Constitutively Active Gα12, Gα13, and Gαq Induce Rho-dependent Neurite Retraction through Different Signaling Pathways*

In neuronal cells, activation of a certain heterotrimeric G protein-coupled receptor causes neurite retraction and cell rounding via the small GTPase Rho. However, the specific heterotrimeric G proteins that mediate Rho-dependent neurite retraction and cell rounding have not yet been identified. Here we investigated the effects of expression of constitutively active Gα subunits on the morphology of differentiated PC12 cells. Expression of GTPase-deficient Gα12, Gα13, and Gαq, but not Gαi2, caused neurite retraction and cell rounding in differentiated PC12 cells. These morphological changes induced by Gα12, Gα13, and Gαq were completely inhibited by C3 exoenzyme, which specifically ADP-ribosylates and inactivates Rho. The tyrosine kinase inhibitor tyrphostin A25 blocked the neurite retraction and cell rounding induced by Gα13 and Gαq. However, tyrphostin A25 failed to inhibit the Gα12-induced neuronal morphological changes. On the other hand, inhibition of protein kinase C or elimination of extracellular Ca2+ blocked the neurite retraction and cell rounding induced by Gαq, whereas the morphological effects of Gα12 and Gα13 did not require activation of protein kinase C and extracellular Ca2+. These results demonstrate that activation of Gα12, Gα13, and Gαq induces Rho-dependent morphological changes in PC12 cells through different signaling pathways.

The function of the nervous system depends on the highly specific pattern of connections formed between neurons during development. The specificity of these connections requires neurite extension toward the correct targets guided by the growth cone and remodeling of the initial pattern of connections in response to environmental signals (1). The Rho family of small GTPases (Rac, Cdc42, and Rho) has been demonstrated to play critical roles in the regulation of the cytoskeleton required for neurite extension and retraction. Studies on neuronal cell lines have shown that Rac and Cdc42 are required for the outgrowth of neurites, whereas Rho is required for their retraction (2)(3)(4)(5). The downstream effectors involved in these GTPase-mediated neuronal morphological effects have been elucidated. The p21activated kinase PAK1 was shown to act downstream of Rac and Cdc42 to induce neurite outgrowth (4). On the other hand, we recently revealed that the p160 rhoA -binding kinase ROK␣ induces neurite retraction acting downstream of Rho (5). However, little is known about the signaling pathways upstream of these Rho family small GTPases in neuronal cells.
The activation of a certain heterotrimeric G protein-coupled receptor, such as the lysophosphatidic acid (LPA), 1 sphingosine 1-phosphate, thrombin, and prostaglandin EP3 receptors, was shown to cause Rho-dependent neurite retraction in several neuronal cell lines (6 -9). However, the heterotrimeric G proteins, which are coupled to those receptors for induction of neurite retraction, have not yet been identified. Previous studies demonstrated that pertussis toxin did not inhibit receptormediated neurite retraction (9,10), indicating that this action is not mediated by G i or G o . Furthermore, the activation of G s by cholera toxin or an elevation of cAMP by forskolin failed to induce neurite retraction, but rather suppressed the receptormediated neurite retraction (9,11), suggesting that G s activation is not linked to induction of neurite retraction.
The G 12 family of heterotrimeric G proteins, defined by G␣ 12 and G␣ 13 , is the most recent family to be identified using a homology-based polymerase chain reaction (PCR) strategy (12). Although immediate downstream effectors have not yet been identified, studies with the constitutively active mutants of G␣ 12 and G␣ 13 have resulted in the identification of several novel functions regulated by these G␣ subunits, including transformation of fibroblasts (13,14), activation of the c-Jun N-terminal kinase cascade (15)(16)(17), stimulation of stress fiber formation and focal adhesion assembly (18,19), stimulation of the Na ϩ -H ϩ exchanger (20 -22), activation of phospholipase D (23), and induction of apoptosis (24). These studies also indicated that the Ras or Rho family small GTPases appear to be involved in the downstream responses regulated by G␣ 12 and G␣ 13 .
Rat pheochromocytoma PC12 cells have served as a useful model system for studies of neuronal differentiation and morphology. When PC12 cells are exposed to nerve growth factor (NGF) for several days, they acquire many features of sympathetic neurons, such as an outgrowth of neurites. To investigate the role of the G 12 family and other G␣ subunits in neuronal cell morphology, we microinjected expression plasmids encoding GTPase-deficient mutants of G␣ subunits into the nuclei of NGF-differentiated PC12 cells bearing neurites. We report here that expression of constitutively active mutants of G␣ 12 , G␣ 13 , and G␣ q caused Rho-dependent neurite retraction and cell rounding through different pathways.
Construction of Expression Plasmids-Wild-type mouse G␣ 12 and G␣ 13 were generous gifts from Dr. M. I. Simon (California Institute of Technology). Wild-type rat G␣ i2 was kindly provided by Dr. T. Katada (Tokyo University). The mammalian expression vector pEF-BOS was kindly provided by Dr. S. Nagata (Osaka University). Wild-type mouse G␣ q was generated by reverse transcription-PCR from mouse brain using primers 5Ј-GAGGCACTTCGGAAGAATGA-3Ј and 5Ј-AAGAAC-CAGTTTCTGGGAGG-3Ј, and the PCR product was cloned into the pCR2.1 vector and sequenced completely. The cDNAs of the constitutively active mutants of G␣ 12 (G␣ 12 Q229L (G␣ 12 QL)), G␣ 13 (G␣ 13 Q226L (G␣ 13 QL)), G␣ q (G␣ q Q209L (G␣ q QL)), and G␣ i2 (G␣ i2 Q205L (G␣ i2 QL)) were generated by PCR-mediated mutagenesis (25) and sequenced completely. The cDNA of RhoA V14 was obtained as described previously (5), and the cDNA of RhoA N19 was generated by PCR-mediated mutagenesis. The cDNAs of constitutively active mutants of G␣ subunits were inserted into the pcDNA3 expression vector (Invitrogen), and the cDNAs of RhoA V14 and RhoA N19 were inserted into pEF-BOS.
Cell Culture and Microinjection-PC12 cells were cultured in Dulbecco's modified Eagle's medium containing 10% fetal bovine serum, 5% horse serum, 4 mM glutamine, 100 units/ml penicillin, and 0.2 mg/ml streptomycin under humidified conditions in 95% air and 5% CO 2 at 37°C. For microinjection, cells were seeded at a density 2 ϫ 10 4 onto poly-D-lysine (Sigma)-coated 35-mm dishes, which were marked with a cross to facilitate the localization of injected cells. Microinjection was performed using an IMM-188 microinjection apparatus (Narishige, Tokyo). After cells had been differentiated in serum-free Dulbecco's modified Eagle's medium containing 50 ng/ml NGF for 3 or 4 days, 30 g/ml concentrations of the indicated plasmids were microinjected into the nucleus with 1 mg/ml Texas Red-coupled dextran to visualize injected cells. C3 exoenzyme (100 g/ml) was co-microinjected with expression plasmids. During microinjection, differentiated cells were maintained in Hepes-buffered Dulbecco's modified Eagle's medium (pH 7.4) at 37°C. After microinjection, cells were replaced in NGF-containing serum-free Dulbecco's modified Eagle's medium and incubated for 3 h. To examine the effect of extracellular Ca 2ϩ , microinjected cells were replaced in Hepes-buffered saline containing 140 mM NaCl, 4.7 mM KCl, 5 mM MgCl 2 , 1.2 mM KH 2 PO 4 , 11 mM glucose, 2 mM EGTA, and 15 mM Hepes (pH 7.4). Cells were photographed at ϫ400 magnification under a phase-contrast microscope and by fluorescence of Texas Red-coupled dextran. For quantitative examinations, neurite-retracted cells were defined as the cells that almost completely retracted their neurites and caused rounding of the cell body within 3 h of the microinjection of plasmids. The percentages of neurite-retracted cells were calculated by counting at least 30 Texas Red-positive cells in the same field, and all data were obtained from triplicate experiments.

Expression of Constitutively Active G␣ Subunits in NGFdifferentiated PC12
Cells-To determine the role of the G 12 family of heterotrimeric G proteins in neuronal cell morphology, we expressed constitutively active mutants of G␣ 12 and G␣ 13 in NGF-differentiated PC12 cells by nuclear microinjection of expression plasmids. Replacing a conserved glutamine with a leucine in the G3 region of the G␣ subunit, which corresponds to residue 229 in G␣ 12 and residue 226 in G␣ 13 , has been shown to result in a GTPase-deficient, constitutively active form of the G␣ subunit (26,27). As shown in Fig. 1, microinjection of expression plasmids (30 g/ml) encoding constitutively active G␣ 12 (G␣ 12 QL) and G␣ 13 (G␣ 13 QL) into the nuclei of NGF-differentiated PC12 cells caused retraction of their extended neurites and rounding of the cell body within 3 h. Cells microinjected with the empty vector did not exhibit any morphological changes, indicating that there are no nonspecific effects due to nuclear microinjection itself. We also examined the effects of GTPase-deficient mutants of G␣ q (G␣ q QL) and G␣ i2 (G␣ i2 QL) on differentiated PC12 cell morphology. As shown in Fig. 2, when expressed in differentiated PC12 cells, G␣ q QL mimicked G␣ 12 QL and G␣ 13  expression of G␣ i2 QL neither stimulated outgrowth nor caused retraction of neurites.
Effect of C3 Exoenzyme on Constitutively Active G␣ Subunitinduced Neuronal Morphological Changes-Previous studies have shown that the small GTPase Rho is required for neurite retraction in response to a certain G protein-coupled receptor agonist such as LPA (6 -9). As shown in Fig. 2, microinjection of expression plasmids encoding a constitutively active form of RhoA, RhoA V14 , caused the retraction of neurites and rounding of the cell body. These morphological changes induced by RhoA V14 were quite similar to those induced by G␣ 12 QL, G␣ 13 QL, and G␣ q QL ( Figs. 1 and 2). Therefore, to examine whether the neuronal morphological changes induced by constitutively active forms of G␣ subunits were Rho-dependent, we co-microinjected the constitutively active G␣-encoding plasmids into the cells with C3 exoenzyme from C. botulinum, which has been shown to catalyze ADP-ribosylation of Rho and the specifically suppress the action of Rho (28,29). As shown in Figs. 3 and 4, co-microinjection of C3 exoenzyme (100 g/ml) completely blocked both neurite retraction and cell rounding induced by G␣ 12 QL, G␣ 13 QL, and G␣ q QL. We also co-microinjected expression plasmids encoding the dominant-negative form of RhoA, RhoA N19 , into the cells with the constitutively active mutants of G␣ subunits. Coexpression of RhoA N19 slightly blocked neurite retraction induced by G␣ subunits, but was less effective than co-injection of C3 exoenzyme (data not shown). RhoA N19 would not be an effective inhibitor for complete suppression of G␣ subunit-induced morphological changes due to coexpression of RhoA N19 and G␣ subunits or the requirement of a large amount of RhoA N19 for suppression.
Effect of the Tyrosine Kinase Inhibitor Tyrphostin A25 on Neuronal Morphological Changes Induced by Constitutively Active G␣ Subunits-A previous study showed that the tyrosine kinase inhibitor tyrphostin A25 inhibited stress fiber formation stimulated by LPA, but not by microinjection of constitutively active Rho into quiescent Swiss 3T3 cells, indicating that a tyrosine kinase is involved in the LPA-stimulated stress fiber formation acting upstream of Rho (30). Therefore, we examined the effect of tyrphostin A25 on the neuronal morphological changes induced by constitutively active G␣ subunits. As shown in Figs. 5 and 6, treatment of differentiated cells with tyrphostin A25 (150 M) inhibited the G␣ 13 QL-and G␣ q QLinduced neurite retraction and cell rounding. In contrast, the neurite retraction and cell rounding induced by G␣ 12 QL were not influenced by this tyrosine kinase inhibitor. Thus, tyrphostin A25 specifically inhibited the signaling of G␣ 13 QL and G␣ q QL. In addition, we examined the effect of another tyrosine kinase inhibitor, tyrphostin AG1478, on morphological changes induced by constitutively active G␣ subunits. Treatment of differentiated cells with tyrphostin AG1478 (10 M) also specifically inhibited the G␣ 13 QL-and G␣ q QL-induced neurite retraction and cell rounding, and G␣ 12 QL-induced morphological changes were not inhibited by this inhibitor (data not shown).
Effects of Protein Kinase C Inhibition and Elimination of Extracellular Ca 2ϩ on Neuronal Morphological Changes Induced by Constitutively Active G␣ Subunits-A number of the cellular responses caused by activation of G␣ q have been shown to be mediated by activation of protein kinase C (PKC) or elevation of the intracellular Ca 2ϩ concentration. Therefore, we examined whether activation of PKC was required for the neuronal morphological changes induced by constitutively active G␣ q and G␣ 12 /G␣ 13  ical changes were not significantly altered by PKC depletion of cells. Similar results were obtained by treatment of cells with the PKC inhibitor Ro31-8220 (300 nM) (Fig. 9). These results indicate that inhibition of PKC activity specifically interferes with the signaling pathway of G␣ q for neurite retraction and cell rounding.
Next we examined the role of Ca 2ϩ signaling in the neuronal morphological changes induced by activated G␣ subunits. Differentiated PC12 cells were incubated in a Ca 2ϩ -free medium in the presence of 2 mM EGTA during expression of G␣ subunits. Under these conditions, expression of G␣ q QL failed to induce neurite retraction and cell rounding. In contrast, the neuronal morphological changes induced by G␣ 12 QL and G␣ 13 QL normally occurred in a Ca 2ϩ -free medium with EGTA ( Figs. 8 and 9). These results indicate that Ca 2ϩ influx is required for the G␣ q QL-induced neuronal morphological changes. DISCUSSION Activation of a certain G protein-coupled receptor has been reported to induce Rho-dependent neurite retraction and cell rounding in neuronal cell lines (6 -9). Here we have demonstrated that constitutively active forms of G␣ 12 , G␣ 13 , and G␣ q , but not G␣ i2 , can trigger neurite retraction and cell rounding in NGF-differentiated PC12 cells. These morphological changes were similar to those induced by a constitutively active form of RhoA, RhoA V14 (Fig. 2); and C3 exoenzyme, which specifically ADP-ribosylates and inactivates Rho (28,29), completely inhibited both neurite retraction and cell rounding induced by G␣ 12 QL, G␣ 13 QL, and G␣ q QL (Figs. 3 and 4), indicating that activation of G␣ 12 , G␣ 13 , and G␣ q induces neurite retraction and cell rounding through the Rho-dependent signaling pathway in differentiated PC12 cells.
G␣ 12 and G␣ 13 , the members of the G 12 class of heterotrimeric G proteins, show 67% amino acid identity to each other and often cause similar responses in various cell types, including transformation of fibroblasts, activation of the c-Jun N-terminal kinase cascade, and stimulation of stress fiber formation and focal adhesion assembly (31). We have also shown that both G␣ 12 and G␣ 13 can trigger Rho-dependent neurite retraction and cell rounding. These findings suggest that G␣ 12 and G␣ 13 may interact with a common effector. In this study, however, the tyrosine kinase inhibitor tyrphostin A25 blocked the G␣ 13 QL-induced neurite retraction and cell rounding, whereas the G␣ 12 QL-induced effects were not influenced by this tyrosine kinase inhibitor (Figs. 5 and 6), indicating that a tyrphostin-sensitive tyrosine kinase is involved in the signaling of G␣ 13 , but not in that of G␣ 12 . This finding strongly suggests that G␣ 12 and G␣ 13 interact with different effectors to regulate neuronal cell morphology. The differences in the sensitivity to tyrphostin between G␣ 12 and G␣ 13 were also shown in the signaling of G␣ 12 -and G␣ 13 -stimulated stress fiber formation and focal adhesion assembly in Swiss 3T3 fibroblasts (19). Furthermore, it was previously reported that G␣ 12 and G␣ 13 stimulate Na ϩ -H ϩ exchangers through different mechanisms in COS-7 cells (32,33). Therefore, it is likely that G␣ 12 and G␣ 13 activate different pathways to regulate their cellular functions.
Activation of G␣ q can stimulate the phospholipase C-␤ family, which results in stimulation of PKC activity and elevation of the intracellular Ca 2ϩ concentration. In this study, depletion of endogenous PKC by both TPA and the PKC inhibitor Ro31-8220 specifically diminished the amount of neurite-retracted cells induced by G␣ q QL, whereas the G␣ 12 QL-and G␣ 13 QLinduced morphological changes were not influenced (Figs. 7  and 9). In addition, elimination of extracellular Ca 2ϩ also inhibited the effects of G␣ q QL, but not those of G␣ 12 QL and G␣ 13 QL (Figs. 8 and 9). It has been shown that inositol 1,4,5trisphosphate and inositol 1,3,4,5-tetrakisphosphate, products of phospholipase C activation pathways, activate Ca 2ϩ -permeable channels in plasma membranes (34,35). Recently, G␣ q was reported to activate inositol 1,4,5-trisphosphate-operated Ca 2ϩ -permeable channels (36). The requirement of extracellular Ca 2ϩ for the G␣ q -induced morphological changes could be interpreted by this G␣ q -mediated Ca 2ϩ -permeable channel activation. Therefore, both PKC activation and Ca 2ϩ influx are essential elements in the signaling of G␣ q upstream of Rho. We also examined the involvement of phospholipase C in the G␣ q QL signaling using the phospholipase C inhibitor U-73122, but this compound was cytotoxic for differentiated PC12 cells, and treatment with U-73122 alone caused cell detachment. 2 Interestingly, both PKC activity and Ca 2ϩ influx as a result of phospholipase C activation appeared to be required for LPAinduced neurite retraction in NGF-differentiated PC12 cells (7). Therefore, it is likely that a G q -coupled LPA receptor stimulates phospholipase C activity, and the resultant activation of PKC and Ca 2ϩ influx induces Rho-dependent neurite retraction in PC12 cells. In contrast, prostaglandin EP3 receptorinduced neurite retraction was mediated through a PKC-independent pathway (9), indicating that G 12 or G 13 mediates the action of the EP3 receptor.
Interestingly, the G␣ q QL-induced neurite retraction and cell rounding were also blocked by treatment of cells with the tyrosine kinase inhibitor tyrphostin A25 (Figs. 5 and 6), indicating that a tyrphostin-sensitive tyrosine kinase is involved in the signaling from G␣ q to Rho. This study did not show whether this tyrphostin-sensitive tyrosine kinase acts upstream or downstream of Ca 2ϩ and PKC in the signaling from G␣ q to Rho. Since activation of G␣ q can directly stimulate phospholipase C, which results in stimulation of PKC activity and elevation of the intracellular Ca 2ϩ concentration, this tyrphostin-sensitive tyrosine kinase may act downstream of Ca 2ϩ and PKC. Recently, the novel nonreceptor tyrosine kinase PYK2 has been shown to mediate G q -coupled receptor-stimulated activation of the mitogen-activated protein kinase cascade in PC12 cells, and the activity of this tyrosine kinase appears to be regulated by elevation of the intracellular Ca 2ϩ concentration as well as by PKC activation (38). Therefore, PYK2 can be speculated to be a candidate for the tyrosine kinase, which links the signal of G␣ q to Rho for the induction of neurite retraction and cell rounding. A tyrphostin-sensitive tyrosine kinase was of course involved in the signaling of G␣ 13 to Rho. However, in contrast to G␣ q , G␣ 13 did not require either PKC activation or Ca 2ϩ influx. Two potential explanations for the difference between the signaling pathways of G␣ 13 and G␣ q can be presented. One explanation is that an identical tyrosine kinase mediates the signals of G␣ 13 and G␣ q to Rho, but the pathways of both ␣ subunits to activate the tyrosine kinase are different; G␣ q activates the kinase through PKC and Ca 2ϩ influx, whereas G␣ 13 activates the kinase independent of PKC. The other explanation is that different tyrosine kinases are involved in the pathways of both ␣ subunits. We have summarized these possible pathways of signal transduction of three ␣ subunits for Rho-dependent neurite retraction and cell rounding (Fig. 10).
Expression of a GTPase-deficient form of G␣ q in undifferentiated PC12 cells was recently shown to induce neurite outgrowth during 2-3 weeks using the retrovirus-mediated infection procedure (39). In contrast, our results showed that expression of G␣ q QL in NGF-differentiated PC12 cells triggered neurite retraction within 3 h after microinjection. Therefore, these opposite effects of G␣ q on the regulation of neurites may be due to different conditions of cells in differentiation or to a different time scale for examination of morphological effects.
The data presented here demonstrated that constitutively active mutants of G␣ 12 , G␣ 13 , and G␣ q can induce neurite retraction and cell rounding through different signaling pathways, which, however, finally converge at activation of Rho (Fig. 10). Rho, like other small GTPases, functions as a molecular switch; it is active in its GTP-bound state and inactive in its GDP-bound state. Upstream activation of the cycle is mediated by guanine nucleotide exchange factors, which promote the exchange of GDP for GTP (40). A number of putative guanine nucleotide exchange factors for Rho and other Rho family GTPases have been identified, and some of these demonstrate Rho-specific guanine nucleotide exchange factor activity in vitro, including Lbc, Lfc, and Lsc (3,(41)(42)(43). In addition, they appear to be expressed in the same cell type (42). Therefore, one possibility for the existence of multiple guanine nucleotide exchange factors for Rho in the same cell type may be related to the existence of different signaling pathways from G␣ subunits to Rho activation.
Recent studies have shown the involvement of the Rho family of small GTPases in the regulation of neurite outgrowth in primary neurons (37,44). In embryonic chick dorsal root ganglion, inhibition of Rho with C3 exoenzyme stimulated the outgrowth of neurites (44), suggesting that activation of Rho suppresses neurite outgrowth in primary neurons. Therefore, G␣ 12 , G␣ 13 , and G␣ q may play a negative regulator for neurite FIG. 8. Effect of elimination of extracellular Ca 2؉ on neuronal morphological changes induced by constitutively active G␣ subunits. After differentiated PC12 cells had been microinjected with expression plasmids (30 g/ml) encoding G␣ 12 QL, G␣ 13 QL, or G␣ q QL, they were incubated in Ca 2ϩ -free medium containing EGTA (2 mM) for 3 h. Cells were photographed before (left panels) and 3 h after (middle panels) microinjection under a phase-contrast microscope or by fluorescence of Texas Red-coupled dextran co-microinjected with the expression vectors (right panels). The arrows indicate injected cells. The results shown are representative of three independent experiments. The bar represents 50 m.
FIG. 9. Quantification of effects of inhibition of PKC and elimination of extracellular Ca 2؉ on neurite retraction induced by constitutively active G␣ subunits. After differentiated PC12 cells had been microinjected with expression plasmids (30 g/ml) encoding G␣ 12 QL, G␣ 13 QL, or G␣ q QL, they were incubated in the absence (Control) or presence of 300 nM Ro31-8220 (ϩRo31-8220) or in Ca 2ϩ -free medium containing 2 mM EGTA (Ca 2ϩ -free) for 3 h. The expression plasmids were microinjected into differentiated cells that had been treated with 1 M TPA for 24 h before microinjection to induce downregulation of endogenous PKC (PKC-depleted). The percentages of neurite-retracted cells were determined 3 h after microinjection as described under "Experimental Procedures." Data are the means Ϯ S.E. of triplicate experiments. outgrowth through activation of Rho in primary neurons.
In conclusion, we have here shown that activation of G␣ 12 , G␣ 13 , and G␣ q can trigger Rho-dependent neurite retraction and cell rounding in differentiated PC12 cells through different signaling pathways. This study will contribute to the understanding of the signal transduction between heterotrimeric G protein-coupled receptors and Rho in neuronal cells.