Rac3-Mediated Transformation Requires Multiple Effector Pathways

Our initial characterization of Rac3, a close relative of the small GTPase Rac1, established its ability to promote membrane ruffling, transformation, and activation of c-jun transcriptional activity. The finding that Rac3 is transforming, and its similarity to Rac1, a protein that has a well-established connection to many processes important for cancer progression, prompted further investigation into Rac3 transformation. We used effector domain mutants (EDMs) to explore the relationship among Rac signaling, transformation, and effector usage. All Rac3 EDMs tested (N26D, F37L, Y40C, and N43D) retained the ability to promote membrane ruffling and focus formation. In contrast, only the N43D mutant promoted anchorage independence. This differs from Rac1, where both N26D and N43D mutants were impaired in both types of transformation. To learn more about the signaling pathways involved, we did luciferase reporter assays and glutathione S-transferase pull-down assays for effector binding. We found evidence for a functional link between activation of phospholipase Cβ₂ by Rac3 and signaling to the serum response factor (SRF). Surprisingly, we also found that Rac3 binds poorly to the known Rac1 effectors mixed lineage kinases 2 and 3 (MLK2 and MLK3). Transcription of cyclin D1 was the only pathway that correlated with growth in soft agar. Our experiments show that activation of membrane ruffling and transcriptional activation of c-jun, SRF, or E2F are not sufficient to promote anchorage-independent growth mediated by Rac3. Instead, multiple effector pathways are required for Rac3 transformation, and these overlap partially but not completely with those used by Rac1.

Rac1 is one of the most heavily studied members of the Rho subfamily of Ras-related small GTPases (1). Like Ras proteins, Rho family proteins cycle between active GTP-bound and inactive GDP-bound states. Oncogenic, activated forms of Rac proteins are either constitutively GTP bound (G12V or Q61L) or have greatly increased rates of GDP-GTP cycling (F28L). Activated forms of Rac proteins, whether through the action of guanine nucleotide exchange factors or mutation, promote their biological effects by signaling through an ever-growing list of downstream effector molecules (2, 3).

Cytoskeletal reorganization, specifically the formation of membrane ruffles and lamellipodia, was one of the first functions attributed to activated Rac1 (4). Since then, another major role defined for Rac1 involves coordinating many processes related to the development and progression of cancer (5). Rac1 is essential for progression of the cell cycle and activates several pathways known to be important in cellular proliferation, such as serum response factor (SRF), cyclin D1, and E2F (6, 7). Rac1 also promotes cellular survival by activating the nuclear factor-κB pathway and by preventing anoikis and apoptosis (8–10). Not surprisingly, due to its role in promoting cell cycle progression and survival, Rac1 also promotes transformation in rodent fibroblast models and is required for transformation induced by Ras (11–14). Due to its role in modulating the actin cytoskeleton, Rac1 also promotes motility and invasion, activities that are important for tumor metastasis (15). Given these activities, Rac1 has become a target for inhibitors that may be useful in future cancer therapies (16, 17).

Rac3, a close relative of Rac1, is less well characterized. Rac3 has 92% overall amino acid identity with Rac1, with the majority of the differences occurring in the COOH-terminal membrane targeting region and in regions surrounding and within the Rho insert domain (18, 19). Rac3 has been mapped to chromosome band 17q25.3 near a region that is commonly deleted in breast and ovarian cancers, suggesting possible transcriptional deregulation (20). It has been suggested that Rac3 but not Rac1 is hyperactivated in a subset of breast cancer cells and that inhibition of Rac3 impairs proliferation of these cells (21). In addition, a recent report suggests that Rac3 plays a growth stimulatory role in mouse models of Bcr/Abl-mediated leukemias (22). We have shown previously that Rac3 can promote membrane ruffling, transcriptional activity of c-jun, and cellular transformation in both focus formation and soft agar assays (16). Others have shown that Rac3 can bind and activate the Rac1 effector phospholipase Cβ₂ (PLCβ₂) and promote the activation of p21-activated kinase-1 (Pak1) and c-Jun NH₂-terminal kinase (JNK; refs. 18, 21, 23). Evidence that Rac1 and Rac3 share some of the same effector pathways coupled with the demonstration that Rac3 is transforming in rodent fibroblast models and plays a growth-promoting role in mouse leukemia models suggest that Rac3 may have a role in human cancer. This led us to further investigate the effector binding and biological properties of Rac3 that contribute to the transformation process.

Consistent with its diverse cellular functions, a multitude of effector molecules have been identified that interact with Rac1. These effector molecules fall into several functional classes, such as serine/threonine kinases, lipid kinases, and adaptors/scaffolds, and generally interact with the GTP-bound form of the protein (2, 3). The core effector region (amino acids 26-45) of small GTPases is conserved among Rac family members and serves as the primary site for interaction with effectors (2, 3, 18). Mutations in this region, effector domain mutants (EDMs), were first developed for Ras proteins and have helped to uncouple certain effector pathways, such as Ras activation of Raf, from biological outcomes, such as transformation (24, 25). This has led to a greater understanding of how effectors of Ras contribute to its biology but has also highlighted the need for multiplicity and cooperation among effectors to recapitulate the complex biological effects of the native protein. Following the example of Ras EDMs, mutations in the core effector region of Rho family proteins, such as Rac1 and RhoA, have also been characterized (12, 26).

For Rac1, EDMs were shown to selectively impair signaling pathways, such as transformation (N26D or N43D mutants), membrane ruffling (F37L mutant), and activation of Pak1 (Y40C mutant; ref. 12). We made these same mutations in a Rac3 background to better understand how the effector binding and signaling properties of Rac3 contribute to cellular transformation. In this study, we define a novel pathway linking the Rac effector PLCβ₂ and SRF-mediated transcription. We have also found that transcription of cyclin D1 may be important for promoting anchorage-independent growth but that membrane ruffling and transcriptional activation of c-jun, SRF, or E2F are not sufficient. These results underscore the complexity of Rac3 signaling to effectors necessary for transformation.

Effector domain mutants of Rac3. The mutations N26D, F37L, Y40C, and N43D were introduced into rac3 Q61L or rac3 F28L with the QuickChange Site-Directed Mutagenesis kit (Stratagene, La Jolla, CA) according to the manufacturer's instructions; the F28L mutation was also introduced in this way. The Nucleic Acids Core Facility at the University of North Carolina at Chapel Hill generated all primers. The rac1 and rac3 Q61L constructs have been described previously (16). Restriction enzyme digestion sites were added to the 5′ and 3′ ends of the rac3 EDMs by PCR-based mutagenesis as described previously (16, 27). PCR products of rac3 were digested with SalI and NotI (all restriction enzymes from Invitrogen, Carlsbad, CA) at the sites introduced at the 5′ and 3′ ends and ligated in-frame with the hemagglutinin (HA) epitope tag into the vector pCGN-hyg for expression in mammalian cells. PCR products were also digested with BamHI and EcoRI and ligated in-frame with an enhanced green fluorescent protein (EGFP) tag into the mammalian expression vector pEGFP-C1 (BD Biosciences Clontech, Palo Alto, CA) that was cut previously with BglII and EcoRI or in-frame with a glutathione S-transferase (GST) tag in the bacterial expression vector pGEX-2T (Amersham Biosciences, Piscataway, NJ). All mutations and ligation junctions were verified by forward and reverse sequencing.

Cell culture and transfections. NIH 3T3 mouse fibroblasts were grown in DMEM (Invitrogen/Life Technologies, Gaithersburg, MD) supplemented with 10% calf serum (Invitrogen/Life Technologies) and 1% penicillin/streptomycin (Invitrogen/Life Technologies) and maintained in 10% CO₂ at 37°C. The day before transfection, 5 × 10⁵ (100-mm dish), 2.5 × 10⁵ (60-mm dish), 1 × 10⁵ (per well of a six-well plate), or 3 × 10⁴ (per well of a 12-well plate) cells were plated unless otherwise indicated. Rat intestinal epithelial cells (RIE-1) and COS-7 cells were grown in 5% fetal bovine serum (FBS; Sigma-Aldrich, St. Louis, MO) or 10% FBS, respectively, and 1% penicillin/streptomycin and maintained as described for NIH 3T3 cells. NIH 3T3 cells were transfected by calcium phosphate coprecipitation as described (27) or with LipofectAMINE and PLUS reagent (Invitrogen) or FuGENE6 (Roche, Indianapolis, IN) according to the manufacturer's instructions. COS-7 cells were transfected with FuGENE6 and RIE-1 cells were transfected with LipofectAMINE and PLUS reagent according to the manufacturer's instructions. Stable cell lines were created in RIE-1 and NIH 3T3 cells after transfection with 1 μg pCGN-hyg constructs expressing the Rac3 F28L EDMs as has been described previously (16).

Membrane ruffling activity. NIH 3T3 cells were plated on glass coverslips in six-well plates. For visualization of membrane ruffles, cells were transiently transfected with 1 μg pEGFP-C1 vector, pEGFP-rac3 61L, or pEGFP-rac3 61L EDMs. After 48 hours, localization of Rac3 in live cells was visualized with a Zeiss Axioskop fluorescent microscope (Zeiss, Thornwood, NY). For each condition, at least 200 cells were counted per coverslip and scored as either ruffling or flat. Images were captured under the ×20 objective with MetaMorph digital imaging software (Universal Imaging Corp., Downington, PA).

Transformation assays. For focus formation assays, NIH 3T3 cells were plated in 60-mm dishes and cotransfected in duplicate (calcium phosphate) with either 200 ng pZIP vector or pZIP-raf 22W (NH₂-terminally truncated and active) and 500 ng pCGN-hyg vector, pCGN-rac1/rac3 61L, or pCGN-rac3 61L EDMs. After 14 to 21 days, cells were photographed under the ×4 objective and stained as described previously (16). Stained foci were then counted to determine transforming activity and sorted by focus morphology.

For soft agar assays, NIH 3T3 or RIE-1 cells stably expressing pCGN-hyg vector, pCGN-rac3 28L, or pCGN-rac3 28L EDMs were prepared as described above. Single-cell suspensions (1-5 × 10⁵ cells per 60-mm dish) of each stable cell line were plated in medium containing 0.4% agar on top of a bottom layer containing 0.6% agar. Colonies were allowed to form for 14 to 21 days after which they were photographed under the ×4 objective and counted.

Western immunoblotting. Stable cell lines and transiently transfected cells were harvested 2 days after plating or transfection, respectively, as described previously (16). Samples were prepared in Laemmli sample buffer and 5 to 10 μg protein from each sample were run on 12% SDS-PAGE gels. Proteins were transferred at 100 V for 1 hour to polyvinylidene difluoride (Immobilon-P, Millipore, Bedford, MA). Membranes were blocked in 3% fish gelatin (Sigma, St. Louis, MO), incubated for 1 hour in either 1:1,000 anti-HA antibody (Covance, Philadelphia, PA) or 1:5,000 anti-β actin (Sigma), and washed. Membranes were incubated for 30 minutes in 1:30,000 anti-mouse IgG-horseradish peroxidase (IgG-HRP, Amersham Biosciences), washed, and developed with SuperSignal West Dura Extended Duration substrate (Pierce, Rockford, IL).

Effector pull-down assays. Activated Rac1, Rac3 (both 61L), and Rac3 EDMs attached to a GST tag were purified from a bacterial culture as described previously (28). The concentration of protein bound to glutathione-agarose beads (Sigma) was determined by colorimetric assay (Bio-Rad, Hercules, CA). NIH 3T3 cells in 100-mm dishes were transfected with 3 μg constructs expressing the following Rac effector proteins: pCMV6-Myc-pak1, pcDNA3-HA-mlk3, pBABE-T7-par6 (described previously; ref. 28), pcDNA3.1⁺-plcβ₂, or pcDNA3.1-HA-mlk2 (gift of Lloyd Greene, Columbia University, New York, NY), and harvested 24 to 48 hours later. Cell lysates were collected as described previously (28) and precleared with 50 μg GST beads. Lysates were split and mixed with 10 to 30 μg (depending on effector used) of GST-only, GST-Rac1 61L, GST-Rac3 61L, or GST-Rac3 61L EDMs bound to beads. After incubation for 1 hour at 4°C, beads were washed and resuspended in 20 μL sample buffer. Samples were run on 12% SDS-PAGE gels, transferred, and blocked as described above. Membranes were incubated for 1 hour in 1:1,000 anti-HA, anti-Myc (9E10, Sigma), anti-T7 (EMD Biosciences/Novagen, San Diego, CA), or anti-PLCβ₂ (Santa Cruz Biotechnology, Santa Cruz, CA) and then washed. Membranes were incubated with secondary antibody (either anti-mouse or anti-rabbit IgG-HRP) and developed as described above.

Reporter gene assays. For transient luciferase assays, NIH 3T3 cells in 6- or 12-well dishes were cotransfected with 500 ng pCGN-rac3 61L or 500 ng pCGN-rac3 61L EDMs and 500 ng c-jun, E2F, SRF, or cyclin D1 reporter constructs (29). All transfections were done in at least duplicate. Three hours after transfection by LipofectAMINE and PLUS reagent, the medium was replaced with 0.5% serum DMEM and grown for 20 to 24 hours. For the PLCβ₂-SRF assays, cells were transfected by FuGENE6 in 10% serum for 20 to 24 hours with 5 ng pCGN-rac3 61L or 25 ng pCGN-rac3 61L EDMs and 150 ng of the SRF reporter construct in the presence of either pcDNA3 vector or 100 ng pcDNA3.1⁺-plcβ₂. The cells were then rinsed with 1× PBS and lysed in 1× lysis buffer (Amersham Biosciences), and luciferase activity was measured with enhanced chemiluminescence reagents (Amersham Biosciences) in a Monolight 2010 luminometer (Analytical Luminescence, San Diego, CA).

Phospholipase C activity assay. COS-7 cells were seeded in 12-well culture dishes at a density of 7 × 10⁴ per well and transfected with 100 ng/well pCGN-rac3 61L or pCGN-rac3 61L EDMs and 300 ng/well pcDNA3 vector or pcDNA3.1⁺-plcβ₂. Approximately 24 hours after transfection, culture medium was changed to inositol-free DMEM (ICN Biomedicals, Aurora, OH) containing 1 μCi/well myo-[2-³H(N)]inositol (American Radiolabeled Chemicals, St. Louis, MO), and metabolic labeling was allowed to proceed for 12 to 16 hours. Accumulation of [³H]inositol phosphates was initiated by the addition of 10 mmol/L LiCl to inhibit inositol monophosphate phosphatases. The reaction was stopped after 60 minutes by aspiration of the labeling medium and the addition of 50 mmol/L formic acid followed by neutralization with 150 mmol/L NH₄OH. [³H]Inositol phosphates were quantified by Dowex chromatography as described previously (30).

All Rac3 effector domain mutants tested retain the ability to produce membrane ruffles. To better understand the role of Rac3 effector pathways, we generated mutations in the core effector domain of Rac3 analogous to mutations described previously for Rac1 (12, 31, 32). It has been reported that a F37L mutation in Rac1 impairs the formation of membrane ruffles and lamellipodia (12, 32). To determine the effects of this mutation and others on the ability of Rac3 to produce membrane ruffles, NIH 3T3 cells expressing EGFP-Rac3 or Rac3 EDMs were scored as either ruffling or flat (Fig. 1). For activated Rac3 without any effector domain mutation, ∼62% of cells were characterized as ruffling (Fig. 1B). Whereas no single effector domain mutation caused a complete impairment of ruffle formation, either a N26D or a Y40C mutation led to a decreased percentage of ruffling cells, 33% and 47%, respectively. A N43D mutation led to an increased percentage of cells that were ruffling at 70%. Unlike previous reports for Rac1, a F37L mutation in the effector domain of Rac3 did not lead to impairment of ruffling but instead led to enhanced ruffling, with >75% of the cells that were counted producing membrane ruffles. These results suggest that Rac1 and Rac3 may use different effector pathways to promote cytoskeletal rearrangements, because the F37L effector domain mutation produced the opposite effect in Rac3 and Rac1.

Mutations in the effector domain of Rac3 alter the morphology of transformed foci. We have shown previously that activated GTP-bound Rac3 is capable of promoting cellular transformation by loss of contact inhibition, forming foci of morphologically transformed cells in cooperation with activated Raf (16). We have also noted that the foci formed by Rac3 are more compact in morphology than those formed by Rac1.⁵

To determine if mutations that impair transformation by Rac1 also affect transformation promoted by Rac3, we did a focus formation assay in NIH 3T3 cells transfected with Raf alone, Rac1, Rac3, or Rac3 EDMs alone or with Raf in combination with Rac1, Rac3, or Rac3 EDMs. Unlike Rac1, where both N26D and N43D mutants were impaired in promoting loss of contact inhibition in a focus assay (12), all Rac3 EDMs were capable of forming foci in cooperation with Raf with similar total numbers of foci (Fig. 2; data not shown). However, the morphology of the foci differed depending on the mutation. Small compact foci (“Rac3 like”) were formed predominantly in Rac3 and Rac3 N26D-transfected cells. Larger, spread foci (“Rac1 like”) were formed predominantly in Rac1 and Rac3 F37L-, Y40C-, and N43D-transfected cells. The shift of focus morphology from a Rac3-like phenotype to a Rac1-like phenotype suggests that Rac1 and Rac3 may use different subsets of effectors to promote this type of transformation and that the EDMs of Rac3 tested may shift the profile of effectors used.

Only the N43D mutation retains the ability to promote anchorage-independent growth. Another characteristic of cellular transformation is the ability of cells to grow in an anchorage-independent manner. We have established previously that GTPase-defective, activated Rac3 is capable of promoting anchorage-independent growth by forming colonies in soft agar (16). Here, we establish that a fast cycling mutant of Rac3 (F28L) is also capable of promoting anchorage-independent growth (Fig. 3). We used a soft agar assay to gauge the ability NIH 3T3 cells stably expressing Rac3 EDMs to promote anchorage-independent growth. Immunoblot analysis indicated similar expression levels of activated Rac3 and Rac3 EDMs with the exception of Rac3 F37L, which consistently had decreased expression in several independent stable lines generated, as shown in Fig. 3A. Expression of the F37L mutant was confirmed transiently (Fig. 3A), suggesting that decreased expression was required for selection of stably expressing cells. In contrast to the focus formation assay, where all EDMs retained transforming ability, only the N43D mutant was capable of transformation in soft agar, producing colonies in somewhat greater number but of similar size to Rac3 lacking any effector domain mutation (Fig. 3B and C). A N26D, F37L, or Y40C mutation greatly impaired transformation in soft agar, with few colonies formed, and in those that did form, smaller colony size. For the F37L mutation, it is unclear if transformation is impaired by the consequences of the mutation on effector interaction or by the conistent down-regulation of protein expression in stable lines. The pattern of expression of the EDMs and of colony formation, with only the N43D mutant retaining activity, was the same for soft agar assays done in an epithelial cell line, RIE-1 cells (data not shown). Previously reported data from similar assays done with Rac1 indicated that both the N26D and the N43D mutants were impaired in focus formation and colony formation (12). The fact that all EDMs tested for Rac3 retained transforming activity in focus formation assays combined with the opposite result seen with Rac3 N43D in soft agar assays, where, for Rac1, N43D is impaired but, for Rac3, it is not, again suggests that Rac3 uses an incompletely overlapping set of effector pathways compared with Rac1 to promote transformation.

Rac3 Y40C is impaired in binding to Pak1 and Par6, whereas both Rac3 F37L and Y40C have reduced binding to phospholipase Cβ₂. To determine how mutations in the effector domain of Rac3 affect binding to known and potential effectors, we used a GST pull-down assay, which has been described previously for determining effector interactions with RhoG (28). Lysates from NIH 3T3 cells that were expressing effector proteins (Pak1, Par6, and PLCβ₂) were combined with agarose beads bound with GST alone or GST fused to GTP-bound Rac3 or Rac3 EDMs. Beads bound with GST fused to GTP-bound Rac1 were used as a positive control for effector binding. Binding to Rac3 or Rac3 EDMs was determined by Western blot for the specific effector protein as shown in Fig. 4A. As expected, known Rac3 effectors Pak1 and PLCβ₂ (21, 33) bound to wild-type Rac3. However, Rac3 Y40C displayed impaired binding to Pak1, and both Rac3 Y40C and F37L were diminished in binding to PLCβ₂. Additionally, we show that Rac3 can bind to Par6, a known Rac1 effector/adaptor protein (34), and as with Pak1, only the Y40C mutant is impaired in binding (Fig. 4A). As Pak1 and Par6 both contain a Cdc42/Rac interactive and binding (CRIB) motif that is important for interaction with Rac, this suggests that, like what has been shown for Rac1, Y40C mutation impairs binding of CRIB-containing effectors to Rac3 (31, 35). Interaction of Rac3 with Par6 suggests a possible role for Rac3 in modulating cell/cell junctions and polarity, known functions of Par6, although the role that Rac1 plays in Par6-mediated signaling is incompletely defined (36).

Unlike Rac1, Rac3 is impaired in binding to mixed lineage kinases 2 and 3. GST pull-down assays were used as described above to determine if the known Rac1 effectors mixed lineage kinases 2 and 3 (MLK2 and MLK3; refs. 37, 38) could also bind to activated Rac3. To our surprise, we found that MLK2 bound very weakly to Rac3 compared with binding to Rac1 and that there was no detectable binding of MLK3 to Rac3 despite equivalent loading of the GST fusion proteins (Fig. 4B). This result suggests that Rac3 may differ from Rac1 in its ability to use MLK proteins to promote its downstream effects or that Rac3 may interact with MLK2/MLK3 through additional proteins in vivo to promote its effects.

Mutations in the effector domain of Rac3 differentially impair signaling to Rac3 transcriptional pathways. We have shown previously that Rac3 up-regulates transcription by the c-jun transcription factor (16). Here, we show that transient expression of activated Rac3 also up-regulates the transcription of cyclin D1 and the transcription of E2F- and SRF-responsive elements (Fig. 4C). To determine the effect of specific mutations in the effector domain of Rac3 on transcriptional transactivation, we did luciferase assays to gauge the ability of Rac3 EDMs to up-regulate transcriptional pathways. Activation of SRF transcriptional activity is greatly reduced by Rac3 with a F37L or Y40C mutation. However, for c-jun transcriptional activity, only the Y40C mutant is impaired in signaling. Transcription of cyclin D1 is reduced by N26D, F37L, or Y40C mutations but not by a N43D mutation. All effector domain mutations tested decrease E2F transcriptional activity, with Y40C and N43D showing the greatest impairment. For all transcriptional pathways tested, a Y40C mutation in Rac3 led to decreased activity, suggesting the importance of effectors of Rac3 that are disrupted by this mutation, such as Pak and Par6 (Fig. 4A) or other CRIB motif–containing effectors. When transient expression of Rac3 EDMs was evaluated by Western blot (Fig. 4D), all mutant proteins were expressed, indicating that impaired expression of the Y40C mutant does not account for its reduced transcriptional activity.

Rac3 synergizes with phospholipase Cβ₂ to promote transcriptional activation of serum response factor. A comparison of the effector binding and transcriptional data from Fig. 4 suggests that there is a correlation between reduced binding to the effector PLCβ₂ and reduced activation of SRF-mediated transcription. Although it has been shown that PLCβ₂ is an effector of Rac1, Rac2, and Rac3, as defined by its selective binding to Rac-GTP versus Rac-GDP, the biological functions resulting from its activation by Rac have not been determined (23, 39). To further explore this potential connection, we determined the ability of Rac3 EDMs to activate PLCβ₂ by use of a PLC activity assay. As seen in Fig. 5A, the F37L mutant of Rac3 that was most impaired in binding to PLCβ₂ (Fig. 4A) also was the most impaired in activating PLCβ₂. The N43D mutant, which retained binding to PLCβ₂, also retained ability to activate PLCβ₂ to levels similar to Rac3 without an effector domain mutation. The N26D and Y40C mutants of Rac3 displayed intermediate ability to activate PLCβ₂. Although the Y40C mutant displayed reduced binding to Rac3, its binding was not completely abolished; thus, it is not surprising that it retained some ability to activate PLCβ₂. Although our results in Fig. 4A suggest that the N26D mutant binds to PLCβ₂ with equivalent affinity to Rac3 without an effector domain mutation, we cannot rule out that the mutation has an effect structurally that may affect the efficiency of PLCβ₂ activation, despite robust binding, which may explain its reduced activity in the PLC activity assay. All EDMs were expressed equivalently under these assay conditions (data not shown).

To determine if Rac activation of PLCβ₂ contributes to SRF activity, we did luciferase reporter assays for SRF with Rac3 in the presence or absence of PLCβ₂. As seen in Fig. 5B, either PLCβ₂ alone or Rac3 alone led to roughly 3-fold activation of SRF, but when expressed together SRF-mediated transcriptional activity was boosted to ∼16-fold. The synergistic effect of expressing Rac3 and PLCβ₂ suggests that Rac3 uses PLCβ₂ activity to promote SRF-mediated transcriptional activation. Further bolstering this idea, Fig. 5C shows Rac3 with reduced capability to bind to PLCβ₂ (F37L and Y40C) and also exhibited reduced synergy with PLCβ₂ in SRF-mediated transcription, promoting only marginal increases in activity over the additive contributions of Rac3 and PLCβ₂ alone. In contrast, the N26D and N43D mutants, when transfected with PLCβ₂, promoted synergistic effects of ∼20- and 15-fold, respectively, compared with only 9- and 8-fold, respectively, for the additive contributions of Rac3 and PLCβ₂ alone. These results suggest that PLCβ₂ activation is a novel effector pathway used by Rac3 to promote SRF-mediated transcription.

We have used EDMs to evaluate relationships among Rac3 effector binding, transcriptional activation, membrane ruffling, and transformation (summarized in Fig. 6A). Our results indicate that the ability of Rac3 to promote transformation requires multiple effector pathways. Membrane ruffling and activation of transcriptional pathways involving c-jun, SRF, and E2F are not sufficient to promote transformation. In the case of membrane ruffling, all EDMs tested retained some level of activity, yet only the Rac3 N43D EDM promoted anchorage-independent growth (Fig. 6A). This agrees with published data for Rac1, where mutations that impaired lamellipodia, such as F37L or deletion of the insert region, had no apparent loss of transforming ability (12, 29). For transcriptional activation of c-jun, SRF, and E2F, the Rac3 N26D EDM retained fairly robust activation of these pathways yet was clearly impaired in anchorage-independent growth (Fig. 6A), indicating that activation of c-jun-, SRF-, and E2F-mediated transcriptional pathways are not sufficient to promote transformation. Additionally, the Rac3 N26D EDM also retained robust binding to Pak1, Par6, and PLCβ₂, indicating that these effector pathways are also not sufficient for transformation (Fig. 6A and B). This is in agreement with data from Rac1 EDMs and deletion mutants, showing a lack of correlation among Pak1 binding, JNK or SRF activation, and transformation (12, 31, 32).

Although the core effector domain is 100% identical between Rac1 and Rac3, protein sequence differences do occur in regions COOH terminal to it (18, 19). Effectors and interacting proteins of Rac1, such as Pak1, NADPH oxidase, and PI5K, are known to bind regions outside the core effector domain (40–42). Thus, the importance of other effector binding regions in Rac proteins suggests that differential effector binding between Rac1 and Rac3 could occur. This seems to be the case for in vitro binding of Rac3 to MLK2 and MLK3. Both contain a degenerate CRIB motif that is missing one of two conserved histidine residues, suggesting that MLK2/MLK3 may not interact as strongly with the core effector domain where the CRIB motif binds and may require other regions of Rac for efficient binding (43). JNK activation is correlated with MLK3 activation downstream of Rac1 (37, 38), and it has been shown that Rac3 can activate JNK (18, 21) as well as c-jun-mediated transcription (16). This suggests that Rac3 may preferentially use effectors other than MLKs to activate JNK/c-jun, such as Pak, mitogen-activated/extracellular signal regulated kinase kinase kinase 1/4 (MEKK 1/4), or plenty of SH3s (POSH; refs. 44–46). For Rac1-mediated activation of p38, a scaffolding molecule, JIP2, was shown to assemble a cascade from the Rac1 GEF Tiam1 to MLK3, MKK3, and p38 (47). This suggests that even if Rac3 is impaired in direct binding to MLK3, scaffolding molecules, such as JIP2, could still allow for signaling through MLK3 to its downstream targets.

Our data on Rac3 EDMs and transformation also suggest that Rac1 and Rac3 use an incompletely overlapping set of effectors. A previous study reported that the pattern of impairment was the same for Rac1 in both transformation by loss of contact inhibition or gain of anchorage independence, with both Rac1 N26D and N43D impaired (12). This was not the case for Rac3. The variation in focus morphology between Rac1 and Rac3 coupled with the different effects of the same EDMs in Rac1- and Rac3-mediated transformation suggests that there are differing pools of effectors that Rac1 and Rac3 use to promote transformation. This may reflect differences in innate effector binding between Rac1 and Rac3 and/or different accessibility of effectors due to different localizations of Rac1 and Rac3 within cells.

Here, we describe a new connection between Rac3 activation of PLCβ₂ and SRF-mediated transcription. Before this, it was unclear what role PLCβ₂ plays downstream of Rac proteins (23, 39). Activation of PLCβ isoforms leads to hydrolysis of phosphatidylinositol 4,5-bisphosphate into the second messengers inositol 1,4,5-trisphosphate (IP₃) and diacylglycerol (DAG; ref. 48). DAG then activates protein kinase C, whereas IP₃ leads to an increase in cytoplasmic calcium ion concentration and activation of calmodulin and calmodulin kinase (48, 49). Interestingly, several reports indicate the involvement of Ca²⁺ and calmodulin kinase in the regulation of SRF activity. SRF was reported to regulate increased transcription in response to elevated levels of cytoplasmic Ca²⁺ in a pathway involving calmodulin kinase (50). Additionally, a recent report suggests that the activity of SRF is repressed in the nucleus by histone deacetylase 4 (HDAC4) binding and that association of calmodulin kinase after Ca²⁺ stimulation with the HDAC4/SRF complex resulted in derepression and increased SRF activity (51). SRF is a common target in many growth factor–induced pathways, suggesting that PLCβ₂ could contribute to cellular growth activities induced by Rac activation. Further studies will help to refine the pathway from PLCβ₂ to SRF and to determine the potential involvement of PLCβ₂ in other Rac-mediated signaling pathways and biological functions.

Deregulation of growth control pathways is a major means by which Rho proteins contribute to cellular transformation. Our data suggest a correlation between transcription of cyclin D1 and Rac3-mediated anchorage-independent growth (Fig. 6A and B). Both Rac and cyclin D1 activity are required for transformation induced by Ras (13, 14, 52). Furthermore, cyclin D1 transcription has been correlated with the transforming activity of several Dbl family GEFs, which activate Rac1 and other Rho family proteins, and is required in mouse models for transformation by Ras (52–54). Our results suggest that cyclin D1 may play an important role in transformation induced by Rac3. Further determination of signaling pathways relevant for Rac3-mediated transformation awaits future study.

Grant support: NIH grants T32-CA71341 (P.J. Keller) and CA67771 (A.D. Cox).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

We thank Krister Wennerberg for helpful discussions and advice on the GST-Rac effector pull-down assays, Lloyd Greene for the MLK2 construct, and Laura Brenner for technical assistance in creating the EDMs.