Loss of Gαi proteins impairs thymocyte development, disrupts T-cell trafficking, and leads to an expanded population of splenic CD4+PD-1+CXCR5+/− T-cells

Thymocyte and T cell trafficking relies on signals initiated by G-protein coupled receptors. To address the importance of the G-proteins Gαi2 and Gαi3 in thymocyte and T cell function, we developed several mouse models. Gαi2 deficiency in hematopoietic progenitors led to a small thymus, a double negative (DN)1/DN2 thymocyte transition block, and an accumulation of mature single positive (SP) thymocytes. Loss at the double positive (DP) stage of thymocyte development caused an increase in mature cells within the thymus. In both models an abnormal distribution of memory and naïve CD4 T cells occurred, and peripheral CD4 and CD8 T cells had reduced chemoattractant responses. The loss of Gαi3 had no discernable impact, however the lack of both G-proteins commencing at the DP stage caused a severe T cell phenotype. These mice lacked a thymic medullary region, exhibited thymocyte retention, had a peripheral T cell deficiency, and lacked T cell chemoattractant responses. Yet a noteworthy population of CD4+PD-1+CXCR5+/− cells resided in the spleen of these mice likely due to a loss of regulatory T cell function. Our results delineate a role for Gαi2 in early thymocyte development and for Gαi2/3 in multiple aspects of T cell biology.

proteins [12][13][14] . Lymphocytes express little Gα i1 , and no lymphocyte phenotype has been reported in the Gnai1 −/− mice. In contrast, human and mouse lymphocytes strongly express Gα i2 , and a lesser amount of Gα i3 . Besides varying expression levels, the proteins localize differently within cells as Gα i2 resides predominately on the inner leaflet of the plasma membrane, while Gα i3 associates with plasma and intracellular membranes. Mice lacking Gα i3 and to a much greater extent Gα i2 exhibit immune phenotypes. Three day old neonatal Gnai3 −/− mice have reduced numbers of thymocytes and peripheral T cells, however, cell numbers rapidly return to the normal range 15 . In adoptive transfer experiments Gnai3 −/− thymic progenitor homed less well to the thymus. Otherwise, no other major defects in lymphocyte function have been reported. In contrast, severe immune defects have been described in the Gnai2 −/− mice. They develop a T helper type 1 dominated colitis whose penetrance depends upon the genetic background of the mice 16 . However, C57BL/6 mice kept in a clean mouse facility exhibit little, if any disease. Loss of Gα i2 also reduces thymic progenitor homing, and decreased the chemotactic responsiveness of DN1 cells to CXCL12 15,17 . Double positive (DP) thymocytes more rapidly transit to single positive (SP) cells, which is exaggerated in colitis-prone mice 18,19 . In the thymus, mature SP thymocytes accumulate, in part secondary to reduced egress 19 while in the periphery there is an increase in memory T cells 20 . Purified CD4 and CD8 T cells from Gnai2 −/− mice respond less well to chemoattractants and, functionally, the Gnai2 −/− T cells inefficiently cross endothelial barriers and show reduced migratory rates 21 . They are also poorly retained in lymph nodes following FTY-720 treatment 22 .
Besides underscoring the importance of Gα i2 in chemoattractant receptor signaling, the analysis of the Gnai2 −/− mice has suggested other functional activities for Gα i2 in T cells. For example, compared to wild type T cells, naïve Gnai2 −/− CD4 T cells (mixed background) had an enhanced intracellular calcium response, an exaggerated proliferative response, and augmented cytokine production following T cell receptor (TCR) crosslinking 20 . However, treating wild type CD4 T cells with pertussis toxin failed to reproduce these abnormalities. Furthermore, in Gnai2 −/− colitis prone mice, regulatory T cells did not normally inhibit effector memory CD4 T cells 23 . While intriguing these studies are complicated by variations in mouse genetic backgrounds, and that thymic T cells development occurs in a globally Gα i2 deficient animal or a Gnai2 −/− bone marrow reconstituted wild type animal.
To better understand the consequences of loss of Gα i proteins in T cells we have developed several different mouse models to assess the impact of the loss of Gα i2 and Gα i3 on T cell development, trafficking, and function. These studies confirmed the important of Gα i2 in T cell development and Gα i2 and Gα i3 in both thymocyte and peripheral T cell chemotaxis. They support a role for Gα i2 in maintaining naïve T cells and reveal an unexpected phenotype in the double deficient peripheral T cells.

Results
Generation of mouse models to assess the impact of the loss of Gα i proteins on thymocyte development. We extensively backcrossed the Gnai3 −/− and the Gnai2 fl/fl mice onto a C57BL/6 background.
Using the Gnai2 fl/fl mice, we deleted Gnai2 in hematopoietic progenitors using vav1-cre, and in DP thymocytes using cd4-cre. We crossed the Gnai2 fl/fl cd4-cre mice to Gnai3 −/− mice eliminating Gnai2 and Gnai3 expression at the DP thymocyte stage. We failed to generate viable Gnai2 fl/fl vav1-cre/Gnai3 −/− mice. Next, we compared the thymocyte profiles of the different mice. Shown are representative FACS profiles of purified thymocytes assessed for CD4 and CD8 expression, and SP CD4 thymocytes for CD62L and CD44 expression, which allows the distinction between immature and mature cells ( Fig. 1A and B). The non-conditional loss of Gα i2 decreased DP thymocytes and an increased SP CD4 and CD8 cells as previously reported 18,19 . Also, the % of mature SP CD4 thymocytes was increased. Deleting Gnai2 using cd4-cre led to similar profiles, although the changes were less marked. Conversely, deleting Gnai2 using vav1-cre produced a thymocyte phenotype like that observed in the Gnai2 −/− mice. The loss of Gnai3 had little impact on the thymocyte flow cytometry profiles. The Gnai2 fl/fl cd4cre/Gnai3 −/− (DKO) mice thymocytes resembled that of the Gnai2 −/− and the Gnai2 fl/fl vav1-cre mice.
We consistently observed a small thymus and reduced thymocyte numbers in the C57BL/6 Gnai2 −/− and the Gnai2 fl/fl vav1-cre mice, but not in the Gnai2 fl/fl cd4-cre mice suggesting a role for Gα i2 in progenitor homing and/ or early thymocyte development. To assess early thymocyte development, we examined the expression of CD44 and CD25 on DN thymocytes, which allows the separation of DN thymocytes into 4 consecutive developmental stages termed DN1-DN4 24 . Both the Gnai2 −/− and the Gnai2 fl/fl vav1-cre mice had evidence of a DN1 to DN2 transition block (Fig. 1C). When corrected for the number of thymocytes recovered from the wild type and Gnai2 fl/fl vav1-cre mice, the later had a three-fold excess of DN1 thymocytes (157,000 versus 62,000), yet one-third fewer DN2 thymocytes (83,000 versus 249,000). There also may be a problem in the DN2-DN3 transition as the WT cells expanded 8-fold while the Gnai2 fl/fl vav1-cre DN2 cells only expanded 4-fold. Despite a predicted homing defect, the absolute number of early thymocyte precursors 25 (ETPs, Lin − CD4 − CD25 − CD44 + CD117 + ) present in the Gnai2 fl/fl vav1-cre mice slightly exceeded those in the WT thymus. A breakdown of the DN3 compartment into DN3a and DN3b cells revealed similar percentages in the WT and the mutant mice suggesting no defect in β chain selection (data not shown). When we tested the expansion of DN1 and DN3 cells from the WT, Gnai2 fl/fl cd4-cre, Gnai2 −/− , and Gnai3 −/− mice in the OP-9 DL1 culture system, we found a severe defect with the Gnai2 −/− DN1 cells (Fig. 1D). They expanded poorly and generated few mature cells. In contrast, the Gnai2 −/− DN3 thymocytes exhibited no apparent defect (Fig. 1E). Representative flow cytometry patterns of WT and Gnai2 −/− DN1 thymocytes cultured for various durations on OP9 DL1 cells are shown (Fig. 1F). Rather displaying the results as dot blots we used contour plots due to the low number of Gnai2 −/− cells recovered from the cultures. In contrast to the WT DN1 cells, the Gnai2 −/− DN1 cells failed to generate any DP thymocytes. Overall these results are consistent with a role for Gα i2 in the DN1/DN2 thymocyte transition.  like controls with the exception a mild reduction in mesenteric lymph node cells ( Fig. 2A). Both the Gnai2 fl/ fl vav1-cre mice and the Gnai2 fl/fl cd4-cre mice had an increase in the % of SP thymocytes with a mature phenotype bias ( Fig. 2B and C). In addition, the Gnai2 fl/fl vav1-cre had an increase in the CD4/CD8 ratio not observed in the Gnai2 fl/fl cd4-cre. The Gnai2 fl/fl cd4-cre mice had a normal B cell compartment while, as expected, the Gnai2 fl/ fl vav1-cre mice had a loss of marginal zone B cells and reduced peripheral B cells (Fig. 2D-F). The B cell phenotype observed in the Gnai2 fl/fl vav1-cre mice is like that observed following a B cell specific deletion of Gnai2 26 .

Comparison of T cell compartments in
In both models peripheral T cells had higher ICAM-1 expression and lower CCR7 and CD62L expression (Fig. 2G). These expression changes may arise from the skewing of the memory/naïve CD4 T cell ratio in the mutant mice (Fig. 2H). An increase in memory/naïve CD4 T cell ratio has also been reported in the Gnai2 −/− mixed background mice 20 . Also both strains had an increase in CD62L high CD44 very low cells. This subset is reported to be enriched for naïve CD4 T cells with stem cell-like properties 27 . We checked the responsiveness of DN thymocytes to CXCL12 and CCL19 and mature SP CD4 thymocytes to the same chemokines and to S1P. The DN thymocytes from the Gnai2 fl/fl vav1-cre mice migrated poorly to optimal concentrations of CXCL12, while as expected the Gnai2 fl/fl cd4-cre mice DN thymocytes responded normally. A similar reduction in CXCL12 directed migration occurred when we examined the DN1 subset from the Gnai2 fl/fl vav1-cre mice (data not shown). The mature SP CD4 cells from both strains responded less well to CXC12 and CCL19 than did the controls although the Gnai2 fl/fl vav1-cre CD4 T cells were more impaired. Likely persistent Gα i2 protein expression in the Gnai2 fl/ fl cd4-cre SP thymocytes explains the difference. Thus, a loss of Gnai2 in hematopoietic progenitors reproduces many of the phenotypes reported with the Gnai2 −/− mice, while a cd4-cre mediated deletion predominately affected late thymocyte development and the composition of the peripheral T cell pool.

Analysis of the T cell compartments in the Gnai2 fl/fl cd4-cre/Gnai3 −/− (DKO) mice. Breeding
Gnai2 fl/fl cd4-cre/Gnai3 +/− males and Gnai2 fl/fl /Gnai3 −/− females occasionally generated DKO progeny. The additional loss of Gnai3 did not affect the numbers of thymocytes or splenocytes when compared to wild type mice, but it caused a reduction in the numbers of blood leukocytes and lymph node cells (Fig. 3A). The thymocyte profile again showed an increase in SP cells with a severe skewing of the mature/immature ratio ( Fig. 3B and C). The blood from these mice contained an increased percentage of B220 cells with a reduced percentage of CD4 and CD8 T cells (Fig. 3D). Surprisingly, the spleen contained near normal numbers of CD4 T cells, while lymph nodes had fewer CD4 and CD8 T cells than controls (Fig. 3E). In addition, the number and size of Peyer's patches were sharply reduced. There was no overt evidence of colitis and splenic B cell development proceeded normally (data not shown). The CD4 T cells in the spleen exhibited an unusual phenotype as nearly 70% expressed PD-1 and half of those co-expressed CXCR5 (Fig. 3F). The residual lymph node CD4 T cells had a similar PD-1 + CXCR5 +/− phenotype. There was a near complete loss of naïve CD4 T cells as the splenic and lymph node CD4 T cells lacked CD62L and expressed CD44 ( Fig. 3G and H). There was also a significant loss of regulatory T cells in the periphery (Fig. 3I). Further immunophenotyping revealed similar expression levels of CD25 and TCRβ, but increased levels of ICAM-1 and CXCR3 compared to wild type CD4 T cells (Fig. 3J). Despite the CD4 T cell Tfh-like phenotype, the spleen and lymph nodes contained few B cells with a germinal center phenotype (Fig. 3K). Furthermore, while the WT and DKO mice had similar serum levels of IgM, IgG 1 , IgG 2c , and IgG 3 ; the DKO mice had a 50% reduction in serum IgG 2b and 25% decrease in IgA (data not shown). Thus, the loss of Gnai3 and the additional loss of Gnai2 at the DP stage of thymocyte development led to a reduction in peripheral T cells, but the expansion of an unusual population of CD4 T cells in the spleen.

Response of DKO thymocytes and splenic T cells to chemokines.
We found that 13% of the wild type DP thymocytes migrated to CXCL12 while only 4% of the DKO cells responded (Fig. 4A). The SP CD4 and CD8 thymocytes had an approximately 70, 80, and 90% reduction in their responsiveness to optimal concentrations of S1P, CXCL12, and CCL19, respectively ( Fig. 4B and C). That some SP DKO thymocytes still specifically migrated to chemoattractant likely reflects persistent Gα i2 protein expression as pertussis toxin nearly eliminates all chemokine directed migration (Fig. 4D). The splenic CD4 T cells had a near complete loss in their chemokine responses while the splenic CD8 T cells retained some responsiveness to CXCL12 (Fig. 4E). Consistent with some of the CD8 T cells escaping the CD4-Cre mediated deletion of Gnai2, we found that they retained their sensitivity to pertussis toxin treatment (Fig. 4F). As expected splenic B cells from the DKO mice responded normally to CXCL12 and CCL19 (data not shown). These results suggest that nearly all the peripheral CD4 T cells in these mice lack Gα i2 and Gα i3 , while some CD8 T cells have not deleted the floxed Gnai2.
Reconstitution of irradiated wild type or Rag2 −/− mice with DKO bone marrow. Because of the difficulties in generating the DKO mice we used bone marrow from these mice or controls to reconstituted irradiated CD45.1 mice. We analyzed the mice 8-12 weeks after reconstitution. We fully reconstituted the thymus however, in the spleen CD45.1 cells persisted, particularly so, in the mice reconstituted with DKO bone marrow (Fig. 5A). The cell populations in the thymus, spleen, lymph nodes, Peyer's patches, and bone marrow were assessed by gating on CD45.2 positive cells (Fig. 5B). The thymus phenotyping resembled that of the non-reconstituted DKO mice (data not shown), however, the lymphoid organs and the blood had few if any DKO CD4 and CD8 T cells (Fig. 5C). The population of PD-1 + CXCR5 +/− CD4 T cells observed in the non-reconstituted mice had largely disappeared. As regulatory T cells may persist in irradiated hosts 28 , we performed a similar reconstitution experiment using Rag2 deficient mice as the recipient. In the control and DKO bone marrow reconstituted Rag2 −/− mice, we found nearly equivalent cell numbers in the major lymphoid organs and the blood with the exception of peripheral lymph nodes (Fig. 5D). As in the non-reconstituted DKO mice, the DKO bone marrow reconstituted Rag2 −/− mice had numerous PD-1 + CXCR5 +/− CD4 T cells in the spleen and the blood (Fig. 5E-G). As previously, the numbers of regulatory T cells derived from the DKO bone marrow were reduced in the periphery (Fig. 5H). Phenotypically the CD4 T cells in the Rag2 reconstituted mice resembled the CD4 T cells in the non-reconstituted DKO mice with low levels of CCR7 and CD62L, and increased expression of ICAM-1, CXCR3, and CD44 (Fig. 5I). These CD4 T cells were refractory to chemokines (Fig. 5J). The assessment of thymocyte development appeared similar to that noted in the non-reconstituted DKO mice (Fig. 5K). These data suggest that a population of CD4 T cells deficient in Gα i2/3 expands in the spleen when regulatory T cells are lacking and those present likely to be dysfunctional. Further characterization of the peripheral CD4 + PD-1 + CXCR5 +/− T cells and memory CD4 T cells from the DKO mice. Typical CD4 follicular helper T cells (Tfh) express high levels of PD-1 and CXCR5, and low levels of CCR7 29 as did the CD4 T cells found in the spleens of the DKO mice. Classic Tfh cells also express high levels of ICOS, the transcription factor Bcl-6, and they produce the cytokine IL-21. In contrast, the DKO CD4 PD-1 + CXCR5 + T cells did not have observable increases in ICOS or Bcl-6. We also noted that the Gnai2-deficient Tfh had reduced ICOS and Bcl-6 expression compared to controls (Fig. 6A and B). This suggests some role for Gα i2 in generating elevated ICOS expression and a requirement for Gα i2/3 for functional Tfh cells.
As noted previously the DKO CD4 T cell compartment essentially lacked naïve CD4 cells as most expressed high levels of CD44 and low levels of CD62L. To provide some assessment of the memory CD4 T cell compartment in the DKO mice, we compared the basal cytokine levels in CD4 CD44 high CD62L low T cells isolated from WT, Gnai2 fl/fl cd4-cre, Gnai2 fl/fl vav1-cre or DKO mice. We found that compared to WT cells the DKO CD4 T cells had reduced amounts of all the cytokines tested (Fig. 6C). The Gnai2 fl/fl cd4-cre, or Gnai2 fl/fl vav1-cre memory CD4 T cells had a small reduction in their basal level of intracellular interferon-γ, but little change in the levels of the other cytokines (Fig. 6C). Thus, CD4 memory cells are likely to be functionally impaired in the DKO mice.
Loss of the normal thymus and spleen architecture in the DKO mice. Hematoxylin and eosin stained thymus sections examined by confocal microscopy using a visible light photomultiplier tube revealed small, fragmented medullary regions in the DKO thymus (Fig. 7A). Thick sagittal sections of the thymus were immunostained with antibodies directed at CD4, CD8, UEA-1, and ER-TR7 (Fig. 7B). UEA-1 serves as a marker for thymic medullary epithelial cells 30 , and ER-TR7 delineates the perivascular space surround egress vessels 31 . Stitched multicolor confocal images showed small, scattered medullary regions in the DKO thymus. Higher magnification showed that thymocytes had accumulated in the perivascular spaces surrounding putative egress blood vessels (Fig. 7C). Immunostaining for CD3 and UEA-1 demonstrated that the perivascular thymocytes had a mature phenotype (Fig. 7D) while immunostaining for Gα i2 protein verified the loss of Gα i2 in the DKO thymocytes surrounding the egress vessels (Fig. 7E). Analysis of DKO spleen sections for B220, CD4, CD35, CD169, and Ki67 immunoreactivity revealed small T cell zones in the white pulp, while the red pulp contained numerous CD4 cells (Fig. 7F). The splenic red pulp also contained numerous Ki67 + cells, some of which expressed CD4. Examining the B cell follicles at higher power demonstrated the presence of numerous CD4 cells and the lack of organized germinal centers (Fig. 7G). Thus, in the DKO mice CD4 T cell populate both splenic white and red pulp, although the usual spleen T cell architecture is severely disrupted.

Discussion
Several conclusions can be drawn from the previous studies of T cell phenotypes in non-conditional Gnai2 and Gnai3 knockout mice 15, 16, 18, 20-22, 32, 33 and this study of conditionally deleting Gnai2 in the context of deleting Gnai3, or not. First, Gα i2 can mediate the essential function of Gα i proteins for thymocyte development when Gα i3 is not available, while Gα i3 cannot when Gα i2 is not available. Second, either the loss of Gα i2 in hematopoietic progenitors or the non-conditional loss of Gα i2 severely reduces the size of the thymus, interferes with DN thymocyte differentiation, and impairs thymocyte egress. These phenotypes are likely thymocyte intrinsic. The loss of Gα i2 at the DP stage is less detrimental resulting in an accumulation of mature SP cells. Third, the conditional loss of Gα i2 in either model decreased the responses of peripheral CD4 and CD8 T cells to chemokines, increased the percentage of memory-like CD4 T cells, and enriched for a population of CD4 + CD62L high CD44 verylow cells. Fourth, the loss of Gα i3 combined with the loss of Gα i2 at the DP stage leads to a disorganized thymus with a near absence of a medullary region; decreased DP and increased mature SP thymocytes, and a severe thymocyte egress defect. Fifth, despite the lack of Gα i2 and Gα i3 , mature thymocytes can access the perivascular space of the thymus egress blood vessels. Sixth, despite thymic retention and loss of chemoattractant responsiveness numerous CD4 T cells accumulate in the DKO spleen. Yet when wild type mice are reconstituted with DKO bone marrow these cells are largely eliminated. Finally, functional Tfh and Tregs do not develop in the absence of Gα i2/3 .
Bone marrow derived T cell progenitors redundantly use the Gα i dependent chemokine receptors CXCR4, CCR7, and CCR9 to enter the neonatal thymus 34 . The lack of Gα i2/3 in bone marrow T cell progenitors would likely produce a phenotype like that observed with the triple receptor knockout mice. Unfortunately, we never identified a viable Gnai2 fl/fl vav1-cre/Gnai3 −/− mouse. We did find that 6-12 week old Gnai2 −/− and Gnai2 fl/ fl vav1-cre mice had visibly smaller thymuses than the controls. Others have reported fewer thymocytes in Gnai2 −/− mice post weaning 17 . Surprisingly, we did not find fewer ETPs in the Gnai2 fl/fl vav1-cre mice indicating that Gα i3 compensated for the loss of Gα i2 for entrance into the thymus. The accumulation of DN1 thymocytes in the Gα i2 deficient animals suggested a partial block at the DN1/DN2 transition. Upon entering the thymus early progenitors receive signals from the thymic microenvironment that initiate T cell lineage specification  and progenitor expansion 35 . The DN1/DN2 transition strongly depends upon Notch1 signaling. Poor access to Notch1 ligands due to the migratory defect present in Gnai2 fl/fl vav1-cre DN1 thymocytes could account for this observation. However, directly plating Gα i2 deficient DN1 cells on OP-9DL1 cells argued against this hypothesis. Further implicating Gα i2 in Notch signaling, marginal zone B cell development, which depends upon Notch2, is severely impaired in the Gnai2 fl/fl vav1-cre mice. Direct plating Gα i2 deficient bone marrow progenitors on OP9-DL1 cells does not rescue marginal zone B cell development (I-Y. Hwang, unpublished observation). We are exploring whether the loss of Gα i2 impairs the integrin mediated cell-cell interactions needed to trigger Notch receptor processing, or whether the Gα i2 deficiency causes a cell intrinsic defect in Notch processing due to impaired Adam10 or γ-secretase activity.
As DN1 thymocytes differentiate to DN2 and DN3 cells they move into the thymic cortex and eventually to the subcapsular zone. Whether this migration is chemokine directed remains controversial 3,36 . Once they reach the subcapsular zone, the DN3 cells transition to DN4 cells. The conditional deletion of CXCR4 using cre expressed from the proximal Lck promoter caused a partial block in the transition of DN3 to DN4 cells 37 . We did not find an obvious DN3-DN4 transition block in the Gnai2 fl/fl vav1-cre or the Gnai2 −/− mice, although the loss of Gα i2 did impair the in vitro migration of both DN3 and DN4 thymocytes to CXCL12 (3-5 fold decrease). Arguing that Gα i3 may compensate for the loss of Gα i2 in DN3/4 cells, Gnai2 −/− DN3 thymocytes plated on OP9-DL1 stromal cells expanded like wild type DN3 cells.
From the subcapsular zone the DN4 thymocytes migrate into the cortex and toward the medulla 3 . Coincidently they express CD4 and CD8 and transition to the DP stage. CXCR4 levels begin to decrease reducing the cortical retention signal. The deletion of Gα i2 prior to entrance into the thymus or under the control of cd4-cre had no apparent effect on the appearance of DP cells although DP thymocytes from both strains had reduced response to chemokines. As the thymocyte transit the cortex they undergo positive selection, and subsequently further reduce CXCR4 and increase CCR7 and CCR4 expression. Based on pertussis toxin data entrance into medulla depends upon Gα i signaling although a consensus is lacking on which receptors are required 38 . The non-conditional loss of Gα i2 or its loss in hematopoietic progenitors led to a thymus with small, fragmented medullary regions, while the combined loss of Gα i3 and Gα i2 at the DP stage led to a near absence of thymocytes in the residual medullary regions.
Semi-mature SP thymocytes that successfully undergo negative selection reduce their expression of CD69 and raise their expression of CD62L and the chemoattractant receptor S1PR1, which re-localizes the cells to the corticomedullary junction leading to their Gα i dependent reverse transmigration into the blood via sensing of sphingosine-1 phosphate 39 . The loss of Gα i2 interferes with thymocyte access to the medullary region, but it does not prevent the maturation of SP cells. The loss of Gα i2 reduces their responsiveness to S1P, which may account for their accumulation in the thymus. The DKO thymocytes have a severely attenuated response to chemokines and S1P, yet they also acquire a mature phenotype. Surprisingly, both the Gα i2 deficient and the DKO thymocytes can access the perivenule channels (PVC) surrounding the normal egress blood vessels, although they may have accessed the PVC prior to exhausting their supply of Gα i2 protein.
The cd4-cre and vav1-cre Gnai2 fl/fl mice shared similar peripheral T cell phenotypes. Alteration in thymus egress and responsiveness to chemokines may explain the changes in the CD4 naïve and memory pools we observed. The elevated ICAM-1 expression suggests ongoing immune activation in these mice. However, no overt autoimmunity was noted and the reduction in CD4 ICOS levels does not support ongoing T cell activation. ICOS levels, which normally rise on Tfh cells, did not increase on the Gα i2 deficient Tfh-like cells suggesting that CXCR5 signaling supports ICOS expression on Tfh cells. Finally, we found only minimal changes in the basal cytokine levels in naïve or memory CD4 cells that lacked Gα i2 .
We cannot fully account for the presence of CD4 T cells in the spleen of the DKO mice. Pertussis toxin lck transgenic mice apparently lacked these cells 7 . Yet their presence in the periphery of the DKO mice argues that an occasional thymocyte transmigrated into the blood or lymph, despite the loss of Gα i2/3 proteins; or some thymocytes escaped from the thymus before cre mediated Gnai2 deletion, or retained sufficient Gα i2 protein to escape. Having left the thymus the newly emigrated CD4 T cells that happened to retain Gα i2 expression could access lymph nodes and the splenic white pulp, but as they exhaust their Gα i2 supply they would be trapped in lymphoid organs, or relegated to the blood and splenic red pulp. In the DKO mice those CD4 T cells that escaped the thymus expanded, and only a mild reduction in CD4 T cells occurred in the spleen. They evidently proliferated to fill the splenic niche. Many of these splenic CD4 cells expressed high levels of CD44 and PD-1, low levels of CCR7 and CD62L, and some co-expressed CXCR5. Despite this phenotype, they lacked other features of Tfh cells. Regulatory T cells evidently restricted their expansion as the DKO bone marrow reconstituted wild type mice lacked these cells, while they were abundant in DKO bone marrow reconstituted Rag2 −/− mice. CD4 T cell made up more than 70% of the lymphocyte gate in the spleen and blood of the reconstituted Rag2 −/− mice. The mice exhibited features of a CD4 T cell lymphoproliferative disorder, likely because of the loss of regulatory T cell function.
In conclusion, the homing of thymocyte bone marrow progenitors, thymocyte development, and thymocyte egress partially depends on the availability of Gα i2 , and completely depends upon Gα i2 and Gα i3 . The loss of Gα i2 impairs the expansion of DN1 thymocytes despite the availability of Notch ligands and IL-7. Further investigation of the role of Gα i proteins in Notch signaling in both B and T cell development is warranted. The trafficking of peripheral T cells, their proper localization in lymphoid organs, and their recruitment to inflammatory sites also partially depend upon the availability of Gα i2 , and completely upon Gα i2 and Gα i3 . If peripheral CD4 T cells lack Gα i2/3 they expand and predominately localize in the spleen and blood. Many of these CD4 T cells express high levels of PD-1, ICAM-1, and CD44; and low levels of CD62L, ICOS, and CCR7; and many co-express CXCR5. These cells cannot traffic properly, populating the spleen and accumulating in the blood. Further investigation of mice with more refined deletion of Gα i subunits should provide additional insights to their functional roles in thymocyte development, and mature T cell function.

Material and Methods
Mice and bone marrow reconstitutions. C57BL/6, B6.SJL-Ptprc a Pepc b /BoyJ (CD45.1), C57BL/6 vav1cre, and C57BL/6 cd4-cre mice were obtained from Jackson Laboratory. Mice with targeted deletion in Gnai3 and Gnai2, and Gnai2 fl/fl mice were kindly provided by Dr. Lutz Birnbaumer (NIEHS, NIH) and backcrossed more than 12 times to C57BL/6 mice 26 . Gnai2 fl/fl /cd4-cre and Gnai2 fl/fl/ vav1-cre mice were obtained by crossing the appropriate cre expressing strain with the Gnai2 fl/fl mice and backcrossing to obtain the desired genotype. The Gnai2 fl/fl cd4-cre/Gnai3 −/− (DKO) mice were generated by breeding a Gnai2 fl/fl cd4-cre/Gnai3 +/− male mouse versus a Gnai2 fl/fl /Gnai3 −/− female, the only combination that successfully produced viable progeny. For those experiments that directly compared WT and gene targeted mice, littermate controls were used when possible. Otherwise age and sex matched cd4-cre, vav1-cre, or Gnai2 fl/fl mice served as controls. For bone marrow reconstitution, twenty 7 weeks old CD45.1 mice were irradiated twice with 550 rads for total of 1100 rads and received bone marrow from C57BL/6 CD45.2 mice (control) or from the indicated gene targeted CD45.2 mice. The engraftment was monitored by sampling the blood 28 days later. The mice were used 6-8 weeks after reconstitution. All mice were used in this study were 6-14 weeks of age. Mice were housed under specific-pathogen-free conditions. All the animal experiments and protocols used in the study were approved and carried out in accordance with the guidelines of the NIAID Animal Care and Use Committee (ACUC) at the National Institutes of Health.
Chemotaxis assays. Chemotaxis assays were performed using a transwell chamber (Costar) as previously described 42 . The numbers of cells that migrated to the lower well after 2 h or 3 h incubation were counted using a MACSQuant flow cytometer (Miltenyi Biotec). The percent migration was calculated by the numbers of cells of a given subset that migrated into the bottom chamber divided by the total number of cells of that subset in the starting cell suspension, and multiplying the results by 100. D-erythro-sphingosine 1-phosphate was purchased