Cytokine receptor modulation by interleukin-2 broadly regulates T helper cell lineage differentiation

Defective T_H1 differentiation in Il2^−/− T cells

We initially focused on the role of IL-2 in T_H1 differentiation, a process that is driven by IL-12. IL-12 induces not only Ifng expression¹ but also T-bet⁷, which promotes the survival and proliferation of differentiating T_H1 cells¹⁶. T-bet induces Ifng^{7, 16} and IL-12Rβ2 expression¹⁷, while inhibiting GATA3 binding to target genes, including the Il4 gene¹⁸, and thus T_H2 differentiation. Naïve CD4⁺ T cells do not respond to IL-12 because they lack a key receptor component, IL-12Rβ2³. Like T_H2 differentiation, T_H1 differentiation requires activation via the T-cell receptor (TCR), which induces production of IL-2. We therefore examined the effect of TCR stimulation on IFN-γ production in wild-type (WT) and Il2^−/− T cells. We stimulated Il2^+/+ or Il2^−/− CD4⁺ T cells with anti-CD3 + anti-CD28, gating on proliferating cells based on FSC-SSC (Fig. 1a, upper two panels). Naïve Il2^−/− CD4⁺ T cells stimulated with anti-CD3 + anti-CD28 had far fewer IFN-γ⁺ cells than did Il2^+/+ CD4⁺ T cells (Fig. 1a, second pair of panels), consistent with previous observations that IL-2 induces IFN-γ production^{13, 14}. Moreover, after two rounds of T_H1 polarization, a lower percentage of Il2^−/− CD4⁺ T cells expressed IFN-γ⁺ as compared to Il2^+/+ cells (Fig. 1a, lower two sets of panels). Although IL-2 contributes to survival and accordingly the Il2^−/− population had more dead cells than the Il2^+/+ population after T_H1 differentiation (Supplementary Fig. 1a,b), in Fig. 1a, we gated on viable cells (see Annexin V and 7AAD staining in Supplementary Fig. 1a); thus the decrease in IFN-γ⁺ cells among Il2^−/− T cells could not be attributed solely to diminished viability.

Because IL-2 is a T-cell growth factor, and because IFN-γ expression is known to increase with successive rounds of cell division¹⁹, we tested whether the role of IL-2 in T_H1 differentiation was due solely to its effects on proliferation or whether it also had proliferation-independent effects on IFN-γ production. To follow cell division we labeled cells with CFSE and incubated them under T_H1 conditions for 72 h. We observed the same number of rounds of cell division, but as anticipated, the Il2^−/− T cells were delayed in their division relative to the Il2^+/+ T cells (Fig. 1b, left panels). We compared the percentage of IFN-γ-producing cells in each division in the Il2^+/+ and Il2^−/− cell populations. As anticipated, IFN-γ expression increased with cell division in Il2^+/+ cells, but there was a decrease in the percentage of IFN-γ-producing cells at each cell division in the Il2^−/− T cells relative to the Il2^+/+ T cells (Fig. 1b; summarized in Fig. 1c). Thus, even when Il2^−/− T cells underwent the same number of divisions as Il2^+/+ T cells, they had diminished production of IFN-γ, suggesting both proliferation-dependent and proliferation-independent effects of IL-2 in T_H1 differentiation.

IL-2 augments Il12rb2 and Tbx21 expression

To clarify the role of IL-2 in T_H1 differentiation, we examined Il2^−/− T_H1 differentiated cells for T_H1-associated genes. These cells expressed less Ifng as well as less Il12rb2 and Tbx21 (encoding T-bet) mRNA than did WT T_H1 cells (Fig. 2a); Ccnd3 (encoding cyclin D3) is shown as a control gene whose expression was not altered (Fig. 2a). Because IFN-γ was reported to drive expression of both Il12rb2 and Tbx21 genes and initiate T_H1 differentiation^{3, 20}, we next investigated whether the defective T_H1 differentiation of Il2^−/− T cells was due to defective IL-2-dependent IFN-γ production. Whereas IL-2 substantially reversed the diminished IFN-γ production observed in the Il2^−/− cells (Fig. 2b), the addition of IFN-γ had a smaller effect, as shown by intracellular staining and secreted cytokine (Fig. 2b, Supplementary Fig. 2). Interestingly, in this setting, IL-2 substantially restored Il12rb2 mRNA expression, whereas IFN-γ only had a modest effect (Fig. 2c, upper panel). In contrast, IFN-γ appeared slightly better than IL-2 in increasing Tbx21 mRNA expression (Fig. 2c, middle panel). We also examined the effect on mRNA encoding Il21rb1, the other component of the IL-12 receptor. Although Il12rb1 expression was not decreased in Il2^−/− T cells, the addition of IL-2 increased its expression (Fig. 2c, lower panel), suggesting that while Il12rb1 expression was not limiting, its induction by IL-2 could potentially further increase IL-12 responsiveness. Consistent with prior reports of a role for IFN-γ in the induction of both the Il12rb2 and Tbx21 genes^{3, 20}, IL-2-induced Il12rb2 and Tbx21 mRNA levels were lower in Ifng^−/− CD4⁺ T cells (Fig. 2d), but the fold induction was not decreased (Fig. 2d), indicating IFN-γ-independent induction by IL-2; thus, the effect of IL-2 in T_H1 differentiation is at least partially independent of IFN-γ. This conclusion was further supported by the ability of IL-2 to induce Il12rb2 and Tbx21 mRNA expression in Stat1^−/− T cells (Fig. 2e); the basal levels were lower than in the Stat1^+/+ T cells, but IL-2 still induced expression. IL-2 also induced Il12rb2 mRNA in Tbx21^−/− T cells (Fig. 2f), indicating that it can act independently of T-bet. For unclear reasons, there was increased basal Il12rb2 mRNA expression in the Tbx21^−/− T cells (Fig. 2f). IL-2 and IL-12 were previously shown to cooperatively induce Il12rb2²¹, our data now indicate a direct IL-12-independent effect of IL-2. Consistent with effects of both IL-2 and IFN-γ on T_H1 differentiation, antibodies to IL-2Rα and IL-2Rβ or to IFN-γ decreased Il12rb2 and Tbx21 mRNA expression (Fig. 2g); the antibodies cooperatively inhibited Tbx21 expression, whereas blocking IL-2 or IFN-γ alone decreased Il12rb2 mRNA (Fig. 2g).

IL-2 also induced human IL12RB2 and TBX21 gene (Fig. 3a) and protein (Fig. 3b,c) expression, whereas IFN-γ had little if any effect at 4 h (Fig. 3a-c). We confirmed there was no reproducible significant effect of IFN-γ even at 8 or 24 h time points (Supplementary Fig. 3). Thus in both mice and humans, IL-2 promotes expression of IL-12Rβ2 and T-bet. To examine the role of IL-2 in human T_H1 differentiation, we used blocking antibodies to IL-2Rα (anti-Tac²²) and IL-2Rβ (Mikβ1²³). When human peripheral blood naïve (CD45RA^highCD25⁻HLA⁻DR⁻) CD4⁺ T cells were cultured under T_H1-polarizing conditions, the anti-IL-2R antibodies decreased IL12RB2, TBX21, and IFNG mRNA induction (Fig. 3d) as well as the percentage of IFNγ⁺ CD4⁺ T cells (Fig. 3e). The addition of exogenous IFN-γ only slightly reversed this effect (Fig. 3e). Similar to mouse T_H1 differentiation, during human T_H1 differentiation, T cells undergo cell division (Fig. 3f, left panel) and the percentage of IFN-γ⁺ cells tended to increase with each division (Fig. 3g, left panel). As expected, because IL-2 is a T cell growth factor, the addition of anti-IL-2Rα or IL-2Rβ during T_H1 differentiation decreased proliferation (Fig. 3f, right panel), but this treatment also markedly diminished the percentage of IFN-γ⁺ cells in each division (Fig. 3g), indicating that analogous to mouse T_H1 differentiation, the effect of IL-2 on human T_H1 differentiation is partially independent of its proliferative effect. Consistent with the data in Fig. 3e, IFN-γ had little if any effect in reversing the inhibition mediated by anti-IL-2R antibodies (Fig. 3g). Thus, in mice and humans, IL-2 can promote T_H1 differentiation in a manner at least partially independent of proliferation and IFN-γ.

STAT5A and STAT5B regulate Il12rb2 and Tbx21 expression

Because IL-2 activates STAT5A and STAT5B⁹, we next investigated the role of these STAT5 proteins in Il12rb2 and Tbx21 gene expression. By RT-PCR and flow cytometry, following in vitro deletion of Stat5a and Stat5b in CD4⁺ T cells from mice in which the Stat5a-Stat5b locus was floxed (Fig. 4a, first two panels) Il12rb2 and Tbx21 mRNA expression were partially decreased, with a greater decrease in Il12rb2 expression (Fig. 4a, 3^rd and 4^th panels). Although the decrease in Tbx21 mRNA was not statistically significant, there was a reproducibly observed decrease in T-bet protein (Fig. 4a, 5^th panel). To further investigate the importance of STAT5 in T_H1 differentiation, we examined the effect of retroviral expression of a constitutively-activated form of STAT5A, denoted STAT5A-1*6²⁴. When this was transduced into Il2^−/− T cells, defective T_H1 differentiation was partially reversed (Fig. 4b), consistent with a role for STAT5 in T_H1 differentiation (Fig. 4b). ChIP-Seq was used to look at STAT5 interaction with the Il12rb2 gene. IL-2 induced STAT5A and STAT5B binding to the mouse Il12rb2 gene at multiple TTCN₃GAA GAS motifs (Fig. 4c; sequences are in Supplementary Fig. 4). We cloned the TSS-proximal region spanning GAS1 and GAS2 5′ of the TSS and the GAS3 region 3′ of the TSS in the pGL4.10 luciferase reporter vector and found that reporter activity was induced by IL-2 (Fig. 4d, see GAS1,2,3 construct) and required an intact GAS1 motif (Fig. 4d). Consistent with IL-2-induced IFN-γ-independent Il12rb2 expression (Fig. 2d), IL-2 still induced reporter activity in the presence of anti-IFN-γ (Fig. 4d). The importance of GAS1 was confirmed in preactivated human T cells, with STAT5A and STAT5B binding IL12RB2 at the site (Fig. 4e); note the sequence conservation of the GAS1 motif in humans and mice (bottom of Fig. 4e). No IFN-γ-induced STAT1 binding was detected at this site (Fig. 4e), in keeping with little if any effect of IFN-γ on IL-12Rβ2 expression (Fig. 3a,b). Consistent with the ability of IL-12 to induce expression of its own receptor (Fig. 3a), IL-12 induced Il12rb2 reporter activity, albeit more weakly than IL-2 (Fig. 4f, GAS1,2,3 construct), and this also required an intact GAS1 motif (Fig. 4f).

Analogous to their binding to elements in the Il12rb2 gene, STAT5A and STAT5B also bound to the mouse Tbx21 gene (Fig. 5a), ≈12 kb 5′ of the TSS at a region that binds STAT4²⁵, and this region mediated IL-2-induced luciferase reporter activity (Fig. 5b). As expected, anti-IFN-γ had only a partial effect on reporter activity (Fig. 5b and Supplementary Fig. 5), consistent with IFN-γ-independent effects of IL-2 on T_H1 differentiation, as we observed above. Whereas the non-canonical motifs may partially contribute to reporter activity, mutation of the canonical GAS motif essentially abrogated activity (Fig. 5b, mut3 vs. mut1 and mut2). In the human TBX21 gene, STAT5A and STAT5B bound the same region as STAT4 (Fig. 5c) at a site conserved between human and mouse (Fig. 5d), suggesting the possibility that these different STATs can compete for binding to the same site. STAT1 exhibited only very low binding at this site (Fig. 5c), consistent with higher induction of TBX21 by IL-2 and IL-12 than by IFN-γ (Fig. 3e). In contrast, STAT1 as well as STAT4, STAT5A, and STAT5B bound to the promoter region of human IRF1 (Fig. 5e), another transcription factor important for T_H1 differentiation²⁶. When we deleted Stat5a and Stat5b by Cre-mediated deletion in vitro using mouse cells there was a slight decrease in Irf1 mRNA as well as in Il12rb1 mRNA (Supplementary Fig. 6), consistent with Il12rb1 being regulated by Irf1²⁶. Whereas IL-2 more strongly induced Il12rb2 promoter activity than did IL-12 (Fig. 4f), it less strongly induced Tbx21 reporter activity (Fig. 5f). Both cytokines required the same GAS motif for Tbx21 induction (Fig. 5f, mut3); thus, IL-2 and IL-12 act at least in part via shared elements to regulate expression of both T-bet and IL-12Rβ2. Antibodies to IL-2, IL-2Rα, and IL-2Rβ lowered IL-12-induced Tbx21 activity (Fig. 5g), presumably at least in part by their inhibition of IL-2-induced IL-12Rβ2 expression.

Analysis of ChIP-Seq libraries from T_H1-differentiated cells revealed that the STAT5A and STAT5B peaks seen at the GAS1, GAS2, and GAS3 motifs in the Il12rb2 locus in T cells stimulated with IL-2 (Fig. 4c) were also evident in T_H1 cells (Fig. 6a). STAT4, STAT5A, and STAT5B ChIP-Seq data revealed that both STAT5B and STAT5A exhibited binding at the Il12rb2 GAS1 site that was relatively stronger than that of STAT4; conversely, STAT4 exhibited very strong binding upstream of the Tbx21 gene (the 11^th strongest peak out of 7318 peaks in the STAT4 library), whereas STAT5A and STAT5B binding were relatively weak at this site. Because these data were from the same ChIP-Seq libraries, we can conclude that the relative binding of STAT5A and STAT5B is weaker than STAT4 binding at the Tbx21 locus but stronger than that of STAT4 at the Il12rb2 locus in T_H1 differentiated cells (Fig. 6, b vs. a). Coupled to the functional reporter assays (Fig. 4f, 5f), these data suggest that IL-2 via STAT5 is the dominant signal for Il12rb2 induction, whereas IL-12 via STAT4 is more critical for Tbx21 regulation.

Il12rb2 rescues T_H1 differentiation in Il2^−/− T cells

We next investigated whether enforced expression of Il12rb2 or Tbx21 could rescue defective T_H1 differentiation in Il2^−/− cells. As expected, Il12rb2 retroviral transduction had little effect in Il2^+/+ cells, which already express IL-12Rβ2; however, transduction of Il2^−/− CD4⁺ T cells with the Il12rb2 retrovirus markedly increased IFN-γ protein expression (Fig. 6c,d). In contrast, transduction of Tbx21 did not rescue impaired T_H1 differentiation in Il2^−/− CD4⁺ T cells (Fig. 6e), suggesting IL-2-induced T-bet is not sufficient for T_H1 differentiation. Nevertheless, retroviral transduction of Tbx21 increased IFN-γ production in both T_H17-differentiated Il2^+/+ and Il2^−/− T cells (Fig. 6e), confirming the functionality of the retrovirally-transduced protein, although the basis for the different effect of the Tbx21 retrovirus in T_H1 vs. T_H17-differentiated cells is unclear. Interestingly, transduction of the Il12rb2 retrovirus substantially increased Il12rb2 and Tbx21 mRNA, as compared to a control virus in both Il2^+/+ and Il2^−/− T cells (Fig. 6f). In contrast, the Tbx21 retrovirus significantly increased Tbx21 mRNA but if anything slightly decreased Il12rb2 mRNA expression in Il2^+/+ T cells and only modestly increased Il12rb2 mRNA in Il2^−/− T cells (Fig. 6f). At the protein level, both viruses induced T-bet expression (Fig. 6g). In contrast, for IL-12Rβ2 expression, retroviral transduction of Il12rb2 had a modest shift whereas transduction of Tbx21 had very little if any discernible shift (Fig. 6h).

IL-2 is required for T_H1 responses in vivo

The above data collectively revealed a critical role for IL-2 in T_H1 differentiation in vitro. We next sought to extend these findings by examining an antigen-specific response in vivo. Mice immunized with pigeon cytochrome C (PCC) exhibited a marked increase in cell surface IL-12Rβ2 expression on Il2^+/+ CD4⁺ T cells, but there was little effect in T cells from Il2^−/− mice (Fig. 7a right vs. left panels; Fig. 7b). Similarly, less Tbx21 mRNA was detected in cells from Il2^−/− mice (Fig. 7c) and PCC-induced IFN-γ protein (Fig. 7d) and mRNA (Fig. 7e) expression was also diminished. Thus, IL-2 is required for normal Il12rb2 and Tbx21 expression in vivo.

IL-2 decreases gp130 expression

The above data and a prior study⁴ together establish key roles for IL-2 in T_H1 and T_H2 differentiation, at least in part based on cytokine receptor modulation. IL-2 can also negatively regulate T_H17 differentiation¹⁵. Given the key role of IL-6 in T_H17 differentiation, we investigated whether IL-2 might inhibit expression of IL-6Rα and/or gp130 (encoded by Il6ra and Il6st, respectively), and indeed found that IL-2 stimulation decreased Il6ra and Il6st mRNA (Fig. 8a) and protein (Fig. 8b and Fig. 8c) expression. Consistent with this, under T_H17 conditions, both Il6ra and Il6st mRNA expression were higher in Il2^−/− than in WT T cells (Fig. 8d). The increased expression correlated with a higher level of Il17a mRNA (Fig. 8d). Il12rb2 was expressed even in T_H17-differentiated cells in an IL-2-dependent fashion (Fig. 8d), which might explain the ability of T_H17 cells to produce IFN-γ when IL-12 is present and thus for the cells to exhibit T_H plasticity²⁷. Because IL-2 is required for normal IFN-γ production (Fig. 1a) and IFN-γ negatively regulates the generation of T_H17 cells⁶, it was conceivable that increased IL-17 protein in the absence of IL-2 resulted from the loss of IFN-γ production by these cells. However, we could not detect IFN-γ production by the T_H17 cells in either the WT or Il2^−/− background (Fig. 8e, left panel), indicating that IFN-γ by itself cannot explain the difference in IL-17 production in the Il2^+/+ and Il2^−/− T cells (Fig. 8e right panel). When Il6ra and Il6st were separately introduced into CD4⁺ T cells by retroviral transduction (Fig. 8f), only Il6st could partially rescue the defective T_H17 differentiation caused by IL-2, with an increase in IL-17A protein expression (Fig. 8g). The fact that Il6st retroviral transduction increased IL-17A expression in WT mice (Fig. 8g, iii vs. i) indicated that the level of gp130 is limiting. This was underscored by the fact that IL-2 treatment diminished Il6st expression (Fig. 8a) and that retroviral expression of Il6st into IL-2-treated cells increased T_H17 differentiation (Fig. 8g, vi vs. iv). However, given that IL-2 inhibited T_H17 differentiation even when Il6st was constitutively expressed (Fig. 8g, vi vs. iii), the role of IL-2 extends beyond the regulation of gp130.