Antagonistic Regulation of Neurite Morphology through Gq/G11 and G12/G13*

The induction of neurite retraction and growth cone collapse via G-protein-coupled receptors is involved in developmental as well as regenerative processes. The role of individual G-protein-mediated signaling processes in the regulation of neurite morphology is still incompletely understood. Using primary neurons from brains lacking Gαq/Gα11 or Gα12/Gα13, we show here that G12/G13-mediated signaling is absolutely required for neurite retraction and growth cone collapse induced by the blood-borne factors lysophosphatidic acid and thrombin. Interestingly, the effects of lysophosphatidic acid were mediated mainly by G13, whereas thrombin effects required G12. Surprisingly, lack of Gαq/Gα11 resulted in overshooting responses to both stimuli, indicating that Gq/G11-mediated signaling most likely via activation of Rac antagonizes the effects of G12/G13.

Chemorepellents that are able to induce neuronal growth cone collapse and neurite retraction play important roles as guidance cues during the development of the nervous system (1,2) and are also believed to prevent regeneration in the nervous system after injury (3)(4)(5). The regulation of growth cone morphology and neurite growth has been shown to be mediated by small GTPases of the Rho family. Whereas Rac and Cdc42 activation and inactivation of RhoA promote neurite formation and extension of growth cones, RhoA activation induces growth cone collapse and neurite retraction (2, 6 -8).
The blood-borne factors thrombin and lysophosphatidic acid (LPA) 2 induce growth cone collapse and neurite retraction by activating specific G-protein-coupled receptors (9 -11), and they are involved in the inhibition of regenerative processes after nervous system injuries (12,13). In addition, LPA is involved in various aspects of neural development (14,15). Both thrombin and LPA receptors are coupled to the heterotrimeric G-proteins G q /G 11 , G 12 /G 13 , and G i (16 -18). Activation of G q /G 11 results in an increase in the enzymatic activity of ␤-isoforms of phospholipase C, leading to the formation of inositol 1,4,5-trisphosphate and diacylglycerol. This subsequently results in an increase in the intracellular Ca 2ϩ concentration as well as activation of various protein kinases, including protein kinase C (19). In contrast, activation of G 12 /G 13 leads to stimulation of the Rho/Rho kinase pathway via a subgroup of Rho guanine nucleotide exchange factors (20).
The relative roles of the G 12 /G 13 -and G q /G 11 -mediated signaling pathways in the acute effects of LPA and thrombin on neuronal morphology are rather unclear. Although there is good evidence that G 12 /G 13 -mediated RhoA activation via increased actomyosin contractility plays an important role in LPA-and thrombin-induced neurite retraction and growth cone collapse (9,11,21,22), evidence has also been provided that G q /G 11 -mediated signaling can mediate neurite retraction and growth cone collapse (21,23). However, studies analyzing the role of G-proteins in the regulation of neurite morphology via G-protein-coupled receptors have been based primarily on the use of constitutively active mutants of G-protein ␣-subunits. In addition, regulation of neuronal morphology via G-protein-coupled receptors has been studied mainly in neuronal cell lines, whereas the role of these regulatory processes in primary neurons as well as in the developing nervous system is unclear.
We therefore analyzed the role of the G q /G 11 -and G 12 /G 13mediated signaling pathways in the regulation of neurite morphology by thrombin and LPA in primary hippocampal neurons derived from mice lacking the ␣-subunits of G q /G 11 or G 12 /G 13 selectively in neuronal cells. We found that G 13 mediates neurite retraction and growth cone collapse induced by LPA, whereas G 12 mediates the effects of thrombin on neuronal morphology. Surprisingly, neurons lacking G␣ q /G␣ 11 showed a strongly increased response to LPA and thrombin, indicating that G 12 /G 13 -and G q /G 11 -mediated signaling have opposing effects on neurite morphology.
Primary Cultures from Mouse Hippocampus and Cortex-Cultures of hippocampal and cortical neurons were prepared as described previously (24). Hippocampi and cerebral cortices from embryonic day 17.5 mouse embryos were dissected in ice-cold phosphate-buffered saline containing 30 mM HEPES and 33 mM glucose (pH 7.38), washed once with phosphatebuffered saline, incubated with 0.05% trypsin (Invitrogen) for 15 min at 37°C, and triturated with fire-polished Pasteur pipettes. Cells were seeded in minimum essential Eagle's medium supplemented with 10% fetal bovine serum, 1 mM sodium pyruvate (all from Invitrogen), 25 mM glucose (Sigma), and 25 M glutamic acid (Sigma) on poly-L-lysine (Sigma)coated plastic dishes (Nunc, Falcon) at a density of 2 ϫ 10 4 cells/well (ϳ1 ϫ 10 4 cells/cm 2 , for live cell imaging) or 2 ϫ 10 6 cells/well (2 ϫ 10 5 cells/cm 2 , for G-LISA assays), respectively, and maintained at 37°C in 5% CO 2 . After 4 h, the medium was replaced with neurobasal medium containing B-27 supplement and 0.5 mM L-glutamine (all from Invitrogen), and the cells were cultured for 12-18 h without changing the medium. To inactivate G i , pertussis toxin (100 ng/ml; Calbiochem) was added to the medium 8 -10 h before experiments.
Live Cell Imaging-Growth cone collapse and neurite retraction assay was performed on a Leica DM IRE2 microscope (Leica Microsystems GmbH, Wetzlar, Germany) equipped with a 37°C/ 5% CO 2 environmental control chamber. Live cell images were recorded using a Leica DC 350 FX camera and Leica FW4000 software. Images from randomly selected positions were acquired once before and 2, 5, or 20 min after bath application of LPA (BIOMOL), thrombin (Sigma), or ephrin A5/Fc (R&D Systems). Alternatively, image sequences were recorded at a rate of six images/min 1 min before and 5-6 min after addition of agonists. Before use, ephrin A5/Fc was preclustered with anti-human immunoglobulin G/Fc (Sigma) as described previously (26). Acquired images were analyzed using the NIH ImageJ program (available at rsb.info.nih.gov.ij/) (27), and the percentage of growth cones and average length of neurites before and after stimulation were calculated for 15-20 cells/experiment. Retraction was defined as neurite shortening expressed as percent of the initial length. Results are presented as the means Ϯ S.E. of 3-11 independent experiments. Dose response to LPA was approximated with the following equations: y ϭ Ϫ1.35 ϩ 22.19/(1 ϩ 10 Ϫ6.55Ϫx ) and R 2 ϭ 0.97 for wild-type neurons and y ϭ Ϫ1.35 ϩ 60.54/(1 ϩ 10 Ϫ7.107Ϫx ) and R 2 ϭ 0.98 for G␣ q /G␣ 11 -deficient neurons. Statistical significance was determined by the two-tailed t test; analysis of variance and Bonferroni's correction were used for multiple comparisons.
Determination of RhoA and Rac Activity-RhoA and Rac1 activation assays were performed using G-LISA RhoA and Rac1 activation assay kits (Cytoskeleton, Inc.) according to the manufacturer's instructions. Briefly, cells were incubated for the indicated time periods with vehicle alone (phosphate-buffered saline) or 10 M LPA, immediately washed with ice-cold phosphate-buffered saline, and prepared in ice-cold lysis buffer. Lysates were incubated in microplate wells coated with a RhoAor Rac1-GTP-binding protein, and bound active RhoA/Rac1 was detected using a RhoA/Rac1-specific antibody and chemiluminescence. The data represent -fold stimulation Ϯ S.E. from three to four independent experiments. Statistical significance was determined by the Mann-Whitney U test.

RESULTS
To analyze the role of particular signaling pathways that mediate the regulation of neurite morphology via G-proteincoupled receptors, we prepared hippocampal neurons at embryonic day 17.5 from mice with nervous system-specific G␣ q /G␣ 11 or G␣ 12 /G␣ 13 deficiency (Nes-Cre;Gnaq flox/flox ; Gna11 Ϫ/Ϫ or Nes-Cre;Gna12 Ϫ/Ϫ Gna13 flox/flox , respectively) (24,25). Embryonic hippocampal neurons lacking G␣ q /G␣ 11 or G␣ 12 /G␣ 13 were basically indistinguishable from wild-type neurons. Wild-type and mutant neurons had the same number and average length of neurites at stage 2/3 ( Fig. 1, A and B), and there was no difference in the differentiation of neurons in vitro (Fig. 1C). However, neurons lacking G␣ q and G␣ 11 showed a slightly reduced number of growth cone-positive neurites compared with wild-type or G␣ 12 /G␣ 13 -deficient neurons (Fig. 1D).
Murine wild-type hippocampal neurons at stage 2/3 rapidly respond to LPA and thrombin with growth cone collapse and retraction of their neurites (Figs. 2 (A and B) and 3 (A, B, D, and E) and supplemental Movie 1). In contrast, neurons lacking G␣ 12 /G␣ 13 were completely unresponsive, showing no neurite retraction and growth cone collapse when exposed to 10 M LPA (Figs. 2A and 3 (A and D) and supplemental Movie 2) or 1 unit/ml thrombin (Figs. 2B and 3 (B and E)). G␣ 12 /G␣ 13 -deficient neurons were not generally unresponsive as indicated by the effect of ephrin A5, which acts independently of G-protein-mediated signaling pathways and induced growth cone collapse as well as neurite retraction in both wild-type and G␣ 12 /G␣ 13 -deficient neurons (Figs. 2C and 3 (C and F)).
Unexpectedly, a deficiency of G␣ q /G␣ 11 resulted in an overshooting response to LPA ( Fig. 2A and supplemental Movie 3) and thrombin (Fig. 2B). Whereas both stimuli reduced the percentage of growth cone-positive neurites in wild-type neurons to 10 -30%, no growth cones were detectable after exposure of G␣ q /G␣ 11 -deficient cells to LPA and thrombin (Fig. 3, A and B). Similarly, the effect of LPA and thrombin on neurite retraction was largely increased in the absence of G␣ q /G␣ 11 . Whereas LPA and thrombin reduced wild-type neurite length by ϳ20 and 10%, respectively, the extent of neurite retraction was increased by 2-3-fold in the absence of G␣ q /G␣ 11 (Fig. 3, D and  E). The overshooting response of neurons lacking G␣ q /G␣ 11 was not due to a general overresponsiveness of these cells as demonstrated by the normal effect of ephrin A5 on growth cone and neurite morphology (Figs. 2C and 3 (C and F)). Uncoupling of receptors from G i -type G-proteins by pretreatment of cells with pertussis toxin had no significant effect on LPA-induced neurite retraction or growth cone collapse in wild-type neurons or G␣ q /G␣ 11 -deficient neurons (Fig. 3, A and D). Thus, G 12 / G 13 -and G q /G 11 -mediated signaling pathways obviously play opposing roles in the regulation of neurite morphology via G-protein-coupled receptors. Whereas G 12 and G 13 are critically involved in the induction of growth cone collapse and neurite retraction, G q and G 11 appear to counteract the activity of G 12 /G 13 -mediated signaling.
To test whether individual Gproteins are responsible for the observed effects of G␣ 12 /G␣ 13 and G␣ q /G␣ 11 deficiency on the regulation of growth cone and neurite morphology, we analyzed the effect of LPA and thrombin in neurons lacking only G␣ 12 or G␣ 13 . G␣ 13deficient neurons behaved similarly to G␣ 12 /G␣ 13 -deficient cells and showed almost no response to LPA, whereas neurons lacking G␣ 12 behaved like wild-type cells (Fig. 4,  A and C). Interestingly, thrombin effects were strongly reduced in the absence of G␣ 12 but not in the absence of G␣ 13 (Fig. 4, B and D). This indicates that LPA-induced regulation of neurite morphology involves primarily G 13 , whereas thrombin effects are mediated preferentially by G 12 . Neurons lacking only G␣ 11 showed slightly overshooting responses to both LPA and thrombin (Fig. 4). However, the phenotype was not as excessive as that of neurons lacking both G␣ q and G␣ 11 , indicating that both G-protein ␣-subunits are involved in the effects of LPA and thrombin.
To further analyze the overshooting response of G␣ q /G␣ 11 -deficient neurons to agonists of G-protein-coupled receptors, we determined the dose-response relationship of LPA-induced neurite retraction in wild-type and G␣ q /G␣ 11 -deficient cells (Fig. 5A). In both cell types, LPA induced maximal neurite retraction at concentrations of 3-10 M while being about three times more efficacious in neurons lacking G␣ q /G␣ 11 . However, also the potency of LPA to induce neurite retraction appeared to be slightly increased in the absence of G␣ q /G␣ 11 . Whereas LPA induced neurite retraction in wild-type cells with an EC 50 of 280 nM, the EC 50 of LPA effects in G␣ q /G␣ 11 -deficient neurons was ϳ80 nM.
The massively enhanced effect of LPA in the absence of G␣ q / G␣ 11 could be due to the fact that activation of the G q /G 11mediated signaling pathway exerted a stimulatory effect on neurite and growth cone extension, which under wild-type conditions counteracts the effect of G 12 /G 13 activation on neurite retraction and growth cone collapse. Alternatively, excessive neurite retraction and growth cone collapse in the absence of G␣ q /G␣ 11 could result simply from an overactivation of the G 12 /G 13 -mediated signaling pathway due to altered expression levels of the receptors or an increased coupling efficiency of receptors to G 12 /G 13 in the absence of G␣ q /G␣ 11 . Quantitative PCR showed that the expression levels of LPA1, LPA2, and  LPA4 receptors as well as of G␣ 12 , G␣ 13 , and RhoA were unchanged in G␣ q /G␣ 11 -deficient neurons (Fig. 5B). For LPA3 and LPA5 receptors, no corresponding mRNA was detected.
To test whether coupling efficiency of receptors to G 12 /G 13 was increased in the absence of G q /G 11 , we determined the activity of RhoA, the main effector of G 12 /G 13 , in untreated and LPA-treated wildtype, G␣ 12 /G␣ 13 -deficient, and G␣ q /G␣ 11 -deficient neurons. The increase in RhoA activity was comparable in wild-type cells and cells lacking G␣ q /G␣ 11 , whereas no RhoA activation could be observed in G␣ 12 /G␣ 13 -deficient neurons (Fig. 5C). This indicates that G␣ q / G␣ 11 deficiency does not result in overactivation of G 12 /G 13 -mediated signaling in response to LPA.
Because LPA and thrombin can activate Rac in parallel with RhoA (29,30) and because Rac can antagonize RhoA function (31)(32)(33), we determined LPA effects on Rac activity in wild-type and G␣ q /G␣ 11deficient neurons. Interestingly, the LPA-induced activation of the small GTPase Rac, which could be seen in wild-type neurons, was absent in neurons lacking G␣ q /G␣ 11 (Fig.  5D). This indicates that in wild-type neurons both RhoA and Rac are activated and that the G q /G 11 -mediated Rac activation that promotes neurite and growth cone extension may counteract the effect of G 12 /G 13 -mediated RhoA activation on neurite retraction and growth cone collapse under wild-type conditions.

DISCUSSION
A variety of G-protein-coupled receptors have been shown to be able to mediate the induction of neurite retraction and growth cone collapse (9,11,34,35). Regulation of growth cone and neurite morphology is involved in developmental as well as regenerative processes of the nervous system (12,13). Receptors that are able to stimulate neurite retraction and growth cone collapse couple to G 12 /G 13 , G q /G 11 , and G i -type G-proteins. The fact that pretreatment of neurons with pertussis toxin did not affect the ability of LPA or thrombin to reduce neurite retraction and growth cone collapse (see Fig. 3) (11,34,36,37) strongly indicates that G i -type G-proteins are not involved in these effects. It has been difficult to evaluate the relative roles of G 12 /G 13 and G q /G 11 in the regulation of neurite morphology because there are no appropriate inhibitors available that specifically block either of the G-protein families. Based on the expression of constitutively active mutants of G␣ 12 and G␣ 13 in various cell culture lines, there is good evidence that these G-protein ␣-subunits can mediate growth cone collapse and neurite retraction (21,22). However, the role G q /G 11 plays in these processes has been controversial.
Data involving G q /G 11 in mediating neurite retraction and growth cone collapse in response to LPA and thrombin are based on the use of constitutively active mutants of G␣ q that,  after expression in various cell lines, have been shown to cause neurite retraction similarly to G␣ 12 /G␣ 13 (21,23). However, other groups have reported that expression of constitutively active G␣ q causes massive cell death in neurons and other cells (22,38). Thus, it might be difficult to differentiate the toxic effects of activated mutants of G␣ q from cellular responses specifically induced via physiological G q activation. In addition, blockade of G q /G 11 -mediated signaling in a neuronal cell line has been shown to be without any effect on LPA-induced neurite retraction (39,40).
To clarify the role of different G-protein-mediated signaling pathways in the regulation of neurite morphology, we took a genetic approach to study the regulation of neurite and growth cone morphology via endogenous receptors in primary neurons lacking either G␣ q /G␣ 11 or G␣ 12 /G␣ 13 . Our data clearly show that G 12 and G 13 are required for LPA-and thrombin-induced neurite retraction and growth cone collapse. Surprisingly, we observed overshooting responses in neurons lacking G␣ q / G␣ 11 , which strongly indicates that these G-proteins do not promote neurite retraction and growth cone collapse but are rather involved in a pathway that counterregulates the stimulation of neurite retraction and growth cone collapse via G 12 / G 13 . Interestingly, it has been shown in Drosophila that expression of an activated version of Drosophila G␣ q results in the ectopic midline crossing of neurons (41). Thus, G q /G 11 -mediated signaling may regulate axonal pathfinding by inhibiting repulsive effects and by promoting neurite and growth cone extension.
The excessive response of G␣ q /G␣ 11 -deficient neurons to LPA and thrombin was obviously not due to an enhanced activation of the G 12 /G 13 -mediated pathway as indicated by the unaltered RhoA activation in neurons lacking G␣ q /G␣ 11 . Because the effects of a G-protein-independent activator of neurite retraction and growth cone collapse like ephrin A5 were not affected by G␣ q /G␣ 11 deficiency, unspecific effects can be ruled out. Thus, G q /G 11 -mediated signaling per se has an unexpected inhibitory effect on Rho/Rho kinase-mediated neurite retraction and growth cone collapse. Given the fact that G q and G 11 couple receptors to ␤-isoforms of phospholipase C, resulting in transient increases in [Ca 2ϩ ] i , the promotion of neurite extension via G q /G 11 -mediated signaling is likely to be mediated by Ca 2ϩ . Recent evidence indicates that the spatiotemporal pattern of intracellular Ca 2ϩ signals can have profound effects on axonal growth cones. Whereas small and large elevations in the intracellular Ca 2ϩ concentration result in repulsion, moderate elevation promotes growth cone extension (42,43). Increases in [Ca 2ϩ ] i have been shown to induce activation of Rac in nonneuronal and neuronal cells (44 -47). Consistent with this hypothesis, we observed that the LPA-induced Rac activation that was seen in wild-type neurons was absent in G␣ q /G␣ 11 -deficient neurons. Rac activation is well known to counteract RhoA activation, to inhibit neurite retraction, and to promote growth cone extension. It is therefore most likely that the G q /G 11 -mediated increase in [Ca 2ϩ ] i via activation of the small GTPase Rac inhibits neurite retraction and growth cone collapse.
When we tested the effect of LPA and thrombin on neurite and growth cone morphology in cells lacking only G␣ 12 or G␣ 13 , we observed that G␣ 12 deficiency blocked thrombin-induced effects, whereas lack of G␣ 13 resulted in a loss of LPAinduced effects. This is consistent with observations made in fibroblasts and human embryonic kidney cells, in which LPA effects appear to be mediated by G 13 , whereas thrombin effects involve G 12 (48,49). Evidence has been provided that this is due to the preferential coupling of LPA receptors to G 13 and of thrombin-activated receptors to G 12 in intact cells (49,50).
In this study, we have shown that G 12 and G 13 are absolutely required for neurite retraction and growth cone collapse induced by the blood-borne factors thrombin and LPA. Interestingly, analysis of single G␣ 12 -and G␣ 13 -deficient neurons showed that the effects of thrombin were primarily mediated by G 12 , whereas the LPA effects depended on G 13 . Surprisingly, the overshooting response of G␣ q /G␣ 11 -deficient neurons revealed a so far unknown role of the neuronal G q /G 11 -mediated signaling pathway in antagonizing the effects of G 12 /G 13mediated signaling. Inhibition of G 12 /G 13 -mediated activation of Rho/Rho kinase signaling has been suggested as a strategy to improve neurite growth and sprouting after neural injury (3). Our data clearly support this concept and, in addition, indicate that activation of G q /G 11 -mediated signaling would be of additional benefit. . Characterization of G q /G 11 -mediated inhibition of neurite retraction and growth cone collapse. A, neurite retraction in wild-type (WT), G␣ 12 /G␣ 13 -deficient (12/13 KO), and G␣ q /G␣ 11 -deficient (q/11 KO) hippocampal neurons measured 5 min after exposure to different concentrations of LPA. B, ratio of LPA1/LPA2/LPA4 receptor, G␣ 12 (Gna12), G␣ 13 (Gna13), and RhoA expression in G␣ q /G␣ 11 -deficient hippocampal neurons compared with wild-type neurons. The pairwise fixed reallocation randomization test showed no significant difference between wild-type and G␣ q /G␣ 11 -deficient neurons (p ϭ 0.34 -0.96). C and D, effect of 10 M LPA on RhoA (C) and Rac1 (D) activity in wild-type, G␣ 12 /G␣ 13 -deficient, and G␣ q /G␣ 11 -deficient cortical neurons. Cells were incubated with LPA for 15-60 s (RhoA) or for 15 s (Rac1). n.s., not significant; *, p Ͻ 0.05.