STAT5 is a potent negative regulator of T_FH cell differentiation

It remains unclear how diverse signals combine to specify commitment to T_FH or effector (T_H1, T_H2, T_H17, etc.) CD4⁺ T cell differentiation. Given the importance of STATs in regulating effector CD4⁺ T cell gene programs (Zhu et al., 2010) and conflicting reports of STAT5 regulating Bcl6 or Blimp-1 in B cells (Scheeren et al., 2005; Walker et al., 2007; Duy et al., 2011), we examined the role of STAT5 in T_FH cell differentiation. Because the primary limiting factor for STAT activation and signaling is the availability of activating cytokines, we used retroviral expression vectors (RVs) expressing only GFP (“GFP”), GFP and WT STAT5b (STAT5-WT), or GFP and a constitutively active mutant of STAT5b (STAT5-CA; Onishi et al., 1998). SMARTA TCR transgenic CD4⁺ T cells (SM cells), specific for the lymphocytic choriomeningitis virus (LCMV) epitope GP_66–77 bound by MHC class II I-A^b, were transduced with these RVs. STAT5-CA⁺ cells exhibited increased phospho-STAT5 protein in the absence of IL-2 (93% increase relative to GFP⁺ cells; Fig. 1 A; Onishi et al., 1998). Both STAT5-CA⁺ and STAT5-WT⁺ cells had augmented levels of phospho-STAT5 protein after stimulation with IL-2 (150% and 47% increase relative to GFP⁺ cells, respectively; Fig. 1 A). Sorted transduced cells were adoptively transferred into WT C57BL/6J host mice. Shortly thereafter, host mice were infected with Armstrong strain LCMV.

GFP⁺, STAT5-WT⁺, and STAT5-CA⁺ SM cells all expanded normally in response to acute LCMV infection (Fig. 1 B). However, STAT5-CA⁺ SM cells largely failed to differentiate into T_FH cells (78% fewer T_FH cells; P < 0.0001; Fig. 1 C). Overexpression of WT STAT5 also reduced T_FH cell differentiation (33% fewer T_FH cells; P = 0.0002; Fig. 1 C). Mice that received STAT5-CA⁺ SM cells had fewer GC B cells (71% fewer GC B cells; P < 0.0001; Fig. 1 D), consistent with a substantial loss of T_FH cell help.

One possible mechanism by which STAT5 signaling could impair effector CD4⁺ T cell differentiation was induction of T_reg cell differentiation. In some settings, STAT5 signaling drives T_reg cell differentiation via induction of FoxP3 (Wei et al., 2008). However, we found that SM cells transduced with the STAT5-WT RV or the STAT5-CA RV did not express FoxP3 and did not detectably suppress the endogenous immune response (Fig. 1 E and not depicted). Furthermore, STAT5-WT⁺ and STAT5-CA⁺ SM cells produced high levels of the T_H1-associated cytokine IFN-γ (Fig. 1 F), suggesting that non-T_FH effector cell differentiation was not impaired by enhanced STAT5 signaling. Collectively, these data indicated that STAT5 signaling selectively inhibited T_FH cell differentiation during an acute viral infection.

We hypothesized that STAT5 was a physiological inhibitor of T_FH cell differentiation and consequently that a lack of STAT5 signaling during CD4⁺ T cell priming would enhance T_FH cell differentiation. However, insufficient STAT5 signaling in the thymus results in a loss immunological self tolerance (Malek et al., 2002; Burchill et al., 2007). To avoid this complication, we conditionally deleted STAT5 in mature CD4⁺ T cells by transducing splenic STAT5^fl/fl SM CD4⁺ T cells with a Cre-expressing RV (Cre). Phospho-STAT5 protein was absent in STAT5-deficient (STAT5^fl/fl Cre⁺) cells (Fig. 2 A), in agreement with PCR data (not depicted). STAT5-deficient SM cells or untransduced but similarly treated (control) SM cells were adoptively transferred into C57BL/6J host mice that were subsequently infected with LCMV. STAT5-deficient SM cells expanded as well as control SM cells and expressed surface markers consistent with normal activation (Fig. 2, B and C). Although the STAT5 signaling cytokines IL-2 and IL-7 are drivers of T cell survival and proliferation (Rochman et al., 2009), these results show that deletion of STAT5 from mature CD4⁺ T cells is a viable approach to examine the role of STAT5 in effector CD4⁺ T cell differentiation, consistent with recent CD4 T cell studies using STAT5^fl/fl (Tripathi et al., 2010) or IL-2^−/− CD4 T cells (Liao et al., 2011).

Deletion of STAT5 markedly enhanced SM T_FH cell differentiation (80% more T_FH cells; P < 0.0001; Fig. 2 D). T_FH cell differentiation is a multistep, multistage process (Crotty, 2011), and T_FH cells that have progressed into GCs, GC T_FH cells, can be identified by PD-1 or GL7 staining (Haynes et al., 2007; Yusuf et al., 2010; Kitano et al., 2011; Lee et al., 2011; Goenka et al., 2011). GC T_FH cell abundance was also increased in the absence of STAT5 (55% more GC T_FH cells; P = 0.01; Fig. 2 E). STAT5-deficient T_FH cells expressed normal levels of IL-21, a key T_FH-produced cytokine that sustains the GC reaction (Fig. 2 F; Crotty, 2011). Production of IL-2 in both T_FH and non-T_FH effector cells was also maintained in the absence of STAT5 signaling (not depicted). These data showed that T_FH proliferation, survival, and function were not dependent on STAT5 signaling and, more importantly, that STAT5 was a physiological inhibitor of T_FH cell differentiation.

Recently, we and others found that activated CD4⁺ T cells rapidly bifurcate into T_FH or non-T_FH effector cell differentiation programs during priming (Choi et al., 2011; Kitano et al., 2011). Consequently, we examined the effect of STAT5 deficiency on commitment to T_FH cell differentiation. 4 d after LCMV infection, STAT5-deficient (Cre⁺) SM cells had again expanded as well as control SM cells (Fig. 2 G), yet the absence of STAT5 signaling strongly enhanced Bcl6 expression and T_FH cell differentiation (123% increase in Bcl6 mean fluorescence intensity [MFI; P = 0.0012] and 128% more T_FH cells [P < 0.0001]; Fig. 2, H and I). These results indicate that STAT5 acts early in T cell priming during an acute viral infection to block Bcl6 expression and thereby prevent commitment to T_FH cell differentiation.

A key function of STAT5 in T cells is to mediate signaling by IL-2. Intriguingly, activated CD4⁺ T cells that have recently begun T_FH cell differentiation express lower levels of the high affinity subunit of the IL-2 receptor, IL-2Rα (CD25; Choi et al., 2011). We noted that STAT5 deficiency resulted in a reduction in IL-2Rα expression (82% reduction in CD25^high CXCR5^low non-T_FH SM cells; P < 0.001; Fig. 2 J), consistent with previous studies (Nakajima et al., 1997; Malek et al., 2002). Overall, these results suggested that STAT5 inhibited commitment to T_FH cell differentiation during T cell priming and that IL-2 may induce this pathway.

To directly test the role of IL-2 on commitment to T_FH cell differentiation, we first used SM cells that were heterozygous for deletion of IL-2Rα (CD25^+/−). When transferred into C57BL/6J mice that were subsequently infected with LCMV, activation and expansion of CD25^+/− SM cell was equivalent to that of WT SM cells (Fig. 3, A and B). Strikingly, the twofold reduction in IL-2Rα expression resulted in preferential T_FH cell differentiation by CD25^+/− cells as early as 48 h after LCMV infection (110% more CD25^low Bcl6^high cells; P < 0.001; Fig. 3 C). Similar results were obtained 72 h after LCMV infection (P < 0.01; Fig. 3 D), by which time SM cells had bifurcated into T_FH and T_H1 effector cells.

Next, we transferred WT SM cells into C57BL/6J mice treated with anti–IL-2 neutralizing antibodies and then infected with LCMV. IL-2 neutralization did not impair SM cell expansion (Fig. 3 E) but did significantly enhance commitment to T_FH cell differentiation (86% increase; P = 0.0004; Fig. 3 F). Together, these data demonstrated that IL-2 is a key mediator of STAT5 signaling and inhibition of T_FH cell differentiation during T cell priming and that reduced IL-2 signaling is sufficient to bias T cells away from T_H1 effector cell differentiation and toward T_FH cell differentiation.

Because IL-2 and STAT5 regulate myriad genes in lymphocytes, it was important to identify the STAT5 targets responsible for inhibiting T_FH cell differentiation. Because IL-2 can induce Blimp-1 expression in CD8⁺ T cells (Martins and Calame, 2008; Kalia et al., 2010; Pipkin et al., 2010), we hypothesized that STAT5 signaling in CD4⁺ T cells inhibited T_FH cell differentiation by also inducing expression of Blimp-1.

We tested the ability of STAT5 signaling to inhibit T_FH cell differentiation in the absence of Blimp-1 (encoded by the gene Prdm1) by cotransducing Prdm1^fl/fl SM cells with both STAT5-CA RV (expressing GFP) and Cre RV (expressing the fluorescent protein Ametrine). STAT5-CA⁺ Cre⁺, STAT5-CA⁺ Cre⁻, STAT5-CA⁻ Cre⁺, and STAT5-CA⁻ Cre⁻ (control) SM cells were purified and adoptively transferred into C57BL/6J mice that were subsequently infected with LCMV (Fig. 4 A). All populations of SM cells expanded equivalently (Fig. 4 B). Deletion of Blimp-1 alone (Cre⁺) enhanced T_FH cell differentiation (Fig. 4 C), as previously demonstrated (Johnston et al., 2009). Consistent with the experiments shown in Fig. 2, expression of STAT5-CA in Cre⁻ (Blimp-1 intact) Prdm1^fl/fl SM cells inhibited T_FH cell differentiation (Fig. 4 C). However, STAT5-CA and Cre cotransduced Prdm1^fl/fl SM cells, which possessed constitutive STAT5 signaling but lacked Blimp-1, readily differentiated into T_FH cells (176% more T_FH cells than for Blimp-1–intact STAT5-CA⁺ SM cells; P < 0.0001; Fig. 4 C). GC T_FH cell differentiation was also restored in STAT5-CA⁺ SM cells by the absence of Blimp-1 (104% increase; P < 0.0001; Fig. 4 D). These data indicated that STAT5 inhibition of T_FH cell differentiation was mediated by Blimp-1, consistent with the recent finding that STAT5 can directly bind the Prdm1 promoter in CD4⁺ T cells after IL-2 stimulation (Yang et al., 2011). Additional STAT5 targets may also contribute (Liao et al., 2011), as the contribution of Blimp-1 was not complete (Fig. 4, C and D).

The signals that negatively regulate T_FH cell differentiation have not been well characterized. In this study, we found that STAT5 is a key physiological inhibitor of Bcl6 expression and thereby an inhibitor of T_FH cell differentiation. The absence of STAT5 resulted in increased Bcl6 expression and preferential T_FH cell differentiation. This STAT5 function appears to be primarily induced by IL-2, as reduced IL-2 signaling substantially increased T_FH cell differentiation. Because IL-2 and other STAT5 signaling cytokines are important mediators of T cell proliferation and survival, it was somewhat surprising that IL-2–deprived or STAT5-deficient T_FH cells expanded normally. In agreement with our observations, the size of the antiviral CD4⁺ T cell response was unaffected by STAT5 deficiency in another study (Tripathi et al., 2010).

Our data suggest that STAT5 is not necessary for the CD4⁺ T cell effector response per se, but is required to properly balance T_FH and T_H1 effector CD4⁺ T cell differentiation. Importantly, our finding that a twofold reduction in IL-2Rα expression (CD25^+/−) shifts CD4⁺ T cells toward T_FH cell differentiation demonstrates that small changes in IL-2 availability can have a significant impact on T cell fate decisions in vivo. Bcl6 is also involved in the development of T cell memory (Ichii et al., 2004; Crotty et al., 2010; Pipkin et al., 2010; Pepper et al., 2011). Although in this study we focus on how STAT5 negatively regulates T_FH cell differentiation and the development of GCs (Fig. 1 C), it is also intriguing to consider how these processes may impact CD4 T cell memory.

Given the association of dysregulated T_FH activity with autoantibody production, manipulation of STAT5 activity to attenuate T_FH cell differentiation or function may be a useful tool in the treatment of autoimmune disease. Conversely, manipulation of this pathway may also be a valuable tool to augment T_FH activity and thus the potency of candidate vaccines for a variety of infectious diseases.

C57BL/6J mice, as well as Prdm1^fl/fl mice (Shapiro-Shelef et al., 2003) and CD25-deficient mice (Il2ra^−/−, B6.129S4-Il2ra^tm1Dw/J; Willerford et al., 1995) fully backcrossed to C57BL/6J, were purchased from the Jackson Laboratory. CD4-cre mice were purchased from Taconic. SM TCR transgenic CD45.1⁺ mice with a C57BL/6J background were bred at the La Jolla Institute for Allergy and Immunology (LIAI; McCausland et al., 2007). STAT5^fl/fl mice were generated by L. Hennighausen and colleagues (National Institutes of Health, Bethesda, MD; Cui et al., 2004) and backcrossed to the C57BL/6J background by M. Farrar and colleagues at the University of Minnesota (Minneapolis, MN), resulting in >10 generations of backcrossing to B6. STAT5^fl/fl mice were further backcrossed to SM mice on the C57BL/6J background. 862 of 884 (97.5%) descriptive single nucleotide polymorphisms in backcrossed STAT5^fl/fl mice were consistent with the C57BL/6J background, as determined by whole genome microsatellite analysis performed through the University of California, Los Angeles Southern California Genotyping Consortium on mice from LIAI. STAT5^fl/fl mice, Prdm1^fl/fl mice, and CD25^−/− mice were crossed with SM mice at LIAI.

STAT5 deletion efficiency was determined by quantitative PCR (forward primer, 5′-ATGGACTCACACCCACAAGGA-3′; and reverse primer, 5′-CACTGCTACAAGGCTACACAAAACC-3′). For CD25 SM experiments, CD25^+/+ littermate control SM mice were used with CD25^+/− SM mice. All animal experiments were conducted in accordance with animal protocols approved by the LIAI Institutional Animal Care and Use Committee.

Gene expression experiments were performed with GFP-expressing RV pMIG as well as a modification of pMIG expressing the fluorescent protein mAmetrine1.1 (Ai et al., 2008) instead of GFP (provided by D. Vignali, St. Jude Children’s Research Hospital, Memphis, TN). STAT5-expressing retroviruses were designed using previously described sequences for WT and constitutively active STAT5b (H299R + S711F; Onishi et al., 1998). Cre recombinase–expressing retrovirus has been previously described (Johnston et al., 2009). Transduction of STAT5^fl/fl SM CD4⁺ T cells with Cre RV resulted in a deletion efficiency of ∼93%, as measured by qPCR analysis.

Virions were produced using the Plat-E cell line as previously described (McCausland et al., 2007). For retroviral transduction of CD4⁺ T cells, CD4⁺ T cells were purified from the splenocytes of naive mice by magnetic bead negative selection (Miltenyi Biotec) and suspended in D-10 (DMEM + 10% fetal calf serum, supplemented with 2 mM GlutaMAX [Invitrogen] and 100 U/ml penicillin/streptomycin [Invitrogen]) with 10 ng/ml recombinant human IL-2 and 50 µM β-mercaptoethanol. 2 × 10⁶ cells per well were stimulated in 24-well tissue culture plates precoated with 8 µg/ml anti-CD3 (clone 17A2; Bio X Cell) and anti-CD28 (clone 37.51; Bio X Cell). After 24 h, cells were transduced as described previously (McCausland et al., 2007). Where necessary, cells were cotransduced by simultaneous transduction with two separately prepared retrovirus stocks. After a total of 72 h of stimulation, CD4⁺ T cells were split and transferred into new wells with fresh D-10, IL-2, and β-mercaptoethanol. After an additional 72 h, transduced CD4⁺ T cells were highly purified by sorting for GFP and/or mAmetrine expression on a FACSDiva or FACSAria (BD). Transduction efficiencies ranged from 10–40% before sorting. 2.0 × 10⁵ or 2.5 × 10⁴ transduced SM cells were adoptively transferred into each C57BL/6J host mouse via the retroorbital sinus for day 4 and day 8 experiments, respectively. In some experiments, cells that were not transduced but treated similarly were used as control cells in parallel adoptive transfer experiments. For day 2 and day 3 experiments, 10⁶ and 5 × 10⁵ freshly isolated SM CD4⁺ T cells, respectively, were adoptively transferred into each C57BL/6J host mouse via the retroorbital sinus.

LCMV stocks were prepared and quantified as previously described (McCausland et al., 2007). Infection doses were 10⁶, 5 × 10⁵, 2 × 10⁵, and 5 × 10⁴ plaque-forming units of LCMV Armstrong per mouse for day 2, day 3, day 4, and day 8 experiments, respectively. Infections were performed by intraperitoneal injection.

Mice received either control rat IgG2a or rat anti–IL-2 (clone S4B6). Each mouse was treated with 1.0 mg antibody via intraperitoneal and retroorbital injection 24 h before LCMV infection and then again 24 h after LCMV infection.

Single-cell suspensions of spleen were prepared by standard gentle mechanical disruption. Surface staining for flow cytometry was performed with monoclonal antibodies against SLAM (CD150; BioLegend) and CD4, CD8, CD45.1, CD44, CD62L, CD25, PD-1, CD69, B220, Fas, and GL7 (eBioscience). Surface stains were performed for 30–60 min at 4°C in PBS supplemented with 0.5% bovine serum albumin and 0.1% sodium azide, unless specified otherwise.

CXCR5 staining was performed as described previously (Johnston et al., 2009) for day 8 experiments using purified anti-CXCR5 (BD) for 60 min, followed by biotinylated anti–rat IgG (Jackson ImmunoResearch Laboratories, Inc.), and then by allophycocyanin (APC)- or PE-labeled streptavidin (Invitrogen) at 4°C in PBS supplemented with 0.5% bovine serum albumin, 2% fetal calf serum, and 2% normal mouse serum. For day 2–4 experiments, CXCR5 staining was performed using biotinylated anti-CXCR5 (BD) for 30 min, followed by APC- or PE-labeled streptavidin at 4°C (Choi et al., 2011).

Intracellular cytokine staining was performed as described previously (McCausland et al., 2007) after stimulation with 20 ng/ml PMA (Sigma-Aldrich) and 1 µM ionomycin (Sigma-Aldrich) in the presence of 2 µg/ml brefeldin-A (BD) for 4 h. Directly conjugated antibodies against IFN-γ and IL-2 (BD) were used. For IL-21, staining was performed using an IL-21R–FC chimeric protein (R&D Systems) followed by PE- or APC-labeled anti–human IgG (Jackson ImmunoResearch Laboratories, Inc.; Johnston et al., 2009). Intracellular staining for Bcl6 was performed as previously described (Choi et al., 2011) with an Alexa Fluor 647–conjugated monoclonal antibody to Bcl6 (clone K112-91; provided by E. O’Donnell and D. Ernst, BD) and the FoxP3 ICS kit buffers and protocol (eBioscience).

For phospho-STAT5 FACS, sorted GFP⁺ cells were fixed and permeabilized with Phosflow Lyse/Fix buffer and Phosflow Perm Buffer III (BD). Where indicated, cells were first stimulated with 20 ng/ml of recombinant human IL-2 for 30 min. Cells were stained with anti–phospho-STAT5 antibody (pY694; BD). Samples were acquired using a C6 Flow Cytometer (Accuri) or an LSRII (BD) and analyzed using FlowJo (Tree Star).

Statistical tests were performed using Prism 5.0c (GraphPad Software). P-values were calculated by two-tailed unpaired Student’s t tests with a 95% confidence interval. Error bars depict the standard error of the mean. *, P < 0.05; **, P < 0.01; ***, P < 0.001.

We thank Mark Kroenke, Danelle Eto, Robin Kageyama, Lindsay Crickard, Cheryl Kim, Kurt Van Gunst, and Anthony Jose for technical advice and assistance. We thank Michael Farrar and Lothar Hennighausen for STAT5^fl/fl mice. We thank Dario Vignali for providing mAmetrine1.1 DNA. We thank Erika O’Donnell, David Ernst, and BD for providing anti-Bcl6 antibodies.

This work was supported by a National Institutes of Health (NIH) training grant to R.J. Johnston and by NIH grants and LIAI institutional funds to S. Crotty.

R.J. Johnston is now a postdoctoral research fellow at Genentech. The authors have no additional conflicting financial interests.