Selective Regulation of CD8 Effector T Cell Migration by the p110γ Isoform of Phosphatidylinositol 3-Kinase¹

The directed migration of T cells both in a constitutive environment as well as in the presence of inflammation is critical to the induction of an appropriate immune response (1). Chemoattractants such as chemokines play an important role in the process of T cell localization to specific tissues in a situation-appropriate manner. Differences in temporal and spatial expression of chemokines correlate with the various stages of the T cell immune response. In the absence of inflammation, naive T cells constitutively circulate from the blood into secondary lymphoid organs such as the lymph nodes and spleen. Naive T cells must access the lymphoid organs because it is there that they encounter Ag and subsequently become activated. Lymph node trafficking depends on T cell expression of L-selectin (2) and the chemokine receptor CCR7, which binds the constitutively expressed chemokines CCL19 and CCL21 (3, 4). Signaling through CCR7 results in the activation of the β₂ integrin LFA-1, which mediates the adhesion of T cells to the high endothelial venules through which they must traverse to enter the lymph nodes (5).

Upon activation, naive T cells differentiate into effector T cells, which display distinctly different patterns of trafficking than naive T cells. Effector T cells lose L-selectin (CD62L)³ and CCR7 expression and thus are restricted in their ability to undertake further migration to lymphoid organs (6). In contrast to naive T cells, effector T cells migrate to nonlymphoid sites, such as the liver and lung (7). Though it is not entirely clear whether specific chemokines drive effector T cell migration to particular nonlymphoid tissues, there is a distinct group of chemokines that differentially attracts effector T cells, as the receptors for these chemokines expressed at inflammatory sites are not highly expressed before T cell activation (8). The ability of effector T cells to access nonlymphoid sites allows them to migrate to inflamed sites throughout the body, where they can carry out their role in resolving infection and disease.

The PI3K family of lipid kinases has been proposed to play a major role in regulating the migration of T cells and other leukocytes in response to chemokines and other chemoattractants (9). The p110γ isoform of PI3K is unique among the class I PI3K isoforms in that it is only activated as a result of G protein-coupled receptor stimulation (10, 11, 12). Analysis of three independent p110γ knockout (p110γ^−/−) mouse models demonstrated a dramatic impairment in the recruitment of p110γ^−/− neutrophils and macrophages in response to a variety of inflammatory chemotactic signals (13, 14, 15). In vivo, these defects were significant in that p110γ^−/− neutrophils and macrophages were largely unable to traffic to the site of inflammation in two different models of peritonitis (15). Pharmacological inhibition of p110γ with a p110γ-specific inhibitor inhibited the migration of neutrophils into the peritoneum in response to the chemokine CCL5 (RANTES) (16). A p110γ-specific inhibitor was also shown to reduce the increased number of leukocytes present in systemic lupus erythematosus-afflicted MRL-lpr mice that contributed to sustained inflammation and tissue damage. As a result, these mice were able to recover from disease even if inhibitor treatment was started after the onset of symptoms (17). In a mouse model of collagen-induced arthritis, treatment with the same p110γ inhibitor resulted in a reduction of swelling and less severe cartilage erosion due to reduced infiltration of autoreactive leukocytes (18).

Initial studies with p110γ-deficient mice did not extensively analyze the migration responses of p110γ-deficient T cells. However, a defect in T cell activation in response to anti-CD3 Abs in vitro was noted in one study (15). More recent studies have demonstrated that p110γ-deficient naive T cells exhibit reduced chemotaxis in vitro in response to the CCR7 chemokine receptor ligand CCL19 and the CXCR4 ligand CXCL12, but not in response to the CCR7 ligand CCL21 (19). Similar results were reported by Nombela-Arrieta et al. (20), who also demonstrated a major role for the adapter protein DOCK2 in regulating naive T cell migration independent of PI3K. Imaging studies in vivo also indicate that DOCK2, but not p110γ, plays a major role in regulating migration velocity in lymph nodes, although p110γ^−/− T cells did exhibit increased turning angles (21). Although these studies show that p110γ plays a minor role in the chemotactic responses of naive T cells, effector T cells respond to a different array of chemoattractants that are expressed in inflammatory sites. Because p110γ^−/− neutrophils and macrophages display migration defects into inflammatory sites, we hypothesized that effector T cells may also be dependent on p110γ for efficient migration into inflammatory sites during the course of a primary T cell immune response.

In this report, we use p110γ-deficient mice (13) to demonstrate that p110γ does not play a major role in the migration of naive CD8 T cells in vitro and in vivo, or in activation and subsequent differentiation into effector T cells in vitro and in vivo. However, p110γ-deficient effector CD8 T cells exhibit impaired migration in vitro and exhibit reduced migration into the inflamed peritoneum following vaccinia virus (VV) challenge in vivo. Furthermore, p110γ^−/− mice exhibit increased susceptibility to VV infection when compared with wild-type (WT) mice.

The p110γ knockout (p110γ^−/−) mice were provided by Dr. C. Abrams (University of Pennsylvania, Philadelphia, PA) (13). These mice, which express GFP under the control of the endogenous p110γ promoter, were then highly backcrossed (>10 generations) to the C57BL/6 background. C57BL/6 and B6.SJL (CD45.1⁺) mice were purchased from Taconic Farms. The p110γ knockout mice were bred with CD45.1 mice to generate CD45.1 p110γ^−/− mice. B6.PL-Thy1^a/CyJ (Thy1.1) mice were purchased from The Jackson Laboratory. The p110γ knockout mice were bred with Thy1.1 mice to generate Thy1.1 p110γ^−/− mice. The p110γ knockout mice were also bred with OT-I mice, provided by Dr. K. Hogquist (University of Minnesota, Minneapolis, MN), to generate OT-I p110γ^−/− mice, which have a class I-restricted transgenic TCR specific for the OVA peptide SIINFEKL (22). Mice were used between 8 and 12 wk of age. All experimental protocols involving the use of mice were approved by the Institutional Animal Care and Use Committee at the University of Minnesota.

CCR7 expression was assessed using CCL19-Fc, biotinylated anti-human IgG Fcγ fragment-specific Ab and allophycocyanin-conjugated streptavidin (eBioscience). General PI3K activity was inhibited with wortmannin (Sigma-Aldrich). Granzyme B expression was assessed by flow cytometry with PE-conjugated anti-human granzyme B (Caltag Laboratories). All additional directly conjugated Abs were purchased from eBioscience or BD Pharmingen.

The Western Reserve strain of VV (VV-WR) was provided by Dr. M. Bevan (University of Washington, Seattle, WA). Recombinant VV-GFP-JAW-OVA (VV-OVA) was provided by Dr. J. Yewdell (National Institutes for Health, Bethesda, MD). This virus expresses the OVA_257–264 epitope fused C-terminally to GFP and the transmembrane region of JAW-1. Mice were i.p. infected with viral doses of 2, 5, or 10 × 10⁶ PFU as indicated. Viral titers were determined by plaque assay with 143B cells.

Chemotaxis assays were performed in Transwell chambers with 5-μm polycarbonate membranes (Costar). Recombinant mouse CCL21 (R&D Systems) was diluted to the appropriate concentrations in migration medium (RPMI 1640 supplemented with 1% BSA, 10 mM HEPES, penicillin/streptomycin, and l-glutamine) and added to the lower chamber of the Transwells. The membranes were placed on top, and 1–2 × 10⁶ WT or p110γ^−/− lymph node cells were loaded into the upper chamber in migration medium. The cells were allowed to migrate for 2 h at 37°C in 5% CO₂. Migrated cells were collected, pelleted, and resuspended in 200 μl of ice-cold FACS buffer (HBSS supplemented with 0.1% bovine calf serum and 0.2% sodium azide). Cells were stained with anti-CD3 and anti-CD8 Abs, and a fixed number of PKH26 reference beads (Sigma-Aldrich) was added to samples, which were then analyzed by flow cytometry.

Naive T cell migration in vivo was assessed as previously described (23). Briefly, peripheral lymph node cells from WT (Thy1.2) and p110γ^−/− (Thy1.1) mice were stained with 0.4 μM CellTracker Green CMFDA (Invitrogen Life Technologies), combined at a ratio of 1:1 (3 × 10⁶ cells of each genotype), and injected i.v. into B6 recipients. Recipient peripheral lymph nodes and spleen were harvested 1 h postinjection with stained cells. Donor cells were identified by CellTracker staining, whereas WT and p110γ^−/− T cells were distinguished within an individual recipient by Thy1.1/1.2 expression.

WT and p110γ^−/− mice were infected with VV-WR at the indicated doses. On days 1, 3, and 4 postinfection, mice were sacrificed and peritoneal cells were removed by washing with 5 ml of PBS. Peritoneal cell suspensions were stained with directly conjugated anti-CD3, anti-CD8, and anti-NK1.1 Abs to identify CD8 T cells and NK cells by flow cytometry. For survival experiments, mice were infected with the indicated dose of virus and monitored for a period of up to 14 days. Animals were weighed every other day and were euthanized upon loss of 25% of original body weight.

CD8⁺ CD45.1⁺ OT-I cells (5 × 10⁵ cells/recipient mouse) from either WT or p110γ^−/− mice were labeled with 1.66 mM CFSE and injected i.v. into C57BL/6 recipients 24 h before infection with 2 × 10⁶ PFU VV-OVA. On days 4, 8, and 12 postinfection, recipient spleens, peritoneal cells, and pancreatic lymph nodes were collected and stained with directly conjugated anti-CD45.1, anti-CD8, anti-CD62L, and anti-CD49d Abs for analysis by flow cytometry.

CD8⁺CD45.1⁺ effector cells were generated by i.v. injecting C57BL/6 recipients with 5 × 10⁵ WT or p110γ^−/− OT-I cells and infecting the recipients with 5 × 10⁶ PFU VV-OVA 1 day later. On day 6 postinfection, effectors were harvested from the spleen, rested for 5 h, labeled with CellTracker Green CMFDA, and injected into secondary recipients that had been injected with 5 × 10⁶ PFU VV-WR 2 days prior. At 18 h after the cell transfer, secondary recipients were sacrificed and peritoneal cells and spleens were harvested. Flow cytometry was used to assess the ratio of p110γ^−/− to WT cells in each location. Donor effector cells were identified by staining with directly conjugated anti-CD8 and anti-CD45.1 Abs and WT and p110γ^−/− effector cells were distinguished from one another in the same mouse through differential CellTracker staining.

CD8 effector T cells were generated in vitro as previously described (24). Briefly, splenocytes from WT and p110γ^−/− mice were stimulated with 1 μg/ml anti-CD3 mAb 2C11 for 48 h, then washed and resuspended in medium containing 20 ng/ml recombinant mouse IL-2 (R&D Systems). Cells were fed with fresh medium containing IL-2 every 48 h. Cells were used at days 8–10 of culture.

Conjugate assays were performed as previously described (25). Briefly, T cells were purified from WT or p110γ^−/− OT-I lymph nodes by negative selection using Abs against I-A^b, B220 and mouse IgG. T cell depleted splenocytes were used as APCs, which were left untreated or were pulsed with specified doses of SIINFEKL peptide (Invitrogen Life Technologies) for 30 min at 37°C. For conjugate formation, 5 × 10⁵ T cells were incubated with 5 × 10⁵ APCs for 10 min at 37°C in 96-well round-bottom plates (Costar). Following conjugation, cells were fixed in 1% paraformaldehyde (Electron Microscopy Sciences) for 20 min at room temperature. Cells were washed twice in PBS and stained with anti-Vα2, anti-CD8α, and anti-B220 for analysis by flow cytometry. Conjugates were defined as CD8α⁺Vα2⁺B220⁺ events.

Data were analyzed using the unpaired two-tailed Student’s t test as calculated by Prism software (Graphpad). Graphs show mean values and error bars represent the SD. The p values obtained are indicated for each experiment.

Given the established role for p110γ in mediating leukocyte migration (13, 14, 15, 26), we analyzed the migration of WT and p110γ^−/− naive T cells in vitro in response to CCL21, a chemokine that engages the CCR7 chemokine receptor (3, 4). CCR7 was expressed similarly on both WT (average mean fluorescence intensity = 91.98) and p110γ^−/− (average mean fluorescence intensity = 104.64) naive (CD44^low) CD8 T cells (Fig. 1,A). We tested the ability of these T cells to respond chemotactically to CCL21 by in vitro Transwell chemotaxis assays. At all chemokine concentrations tested, we observed no significant differences in the percentages of WT and p110γ^−/− CD8 T cells that were capable of migrating in response to a CCL21 gradient (Fig. 1,B). Finally, we analyzed the relevance of p110γ inactivation to naive CD8 T cell migration in vivo. Lymph node T cells from WT (Thy1.2) and p110γ^−/− (Thy1.1) mice were isolated, stained with CellTracker Green, then injected into C57BL/6 (Thy1.2) recipients. After 1 hour, recipients were sacrificed, and cell suspensions of the spleen, peripheral and mesenteric lymph nodes were analyzed for the presence of donor WT and p110γ^−/− CD8 T cells. Consistent with our in vitro results, p110γ^−/− CD8 T cells were able to migrate into peripheral lymph nodes, the mesenteric lymph node, and spleen as efficiently as WT CD8 T cells (Fig. 1 C). Taken together, these data indicate that p110γ is not required for naive CD8 T cell migration in vitro or in vivo.

Previous studies have suggested that p110γ^−/− T cells have an impaired ability to proliferate in response to in vitro stimulation by anti-CD3 Abs (15). We examined various aspects of CD8 T cell responses to either Ag or anti-CD3 Ab as a mimic of TCR stimulation. p110γ^−/− T cells expressing the OT-I transgenic TCR could form conjugates with SIINFEKL-loaded APCs as well as WT OT-I T cells at all doses of Ag tested (Fig. 2,A). This suggests that initial interactions of T cells with APCs critical for T cell activation are not impaired in the absence of p110γ. We also generated WT and p110γ^−/− CD8 effector T cells using a well characterized in vitro system that uses anti-CD3 mAb stimulation and IL-2 (24). p110γ^−/− CD8 T cells expanded similarly to WT T cells (Fig. 2 B). In addition, both WT and p110γ^−/− effector T cells at day 8 of culture were CD8⁺ and expressed low levels of CD62L and high levels of CD44 (data not shown). Thus, these results suggest that p110γ^−/− T cells do not have inherent defects in TCR-dependent activation responses.

We used a VV infection model and an adoptive transfer approach to study the activation, proliferation and differentiation of p110γ^−/− OT-I T cells in a WT environment in vivo. Naive WT and p110γ^−/− OT-I cells were harvested from the peripheral lymph nodes of donor mice and stained with CFSE before injection into C57BL/6 recipients. These recipient mice were subsequently challenged 24 h after transfer with an i.p. injection of 2 × 10⁶ PFU VV-OVA. The peak of clonal expansion of WT CD8 T cells occurred at day 4 in the spleen and peritoneum-draining pancreatic lymph node (Fig. 3,A) and there was no significant difference in the number of WT and p110γ^−/− OT-I T cells that was found in either of these locations. Additionally, both WT and p110γ^−/− OT-I effector T cells exhibited a uniform down-regulation of CD62L (Fig. 3,B, left panels) and diluted CFSE to a comparable degree (data not shown). Expression of α₄ integrin, which is critical for migration of effector cells into the peritoneum (27, 28), was equivalent on WT and p110γ^−/− OT-I effector cells in the spleen and pancreatic lymph node as well (Fig. 3,B, right panels). Finally, both WT and p110γ^−/− OT-I effectors in the spleen and peritoneum produced similar amounts of IFN-γ upon restimulation and expressed comparable levels of granzyme B (Fig. 3 C), indicating that p110γ is not required for CD8 T cell differentiation into effectors with cytotoxic potential. Thus, these results show that when p110γ^−/− OT-I T cells are challenged in a WT environment, they become activated, proliferate, and differentiate into effector CD8 T cells just as well as WT OT-I cells.

We performed a series of experiments over a short time course (day 1–4 postinfection) to study the endogenous CD8 T cell response to VV infection in WT and p110γ^−/− mice. In a noninflamed WT peritoneum, there were relatively few resident CD8 T cells. However, there were even fewer resident CD8 T cells in the peritoneum of uninfected p110γ^−/− mice (Fig. 4,A). The i.p. infection of WT mice with VV-WR resulted in an influx of CD8 T cells into the peritoneum (Fig. 4,B). However, there was a reduced influx of CD8 T cells into the peritoneum of p110γ^−/− mice infected with VV-WR at all time points examined (Fig. 4,B). As a control, we tracked the influx of NK cells into the inflamed p110γ^−/− peritoneum (Fig. 4,C), as previous studies have suggested that p110γ is important in mediating NK cell migration in response to inflammatory chemokines in vitro (29). Accordingly, we observed a reduction in the number of NK cells that migrated to the p110γ^−/− peritoneum. The phenotype of CD8 T cells in the spleen and peritoneum following VV-WR infection of WT and p110γ^−/− mice was similar; CD62L expression decreased and CD44 expression increased over time following virus infection (Fig. 4 D). These results show a reduced influx of endogenous CD8 T cells in p110γ^−/− mice following VV-WR infection, despite a normal activation phenotype.

Although the influx of endogenous p110γ^−/− CD8 T cells to the inflamed peritoneum is clearly impaired, differences in the p110γ^−/− vs WT inflammatory environments may affect the ability of various leukocyte subsets to migrate into the peritoneum. To determine whether there are T cell-intrinsic defects in the ability of p110γ^−/− CD8 effector T cells to migrate into an inflammatory site in vivo, we used an adoptive transfer approach to study the influx of WT and p110γ^−/− OT-I effector T cells into the peritoneum of WT mice following VV-OVA infection. The CD8 effector T cell response in the peritoneum was slightly delayed compared with the spleen and pancreatic lymph node (Fig. 3,A) in that the peak of the response occurred at day 8 postinfection (Fig. 5,A), consistent with the model that CD8 T cells are first primed in the lymph nodes and then migrate into the inflamed peritoneum (30 and data not shown). There was a significant reduction in the number of p110γ^−/− OT-I effector T cells compared with WT OT-I effector T cells that migrated to the peritoneum at day 8. This response is not due to a delayed kinetics of the p110γ^−/− effector T cell response because there was no additional accumulation of p110γ^−/− OT-I T cells in the peritoneum after day 8. In contrast to the endogenous CD8 T cell response (Fig. 4), we did not observe a difference in WT and p110γ^−/− OT-I CD8 T cell numbers in the peritoneal cavity on day 4 postinfection (Fig. 5 A). This may be due to differences in the magnitude of the endogenous CD8 T cell response to VV-WR compared with the OT-I T cell response to VV-OVA. Both WT and p110γ^−/− OT-I T cells in the peritoneum down-regulated CD62L and p110γ^−/− OT-I effector T cells expressed just as much α₄ integrin as WT OT-I effectors (data not shown).

We next directly compared the migratory ability of p110γ^−/− and WT effector cells in vivo. To complete the comparison, we transferred naive WT and p110γ^−/− OT-I cells expressing the CD45.1 congenic marker into C57BL/6 recipients and infected these mice with VV-OVA to generate OT-I effector T cells. On day 6 postinfection, WT and p110γ^−/− OT-I effector T cells were harvested from the spleens of donor mice, stained with CellTracker Green (Fig. 5,B, top left), and injected into secondary C57BL/6 recipients that had been infected with 5 × 10⁶ PFU VV-WR i.p. 48 h before the transfer. Donor CD8⁺CD45.1⁺ cells were identified by flow cytometry analysis (Fig. 5,B, bottom left) and WT and p110γ^−/− OT-I cells were distinguished from one another in the peritoneum and spleen by differential CellTracker green staining (Fig. 5,B, right panels). Compared side by side in the same animal, only about half as many p110γ^−/− OT-I effector T cells were found in the peritoneum compared with WT OT-I effector cells 18 h after transfer (Fig. 5,C). Conversely, there were more p110γ^−/− OT-I effector T cells in the spleen than WT OT-I effectors (Fig. 5 C), suggesting that those effectors that cannot make it to the peritoneum were accumulating in the circulation. These results show that p110γ^−/− CD8 effector T cells have an impaired ability to migrate into an inflammatory site.

We also analyzed the ability of in vitro generated p110γ^−/− CD8 effector T cells (24) to migrate in vitro in response to several chemoattractants. We analyzed the migration of WT and p110γ^−/− CD8 effector T cells to a classic inflammatory chemokine, CCL5 (RANTES), as well as the proinflammatory lipid leukotriene B₄ (LTB₄), which has been shown to selectively attract CD8 effector T cells (31). Previous research has also demonstrated the importance of p110γ activity downstream of the LTB₄ receptor (32). The migration in vitro of p110γ^−/− CD8 effector T cells in response to both LTB₄ and CCL5 was reduced when compared with WT CD8 effector T cells (Fig. 6). We consistently observed a more pronounced impairment of migration of p110γ^−/− CD8 effector T cells in response to LTB₄. Although in vitro generated effector CD8 T cells were not highly responsive to CCL21, we observed no significant difference in the ability of WT and p110γ^−/− CD8 effector T cells to migrate in response to this chemokine (Fig. 6).

Given the cellular defects that exist in the absence of p110γ^−/− (13, 14, 15, 26), we used our VV model to test the ability of p110γ^−/− mice to clear and resolve infection. We infected WT and p110γ^−/− mice with various doses of VV-WR, and monitored mice over time for weight loss and illness. Similar to WT mice, p110γ^−/− mice could control low doses of VV (2 × 10⁶ PFU), which is evident in that they maintained their starting weight and survived the infection (Fig. 7, left panels). However, at a higher viral dose of 5 × 10⁶ PFU, p110γ^−/− mice rapidly lost weight and became visibly ill, resulting in an average time to death of 5 days postinfection (Fig. 7, right panels). In contrast, WT mice did not show the same pattern of illness and death until the viral dose was doubled to 10 × 10⁶ PFU (data not shown).

In this report, we present evidence that the p110γ isoform of PI3K plays a selective role in regulating the migration of effector CD8 T cells. Consistent with previous reports (19, 20), we found that there were no differences between WT and p110γ^−/− naive CD8 T cells in their ability to migrate in vitro in response to the CCR7 ligand CCL21. We also observed no significant difference in the migration of naive WT and p110γ^−/− CD8 T cells to peripheral lymph nodes in vivo, a process that is dependent on CCR7 function (3, 4). These results are in general agreement with previous studies demonstrating that the Rac nucleotide exchange factor DOCK2 is a critical mediator of T cell migration under noninflammatory conditions, whereas p110γ only plays a minor role (20). However, although we did not observe any significant defects in lymph node homing in vivo, the previous study did demonstrate a 15–30% reduction in the migration of p110γ^−/− T cells to lymph nodes in vivo (20). The reason for this discrepancy is not known, but may relate to differences in the T cell populations analyzed (bulk T cells vs CD8 T cells). Although p110γ^−/− T cells exhibit minimal defects in migration both in vitro and in vivo, it is notable that general PI3K inhibitors, such as wortmannin, effectively block in vitro and in vivo migration of naive T cells (19 and data not shown). This finding suggests that multiple isoforms of PI3K likely contribute to naive T cell migration. Likely candidates are p110δ and p110γ, as mice deficient in both of these PI3K isoforms exhibit dramatic defects in T cell development (33, 34).

Previous studies of p110γ^−/− mice also demonstrated impaired proliferation of p110γ^−/− T cells in response to anti-CD3 stimulation in vitro (15). We extended these previous findings by specifically assessing the activation of OT-I TCR transgenic T cells in response to challenge in vivo with VV-OVA. Our results indicated that similar to WT OT-I T cells, p110γ^−/− OT-I T cells expanded vigorously in response to VV-OVA challenge in vivo in both the spleen and pancreatic lymph node. Furthermore, the extent of OT-I clonal expansion was comparable at all time points examined. Activated p110γ^−/− OT-I CD8 T cells also exhibited a cell surface phenotype similar to that of control OT-I CD8 T cells, characterized by low levels of CD62L, and elevated levels of CD44 and α₄ integrin. Furthermore, Ag-activated p110γ^−/− OT-I T cells expressed both IFN-γ and granzyme B at levels similar to Ag-activated WT OT-I T cells. We also used a previously published technique using anti-CD3 Ab stimulation and long-term culture in IL-2 to generate polyclonal CD8 effector T cells (24). Using this system, expansion and phenotypic differentiation of p110γ^−/− T cells was similar to that observed with WT T cells. Thus, unlike the p110δ isoform of PI3K (35), p110γ does not appear to be critical for naive CD8 T cell activation and differentiation into effector T cells either in vivo or in vitro.

Unlike naive T cells, effector T cells can migrate effectively into inflammatory sites in peripheral tissue. Because other p110γ^−/− leukocytes exhibit defects in migration in vitro to chemokines expressed under inflammatory conditions (13, 14, 15, 26) and also show reduced ability to migrate into inflammatory sites in vivo (15), we hypothesized that CD8 effector T cells might also exhibit a similar impairment in migration in response to inflammation. When we specifically studied the endogenous CD8 p110γ^−/− T cell response to VV-WR, we observed that the absolute number of CD8 p110γ^−/− effector T cells that were present in the peritoneum at days 1, 3, and 4 postinfection were consistently lower than the comparable WT CD8 effector T cell population. A similar impairment in the migration of p110γ^−/− CD8 effector T cells was observed when we adoptively transferred naive WT and p110γ^−/− OT-I T cells into WT recipients, infected them with VV-OVA, and then assessed the influx of OT-I effector T cells in the peritoneum at the peak of the response. In addition, we transferred WT and p110γ^−/− OT-I effector T cells into secondary WT recipients that had been infected with VV-WR before T cell transfer. In these recipients, we also observed a reduced ability of p110γ^−/− CD8 effector T cells to migrate into the peritoneum. Together, these results suggest that p110γ^−/− CD8 effector T cells have an intrinsic impairment in their ability to migrate into the inflamed peritoneum. Additional evidence in support of this hypothesis comes from our in vitro studies indicating in vitro generated p110γ^−/− CD8 effector T cells have reduced migration to two different chemoattractants, CCL5 and LTB₄.

A selective role for p110γ in the migration of CD8 effector T cells, but not naive T cells, is not due simply to increased p110γ expression as a result of naive T cell differentiation into effector T cells. There were no differences in expression of GFP, which is under the control of the endogenous p110γ promoter in the p110γ^−/− mice used in this study (13), between naive p110γ^−/− CD8 T cells and in vitro generated p110γ^−/− CD8 effector T cells (data not shown). It is possible that some chemokine receptors, but not others, are coupled to p110γ-dependent pathways that regulate cell migration. This idea that proteins can be selectively involved with certain receptors of a certain class, but not others, is supported by evidence that the class IA PI3K isoforms α and β exhibit differential patterns of coupling to certain tyrosine kinase receptors (36). However, it is also possible that there are cell type-specific differences in the use of p110γ downstream of chemokine receptors. It is interesting to note that although CCR7-mediated migration of naive T cells is not dependent on p110γ, other studies have shown that CCR7-mediated migration of dendritic cells is impaired in the absence of p110γ (26). Thus, p110γ can be used by distinct cell types to mediate migration induced by the same chemokine receptor. Although the migration of naive T cells in response to chemokines that bind CCR7 is not dependent on p110γ, other responses of naive T cells downstream of CCR7, such as chemokine-mediated phosphorylation of protein kinase B, are clearly dependent on p110γ (20). Thus, p110γ is functional downstream of CCR7 in naive T cells, even though it does not regulate naive T cell migration in response to CCR7.

These studies also support the increasing interest in using p110γ inhibition as a clinical strategy for treating disease that results from an unwanted, persistent inflammatory response (17, 18). Through elucidating a specific role for p110γ in the CD8 effector T cell response, our data argues that this therapeutic approach could be used for a more broad range of conditions, specifically including those that have been shown to result from CD8 T cell-induced damage, such as graft rejection, multiple sclerosis, and type I diabetes (37, 38, 39). However, the increased susceptibility of p110γ^−/− mice to VV challenge suggests that such inhibitors may also have more global immunosuppressive effects. Future work will ultimately be needed to determine whether a safe and effective therapeutic course of p110γ inhibition can be used to successfully ameliorate disease symptoms without inducing a detrimental degree of global immunosuppression.

We thank S. Highfill for technical assistance, M. Prlic for assistance with vaccinia virus experiments, and M. Thomas for helpful discussion and advice.

The authors have no financial conflict of interest.