Cdc42 stimulates RNA splicing via the S6 kinase and a novel S6 kinase target, the nuclear cap-binding complex.

Cdc42 is a low molecular weight GTP-binding protein that plays a key regulatory role in a variety of cellular activities. The importance of the coordination of different cell functions by Cdc42 is underscored by the fact that a constitutively active Cdc42 mutant induces cellular transformation. In this study, we describe a novel function for Cdc42: its ability to stimulate pre-messenger RNA splicing. This activity is dependent on cysteine 37 in the effector loop of Cdc42 but is not dependent on cell growth. A likely candidate protein for mediating the Cdc42 effects on pre-mRNA splicing is the nuclear RNA cap-binding complex (CBC), which plays a key role in an early step of cap-dependent RNA splicing. Activation of the CBC by Cdc42 can be inhibited by rapamycin. Additionally, phosphatidylinositol 3-kinase and the Cdc42 effector, pp70 S6 kinase, stimulate the RNA cap-binding activity of the CBC. S6 kinase may directly target the CBC in vivo as it can phosphorylate the 80-kDa subunit of the CBC, CBP80, at residues that are subject to a growth factor-dependent and rapamycin-sensitive phosphorylation in vivo. Together these data suggest the involvement of a Cdc42-S6 kinase pathway in the regulation of RNA splicing, mediated by an increase in capped RNA binding by the CBC, as well as raise the possibility that the effects of Cdc42 on cell growth may be due in part to its regulation of RNA processing.

Cdc42 is a low molecular weight GTP-binding protein that coordinates multiple cellular events by a regulated binding and hydrolysis of GTP. Transcriptional regulation (1)(2)(3)(4)(5), vesicular trafficking (6), actin cytoskeletal arrangement and cell cycle progression (7,8) all represent activities in which Cdc42 has been implicated. Additionally, it has been shown that Cdc42 plays an important role in cell growth control, as a constitutively active form of Cdc42, Cdc42 F28L, induces malignant transformation (9). We have recently shown that an interaction between Cdc42 and the coatomer subunits of COP-coated ves-icles is somehow involved in the transformation signal (6). However, there appears to be a requirement for other Cdc42 targets as well, as deletion of the insert region of Cdc42 blocks transformation (10) independent of any interaction with the coatomer complex. What this suggests is that Cdc42 may be orchestrating distinct cellular functions in order to coordinate regulated cell growth.
In a previous study, we observed that activated forms of Cdc42 were capable of stimulating the nuclear cap-binding complex (CBC) 1 to bind capped RNAs (11). In binding to RNAs containing a m 7 G cap structure, the CBC has been implicated in a number of fundamental aspects of RNA processing including pre-mRNA splicing (12)(13)(14)(15), U snRNA export (16), and polyadenylation (17). Importantly, the CBC has also been shown to be a nuclear end point for extracellular stimuli such as growth factors and cell stress stimulants (11). In this study we have investigated the possibility that Cdc42 might be involved in the signal regulation of RNA processing via the CBC. Specifically, we find that activated alleles of Cdc42 can stimulate splicing activity in a manner that is dependent on amino acid 37 of the effector loop. As the CBC plays a pivotal role in cap-dependent pre-mRNA splicing, we have investigated the Cdc42 pathway leading to CBC activation and find that this pathway proceeds via the Cdc42 effector, pp70 S6 kinase (S6K). Rapamycin inhibits the Cdc42-induced activation of the CBC, whereas S6K, together with PI3 kinase, will induce the CBC to bind capped RNAs. Additionally, S6K can phosphorylate the CBC in vitro at a site that undergoes a rapamycin-sensitive, growth factor-dependent phosphorylation in vivo. Together, these data suggest a mechanism by which Cdc42 might influence pre-mRNA splicing, through the S6K-mediated activation of the CBC.
Photoaffinity Labeling and in Vitro Kinase Assays-CBC cap-binding activity was assayed by measuring the incorporation of [␣-32 P]GTP into CBP20 by UV cross-linking as established previously (11). For in vitro kinase assays, S6K was incubated with CBC as indicated in the text in a buffer containing 10 mM Hepes, pH 8.0, 20 M ATP, 5 mM MgCl 2 , and 10 Ci/sample [␥-32 P]ATP for various times at room temperature.
Pre-mRNA Splicing Reactions-Splicing extracts were prepared from serum-starved (1% calf serum for 24 h) NIH 3T3 cells stably expressing different Cdc42 constructs, and splicing reactions were then carried out as described previously (11).

RESULTS AND DISCUSSION
Previously, we have demonstrated that activated forms of Cdc42 can promote the nuclear CBC to bind capped RNA substrates (11). One of the important functional outcomes of this binding event is an enhancement of capped RNA splicing (12). This leads to the intriguing suggestion that Cdc42-stimulated signaling pathways can promote cap-dependent RNA splicing. To test this possibility, splicing lysates were prepared from NIH 3T3 cells that were stably transfected with a constitutively active form of Cdc42, Cdc42 F28L (9), or vector alone, after a 24-h growth period in low serum. The activity of these splicing lysates was then assessed in an in vitro splicing assay using either a specific m 7 GpppG-capped probe or a nonspecific ApppG-capped probe. As shown in Fig. 1A, Cdc42 F28L facilitates the generation of RNA splicing products as compared with vector alone, and this effect is greatly enhanced when the RNA probe contains a specific m 7 GpppG cap.
We have shown that the Cdc42 F28L mutant causes the transformation of NIH 3T3 cells (9), raising the possibility that Cdc42 F28L-mediated effects on splicing are not a direct outcome of a signaling pathway initiated by Cdc42 but rather reflect the fact that cells expressing Cdc42 F28L are able to grow in low serum. Therefore, we sought a Cdc42 mutant that would uncouple RNA splicing and cell growth. Splicing experiments were performed using lysates from NIH 3T3 cells that stably expressed either Cdc42 F28L or two effector loop mutants of Cdc42 in the activated background, Cdc42 F28L,C37A and Cdc42 F28L,Y40C (Fig. 1B). In contrast to Cdc42 F28L, neither of the effector loop mutants will support the growth of cells in low serum (data not shown). The Cdc42 F28L,Y40C double mutant activates the splicing machinery, whereas Cdc42 F28L,C37A does not (Fig. 1B). Thus, the activation of splicing by Cdc42 is not dependent on cell growth. Furthermore, this mutational analysis suggests that the Cdc42 targets known to utilize position 40 in the effector loop, such as the CRIB (Cdc42/Rac-interactive binding) domain-containing proteins PAK, ACK, and WASP, are not involved in signaling to Cdc42 Stimulates RNA Splicing 37308 the splicing machinery. The identification of Cdc42 effectors for which position 37 is necessary has not been described to date, but it would appear that an effector of this class plays a pivotal role in the Cdc42-RNA processing pathway.
Given that the CBC is a downstream target for Cdc42, it seems likely that the regulation of CBC activity underlies the effects of Cdc42 on pre-mRNA splicing. To further explore how signals are conveyed from Cdc42 to the CBC, we tested the abilities of two small molecule inhibitors, rapamycin and wortmannin, to inhibit the activation of the CBC by Cdc42. These drugs inhibit FRAP and PI3 kinase, respectively. HeLa cells were transiently transfected with Cdc42 F28L, and 30 min prior to harvesting, either rapamycin or wortmannin was added. The cells were fractionated and the nuclear fractions were assayed for CBC activity by measuring the photocatalyzed incorporation of [␣-32 P]GTP into CBP20 as has been described previously (11). Rapamycin reduced CBC activity in the presence of Cdc42 F28L, whereas wortmannin did not ( Fig.  2A) suggesting that FRAP, but not PI3 kinase, is downstream of Cdc42 in the CBC pathway.
FRAP is thought to play a pivotal role in signaling pathways that regulate mRNA translation and is well documented as an upstream activator of S6K (18 -21). This suggested that Cdc42 might signal to the CBC via a FRAP-S6K pathway. However, PI3 kinase is also an upstream activator of S6K (22), but wortmannin will not block the Cdc42 activation of the CBC. To further assess the role of PI3 kinase and/or S6K in the activation of the CBC, COS-7 cells were transiently transfected with wild type S6K, a constitutively active form of PI3 kinase, the myristoylated 110-kDa subunit, or the two in tandem. As shown in Fig. 2B, wild type S6K did not show a detectable activation of the CBC. In contrast, constitutively active PI3 kinase will activate the CBC, and this activation is enhanced when S6K is coexpressed with PI3 kinase. Taken together, these data show that although PI3 kinase is not necessary for the Cdc42-mediated activation of the CBC, it can signal to the CBC, probably through its ability to stimulate S6 kinase activity. The ability of Cdc42 to stimulate S6 kinase activity, independent of PI3 kinase activation, thus represents the most likely pathway by which Cdc42 regulates the CBC.
The 80-kDa subunit of the CBC, CBP80, contains two potential S6K phosphorylation consensus sites within its N terminus in a region that coincides with the bipartite NLS (designated NLS1 and NLS2, see Fig. 4C). Thus we examined the ability of S6K to phosphorylate CBP80 in vitro (Fig. 3, A and B). S6K phosphorylated CBP80 to a level similar to that observed when equimolar amounts of GST-S6 were substituted for the CBC as the substrate in the assay (data not shown). To determine whether the phosphorylation occurred within the NLS region, an NLS mutant of CBP80 was examined. This mutant, CBP80 K17A,R18A, has been shown previously to inhibit the nuclear localization of CBP80 (23) and would no longer contain the S6K consensus site at NLS2. Indeed, this mutant fails to serve as a substrate for S6K phosphorylation (Fig. 3C).
Thus we are presented with an interesting possibility for how signaling from Cdc42 to the CBC proceeds, via a direct S6K-catalyzed phosphorylation of CBP80. Previously, we had

FIG. 4. CBP80 undergoes a growth factor-responsive and rapamycin-sensitive phosphorylation within the N-terminal NLS in vivo.
A, PC12 cells were metabolically labeled with [ 32 P]orthophosphoric acid for 3 h and then stimulated with NGF (100 ng/ml) for 15 min. CBP80 was then isolated by immunoprecipitation, and its phosphorylation state was assessed by autoradiography. B, PC12 cells were metabolically labeled for 3 h with [ 32 P]orthophosphoric acid. Thirty minutes prior to NGF treatment, 50 ng/ml rapamycin (Rap) was added as indicated. The cells were harvested and fractionated, and the cytosolic and nuclear lysates were immunoprecipitated using a specific CBP80 antiserum. Immunoprecipitated proteins were detected by autoradiography. C, N-terminal CBP80 sequence. The bipartite NLS is underlined and labeled as NLS1 or NLS2, and alanine mutations were made as shown. D, wild type and mutant CBP80 constructs were transfected into HeLa cells, and the proteins were expressed for 24 h. The cells were metabolically labeled with [ 32 P]orthophosphoric acid for 3 h and treated with heregulin for 15 min. Lysates were prepared, and the transfected CBP80 proteins were isolated using an ␣-KT 3 antibody. The immunoprecipitated proteins were separated by SDS-PAGE, and CBP80 was detected by Western blotting using a specific CBP80 antibody (top panel) and by autoradiography (bottom panel).

Cdc42 Stimulates RNA Splicing 37309
shown that the CBC is strongly activated by NGF in PC12 cells and by heregulin in HeLa cells (11). Therefore, PC12 cells were metabolically labeled with [ 32 P]orthophosphoric acid in the presence or absence of NGF. The labeled cells were fractionated, and the CBC was immunoprecipitated from cytosolic and nuclear lysates using a CBP80 antiserum or preimmune serum. Fig. 4A shows that nuclear CBP80 is phosphorylated in a NGF-dependent manner in cells. Phosphorylation of CBP20 was not observed. Additionally, the NGF-dependent phosphorylation of CBP80 was blocked by a 30-min pretreatment of cells with rapamycin (Fig. 4B). N-terminal alanine mutant constructs of CBP80 were generated to confirm that the in vivo phosphorylation occurs within the S6K phosphorylation consensus sites (Fig. 4C). These CBP80 constructs were transiently transfected into HeLa cells that were then metabolically labeled and treated with heregulin to promote CBC activation. The mutant constructs were isolated from cell lysates and their phosphorylation states were assessed. The in vivo phosphorylation of CBP80 was reduced in the S7A and the T21A,S22A mutants and completely blocked in the CBP80 S7A,T21A,S22A mutant (Fig. 4D). All mutant constructs showed wild type nuclear localization as assessed by immunofluorescence (data not shown). Thus, the in vivo phosphorylation of CBP80 is consistent with a direct phosphorylation by S6K.
Finding that CBP80 is phosphorylated at two positions in cells was unexpected based on our in vitro phosphorylation experiments with S6K and the CBC. Phosphorylation of CBP80 by S6K may be restricted to the NLS2 site, and another, yet to be identified kinase could be responsible for the phosphorylation at NLS1. The CBC is nuclear in the steady state, however, it has been suggested that the CBC accompanies the RNA particle through the nuclear pore (24), and releases its cargo in the cytosol to the cytosolic cap-binding protein eIF-4E before rapidly shuttling back to the nucleus. Although mainly cytosolic, the S6K has also been reported in the nucleus, both as a longer isoform containing a NLS sequence (25)(26)(27) and as a kinase capable of the nuclear phosphorylation of the transcription factor, CREM (28). Given that the phosphorylation is occurring within the NLS of CBP80, it might be anticipated that a cytosolic phosphorylation would block importin binding to the CBC and thus nuclear localization. Such has been found to be true for diacylglycerol kinase (29). However, we have not found phosphorylation of CBP80 by S6K to block an interaction with the importins in vitro nor have we observed that aspartic acid mutants, designed to mimic a phosphorylated CBP80, lose their nuclear localization. 2 Thus, at present both cytosolic and nuclear phosphorylations of CBP80 by S6K remain a viable mechanism.
S6K has been previously identified as an effector for Cdc42 (30) although this is the first report to suggest a functional consequence of this pathway, RNA splicing. FRAP, in addition to activating S6K, participates in the growth factor-regulated activation of the translation initiation complex, eIF-4F via the phosphorylation of 4E-BP1 (31)(32)(33). It is interesting that we have identified a potentially new role for these translational signaling players in CBC activation and RNA processing. This observation suggests a necessary coordination among aspects of gene expression that require the m 7 G cap structure: cap-dependent RNA splicing, export, and translation.
The work reported here also raises some new possibilities regarding how Cdc42 controls cell growth. It has recently been suggested that the involvement of Cdc42 in vesicular trafficking through coatomer interactions is necessary but not sufficient for Cdc42-induced transformation (6). The fact that the Cdc42 F28L,C37A double mutant is both transformation-defective and unable to stimulate RNA splicing suggests an interesting connection between the ability of Cdc42 to mediate RNA processing and induce malignant transformation. Thus, we are currently examining the possibility that CBC mutants, such as phosphorylation-defective CBC, will block Cdc42-induced transformation. Such data would underscore both the importance of Cdc42 in coordinating diverse cellular functions including RNA splicing and the role of CBC-mediated RNA processing in regulated cell growth.