Unique in Vivo Associations with SmgGDS and RhoGDI and Different Guanine Nucleotide Exchange Activities Exhibited by RhoA, Dominant Negative RhoAAsn-19, and Activated RhoAVal-14 *

We compared the in vivocharacteristics of hemagglutinin (HA)-tagged RhoA, dominant negative RhoAAsn-19, and activated RhoAVal-14 stably expressed in Chinese hamster ovary (CHO) cells. Proteins co-precipitating with these HA-tagged GTPases were identified by peptide sequencing or by Western blotting. Dominant negative RhoAAsn-19 co-precipitates with the guanine nucleotide exchange factor (GEF) SmgGDS but does not detectably interact with other expressed GEFs, such as Ost or Dbl. SmgGDS co-precipitates minimally with wild-type RhoA and does not detectably associate with RhoAVal-14. The guanine nucleotide dissociation inhibitor RhoGDI co-precipitates with RhoA, and to a lesser extent with RhoAVal-14, but does not detectably co-precipitate with RhoAAsn-19. Wild-type RhoA is predominantly in the [32P]GDP-bound form, RhoAVal-14 is predominantly in the [32P]GTP-bound form, and negligible levels of [32P]GDP or [32P]GTP are bound to RhoAAsn-19 in 32P-labeled cells. Immunofluorescence analyses indicate that HA-RhoAAsn-19 is excluded from the nucleus and cell junctions. Microinjection of SmgGDS cDNA into CHO cells stably expressing HA-RhoA causes HA-RhoA to be excluded from the nucleus and cell junctions, similar to the distribution of RhoAAsn-19. Our findings indicate that the expression of RhoAAsn-19 may specifically inhibit signaling pathways that rely upon the SmgGDS-dependent activation of RhoA.

Dominant negative Rho mutants are used to define the signaling pathways that regulate Rho activity (12)(13)(14)(15). Substitution of asparagine for threonine at amino acid 19 in RhoA is believed to induce a dominant negative function because of the increased association of RhoA Asn-19 with a GEF. Competitive binding of a specific GEF by RhoA Asn-19 may inhibit the GEF from interacting with endogenous RhoA. This event would disrupt any signal transduction pathway that normally utilizes the GEF. This model of how RhoA Asn-19 induces a dominant negative phenotype is based on previous studies using mutant Ras proteins with the analogous substitution of asparagine for serine at amino acid 17 (16,17). These studies support the model that Ras Asn-17 has a dominant negative function because it competitively binds a GEF needed for the activation of endogenous Ras (16,17). However, the GEFs that are preferentially bound by Ras  or RhoA Asn-19 in vivo have not been characterized. The potential dependence of endogenous RhoA on the proteins competitively bound by RhoA Asn-19 provides a strong rationale for identifying the proteins associating with RhoA Asn-19 in vivo. For this reason, we investigated the in vivo characteristics of hemagglutinin (HA)-tagged wild-type and mutant RhoA stably expressed in Chinese hamster ovary (CHO) cells.

EXPERIMENTAL PROCEDURES
Cells and Plasmids-The generation of the clonal CHO-m3 sublines stably expressing HA-tagged wild-type or mutant RhoA was described previously (12). HA-RhoA is expressed by the m3WTRho-1 and m3WTRho-11 cell lines. Activated HA-RhoA Val-14 is expressed by the m3CARho-1 and m3CARho-4 cell lines, and dominant negative HA-RhoA Asn-19 is expressed by the m3DNRho-2, m3DN-Rho-4, and m3DN-Rho-6 cell lines. The m3Zeo-2 cell line is a clonal CHO-m3 subline that does not express HA-tagged Rho proteins (12). All CHO-m3 sublines used in this study also express transfected human M 3 muscarinic acetylcholine receptors (mAChR) (12). The cells were cultured in complete medium (12) to promote exponential proliferation. Exponentially proliferating Jurkat and HeLa cells were provided by Drs. John Noti and Robert Aronstam (Guthrie Research Institute), and freshly collected rat brain tissue was a gift from Dr. Seong-Woo Jeong (Guthrie Research Institute). The pSR␣-smg GDS plasmid containing the full-length Smg-GDS coding sequence (18)  ) and scraped from the culture dishes after the addition of ice-cold extraction buffer (50 mM Hepes, 500 mM NaCl, 15 mM MgCl 2 , 0.5% Triton X-100, 1 mM phenylmethylsulfonyl fluoride (PMSF), 4 M leupeptin, 0.3 M aprotinin, pH 7.4). The cell lysates were centrifuged (13,000 ϫ g, 10 min, 4°C), and the resulting supernatants were incubated (90 min, 4°C) with HA antibody (5 g/ml) (Babco, Berkeley, CA) and Protein A-agarose (Life Technologies, Inc.). The 32 P-labeled nucleotides were eluted from the immunoprecipitated HA-RhoA proteins and separated by thin layer chromatography as described previously (21).
Isolation and Sequencing of Co-precipitated Proteins-HA-RhoA Asn-19 was immunoprecipitated from 4 ϫ 10 8 m3DNRho-4 cells lysed in 1.5 ml of lysis buffer, using the HA antibody as described above. The immunoprecipitate was subjected to SDS-PAGE, and the resulting gel was stained with Coomassie Blue. The Coomassie Blue-stained 60-kDa protein that co-precipitates with HA-RhoA Asn-19 was collected in a piece of excised gel and frozen at Ϫ80°C until it was enzymatically digested and sequenced at the Protein Core Facility, Mayo Clinic (Rochester, MN), using previously described techniques (23).
Microinjection and Immunofluorescence-The cells were cultured for 2 days on glass coverslips etched with labeled grids (Eppendorf, Hamburg, Germany) before being microinjected as described previously (24) using an Eppendorf Model 5171 micromanipulator and Model 5246 transjector. The cells were microinjected with plasmids coding for fulllength SmgGDS, Ost-␣, or HA-Dbl at a concentration of 100 ng of DNA/l of injection buffer (24). The success of injection was determined by co-injecting some cells with the pEGFP-N1 plasmid containing the green fluorescent protein (GFP) coding sequence (CLONTECH Laboratories, Palo Alto, CA), at a concentration of 5 ng of DNA/l of injection buffer. Control cells were microinjected either with injection buffer or the pEFGP-N1 plasmid alone. The cells were cultured in complete medium (12) for 18 -24 h after injection, and then immunofluorescently labeled with HA or RhoA antibody (25). The microinjected cells were identified by their location on the grids on the coverslips or by their expression of GFP prior to fixation for immunofluorescent labeling.

RESULTS
The in vivo guanine nucleotide exchange activities of the HA-tagged wild-type and mutant RhoA proteins were determined by measuring the amounts of [ 32 P]GTP and [ 32 P]GDP bound to these proteins in 32 P-labeled cells (Fig. 1A). Wild-type RhoA is in the predominantly [ 32 P]GDP-bound form, whereas activated RhoA Val-14 is in the predominantly [ 32 P]GTP-bound form in the 32 P-labeled cells (Fig. 1A). In contrast, only minimal levels of [ 32 P]GDP and [ 32 P]GTP are bound to RhoA Asn-19 in the 32 P-labeled cells (Fig. 1A, lane 3). Both wild-type RhoA and activated RhoA Val-14 were found to bind significant levels of  (Fig. 1B).
A 60-kDa protein preferentially associates with dominant negative RhoA Asn-19 (lane 3, Fig. 2A). Enzymatic digestion and peptide sequencing of this 60-kDa protein yielded a 17-amino acid peptide identical to residues 320 -336 of SmgGDS (Fig.  2C). The high performance liquid chromatography fraction containing this peptide also contained a small amount of a secondary peptide. The sequence of this secondary peptide corresponds to residues 379 -393 of SmgGDS (Fig. 2C). These results indicate that SmgGDS is the 60-kDa protein that preferentially associates with dominant negative RhoA Asn-19 . This conclusion is supported by previous reports that SmgGDS is a 61-kDa (558 amino acids) GEF that physically associates with RhoA (3-6). SmgGDS and a 57-kDa protein preferentially coprecipitate with RhoA Asn-19 in all three sublines expressing dominant negative RhoA Asn-19 (lanes 3-5, Fig. 2D). Small amounts of these proteins also co-precipitate with HA-tagged wild-type RhoA expressed in m3WTRho-1 and m3WTRho-11 cells (lanes 1 and 2, Fig. 2D).
We found that both Dbl and Ost are expressed by m3DNRho-4 cells (Fig. 2E) and by the other CHO-m3 sublines (data not shown). Interestingly, dominant negative RhoA Asn-19 does not detectably co-precipitate with these GEFs nor with other proteins of similar relative molecular masses (95-115 kDa) (Fig. 2, A and D). The co-precipitation of Dbl or Ost with RhoA Asn-19 was also not observed when Western blots of immunoprecipitated RhoA Asn-19 were probed with Dbl or Ost antibodies (n ϭ 3, data not shown).
Wild-type RhoA is diffusely distributed throughout the nucleus and cytoplasm, with increased localization at cell junctions and in the perinuclear area of some cells (Fig. 3A). Activated RhoA Val-14 is similarly distributed throughout the cell, although some cells appear to have greater levels of RhoA-Val-14 in the nucleus compared with the cytoplasm (Fig. 3B). This appearance may result from increased nuclear localization of RhoA Val-14 or because increased cell spreading causes a more diffuse and apparently less intense staining of cytoplasmic RhoA Val-14 . Dominant negative RhoA Asn-19 is confined to the perinuclear area (Fig. 3C). Microinjection of SmgGDS cDNA into m3WTRho-1 cells resulted in the exclusion of HA-RhoA from the nucleus and from cell-cell junctions (Fig. 3E). In contrast, wild-type RhoA remains in the nucleus and at junctions after the cells are microinjected with Ost-␣ cDNA (Fig.  3F). Microinjection of HA-Dbl cDNA into CHO-m3 cells did not alter the junctional and nuclear staining of endogenous RhoA detected by RhoA antibody (data not shown). Microinjection of buffer or GFP cDNA also did not alter the distribution of HA-RhoA or endogenous RhoA in the cells (data not shown). DISCUSSION This study demonstrates that dominant negative RhoA Asn-19 has increased association with SmgGDS and reduced interaction with RhoGDI. RhoA Asn-19 also does not detectably bind GTP in the CHO-m3 sublines. This result is consistent with in vitro studies demonstrating that H-Ras Asn-17 has a 40-fold lower affinity for GTP compared with that of wild-type H-Ras (16). The binding of GTP by small GTPases may promote their disassociation from GEFs such as SmgGDS (6,26). Thus, RhoA Asn-19 may remain associated with SmgGDS because RhoA Asn-19 cannot bind GTP. These possibilities are consistent with a previously developed model suggesting that Ras Asn-17 exerts a dominant negative phenotype by nonproductively associating with a GEF (16,17).
These findings provide an explanation for the dominant negative function of RhoA  . The association of SmgGDS with RhoA Asn-19 may prohibit SmgGDS from interacting with endogenous RhoA. This event may disrupt signaling pathways that involve the SmgGDS-dependent activation of RhoA. RhoA Asn-19 expression diminishes myosin reorganization induced by stimulating M 3 mAChR in the CHO-m3 sublines (12). It is possible that M 3 mAChR stimulation cannot activate RhoA when Smg-GDS is competitively bound by RhoA Asn-19 , resulting in diminished myosin reorganization. We found that microinjection of SmgGDS cDNA into m3WTRho-1 cells does not mimic myosin reorganization induced by M 3 mAChR stimulation (data not shown). This result is expected because M 3 mAChR-mediated myosin reorganization depends upon other signals in addition to RhoA activation, including protein kinase C activation (12). RhoA or SmgGDS may have to be modified by mAChR-mediated signals before these proteins can participate in myosin reorganization. This possibility is consistent with a previous report that the SmgGDS-dependent activation of Rap1B is enhanced by protein kinase A activation (27).
The perinuclear distribution of dominant negative RhoA  suggests that it accumulates in the endoplasmic reticulum, late endosomes, or the Golgi apparatus. Although the co-localization of SmgGDS to this region has not been determined, other proteins that physically interact with Smg GDS, including Rap1B (28) and SMAP (29), also accumulate in these perinuclear organelles. We found that microinjection of SmgGDS cDNA causes wild-type HA-RhoA to be excluded from the nucleus and from cell junctions. The absence of HA-RhoA from the junctions of cells overexpressing SmgGDS is consistent with previous reports that SmgGDS causes small GTPases to dissociate from plasma membranes (5,30). Previous studies indicate that signals transduced by surface receptors cause RhoA to enter the nucleus (31). Our results suggest that interactions with SmgGDS may prohibit RhoA from entering the nucleus.
It is somewhat surprising that dominant negative RhoA Asn-19 interacts with SmgGDS but does not detectably associate with other GEFs such as Dbl or Ost. It is possible that Dbl and Ost are expressed at significantly lower levels than SmgGDS, causing the co-precipitation of RhoA Asn-19 with Dbl or Ost to be significantly less detectable than its co-precipitation with SmgGDS. We are currently addressing this possibility by comparing the expression of SmgGDS with other GEFs in the CHO-m3 cell lines.
SmgGDS can activate several other small GTPases in addition to RhoA in vitro, including Rac1 (3, 5, 6), Rac2 (6), Cdc42 (3, 6), K-Ras (3,4), and Rap1B (3,4). The binding of SmgGDS by RhoA Asn-19 could conceivably diminish the interaction of SmgGDS with these different GTPases. This event would disrupt signaling pathways that involve the SmgGDS-dependent activation of these other GTPases. However, several lines of evidence suggest that the expression of RhoA Asn-19 does not inhibit the activity of GTPases different from RhoA. First, these different GTPases are believed to be selectively activated by GEFs other than SmgGDS in vivo (2,6). Thus, the activity of these GTPases would only be minimally affected by their inability to interact with SmgGDS that is competitively bound by RhoA  . Secondly, SmgGDS is significantly more active toward RhoA than toward other GTPases (3)(4)(5). The potential loss of SmgGDS interactions would therefore affect RhoA more profoundly than other GTPases. Finally, we did not detect the co-precipitation of SmgGDS with wild-type Rac1, activated Rac1 Val-12 , or dominant negative Rac1 Asn-17 stably expressed in other CHO-m3 sublines. 2 This finding indicates that Rac1 does not interact with SmgGDS in these cells, even though Rac1 can be activated by SmgGDS in vitro (3,5,6).
In contrast to dominant negative RhoA Asn-19 , activated RhoA Val-14 has reduced association with both SmgGDS and RhoGDI. This result is consistent with reports that SmgGDS and RhoGDI have reduced affinity for the GTP-bound form of small GTPases (6,26). We did not detect any proteins that co-precipitate more effectively with RhoA Val-14 compared with wild-type RhoA. The interaction of RhoA Val-14 with potential effectors may be too labile to be detected by our immunoprecipitation techniques. Alternatively, the effectors that interact with RhoA Val-14 may not be solubilized by our immunoprecipitation techniques.
These findings help define how the expression of RhoA Val-14 or RhoA Asn-19 alters certain signaling pathways. The predom-inance of RhoA Val-14 in the GTP-bound form indicates that RhoA Val-14 may activate any signaling pathway that involves RhoA activation. In contrast, the preferential interaction of RhoA  with SmgGDS indicates that the expression of RhoA Asn-19 most likely disrupts signaling pathways that rely upon the SmgGDS-dependent activation of RhoA.