G 13 Stimulates Na -H Exchange through Distinct Cdc42-dependent and RhoA-dependent Pathways*

Activity of the ubiquitously expressed Na (cid:49) -H (cid:49) exchanger subtype NHE1 is stimulated upon activation of receptor tyrosine kinases and G protein-coupled recep-tors. The intracellular signaling pathways mediating re- ceptor regulation of the exchanger, however, are poorly understood. Using transient expression of dominant in- terfering and constitutively active alleles in CCL39 fi-broblasts, we determined that the GTPases Ha-Ras and G (cid:97) 13 stimulate NHE1 through distinct signaling cascades. Exchange activity stimulated by constitutively active RasV12 occurs through a Raf1- and mitogen-acti-vated protein kinase kinase/extracellular signal-regu- lated kinase kinase (MEK)-dependent mechanism. Constitutively active G (cid:97) 13QL, recently shown to stimulate the Jun kinase cascade, activates NHE1 through a Cdc42- and MEK kinase (MEKK1)-dependent mechanism that is independent of Rac1. Constitutively active Rac1V12 does stimulate NHE1 through a MEKK1-de-pendent mechanism, but dominant interfering Rac1N17 does not inhibit G (cid:97) 13QL-mediated or constitutively active Cdc42V12-mediated stimulation of the exchanger. Conversely, Cdc42N17 does not inhibit Rac1V12 activation of NHE1, suggesting that Rac1 and Cdc42 independ- ently regulate a MEKK1-dependent activation of the exchanger. Rapid ( < 10 min) stimulation of NHE1 with a z chimera also was inhibited by a kinase-inac- tive MEKK. G (cid:97) 13QL, but not RasV12, also stimulates NHE1 through a RhoA-dependent pathway that is inde- pendent of MEKK, and microinjection of mutationally active G (cid:97) 13 results in a Rho phenotype of increased stress fiber formation. These findings indicate a new target for Rho-like proteins: the regulation of H (cid:49) exchange and intracellular pH. Our findings also suggest that a MEKK cascade diverges to regulate effectors other than transcription factors.

The ubiquitously expressed Na ϩ -H ϩ exchanger subtype NHE1 1 plays a major role in intracellular pH (pH i ) homeostasis and in cell volume regulation (1). NHE1 activity is stimulated by hormones, cytokines, and growth factors, resulting in an increase in pH i . Hyperosmotic shock (2) and cell adhesion (3) also activate NHE1. Increases in NHE1 activity are associated with increased cell proliferation (4,5), differentiation (6,7), and neoplastic transformation (8 -10). Receptor (11, 12), but not osmotic (2), activation of NHE1 is associated with increased phosphorylation of the exchanger on serine residues, suggesting kinase-dependent regulatory mechanisms. Although activation of protein kinase C stimulates NHE1, growth factors and vasoactive agents can stimulate the exchanger independently of this kinase. Mutational activation of three GTPases, Ha-Ras (8,9), G␣ q (13)(14)(15), and G␣13 (13)(14)(15), stimulates NHE1 activity. Of these GTPases, only G␣ q activates the exchanger through a protein kinase C-dependent mechanism (14). 2 The downstream signaling events mediating Ha-Ras and G␣13 stimulation of NHE1 have not been identified. Ha-Ras and G␣13 regulate two parallel MAP kinase signaling cascades, and these cascades include serine/threonine protein kinases that could potentially modulate exchange activity.
Stimulation of one of these cascades, the extracellular signalregulated kinase (ERK) cascade, by activated growth factor receptor tyrosine kinases is mediated by the GTPase Ras. GTPbound Ras recruits an immediate downstream effector, Raf1, to the plasma membrane, where it is activated by an unidentified mechanism (16,17). The serine/threonine kinase Raf1 activates the MAP kinase kinases MEK1 and MEK2 (18). These dual-specificity kinases phosphorylate and activate the MAP kinases ERK1 and ERK2 (19). Activated ERKs regulate a wide range of cytosolic and nuclear proteins involved in cell proliferation and neoplastic transformation, including phospholipase A 2 , p90 rsk , c-Myc, and c-Fos (19,20). Heterotrimeric G proteins, both ␣ (21-23) and ␤␥ (23)(24)(25) subunits, also regulate the ERK cascade, although their action on ERK is more cell type-specific than that of Ras. For example, mutational activation of the ubiquitously expressed G␣13 subunit stimulates growth and neoplastic transformation in Rat1 (22) and NIH3T3 (26,27) fibroblasts; however, it enhances epidermal growth factor-stimulated ERK activity in the former cell type (22), but does not affect this kinase in the latter (26,27). In already transformed COS-7 cells, mutationally activated G␣13 has been found both to inhibit 3 and to have no effect (24, 28) on ERK activity. In contrast to its cell-specific effects on the ERK cascade, G␣13 consistently stimulates NHE1 activity in a wide range of cell types, suggesting a divergence in its actions on ERK and the exchanger.
Recently, a parallel MAP kinase cascade, the Jun kinase (JNK) or stress-activated protein kinase cascade, has been described (18,29,30). Identified substrates of JNK/stress-activated protein kinase are the transcription factors c-Jun and AFT2 (31). The JNK cascade is activated by epidermal growth factor (32) and UV irradiation (33) through a Ras-dependent mechanism and by cytokines in a Ras-independent manner (30,34). Recently, G␣13 was found to activate the JNK cascade through a mechanism that is interrupted by the dominant interfering allele RasN17 (28). Extracellular signals activating JNK are mediated through activation of MEK kinase (MEKK1), a mammalian homolog of the STE11 kinase involved in the yeast pheromone mating pathway (35). MEKK1 directly activates the stress-activated protein-kinase kinase SEK1 (JNK kinase), a direct upstream regulator of JNK (36,37). Upstream, although probably not direct regulators of MEKK1, are two members of the Rho subfamily of GTPases, Rac1 and Cdc42 (38,39). These GTPases are also involved in the organization of the actin cytoskeleton. Rac1 regulates membrane ruffling and lamellipodia (40,41), and Cdc42 controls the formation of filopodia (41,42). Although morphological studies suggest that Cdc42 acts upstream of Rac1 and Rac2 (41,42), their sequential regulation of JNK remains unclear. Coexpression of a dominant negative Rac1N17 with constitutively activated Cdc42V12 does not inhibit (38), or only partially inhibits (39), JNK activation, indicating that these GTPases may independently regulate the JNK cascade.
Ha-Ras and G␣13 are highly oncogenic, and their regulation of the ERK and JNK cascades may be critical for their transforming actions. The ability of these GTPases to constitutively stimulate NHE1 activity and the correlation of increased exchange activity with transformed phenotypes indicate the importance of determining the mechanisms mediating GTPase regulation of NHE1. In this study, we used dominant interfering alleles of kinases and GTPases to determine the mechanisms by which Ha-Ras and G␣13 activate NHE1. Although Ha-Ras and G␣13 regulate both the ERK and JNK cascades, our results indicate that Ha-Ras stimulates the exchanger through a Raf1-and MEK1-dependent mechanism that is independent of MEKK1. In contrast, G␣13 activates NHE1 through a MEKK1-dependent mechanism that requires Cdc42, but not Rac1. G␣13, but not Ras, also stimulates exchange activity through a Rho-dependent mechanism that is independent of MEKK1.

MATERIALS AND METHODS
Cell Culture and Transfections-Chinese hamster lung CCL39 fibroblasts were cultured in Dulbecco's modified Eagle's medium supplemented with 5% heat-inactivated fetal bovine serum. 18 h prior to transfections, cells were plated at a density of 0.8 ϫ 10 Ϫ6 in 60-mm dishes. Cells were transfected using the Lipofectamine method (Life Technologies, Inc.) with 1-3 g of DNA. pcDNA empty vector was used to maintain total transfected DNA constant. 18 h after transfection, cells were reseeded in serum-containing Dulbecco's modified Eagle's medium onto glass coverslips, allowed to adhere, and then maintained for an additional 18 -24 h in serum-free Dulbecco's modified Eagle's medium until used for measuring NHE1 activity.
NHE Activity-For pH i determinations, cells were transferred to a nominally HCO 3 Ϫ -free HEPES-buffered medium (46) and loaded with a 1 M concentration of the acetoxymethyl ester derivative of the pHsensitive dye 2Ј,7Ј-bis(2-carboxyethyl)-5(6)-carboxyfluorescein (Molecular Probes, Inc.) for 15 min at 37°C. 2Ј,7Ј-Bis(2-carboxyethyl)-5(6)carboxyfluorescein fluorescence was measured using a Shimadzu RF5000 spectrofluorometer by alternately exciting the dye at 500 and 440 nm at a constant emission of 530 nm. Fluorescence ratios were calibrated with 10 M nigericin in 105 mM KCl (47). Cells were acidloaded by the application (10 min) and removal of 20 mM NH 4 Cl (48). Rates of recovery from this acid load (dpH i /dt) were determined by evaluating the derivative of the slope of the pH i tracing at pH i intervals of 0.05. Data represent the mean Ϯ S.E. of the indicated number of separate cell transfections.
Immunoblotting-Immunoblot analyses were made from cells prepared for pH i determinations to ensure that samples used for Western blots and pH i determinations had similar levels of protein expression. After the glass coverslips were removed for fluorescence measurements, cell lysates were prepared from the remaining adherent cells. Samples were normalized for protein content, and 25 g of protein was resolved by SDS-polyacrylamide gel electrophoresis. Resolved proteins were transferred to nylon-supported nitrocellulose (Alameda Chemicals & Science, Inc.). After nonspecific binding sites were blocked, the filters were incubated with an anti-Myc antibody at a 1:1000 dilution. Immunopositive proteins were detected by chemiluminescence (Amersham Corp.).
Microinjection and Immunofluorescence-CCL39 cells were microinjected with plasmids in the nucleus, incubated in serum-free Dulbecco's modified Eagle's medium for ϳ5 h, and fixed in 4% formaldehyde in Ca 2ϩ /Mg 2ϩ -free phosphate-buffered saline. Immunofluorescence procedures using anti-Myc antibodies to visualize injected cells were carried out essentially as described previously (16). Cells were also stained with fluorescein isothiocyanate-labeled phalloidin (Sigma) at 0.5 g/ml.

Ras and G␣13 Use Distinct Kinase Cascades to Stimulate NHE1-Both
Ras and G␣13 regulate ERK and JNK. To determine whether the kinase cascades leading to ERK and JNK activation mediate stimulation of NHE1, we transiently expressed constitutively active kinases and GTPases and dominant negative kinases in CCL39 fibroblasts. NHE1 activity was determined by monitoring the rate of pH i recovery from an acute acid load induced by 20 mM NH 4 Cl. We first studied the role of Raf1, an immediate downstream effector of Ras (49). Expression of constitutively active v-Raf, RasV12, and ␣13QL stimulated NHE1 activity (Fig. 1A). The resting pH i of 6.97 Ϯ 0.02 (mean Ϯ S.E.) in vector controls also increased, to 7.18 Ϯ 0.02 with v-Raf, to 7.11 Ϯ 0.02 with RasV12, and to 7.12 Ϯ 0.01 with ␣13QL (n ϭ four separate transfections). We then cotransfected a kinase-deficient Raf1 (Naf) allele with these GTPases. This dominant interfering kinase specifically blocks activation of endogenous Raf1 activity (43). Naf completely inhibited RasV12-activated exchange activity (Fig. 1A) and decreased RasV12-induced resting pH i to 6.97 Ϯ 0.03 (n ϭ 4). Coexpression of wild-type Raf1 at a 1:1 ratio with Naf rescued the Naf-inhibited activation by RasV12 (Fig. 1A). Naf, however, had no effect on ␣13QL-stimulated exchange activity (Fig. 1A) or on ␣13QL-induced resting pH i (7.11 Ϯ 0.03 in the presence of Naf; n ϭ 4). In three separate transfections, coexpression of Naf also failed to inhibit activation of NHE1 by mutationally active G␣ q RC (data not shown). These findings suggested that Ras, but not G␣13 or G␣ q , stimulates NHE1 through a Raf1dependent mechanism.
Although Raf1 may act on several substrates, the MAP kinase kinase proteins MEK1 and MEK2 are its preferred targets (18). We used a catalytically inactive MEK-Km allele (35) to investigate MEK-dependent activation of NHE1. Coexpression of MEK-Km completely inhibited RasV12 activation of NHE1 ( Fig. 1B) and lowered RasV12-induced resting pH i from 7.13 Ϯ 0.02 to 6.97 Ϯ 0.03 (n ϭ 4). Ras therefore activates NHE1 by the same pathway as it activates ERK. MEK-Km, however, had no effect on ␣13QL-induced NHE1 activity (Fig. 1B) or on resting pH i (data not shown).
Ras (29,30) and G␣13 (28) also activate the JNK cascade, which suggests that another signaling pathway may mediate the effect of these GTPases on NHE1. The lack of suitable dominant interfering or constitutively active alleles of JNK and its upstream regulator JNK kinase limited our study to the role of MEKK1, an upstream regulator of JNK (36,37), in stimulating the exchanger. Expression of a constitutively active carboxyl-terminal truncated MEKK1 (MEKK1⌬) increased NHE1 activity (Fig. 1C) and increased the resting pH i from 7.00 Ϯ 0.02 in vector controls to 7.16 Ϯ 0.04 (n ϭ 3). Coexpression of a kinase-inactive MEKK1-Km with RasV12 had no effect on Rasstimulated exchange activity (Fig. 1C) or on Ras-induced increases in pH i (7.12 Ϯ 0.02 in the absence and 7.11 Ϯ 0.01 in the presence of MEKK1-Km; n ϭ 4). Coexpression of MEKK1-Km with ␣13QL, however, completely inhibited G13stimulated exchange activity (Fig. 1C) and reduced the G13induced resting pH i from 7.12 Ϯ 0.02 to 7.00 Ϯ 0.03 (n ϭ 5). G␣13, but not Ras, therefore uses a MEKK1-dependent pathway to stimulate NHE1.
Members of the Rho Family of GTPases Couple to the Activation of NHE1-Recently, two members of the Rho subfamily of GTPases, Rac1 and Cdc42, were determined to activate the JNK cascade (38,39). A third member of this family, Rho, does not couple to this signaling pathway (39,50). To examine whether these GTPases can regulate NHE1 activity, we transiently expressed the constitutively active alleles Rac1V12, Cdc42V12, and RhoAV14 in CCL39 fibroblasts. In three separate transfections, expression of each constitutively active GTPase resulted in an increase in exchange activity (Fig. 2). Additionally, the resting pH i of 6.98 Ϯ 0.01 in vector controls increased to 7.14 Ϯ 0.02 with Rac1V12, to 7.18 Ϯ 0.04 with Cdc42V12, and to 7.09 Ϯ 0.02 with RhoAV14. Coexpression of MEK-Km had no effect on Rac1V12 or Cdc42V12 stimulation of NHE1 (Fig. 2). Coexpression of MEKK1-Km with these alleles, however, completely inhibited their activation of the exchanger (Fig. 2) and reduced resting pH i to 7.01 Ϯ 0.02 with Rac1V12 and to 7.02 Ϯ 0.03 with Cdc42V12. In contrast, MEKK1-Km failed to reduce increases in either NHE1 activity (Fig. 2) or resting pH i induced by RhoAV14, confirming the specificity of its inhibitory effects on Rac1V12 and Cdc42V12. Hence, the mechanism whereby these small G proteins activate the ex-FIG. 1. Distinct kinases mediate RasV12 and ␣13QL stimulation of NHE1. The rate of pH i recovery (dpH i /dt ϫ 10 Ϫ4 pH/s) from an acute acid load was determined in CCL39 cells transiently expressing empty vector (pcDNA) or the indicated kinases and GTPases. Data are expressed as the recovery rate at pH i 6.75. A, Raf1 regulation of NHE1 activity was determined by measuring recovery rates in cells expressing pcDNA or constitutively active v-Raf, RasV12, and ␣13QL alone (control) or coexpressed with a dominant interfering Raf1 (ϩ Naf). Recovery rates also were determined in cells coexpressing RasV12, Naf, and wild-type Raf (ϩ Naf ϩ wtRaf). B, MEK-regulated NHE1 activity was determined in cells expressing pcDNA, RasV12, and ␣13QL in the absence (control) or presence (ϩ MEK-Km) of a dominant interfering MEK. C, the role of MEKK1 in stimulating NHE1 was determined in cells expressing pcDNA or constitutively active MEKK1⌬, RasV12, and ␣13QL alone (control) or coexpressed with a dominant interfering MEKK1 (ϩ MEKK-Km). Data represent the mean Ϯ S.E. of recovery rates in four to five separate transfections. changer is different, as Rac1 and Cdc42, but not RhoA, act through a MEKK1-dependent pathway. The MEKK1-independent effects of both Ras and Rho on NHE1 suggested that a Rho-mediated pathway may be an additional mechanism linking Ras to the exchanger. In three separate transfections, however, coexpression of a dominant interfering RhoAN19 had no effect on Ras-induced increases in exchange activity or resting pH i (data not shown).
Cdc42 and RhoA, but Not Rac1, Mediate G␣13 Activation of NHE1-Because all three members of the Rho family stimulated NHE1 activity, we examined their role in mediating G␣13 actions. Expression of a dominant interfering Rac1N17 had no effect on the actions of ␣13QL on exchange activity (Fig. 3A) or on resting pH i (7.12 Ϯ 0.01 in the absence and 7.13 Ϯ 0.02 in the presence of Rac1N17; n ϭ 6). Dominant interfering Cdc42N17 (Fig. 3B) and RhoAN19 (Fig. 3C), however, completely blocked NHE1 stimulation by ␣13QL. The resting pH i also decreased, from 7.12 Ϯ 0.01 with ␣13QL alone to 6.97 Ϯ 0.02 (n ϭ 4) with coexpression of Cdc42N17 and to 6.96 Ϯ 0.03 (n ϭ 4) with coexpression of RhoAN19. The inhibitory actions of Cdc42N17 and RhoAN19 were specific, as neither dominant interfering allele blocked RasV12 activation of the exchanger (data not shown). Rac1N17 was expressed at levels comparable to Cdc42N17 (Fig. 3D), suggesting that its inability to block ␣13QL activation of NHE1 was not attributed to the lack of protein expression. Rac1N17 also failed to inhibit NHE1 activity stimulated by RasV12 (data not shown). Although we found no condition that could serve as a positive control to confirm the dominant interfering function of Rac1N17, this plasmid construct was previously used to block the transforming action of RasV12 (44). Hence, the inhibition of ␣13QL-stimulated NHE1 activity by Cdc42N17, but not by Rac1N17, suggests that Rac1 may not act downstream of Cdc42, as was previously suggested by morphological studies (41,42). Recent reports on Cdc42 and Rac1 activation of JNK also failed to establish a clear hierarchal action of these Rho-related GTPases (38,39). To further confirm the independent actions of the Rho family of GTPases on NHE1, we examined the ability of their dominant interfering alleles to inhibit stimulation by their constitutively active alleles. Rac1N17 had no effect on Cdc42V12; Cdc42N17 had no effect on Rac1V12 or RhoAV14 stimulation; and RhoAN19 had no effect on Cdc42V12 stimulation (Fig. 4).
Members of the Rho family have distinct effects on the actin cytoskeleton. Rac1 regulates membrane ruffling and lamellipodia (40,41); Cdc42 controls the formation of filopodia (41,42); and RhoA induces stress fiber formation (40). We next examined the effect of ␣13QL on the actin cytoskeleton by plasmid microinjection followed by immunofluorescence microscopy. CCL39 cells expressing ␣13QL were contracted and showed a marked increase in stress fiber formation (Fig. 5, A and B), which was very similar to the phenotype induced by expression of RhoAV14 (Fig. 5, C and D). In contrast, expression of Rac1V12 caused cell spreading and stimulated lamellipodia formation (Fig. 5, E and F), whereas Cdc42V12-expressing cells showed enhanced formation of filopodia and, to a much lesser extent, of lamellipodia as well (Fig. 5, G and H). These results suggest that of the Rho family members, RhoA may be the primary target of G␣13 activation.
Acute Activation of NHE1 by a G␣13/␣ z Chimera Is Mediated by MEKK1-We have primarily studied NHE1 stimulation by the mutationally activated ␣13QL, which allows us to determine effects ascribed to a single GTPase. The effects of ␣13QL, however, could reflect an indirect response because this constitutively active G␣ subunit was expressed for 48 h before NHE1 activity was measured. To study rapid effects of G␣13, we used an ␣13/␣ z chimeric protein that allows specific receptor-mediated activation of G␣13. This chimera was constructed by substituting the five carboxyl-terminal residues of ␣13, which are thought to specify receptor recognition, with cognate residues of G␣ z (13). Because the D 2 -dopamine receptor (D 2 R) activates G␣ z , but not G␣13, we previously used this chimeric allele in HEK293 cells to demonstrate rapid stimulation of FIG. 3. ␣13QL stimulation of NHE1 activity is inhibited by Cdc42N17 and RhoAN19, but not Rac1N17. The rates of pH i recovery at the indicated pH i values were determined in CCL39 cells transiently expressing empty vector (pcDNA) or ␣13QL or coexpressing ␣13QL and the indicated dominant interfering Rho family proteins. Data represent the mean Ϯ S.E. of recovery rates in four to six separate transfections. A, coexpression of Rac1N17 had no effect on ␣13QL stimulation of NHE1 activity. B, coexpression of Cdc42N17 blocked ␣13QL-stimulated exchange. C, coexpression of RhoAN19 also blocked ␣13QL-stimulated exchange. D, expression of Myc-tagged Rac1N17 and Cdc42N17 was determined by immunoblotting with an anti-Myc antibody. NHE1 by activation of the D 2 R (13). In CCL39 cells, activation of an ␣13/␣ z chimera also resulted in a rapid stimulation of the exchanger (Fig. 6). Quinpirole, a D 2 R agonist, had no effect on NHE1 activity in cells expressing empty vector (pcDNA) or the D 2 R. In cells coexpressing the D 2 R and the ␣13/␣ z chimera, however, quinpirole stimulated exchange activity (Fig. 6A), inducing a rapid increase in the rate of pH i recovery from an acid load (Fig. 6B). Stimulation of NHE1 by quinpirole activation of ␣13/␣ z was completely inhibited by MEKK1-Km, whereas MEK-Km had no effect (Fig. 6C). These findings indicate that rapid as well as constitutive activation of NHE1 by G␣13 can be mediated by MEKK1. This mechanism of acute stimulation of the exchanger suggests that nuclear transcription factors, the previously described targets of a MEKK1-dependent signaling cascade, are probably not involved in a G␣13-MEKK1-NHE1 pathway. Coupling to NHE1 may therefore represent a divergence in MEKK1 signaling.

DISCUSSION
The ubiquitously expressed Na ϩ -H ϩ exchanger NHE1 is one of several ion exchangers involved in cytoplasmic pH homeostasis. Activation of NHE1 increases the rate of H ϩ efflux from the cell, resulting in a rise in pH i . Although the kinetics of NHE1 activation by growth factors, hormones, and cytokines has been extensively studied (1), post-receptor signaling mechanisms regulating the exchanger remain largely unknown. This study focuses on the signaling pathways mediating activation of NHE1 by Ha-Ras and G␣13. Although these GTPases regulate similar kinase cascades, our findings indicate they use distinct mechanisms to stimulate NHE1 activity (Fig. 7). It is not surprising that Raf1 and MEK, which act downstream of Ras, mediate RasV12 activation of the exchanger. ERK, a selective substrate for MEK (19), is also a likely component in Ras activation of NHE1, although there is currently no direct evidence for an ERK-dependent regulation of the exchanger. G␣13 also regulates ERK (22); however, it stimulated NHE1 activity independently of the Ras/Raf pathway. It is unknown how G␣13 regulates ERK; if the G␣13 signal acts directly on ERK, then the dominant interfering alleles of Raf1 and MEK used in this study would not block an ERK-mediated activation of NHE1.
Our findings identify NHE1 as a previously undescribed effector of a MEKK1 signaling pathway. A role for MEKK1 in activating the exchanger was determined by several experiments. First, a constitutively activated amino-terminal truncated MEKK1 increased NHE1 activity and pH i . Second, a kinase-inactive MEKK1-Km inhibited NHE1 activation by constitutively activated Rac1V12, Cdc42V12, and ␣13QL. This inhibition was specific, as MEKK1-Km had no effect on NHE1 activity stimulated by RasV12 or RhoAV14. Third, acute stimulation of the exchanger by D 2 R activation of an ␣13/␣ z chimeric protein was also inhibited by MEKK1-Km. This latter finding argues against a role for transcriptional regulation in a G␣13-MEKK1-NHE1 pathway, suggesting a divergence in the  (37), is stimulated by Ras through a Rac-dependent mechanism (38,39). Ras activation of NHE1, however, occurs independently of Rac and MEKK1. A likely explanation for this is that perhaps a critical threshold of MEKK1 activation must be attained to signal downstream to the exchanger. Ras stimulates only a modest increase in JNK activity (29,38,39) that is much less potent than its activation of ERK. Additionally, we consistently find that JNK activation by ␣13QL is 3-4 times greater than that by RasV12. 4 The distinct mechanisms whereby Ras and G␣13 stimulate NHE1 are not consistent with recent findings that G␣13 stimulates JNK through a Ras-dependent mechanism (28). A Ras-mediated activation of JNK by G␣13, however, could function in regulating effectors other than NHE1.
Our findings also identify NHE1 as a previously undescribed downstream effector of the Rho family of GTPases. Although three members of this family, Rac1, Cdc42, and RhoA, stimulated NHE1 activity, their upstream regulation and their coupling to the exchanger were distinct. Rac1 and Cdc42, recently shown to stimulate JNK activity (38,39,52), activated NHE1 through a MEKK1-dependent mechanism, whereas RhoA coupled to the exchanger independently of MEKK1. G␣13 activation of NHE1 was mediated by Cdc42 and Rho, but not by Rac, indicating that G␣13 does not regulate Rac. Dbl, a tissuespecific proto-oncogene, functions as a guanine nucleotide exchange factor for Cdc42 and RhoA, but not Rac1 (53). Although speculative, our findings suggest that perhaps a Dbl homolog expressed in CCL39 cells may act downstream of G␣13. If Cdc42-and Rho-mediated pathways are divergently regulated by G␣13, it is uncertain why a dominant interfering allele of either monomeric GTPase completely blocked ␣13QL activation of NHE1. One possibility is that G␣13 activation of the exchanger requires a cooperative action of Cdc42 and RhoA. Another explanation is that Cdc42N17 or RhoAN19 could competitively bind to a guanine nucleotide exchange factor common to Cdc42 and RhoA and thus indirectly inhibit activation of endogenous proteins.
Our morphological studies suggest that RhoA is the primary target of G␣13, as expression of constitutively activated ␣13QL increased stress fiber formation, similar to a RhoAV14-induced phenotype. How acute activation of G␣13 regulates the actin cytoskeleton is currently unknown, and we cannot exclude the possibility that transient activation of G␣13 might also induce a Cdc42 phenotype. Concertina, a Drosophila homolog of G␣13, also functions in cytoskeletal reorganization (58). Concertina is critical for early gastrulation, as it coordinates cell shape changes during ventral furrow formation. Hence, regulation of the actin cytoskeleton may be a phylogenetically conserved function of G␣13.
It is currently unknown whether the Rho-like GTPases divergently regulate NHE1 and the actin cytoskeleton or whether these effectors lie within a single signaling pathway. If the latter is true, then a critical question is whether activation of NHE1 occurs upstream or downstream of cytoskeletal reorganization. Cytoplasmic pH may be a potent modulator of cytoarchitecture, as the bundling and cross-linking of actin filaments are pH-dependent processes (54). In Dictyostelium, changes in pH i over the physiological range seen in our study (6.9 -7.2) produce dramatic reorganization of the actin cytoskeleton (54). Additionally, pH i -influenced cytoskeletal remodeling has been suggested to modulate cell motility (55), differentiation (56), and protein synthesis (57). NHE1 regulation could also occur downstream of cytoskeletal reorganization by the Rho-like GTPases, as osmotically induced changes in cell shape regulate Na ϩ -H ϩ exchange (2).
At this point, it is not clear how MEK-, MEKK-, and Rhomediated pathways couple to NHE1. Ultimately, these pathways must converge at the exchanger or at an upstream regulator of the exchanger. Using interaction cloning, we recently identified a novel protein, NIP1, that coprecipitates with NHE1 in vivo and regulates its activity. 5 Overexpression of NIP1 interferes with Ras and G␣13 activation of the exchanger, suggesting that the signaling pathways identified in this study may converge at NIP1 to regulate NHE1. The functional consequence of increased NHE1 activity remains unresolved. Constitutive activation of the GTPases and kinase cascades we studied induces neoplastic transformation, and transformed cells have a higher exchange activity and resting pH i than nontransformed cells (8 -10). The Rho and JNK pathways may also participate in cell volume regulation, and NHE1 is activated by osmotic challenge (2). Whether activation of NHE1 plays either an obligatory or a permissive role in the cellular actions regulated by these three distinct signaling pathways remains to be determined.