Characterization of RAC3, a novel member of the Rho family.

The small GTP-binding proteins Rac1 and Rac2 are critically important in regulating multiple signal transduction pathways in eukaryotic cells. Here we report the isolation of a novel third Rac family member, Rac3. Rac3 differs from Rac1/2 at its carboxyl-terminal end, a domain associated with subcellular localization and binding to specific cellular regulators. RAC3 mRNA expression patterns differ from those of RAC2, which is hematopoietic specific and also from those of RAC1. The RAC3 gene was mapped to chromosome 17q23-25, a region frequently deleted in breast cancer. Rac3 protein levels are not affected by organization of the actin cytoskeleton but remarkably, are serum-inducible. Rac3 is an active GTPase, and this activity is regulated by Bcr. When constitutively activated, Rac3 is able to stimulate efficiently the c-Jun amino-terminal kinase signaling pathway. These findings support a role for Rac3 in intracellular signaling.

The small G proteins Rac1 and Rac2 are highly related GTPases belonging to the Rho subfamily of Ras proteins (1)(2)(3). Rho family members regulate the organization of the actin cytoskeleton. Specifically, activation of Rac is associated with actin reorganization accompanying the appearance of lamellipodia and membrane ruffles in fibroblasts (4,5). In neuroblastoma cells, Rac1 stimulates the formation of lamellipodia leading to neurite development (6). Activated Rac stimulates F-actin uncapping and morphological transformation of platelets but inhibits receptor-mediated endocytosis (7,8). In neutrophils and macrophages, Rac is essential for activation of a multiprotein complex that produces superoxide in phagocytic cells, the NADPH oxidase (9 -11). Moreover, Rac1 regulates intracellular reactive oxygen species production in fibroblasts and the activity of the redox-dependent transcription factor nuclear factor-B (12,13).
Rac participates in signal transduction from the membrane to the nucleus via two distinct mitogen-activated protein kinase cascades. Exposure of cells to endotoxins, proinflammatory cytokines, or hyperosmolarity effects the activation of Rac and subsequently p38, ultimately resulting in activation of activating transcription factor 2 (14). A second stress-activated pathway involving Rac, the JNK 1 /stress-activated protein kinase pathway, leads to c-Jun and activating transcription factor 2 activation (14 -17). Activation of Rac also results in activation of the serum response factor, a transcription factor that binds to serum response elements found in the promoters of many growth factor-regulated genes (18). On a cellular level, Rac GTPases play an important role in cell cycle progression in response to mitogenic stimulation (17). Interestingly, overexpressed Rac1 complexes with and activates the 70-kDa ribosomal S6 kinase (pp70 S6k ), which is necessary for the progression of cells through the G 1 phase of the cell cycle (19). Other experiments have demonstrated that Rac is essential for cellular transformation by Ras (20,21).
A number of Rac-interacting regulators and effectors have been identified in vitro. Our laboratory was the first to isolate Bcr and Abr, large multidomain proteins with GTPase-activating protein activity toward Rac and the related Cdc42 (22)(23)(24). We have shown that ablation of the Bcr protein in vivo in null mutant mice results in deregulation of the production of superoxide generated by the NADPH oxidase, and we have implicated Bcr in the specific regulation of Rac2 in vivo (25). To date, no evidence for regulation of specific Rac1 functions by Bcr in vivo has been found (25). Unlike Rac2, Bcr is not hematopoietic specific (1-3, 22, 26), and therefore it is likely that in vivo Bcr also regulates other Rho family GTPases. Here we report the somewhat unexpected existence of a third human Rac species, Rac3. The finding that Rac3 is closely related to Rac1 and Rac2 yet is clearly distinct will allow us to investigate the specific role of this Rac species in the many processes attributed to Rac proteins.

EXPERIMENTAL PROCEDURES
Cloning and Sequencing of RAC3-A 350-base pair fragment from the coding region of mouse rac1 was used to screen a K562 gt10 cDNA library. Four cDNAs were isolated and partially sequenced. The 5Јuntranslated and coding region of cDNA clone R10 of 1.0 kb was sequenced completely on both strands.
Cell Lines-The K562, DU4475, HL60, SCaBER, Swiss 3T3, and COS-1 cell lines were obtained from the ATCC, Rockville, MD. The cell line GM04155 was from NIGMS Human Genetic Mutant Cell Repository, Camden, NJ, and 5838 was kindly provided by Dr. T. Triche. Spodoptera frugiperda Sf9 insect cells were from Invitrogen, San Diego.
Expression of mRNA-RNAs were isolated using the LiCl-urea method (27). Total RNAs (15 g) were run on guanidine thiocyanateagarose gels (28), blotted to Hybond-N (Amersham Corp.), and hybridized as described (29). A multiple tissue Northern blot (CLONTECH) was used to analyze tissue distribution of RAC3 in human. For detecting RAC3, a 0.24-kb BamHI probe from the 3Ј-untranslated region was used and blots were washed to 0.01 ϫ SSC at 65°C. RAC1 was detected with a 0.21-kb NcoI-EcoRI probe and RAC2 with a 0.28-kb MscI-PstI probe, both of which included 3Ј-untranslated sequences.
Chromosomal Localization-Somatic cell hybrid and regional mapping panels for chromosome 17 were from the NIGMS Human Genetic * This work was supported in part by funds provided by the Breast Cancer Fund of the State of California through the Breast Cancer Research Program of the University of California Grant 1RB-0001 and by Public Health Service National Institutes of Health Grants CA47456 and CA 50248. 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.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBank TM /EBI Data Bank with accession number(s) AF008591.
‡ To whom correspondence should be addressed: Section of Molecular Carcinogenesis, Department of Pathology, Ms 103, Childrens Hospital Los Angeles Research Institute, 4650 Sunset Blvd., Los Angeles, CA 90027.
Mutant Cell Repository. DNAs were digested with EcoRI and hybridized with a 0.45-kb XbaI fragment from the 3Ј-untranslated region of RAC3.
In addition, total cellular extracts were prepared from GST-Racexpressing Sf9 cells. RAC1 was also inserted into pGEX-2T (Pharmacia Biotech Inc.) and expressed as a GST-fusion protein in Escherichia coli. Bacteria were sonicated in buffer A without Triton X-100 and purified on glutathione-agarose (see above).
GTPase-activating protein Assay-GTPase-activating protein activity was determined by measuring GTP hydrolysis as described (22). Briefly, Rac1 and Rac3 were loaded with [␥-32 P]GTP by incubation of the GST-Rac fusion protein in 25 mM Tris, pH 7.5, 4.7 mM EDTA, 1 mM dithiothreitol, 0.1 mM MgCl 2 , 0.1 mg/ml bovine serum albumin, 20 M unlabeled GTP, and 10 Ci [␥-32 P]GTP (5,000 Ci/mmol, Amersham) at 30°C for 4 min. Equally loaded aliquots (as measured by cpm bound) of Rac1 and Rac3 were used for GTPase assays. 200 ng of GST-Bcr fusion protein was added to half of each sample. Hydrolysis was initiated by the addition of MgCl 2 and GTP to final concentrations of 17 mM and 170 M, respectively, and [␥-32 P]GTP remaining bound to protein was determined by filtration analysis (29).
Western Blotting and Immunodetection-Antibodies specific to Rac3 were raised in rabbit against amino acid residues 182-192 and were affinity purified (Zymed Laboratories). For immunoblotting, Rac3 antibody was used at a final concentration of 250 ng/ml. Antibodies against JNK1 (C-17) and GST were from Santa Cruz. Proteins were separated on sodium dodecyl sulfate-polyacrylamide gel electrophoresis, transferred to Hybond-ECL nitrocellulose (Amersham), probed with antibodies as described (30), and visualized using ECL (Amersham).
Rac3 Protein in Swiss 3T3 Cells-Swiss 3T3 cells were treated for 2 h with 10 Ϫ5 or 10 Ϫ6 M cytochalasin B (Sigma) dissolved in dimethyl sulfoxide (Sigma). The final concentration of dimethyl sulfoxide in the culture medium was 0.1%. Induction of Rac3 was studied by serum starving Swiss 3T3 cells for 24 h in Dulbecco's modified Eagle's medium plus 0.1% bicarbonate and stimulating with 10% fetal bovine serum for 1 h. Protein concentrations were determined using the BCA method (Pierce). 10 g of total cellular protein was loaded per lane.
Transfection-For eukaryotic expression studies, RAC3 was cloned into an SV40-based mammalian vector. To generate a constitutively activated Rac3 protein (V12Rac3), the glycine at codon 12 was replaced by valine using polymerase chain reaction. The sequence of the mutated construct was verified by sequencing. RAC3 was transferred into cells using lipofection. After 66 h, cells were lysed in 2 ϫ Laemmli sample buffer.
JNK Assay-COS-1 cells were transfected with V12RAC3 or were mock transfected. 27 h after transfection, the percentage of fetal bovine serum was lowered from 10 to 0.5%. Lysates were prepared after 18 h. As positive control, COS-1 cells were stimulated with 0.4 M sorbitol for 1 h to induce osmotic shock. Cells were lysed in Triton-lysis buffer (30). Endogenous JNK was immunoprecipitated from cleared lysates by incubation with JNK1 antiserum for 2 h at 4°C. Immunocomplexes bound to protein A-agarose beads were washed four times with lysis buffer and once with kinase buffer (31). Immunoprecipitated JNK activity was determined using a method reported previously (15). GST-c-Jun (1-79) was kindly provided by Dr. G. M. Bokoch. Samples were analyzed by sodium dodecyl sulfate-polyacrylamide gel electrophoresis and autoradiography. The relative amount of GST-c-Jun phosphorylation was quantitated by scintillation counting of excised bands. The presence of Rac3 and similar levels of JNK1 protein were confirmed by Western blot analysis.

RESULTS
Isolation of a RAC-related cDNA Clone-The human genome contains many RAC-related sequences. 2 To identify expressed sequences, a 0.35-kb murine rac1 cDNA fragment was used to screen a human K562 cell line cDNA library. Four clones were isolated, of which two represented RAC1 and one RAC2. The fourth, R10, with an insert of 1.0-kb, was sequenced (Fig. 1). Data base searches revealed that its nucleotide sequence was homologous to but clearly distinct from that of RAC1 (77% identity), RAC2 (83%), or RHOG (69%). Because of its high degree of homology to RAC1/2, we named our clone RAC3 (Fig.  2).
The R10 cDNA displayed an open reading frame, able to encode a protein of 192 amino acids. The clone also contained 70 base pairs of 5Ј-noncoding and 450 base pairs of 3Ј-untranslated sequences. Interestingly, this 3Ј region is well conserved between man and rodents (also see Fig. 3, lane 8), suggesting functional significance. Similar to mouse rac1, the 5Ј-untranslated region of RAC3 is exceptionally GC-rich (87%) albeit much shorter, with the potential to form hairpin loops (2).
Chromosomal Localization of RAC3-The 3Ј-untranslated region of RAC3 was chosen to prepare a highly specific probe. This probe detected a single band on Southern blots (see Fig. 3, lane 7) and did not cross-hybridize to the RAC1 or RAC2 cDNAs (not shown). A panel of somatic cell hybrid DNAs from hybrids that each retain one intact human chromosome was then used to localize RAC3 chromosomally. Only a single hybrid contained sequences hybridizing to the RAC3-specific probe, locating RAC3 on chromosome 17 (not shown). RAC1 and RAC2 were present on different chromosomes (not shown). To sublocalize RAC3 on chromosome 17, a regional mapping panel for human chromosome 17 was utilized. Human RAC3 was detected as a single EcoRI fragment of around 10 kb in total human DNA (Fig. 3, lane 7), and somatic cell hybrids containing this fragment localized RAC3 to the q23-25 region of chromosome 17 (Fig. 3, lanes 2, 3, 5, and 6).
Expression of Human RAC3-The RAC3-specific probe was used to investigate RAC3 mRNA levels. In all of the human cell lines studied, a single mRNA of around 1.0 kb was detected (Fig. 4C), which is in close agreement with the size of the R10 cDNA. It was relatively abundant in the chronic myeloid leukemia cell line K562 and also expressed, albeit at a lower level, in the Ewing sarcoma cell line 5838, the promyelocytic cell line HL60, and the breast cancer cell line DU4475 (Fig. 4C, lanes  2-5). In concordance with previous reports, RAC2 mRNA of 1.4 kb was detected mainly in cell lines of hematopoietic origin (Fig. 4B), whereas two RAC1 mRNAs, of 2.4 and 1.1 kb, were present in all of the cell lines (Fig. 4A). To investigate the tissue-specific distribution of RAC3, the RAC3-specific probe was hybridized to a human multiple tissue Northern blot. Interestingly, the highest RAC3 expression levels were found in brain, although it was also detected in heart, placenta, and pancreas (Fig. 4E). A second, larger RAC3 mRNA was detected only in placenta (Fig. 4E, lane 3).
GTPase Activity of Rac3-Not all Rho family members have GTPase activity (32). To examine whether Rac3 was a functional GTPase it was expressed as a GST-fusion protein and compared with GST-Rac1. Both proteins were loaded with [␥-32 P]GTP, and the rate of hydrolysis was measured using a filter binding assay. As shown in Fig. 5, the intrinsic GTPase activity of Rac3 was indistinguishable from that of Rac1. To test whether Bcr can stimulate the GTPase activity of Rac3, baculovirus-expressed GST-Bcr was incubated with [␥-32 P]GTP loaded GST-Rac3 and the rate of GTP hydrolysis measured. As shown in Fig. 5, Bcr stimulated the GTPase activities of Rac3 and Rac1 in a similar fashion.
Regulation of Rac3 Expression-Previous studies have shown that Rac1 expression levels can be modulated by the state of the actin network (2). To investigate this, endogenous Rac3 protein was analyzed in Swiss 3T3 cells. Treatment of these cells with 10 Ϫ5 or 10 Ϫ6 M cytochalasin B had no apparent effect on Rac3 levels (Fig. 6C, lanes 2-4), although this treatment caused a prominent morphological change in the cells caused by disruption of the actin cytoskeleton (not shown). Similarly, Rac3 levels were unaffected by disattachment from the substratum (lane 5). However, deprivation of cells from serum for 24 h caused a decrease in Rac3 levels (Fig. 6C,  compare lanes 6 and 7). Stimulation of the serum-starved cells with 10% fetal bovine serum for 1 h resulted in a return to normal Rac3 levels (lane 8).
V12Rac3 Activates Endogenous JNK-To investigate whether activation of Rac3 would affect signaling, it was mutated to a constitutively activated form by replacing glycine at position 12 with a valine. Overexpression of V12Rac3 in COS-1 cells and the activation of endogenous JNK was examined by phosphorylation of GST-c-Jun. As shown in Fig. 7 (bottom), mock-transfected, V12Rac3-transfected, and osmotically shocked COS-1 cells contained comparable levels of endogenous JNK. Exposure of cells to 0.4 M sorbitol resulted in a maximal stimulation of 4-fold of endogenous JNK activity. Overexpression of the constitutively active Rac3 resulted in the activation of JNK as measured by the 2-fold increased phosphorylation of the GST-c-Jun substrate (Fig. 7, top). DISCUSSION We have identified a new member of the Rho family, Rac3, which has a high degree of amino acid identity with human Rac1 and Rac2 (92 and 89%, respectively). This raises the question of why three highly related proteins exist. Despite this close degree of relatedness, it is likely that each Rac species plays a distinct regulatory role in vivo, although there may be overlap both in vitro and in vivo. Differences among the three could be found in the regulation of their GTPase activity through interaction with other proteins or at the level of gene expression.
With respect to protein structure, experiments with mutant Rac proteins have delineated domains of functional importance for the interaction of both Rac1 and Rac2 with effectors and other regulatory molecules. Two effector sites are needed for in vitro activation of the NADPH oxidase component p67 phox , Rac-induced actin polymerization in fibroblasts, and interaction with p65 PAK in vitro (33). The amino-terminal effector site is identical in all three Rac proteins, and the carboxyl-terminal effector site in Rac3 is 94 and 91% identical to the homologous region in Rac1 and Rac2, respectively ( Fig. 2 and Ref. 33). This suggests that Rac3 is also capable of binding p67 phox and p65 PAK , at least in vitro. In addition, the carboxyl-terminal residues have been identified as an important region for NADPH oxidase and JNK activation (34 -36). Our experiments on the functional activity of Rac3 show that it is a GTPase. Like Rac1, it interacts with Bcr, which enhances its GTPase activity in vitro. COS-1 cells overexpressing a constitutively activated Rac3 show a physiologically significant 2-fold activation of endogenous COS-1 JNK compared with its 4-fold activation by osmotic stress. This demonstrates that Rac3 is able to participate in the stress activation pathway.
The greatest divergence among Rac3, Rac1, and Rac2 occurs at residues 180 -192, which are also hypervariable regions in the Ras proteins (37). Indeed, the only functional difference demonstrated to date between Rac1 and Rac2 using an in vivo assay is that p67 phox interacts 6-fold better with Rac2 than with Rac1 (38). Therefore, this region may specify differences in in vivo binding activities among the three Rac proteins. All three Rac proteins contain a Cys-A-A-X-COOH sequence, in which A is an aliphatic residue and X is any amino acid. In vivo in Rac1/2, the cysteine residue becomes geranylgeranylated, followed by proteolytic removal of the A-A-X residues and carboxyl methylation of the isoprenylated cysteine (39). This posttranslational modification is important for specific intracellular localization and interaction with target proteins (19,40,41). Thus, the identified differences between Rac3 and Rac1/2 in their very carboxyl-terminal end may define differences in subcellular localization and/or binding to specific regulatory molecules.
Prominent differences among Rac1, 2, and 3 were found in expression levels in tissues and cell lines. The highest RAC3 mRNA expression in human tissues studied was in brain. In contrast, little or no RAC2 mRNA is found in brain (1-3). Both Bcr and its cognate Abr have GTPase-activating protein activity toward Rac proteins (22)(23)(24). Because the expression levels of both BCR and ABR are highest in brain (22), Rac3 may be a main target of these GTPase-activating proteins.
Using highly specific Rac3 antisera, Rac3 protein could be identified unambiguously. Previous experiments have shown rac1 mRNA expression levels are affected by the organization of the actin cytoskeleton (2). However, treatment with cytochalasin B, which disrupts actin cytoskeletal organization, did not have a significant effect on Rac3 protein levels, suggesting a functional difference between Rac3 and Rac1. Alternatively, it is possible that Rac3 and Rac1 protein levels are unaffected by this treatment. Surprisingly, Rac3 expression was down-regulated in quiescent fibroblasts and was clearly induced by serum stimulation. This suggests the Rac3 promoter is serum-responsive. To date, of the extended Rho family, only RhoB was shown to be serum-inducible; there have been no reports indicating that either Rac1 or Rac2 expression is modulated by serum. RhoB is closely related to RhoA and RhoC on an amino acid level. In contrast to RhoA and RhoC, RhoB does not have a clearly defined function in actin cytoskeletal reorganization, but rather it appears to be involved in cell proliferation (42,43).
In view of the increasing evidence for critical roles of Rho family members in actin reorganization and cellular signaling including cancer, it will be of interest to define the specific role of Rac3 in these processes. This, in addition to our finding that RAC3 is located at chromosome 17q23-25, a region frequently deleted in breast cancer (44,45), provides further impetus to investigate Rac3 expression and activity in this type of malignancy.