Cannabinoid Receptor-induced Neurite Outgrowth Is Mediated by Rap1 Activation through G (cid:1) o/i -triggered Proteasomal Degradation of Rap1GAPII*

The G (cid:1) o/i -coupled CB1 cannabionoid receptor induces neurite outgrowth in Neuro-2A cells. The mechanisms of signaling through G (cid:1) o/i to induce neurite outgrowth were studied. The expression of G (cid:1) o/i reduces the stability of its direct interactor protein, Rap1GAPII, by targeting it for ubiquitination and proteasomal degradation. This results in the activation of Rap1. G (cid:1) o/i -induced activation of endog- enous Rap1 in Neuro-2A cells is blocked by the proteasomal inhibitor lactacystin. G (cid:1) o/i stimulates neurite outgrowth that is blocked by the expression of dominant negative Rap1. Expression of Rap1GAPII also blocks the G (cid:1) o/i -in- duced neurite outgrowth and treatment with proteasomal inhibitors potentiates this inhibition. The endogenous G (cid:1) o/i -coupled cannabinoid (CB1) receptor in Neuro-2A cells stimulates the degradation of Rap1GAPII; activation of Rap1 and treatment with pertussis

The differentiation process in neurons is a complex phenomenon involving multiple changes. These include both changes in electrophysiological characteristics as well changes in morphology characterized by dendritic and axonal outgrowths. A general term "neurite" is used to define these outgrowths which are morphological characteristics of the neuronal differentiation process. Regulation of neurite outgrowth is tightly controlled and many neurotransmitters are induced in this process. Neurite outgrowth in cortical neurons is regulated by D2 dopamine receptors (1). Serotonin 1B receptors are known to enhance neurite outgrowth in thalamic neurons of rodents (2).
Growth cones from rat cerebellar neurons are guided by the chemoattractant SDF-1, and Xenopus neuron outgrowth is guided by the GABA-B receptors (3). A common feature of all these receptors is that they couple through the G␣ o/i pathway. It has been known for over a decade that G␣ o/i can induce neurite outgrowth (4) and that G␣ o/i is one of the more abundant proteins present in the neuronal growth cones (5). The mechanisms by which the G␣ o/i signals are transduced have never been clarified. We and others (6,7) have found that G␣ o interacts with RapGap and this results in the activation of Rap. The mechanisms by which G␣ o activates Rap and the consequences of this activation are unknown. To determine both the biological consequences of activation of G␣ o /Rap pathway and the mechanisms by which this occurs we searched for a cellular system where we could study the biochemical mechanisms within the context of the biological effects. Here we describe how the endogenous cannabinoid receptors in Neuro-2A cells use the G␣ o/i pathway to induce neurite outgrowth.

EXPERIMENTAL PROCEDURES
Cell Culture-COS-7 and HEK-293T cells were maintained in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum and 1% penicillin-streptomycin and Neuro-2A cells were maintained in 44% Dulbecco's modified Eagle's medium, 44% F12 media and 10% fetal bovine serum, 1% penicillin-streptomycin, and 1% glutamine. FuGENE 6 (Roche Applied Science) was used for the introduction of exogenous DNA into COS-7 cells according to the manufacturer's instructions. Briefly 100,000 cells were plated per well in a 6-well plate, and 24 h later each well was transfected with 2 g DNA and 6 l of FuGENE. For HEK-293T cells, 2 ϫ 10 6 cells were plated on 60-mm dishes and 24 h later transfected with 4 g of DNA using Lipofectamine 2000 (10 l). For imaging experiments Neuro-2A cells (5 ϫ 10 5 ) were plated in 35-mm Mattek® plates and were transfected using Lipofectamine 2000® the following day. Neuro-2A cells were obtained from ATCC (catalog number CCl-131).
Coimmunoprecipitation-293T and COS-7 cells were transfected with the indicated constructs using Lipofectamine 2000 (Invitrogen) and Fu-GENE 6 (Roche Applied Science), respectively. 293T cells were harvested 48 h posttransfection, and immunoprecipitations were done as described previously (6). Samples were separated by SDS-PAGE and immunoblotted with a G␣ o -specific antibody (Santa Cruz Biotechnology).
GST Pull-downs-Human Rap1GAPI and Rap1GAPII were cloned by PCR from Jurkat cells and the N terminus of each was subcloned into the pGEX4T2 expression vector (Amersham Biosceinces). Bacterially expressed purified protein was bound to glutathione-Sepharose and incubated with 293T cell lysates expressing G␣ subunits followed by SDS-PAGE and immunoblotting with G␣ o or G␣ i2 antibodies (Santa Cruz Biotechnology).
Real-time RT 1 -PCR-Total RNA was isolated from cells using TRIzol reagent (Invitrogen), and real-time RT-PCR was performed on a Ligh-tcycler (Roche Applied Science) using the QuantiTect SYBR Green RT-PCR system (Qiagen) according to manufacturer's instructions. The primers used were as follows: Rap1GAPII, 5Ј-ATTACTACTCACTGGA-CACTGC-3Ј (forward) and 5Ј-GCCAAACTTGAAGTTATTGC-3Ј (reverse); RGSZ1, 5Ј-ATGGCCTGTGAGGAGCTG-3Ј (forward) and 5Ј-GA-AGCTGTGCGTCATCGA-3Ј (reverse); actin, 5Ј-CGGAACCGCTCATT-GCC3Ј (forward) and 3Ј-ACCCACACTGTGCCCATCTA-3Ј (reverse). Each reaction contained a 1 M concentration of each gene-specific primer, reaction mix, and 50 ng of total RNA. The annealing temperature was 55°C and the extension time was 15 s. Data were analyzed using Lightcycler data analysis software. Crossing points were calculated using the second derivative maximum method. All reactions were done in triplicate and data for Rap1GAPII were normalized against actin RNA levels for each sample.
Ubiquitination Assays-HA-tagged ubiquitin vectors were used to follow the ubiquitination of proteins. COS-7 cells were cotransfected with the indicated plasmids and 48 h later harvested in 100 l of denaturing lysis buffer (1ϫ TBS (100 mM Tris-HCl, pH 7.5, 150 mM NaCl), 2% SDS) preheated to 95°C. Lysates were heated at 95°C for 10 min followed by the dilution in 900 l of non-denaturing lysis buffer (1ϫ TBS, 1% Triton X-100) and immediately placed on ice. Samples were sonicated for 15 s to reduce the viscosity, followed by preclearing with the addition of 50 l of protein-G-agarose. Cell lysates were centrifuged, and supernatants were immunoprecipitated with 5 g of M2-FLAG antibody overnight followed by the addition of 15 l of protein-Gagarose for 4 h. Samples were separated by SDS-PAGE and immunoblotted with an anti-HA antibody (Santa Cruz Biotechnology).
Antibodies-Antibodies specific for Rap1GAPI/II and Rap1GAPII were made by immunizing rabbits with a keyhole limpet hemocyaninconjugated peptide specific for each protein (Sigma-Genosys). The sequence for the synthesized peptides for Rap1GAPI/II was N-CNP-TARIYRKHFLGKEH-C and for Rap1GAPII was N-MAQLRPAVP-PGRPRRGSLPA-C. Antibodies were affinity-purified on peptide-conjugated columns (Pierce).
Rap Activation Assays-Levels of activated Rap were measured by use of the RalGDS binding domain as described by Carey and Stork (8). The RalGDS-RBD (RalGDS-Rap1 binding domain) assay was performed using bacterial expressed GST-RalGDS-RBD fusion protein pre-bound to glutathione beads to detect active GTP-bound form of Rap1. RalGDS-RBD domain was PCR cloned from Jurkat cDNA and ligated into the pGEX4T2 vector. RalGDS-RBD was expressed as a GST fusion protein in bacteria, extracted in bacterial lysis buffer containing DTT and protease inhibitors, and then incubated with glutathione beads. RalGDS-RBD-bound beads were incubated with total cell lysates in presence of DTT and protease inhibitor mixture for 1 h at 4 o C. After washing, GTP-bound Rap1 was separated from beads by boiling the samples in loading buffer. The samples were then resolved by SDS-PAGE, transferred to nitrocellulose paper, and subjected to immunoblotting using rabbit polyclonal anti-Rap1 antibody (Santa Cruz Biotechnology).
Imaging-Six hours after transfection Neuro-2A cells were serumstarved for 18 h. To inhibit proteasomal activity, cells were incubated with 20 M lactacystin (Sigma, catalog number L-6785) for 2 h prior to imaging. Living cells were imaged with a Zeiss 510 confocal laserscanning microscope. Multiple (3)(4)(5) 0.45-m Z section slices were obtained for each condition. Representative images are shown. Statistical validation of the image shown is included in the figure legend to Fig. 4. TIFF images were processed with Adobe Photoshop.
Neurite Outgrowth Assays-Neuro-2A cells were plated in 6-well dishes. For dominant negative experiments, cells were cotransfected with dominant negative mutant of Rap1 or control bicistronic vector, IRES (Clontech BD Sciences) containing GFP or GFP-G␣ o . For siRNA experiments, HPP-grade siRNA for Rap1 (DNA target sequence AAG-CAAGTCGAGGTAGATTGC with dTdToverhangs) was synthesized by Qiagen. Cells were transfected with siRNA for Rap1 or control oligonucleotides for 4 h with Qiagen TransMessenger transfection reagent. Four hours after adding siRNA, cells were washed with 1ϫ PBS and then transfected with bicistronic vector containing GFP or GFP-G␣ o . Twenty-four hours posttransfection GFP-positive cells were counted for neurite outgrowth. The cells displaying neurite outgrowth were those that had cellular projections of length two times greater than the cell diameter. For each culture condition randomly chosen regions of the plate containing 100 cells were scored under a phase contrast microscope (Nikon TMS). All assays were done in triplicate and in two or three independent experiments, and the indicated values are represented as mean Ϯ S.E.
For studies examining the effect of antagonist, pertussis toxin, or lactacystin, ϳ1-5 ϫ 10 3 cells were plated in 6-well plates and treated with 15 ng/ml pertussis toxin for 16 h or with 20 mM lactacystin (proteasomal inhibitor) or 10 mM SR141716A (CB1 receptor antagonist) for 2 h followed by treatment with 10 M HU-210 for 16 h. Neurite outgrowth was quantitated as described above. All assays were done in triplicate and data represented as mean Ϯ S.E.

Cannabinoid CB1 Receptor-induced Neurite Outgrowth in Neuro-2A Cells through G␣ o/i and Rap-
The mouse neuroblastoma cell Neuro-2A can be induced to differentiate with outgrowth of neurites under a variety of stimuli including retinoic acid, the ganglioside GM1, low serum (9), or inhibition of protein kinase C (10). We initially screened the Neuro-2A cells for endogenous G␣ o/i -coupled receptors and found that these cells express the cannabinoid CB1 receptor. Treatment of serumstarved Neuro-2A cells with the CB1 receptor agonist HU-210 resulted in a significant increase in neurite outgrowth in a concentration dependent manner. Addition of the antagonist SR 141716A inhibited CB1 receptor agonist-induced neurite outgrowth (Fig. 1A). To establish that the CB1 receptor effects were due to coupling to a G␣ o/i pathway we treated the cells with pertussis toxin, a blocker of signaling through this pathway, and tested for the effects of the CB1 receptor agonist. Pretreatment with pertussis toxin fully blocked the CB1 receptor-induced neurite outgrowth ( G␣ o/i Interacts with Rap1GAPII and Targets It for Degradation-Initially, we had cloned Rap1GAP from a chick dorsal root ganglion library that had been used for the yeast twohybrid screen. Subsequently we cloned human Rap1GAP from Jurkat cells and found that there were two human isoforms of Rap1GAP which are splice variants. The second isoform (Rap1GAPII) has 6 additional amino acids (NTDLFE) at the N terminus after the initiation methionine. This N-terminal extension results in the appearance of a GoLoco motif that has been shown to be involved in interaction with G␣-subunits (11). We determined the interactions between Rap1GAPI and -II and G␣ o and G␣ i2 . For this, we transfected 293T cells with FLAG-tagged Rap1GAPI, Rap1GAPII, a N-terminal truncated version of Rap1GAP, and G␣ o . The FLAG-M2 antibody was used to pull down the expressed RapGAP, and the immunoprecipitate was tested for the presence of G␣ o subunits. Rap1GAPII preferentially bound G␣ o subunits, and the WT G␣ o protein appears to interact with Rap1GAPII more avidly than the activated form of G␣ o ( Fig. 2A) as would be expected due to the presence of the GoLoco motif (11). We also immunoblotted the immunoprecipitates with a G␤ antibody to determine whether Rap1GAPII was able to interact with the heterotrimeric G protein complex but were not able to detect G␤ suggesting that the G␣ o associated with Rap1GAPII is likely to be in a different pool from that which is associated with G␤␥ (data not shown). Structural data also support this mutual exclusion, since the GoLoco motif binds to the same region of G␣ subunits as G␤␥ (11). We also conducted GST pull-down experiments where GST-tagged RapGAP was purified, bound to glutathione-Sepharose, and used to isolate G␣ o and G␣ i (Fig.  2B). These experiments indicated that G␣ o/i subunits do interact with Rap1GAPII. To determine whether the observed biochemical interactions also occur in live cells where the G␣ o /Rap pathway might be operative, we investigated the effect of G␣ o co-expression on CFP-tagged Rap1GAPs in Neuro-2A cells. By themselves, both CFP-Rap1GAP1 and CFP-Rap1GAPII showed diffuse cytosolic fluorescence with negatively imaged organelles (Fig. 2C, panels i and iii). Expression of G␣ o led to the selective accumulation of CFP-tagged Rap1GAPII to perinuclear vesicles (please compare Fig. 2C, panel ii versus panel iv) indicating a functional effect of G␣ o on Rap1GAPII. However, we were unable to detect any measurable FRET between YFP-tagged G␣ o and CFP-tagged Rap1GAPII (data not shown).
G␣ o Degradation of Rap1GAPII Activates Rap and Induces Neurite Outgrowth-During our coexpression studies when we immunoblotted for both G␣ o and Rap1GAPII we consistently found that expression of G␣ o appeared to reduce the levels of Rap1GAPII proteins. We had found this to be true in both 293T cells and COS-7 cells. To explore the reasons for this we expressed Rap1GAPI, Rap1GAPII, and the N-terminal truncated RapGAP in control, WT G␣ o , and activated G␣ o expressing cells. We found that coexpression of G␣ o and Rap1GAPII resulted in a significant decrease in the levels of Rap1GAPII (Fig. 3A). The decrease in levels of Rap1GAPII appears to be specific for members of the G␣ o/i family, since coexpression of G␣ s or G␣ q did not affect the levels of Rap1GAPII (Fig. 3B). Under conditions where the Rap1GAPII protein levels were decreased, we did not find any changes in the Rap1GAPII mRNA levels as assayed by real-time PCR (Fig. 3C). Treatment of Rap1GAPII-expressing cells with inhibitors of various proteases indicated that the proteasomal inhibitors (MG-132, lactacystin, and PSI) blocked the disappearance of Rap1GAPII indicating that decrease in Rap1GAPII protein levels is due to proteasomal degradation (Fig.  3D). Such proteasomal degradation of RapGAP has been observed in thyroid cells (13). Since RapGAP may be ubiquitinated for targeting for proteasomal degradation, and since ubiquitination is often a regulated process, we determined if the Rap1GAPII ubiquitination was affected by G␣ o . For this we expressed FLAG-tagged Rap1GAPPII and HA-tagged ubiquitin in the absence and presence of G␣ o . The presence of G␣ o increased the ubiquitination of Rap1GAPII (Fig. 3E). Although Rap1GAPII is ubiquitinated, the extent of ubiquitination appears to be far less than that of a well known ubiquitinated protein, Traf2 (Fig.  3F). Taken together these data indicate that G␣ o promotes the ubiquitination of Rap1GAPII and thus targets it for proteasomal degradation.
These experiments raised two questions: can the effect of G␣ o be observed on native Rap1GAPII, and does the degradation of Rap1GAPII stimulate Rap activity? To address these issues we transfected G␣ o into Neuro-2A cells and tested for its effects on the levels of Rap1GAPII and on the activation state of Rap in the presence and absence of proteasomal inhibitors. When G␣ o was expressed in Neuro-2A cells, there was a decrease in the endogenous level of Rap1GAPII, and this decrease could be blocked by treating the cells with MG132 (Fig. 4A). We also determined the activation state of Rap in cells transfected with G␣ o in the presence and absence of proteasomal inhibitor. G␣ o stimulated the activation of Rap-1, and this effect was blocked by treating the cell with a proteasomal inhibitor (Fig. 4B). Next, we determined whether we could observe the proteasome sensitivity of G␣ o activation of Rap in live cells. For this, cells were transfected with GFP-tagged RalGDS-RBD that has been previously used to measure Rap activation in intact cells (14). Neuro-2A cells were transfected with the fluorescent probe without or with G␣ o (or G␣ q as control) and then treated without or with lactacystin. The cells were imaged by confocal microscopy. The expression of G␣ o , but not G␣ q , led to a strong accumulation of fluorescence in the perinuclear vesicles (Fig.  4C, panel ii versus panel iv). This indicates that Rap1 is being activated in these vesicles. When the cells were treated with the proteasomal inhibitor lactacystin, the G␣ o -induced accumulation of fluorescence was not observed (  Figs. 1-3 predict that inhibiting the proteasome should selectively enhance the effect of Rap1GAPII. When the Rap1GAPI expressing cells were treated with MG-132 no further significant inhibition of neurite outgrowth was observed. However, when Rap1GAPII-expressing cells were treated with MG-132, there was a marked increase in the inhibition of neurite outgrowth with a close to complete blockage of G␣ o -induced neurite outgrowth (Fig. 4D). Taken together the experiments in Fig. 4 indicate that G␣ o by inducing the degradation of Rap1GAPII activates Rap to induce neurite outgrowth.
Cannabinoid receptor activity, through the G␣ o/i pathway, stimulates the degradation of Rap1GAPII, the activation of Rap, and neurite outgrowth.
We next determined whether the G␣ o -Rap1GAPII-Rap1 pathway for the regulation of neurite outgrowth could be regulated in a receptor-dependent manner. We also tested whether the pathway and the biological effects could be observed in a completely native setting. Treatment of serumstarved Neuro-2A cells with the CB1 receptor agonist resulted in a time-dependent activation of Rap1. Activation is substantial (ϳ2-fold) within 1 h and maximal by 2 h. If the ligand is present, the activation persists for 16 h (Fig. 5A). These results suggest that the CB1 receptor can activate Rap relatively fast. Since the CB1 receptor is known to couple through the G␣ o/i pathway, we tested whether receptor stimulation of Rap1 activation is blocked when the cells are treated with pertussis toxin. The CB1 receptor agonist did not stimulate Rap1 activation or the degradation of Rap1GAPII in cells treated with pertussis toxin (Fig. 5B, left panels). Since the coupling between G␣ o and Rap is due to the degradation of Rap1GAPII we tested whether proteasomal inhibitor treatment blocked receptor activation of Rap. When CB1 receptor agonist was added to lactacystin-treated cells, the degradation of Rap1GAPII and activation of Rap was blocked. (Fig. 5B, right panels). We next tested the role of Rap in CB1 receptor-induced neurite outgrowth. For this, cells were transfected with Rap1 specific siRNA or control siRNA. The siRNA for Rap1 inhibited receptor regulated neurite outgrowth (Fig. 5C). Since both G␣ o/i and Rap are downstream of the CB1 receptor for neurite outgrowth, and given the connectivity between G␣ o/i and Rap1, both proteasomal inhibitors and Rap1GAPII would be predicted to block CB1-receptor induced neurite outgrowth. The experi-  Fig. 5, D and E, indicate that as predicted, treatment of cells with lactacystin or transfection with excess Rap1GAPII blocks receptor induced neurite outgrowth. These results indicate that in a fully native system the CB1 receptor through G␣ o promotes the degradation of Rap1GAPII to activate Rap and induce neurite outgrowth. DISCUSSION Typically, signaling through heterotrimeric G protein pathways involves information transfer through non-covalent protein-protein interactions or through protein phosphorylation/ dephosphorylation reactions. Such regulation is often sufficient for acute signal transmission and for short term integration between signaling pathways. However, these mechanisms do not readily allow for a signaling network to propagate or integrate signals across time scales. The mechanism that we describe here allows for such capability. The regulated degradation of RapGAP occurs in a time scale of 1-2 h in two native systems (Neuro-2A cells and rat hippocampal slices). 2 There is an interesting similarity both temporally as well as in configuration to another well known form of long term regulation studied in heterotrimeric G protein systems: heterologous desensitization. The phenomenon we describe here may lead to 2 J. D. Jordan and R. Iyengar, unpublished observations. heterologous sensitization. Signal flow, through both heterotrimeric and small G protein pathways, is regulated by the activity of the corresponding GAP proteins. From a systems design perspective it would be advantageous to regulate the levels of these crucial regulators and thus set a threshold for signal flow through these pathways. The G␣ o/i -mediated targeting of Rap1GAPII for proteasomal degradation indicates that such a mechanism, for signal flow as well as integration through threshold control, exists. Thus, a relatively brief activation of the G␣ o/i -coupled receptor and the consequent activation of the heterotrimeric G protein G␣ subunit could result in a prolonged activation of the small GTPase downstream of the heterotrimeric G protein. Duration and magnitude of activation of the small G protein often play a defining role in the conversion of biochemical signals into biological effects. A classical example is the persistent activation of Ras and the triggering of proliferation. Here it appears that Rap1 plays a crucial role in the G␣ o/i pathway-induced neurite outgrowth. The mechanism that we elucidate in this study would allow G␣ o/i to activate Rap1 for prolonged periods of time. These periods would be largely defined by the rates of degradation and synthesis of Rap1GAPII in cells where this isoform predominates, as is the case with Neuro-2A cells. Thus we would like to propose that signal dependent targeting of regulators for degradation by the proteasome may be one mechanism by which short duration signals in G protein pathways may be converted to long duration downstream signals.
Signal transmission through regulation of GAP activity assumes the presence of a constitutively active guanine nucleotide exchange factor that would turn on the small GTPase. In Neuro-2A cells grown to 50% confluence were starved overnight in serum-free medium. A, cells were stimulated with the CB1 receptor agonist, HU-210 (CB1R-Ag), for the indicated time periods. Cells were harvested in lysis buffer containing multiple protease inhibitors. Rap1GTP was pulled down using Ral-RBD and immunoblotted using polyclonal anti-Rap1 antibody. Cell lysates were also used for determination of total Rap-1 by immunoblotting. B, Neuro-2A cells were pretreated either with pertussis toxin (PT) at 15 ng/ml overnight or lactacystin (Lact) at 20 M for 2 h, and then cells were stimulated with CB1R-Ag for 2 h. Cells were harvested in lysis buffer containing protease inhibitors. Rap-1-GTP was pulled down using Ral-RBD and immunoblotted using polyclonal anti-Rap1 antibody. Cell lysates were also used for determination of total Rap-1 by immunoblotting. RapGAPII was first immunoprecipitated with rabbit polyclonal anti-RapGAPII antibody and then detected by immunoblotting with goat antiRapGAP antibody (Santa Cruz Biotechnology). C, control. C, Neuro-2A cells transfected with the control non-silencing siRNA or Rap1 siRNA were stimulated with different concentrations of HU-210. Data represent mean Ϯ S.E. (n ϭ 3-5). D, serum-starved cells were treated with 20 M lactacystin for 2 h prior to the addition of CB1-R agonist. Cells were scored for neurite outgrowth 16 h later. E, cells were transfected with or without 0.1 g Rap1GAPII cDNA, cultured in serum-free medium, and stimulated with CB1R-Ag (1 M HU-210) for 16 h and scored for neurite outgrowth.
our system this appears to be the case. However, it is well known that Rap is activated by several signaling pathways including the integrin pathway and G s /cAMP pathway (15). In such cases the simultaneous or prior activation of the G␣ o/i pathway could play an interesting role in defining a graded switch allowing the cell to regulate both the duration and extent of Rap activation. Such a configuration would differ significantly from switches that arise from positive feedback where the input signal switches the system from the inactive to the active state, but the level of activation is set by the ratios of the components in the feedback loop and their activities (16). One advantage of having a graded switch might be to impart specificity in the engagement of down stream machinery. These potential capabilities remain to be experimentally explored.
Our data indicate that there may be variations in the classical heterotrimeric G protein cycle that allows for signal propagation across time scales. This proposed model for signal flow is summarized in the schematic in Fig. 6. The data in Figs. 2 and 3 indicate that it is the unactivated G␣ subunit that targets the Rap1GAPII for degradation. However, we find that the G␣ o/i -coupled CB1 cannabinoid receptor stimulates the degradation of Rap1GAPII and thus activates Rap1. These seemingly contradictory results may be explained if the GTP-bound G␣ subunit weakly binds Rap1GAPII, and this interaction is strengthened by the hydrolysis of the GTP. The GDP bound G␣ subunit would now have a higher affinity for Rap1GAPII, which acts as a GDI for the G␣ subunit preventing it from exchanging the GDP for GTP and thus keeping it in the inactive state. This higher affinity interaction would then allow the complex to stabilize such that G␣ could target Rap1GAPII for degradation by the proteasome. Support for this model comes from previous observations (17) and our own observations with Rap1GAPII, that G␤␥ subunits are not associated with the complexes between inactive G␣ subunits and proteins containing the GoLoco motif such as Rap1GAPII. Structural data indicate that proteins containing the GoLoco motif bind to the same region of G␣ as the G␤␥ subunits (12). Thus it appears likely that the G␣ subunit in GDP bound form could play a role in regulating signal flow prior to re-association with G␤␥ and returning to the resting state. Further experiments are needed to unequivocally establish such a scheme; however, our observations are supportive of a mechanism for signal flow from heterotrimeric to small GTPases as is described in the model in Fig. 6. We do not as yet understand the mechanisms by which G␣ o promotes the ubiquitination of Rap1GAPII. The simplest model would involve the presentation of Rap1GAPII to an E3 ligase. We have not as yet been able to identify the E3 ligase for Rap1GAPII and experiments to define the interactions and relationships between G␣ o , Rap1GAPII, and the E3 ligase await this identification.
There are some differences in our studies as compared with previous studies. Mochizuki et al. (7) found that Rap1GAPII bound more avidly to the activated form G␣ i2 as compared with WT G␣ i2 . We do not understand the basis for the different observations. However, that study was conducted only with exogenously expressed components and the observed interactions could be related to context of the pathway created by the exogenous expression of multiple components including a high level of receptors. Rap1GAP has also been reported to interact with G␣z, although it has not been specified if this interaction occurs with Rap1GAPI or Rap1GAPII (18).
The observations in this study highlight the multifaceted regulatory capability of the G␣ o/i subunits, in contrast to G␣ s and G␣ q family members. G␣ o/i subunits in addition to inhibiting adenylyl cyclase (19) and stimulating STAT3 through c-Src (20) and regulating neuronal function through GRIN (21) now appear capable of activating and modulating signal flow through the Rap1 pathway. It is interesting that in neuronal systems the G␣ o/i pathway can often synergize or positively affect the cAMP pathway. This often occurs through G␤␥ regulation of adenylyl cyclase 2 (22). Our present observations suggest that it can also occur through G␣ subunits, since G␣mediated degradation of RapGAP can enhance or synergize cAMP-mediated Rap activation. Thus even well studied pathways such as the classical heterotrimeric pathway may as yet hold surprises and the complexity of upstream regulatory mechanisms within signaling networks continues to grow. Such complexity may have important biological functions. Recent studies indicate that the sequential activation of Cdc42 and Rap plays an important role in determining which neurites become axons (23). CB1 receptors, like D2-dopamine receptors, are often presynaptic receptors that modulate transmitter release. It is possible that they are also present in growing axons and thus contribute to the activation of Rap to specify polarity during hippocampal development. Further studies are needed to determine whether such mechanisms are operative during the development of the hippocampus.