A Candidate Target for G Protein Action in Brain*

An effector candidate for G protein action, GRIN1, was identified by screening a cDNA expression library with phosphorylated GTPγS-Gzα as a probe. GRIN1 is a novel protein without substantial homology to known protein domains. It is expressed largely in brain and binds specifically to activated Gzα, Goα, and Giα through its carboxyl-terminal region. The protein KIAA0514 (GRIN2) is homologous to GRIN1 at its carboxyl terminus and also binds to activated Goα. Both GRIN1 and Goα are membrane-bound proteins that are enriched in the growth cones of neurites. Coexpression of GRIN1 or GRIN2 with activated Goα causes formation of a network of fine processes in Neuro2a cells, suggesting that these pathways may function downstream of Goα to control growth of neurites.

It is likely that several targets for the action of G protein 1 ␣ subunits remain to be identified. Members of the G i ␣ subfamily of these proteins, particularly G o ␣ and G i ␣ 1 , constitute roughly 1% of brain membrane protein, yet the only known effectors for G o ␣, the three G i ␣ proteins, and G z ␣ are certain isoforms of the enzyme adenylyl cyclase (1,2). We have thus attempted to detect novel effectors that lie downstream of selected G protein ␣ subunits, initially by utilizing 32 P-labeled ␣ subunits activated (essentially irreversibly) with GTP␥S to probe cDNA expression libraries. We have taken advantage of the fact that G z ␣ can be phosphorylated by protein kinase C at a site near its amino terminus that does not interfere with interaction between G z ␣ and adenylyl cyclase (3). We have also appended a site for phosphorylation by cyclic AMP-dependent protein kinase to the carboxyl terminus of G o ␣; this region of G protein ␣ subunits is also not involved in interactions with known effectors. This strategy led to isolation of a novel cDNA, initially designated Z-16, and detection of a homolog, KIAA0514. The protein products of these cDNAs interact selectively with GTP␥Sor GDP-AlF 4 Ϫ -bound forms of G i ␣ subfamily members in vitro, and they cause extension of neurites in Neuro2a cells when coexpressed with activated forms of G o ␣. We thus tentatively refer to these two proteins as (GRIN1) (Z-16) and GRIN2 (KIAA0514) (G protein-regulated inducer of neurite outgrowth).

EXPERIMENTAL PROCEDURES
Preparation of Phosphorylated G z ␣ and G o ␣-G z ␣ and protein kinase C␣ (PKC␣) were purified using a recombinant baculovirus-Sf9 cell expression system as described previously (2,3). The recombinant catalytic subunit of cAMP-dependent protein kinase (protein kinase A) was expressed and purified from Escherichia coli (4). A tag encoding a phosphorylation site for protein kinase A (LRRASLG) followed by six histidine residues was added to the carboxyl terminus of G o ␣ by polymerase chain reaction. The protein was then coexpressed with protein N-myristoyl transferase in E. coli and purified as described (5). Other recombinant G protein ␣ subunits were prepared as described previously (2,6). G z ␣ (250 g) was incubated with GTP␥S at 30°C for 60 min in the presence of 5 mM EDTA and 2 mM MgSO 4 . Activated G z ␣ was then phosphorylated with PKC␣ (4 g) in buffer containing 25 mM Tris-HCl (pH 7.5), 5 mM MgSO 4 , 0.125 mM CaCl 2 , 1 mM DTT, 3 M [␥-32 P]ATP (40,000 cpm/pmol), and 20 g/ml phosphatidylserine-diolein (Sigma) for 30 min at 30°C. Tagged G o ␣ (600 g) was similarly activated with GTP␥S and phosphorylated with the catalytic subunit of protein kinase A (3 g) at 30°C for 30 min as described by Baude et al. (4). After phosphorylation, free [␥-32 P]ATP was removed by gel filtration (PD-10 column; Amersham Pharmacia Biotech) in Buffer A (20 mM NaHepes (pH 7.4), 100 mM NaCl, 3 mM MgCl 2 , 1 mM EDTA, 0.05% C 12 E 10 (polyoxyethlene 10-lauryl ether), 1 mM DTT, 10 mM ␤-glycerophosphate), and the proteins were used as probes for screening cDNA expression libraries.
Isolation of Z-16 cDNA-Candidate effector proteins were sought in a EXlox library from 16-day mouse embryo cDNA (Novagen) with BL21/DE3 as the host E. coli strain. Approximately 1 ϫ 10 6 clones were screened using phosphorylated G z ␣ as a probe. Phage were plated at a density of 3-5 ϫ 10 4 per 150-mm plate and incubated at 37°C for 7 h. The plates were overlaid with Hybond-C filters (Amersham Pharmacia Biotech) that had been saturated with 10 mM isopropyl-␤-D-thiogalactopyranoside. They were further incubated at 37°C for 3.5 h to induce expression of proteins encoded by cDNAs. The filters were then rinsed with Buffer A at room temperature and blocked with Buffer A containing 5% dry milk at 4°C overnight. The filters were probed with 50 nM phosphorylated GTP␥S-bound G z ␣ in Buffer A with 5% dry milk for 4 h at 4°C. The filters were washed three times with Buffer A for 5 min at 4°C, air dried, and exposed to film at Ϫ70°C for 2 days. Secondary and tertiary screens were carried out under identical conditions. Two clones, Z-13 and Z-16, were isolated and cDNA inserts were sequenced. Z-13 had a 0.8-kb insert encoding mouse nucleobindin (7). Z-16 had a 1.8-kb insert encoding 273 amino acid residues of previously undescribed sequence.
To obtain a full-length Z-16 cDNA, a mouse brain gt11 library (generously provided by Dr. Melvin Simon, California Institute of Technology) was screened by plaque hybridization using a fragment from the 5Ј end of the initial Z-16 clone as a probe. An additional 0.8 kb of coding sequence was isolated; the reading frame was open throughout. Two rounds of 5Ј-rapid amplification of cDNA ends reactions were then performed with a mouse brain cDNA library (CLONTECH), using oligonucleotides based on sequences at the 5Ј end of the Z-16 cDNA. The final Z-16 cDNA contains an ATG codon that agrees well with Kozak's translation initiation criteria and an in-frame, upstream stop codon. A related cDNA clone, designated KIAA0514 (8), was identified in public data bases and was generously supplied by Dr. T. Nagase (Kazusa DNA Research Institute).
Northern Analysis-Northern blots of RNA from various tissues or brain regions (CLONTECH) were probed with GRIN1 (Z-16) or GRIN2 (KIAA0514) cDNA fragments labeled with [␣-32 P]dATP (random primer labeling; Stratagene, Prime-it II). Blots were hybridized with probe (5 ϫ 10 6 cpm/ml) in ExpressHyb (CLONTECH) at 68°C for 2 h. The blots were then washed in 2ϫ SSC and 0.1% SDS for 45 min at room temperature, followed by a high stringency wash with 0.1ϫ SSC containing 0.1% SDS for 30 min at 50°C. Then, the blots were finally subjected to phosphorimaging analysis (TR2040S imaging plates and BAS1500 scanner, Fuji Medical System).
Antisera-Two peptides were synthesized based on the amino acid sequence of GRIN1: P1 ( 550 SSAQPQRDTRSIGSLPER 567 ) and P2 ( 747 EVEVLGMAIQKHLERQIE 764 ); these peptides also included an additional cysteine residue at their amino termini and a tyrosine residue at their carboxyl termini. The cysteine residue was utilized to facilitate cross-linking of peptide to keyhole limpet hemocyanin (Sigma) with m-maleimidobenzoyl-N-hydroxysuccinimide ester. Antisera were produced in New Zealand White rabbits. Antisera T116 and T114 were generated against peptides P1 and P2, respectively. The specificities of the antisera were confirmed by immunoblotting. An antiserum specific for G o ␣ (U1901) was generated in a rabbit against a synthetic peptide with the amino acid sequence of G o ␣ ( 94 EYGDKERKADSKMVC 108 ) conjugated to keyhole limpet hemocyanin. Other rabbit polyclonal antisera against various G␣ subunits have been described previously (2, 9 -11), as has mouse monoclonal antibody (mAb2A) against G o ␣ (12).
Purification of Recombinant GRIN1 and GRIN2 Proteins-GRIN1 and GRIN2 were subcloned into the pFastBacHTb vector (Life Technologies, Inc.), and recombinant baculoviruses encoding His 6 -GRIN1 or His 6 -GRIN2 were generated according to the manufacturer's protocol. Membranes from Sf9 cells infected with baculoviruses encoding His 6 -GRIN1 or His 6 -GRIN2 were prepared and extracted with 1% C 12 E 10 . His 6 -GRIN1 and His 6 -GRIN2 were then purified from these extracts using Ni-NTA chromatography. The proteins were further purified by Mono Q and Superdex 200 column chromatography in 20 mM NaHepes (pH 8.0), 5 mM MgCl 2 , 2 mM EDTA, 50 mM NaCl, 1 mM DTT, and 0.5% C 12 E 10 .
Binding of Purified GRIN1 or GRIN2 to G Protein ␣ Subunits-Hexa-histidine-tagged GRIN1 or GRIN2 purified from SF9 cells (1.5 g) was mixed with 2 g (500 nM) of recombinant G o ␣ or G z ␣ bound with GDP, GTP␥S, or GDP-AlF 4 Ϫ , as indicated, in 100 l of Buffer B (50 mM NaHepes (pH 8.0), 5 mM MgCl 2 , 10 mM ␤-mercaptoethanol, 0.1% C 12 E 10 ) and incubated on ice for 30 min. NaF (10 mM) and AlCl 3 (30 M) were included in the buffer to prepare G␣-GDP-AlF 4 Ϫ . Ni-NTA resin (Qiagen) was equilibrated with Buffer B, and 25 l was added to the mixture of proteins, followed by further incubation on ice for 5 min. The resin was collected by brief centrifugation, and the supernatant was saved as the flow-through fraction. The resin was washed twice with 150 l of Buffer B containing 500 mM NaCl and 10 mM imidazole. NaF and AlCl 3 were included in the wash buffer when they were present initially. The resin was finally eluted twice with 50 l of Buffer B containing 200 mM imidazole. Fractions were analyzed by immunoblotting after SDS-PAGE.
Expression Vectors-An expression vector for GRIN1 was constructed by adding a FLAG tag at the amino terminus of GRIN1 by polymerase chain reaction and subcloning the resultant GRIN1 cDNA (KpnI-HindIII fragment, 3.6 kb) into the pCMV5 vector. For GRIN2, a FLAG tag or a hexahistidine tag was added at the amino terminus, and the resultant cDNA (EcoRI-HindIII fragment, 1.4 kb) was subcloned into pCMV5.
Cell Culture and Transfection-Simian kidney COS-M6 cells, simian renal epithelium MA104 cells, and murine neuroblastoma Neuro2a cells were cultured in Dulbecco's modified Eagle's medium supplemented with 25 mM glucose and 10% heat-inactivated fetal bovine serum at 37°C in an atmosphere of 10% CO 2 . Cells were transiently transfected with appropriate expression plasmids using Lipo-fectAMINE (Life Technologies, Inc.). Cell lysates were fractionated into particulate and soluble fractions as described (13,14).
Immunoprecipitation-A crude membrane fraction of COS cells expressing FLAG-tagged GRIN1 was extracted with immunoprecipitation buffer (50 mM sodium phosphate (pH 7.2), 100 mM NaCl, 1 mM DTT, and 0.5% C 12 E 10 supplemented with fresh protease inhibitors). The extracts were centrifuged at 100,000 ϫ g for 20 min at 4°C. The resulting supernatants were incubated with 200 nM G protein ␣ subunits either in the presence or absence of 10 mM NaF and 30 M AlCl 3 for 10 min on ice, mixed (50 g of protein) with 5 l of anti-FLAG monoclonal antibody M2 (3 mg/ml; Eastman Kodak Co.) in a total volume of 50 l, and incubated at 4°C overnight. Fixed Staphylococcus aureus bacteria (5 l of a 10% suspension; Pansorbin, Calbiochem) were added and incubated on ice for an additional 1 h. The extracts were then centrifuged at 13,000 ϫ g for 1 min, and the precipitates were suspended in 100 l of immunoprecipitation buffer. The suspensions were layered over 0.9 ml of immunoprecipitation buffer containing 20% sucrose (w/v) and centrifuged for 6 min. The pellets were washed with 0.5 ml of phosphatebuffered saline, suspended in SDS-PAGE sample buffer, resolved by

SDS-PAGE, and analyzed by Western blotting.
Subcellular Fractionation of Mouse Brain-Subcellular fractionation of adult mouse brain was performed based on the procedure described in Ref. 15. Mouse brain was homogenized in Buffer C (10 mM Tris-HCl (pH 8.0), 1 mM DTT, 11% sucrose with protease inhibitors) and centrifuged at 600 ϫ g for 10 min. The supernatants were collected and centrifuged at 11,700 ϫ g for 20 min. The pellets were resuspended in Buffer C and loaded onto a discontinuous sucrose gradient (10 -40%) and centrifuged at 150,000 ϫ g for 2 h. Fractions at the interfaces of the gradient were analyzed by immunoblotting. Plasma membrane is concentrated at the 25-30% sucrose interface. Subcellular fractions of embryonic mouse brain were obtained as described (16,17).
Immunocytochemistry-To detect endogenous GRIN1, Neuro2a cells were plated on laminin-coated coverslips at 50% confluency and grown for 2 days to generate neurites. For transient transfection, Neuro2a cells were plated on nontreated coverslips at 60 -70% confluency and allowed to grow for 1 day. Cells were then transfected with expression plasmids and incubated for 48 h.

RESULTS
Cloning of Z-16 cDNA-Phosphorylated GTP␥S-G z ␣ was utilized to screen a mouse embryo EXlox expression library. Two clones, designated Z-13 and Z-16, were eventually isolated based on clearly positive signals in several successive screens. The product of the Z-13 cDNA bound GDP-G z ␣ and GTP␥S-G z ␣ equally well, whereas the product of the Z-16 cDNA bound GTP␥S-G z ␣ selectively (Fig. 1). Sequencing and data base searches revealed that Z-13 encoded the protein nucleobindin. Interactions of G i ␣ 2 with nucleobindin have been detected previously (18). Nucleobindin is described as a secreted protein that interacts with DNA (7). We did not pursue nucleobindin further because of these characteristics and its failure to recognize GTP␥S-bound G z ␣ selectively. The protein product of the Z-16 cDNA also bound to phosphorylated protein kinase A-tagged, GTP␥S-bound G o ␣.
Sequencing of the 1800-base pair insert designated Z-16 revealed that it had an open reading frame encoding 273 amino acid residues fused to the T7 gene 10 product, as anticipated. The sequence was incomplete at the 5Ј-end. An apparently full-length cDNA was obtained by screening a mouse brain cDNA library and by 5Ј-rapid amplification of cDNA ends, using a 5Ј-stretch of randomly primed brain cDNA library ( Fig.  2A). The deduced amino acid sequence specifies a protein containing 827 residues, with M r ϭ 84,700. Nearly 45% of the residues have small side chains: Ala (11%), Gly (9.8%), Pro (10.5%), and Ser (13.4%). Data base searches revealed that KIAA0514, a cDNA clone isolated from a human brain cDNA library, and Z-16 encode homologous ϳ100 -150 amino acid residue domains near their carboxyl termini (Fig. 2B). No other matches or homologies were detected, except for a few previously unidentified Z-16 expressed sequence tags (AI427420, W54141, and AI413422). In addition, related genes were not detected by Southern blot analysis of mouse genomic DNA (data not shown). Because of the functional properties of the proteins encoded by the Z-16 and KIAA0514 cDNAs (described below), we refer to these two proteins as GRIN1 and GRIN2, respectively.
Northern and Western Analysis of GRIN1-Northern blot analysis of GRIN1 was performed using several regions of the cDNA as probes. Messenger RNA for GRIN1 (about 4 kb) was detected in brain but not in heart, liver, spleen, lung, skeletal muscle, kidney, or testis (Fig. 3A). GRIN1 mRNA is widely distributed in the central nervous system; the highest concentration was detected in the spinal cord (Fig. 3B). Messenger RNA for GRIN2, which is about 8 kb, is also specifically expressed in brain. In contrast to GRIN1, expression of GRIN2 was only detected in cerebellum (Fig. 3B).
Two antibodies specific for GRIN1 (designated T114 and T116) were prepared using GRIN1 peptides as immunogens. In lysates of COS cells transfected with GRIN1 cDNA, both anti- Ϫ -bound G protein ␣ subunit, respectively, was mixed with extract from COS cells expressing Flag-tagged GRIN1.
bodies detected an immunoreactive band with an apparent molecular weight of 110,000. Consistent with the results of Northern analysis, this 110-kDa band was also detected in mouse brain homogenates but not in homogenates of other tissues (Fig. 3C). The appearance of GRIN1 as a doublet band in brain lysate may reflect proteolysis during sample preparation, or it may indicate the existence of splice variants of the protein. This immunoreactive band was not detected if antisera were first incubated with peptides used as immunogens or with expressed GRIN1 protein (data not shown). We thus conclude that this band represents GRIN1 protein and that the protein migrates anomalously during SDS-polyacrylamide gel electrophoresis. Among several cell lines tested, endogenous GRIN1 was detected in mouse neuroblastoma Neuro2a (see Fig. 5B) and rat pheochromocytoma PC12 cells as a doublet similar to its appearance in brain lysate (data not shown).
The peptide used to generate antibody T114 shares 13 amino acid residues (of 18) with GRIN2. This antibody, but not T116, recognized the GRIN2 product expressed in COS or Sf9 cells (data not shown). GRIN2 (461 amino acid residues) has a calculated molecular weight of 47,600 but migrates with an apparent molecular weight of 65,000 during SDS-PAGE.
Interaction of GRIN1 and GRIN2 with Various G Protein ␣ Subunits-Baculoviruses encoding full-length GRIN1 or GRIN2 (with hexahistidine tags at the amino terminus) were generated, and recombinant GRIN1 or GRIN2 protein was purified from Sf9 cells using Ni-NTA, Mono-Q, and Superdex 200 column chromatography. Purified GRIN1 or GRIN2 was mixed with 500 nM G o ␣ or G z ␣ bound with either GDP, GDP-AlF 4 Ϫ , or GTP␥S, and interactions were analyzed by co-elution from Ni-NTA columns (Fig. 4A). Both GRIN1 and GRIN2 interacted with the GTP␥S-or GDP-AlF 4 Ϫ -bound form of G o ␣ and G z ␣; interactions with the GDP-bound forms of these two proteins were not detected.
Interactions between GRIN1 and various G␣ proteins were also examined using full-length Flag-tagged GRIN1 expressed in COS cells. Detergent extracts containing GRIN1 were then mixed with G protein ␣ subunits, and complexes were precipitated with an anti-Flag monoclonal antibody (Fig. 4B). Specific interactions were detected between GRIN1 and members of the G i subfamily of G␣ subunits, including G o ␣, G z ␣, and G i ␣ 1 . A weak interaction was detected with G 12 ␣, but none was observed with G q ␣ or G s ␣. Again, GRIN1 interacted preferentially with the GTP␥S or the GDP-AlF 4 Ϫ -bound forms of the G␣ proteins compared with the GDP-bound proteins. Subcellular Distribution of GRIN1; GRIN1 Is Found in Growth Cones with G o ␣-The subcellular distribution of GRIN1 was first examined in transfected COS cells. Exogenously expressed GRIN1 and G o ␣ were largely confined to the particulate fractions, and only small amounts appeared in the cytosol (Fig. 5A). Similarly, endogenous GRIN1 and G o ␣ in Neuro2a cells and brain are predominantly membrane-bound (Fig. 5, B and C). The amino acid sequence of GRIN1 contains no obvious hydrophobic domain sufficient to explain membrane localization. There are also no consensus sequences for covalent modification by myristoylation or prenylation.
Co-localization of GRIN1 and G o ␣ in Neuro2a cells was also demonstrated by immunofluorescence. Neuro2a cells were induced to extend neurites by culture on laminin-coated dishes. Permeabilized cells were then incubated with both rabbit polyclonal anti-GRIN1 antibodies (T116) and a mouse monoclonal antibody specific for G o ␣ (mAb2A). Oregon Green-conjugated goat anti-rabbit IgG or Texas Red-conjugated goat anti-mouse IgG were used as secondary antibodies to detect GRIN1 and G o ␣, respectively. The specificity of observed fluorescence was confirmed by competition with the immunogenic peptide or GRIN1 protein (data not shown). The pattern of immunofluorescence for GRIN1 and G o ␣ in differentiated Neuro2a cells is similar, and the two proteins are predominantly found at the plasma membrane (Fig. 5, D-F).
G o ␣ and GAP43 are enriched in growth cone membranes (19). Growth cone membrane fractions were prepared as described by Pfenninger et al. (16) and Simkowitz et al. (17) and analyzed by immunoblotting using GRIN1 antisera. Fig. 6A shows that GRIN1 is highly enriched in the growth cone membrane fraction (lane 4), as are GAP43 and G o ␣. By contrast, G z ␣ and the low-affinity NGF receptor are found in all membrane fractions. Furthermore, immunofluorescence of differentiated Neuro2a cells shows that GRIN1 and GAP43 are concentrated in putative growth cone membranes (Fig. 6, B-D, arrows).
GRIN1 and GRIN2 Cause Morphological Changes in MA104 and Neuro2a Cells-GRIN1 or GRIN2 was cotransfected into MA104 (simian renal epithelium) cells with either wild type G o ␣ or a constitutively active (GTPase-deficient) mutant of the protein (G o ␣ Q205L), and the transfected cells were examined by immunofluorescence microscopy. As shown in Fig. 7A, MA104 cells transfected with only wild type G o ␣, G o ␣Q205L, or GRIN1 had a relatively unperturbed morphology. However, cells spread irregularly on coverslips and generated many fine, neurite-like processes if they were cotransfected with GRIN1 and G o ␣Q205L. This effect was much less apparent if wild type G o ␣ was expressed with GRIN1. Expression of GRIN2 alone caused extension of processes in some cells, and co-transfection of G o ␣ or G o ␣Q205L increased both the frequency and extent of these changes. Similar experiments were performed with Neuro2a cells (Fig. 7, B and C). Again, co-transfection of GRIN1 or GRIN2 with G o ␣Q205L caused long neurites to appear. In addition, these neurites displayed many hair-like processes. Co-transfection of GRIN1 or GRIN2 with wild type G o ␣ caused less extensive changes. In contrast to MA104 cells, expression of G o ␣Q205L alone induced formation of neurites, albeit with lower efficiency. This effect is perhaps explained by the endogenous content of GRIN1 in Neuro2a cells. These results indicate that G o ␣ and GRIN1 interact (directly or indirectly) in vivo and that this causes formation and extension of neurite-like processes. Similar effects of G o ␣Q205L have been described in PC12 cells (20). DISCUSSION We screened a mouse embryo cDNA expression library to search for proteins capable of interacting with phosphorylated GTP␥S-G z ␣. Two clones were isolated, designated Z-13 and Z-16. Z-13 encoded nucleobindin, the interactions of which with G i ␣ 2 were detected previously using a yeast two-hybrid screen (18). We did not pursue this lead further because of the apparently similar affinity of GDP-G z ␣ and GTP␥S-G z ␣ for nucleobindin and the presence of a signal sequence in the protein. By contrast, the protein now designated GRIN1 interacts preferentially with activated forms of G protein subunits in the G i ␣ subfamily (G i ␣, G o ␣, G z ␣). Furthermore, GRIN1 is specifically expressed in brain and shares with G o ␣ substantial enrichment in membranes from neuronal growth cones. We thus hypothesize that GRIN1 may function as a downstream effector for G o ␣.
A homolog of GRIN1-GRIN2 (KIAA0514), was isolated and sequenced previously as a newly identified brain cDNA by Nagase et al. (8). The regions of GRIN1 and GRIN2 that are similar are at the carboxyl termini of both proteins, and we had shown independently that this is the G o ␣-binding domain of GRIN1 (data not shown). Significantly, GRIN2 also interacts preferentially with activated members of the G i subfamily of G protein ␣ subunits. Although GRIN1 is widely distributed throughout the central nervous system, GRIN2 is apparently restricted to the cerebellum.
G o ␣ is the most abundant G protein ␣ subunit in mammals. It is expressed predominantly in brain and is enriched in neural growth cones (19). Despite these interesting properties, physiological roles for G o ␣ have not been identified, other than its interactions with G protein ␤␥ subunits and appropriate receptors. The heterotrimeric G o protein is responsible for receptor-mediated inhibition of voltage-sensitive N-type or P/Qtype Ca 2ϩ channels in presynaptic nerve terminals, but this effect appears to be mediated by the G protein ␤␥ subunit complex (21). G o ␣ is a weak inhibitor of some isoforms of adenylyl cyclase (1), but the physiological significance of this is difficult to evaluate. G o ␣ has also been hypothesized to regulate neurite extension. Binding of GTP␥S to G o ␣ is stimulated by GAP43 (neuromodulin), an abundant growth cone protein that is important for neural pathfinding (19). The expression of both GAP43 and G o ␣ starts in brain regions when differentiated neurons begin to extend neurites (22). Furthermore, expression of constitutively activated mutant forms of G o ␣ stimulates neurite outgrowth in neuronal cell lines (20). The molecular mechanism for this phenomenon has not been defined.
We have shown herein that either GRIN1 or GRIN2 induces extensive outgrowth of neurites from Neuro2a cells when coexpressed with activated forms of G o ␣. This result implies interaction between proteins in vivo, although not necessarily a direct one. However, the fact that G o ␣ does interact directly with both GRIN1 and GRIN2 in vitro suggests that these latter proteins may function physiologically as downstream targets for G o ␣ and/or other members of the G i ␣ subfamily to regulate neurite outgrowth. The amino acid sequences of GRIN1 and GRIN2 show no significant homology with known kinases, enzymes that generate second messengers, or other identified effectors for G protein action. Although GRIN1 has a prolinerich domain (residues 590 -710), we detected no other provocative signatures. Recent evidence suggests that Rho family GTPases are important components of signaling pathways that control axonal growth and guidance (23). Rho itself is involved in collapse of growth cones and retraction of neurites. By contrast, Cdc42 and Rac1 stimulate the formation and advance of growth cones through formation of filopodia and lamellipodia (24). It is thus possible that Cdc42 or Rac1 are downstream components of signaling pathways that include G o ␣ and GRIN1 or GRIN2. Perhaps relevant is the fact that the G proteins G 13 ␣ and G 12 ␣ have recently been shown to control the activity of a guanine nucleotide exchange factor, p115, that activates Rho (25,26).