Critical roles of mTOR Complex 1 and 2 for T follicular helper cell differentiation and germinal center responses

T follicular helper (Tfh) cells play critical roles for germinal center responses and effective humoral immunity. We report here that mTOR in CD4 T cells is essential for Tfh differentiation. In Mtorf/f-Cd4Cre mice, both constitutive and inducible Tfh differentiation is severely impaired, leading to defective germinal center B cell formation and antibody production. Moreover, both mTORC1 and mTORC2 contribute to Tfh and GC B cell development but may do so via distinct mechanisms. mTORC1 mainly promotes CD4 T cell proliferation to reach the cell divisions necessary for Tfh differentiation, while Rictor/mTORC2 regulates Tfh differentiation by promoting Akt activation and TCF1 expression without grossly influencing T cell proliferation. Together, our results reveal crucial but distinct roles for mTORC1 and mTORC2 in CD4 T cells during Tfh differentiation and germinal center responses. DOI: http://dx.doi.org/10.7554/eLife.17936.001


Introduction
T follicular helper (Tfh) cells belong to a special subset of CD4 T cells that are essential for germinal center (GC) formation, Ig-class switch and hypermutation, memory B cell and long-lived plasma cells generation, and establishment of long-term protective immunity (King, 2009;Tangye et al., 2013;Craft, 2012;Crotty, 2014;Vinuesa et al., 2016). Differentiation of Tfh cells is considered a multistage and tightly regulated process (Lu et al., 2011). After engagement with antigenic peptide-MHC complexes, which dendritic cells present through their T cell receptors (TCRs) and subsequent activation, a portion of CD4 T cells upregulate ICOS, Bcl-6, and CXCR5 expression, migrate to the T cell-B cell border in the spleen and lymph nodes (LNs) (Kerfoot et al., 2011;Suan et al., 2015). Engagement of T cells with matured B cells via CD40-CD40L and ICOS-ICOSL interactions promotes GC-responses (Liu et al., 2015;Pratama et al., 2015;Awe et al., 2015). Subsequently, the differentiating Tfh cells migrate into B cell follicles and further differentiate into GC-Tfh cells to direct generation of GC B cells. GC-Tfh cells produce cytokines such as IL-21 and IL-4 to regulate Ig-class switch and hypermutation in B cells (Schmitt et al., 2009).
The role of mTOR in Tfh differentiation is largely unclear. A recent report suggests that mTORC1 inhibits Tfh differentiation based on data from rapamycin inhibition and shRNA knockdown (Ray et al., 2015). However, another study has found that a hypomorphic mutation of mTOR causes reduced Tfh responses (Ramiscal et al., 2015). While these studies support a role of mTOR in Tfh responses, they do not distinguish between the roles of mTORC1 and mTORC2 or rule out potential effects from other immune cell lineages. Using mice selectively deficient of mTOR, Raptor/mTORC1, and Rictor/mTORC2 in T cells, we report here that mTOR deficiency caused severe decreases in constitutive Tfh and GC-B cells in the mesenteric lymph nodes (mLNs) and Peyer's patches (PPs), correlated with drastic decreases in virtually all serum IgG subtypes in unchallenged naïve mice. Moreover, mTOR-deficient CD4 T cells fail to differentiate to Tfh cells following antigen immunization, resulting in impaired GC-B cell formation and antigen-specific IgG responses. Both mTORC1 and mTORC2 contribute significantly to Tfh differentiation but appear to do so via distinct mechanisms. mTORC1-deficient CD4 T cells are defective in proliferation and cannot reach divisions that Tfh differentiation may require. In contrast, mTORC2-deficient CD4 T cells do not display obvious impairment in proliferation; instead, they display impaired Akt activation because of decreased phosphorylation at S473, subsequently increased cell death, impaired GSK3b phosphorylation at S9 and inactivation, and decreased b-catenin and TCF1 expression. Importantly, overexpression of a phosphomimetic mutant of Akt, Akt S473D, or TCF1 can partially restore Tfh differentiation, suggesting an mTORC2/Akt/TCF1 axis for Tfh differentiation.

Deficiency of mTOR in T cells impaired constitutive Tfh differentiation and GC B cell responses
Because oral antigens and gut-microbiota-derived antigens stimulate T and B cells, the PPs contain constitutive GCs. About one-third of PP CD4 T cells from Mtor f/f (WT) mice were CXCR5 + PD-1 + Tfh Mtor f/f Figure 1. Critical role of mTOR for constitutive Tfh and GC responses. We collected sera, mLNs, and PPs from 2-3-month-old Mtor f/f and Mtor f/f -Cd4Cre for analysis. (a) Representative dot-plots of CXCR5 and PD1 staining in gated CD4 + TCRb + T-cells from mLNs and PPs. (b) Scatter plots represent mean ± SEM of Tfh percentages (left panel) and numbers (right panel). (c) Representative dot-plots show GL7 and Fas staining in gated CD93 -B220 + IgM -IgD -B cells from mLNs and PPs. (d) Scatter plots represent mean ± SEM of GC-B cell percentages (left panel) and numbers (right panel). (e) Relative serum IgM, IgG, and IgG subtypes (n ! 5) and fecal IgA (n = 19) levels measured by ELISA. Data represent or are calculated from at least five experiments (a-d) or two experiments (e). *p<0.05; **p<0.01; ***p<0.001 determined by unpaired two-tailed Student t-test. DOI: 10.7554/eLife.17936.002 cells (Figure 1a and b), and about 10% to 25% B220 + CD93 -IgM -IgD -B cells were GL7 + Fas + GC B-cells (Figure 1c and d). Tfh and GC-B cells could also be detected in mLNs but at lower frequencies than PPs in WT mice. Both percentages (Figure 1a and b) and absolute numbers (right panel, Figure 1b) of Tfh cells substantially decreased in mLNs and PPs from Mtor f/f -Cd4Cre mice compared with their littermate controls. Concordantly, GC B-cell percentages (Figure 1c and d) and absolute numbers (right panel, Figure 1d) in Mtor f/f -Cd4Cre mLNs and PPs also dramatically decreased, correlated with reduced serum IgG but increased serum IgM levels ( Figure 1e). Interestingly, IgG1, and IgG2b but not IgG3 levels decreased in mTOR deficiency mice, suggesting that the IgG3 class-switch occurred independently of mTOR signaling in CD4 T cells. Additionally, IgA secreted in the intestinal lumen decreased (Figure 1e), which was consistent with impaired GC-responses in PPs. Thus, mTOR deficiency in T cells severely compromised constitutive Tfh and GC responses in PPs and mLNs as well as overall humoral immunity.

Contribution of mTORC1 and mTORC2 to constitutive Tfh and GC B cell responses
To further investigate the contribution of mTORC1 and mTORC2 to constitutive Tfh and GC B cell responses, we examined Rptor f/f -Cd4Cre or Rictor f/f -Cd4Cre mice and their littermate controls in a manner similar to that described in the previous section. Both Rptor f/f -Cd4Cre (Figure 2a,b) and Rictor f/f -Cd4Cre mice (Figure 2c,d) contained fewer Tfh cells in mLNs and PPs compared to their respective controls. To rule out the possibility that defective Tfh differentiation of Rptor f/f -Cd4Cre T cells resulted from abnormal T cell development after Rptor deletion in developing thymocytes, we adoptively transferred a mixture of CD45.1 WT and CD45.2 Rptor f/f -Rosa26 CreER CD4 T cells into Rag2 deficient mice. Recipients were injected with tamoxifen on 7, 8, and 11 days after reconstitution, then were examined on day 14. CXCR5 + PD1 + Tfh cell percentages within CD45.1 + WT and CD45.2 + Rptor f/f -Rosa26 CreER CD4 T cells were similar in recipients without tamoxifen injection. However, in tamoxifen-treated recipients, CXCR5 + PD1 + Tfh cell percentages in Rptor f/f -Rosa26 CreER CD4 T cells were obviously decreased compared with WT controls in the same recipients or with Rptor f/f -Rosa26 CreER CD4 T cells in mice without tamoxifen injection ( and Rictor f/f -Cd4Cre mice (Figure 2j) also decreased compared with controls. Interestingly, mTORC1 deficiency caused reduced IgG1 and IgG2b levels without obviously affecting IgG3 (Figure 2i), while mTORC2 deficiency resulted in decreases in IgG1, IgG2b, and IgG3 levels in serum ( Figure 2j). However, unlike mTOR-deficient mice, neither Rptor f/f -Cd4Cre nor Rictor f/f -Cd4Cre mice had reduced fecal IgA levels (Figure 2i,j). Together, these observations suggested that both mTORC1 and mTORC2 contributed to constitutive Tfh and GC responses in mLNs and PPs and may synergistically or redundantly promote intestinal IgA responses.

Effects of mTORC1 and mTORC2 deficiency on regulatory T cells
Regulatory T cells (Tregs) actively suppress immune responses and are essential for maintaining selftolerance. It has been reported that mTOR deficiency causes relative enrichment of Tregs over conventional CD4 T cells (Tcon) (Delgoffe et al., 2009) and that mTORC1 but not mTORC2 is critical for Treg suppressive function (Zeng et al., 2013). In Rptor f/f -Cd4Cre mice, CD4 + Foxp3 + Treg percentages as well as Treg to Tcon ratios were similar to WT controls in the spleen, mLN, and PPs (Figure 3a-c). Treg numbers in Rptor f/f -Cd4Cre mice were not obviously altered in the spleen and mLNs, but they were decreased in the PPs compared with WT controls (Figure 3d). The decrease of total PP cell numbers (14.17 ± 3.22 million in Rptor f/f vs 6.13 ± 0.72 million in Rptor f/f -Cd4Cre, p<0.05) was a major contributing factor for decreased PP Treg numbers in Rptor f/f -Cd4Cre mice. In Rictor f/f -Cd4Cre mice, Treg percentages and numbers were decreased in the spleen and PPs, but they were not significantly decreased in mLNs (Figure 3g,h and j), correlated with decreased total T cell numbers in the spleen and PPs in Rictor f/f -Cd4Cre mice (data not shown). However, Treg to Tcon ratios were not significantly skewed ( Figure 3i). It has been recently reported that T follicular regulatory cells (Tfr) play important roles in suppressing GC responses (Sage and Sharpe, 2015). The percentages of the Foxp3 + population within PP Tfh cells were decreased and Foxp3 + Treg numbers were more severely decreased in Rptor f/f -Cd4Cre mice, suggesting that mTORC1 deficiency had a stronger effect on Tfr development/homeostasis than on conventional Tregs (Figure 3e,f). In Rictor f/f -Cd4Cre mice, PP Tfr percentages within Tfh cells were not decreased, but their numbers were decreased due to reduced Tfh numbers (Figure 3k,l). Together, our data suggest that deficiency of either mTORC1 or mTORC2 did not increase the relative abundance of Tregs within CD4 T cells in the spleen, mLNs, and PPs or the Tfr ratios within Tregs in the PPs. Together with decreased total Treg and Tfr numbers in Rptor or Rictor deficient PPs, these data suggest that the decreased constitutive GC-B cell response was likely not caused by change of Treg or Tfr numbers in these mice. However, whether mTORC1 and mTORC2 may play a role in Tfr function remains to be determined.

Impaired inducible Tfh and GC B cell development in the absence of mTOR in T cells
To examine the role of mTOR in Tfh and GC responses during immune responses, we immunized Mtor f/f -Cd4Cre mice and their littermate controls with a T-cell-dependent antigen, NP17-CGG, in alum. On day 21 after immunization, fewer splenic Tfh cells were present in Mtor f/f -Cd4Cre mice than in WT controls in both percentages and numbers (Figure 4a,b) with concomitantly reduced GC B-cells (Figure 4c,d). Induction of Bcl-6, a transcription factor crucial for GC B cell differentiation (Fukuda et al., 1997), was blunted in splenic and mLN B cells from Mtor f/f -Cd4Cre mice after immunization ( Figure 4e). Moreover, both total (NIP26) and high-affinity (NIP7) serum NIP-specific IgM and IgG levels were much lower in Mtor f/f -Cd4Cre mice than in WT mice 7, 14, and 21 days after immunization ( Figure 4f). Thus, mTOR in CD4 T cells was essential for inducing Tfh differentiation and GC B cell formation following immunization and for T-cell-dependent antigen-induced humoral immune responses.

Critical roles for mTORC1 and mTORC2 in Tfh differentiation following antigen immunization
To determine the role of mTORC1 and mTORC2 in antigen-induced Tfh/GC-B cell responses, we immunized Rptor f/f -Cd4Cre, Rictor f/f -Cd4Cre mice, and their respective control mice with NP17-CGG in alum. Both Rptor f/f -Cd4Cre (Figure 5a Rptor-deficient mice, neither total nor high-affinity NIP-specific IgM levels decreased in Rictor f/f -Cd4Cre mice (Figure 5j). While Rictor f/f -Cd4Cre mice manifested a trend of mildly decreased NIP specific total and high-affinity IgG levels, only decreases of total IgG on day 7 and high-affinity IgG on day 14 were statistically significant ( Figure 5j). Because NIP-specific antibody responses in Rictordeficient mice were not as severe as in Rptor-deficient mice, we further examined GC formation in Rictor f/f -Cd4Cre mice on day 14 after immunizing them via immunofluorescence microcopy using PNA and Thy1.2 to detect GC B cells and T cells, respectively. As shown in Figure 5k,l, GC numbers were obviously less and sizes of individual GCs were noticeably smaller in Rictor f/f -Cd4Cre spleens than in controls, confirming impaired GC responses in the absence of mTORC2 in T cells. Together, both mTORC1 and mTORC2 contributed to Tfh/GC B cell responses, in which mTORC1 appeared to play a more important role than mTORC2.
One potential mechanism that contributes to decreased Tfh responses of Rptor or Rictor deficient CD4 + T cells was increased death. Rptor deficient OT2 T cells, including both Tfh and non-Tfh populations, had similar percentages of dead cells compared with WT controls (Figure 6e). In contrast, Rictor deficient OT2 T cells showed increased death rates (Figure 6j). Thus, Rictor but not Raptor deficiency led to increased death of CD4 + T cells during immune responses.

mTORC1 and mTORC2 promote Tfh differentiation via distinct mechanisms
To understand whether the impaired Tfh differentiation of Rptor-or Rictor-deficient CD4 T cells was associated with impaired proliferation, we adoptively transferred CFSE-labeled WT and Rptor f/f -Cd4Cre-OT2 cells into WT recipients and followed with OVA 323-339 peptide immunization. WT OT2 T cells demonstrated vigorous proliferation following immunization. In the absence of Raptor, OT2 T cells displayed an obvious defect in proliferation 3, 5, and 7 days after immunization (Figure 7a). Similarly, proliferation of Raptor-deficient OT2 T cells was defective following in vitro TCR stimulation (Figure 7-figure supplement 1a) without severe defects in upregulation of T cell activation markers CD69 and CD44 with the exception of decreased CD25 upregulation (Figure 7-figure supplement 1b), which was consistent with the role of mTORC1 for cell cycle entry after T cell activation (Yang et al., 2013). CXCR5 + or PD1 + Tfh cells, differentiated from WT naïve CD4 T cells, began to appear after at least six divisions (Figure 7b-e). While delayed in proliferative expansion, some Raptor-deficient OT2 cells eventually reached more than six divisions 7 days after immunization. Relative serum NIP-specific IgG and IgM levels from Rptor f/f -Cd4Cre (i; n = 6) and Rictor f/f -Cd4Cre (j; n = 9) indicated days after immunization detected by ELISA with NIP7-or NIP26-BSA coated plates. (k, l) Impaired GC formation in Rictor-deficient mice. 14 days after immunization, we stained spleen thin sections from Rictor f/f -Cd4Cre and Rictor f/f mice with PNA, B220, and Thy1.2. Representative immunofluorescence images are shown (k; 50x and 200 x). Scatter plots depict mean ± SEM of GC numbers per view (l. left; 8 total views for WT and 10 views for KO) and sizes of the PNC + Thy1.2 -GCs (l. right). Data shown represent or are calculated from three experiments. *p<0.05; **p<0.01; ***p<0.001 determined by two-tailed unpaired Student t-test.    The following figure supplement is available for figure 6:   Rptor f/f Cd4Cre-OT2 Rptor f/f -OT2 Rptor f/f -OT2 Rptor f/f Cd4Cre-OT2 Rictor f/f Cd4Cre-OT2 Rictor f/f Cd4Cre-OT2 Rictor f/f Cd4Cre-OT2 Rictor f/f Cd4Cre-OT2 Day 7 Rictor f/f Cd4Cre-OT2 Rictor f/f -OT2 Rictor f/f -OT2

Rictor f/f -OT2
Rictor f/f -OT2 Rictor f/f -OT2 Figure 7. Effects of mTORC1 and mTORC2 deficiency on proliferation-associated Tfh differentiation. We injected CD45.1 + CD45.2 + congenic mice iv with 1.5 Â 10 6 CD45.2 + Va2 + CD4 + WT, Rptor f/f -Cd4Cre (a-e), or Rictor f/f -Cd4Cre (f-j) naïve OT2 T cells on day -1 and immunized them with OVA 323-339 peptide in CFA on day 0, harvesting dLNs on indicated days after immunization. (a) Overlaid histograms showing CFSE intensity in CD45.1 -CD45.2 + CD4 + TCRVa2 + donor-derived WT and Rptor-deficient OT2 T cells 3, 5, and 7 days after immunization. (b, d) Representative dot plots Figure 7 continued on next page However, the PD1 + /PD1and CXCR5 + /CXCR5ratios in individual divisions were much lower for Raptor-deficient OT2 T cells than for WT controls. Thus, Raptor/mTORC1 not only promoted proliferation of CD4 T cells to reach the needed divisions after activation to become Tfh cells but also played an important role for expanded T cells to differentiate into Tfh cells. Unlike Raptor/mTORC1 deficiency, Rictor-deficient OT2 T cells proliferated and upregulated CD69 and CD25 at similar rates/levels compared with controls following in vitro TCR stimulation (Figure 7-figure supplement 1c,d). CD44 levels could also be upregulated, though they remained lower than stimulated WT controls. Furthermore, after immunization, proliferation of adoptively transferred Rictor-deficient OT2 T cells was not obviously impaired compared to WT controls in recipient mice (Figure 7f). However, their CXCR5 + /CXCR5ratios in individual divisions after six divisions were decreased compared with WT controls (Figure 7g,h). The PD1 + /PD1ratios in Rictor-deficient OT2 T cells were not different from WT control in division 6 but displayed a tendency of decline after division 7, although the decreases were only statistically significant after division 10 (Figure 7i,j). Together, these observations suggested that Tfh differentiation, rather than T cell activation and proliferation, selectively required Rictor/mTORC2.

mTORC2 promotes Tfh differentiation via activating Akt and upregulating TCF1
Because mTORC1-deficient T cells were defective in proliferation following T cell activation, we focused on investigating mechanisms by which mTORC2 promoted Tfh differentiation. Recent studies have revealed that the PI3K/Akt pathway plays an important role in Tfh differentiation (Gigoux et al., 2009;Stone et al., 2015;Rolf et al., 2010). Following TCR engagement, we observed apparently decreased Akt phosphorylation at Ser473, an mTORC2-dependent event, but not S6 phosphorylation, an mTORC1-dependent event, in Rictor/mTORC2-deficient T cells (Figure 8a, b). Concordant with decreased Akt phosphorylation, GSK-3b phosphorylation at Ser9 residue, an Akt mediated event, also decreased in these cells (Figure 8a). Thus, mTORC2 deficiency resulted in impaired Akt activation in CD4 T cells following TCR stimulation.
To examine whether decreased Akt S473 phosphorylation contributed to defective Tfh differentiation, we adoptively transferred WT or Rictor-deficient OT2 T cells (CD45.2 + ) transduced with retroviruses coexpressing either the Akt S473D phosphomimetic mutant and GFP or GFP alone (Migr1) into WT C57BL6/J recipient mice, followed by immunization with OVA 323-339 peptide in CFA. Seven days after immunization, we observed a two-fold increase of PD1 + CXCR5 + Tfh cells within GFP + -TCRVa2 + Vb5 + CD4 + Rictor-deficient OT2 T cells expressing AktS473D compared with those expressing GFP alone (Figures 8c,d). Such an effect of AktS473D on Tfh differentiation was correlated with its ability to upregulate Bcl-6 expression ( Figure 8e) and to improve cell survival (Figure 8-figure  supplement 1). Of note, Rictor-deficient OT2 T cells expressing AktS473D were still about 50% fewer Tfh cells than WT OT2 T cells (Figure 8c,d). Together, these observations suggested that mTORC2 promoted Tfh differentiation via both Akt-dependent and -independent mechanisms.

Discussion
Understanding regulation of Tfh cell differentiation is crucial for developing new strategies to elicit protective immunity effectively and to improve treatment of autoimmune diseases. Although mTOR has proven important for Th1, Th2, and Th17 differentiation, its role in Tfh differentiation has been controversial. A recent study showed that inhibition of mTORC1 with rapamycin or reduction of mTOR expression with shRNA promotes Tfh differentiation following lymphocytic choriomeningitis virus infection, concluding that mTORC1 negatively controls Tfh differentiation (Ray et al., 2015). However, another study has found that a hypomorphic mutation of mTOR reduces Tfh and GC responses in mice immunized with sheep red blood cells, concluding that mTOR promoted Tfh differentiation (Ramiscal et al., 2015). Additionally, rapamycin has also been found to inhibit GC formation and Ig class-switch in B cells upon influenza infection, although whether rapamycin directly acts on T cells, B cells, or both is unclear (Keating et al., 2013). While these two studies suggest a positive role for mTOR in Tfh differentiation, they neither firmly establish that mTOR intrinsically controls CD4 T cell differentiation in Tfh cells nor determine the roles of mTORC1 and mTORC2 in Tfh differentiation.
In this report, we demonstrate that T cell-specific ablation of mTOR severely decreases constitutive Tfh and GC B cell formation in mLNs and PPs and reduced serum IgG and fecal IgA levels. Moreover, mice with T cell-specific mTOR deficiency fail to mount effective Tfh and GC responses following immunization with a T cell-dependent antigen. Our study provides genetic evidence that mTOR plays an intrinsically crucial role in CD4 T cells for their differentiation to Tfh cells and for promotion of GC responses. Our data also reveal that both mTORC1 and mTORC2 are important for Tfh differentiation. We have shown that ablation of either Raptor/mTORC1 or Rictor/mTORC2 in T cells reduces Tfh differentiation and GC B cell formation in mLNs and PPs under the steady state as well as in the spleen following immunization with NP-CGG. Although both mTORC1 and mTORC2 contribute to Tfh differentiation, mTORC1 deficiency appears to exert greater impact on Tfh differentiation than mTORC2 deficiency in antigen-induced responses, as NP-specific antibody responses are more severely affected in mTORC1-deficient mice than in mTORC2-deficient mice.
Using adoptive transfer of OT2 CD4 T cells, we have shown that, following antigenic stimulation, differentiation of naïve OT2 CD4 T cells to Tfh cells occurs after six divisions. Most mTORC1 deficient CD4 T cells fail to proliferate, suggesting that an important function of mTORC1 is to promote CD4 T cell proliferation, which is consistent with its role in causing T cells to enter into the cell cycle (Yang et al., 2013). In addition, beyond its role in cell proliferation, mTORC1 likely promotes Tfh differentiation, because some mTORC1-deficient CD4 T cells can reach the required cell divisions but remain defective for turning into Tfh cells. mTORC1 has been shown to be able to upregulate ICOS expression in CD4 T cells to prevent anergy (Xie et al., 2012). Given the importance of ICOS expression in Tfh differentiation (Kang et al., 2013), mTORC1 could operate as a positive feedback mechanism to promote Tfh differentiation by upregulating ICOS expression.
Our data provide the first evidence that mTORC2 also plays important roles for Tfh differentiation and GC-formation. Unlike mTORC1 deficiency, mTORC2 deficiency does not obviously affect T cell proliferation in vitro and in vivo, suggesting that impaired proliferation might not cause defective Tfh differentiation of mTORC2 CD4 T cells. Different from mTORC1, mTORC2 promotes T cell survival during immune responses. Our data suggest that mTORC2 may promote Tfh differentiation at least by increasing Akt activity. mTORC2-deficient CD4 T cells contain decreased Akt phosphorylation at S473, accompanying reduced enzymatic activity reflected by decreased GSK3b phosphorylation at Ser9 and increased CD4 T cell death. Importantly, reconstitution of Akt S473 phosphomimetic mutant AktS473D into mTORC2-deficient CD4 T cells can improve their survival, restore Bcl-6 expression in these cells, and partially reverse their defect in Tfh differentiation. Akt itself phosphorylates numerous substrates to control their activities and subcellular localization. Among them, Foxo1 is known to inhibit Tfh differentiation by directly suppressing Bcl-6 transcription (Stone et al., 2015). Because Akt phosphorylates Foxo1, leading to its sequestration in the cytosol, decreased Akt activity could relieve Bcl-6 from Foxo1-mediated suppression. Thus, mTORC2 may increase Akt activity to relieve Foxo1-mediated repression of Bcl-6 expression during Tfh differentiation. In addition to Akt, mTORC2 phosphorylates several other substrates such as PKCa, PKCq, and SGK1. Whether these substrates contribute to mTORC2-mediated Tfh differentiation remains a question for future investigation.
Our data suggested another potential Akt-mediated mechanism is the regulation of the Wnt/bcatenin/TCF1 axis. Recent studies have demonstrated that TCF1 plays critical roles in Tfh differentiation by directly controlling expression of Tfh-promoting genes such as Bcl-6 and ASCL2 Xu et al., 2015;Choi et al., 2015). GSK3b negatively controls Wnt/b-catenin signaling by directly phosphorylating b-catenin at multiple sites, leading to the degradation of b-catenin (Staal et al., 2008). Akt-mediated phosphorylation negatively controls GSK3b activity at Ser9 (32). Decreased b-catenin levels and TCF1 expression suggest that the Wnt/b-catenin/TCF1 axis is impaired in mTORC2-deficient CD4 T cells, which could be because decreased GSK3b phosphorylation in mTORC2-deficient CD4 T cells causes increased GSK3b activity. Reconstituting these cells with full-length TCF1 partially restores their ability to differentiate into Tfh cells, suggesting that mTORC2 promotes TCF1 expression to enhance Tfh differentiation. It was noteworthy that the TCF1 p33 isoform failed to rectify Rictor deficiency-caused Tfh defects, highlighting a requirement for TCF1-b-catenin interaction.
Our data are consistent with previous studies that have demonstrated important roles in Tfh differentiation and GC-responses for PI3K signaling, which is involved in the induction of key molecules such as Bcl-6, IL-4 and IL-21 Gigoux et al., 2009;Rolf et al., 2010;Bauquet et al., 2009;Choi et al., 2011. Upregulation of PI3K signaling by miR-17-92 promotes Tfh cell differentiation (Kang et al., 2013;, while PTEN's negative control of PI3K signal or Foxp1's downregulation of ICOS inhibits Tfh differentiation Rolf et al., 2010). Because both mTORC1 and mTORC2 are activated downstream of PI3K, our data suggest that mTORC1/mTORC2 are critical downstream effectors of the PI3K/Akt pathway for Tfh differentiation. Further studies should illustrate how different receptors and intracellular signals dynamically activate and regulate mTOR during Tfh differentiation and how altered mTOR signaling may either contribute to pathogenesis of autoimmune diseases caused by uncontrolled Tfh and GC responses or improve efficacy during vaccination.

Flow cytometry
We used standard protocols to prepare single-cell suspensions from the spleen, mLNs, and PPs of mice. After lysis of red blood cells with the ACK buffer, we resuspended cells in IMDM-10 and then stained them with antibodies in PBS containing 2% FBS. We performed intracellular staining for Foxp3, Bcl-6, b-catenin, and TCF1 using the eBioscience Foxp3 Staining Buffer Set and intracellular staining for AKT S473 and S6 using the BD Biosciences Cytofix/Cytoperm and Perm/Wash solutions. We collected all flow cytometry data using a FACS Canto-II (BD Biosciences) and analyzed them using FlowJo.

Total and subtype Ig quantification
We collected and weighed fresh fecal pellets from mice and resuspended them in PBS at 100 mg/ ml. After vortexing the pellets for 5 min and centrifuging them at 3000 g for 10 min, we transferred the supernatant to new tubes containing a protease inhibitor cocktail and stored them at À80˚C until use. We used ELISA to measure the total IgA levels in fecal preparations and serum IgM, total IgG, IgG1, IgG2b, IgG2c, and IgG3 levels. In brief, we added100ml of appropriately diluted fecal or serum samples to 96-well plates precoated with anti-mouse Igk and Igl antibodies (2 mg/ml; SouthernBiotech, Birmingham, AL) in 0.1 M carbonate buffer (pH 9.0) at 4˚C overnight. We determined total and subtype Ig concentrations using HRP-conjugated goat anti-mouse total or Ig subtype antibodies (SouthernBiotech). We computed Ig relative levels by the OD450 values.

Immunization, antibody responses, and GC detection
We immunized the mice with a single i.p. injection of 20 mg of 4-hydroxy-3-nitrophenylacetyl conjugated chicken gamma globulin (NP 17 -CGG) in alum as previously described (Ci et al., 2015). We collected serum on day 7, 14, and 21 postimmunization. To measure NIP-specific IgM and IgG levels, we appropriately diluted the serum and added it into ELISA plates (Corning, New York, NY) precoated with 50 ml 2 mg/ml NIP 7 -BSA or NIP 26 -BSA in 0.1 M carbonate buffer (pH 9.0) at 4˚C overnight. After incubation and multiple washes, we used HRP-conjugated goat anti-mouse IgM and IgG to detect NIP-specific IgM and IgG, respectively.
To visualize GC, we fixed spleens from mice 14 days after immunization with 4% PFA for 24 hr and incubated them in 30% sucrose solution for another 24 hr. After freezing the spleens in OCT embedding medium, we cut 10 mm sections and blocked them with PBS containing 3% BSA for 30 min before incubating them with biotinylated Peanut Agglutinin (PNA, Vector Laboratories, Burlingame, CA) at room temperature for 1 hr. We washed slides with PBS and stained them with Violet 421-anti-B220 (Biolegend, clone RA3-6B2, 2 mg/ml), FITC-anti-Thy1.2 (Biolegend, clone 30-H12, 5 mg/ml), and Alex 594-conjugated streptavidin (2 mg/ml) overnight at 4˚C. Finally, we covered the stained sections with slow-fade diamond antifade mountant (Life Technologies, Carlsbad, CA). We collected fluorescence images using a Zeiss Axio imager wide-field fluorescence microscope with 5X and 20X objectives and used Photoshop (Adobe Systems, San Jose, CA) for postacquisition brightness and contrast processing. We used MetaMorph image analysis software to quantify germinal center (PNA + Thy1.2 -) sizes.

Western blot
We rested splenocytes in DPBS for 30 min and then stimulated or unstimulated them with anti-CD3 (500 A2) for different times, after which we immediately lysed them in lysis buffer (1% nondiet P-40, 150 mM NaCl, 50 mM Tris, pH 7.4) with freshly added protease and phosphorylases inhibitors. We subjected cell lysates were subjected to immunoblotting analysis using indicated antibodies.

Retroviral transduction
We used the Phoenix-Eco packaging cell line to make retroviruses for Migr1, AKT S473D, TCF1 p45 and p33 isoforms, and TCF1-Loop34. Phoenix-Eco cells, kindly provided by Dr. Gary Nolan from Stanford University, were authenticated by STR profiling and were tested free of mycoplasma contamination. We stimulated three million splenocytes in 24-well plates in 1 ml IMDM-10 with anti-CD3 (1 mg/mL) and anti-CD28 (500 ng/mL) for 40 hr. After replacing 500 ml cultural medium with retroviral supernatants containing polybrene (5 mg/mL final concentration), we spin-infected cells at 22˚C, 1250 g, for 1.5 hr. After incubating culture supernatants at 37˚C for 6 hr, we replaced them with fresh IMDM-10 and cultured cells for an additional 48 hr before use.

Statistical analysis
We present data as mean ± SEM; we determined statistical significance using the two-tailed Student t test. We define p values as follows: *p<0.05; **p<0.01; ***p<0.001.