Rac and Cdc42 Are Potent Stimulators of E2F-dependent Transcription Capable of Promoting Retinoblastoma Susceptibility Gene Product Hyperphosphorylation*

The Rho family of GTPases plays an important and diverse role in reorganization of the actin cytoskeleton, transcriptional regulation, and multiple aspects of cell growth. Our study has examined their potential links to the cell cycle machinery. We find that constitutively active mutants of Rac and Cdc42, but not Rho, are potent inducers of E2F transcriptional activity in NIH 3T3 fibroblasts. Furthermore, activated Rac and Cdc42, but again not Rho, are capable of inducing cyclin D1 accumulation and pRB hyperphosphorylation in serum-deprived cells, outlining one route leading to enhanced E2F-mediated transcription. The inhibitory effect of the cyclin-dependent kinase inhibitors, p16 ink4 , p21 cip1 , and p27 cip on Rac/Cdc42-mediated E2F transcription corroborates a role for pRB family members and their functional inactivation by cyclin-dependent kinases in generating E2F activity. While the up-regulation of E2F transcriptional activity by Rac or Cdc42, not Rho, suffices for entry into S phase and DNA synthesis in Rat-1 R12 cells, this is clearly not the case in NIH 3T3, where additional requirements must exist.

The Rho family of GTPases currently comprises Rho (A, B, C), RhoE, RhoG, RhoL, Rac (1, 2), Cdc42Hs, and TC10. Several family members have been demonstrated to serve essential and specialized functions in actin cytoskeleton organization in response to extracellular growth factor signaling. As specific examples, it is known that platelet-derived growth factor induces membrane ruffles and lamellipodia via Rac, bradykinin induces filopodia formation through Cdc42, and lysophosphatidic acid induces stress fibers and focal adhesions mediated by Rho (1)(2)(3). The activated form of each of these GTPases elicits the same cytoskeletal rearrangements (1)(2)(3)(4). In Swiss 3T3 cells it was demonstrated by microinjection studies that Cdc42, Rac, and Rho cross-talk and may function in a more or less linear, hierarchical cascade: Cdc42 3 Rac 3 Rho, to regulate cytoskeletal rearrangement (4).
The function of this family of GTPases is, however, not limited to control of actin polymerization. All three members have been shown to play important roles in regulation of cell growth as well. First, they are weak oncogenes on their own. Various rodent fibroblast cell lines expressing constitutively active mu-tants of Rac, Rho, and Cdc42 display weak focus-forming activity, soft agar growth, and increased saturation density (5)(6)(7)(8)(9). Second, they have also been demonstrated to play a role in Ras-mediated transformation, since dominant negative mutants of Rho family GTPases inhibit Ras transformation (5)(6)(7)(8). Third, activated forms of Rac, Rho, and Cdc42 are capable of inducing DNA synthesis in quiescent Swiss 3T3 cells (10). In these cells, not only are they sufficient to induce G 1 progression into S phase, but they have also been demonstrated to be required for serum-mediated S phase entry, underscoring again their importance in normal growth control (10).
In addition to a role of Rho family GTPases in regulating actin polymerization and proliferation, substantial evidence indicates that they can also regulate gene expression. Rac, Cdc42, and Rho can, evidently via distinct pathways, induce serum response factor activation leading, in concert with ternary complex factor, to elevated transcription from a serum response element, present, e.g. in the promoter of the immediate early gene c-fos (11). Rac and Cdc42, but in most cell types not Rho, also activate the Jun N-terminal kinase and p38/ HOG1 kinase, which in turn leads to increased transcriptional activity of c-Jun and ATF-2 (12)(13)(14)(15).
The accumulating evidence implicating Rho GTPases in multiple aspects of cell proliferation suggests that some of their activities may be linked to the cell cycle. The mammalian cell cycle transition from G 1 to S is regulated at least in part through the phosphorylation and concomitant inactivation of a family of proteins termed the "pocket" proteins (reviewed in Weinberg (16)). These include p105 RB (the retinoblastoma protein), p107, and p130, which are phosphorylated in a temporal manner by a distinct set of complexes between cyclins and cyclin-dependent kinases (Cdks). 1 Especially cyclins D, E, and A in specific complexes with Cdk4, Cdk6, and Cdk2 have been implicated in the sequential phosphorylation of RB family members (reviewed in Weinberg (16) and Sherr (17)). Regulation of Cdk activity is fairly complex; in addition to the positive regulation through cyclin-association, negative regulation is carried out by a family of Cdk inhibitors. These can be divided into specific inhibitors (e.g. p16 ink4 ) targeting cyclin D/Cdk4(6) complexes, and broad spectrum inhibitors (e.g. p21 cip1 and p27 kip1 ) which inhibit all cyclin/Cdk complexes (reviewed in Sherr and Roberts (18)). One important consequence of the phosphorylation of pocket proteins is the concomitant release of transcription factors of the E2F family, the activity of which appears to be essential for subsequent G 1 /S progression, and which leads to induction of genes associated with DNA replication (19). At least in some cell systems, ectopic overexpression of selected members of the E2F family is capable of inducing quiescent cells to enter into DNA synthesis (20 -22). However, relatively little is known about how canonical signal transduction cascades interface with the general cell cycle machinery as outlined above, and the link between Rho family GTPases and the cell cycle has so far remained elusive. This study demonstrates that although activated Rac and Rho induce neither DNA synthesis nor foci in NIH 3T3 clone 7 cells, Rac (and Cdc42) are potent inducers of E2F transcriptional activity in NIH 3T3 fibroblasts, while Rho is not. Rac and Cdc42 also converge on the RB pathway, since their activated forms can induce pRB hyperphosphorylation, and their ability to up-regulate E2F-dependent transcription can be inhibited by the specific Cdk inhibitor p16 ink4 .

EXPERIMENTAL PROCEDURES
Cell Culture and Transfection-NIH 3T3 clone 7 was cultured in Dulbecco's modified Eagle's medium (DMEM) with 10% fetal calf serum (FCS), while the UNC line of NIH 3T3 (23) (kindly provided by C. Der, University of North Carolina, Chapel Hill, NC) was grown in DMEM containing 10% donor calf serum and ATCC NIH 3T3 propagated in DMEM with 10% newborn calf serum. R12 cells are a subclone of Rat-1 cells stably expressing the tet-VP16 transactivator (24) (kind gift of D. Resnitzky and S. Reed). These cells were routinely cultured in DMEM including 10% FCS.
Transfections for reporter assays were generally done by the calcium phosphate co-precipitation method according to Chen and Okayama (25). One g of reporter and 2 g of pEXV, or 0.2 g of CMV-based, GTPase expression plasmid was used for each 60-mm dish, except when examining the effect of Cdk inhibitors, where 2 g of pEXV Rho family expression vector was used together with 1 g of Cdk inhibitor expression vector. The same total amount of DNA was always applied to cells and wherever required, the empty vector pRK5, was used to normalize the DNA dose. All reporter transfections were carried out in duplicate, and a minimum of two independent DNA preparations were compared to ensure reproducibility. For reporter assays, the calcium phosphate precipitate was left for 18 -24 h, and after washing twice with phosphate-buffered saline the medium was replaced with DMEM containing 0.3% FCS (for clone 7) or 0.5% donor calf serum (for UNC) for another 48 h. UNC NIH 3T3 cells were electroporated using a Genepulser II apparatus (Bio-Rad) with 10 g of expression plasmid in a volume of 50 l of phosphate-buffered saline and applying the settings 0.27 kV, 125 F, and 1000 ohm.
Microinjection, BrdUrd Labeling, and Cell Staining-R12 cells were brought to quiescence by incubation for 48 h in serum-free DMEM and subsequently microinjected directly into the cell nuclei using a Zeiss AIS microinjection system essentially as described previously and employing 25 g/ml expression plasmid for each Rho family member or 100 g/ml pGL3 6xE2F reporter plasmid (26). Cells positively expressing the microinjected DNA were verified either by indirect immunofluorescence of the expressed protein or by co-injecting rabbit IgG (1 mg/ml) as a microinjection marker.
When assaying S phase entry, the injected cells were evaluated 30 -36 h following the injection. BrdUrd (100 M) was added for the last 24 h, after which the cells were subjected to double-immunofluorescence analysis visualizing both incorporated BrdUrd as well as expression of the protein encoded by the microinjected plasmid. This double staining, as well as in situ staining for luciferase, was carried out essentially as described previously (26).
Plasmids-The expression plasmids pEXV3 Rac1, pEXV3 Rac1V12, pEXV3 RhoAV14, pRK5 RacL61, pRK5 RacL61 40C, and pRK5 RacL61 37A (kindly provided by A. Hall) all encode Myc -tagged proteins, and the plasmids have been previously described (1,3,27). The plasmids pRK5 and pRK5 Cdc42V12, of which the latter directs expression of a Myc-tagged version of Cdc42V12 from the CMV promoter, were obtained from B. Schaffhausen. The Rho expression vector pCMV5 HA-RhoL63 was generously donated by M. Schwartz. Long terminal repeat expression vectors for RhoL63 and Cdc42V12 (pZIPNeoSV(X)) were provided by C. Der. Oncogenic H-Ras (D12) was expressed from the plasmid pEXV3 RasD12 again in a Myc-tagged form (10); it was kindly provided by H. Paterson. CMV v-ras encoding the H-Ras protein carrying the individually activating G12R and A59T mutations, was constructed as follows. The previously described murine sarcoma virus long terminal repeat-driven v-H-Ras expression plasmid pBW1423 (28) was digested with BamHI to excise the v-H-ras cDNA, which was cloned into the BamHI site of the pCMVNeoBam expression vector (29), generating CMV v-ras.
The E2F-responsive reporter plasmid pGL3-TATA-6xE2F (referred to as pGL3 6xE2F) contains six E2F sites upstream of a minimal TATA box and a luciferase gene. It was generously donated by A. Fattaey, and has been previously evaluated in terms of E2F responsiveness (26). All plasmids were purified by two consecutive bandings in a CsCl gradient according to standard procedures.
No internal standard was used in the transfections, since all promoters we tested (giving measurable activity) responded to Ras or Rac expression to varying extents. We obtained consistent data using several plasmid constructions based on different promoters or different activating mutations, several plasmid preparations of these, and monitored protein concentration for yield in the cell extracts as well as expression of the tagged, exogenous protein by blotting.
Luciferase Assays-Luciferase assays were performed essentially according to the manufacturer's protocol (Promega). Briefly cells on 60-mm dishes were incubated with 400 l of 1ϫ reporter lysis buffer (Promega) for 15 min, scraped, and transferred into an Eppendorf tube, and after a high speed centrifugation, the supernatant was transferred to a new tube. A total of 20 l of supernatant was mixed with 100 l of luciferase assay reagent (Promega), and after a 15-s incubation, the emitted light was measured for a 5-s period on a Berthold Lumat LB 9501 luminometer.
Protein Extraction and Western Blotting-For preparation of lysates, the cells were washed twice in phosphate-buffered saline and scraped off with a cell scraper, and the cells were collected by low speed centrifugation. The cell pellet was extracted for 15 min on ice with a lysis buffer containing 50 mM Tris base, 150 mM sodium chloride, 0.5% Nonidet P-40 supplemented with the phosphatase and protease inhibitors 1.0 mM sodium fluoride, 0.1 mM sodium orthovanadate, 10 mM ␤-glycerophosphate, and 0.1 mM phenylmethylsulfonyl fluoride, 5 g/ml leupeptin. Subsequently extracts were spun in an Eppendorf microcentrifuge for 5 min, and the supernatant lysates were transferred to an equal volume of 2ϫ SDS loading buffer (5% SDS, 25% glycerol, 0.0625 M Tris, pH 6.8, 0.0075% bromphenol blue, 0.7 M ␤-mercaptoethanol), after which the samples were boiled for 3 min.
Western blotting was done according to standard procedures. Equal amounts of total protein, as measured by the Bio-Rad protein assay (Bio-Rad), was loaded in each lane. Approximately 20 g of total protein were loaded when analyzing the pRB phosphorylation state. The signal on Western blots was visualized with enhanced chemiluminescence (ECL or ECL plus) reagents according to the manufacturer (Amersham Life Science).
Thymidine Incorporation-LipofectAMINE (Life Technologies, Inc.)transfected NIH 3T3 clone 7 cells were incubated with DNA for 5 h, serum was added, and the cells were further incubated for 19 h. Medium was changed to 0.3% FCS, and after 24 h of serum starvation, [H]thymidine (final concentration, 3 M, 0.29 Ci/mmol) was added; the control received serum to 9%. After 24 h of incubation, cell extracts were made, and reporter activity and incorporated thymidine (measured after trichloroacetic acid precipitation and collection on glass fiber filters) were determined.

Activated Rac and Cdc42, but Not Rho, Stimulate E2F-mediated Transcription in NIH 3T3-Since the Rho family of
GTPases has been implicated in numerous aspects of cell proliferation, we wanted to examine whether they can directly regulate components of the cell cycle machinery, in particular the E2F family of transcription factors that play a pivotal role in controlling G 1 progression into S phase. In order to measure directly the total E2F-dependent transcriptional activity, we chose a previously characterized and evaluated synthetic reporter construct, pGL3 6xE2F, containing six E2F sites immediately upstream of a minimal TATA box and a luciferase reporter gene (kind gift from A. Fattaey, described further in Lukas et al. (26)). This particular reporter construct provides a sensitive and specific probe for quantitatively measuring E2F transcriptional activity in transiently transfected cells. Expression vectors pEXV RacV12, pRK5 RacL61, pRK5 Cdc42V12, pEXV RhoV14, pCMV5 RhoL63, pEXV RasD12, and CMV v-ras encoding constitutively active versions of Rac, Cdc42, Rho, and Ras were cotransfected with reporter by calcium phosphate precipitation into NIH 3T3 (clone 7) murine fibroblasts, and after transfection the cells were subjected to serum starvation for 48 h in DMEM containing 0.3% FCS. Importantly, we found that activated Rac, Cdc42, and Ras potently stimulated transcription from E2F sites, but activated Rho displayed at most marginal induction of E2F activity (Fig. 1). The two sets of expression vectors, driven by either CMV or the SV40 early promoter/enhancer, and encoding two distinct activated variants of each GTPase, all produced a consistent induction of E2F activity for Rac, Cdc42, and Ras, but none for Rho (Fig. 1). In addition, the long terminal repeat expression constructs, pZIPNeoSV(X) Cdc42V12 and pZIPNeoSV(X) RhoL63, gave as expected significant E2F activation for Cdc42 but none for Rho (data not shown). The nonactivated, cellular version of Rac displayed only a very slight activation of E2F-dependent transcription, emphasizing the importance of GTP-loading on Rac in triggering downstream pathways leading to E2F-mediated transcription (Fig. 1).
The failure for activated Rho to up-regulate E2F transcription was not because of substantially lower levels of expression, since Western blotting with the monoclonal anti-Myc epitope tag antibody demonstrates that RacV12, Cdc42V12, RhoV14, and RasD12 are expressed at comparable levels ( Fig. 1). A dose-response curve for each construct had established the optimal dose of expression vector for reporter assays, and none of the doses of expression construct within a wide range revealed any significant E2F transcriptional activation when Rho was examined. It was observed that especially Cdc42V12 at high dose when expressed from the CMV vector displayed some toxicity, since the fold induction of E2F activity dropped substantially at high dose (data not shown). This is consistent with previous reports that activated, GTPase-defective mutants of Cdc42, when overexpressed in NIH 3T3 cells can be cytotoxic, and therefore preclude the isolation of stable cell lines, perhaps through the G 1 arrest caused by a downstream Cdc42 effector, the p38 kinase (32,33).
A number of controls were performed to further validate the E2F reporter assay outlined in Fig. 1. The corresponding parental construct lacking all six E2F sites showed extremely low activity, but no activation with Rac or Cdc42 (data not shown). In addition, the ability of activated Rac or Cdc42 to promote E2F-mediated transcription was confirmed in two distinct NIH 3T3 cell lines (ATCC and UNC) with only minor differences, reflecting individual degrees of starvability and type of serum used for their cultivation. Furthermore, the method of gene transfer, whether using calcium phosphate co-precipitation, lipofection, or electroporation, did not significantly affect the results, but produced quantitatively similar responses for each GTPase examined.
RacV12 and Cdc42V12 Induce pRB Hyperphosphorylation and Moderate Cyclin D1 and p21 cip1 Accumulation-Because activation of E2F-mediated transcription often is a result of dissociation from inhibitory complexes with pocket proteins due to phosphorylation of these by Cdks, we investigated if expression of activated Rho family members in NIH 3T3 cells can influence this phosphorylation event. In order to obtain the maximum signal, we chose a particular line of NIH 3T3 cells (UNC) which can be electroporated to high transfection efficiency. The high frequency of expressors after transient transfection of these cells by electroporation allowed us to directly analyze the phosphorylation status of endogenous pRB protein by electrophoretic mobility shifts on SDS-polyacrylamide gel electrophoresis gels. Subsequent to electroporation, the cells were starved in 0.2% donor calf serum for 24 h, inducing the accumulation of pRB in its hypophosphorylated form, thus highlighting any potential effects on pRB phosphorylation. Clearly, the expression of RacV12, Cdc42V12, or oncogenic Ras induced the appearance of the hyperphosphorylated form of pRB (Fig. 2), which has often been correlated with inactivation of its growth suppressive function. In contrast, RhoV14 displayed at most a slight induction of pRB hyperphosphorylation. Rac in its nonactivated form failed to induce the appearance of hyperphosphorylated pRB, just as in the E2F reporter assays, stressing the importance of GTP loading. Collectively, these data support the finding that Rac, Cdc42, and Ras in their activated form, but not Rho, can promote E2F transcription.
Using the same transfection protocol, we also investigated if Rho family members can trigger the accumulation of cyclin D1 protein, which, in complex with Cdk4 or Cdk6, is considered a physiological pRB kinase and to be a key mediator for the release of transcriptionally active E2F in response to several growth factors. As demonstrated in Fig. 2 by immunoblotting   FIG. 1. Induction of E2F transcriptional activity by Rac and Cdc42. Upper, expression vectors for either wild type or various activated versions of Rac (pEXV Rac, pEXV RacV12, pRK5 RacL61), Cdc42 (pRK5 Cdc42V12), Rho (pEXV RhoV14, pCMV5 RhoL63), Ras (pEXV RasD12) or an empty expression vector (pRK5) were cotransfected with the E2F-responsive reporter construct pGL3 6xE2F into NIH 3T3 clone 7. Subsequently, the cells were serum-starved for 48 h, and luciferase reporter activity was measured. The bars represent the mean of two determinations using two independently prepared DNAs. At least two additional experiments showed similar fold activation. Lower, cell lysates from NIH 3T3 cells transiently transfected by electroporation were subjected to Western blotting analysis in order to compare expression levels. The blot was probed with antibody 9E10 reactive toward the Myc epitope tag. Comparable amounts of total protein were loaded in each lane.
with cyclin D1-specific antibody, Rac and Cdc42 when activated, but again not Rho, caused a moderate accumulation of cyclin D1 protein compared with the potent induction observed with v-H-Ras, and the almost undetectable level in serumdeprived NIH 3T3 fibroblasts. The up-regulation of cyclin D1 protein elicited by oncogenic Ras has been reported from several investigators (34,35), and serves as a control to further validate our assay system. Since oncogenic Ras at high expression levels also has been shown to up-regulate p21 cip (36), we examined whether this was the case for Rho family members. Western blot analysis demonstrated that RacV12 and Cdc42V12, but not RhoV14, both are capable of moderately up-regulating p21 cip1 levels, whereas v-Ras is a much stronger inducer (Fig. 2).
RacV12-mediated E2F Transcriptional Activation Is Abrogated by p16 ink4 , p21 cip1 , and p27 kip1 -Since pRB phosphorylation and its concomitant functional inactivation is generally believed to be mediated by Cdks, in particular complexes between D-type cyclins and Cdk4/6, we examined if a specific inhibitor of these complexes, p16 ink4 had any effect on the released free E2F activity. The results of reporter assays in NIH 3T3 clone 7 fibroblasts demonstrate that RacV12 activation of E2F-mediated transcription is sensitive to p16 ink4 , as seen by a substantial reduction in transactivation potential when p16 ink4 is coexpressed (Fig. 3). A loss-of-function mutant of p16 ink4 , P114L, caused only a slight reduction of RacV12 stimulation of E2F activity, demonstrating that the p16 ink4 abrogating effect is dependent on a functional inhibitor capable of binding Cdk4/6 ( Fig. 3). We also tested if the broad spectrum Cdk inhibitors, p27 kip1 and p21 cip1 could affect Rac-mediated induction of E2F-dependent transcription, and observed that they reduced RacV12-mediated E2F transcription as much as, or for p21 cip1 even more than, p16 ink4 . Western blot analysis of Rac showed no modulation of Rac expression by the inhibitors (Fig. 3). We found similar inhibitory effects of p16 ink4 , p21 cip1 , and p27 kip1 on Cdc42V12-mediated E2F transcriptional activation as for RacV12 (data not shown).
In Rat-1 Cells, but Not NIH 3T3, RacV12-and Cdc42V12mediated E2F Up-regulation Is Associated with S Phase Entry Inhibitable by p16 ink4 -In its ultimate consequence, E2F transcriptional activation may culminate in S phase entry and DNA replication (20). We examined this possibility in NIH 3T3 cells and one other well characterized, pRB-positive fibroblast line, Rat-1 R12. The latter cell line has the additional advantage that it tolerates rigorous serum deprivation causing the majority of the cells to accumulate in G 0 as measured by BrdUrd incorporation (26). When we tested the ability of Rho family members to induce DNA synthesis from quiescence in NIH 3T3 using combined microinjection of expression plasmid and subsequent BrdUrd labeling, it became clear that RacV12, Cdc42V12, and RhoV14 all failed to significantly promote DNA synthesis ( Fig. 4D and data not shown) Each Rho family member was nevertheless abundantly expressed (Fig. 4A, bottom  panel, and data not shown), and the cells displayed all the morphological features characteristic of each GTPase (1,3,4). For example, after microinjection, activated Rac gave the expected membrane ruffling response. Consistent with the microinjection data in NIH 3T3, we found that two-parameter flow FIG. 2. Activated Rac, Cdc42, and Ras promote pRB hyperphosphorylation, accompanied by cyclin D1 and p21 cip1 accumulation. Top, UNC NIH 3T3 cells were electroporated with expression plasmids encoding activated Rac, Cdc42, Rho, or Ras, after which they were serum-starved for 24 h. Lysates were prepared and analyzed by SDS-polyacrylamide gel electrophoresis and Western blotting, which was followed by probing with an pRB-specific antibody to visualize differentially phosphorylated species of endogenous pRB. Equal amounts of protein were loaded in each lane. Similar results were obtained in three independent experiments. Middle, the same lysates described in the top panel were subjected to Western blotting analysis using a cyclin D1-specific monoclonal antibody to detect endogenous cyclin D1 protein. Bottom, the lysates described in the top panel were analyzed by immunoblotting with p21 cip1 -specific antiserum.

FIG. 3. Cyclin-dependent kinase inhibitors interfere with Racmediated E2F stimulation.
Upper, reporter assays were carried out as described in the legend for Fig. 1. The pEXV RacV12 expression vector (or vector control, pRK5) were transfected with plasmids coding for either wild type p16 ink4 , a loss-of-function mutant of p16 ink4 (P114L), p21 cip1 , or p27 kip1 , and E2F-dependent luciferase activity was measured with the pGL3 6xE2F reporter. Lower, Rac expression in the samples was verified by Western blotting for the Myc tag.
cytometry analysis under conditions perfectly matching the E2F transactivation assays failed to reveal any substantial cell cycle effects of any Rho family member even under serum deprivation, since ⌬G between control and Rho-family expressing cells always was less than 5%. Compatible with our observations, we and others have failed to induce significant DNA synthesis upon microinjection of an E2F-1 expression plasmid into NIH 3T3 cells (Fig. 4D) (37), 2 although the same dose of this plasmid induces massive BrdUrd incorporation in other cell systems, e.g. R12 (Fig. 4D).
To verify the observation that activated Rac induces E2F activity without inducing DNA synthesis in NIH 3T3 cells, transiently transfected cells were investigated for extent of thymidine incorporation compared with E2F-directed luciferase activity. Serum stimulation, as well as transfection with activated Ras, induced thymidine incorporation compared with the vector control. RacV12 in contrast, although inducing luciferase, showed background incorporation and activated Rho induced neither (Fig. 5). In these cells, no focus forming activity of the non-Ras GTPases was detectable: focus forming activities were lower than 0.3-1% of that of v-Ras (Fig. 5, inset).
In sharp contrast, we found that both activated Rac and Cdc42, but significantly not Rho, are capable of efficiently stimulating DNA synthesis in quiescent R12 (Fig. 4, B and D, and data not shown), as reported originally for Rac, Cdc42, as well as Rho in Swiss 3T3. The induction of DNA synthesis by RacV12 or Cdc42V12 could be efficiently blocked by coexpression of the specific Cdk inhibitor, p16 ink4 (data not shown), which is in perfect agreement with our data from reporter assays in NIH 3T3. We observed abundant expression and correct localization of Rac, Cdc42 and Rho by cell staining with the Myc epitope tag antibody (Fig. 4A, top panel, and data not shown), and consistent morphological alterations. We subsequently sought to determine whether like in NIH 3T3, expression of RacV12/Cdc42V12 influences the transcriptional activity of endogenous E2F. We chose a microinjection approach and in situ analysis because of the extremely poor efficiency of gene transfer by conventional techniques in R12 cells. By coinjection of the previously described E2F-responsive reporter, pGL3 6xE2F together with Rho family expression plasmid into quiescent R12 cells and staining for luciferase with the appropriate antibody, we observed that RacV12 and Cdc42V12, but not RhoV14, yield significant in situ activation of the E2F reporter ( Fig. 4C and data not shown). After injection with reporter alone, the R12 fibroblasts displayed only a very low level of background luciferase staining in few isolated cells, reflecting most likely the cells that failed to exit the cell cycle under the starvation conditions. Thus the in situ E2F transactivation in R12 clearly parallels the E2F stimulation demonstrated by reporter assays in NIH 3T3, corroborating the main message, that activated Rac and Cdc42, but not Rho, are efficient inducers of E2F activity.

DISCUSSION
Although an increasing number of signal transduction pathways are being characterized and the molecular details and intricate complexity of the cell cycle is better understood, the interface between canonical signal transduction pathways and cell cycle regulation remains poorly described. Clearly, there must be mechanisms whereby signal transduction cascades operating from cell surface receptors toward transcriptional events in the nucleus, can communicate with the intrinsic cell FIG. 4. RacV12 is capable of inducing E2F transcriptional activity and S phase entry in R12 cells. A, membrane ruffles formed in R12 and NIH 3T3 clone 7 cells following microinjection of RacV12 expression plasmid (25 g/ml) into quiescent cells. Cells were fixed 30 h after injection and immunostained with 9E10 (anti-Myc) monoclonal antibody. B, RacV12 induces BrdUrd incorporation in Rat-1 R12 fibroblasts. Quiescent R12 cells were microinjected with RacV12 expression plasmid (25 g/ml; left panels) or empty vector (25 g/ml; right panels) together with an injection marker (nonimmune purified rabbit IgG, 1 mg/ml). 36 h after injection, the injected cells were revealed by anti-IgG immunostaining and assayed for BrdUrd incorporation. Arrows in both panels point to identical cells within the field. C, RacV12 activates the E2F-responsive reporter construct in a microinjection assay. Quiescent R12 cells were injected with RacV12 expression plasmid (25 g/ml; top panel) or empty vector (25 g/ml, bottom panel) together with pGL3 6xE2F reporter plasmid (100 g/ml). 24 h after injection, the injected areas were re-found by means of grid coordinate system and immunostained with purified anti-luciferase antibody. Arrows in the top panel point to the cells with significant induction of luciferase. D, quantitation of the ability of RacV12 and E2F-1 to initiate S phase entry in R12 and NIH 3T3 clone 7 cells. RacV12 and control (empty vector) plasmids were injected and assayed essentially as described in panel B. E2F-1 expression plasmid (10 g/ml) was injected into quiescent cells and assayed for BrdUrd incorporation 20 -24 h after injection.
FIG. 5. RacV12 expression fails to induce DNA synthesis in NIH 3T3 cells. Expression vectors for the indicated activated GTPases or empty expression vector (pRK5, Vector, and Serum) were cotransfected with the E2F-responsive reporter construct pGL3 6xE2F into NIH 3T3 clone 7. Cells were starved for 24 h and [ 3 H]thymidine was added; one set (Serum) received serum as well. After a further incubation of 24 h, luciferase as well as 3 H incorporation into macromolecules were determined. Inset, focus-forming activity was measured as described previously (54). The v-Ras gave 177 foci per 1000 G418 colonies, and 13,900 G418 colonies per g of plasmid.
cycle machinery involving modulators of Cdk activity, and induce pocket protein phosphorylation, E2F transcriptional activity and, when appropriately cued, ultimately DNA replication. While there is a plethora of published evidence implicating especially Rac, but on occasion also Cdc42 and Rho, in multiple aspects of cellular mitogenic signaling, the actual links to cell cycle have remained unclear (5-10, 38, 39). We have addressed this issue and uncovered some of the links between this family of small GTPases and cell cycle progression. We demonstrate that activated Rac and Cdc42, but not Rho, can stimulate E2F-mediated transcription which is believed to be an essential step in induction of appropriate E2F target genes required for S phase entry and DNA replication. Qualitatively similar results were obtained in luciferase reporter assays in three different NIH 3T3 cell lines as with in situ cell staining in R12 cells using luciferase antibody and the same E2F reporter. This result establishes that the Rac/Cdc42mediated stimulation of E2F activity is not restricted to NIH 3T3 cells, but also occurs in at least one other cell type. Since the phenotypes of RacV12 expression and E2F overexpression are similar, notably anchorage-independent growth and enhanced saturation density (5,7,40), we suggest that E2F stimulation contributes to the oncogenic potential of activated Rac. The ability of activated Rac and Cdc42 to up-regulate E2F transcriptional activity is consistent with published reports for the related small GTPase Ras, which, in its oncogenic form, also induces E2F-dependent transcription (41,42) (Fig. 1). For Ras it is generally assumed that a significant contribution to its mitogenic potential comes from its ability to transcriptionally up-regulate cyclin D1 (34,35,41,(43)(44)(45). Ectopic expression of cyclin D1 accelerates G 1 progression (46), and it bypasses the need for Ras in proliferation of exponentially growing fibroblasts (47). In this study, we demonstrate that also Rac and Cdc42 are capable of inducing cyclin D1 accumulation, albeit less potently than v-H-Ras. This is concordant with the published data indicating that Rac can up-regulate transcription from the cyclin D1 promoter (48) (our data not shown). Whether this induction of accumulation of cyclin D1 is the mediator of the Rac and Cdc42 effects on proliferation awaits further investigation; however, the coordinated induction of cyclin D1 and p21 cip1 , molecules with opposing effects, makes it possible that a modest cyclin D1 induction, in a low p21 cip1 environment, is biologically significant. The ability of Rac to activate p65 pak correlates with its cyclin D1 promoter activation in NIH 3T3 (48), however, neither p65 pak nor Jun Nterminal kinase activation correlates with the ability of Rac to induce G 1 progression in Swiss 3T3 (27). This suggests that Rac, and probably Cdc42 as well, at least in some cell types, almost certainly have additional means of inducing S phase entry beyond those associated with cyclin D1 up-regulation. The participation of cyclin D1, as well as other speculative targets for Rac and Cdc42 in generating E2F activity, is outlined in the model in Fig. 6.
As for most or all previously characterized mitogenic stimuli (30), we show here that Rac and Cdc42 also converge on a pathway leading to functional inactivation of the pRB tumor suppressor. Application of our high efficiency transfection protocol utilizing NIH 3T3 fibroblasts indicates that Rac and Cdc42 in their activated form are capable of inducing pRB hyperphosphorylation, a step presumed to be required for release of active E2F protein and subsequent cell cycle progression (16,17). Their common strategy for targeting of pRB is confirmed both by the ability of the specific Cdk inhibitor p16 ink4 to significantly reduce their capacity to activate transcription from E2F sites, as well as by their up-regulation of cyclin D1 protein, one key partner for the physiological pRB kinases Cdk4/6. The v-H-Ras-induced hyperphosphorylation of endogenous pRB in our transient transfection assays is similar to that seen with inducible systems (35,49). We find that Rac-induced DNA synthesis in R12 cells is inhibited by p16 ink4 , as was seen for Ras induction of DNA synthesis in REF52 cells (50). Thus our data support the emerging central significance of pRB phosphorylation as a prerequisite for completion of G 1 and eventually S phase entry.
While we conclude that Rac and Cdc42, but not Rho, are entirely sufficient to promote S phase entry and DNA synthesis from quiescence in R12 cells in a similar manner as it was reported originally for Rac, Cdc42, and Rho in Swiss 3T3 cells, they all three notably fail to induce DNA synthesis as well as focus formation in NIH 3T3 cells, despite their ability to generate high E2F activity. Apparently, E2F activity is a critical requirement for S phase, but may not always suffice. Indeed, there are at least four independent reports showing that E2F transcriptional activity not always triggers S phase entry. First, dominant negative Cdk2 can prevent induction of DNA synthesis by Simian virus 40 large T antigen despite E2F transcriptional activity (51). Second, v-Mos-expressing cells when serum-deprived maintain transcriptionally active E2F, while actually undergoing G 1 arrest (52). Third, a certain subline of NIH 3T3 is arrested by the v-abl oncogene despite having phosphorylated pocket proteins, and the arrest cannot be overcome by E2F-1 (53). Evidence indicates that in this case only cyclin A can rescue the cells from growth arrest. Fourth, microinjection of E2F-1 expression plasmid was found to be insufficient to promote DNA synthesis in NIH 3T3 cells (although activating at least some E2F target genes), while it was able to do so in other cell lines (37) (Fig. 4D).
In many cell systems, it appears that E2F-1 is sufficient to cause S phase entry from quiescence (20 -22) (Fig. 4D). It remains unknown what particular genetic feature makes NIH 3T3 cells at least partially refractory to E2F-1 stimuli, but we speculate that they have excessive levels of one or more Cdk inhibitors, notably those able to inhibit post-E2F G 1 Cdk activities such as cyclin E/Cdk2 and/or cyclin A/Cdk2. The downregulation of these putative Cdk inhibitors must then require distinct mitogenic stimuli. Oncogenic Ras, while mitogenic in a FIG. 6. Model that shows how Rac and Cdc42 are connected to key regulators of the cell cycle. The arrows from Rac/Cdc42 to other pathways resulting in Cdk activation are speculative, since we do not yet have any evidence for their connection with other positive modulators of Cdk activity than cyclin D1 but suspect such may exist. Enhanced Cdk activity could for example arise from down-regulation of a Cdk inhibitor such as p27 kip1 , dephosphorylation of inhibitory Cdk phopshotyrosines via Cdc25 or phosphorylation of essential threonines via Cdk-activating kinase. The arrow from E2F activation to G 1 progression/DNA synthesis denotes that E2F-1 in some cell types is sufficient for initiation of DNA replication, while in other cell systems, there must be yet unknown additional requirements, depicted with the question mark. majority of cell systems, causes growth arrest in Schwann cells, and even so in NIH 3T3 cells at high dose, possibly as a consequence of Ras-mediated up-regulation of p21 cip1 levels (36). It is conceivable, given the great degree of cell type specificity in determining whether E2F transcriptional activation is accompanied by S phase, that the differential responses to Rho, i.e. DNA synthesis in Swiss 3T3 versus neither E2F activation nor DNA synthesis in R12 and NIH 3T3, may also be attributable to cell type variation. Since the Rho family of GTPases conceivably is involved in signaling from receptors via kinase cascades to transcriptional targets in the nucleus, it will be of particular interest to identify potential stimuli, or cellular processes if any, that physiologically act to engage their cell cycle modulatory activities.