Inhibition of Gαi2 Activation by Gαi3 in CXCR3-mediated Signaling*

G protein-coupled receptors (GPCRs) convey extracellular stimulation into dynamic intracellular action, leading to the regulation of cell migration and differentiation. T lymphocytes express Gαi2 and Gαi3, two members of the Gαi/o protein family, but whether these two Gαi proteins have distinguishable roles guiding T cell migration remains largely unknown because of a lack of member-specific inhibitors. This study details distinct Gαi2 and Gαi3 effects on chemokine receptor CXCR3-mediated signaling. Our data showed that Gαi2 was indispensable for T cell responses to three CXCR3 ligands, CXCL9, CXCL10, and CXCL11, as the lack of Gαi2 abolished CXCR3-stimulated migration and guanosine 5′-3-O-(thio)triphosphate (GTPγS) incorporation. In sharp contrast, T cells isolated from Gαi3 knock-out mice displayed a significant increase in both GTPγS incorporation and migration as compared with wild type T cells when stimulated with CXCR3 agonists. The increased GTPγS incorporation was blocked by Gαi3 protein in a dose-dependent manner. Gαi3-mediated blockade of Gαi2 activation did not result from Gαi3 activation, but instead resulted from competition or steric hindrance of Gαi2 interaction with the CXCR3 receptor via the N terminus of the second intracellular loop. A mutation in this domain abrogated not only Gαi2 activation induced by a CXCR3 agonist but also the interaction of Gαi3 to the CXCR3 receptor. These findings reveal for the first time an interplay of Gαi proteins in transmitting G protein-coupled receptor signals. This interplay has heretofore been masked by the use of pertussis toxin, a broad inhibitor of the Gαi/o protein family.

control mice and few or no T cells in the peripheral lymphoid tissues because of a defect in thymic emigration (11). Thymocyte emigration appears to be controlled by either G␣ i3 or G␣ i2 , because null mutation of either of them does not affect this process significantly (12,13). In the periphery, however, their function differs considerably. G␣ i2 -deficient mice display impairment in lymph node development in several anatomic locations and in Peyer's patch formation (12,14). The mice develop inflammatory bowel disease, at least in part, due to a lack of transforming growth factor-␤ responses in T cells that causes Th1-skewed hyperimmune responses in the colon (12,15). In contrast, G␣ i3 -deficient mice develop no overt phenotype (16).
In this study, we investigate unique properties of G␣ i2 and G␣ i3 in chemotaxis of activated T cells induced by CXCR3 ligands, because expression of the CXCR3 receptor has been associated with pathophysiology of Th1-type autoimmune diseases in both humans and experimental models (17)(18)(19). The receptor has the following three known ligands: monokine-induced IFN-␥ (MIG)/CXCL9, IFN-␥-inducible 10-kDa protein (IP-10)/CXCL10, and IFN-␥-inducible T cell ␣-chemoattractant/CXCL11. These three chemokines exhibited discernible biological effects in vivo and in vitro (20 -24), but they all activated exclusively G␣ i2 through the CXCR3 receptor, as indicated by a lack of CXCR3-mediated responses in G␣ i2Ϫ/Ϫ T cells, regardless of the agonist employed. Interestingly, we found that G␣ i3 was negatively involved in the CXCR3-mediated signaling. This translated to cellular responses, where a lack of G␣ i3 significantly enhanced T cell migration toward CXCR3 agonists. G␣ i3 appears to hinder G␣ i2 activation via the N terminus of the second intracellular loop. A mutation in this region abolished both activation of G␣ i2 and the binding of G␣ i3 to the receptor. This work begins to unravel an interplay between individual G␣ i proteins that has previously been masked by the use of PTX.

MATERIALS AND METHODS
Animals-G␣ i2 -knock-out (KO), G␣ i3 -KO, and wild type (WT) control mice on the mixed 129Sv/C57BL/6 background were generated by gene targeting and backcrossed with C57BL/6 (B6) mice for six times as described (12,25). Both female and male mice were used at 4 -6 weeks of age unless otherwise indicated. The mice were housed in conventional cages at the animal facilities of the Massachusetts General Hospital in accordance with institutional guidelines.
T Cell Preparation-Single-cell suspensions prepared from lymph nodes and spleens were treated with a mixture of rat anti-mouse monoclonal antibodies (Abs) against CD19, CD32, and CD16 followed by depletion of Ab-bound cells with BioMag goat anti-rat IgG (Polysciences Inc., Warrington, PA) per the manufacturer's instructions. This preparation isolated an ϳ90% pure population of T cells. CXCR3 expression on these T cells was induced by stimulation for 2 days at 37°C and 5% CO 2 with 2.0 g/ml concanavalin A (ConA) in a complete RPMI 1640 medium (10% fetal calf serum, 2 mM L-glutamine, 100 units/ml penicillin, 100 g/ml streptomycin, and 50 M ␤-mercaptoethanol). The cells were continuously cultured for 2 more days in the medium with replenishment of 15% ConA supernatant, which gave rise to a maximal level of CXCR3 expression.
Cell Membrane Preparation-Cell membrane was prepared as described with some modifications (26). Briefly, T cells stimulated with or without ConA as above were collected, suspended in ice-cold hypotonic buffer (10 mM Tris-HCl, pH 7.6, 5 mM EDTA, 3 mM EGTA, 1ϫ protease inhibitor mixture, and 1 mM PMSF), and Dounce-homogenized. In some cases, the cells were treated with 10 ng/ml PTX for 1 h at 37°C before suspension in hypotonic buffer and homogenization. Disrupted cells were centrifuged at 500 ϫ g at 4°C for 10 min to remove nuclei and unbroken cells. The remaining supernatant was spun at 32,000 ϫ g for 25 min at 4°C to pellet the cell membrane that was then suspended in membrane solubilization buffer (20 mM glycine, pH 7.2, 1 mM MgCl 2 , 250 mM sucrose, 1ϫ protease inhibitor mixture, and 1 mM PMSF). The suspended cell membrane was left on ice for at least 2 h with periodic vortexing and then frozen until use.
GTP␥S Assay-Cell membrane equivalents of 20 g of protein were first incubated with the indicated concentrations of chemokines on ice for 1 h to prime G␣ i activation by the receptor (27). G protein activation was initiated by placing samples at room temperature for 2 h in the presence of 0.3 nM [ 35 S]GTP␥S (1250 Ci/mmol; PerkinElmer Life Sciences) in a GTP-binding buffer (50 mM Tris-HCl, pH 7.4, 100 mM NaCl, 1 mM EGTA, and 5 mM MgCl 2 ) supplemented with fresh 0.1% bovine serum albumin, 1 mM dithiothreitol, and 2 mM GDP (28). Activation of G␣ i proteins at room temperature instead of 37°C minimizes background radioactive incorporation (28). The concentrations used for each CXCR3 ligand were empirically determined based on the most robust migratory response and our preliminary tests. All chemokines were obtained from PeproTech Inc. (Rocky Hill, NJ). Where indicated, a myristoylated G␣ i3 protein or control G␣ o protein (Calbiochem) at varying concentrations was preincubated with the cell membrane for 1 h on ice to allow the added G␣ protein to interact with the ␤␥ subunit in the membrane prior to chemokine stimulation (27). To determine the G␣ i protein in association with the receptor, 20 g/ml monoclonal Ab specific for G␣ i2 (IgG2b, Chemicon, Temecula, CA) or an isotype-matched control Ab was added to the samples, along with the chemokine to block G␣ i2 protein activation (29). The samples were then transferred to a 96-well plate and suctioned through a glass fiber filter by a TomTec cell harvester (model MACH 3 M, Hamden, CT). The filters were washed five times with a cold GTP washing buffer (20 mM Tris, pH 8.0, 100 mM NaCl, and 25 mM MgCl 2 ) and dried, and radioactive count was determined on a 1450 MicroBeta scintillation counter (Wallac, Turku, Finland). Percentages of specific GTP␥S incorporation in activated T cell membranes were calculated as ((counts/min of sample) Ϫ (counts/min of background control))/((counts/min control without a stimulus) Ϫ (counts/min of background control)) ϫ 100, where background controls contained 1,000 times excess cold GTP␥S relative to radioactive [ 35 S]GTP␥S.
Migration Assays-The Transwell system with 3-m pore size polycarbonate membrane (Corning Glass, Corning, NY) was used for all transmigration assays, because T cells are relatively small in size. This pore size significantly reduced the basal levels of T cell migration in our study. T cells at 3 ϫ 10 6 cells/ml in Aim-V medium (Invitrogen) were added to the upper chamber at 0.3 ml per well, with the indicated concentrations of chemokines in 0.6 ml of Aim-V medium in the lower chamber. Freshly isolated T cells or T cells activated by ConA for 4 days were allowed to migrate for 4 -6 h at 37°C and 5% CO 2 . At the end of the migration, the cells in the lower chamber were collected and counted on a hemocytometer.
Western Blot Analysis-To examine G␣ i2 expression in T cell membranes, T cells with or without stimulation were collected and lysed in cold nondenaturing lysis buffer (50 mM Tris-HCl, pH 7.5, 300 mM NaCl, 5 mM EDTA, 1% Triton X-100, 1 mM PMSF, and 1ϫ protease inhibitor mixture), followed by centrifugation at 500 ϫ g for 10 min to remove nuclear fragments and unbroken cells. Crude cell membrane was pelleted from the supernatant by centrifugation at 32,000 ϫ g for 25 min. Protein samples of ϳ100 g were separated on a 10% SDS-PAGE, transferred onto a nitrocellulose membrane, and immunoblotted with anti-G␣ i2 monoclonal Ab (Chemicon, Temecula, CA). G␣ i2 protein was visualized by incubation of the membrane with sheep anti-mouse Ab conjugated with horseradish peroxidase (Amersham Biosciences) followed by reaction with SuperSignal West Pico ECL reagents (Pierce) and exposure to Kodak film (Eastman Kodak Co.). For assessment of G␣ i3 protein expression, G␣ i2 protein was first depleted by anti-G␣ i2specific Ab because of cross-reaction of anti-G␣ i3 Ab with G␣ i2 protein and similar molecular mass of these two G proteins. Briefly, T cell membranes prepared above were incubated for 1 h on ice with anti-G␣ i2 Ab-conjugated protein G-agarose beads that had been treated previously with 1 mg/ml bovine serum albumin to block nonspecific binding. The beads were removed by centrifugation, and the depletion was repeated three times. The resultant G␣ i2 -depleted T cell membranes were subjected to Western blotting analysis using anti-G␣ i3 Ab (BIOMOL International LP) as detailed above. The same membranes were re-probed, after stripping, with anti-G protein pan-␤ subunit Ab (Upstate Biotechnology, Inc., Lake Placid, NY) for equal protein loading controls.
Flow Cytometry-To determine chemokine receptor expression, T cells with or without stimulation with ConA or transfected COS-7 cells were incubated with specific Abs followed by analysis on a FACSCalibur cytometer equipped with a CellQuest software (BD Biosciences). A murine CXCR3-specific Ab (clone 4c4) was a generous gift from Dr. Dominic Picarella (Millenium Pharmaceuticals) (30). Fluorescence-conjugated rat anti-mouse Abs specific for CCR7 or CXCR4 and mouse anti-human CXCR3 Ab were purchased from Biolegend (San Diego, CA), R&D Systems (Minneapolis, MN), or Pharmingen, respectively.
In Vitro Transcription and Translation-G␣ i2 and G␣ i3 proteins were synthesized and 35 S-labeled by the TNT Quick-Coupled in Vitro Transcription/Translation Systems per the manufacturer's instructions (Promega, Madison, WI). In brief, 1 g of a G␣ i2 or G␣ i3 plasmid was mixed with the rabbit reticulocyte lysate provided, 100 M myristic acid, and [ 35 S]methionine (1000 Ci/mmol) for 75 min at 30°C. The resulting product was verified by SDS-PAGE on the basis of the size and compared in the presence or absence of the specific plasmid.
Immunoprecipitation-To study CXCR3-associated G␣ i proteins, COS-7 cells were transfected with pcDNA plasmids containing either WT or mutated CXCR3 receptors or a control HA-tag⅐S1P 1 (sphingosine 1-phosphate 1) receptor using Lipofectamine according to the manufacturer's instructions (Invitrogen). Three mutants were as follows: 1) an R149N-CXCR3 that had asparagine (Asn) substitution for arginine (Arg) at position 149; 2) LLL-CXCR3 generated by replacement of alanines for leucines (Leu) at positions 332-334 in the C terminus; and ST(Ϫ)-CXCR3 constructed by substitution of all serines (Ser) and threonines (Thr) in the C terminus with alanines (24,31,32). The transfected cells were lysed, and the cell membrane was prepared and suspended in a GPCR solubilization buffer (1 mM MgCl 2 , 250 mM sucrose, 1ϫ protease inhibitor mixture, and 1 mM PMSF in phosphate-buffered saline) as described (24,31,32). Cell membranes with equivalent of 60 g of protein were incubated with 3 l of in vitro synthesized 35 Slabeled G␣ i2 or G␣ i3 for 1 h on ice to allow the added G␣ i protein to interact with the ␤␥ subunit in the membrane, after which the cell membrane was either left unstimulated or stimulated for 2 h on ice by indicated chemokines and activated at room temperature as described (27). The stimulated membranes were then cross-linked for 2 h on ice with 20 M dithiobis-sulfosuccinimidyl propionate (Pierce). The cross-linker was quenched by the addition of 50 M Tris, pH 7.5, for 15 min on ice. To the chemokine-stimulated membranes, anti-CXCR3 Ab or control Ab was added, and the Ab-bound receptor was pulled down by pre-cleaned protein G-agarose beads (Invitrogen). For HA-S1P 1 -transfected cells, immunoprecipitation using anti-HA Ab was carried out in place of anti-CXCR3 Ab. The resultant beads were washed extensively with a cold GPCR washing buffer (1 mM EGTA, 5 mM MgCl 2 , and 0.1% bovine serum albumin in phosphate-buffered saline). Radioactivity of the resultant beads was quantified by scintillation counting on the 1450 MicroBeta scintillation counter to determine the amount of 35 S-G␣ i2 or 35 S-G␣ i3 in association with the CXCR3 receptor.
Statistical Analysis-The Student's two-tailed t test was used to analyze the significance of experimental groups compared with relevant controls.

Altered Migration toward CXCR3 Ligands of G␣ i2 -or G␣ i3deficient T Cells-CXCR3 agonists induce chemotactic responses of activated T cells in vitro
with a potency of CXCL11 Ͼ CXCL10 Ͼ CXCL9 in a PTX-sensitive manner (24,34). These three chemokines also mediate distinct biological effects in vivo (20,21,23,35). In an attempt to understand whether the varying responses evoked by these agonists resulted from activation of different G␣ i proteins, we tested the role of G␣ i2 and G␣ i3 in the migration of activated T cells toward CXCL9, CXCL10, or CXCL11 (36). Stimulation of WT T cells with increasing concentration gradients of CXCR3 ligands elicited a weak chemotactic response to CXCL9 (data not shown) but vigorous responses to CXCL10 or CXCL11 in a dose-dependent manner ( Fig. 1, A, B, and E), mirroring previous investigations (37). However, migration toward these three individual chemokines was all reduced to control levels in cells lacking G␣ i2 over a wide range of agonist concentrations tested ( Fig. 1, A, B, and E). The lack of chemotactic responses toward CXCR3 agonists was not due to aberrant cell viability or migration of these cells, because they migrated similarly as WT counterparts in response to CCL19 or CCL21 (data not shown), two ligands for the CCR7 receptor (Fig. 1H). These results indicate that CXCR3-mediated migration depends predominantly on G␣ i2 , irrespective of the ligands.
Unexpectedly, the cell migration toward CXCL9 and CXCL11 was not attenuated but instead drastically increased in the absence compared with presence of G␣ i3 (Fig. 1, A and E). The migration toward CXCL10 was also increased, albeit to a lesser extent (Fig. 1, B and E). However, such increased chemotaxis could not be induced by CCL21 (data not shown) or CCL19 in the cells, regardless of whether or not the cells were activated (Fig. 1, G and H). We also observed no augmented chemotactic response in G␣ i3 Ϫ/Ϫ cells stimulated with CXCL12, a ligand for another G␣ i -dependent receptor CXCR4. In fact, CXCL12-provoked chemotaxis was diminished in a lack of either G␣ i2 or G␣ i3 (Fig.  1F). The inflammatory chemokines MCP-1 and MIP-1␣ were unable to elicit a significant migratory response given our T cell activation conditions (data not shown). Together, these data stress that the increased migration toward CXCR3 agonists in the absence of G␣ i3 is receptor-specific.
G␣ i2 and CXCR3 Expression in T Cells-The opposing responses of G␣ i2 -and G␣ i3 -deficient T cells to CXCR3 ligands argue for G␣ i3 antagonizing the function of G␣ i2 . To conclude this, we must rule out that the increased migration toward CXCR3 agonists in G␣ i3 Ϫ/Ϫ cells resulted from increased expression of G␣ i2 that could potentially provide a compensatory mechanism to offset the lack of G␣ i3 (38). Fig. 2A clearly showed that G␣ i2 expression was not increased in activated G␣ i3 Ϫ/Ϫ T cells compared with WT T cells (lanes 3 versus 1, lower panel), as was the case for G␣ i3 expression in activated G␣ i2 Ϫ/Ϫ T cells (lanes 5 versus 1, lower panel). Neither G␣ i2 nor G␣ i3 expression was increased in freshly isolated KO T cells compared with WT T cells ( Fig. 2A, top panel). We also detected comparable levels of the CXCR3 receptor on these three groups of T cells (Fig. 2B). The expression of other receptors, including CCR7 on stimulated and unstimulated T cells and CXCR4 on resting T cells, was also comparable in these cells, despite some variations (Fig. 2B). To exclude another possibility that the increased chemotactic response was ascribed to coupling of the CXCR3 receptor to other G protein family members in the absence of G␣ i3 , the cells were treated with 100 ng/ml PTX overnight before they were assayed for chemotaxis. The pretreatment ablated CXCR3-mediated chemotactic responses in both WT and G␣ i3 -deficient T cells (Fig. 1, C and  D), confirming involvement of PTX-sensitive G␣ i2 protein only in the increased migratory response in the absence of G␣ i3 . Our investigation thus concludes that G␣ i3 participates as an inhibitor in CXCR3-stimulated signaling, and null mutation of G␣ i3 directly accounts for the increased migration of G␣ i3 Ϫ/Ϫ T cells toward CXCR3 ligands.
Altered GTP␥S Incorporation in the Absence of G␣ i2 or G␣ i3 -Our migration data show involvement of both G␣ i2 and G␣ i3 in CXCR3-mediated signaling, presumably with G␣ i2 activating downstream effectors to drive T cell migration. Yet G␣ i3 -mediated inhibition of G␣ i2 activity in a receptor-specific fashion has never been explored. So in the subsequent studies, we set out to determine whether G␣ i3 directly blocked G␣ i2 activation or diverted signals downstream of G␣ i2 . For downstream signaling disruption, we would expect G␣ i3 activation to be an indispensable receptor-stimulated event in the G protein-signaling cascade, which can be readily measured by a GTP␥S incorporation assay. All three CXCR3 chemokines at the concentrations optimal for cell migration were able to stimulate significant GTP␥S incorporation in WT T cell membranes as reported previously (Fig.  3A) (39). In contrast, T cells from G␣ i2 Ϫ/Ϫ mice failed to incorporate GTP␥S significantly above controls following stimulation with CXCL9 or CXCL11 under similar conditions (Fig. 3A). A small increase in GTP␥S incorporation was seen in the cells stimulated with CXCL10, consistent with lesser G␣ i3 inhibition in response to this chemokine ( Fig. 3A and Fig. 1B). The lack of GTP␥S incorporation in G␣ i2 Ϫ/Ϫ T cell membranes was not a failure of these cells to incorporate the GTP analog, because adenosine elicited comparable GTP␥S incorporation in T cells isolated from all these mice (Fig. 3A) (40). Instead, the result argues strongly that G␣ i3 cannot be activated by CXCR3 agonists, despite its ability to inhibit G␣ i2 function. Moreover, GTP␥S incorporation in these T cells proportionally correlated with their migratory responses stimulated by CXCR3 agonists, increasing G protein activation in G␣ i3 Ϫ/Ϫ T cells and lacking such activation in G␣ i2 Ϫ

/Ϫ T cells as compared with WT T cells. The increased GTP␥S incorporation in G␣ i3
Ϫ/Ϫ cells was solely contributed by G␣ i2 activation, as demonstrated by complete abrogation of the response following PTX treatment (Fig. 3B). The findings not only indicate that G␣ i2 is the predominant G protein activated by the CXCR3 receptor, but also suggest that CXCR3 receptor activation cannot elicit exchange of GDP for GTP in G␣ i3 protein, arguing that G␣ i3 interferes with G␣ i2 signaling without being activated.

Inhibition of CXCR3-mediated G␣ i2 Activation by G␣ i3 -The consistent and significant increases in GTP␥S incorporation in G␣ i3
Ϫ/Ϫ T cell membranes and in the migration of these T cells argue strongly for inhibition of G␣ i2 activation by G␣ i3 . Moreover, inhibition of G␣ i2 activation without a concomitant rise in G␣ i3 activation raises an intriguing possibility that G␣ i3 may interact with the receptor after ligand stimulation but remain inactive and continue its association with the receptor, by which it blocks G␣ i2 interaction with the receptor. To address this, we first used an Ab directed against the C terminus of the G␣ i2 protein to block the receptor-G protein interaction in the GTP␥S incorporation assay. As shown in Fig. 3C, pretreatment with G␣ i2 -specific Ab but not control Ab completely abolished CXCL10-and CXCL11-stimulated GTP␥S incorporation in G␣ i3 Ϫ/Ϫ T cell membranes. A specific and complete blockade of G protein activation by anti-G␣ i2 Ab further rules out activation of other G proteins by the CXCR3 receptor. Second, in vitro synthesized G␣ i3 protein was added to the GTP␥S incorporation assay prior stimulation with CXCL11 to block G␣ i2 activation. Cell membranes prepared from activated G␣ i3 Ϫ/Ϫ T cells were used in the assay to ensure no competition from endogenous G␣ i3 . As shown in Fig. 3D, the addition of 50 nM G␣ i3 resulted in a significant decrease in GTP␥S incorporation of CXCL11-stimulated membranes, with 100 nM exogenous G␣ i3 protein leading to complete inhibition of G␣ i2 activation. Under similar conditions, however, addition of G␣ o , another member of the G␣ i/o family, did not affect G␣ i2 activation, indicating a specific blockage of CXCR3 activation by G␣ i3 protein.
Association of G␣ i3 with the CXCR3 Receptor-We next determined whether G␣ i3 could physically interact with the CXCR3 receptor following stimulation with CXCR3 ligands. Co-immunoprecipitation of an activated G protein with its receptor has rarely been documented due to transient interactions between the activated receptor and a G protein.
Because G␣ i3 is not activated by the CXCR3 receptor, G␣ i3 would be retained by the receptor after stimulation. To test this possibility, COS-7 cells were transiently transfected with constructs encoding the CXCR3 receptor or its variants that have been shown to alter migration and GTP␥S incorporation (24). Cell surface expression of these receptors was compared by flow cytometry (Fig. 4A). The results showed comparable levels of the ST(Ϫ)-CXCR3 variant and WT CXCR3 and slightly but similarly decreased levels of the DNY-CXCR3 and LLL-CXCR3 variants on the cells. To evaluate the effects of these mutations on G␣ i3 binding, cell membrane fractions prepared from these transfected cells were incubated with in vitro synthesized 35 Slabeled G␣ i2 or G␣ i3 and stimulated with or without CXCL11, as the chemokine induced the most robust response among the three CXCR3 agonists. Immunoprecipitation of the receptor using anti-CXCR3 Ab revealed G␣ i3 interacting with WT CXCR3 receptor after stimulation with CXCL11, with the presence of 35 S-labeled G␣ i3 only in CXCL11-stimulated but not in nonstimulated immunoprecipitates (Fig. 4B). Co-immunoprecipitation of G␣ i3 with the C-terminal mutants LLL-CXCR3 and ST(Ϫ)-CXCR3 was also evident in CXCL11-stimulated cell membranes. However, the interaction between G␣ i3 and the receptor apparently did not occur when the G protein-binding motif DRY was mutated to DNY (R149N mutant) in the N-terminal portion of the second intracellular loop (Fig. 4B). The finding that this motif is required for interaction between G␣ i3 and the CXCR3 receptor recapitulates its critical role in proper migratory responses of the CXCR3 receptor and other chemo- * and ** are statistical significance (p Ͻ 0.05) or high significance (p Ͻ 0.01), respectively, in the presence versus absence of a specific G␣ i protein. B, inhibition of G protein activation by PTX treatment. Activated T cells prepared from indicated mice were treated with 10 ng/ml PTX for 1 h before the cell membranes were prepared and assayed for GTP␥S incorporation induced by either 10 ng/ml CXCL10 or 100 ng/ml CXCL11 as A. Data represent three separate experiments with each sample in triplicate as A. C, blockade of G protein activation in G␣ i3 Ϫ/Ϫ T cell membrane by anti-G␣ i2 Ab. G protein activation in cell membrane prepared from G␣ i3 Ϫ/Ϫ T cells was assayed in the presence of either 3 g of anti-G␣ i2 monoclonal Ab or isotype-matched control Ab (Control Ab) during chemokine stimulation as in A. The data represent three independent experiments with each sample in triplicate as A. ** statistical significance (p Ͻ 0.01) in the presence versus absence of a specific anti-G␣ i2 Ab. D, inhibition of G␣ i2 activation in G␣ i3 Ϫ/Ϫ T cell membranes by exogenous G␣ i3 protein. Cell membranes prepared from G␣ i3 Ϫ/Ϫ T cells were preincubated with indicated concentrations of in vitro synthesized G␣ i3 or G␣ o protein before stimulation of the membrane by 100 ng/ml CXCL11 and assaying G protein activation as in A. The data represent three separate experiments with each sample in triplicate. * and **, statistical significance (p Ͻ 0.05 or p Ͻ 0.01, respectively), in the presence of G␣ i3 versus G␣ o protein.
In parallel experiments, immunoprecipitation of the CXCR3 receptor did not pull down G␣ i2 as expected, presumably because of its rapid dissociation from the CXCR3 receptor upon activation. The 35 S-G␣ i2 could interact well with the S1P 1 receptor, another PTX-sensitive GPCR (Fig. 4B, right panel), ruling out that undetectable 35 S-G␣ i2 in CXCR3 immunoprecipitates resulted from a lack of binding function of the protein.
Furthermore, significant GTP␥S incorporation induced by CXCL11 was seen in the cell membranes prepared from COS-7 cells transfected with WT CXCR3 receptor and all variants, except for the R149N variant that was totally ineffective (Fig.  4C). The inability of the R149N mutant to activate G␣ i2 and to interact with G␣ i3 suggests that G␣ i3 may compete for or sterically interfere with G␣ i2 binding to the CXCR3 receptor directly or indirectly via the DRY domain, by which it controls the amplitude of a migratory response elicited by the receptor.

DISCUSSION
Our results clearly show that a deletion of G␣ i3 augments G␣ i2 -mediated migratory responses toward CXCR3 ligands, whereas deletion of G␣ i2 ablates the response. The G␣ i3mediated inhibition of CXCR3 signaling is not reliant on the exchange of GDP for GTP by G␣ i3 , but it does require ligand-stimulated interaction between G␣ i3 and the CXCR3 receptor. Presumably, a conformational change of the receptor caused by ligand binding provides conditions that are optimal for G␣ i3 binding to the receptor. However, the receptorbound G␣ i3 appears unable to undergo a conformational change required for dissociation of GDP and/or binding to GTP, thus preventing it from full activation and leaving the receptor. By occupancy of the receptor, G␣ i3 blocks G␣ i2 activation. It has been shown that the DRY motif at the N terminus of the second intracellular loop is key for proper G protein recognition and interaction (31). Our data suggest that this motif may be also essential for the blockade induced by G␣ i3 , which requires further investigation. Substitution of Arg 149 amino residue to Asn ablates not only activation of G␣ i2 but also binding of G␣ i3 to the receptor. Under similar conditions, receptors carrying mutations at the C terminus can activate G␣ i2 and preserve G␣ i3 interaction similarly to the WT CXCR3 receptor (Fig. 4B). Lack of G␣ i3 binding to and G␣ i2 activation by the R149N mutant cannot be ascribed to the levels of receptor expression on the cells. Although the R149N-CXCR3 mutant was expressed on the cells at levels slightly lower than the WT CXCR3 receptor, its expression was comparable with the LLL-CXCR3 mutant (Fig. 4A) (24). The latter could activate G␣ i2 and bind with G␣ i3 as the WT CXCR3 receptor (Fig. 4, B and C). Murine G␣ i2 and G␣ i3 are 83% identical in amino residue sequence. This structural similarity may be the basis for their competition for the same binding domain of the CXCR3 receptor or for G␣ i3 interference with G␣ i2 interaction of the receptor. A subtle difference in the receptor conformation induced by a specific agonist is likely to determine whether the activated receptor can interact and activate a given G␣ i protein or interact with a G␣ i protein without inducing its full activation. Therefore, function of a given G␣ i protein as an activator or inhibitor is receptor-specific, and the interplay between these two G proteins may be essential in the control of the amplitude of a GPCR-induced signal. To the best of our knowledge, this is the first description of a G protein interacting with a receptor after ligand binding for the sole purpose of diminishing the other G protein interaction with the receptor. This finding gives novel insight into regulation of GPCR signaling by heterotrimeric G proteins.
GPCRs initiate a signaling cascade through activation of heterotrimeric G proteins following stimulation by an extracellular agonist (43). To meet the complexity and versatility of cell migration in the body, many chemokine receptors have evolved to recognize more than one chemokine, such as the CXCR3 and CCR7 receptors, or one chemokine receptor can be activated by multiple chemokines like CCL5 (regulated on activation normal T cell expressed and secreted), CCL3 (MIP-1␣), MCP-2, and MCP-3 (44). Our finding that a migratory response is tied proportionally to the amplitude of G protein activation indicates that stoichiometry of G protein activation is directly linked to the propensity of a cell to migrate. The precise activation level of the G␣ i2 protein stimulated by ligation of the CXCR3 receptor is determined by the affinity of the ligand as well as the interplay of G␣ i2 and G␣ i3 with the receptor. In accordance with CXCR3 signaling dependent on G␣ i2 and inhibition of the signaling by G␣ i3 , our in vivo study showed that similar to T cells from mice lacking the CXCR3 receptor, G␣ i2deficient T cells failed to elicit an acute graft-versus-host defense reaction after being transfused into full major histocompatibility complex-mismatched Balb/c SB-17 severe combined immunodeficiency mice (30,45). 4 In contrast, transfer of G␣ i3 -deficient T cells into the severe combined immunodeficiency mice stimulated an aggravating graft-versus-host defense response compared with WT T cells. Thus, an interplay among different heterotrimeric G proteins is likely to have a role to play in vivo, introducing another level of complexity in the control of cell migration systemically. The complex regulation of GPCR signaling at multiple levels is crucial in temporal and spatial regulation of cell migration in the body.
Determination of receptors linked to G␣ i proteins has been greatly aided by the use of PTX. PTX has also allowed insight into downstream effectors that are either partially or fully involved in propagating a signal by G␣ i proteins. Although PTX has provided extensive and important details about the agonist and signal transduction mechanisms, the specifics about individual G␣ i/o family members has gone largely unresolved because PTX blocks receptor interaction by several G␣ i/o family members. With the advent of the genetic manipulation of mice, we can tease out the specifics of G␣ i1 , G␣ i2 , and G␣ i3 in a migratory response stimulated by a specific receptor. Studies of B cell migration in G␣ i2 Ϫ/Ϫ mice suggest an irreplaceable role of G␣ i2 in chemotactic responses to chemokines CXCL12 and CXCL13 (14). On the other hand, G␣ i2 and G␣ i3 are redundant in thymic emigration as indicated by no significant defects in thymic exportation in G␣ i2 -and G␣ i3 -deficient mice (12,13). Although the proportions and the numbers of CD4 and CD8 single positive thymocytes were increased in G␣ i2 Ϫ/Ϫ mice as lck-PTX-transgenic mice, the increases were not caused by a defect in thymic egress, rather by an accelerated transition from the double positive to single positive thymocytes as shown by our previous investigations (12,46). In support, nearly normal numbers and percentages of CD4ϩ and CD8ϩ T cells were seen in the spleen of G␣ i2 Ϫ/Ϫ mice (11)(12)(13). Likewise, mature thymocytes in G␣ i3 Ϫ/Ϫ mice populated lymphoid tissues normally in the periphery, and there was no aberrant accumulation of single positive thymocytes in the mice (13). The redundancy of G␣ i2 and G␣ i3 in thymic egress was also consistent with our recent study showing that G␣ i2 -KO and G␣ i3 -KO T cells migrated indistinguishably from WT T cells in response to an increasing concentration of sphingosine 1-phosphate (S1P), a lipid mediator controlling T cell egress (33). 5 How the redundancy works mechanistically is not fully understood at present. Association with the same subunit of G␤␥ may be one of the common mechanisms whereby G␣ i2 and G␣ i3 can compensate for the absence of one to another. Hwang et al. (42) have shown that functional silence of either G␣ i2 or G␣ i3 by specific short interfering RNA has little effect on the migration of macrophages toward an increasing concentration gradient of C5a or C3a. Deletion of G␤2, however, ablated C5a-or C3a-provoked migration, suggesting that both G␣ i2 and G␣ i3 can transduce the migration signal as long as the G␤2 subunit is presented (42). On the contrary, the CXCR4 receptor requires both G␣ i2 and G␣ i3 for a full response. Absence of either gene impaired T cell migration induced by CXCL12 (Fig. 1F), although decreased expression of the CXCR4 receptor on these cells may also be partially involved. We should point out that a slight increase or decrease (10 -15%) in the G␣ i2 expression levels in G␣ i3 Ϫ/Ϫ T cells or G␣ i3 in G␣ i2 Ϫ/Ϫ T cells was observed in some mice ( Fig. 2A). However, the levels of G␣ i protein expression were not correlated with migratory changes for any of the studied chemokines. In particular, the slight decrease in G␣ i2 expression in G␣ i3 -KO T cells was discordant with the increased activity for the CXCR3 receptor. Similarly, although the levels of G␣ i2 and G␣ i3 expression were slightly decreased in activated as compared with nonactivated T cells, the decrease affects little their chemotactic response to CCL19 stimulation (Fig. 1, G and H). This is probably attributed to the following two reasons. First, G␣ i2 or G␣ i3 is a lot more abundant than any individual chemokine receptor in a cell. Second, constant recycling of G␣ i proteins from an active to inactive status further increases the size of a G␣ i protein pool that is already far in excess of any individual chemokine receptor. Therefore, distinct function of G␣ i2 and G␣ i3 as described in this study is unlikely ascribed to a slightly altered level of G␣ i2 or G␣ i3 protein. Nevertheless, this does not exclude that the absolute levels of G␣ i2 and G␣ i3 can affect activation of a given GPCR, when the relative levels of G␣ i2 and G␣ i3 differ drastically. Our data thus argue that G␣ i2 and G␣ i3 proteins may play overlapping, distinct, antagonizing, and additive roles depending on a specific receptor. The G␣ i2 -and G␣ i3null mutation mice provide a unique system to explore fundamental signaling differences of a specific GPCR coupling to G␣ i2 , G␣ i3 , or both and potential interplay between these G␣ i proteins.