphology and number (O’keefe ). We examined localization of C3G in cells treated with isoginkgetin (IGK), a prespliceosome complex inhibitor (O’Brien ). Reversible association of C3G with speckles was seen on treatment of cells with IGK (Figure 3A), suggesting a role for C3G in splicing.

FIGURE 3:

C3G localization to speckles increases on inhibition of splicing and is lost on overexpression of Clk1. (A) MDA-MB-231 cells were either left untreated (UT) or treated with IGK. Cells were also allowed to recover for 6 h post drug washout (IGK-Rec). Cells were fixed and subjected to immunofluorescence with antibodies against C3G and SC35. Images of a single Z-section through the center of the nucleus acquired on a confocal microscope are shown in the panels and coefficient of colocalization between C3G and SC35 is shown in the bar diagram. ****p < 0.0001. (B) MDA-MB-231 cells expressing GFP-Clk1 or GFP-mClk1 construct were left untreated or subjected to α-amanitin treatment, fixed, and immunostained with antibody against C3G. Panels show confocal images of cells expressing C3G and transfected GFP tagged constructs. Arrows indicate GFP-expressing cells in the C3G panels.

Formation of nuclear speckles is dependent on phosphorylation of proteins by the kinase Clk1, and exogenous expression of Clk1 causes redistribution of SR proteins out of speckles (Colwill ). GFP-Clk1 and its catalytic mutant (GFP-mClk1) were expressed in MDA-MB cells and examined for C3G expression in normal and amanitin-treated cells. Just as described for other SR/SF proteins, expression of Clk1, but not its kinase inactive variant, resulted in loss of speckle pattern of staining for C3G in amanitin-treated cells (Figure 3B). These results suggested that localization of C3G to the speckles is dynamic, and coupled to their formation and loss, which can be modulated by altering Clk1 activity. Loss of speckled pattern of C3G in the nucleus on increasing Clk1 activity provides additional evidence that C3G localizes to nuclear speckles and not other structures that could also show a speckled pattern.

C3G is associated with nuclear speckles dependent on intact chromatin and RNA

Molecules generally localize to speckles through association with proteins or RNA (Lallena and Correas, 1997; Dye and Patton, 2001). The association of C3G with speckles was examined by carrying out detergent extraction as well as sensitivity to DNase I and RNase A treatments (Figure 4A). When AMA-treated MDA-MB cells were extracted with detergent prior to fixation, C3G as well as SC35 were intact. Treatment with DNase I alone did not alter speckle localization of C3G or SC35. When treatment with DNase I was followed by 0.4 M NaCl, SC35-positive intranuclear foci were intact, but C3G staining was greatly reduced as most of the chromatin was lost from the nucleus (Figure 4A). When digestion was followed by 2M NaCl treatment that removes weakly associated proteins and retains only the nuclear matrix, C3G was totally lost from speckles. SC35 foci were intact, indicating tight association with nuclear matrix. Removal of chromatin was confirmed by the loss of 4’,6-diamidino-2-phenylindole (DAPI) staining. Cells treated with 0.4 and 2M NaCl, without DNase treatment, had their DNA intact and did not lose speckle localization of C3G or SC35.

FIGURE 4:

C3G localization to nuclear speckles is dependent on the presence of intact chromatin and RNA. (A, B) Amanitin-treated MDA-MB-231 cells were left untreated or subjected to various treatments as indicated and described under Materials and Methods, fixed, and immunostained for C3G and SC35. Quantitation of the extent of colocalization between C3G and SC35 on each of these treatments is shown in the bar diagrams. NS, nonsignificant; ***p < 0.001; ****p < 0.0001. (C) MDA-MB-231 cells were treated with α-amanitin and also exposed to nocodazole or cytochalasin D for 4 h prior to fixation. Immunofluorescence was carried out to detect C3G and SC35.

On RNase A treatment, weak speckle localization of C3G was seen, with some diffused nuclear staining, though most of the RNA was lost from the cells (Figure 4B and Supplemental Figure S1C). When RNase treatment was followed by 0.4 and 2 M NaCl extraction, the foci formed by C3G were totally reduced, whereas SC35 foci were intact (Figure 4B). The efficacy of RNase A digestion was confirmed by the absence of staining with anti m3G antibody that labels capped RNAs (Supplemental Figure S1C). These results indicated that localization of C3G to speckles was dependent on the presence of intact chromatin and RNA in cells.

Localization of proteins to nuclear speckles and many nuclear functions are dependent on actin (Galganski ; Misu ). We examined the requirement of intact cellular microtubules and microfilaments for localization of C3G to speckles on amanitin treatment. C3G staining in large foci formed after α-amanitin treatment was partially reduced when MDA-MB cells were treated with nocodazole, an inhibitor of microtubule polymerization. There was no effect on SC35 foci. When treated with cytochalasin D (an inhibitor of actin polymerization), C3G as well as SC35 showed a weak diffuse pattern, without prominent localization to speckles (Figure 4C), indicating that nuclear actin dynamics regulates localization of C3G as well as SC35 to speckles. The efficacy of cytochalasin D and nocodazole treatments was verified by staining similarly treated parallel cover slips for F-actin and α-tubulin, respectively (unpublished data).

Rap1, a target of C3G localizes to speckles

Since C3G is a GEF for Ras family GTPases, we examined whether Rap1, a target of C3G, localizes to nuclear speckles. Localization of Rap1 to the nucleus has been shown earlier (Lafuente ). Endogenous Rap1 showed nonuniform staining throughout the cell when methanol-fixed cells were examined by immunofluorescence, similarly to that shown earlier (Bivona ), but in amanitin-treated cells, prominent localization was seen at nuclear speckles (Figure 5A). We examined localization of activated (GTP-bound) Rap1 by expressing a GFP fusion protein of Ral-GDS-RBD, which binds specifically to activated molecules of Rap1 in cells and can be observed as enhanced GFP signals (Bivona ). In methanol-fixed cells, prominent GFP fluorescence was seen at nuclear speckles, specifically in α-amanitin–treated cells, in addition to diffuse signals seen throughout the cell (Figure 5B). In untreated cells, intense GFP signals were seen in punctate cytoplasmic structures, which are likely to be endosomes/multivesicular bodies as described earlier (Pizon ). GFP-RalGDS-RBD was prominently nuclear in formaldehyde-fixed cells but showed increased colocalization with SC35 on amanitin treatment (Supplemental Figure S8). These results suggest a function for C3G and its target, Rap1, in nuclear speckles.

FIGURE 5:

Rap1 a target of C3G localizes to nuclear speckles. (A) MDA-MB-231 cells were treated with (AMA) or left untreated (UT) and immunostained for Rap1 and SC35. BL indicates secondary antibody control where cells were processed without addition of primary antibodies. Colocalization of Rap1 staining with that of SC35 is shown in the bar diagram. ***p < 0.001. (B) GFP-RalGDS-RBD transfected MDA-MB-231 cells were treated with or without α-amanitin, fixed with formaldehyde, and immunostained with SC35. (C) MDA-MB-231 cells transfected with GFP-Rap1GAP were treated with or without amanitin, fixed with methanol, and immunostained for expression of SC35. Arrows in SC35 panel show GFP-Rap1GAP-expressing cells. Bar diagram shows quantitation of number of speckles per nucleus in expressing and nonexpressing cells using data obtained from large number of cells from three independent experiments. ***p < 0.001.

This was further validated by examining SC35 speckles in cells expressing GFP-Rap1GAP, a protein known to inhibit Rap1 activation-dependent downstream signaling. We compared structure and number of SC35 speckles in MDA-MB cells expressing GFP-Rap1GAP (in normally growing and under conditions of transcription inhibition) with those that do not express GFP-­Rap1GAP. GFP-Rap1GAP–expressing cells show more compacted and significantly fewer speckles compared with nonexpressing cells (Figure 5C). Difference in speckle morphology and number were not seen under conditions of amanitin treatment, with both expressing and nonexpressing cells showing fewer and more rounded speckles. Similar results were obtained in HeLa cells expressing Rap1GAP-GFP. GFP-expressing cells were used as control and showed no difference in speckle number. The effect of Rap1GAP on speckles was not due to its effect on cell morphology, as Rap1GAP-expressing cells were not morphologically different from nonexpressing or GFP-expressing cells under the conditions of our experiments when stained with Rh-phalloidin (Supplemental Figure S9).

C3G is required for pre-mRNA splicing

Earlier work from our laboratory showed that C3G translocates to the nucleus in response to LiCl, an inhibitor of GSK3β (Shakyawar ). C3G showed enhanced localization to speckles in MDA-MB cells on LiCl treatment, which causes increased expression of many genes (Figure 6A). SC35 localization was also altered on LiCl treatment as shown earlier (Hernández ). Efficacy of LiCl treatment was tested by immunostaining for β-catenin, which shows nuclear translocation in response to LiCl treatment (unpublished data).

FIGURE 6:

C3G associates with nuclear speckles on myocyte differentiation and is required for splicing activity. (A) MDA-MB-231 cells grown on glass coverslips were subjected to LiCl treatment prior to fixation and immunostained for expression of C3G and SC35. Images show confocal sections of immunostained cells and quantitation of colocalization is shown in bar diagram. ***p < 0.001. (B) C2C12 cells were grown in the presence of growth medium (GM) or differentiation medium (DM) for 72 h, fixed, and immunostained for expression of C3G and SC35. The colocalization coefficient between C3G and SC35 in myoblasts and myotubes was quantitated using large numbers of cells from three independent experiments and is shown in the bar diagram. ***p < 0.001. (C) Immunoblot of lysates from control (Con) and C3G KO clone (KO), probed for the expression of C3G; actin was used as loading control. Numbers indicate relative amount of C3G in KO cells compared with control cells. (D) SC35 staining is altered in C3G KO cells. Control and C3G KO clones of C2C12 were treated with or without α-amanitin, fixed, and immunostained to detect C3G and SC35. Panels show images acquired on a confocal microscope. (E) Effect of C3G KO on cellular splicing activity. Control and C3G KO clones transiently transfected with pTN24 vector for 48 h and lysates were assayed for luciferase and β-gal activity. The bar diagram shows the relative luciferase-to-β-gal ratio averaged from three independent experiments. ***p < 0.001. (F) Schematic showing positions of constitutive and variable exons in CD44 pre-mRNA and the isoforms containing the v6 exon. Arrows show primers used to detect all endogenous CD44 isoforms and those that include c5-v6 exon. (G) Quantitation of the relative expression of total CD44 and c5-v6–containing variants in WT MDA-MB cells and knockout clones determined by real-time PCR. Isoginkgetin-treated cells were used as a control to detect variation in CD44 splice form generation under conditions of inhibiting splicing. ***p < 0.001.

Inhibition of GSK3β promotes myogenic differentiation and SC35 speckles show reorganization during myocyte differentiation, becoming more misshapen and larger (van der Velden ; Homma ). C3G is required for C2C12 differentiation and translocates to nuclei on differentiation of myocytes to form myotubes (Sasi Kumar ). Colocalization studies showed that C3G localization in SC35-positive speckles of myotubes is higher compared with that seen in undifferentiated cells (Figure 6B).

Splicing activity and structure of nuclear speckles changes during myocyte differentiation (Homma ). On the basis of its dynamic localization to speckles, we hypothesized a role for C3G in mRNA processing. We examined the role of cellular C3G in regulating activity in nuclear speckles by using a clone of C2C12 cells where C3G was knocked down using CRISPR/Cas9 technology (Figure 6C). Unlike control cells, these cells do not fuse to form myotubes when grown in differentiation medium (Shakyawar ). Compared to normal C2C12, cells with reduced cellular C3G showed very poor staining for SC35 speckles (Figure 6D). In amanitin-treated knockout (KO) cells, SC35 speckles were weakly stained and less distinct throughout the nucleus compared with wild-type (WT) cells.

We went on to determine whether loss of C3G impacted cellular splicing activity using a double reporter splicing assay plasmid, pTN24 (Nasim ). Ratio of the activities of the two reporter genes, luciferase and β-gal, gives information on the extent of splicing activity in cells. This plasmid was expressed in WT and C3G KO C2C12 cells and splicing efficiency determined. C3G KO cells showed significantly lower splicing activity, suggesting a role for C3G in regulating RNA splicing (Figure 6E).

The CD44 gene encodes a large number of variant proteins generated due to alternative splicing, which have functions in growth and motility of cells. Expression of some splice forms is associated with tumorigenesis (Screaton ; Afify ; Loh ). To investigate the role of C3G in regulating splicing of endogenous genes, we examined transcript levels of CD44c5-v6–containing isoforms in WT and C3G KO MDA-MB-231 cells by real-time PCR using isoform-specific primers (Figure 6F). KO clones that showed a 70% or more decrease in C3G RNA and protein levels were used. Expression of total CD44 mRNA, and variants with exon c5-v6, was tested. RNA from Isoginkgetin-treated cells was used as a control for inhibition of splicing activity. We observed an increase in expression of isoform with c5-v6 exons relative to total CD44 transcripts in KO clones compared with WT cells after normalizing with actin (Figure 6G). Isoginkgetin-treated cells also showed a significant increase in c5-v6–containing transcripts. These results suggest that C3G plays a role in regulating alternative splicing of cellular genes. Overexpression of C3G using a viral expression system did not alter splicing activity of CD44 compared with cells expressing control virus (Supplemental Figure S10). It was observed that infection with adenovirus itself altered CD44 splicing compared with uninfected cells.

The reduced levels of SC35 seen in C3G knockout cells could be either a consequence of inefficient assembly into speckles or due to changes in expression level of SC35. We therefore examined total protein levels of SC35 and a few other splicing factors in WT C2C12 cells and C3G KO cells. As shown in Figure 7A, expression of many of the tested splicing factors was significantly lower in cells lacking C3G. A similar reduction was also observed in C3G KO clones of MDA-MB-231 cells (generated using CRISPR/Cas9 technology), indicating that the effect was not restricted to a particular cell type (Figure 7B). These results suggested that cellular C3G functions to regulate expression of splicing factors.

FIGURE 7:

Altered expression of splicing factors in C3G knockout cells. (A) Lysates of C2C12 control and C3G KO clones were subjected to Western blotting and probed for C3G, U5 snRNP200, U5 snRNP116, and SC35. Actin was used as loading control. Numbers indicate relative protein levels in KO compared with control. (B) Whole-cell lysates of MDA-MB-231 control and C3G KO clones probed for the expression of indicated proteins. Bar diagram shows relative levels of the proteins averaged from three independent experiments. *p < 0.05; **p < 0.01; ***p < 0.001. (C) Transcript levels of various splicing factors in control and C3G knockout cells determined by real time PCR. ***p < 0.001. (D) Effect of C3G overexpression on splicing factor levels. MDA-MB-231 and C2C12 cells were transiently transfected with C3G-Flag construct for 30 h. Lysates were subjected to immunoblotting and probed for indicated proteins. (E) Effect of C3G overexpression on splicing activity. HEK-293T cells were cotransfected with pTN24 along with C3G-Flag or control vector. After 48 h, lysates were assayed for luciferase and β-gal activity. (F) Schematic figure depicting dynamic localization of C3G and its effector GTPase Rap1 to nuclear speckles. Inhibition of transcription and splicing as well as physiological stimuli enhance localization of C3G to SC35-positive speckles, whereas expression of active Clk1 disrupts amanitin-induced C3G localization to speckles. Active form of Rap1 is present in nuclear speckles, and inactivation of Rap1 mediated by expression of GFP-Rap1GAP resulted in altered morphology of speckles and significant decrease in number. Cells lacking C3G due to CRISPR/Cas9-mediated knockdown show reduced levels of splicing factors and altered splicing activity.

The transcript levels of some of these splicing factors in MDA-MB C3G KO cells were not significantly different when compared with WT cells (Figure 7C). Presuming that overexpression of C3G would have some consequence on splicing factor levels, we examined their levels in lysates of MDA-MB-231 and C2C12 cells overexpressing C3G. None of the proteins showed any significant change on C3G overexpression (Figure 7D). We also examined the effect of enhancing cellular levels of C3G on splicing activity and observed that coexpression of C3G with TN24 caused a significant increase in splicing activity (Figure 7E).

DISCUSSION

Molecules that exchange between the nuclear and cytoplasmic compartments regulate important processes in the nucleus. As a follow-up of our study identifying nuclear localization of C3G in response to certain physiological stimuli, we carried out a detailed investigation into nuclear functions of C3G. In this study, we identify a novel function for C3G in the nucleus, where it localizes to SC35-positive speckles and regulates splicing activity. Localization of C3G to speckles is a feature conserved in cells derived from various tissue types, as well as in murine and human cells.

On inhibition of nuclear export, C3G is retained in the nucleus, with prominent localization seen with SC35 speckles in several cell types. Just as in the case of a variety of other molecules involved in RNA splicing activity (Carmo-Fonseca ), C3G shows enhanced localization to speckles on inhibition of transcription or splicing. Cell fractionation indicated that inhibition of transcription did not particularly increase nuclear levels of C3G, suggesting that nuclear C3G exchanged between nucleoplasm and speckles. The reversibility of release from speckles seen on removal of transcription or splicing activity inhibition, similarly to other splicing factors, suggests a common mechanism of dynamic exchange between speckles and nucleoplasm. The fact that C3G is not prominently seen in speckles of exponentially growing cells indicates that it is in constant flux between nucleoplasm and speckles.

In response to heat shock, transcriptional reprogramming occurs, and general RNA transcription and splicing are also inhibited (Velichko ). It therefore came as a surprise that on heat shock, C3G did not localize to SC35-positive speckles and appears to be present in alternate granular structures in the nucleus that showed partial colocalization with U1 snRNP70. It is reported that a few splicing factors, but not SC35, relocalize to nuclear stress granules that are sites of HSF1 localization and activity on heat shock (Weighardt ; Chiodi ; Denegri ). Splicing factors are sequestered in speckles and made available in response to specific stimuli (Spector and Lamond, 2011). If C3G is moving to nuclear stress granules in response to heat shock, when enhanced transcription of chaperones takes place, then it is tempting to speculate that C3G enables gene specific transcription in response to heat stress.

Though C3G localizes to speckles, it does not show superimposition with SC35 and is present in a spatially heterogenous manner (more in peripheral regions, compared with SC35), suggestive of being involved in distinct functions within the speckle (Mintz and Spector, 2000). C3G is released from speckles on extraction of chromatin, or RNA from nuclei, and differs from SC35 with respect to its extractability from speckles. Removal of C3G from speckles on mild extraction conditions is in contrast to that seen for several other SR factors, suggesting that C3G is weakly associated with speckles. SC35 and snRNPs differ in their dependence on RNA for retention in speckles (Spector ). Some proteins, like PIP2 and DGK θ, are dependent on RNA, but not on DNA, for association with speckles (Osborne ; Tabellini ). Proteins in nuclear speckles are known to have specific sequence elements that enable their targeting by binding to RNA and other proteins (Cáceres ; Eilbracht and Schmidt-Zachmann, 2001). In the primary sequence of C3G, we could not identify any such features. It is possible that C3G resides in speckles through its ability to interact with other macromolecules present. Since C3G shows properties similar to some snRNPs with respect to extractability, it is possible that localization of C3G to speckles is mediated by snRNPs.

Activation of Clk1 causes phosphorylation of many speckle-localized proteins and disruption of the speckle. Loss of speckled pattern of C3G in the nucleus dependent on activity of Clk1 provided additional evidence for localization of C3G to speckles and not other nuclear structures. The possibility of C3G being a substrate of Clk1 is being explored as it is known that Cdk5 phosphorylates C3G (Utreras ). Since C3G does not possess an RS motif, it will be interesting to identify interacting partners of C3G that enable it to move in and out of speckles in response to Clk1-mediated phosphorylation.

Rap1, the small GTPase target of C3G, localizes to the nucleus and regulates gene expression (Lafuente ), but its association with nuclear speckles was not known. We show that active Rap1 is a component of nuclear speckles, and inactivation of Rap1-dependent signaling alters speckle number and morphology. These changes appear similar to the effect of transcriptional and splicing inhibitors, suggesting that RNA transcription and/or splicing may be dependent on Rap1 activity. Rap1 plays a role at cell junctions (Ando ) and many junctional proteins and those that regulate actin cytoskeleton localize to nuclear speckles (Saitoh ). C3G signals to cytoskeleton-dependent functions in the cell and associates with cytoskeletal elements (Martín-Encabo ; Radha ). Nuclear assembly and various nuclear functions are regulated by cytoskeletal organization (Krauss ). Monomeric actin and molecules that regulate actin dynamics are localized to nuclear speckles (Gieni and Hendzel, 2009; Belin ; Galganski ), and F-actin disruption results in reduction of speckle size and motility (Zhang ). Our results show that C3G localization to speckles is partially compromised on loss of intact MTs and is totally lost on loss of intact F-actin. In contrast, speckle localization of SC35 is not affected by MT disruption but is lost on actin disruption. These results throw light on how C3G and SC35 are dynamically exchanged between speckles and the nucleoplasm.

Inhibition of GSK3β induces import of C3G into the nucleus, where it localizes to nonheterochromatin domains (Shakyawar ). We observed that under these conditions, C3G localizes to speckles, suggesting that localization of nuclear C3G to speckles may be dependent on specific physiological stimuli. Since interchromatin granule/nuclear speckle domains serve as sites of transcriptional activity, it is possible that GSK3β signaling regulates nuclear speckle localization of C3G to regulate gene expression. This has been demonstrated earlier for SC35 wherein GSK3β caused enrichment of SC35 in nuclear speckles (Hernández ). The authors showed that GSK3β inhibition also alters Tau mRNA splicing. It is therefore possible that inhibition of GSK3β causes nuclear translocation of proteins involved in splicing and their localization to nuclear speckles. A very recent study showed that GSK3β phosphorylates multiple splicing factors and contributes to alternative splicing (Shinde ).

C3G shows enhanced localization to speckles in differentiated myotubes, and knockdown resulted in reduced intensity of SC35 speckles. The functional consequence of localization to speckles is a role for C3G in regulating splicing activity. While C3G overexpression enhanced splicing activity of an artificial construct that was coexpressed, it did not affect splicing of endogenous CD44 in cells infected with C3G adeno viral vector compared with control vector. In fact, it was seen that adenoviral infection itself caused altered splicing activity. Knockdown of C3G resulted in reduced splicing when examined using a splicing reporter construct. Loss of C3G affected the expression of splice forms of cellular CD44 containing variable exon c5-v6, suggesting that this variable exon is not spliced out. IGK-treated cells showed increased expression of the CD44 isoform. Earlier reports have shown exon skipping by inhibitors of splicing, suggesting that they can trigger gene-specific splicing alterations (Salton and Misteli, 2016). Expression of v6 exon–containing isoform is an indicator of metastatic potential of cells (Indinnimeo ; Su ). Lower C3G levels have been associated with cervical cancers (Okino ). It would therefore be interesting to learn whether C3G levels regulate the expression of CD44v6 to induce cell transformation.

Alternate splicing is an important mechanism that enables myogenic differentiation (Bland ). Our earlier work showed that C3G is required for C2C12 differentiation. Therefore, one of the required properties of C3G to enable myogenic differentiation may be its function in alternate splicing. We observed that cells lacking C3G show reduced presence of SC35 in speckles. C3G could be playing an adaptor function enabling other proteins home to speckles and also playing an independent catalytic function in regulating splicing activity. Surprisingly, examination of cell lysates having low levels of C3G due to CRISPR/Cas9-mediated knockdown showed dramatic decrease in several splicing factors, which could explain reduced splicing activity in these cells. This decrease was not due to lower transcript levels. The splicing factors regulated by C3G are ubiquitous components of speckles and essential for constitutive splicing. One likely explanation for reduction in several of these factors is proteasomal degradation specific to nuclear speckles (Baldin ). If C3G functions to inhibit this activity, then reduction of several cellular splicing factor protein levels would be expected. If C3G affects splicing activity in cells, then its loss could result in reduced levels of many proteins that are generated from spliced transcripts. These possibilities are being investigated. Though C3G overexpression did not alter protein levels of splicing factors, it did show a significant, but small, increase in splicing activity of an artificial construct. This could be because C3G is overexpressed in only 30–40% of cells.

On the basis of multiple lines of evidence, we describe a novel function for C3G in the nucleus. Localization to splicing factor-rich nuclear speckles; dynamic and reversible localization in response to cellular transcription and splicing activity; dependency of localization on intact DNA and RNA; presence of activated Rap1, a target of C3G in speckles; and its ability to regulate splicing enabled us to show that regulation of RNA splicing is an important function of nuclear C3G (Figure 7E). Ras family GTPases and exchange factors that activate them are involved in multiple signaling pathways regulating cell functions in response to external stimuli. Several of them are known to localize to the nucleus, but their functions in the nucleus are poorly understood. This is the first example of a Ras family GTPase and its exchange factor localizing to and functioning in nuclear speckles.

MATERIALS AND METHODS

Antibodies and reagents

Antibodies against C3G (H300, C19), Calnexin, NF-kB, Rap1, and GAPDH were purchased from Santa Cruz Biotechnology. Polyclonal antibody (C9) raised in our laboratory was used to detect C3G specifically (Radha ). Anti-SC35 (S4045-100UL) and anti-­snRNP200 (HPA029321) were from Sigma. Anti-SC35 (ab204916), anti-DHX38 (ab154801), and anti–Lamin B1 (ab16048) were from Abcam. Anti–Flag M2 monoclonal antibody (F3165) was from Sigma. RNA polymerase-II H5 (Warren ) and β-actin antibodies were from Berkeley Antibody Company and Millipore, respectively. U1 snRNP70 (Lerner and Steitz, 1979), U2 snRNP B″ (Habets ), and Sm snRNP(Y12) (Lerner ) antibodies were gifted by David Spector (Cold Spring Harbor Laboratory [CSHL], Cold Spring Harbor, NY). α-Amanitin, DRB, cytochalacin D, nocadazole, rhodamine-phalloidin, and LiCl were from Sigma. Leptomycin B was from Santa Cruz Biotechnology. Isoginkgetin was from Merck Millipore. Luciferase reagent was from Promega. Lipofectamine 3000 was from Invitrogen. Horseradish peroxidase–conjugated antibodies were from Amersham GE. Fluorophore conjugated secondary antibodies were from Millipore and Amersham GE.

Plasmid constructs

Expression vector for Flag-tagged human C3G was a kind gift from S. Tanaka, The Rockefeller University, New York (Tanaka ). Deletion constructs of C3G have been described earlier (Ichiba ). C3G-GFP expression construct has been described earlier (Dayma and Radha, 2011). GFP-SC35 construct was a kind gift from V. Parnaik, Centre for Cellular & Molecular Biology, Hyderabad, India (Tripathi and Parnaik, 2008), and pTN24 plasmid was gifted by I.C. Eperon, University of Leicester, U.K. GFP-Clk1 and GFP-mClk1(K190R) plasmids were gifted by David Spector (CSHL). GFP-RalGDS-RBD and GFP-Rap1GAP constructs were gifted by P. J. Stork, Oregon Health & Science University (Carey and Stork, 2002) and Patrick Casey, Duke University School of Medicine (Meng and Casey, 2002), respectively. Adenoviral vector expressing human C3G was generated using an AdEasy System provided by Bert Vogelstein (Howard Hughes Medical Institute).

Cell culture, transfections, and treatments

MDA-MB-231, MCF-7, HEK293T, and HeLa cells were cultured in DMEM with 10% fetal bovine serum (FBS). C2C12 cells were maintained in DMEM with 20% FBS under standard conditions, and differentiation was carried out as described earlier (Kumar ). Lipofectamine 3000 was used for transfections in cells as per manufacturers protocol. Treatments of α-amanitin, 50 µg/ml, 5 h; LMB, 37 nM, 6 h; nocodazole, 1 µg/ml, 4 h; cytochalasin D, 0.2 µg/ml, 4 h; IGK 50 µM, 15 h; LiCl, 50 mM, 24 h; and DRB 25µg/ml, 3 h, were given by incubating exponentially growing cells with respective drug(s) in cell culture medium. Transient heat shock treatment of cells was carried out in an incubator at 42°C for 2 h after addition of prewarmed medium.

Generation of C3G KO clones in C2C12 cells has been described earlier (Shakyawar ). A similar strategy was applied to generate C3G KO clones in MDA-MB-231 cells using commercially available CRISPR/Cas9 KO plasmid (sc-401616) and HDR plasmid (sc-401616-HDR).

Immunofluorescence, image analysis, and Western blotting

Immunofluorescence was performed as described earlier (Shivakrupa ). Costaining for two antigens was performed by sequential incubation of primary and corresponding secondary antibodies. Parallel cover slips processed without addition of primary antibodies were used as Blank (BL) to show absence of nonspecific staining from secondary antibodies. Confocal Z stacks were captured on a Leica TCS SP8 Confocal microscope (Leica Microsystems, Germany), and data were analyzed using the Leica Application Suite and ImageJ software’s. Constant image acquisition parameters were used for capturing images of all samples from a given experiment. All confocal images were captured under the 63× or 100× objectives of the microscope and were digitally processed for presentation using Adobe Photoshop CS6 software. Quantitation of SC35-positive nuclear speckles was carried out with ImageJ software. Images of cells were imported into ImageJ, and a digital brightness threshold was applied to each image. A particle analysis tool available in the ImageJ software was applied, and signal intensity of nuclear speckles measuring 0.3–10 µm2 in size were obtained as output. Pearson’s colocalization coefficient was obtained using the Coloc2 tool of ImageJ.

Western blotting was performed as described (Radha ). Vilber-Lourmat Chemiluminescence System (Germany) or Carestream XBT autoradiogram sheets were used for detection of enhanced chemiluminescence (ECL) signal. ImageJ software was used for quantitation of the Western blots, and values were normalized to loading control.

Cell fractionation and in situ extraction

Whole-cell (W), cytoplasmic (C), and nuclear (N) fractions of cells were prepared as described (Radha ). In situ extraction of cells plated on coverslips was performed as described (De Conto ) with minor modifications. Cells grown on coverslips were rinsed twice with TM buffer (50 mM Tris-HCl, pH 7.5, 3 mM MgCl2), followed by 10-min incubation on ice with TM buffer containing 0.4% Triton X-100, 0.5 mM CuCl2, 0.2 mM phenylmethylsulfonyl fluoride (PMSF), and protease inhibitor (detergent extraction buffer). Cells were washed and treated with DNase I (40 U/ml) or RNase A (20 μg/ml) for 30 min at 37°C. Cells were subjected to 0.4 and 2 M NaCl treatment sequentially for 5 min on ice to remove chromatin and RNA. Coverslips were fixed with methanol at each stage and immunostained.

Splicing activity assay

In vivo splicing activity assays was performed as described (Nasim ). pTN24 reporter construct encodes β-gal and luciferase, separated by an intronic sequence (derived from human αs-tropomyosin gene) that contains stop codons. The ratio of luciferase to β-gal reflects the splicing activity in cells. Cells were transfected with the dual reporter plasmid pTN24 in addition to indicated constructs. After 48 h of transfection, lysates were prepared using reporter lysis buffer and assayed for luciferase activity (Turner designs luminometer TD 20/20), and β-gal assay as per Promega protocols. Luciferase to β-gal ratios were normalized with control to determine relative in vivo splicing activity.

Splicing of endogenous genes in MDA-MB-231 cells was determined by detecting CD44 isoforms by real-time PCR as described by Loh . Primers were designed to detect total CD44 transcripts and forms with CD44c5-v6 exons. Ratio of CD44c5-v6 to total CD44 reflects specific exon inclusion/exclusion splicing activity in cells.

RNA isolation and real-time PCR

RNA was prepared using Trizol reagent (Invitrogen). cDNA was made using a Superscript first strand synthesis kit (Invitrogen). For quantitative real time PCR (RT-PCR), we performed PCR using Power SYBR Green PCR Master Mix (Applied Biosystems) with ViiA 7 Real Time PCR System (Applied Biosystems) according to the manufacturer’s instructions. Quantification data are represented as means ± SD of three independent experiments. Primer sets for quantitative RT–PCR were as follows: C3G forward 5′ TCCTCCTTCCGAGCCTAC 3′, and reverse 5′ CCACCGCTTGGAGAAGTT 3′; Actin forward 5′ CTTCCTTCCTGGGCATGGAG 3′ and reverse 5′ CTTCATTGTGCTGGGTGCCA 3′; SC35 forward 5′ CCCGATGTGGAGGGTATGAC 3′ and reverse 5′ GAGACTTCGAGCGGCTGTAG 3′; SNRNP200 forward 5′ CCAAGCTGACCGTTCTCTCAT 3′ and reverse GCCTTGTCTCCCATACGGG; U5 SNRNP116 forward 5′ CAATATCATGGACACTCCAGGAC 3′ and reverse 5′ CGGTCAATCTTGTTGATGCACA 3′; DHX38 forward 5′ TCCCTGAAACGGAGAGAGC 3′ and reverse 5′ CCCGAGCAGACCGATAATGT 3′; U2 SNRNP B” forward 5′ GGTCCCTGATTACCCTCCAAA 3′and 5′ ATCCCTGTAAAGCATCCCTGG 3′; U1 SNRNP70 forward 5′ GACGGGAAAAGATTGAGCGG 3′ and 5′ GCCACGAAGAGAGTCTTGAAGG 3′; CD44 total forward 5′ GAAGAAAGCCAGTGCGTCTC 3′and reverse 5′ GTGCTCTGCTGAGGCTGTAA 3′; CD44c5-v6 forward 5′ CAGGCAACTCCTAGTAGTAC 3′ and reverse 5′ TTTGCTCCACCTTCTTGACTCC 3′.

Click here for additional data file.