Drivers of IL-22 secretion by ILC3s
Secretion of the cytokine interleukin-22 (IL-22) by innate lymphoid cells and intestinal T cells in the gut promotes epithelial integrity and provides protection against intestinal pathogens. Bauché et al. examined which cytokines and signal transducer and activator of transcription (STAT) proteins activate IL-22 production by group 3 innate lymphoid cells (ILC3s) in mice. They found that IL-2 acting through STAT5 contributed to the induction of IL-22 in ILC3s, bolstering the ability of T cell–deficient mice to fend off Citrobacter rodentium infection. IL-23 activated both STAT3 and STAT5 in ILC3s, enabling binding of a STAT3/STAT5 complex to the IL-22 promoter. These studies revealed that ILC3s and T cells use distinct sets of cytokines and STATs to achieve activation of IL-22 expression.
Abstract
Signal transducer and activator of transcription (STAT) proteins have critical roles in the development and function of immune cells. STAT signaling is often dysregulated in patients with inflammatory bowel disease (IBD), suggesting the importance of STAT regulation during the disease process. Moreover, genetic alterations in STAT3 and STAT5 (e.g., deletions, mutations, and single-nucleotide polymorphisms) are associated with an increased risk for IBD. In this study, we elucidated the precise roles of STAT5 signaling in group 3 innate lymphoid cells (ILC3s), a key subset of immune cells involved in the maintenance of gut barrier integrity. We show that mice lacking either STAT5a or STAT5b are more susceptible to Citrobacter rodentium–mediated colitis and that interleukin-2 (IL-2)– and IL-23–induced STAT5 drives IL-22 production in both mouse and human colonic lamina propria ILC3s. Mechanistically, IL-23 induces a STAT3-STAT5 complex that binds IL-22 promoter DNA elements in ILC3s. Our data suggest that STAT5a/b signaling in ILC3s maintains gut epithelial integrity during pathogen-induced intestinal disease.
INTRODUCTION
The Janus kinase (JAK) and signal transducer and activator of transcription (STAT) signaling proteins are important regulators of lymphoid cell development and function (1). JAK-STAT signaling transduces a variety of immune stimuli including cytokines such as interleukin-23 (IL-23) and IL-2. In lymphoid cells, IL-23 binding to its receptor, a heterodimer consisting of IL-23R and IL-12Rβ chains, activates a JAK2-STAT3 pathway (2), whereas IL-2 binding to the IL-2R (consisting of α, β, and γ chains) activates a JAK1-STAT5 pathway (3, 4). Activation of STAT3 versus STAT5 promotes distinct T cell differentiation programs. Although STAT3 signaling induces T helper 17 (TH17) cell differentiation by transactivating RORγt, which drives IL-17 expression and tissue inflammation, STAT5 induces the expression of Foxp3 in CD4+ T cells associated with tissue homeostasis (5–7). Because STAT3 and STAT5 are both up-regulated in the colon of patients with inflammatory bowel disease (IBD) and in preclinical models of experimental colitis (8), pharmacologic modulation of those factors may be a promising therapeutic strategy for the treatment of IBD. However, treatment with sunitinib (a multitargeted tyrosine kinase inhibitor) or pacritinib (a JAK2 inhibitor) led to severe diarrhea in patients (9, 10) and death of mice during Citrobacter rodentium infection, a mouse model of infectious colitis (11). Moreover, alterations (including deletions, mutations, and single-nucleotide polymorphisms) in the loci coding for Stat3 and Stat5, as well as Il2r, Il23r, and Jak2, have been associated with an increased risk of developing IBD and/or colon cancer (12–14), suggesting a crucial role of these pathways in gut homeostasis.
Growing numbers of studies revealed that a subset of lymphoid cells, called group 3 innate lymphoid cells (ILC3s), play an important role in maintaining intestinal barrier integrity through the secretion of IL-22 (15). IL-22 is an IL-10 family member cytokine that promotes epithelial renewal through STAT3 activation in intestinal epithelial cells. Engagement of IL-22 with a heterodimeric receptor consisting of IL-22Rα and IL-10Rβ induces the secretion of antimicrobial peptides and maintains fucosylation of the intestinal epithelium (16–18). Mice lacking IL-22–producing ILC3s are highly susceptible to experimental colitis (19). IL-22 expression is driven by the transcription factors including RORγt, aryl hydrocarbon receptor (AhR), and STAT3 (11, 20, 21). In addition to STAT3, STAT5 may also contribute to the function and/or development of ILC3s (11, 22). In this study, we explored the roles of STAT5a and STAT5b during C. rodentium infection, a model of inflammatory colitis. We found that both IL-2 and IL-23 can induce STAT5 to promote IL-22 production in ILC3s. This is in contrast to CD4+ T cells, which do not depend on STAT5 for IL-22 secretion. Moreover, upon IL-23, but not IL-2, stimulation, STAT5 and STAT3 proteins interact and bind to the same region of the IL-22 promoter. Our findings highlight an unappreciated role of STAT5 in ILC3s during pathogen-induced colitis.
RESULTS
STAT5 signaling in ILC3s, but not in T cells, is required for resistance to C. rodentium–induced colitis
The balance between STAT3 and STAT5 signaling is essential to maintain the development and function of intestinal lymphoid populations including TH17 cells, regulatory T cells (Tregs), and ILC3s. Recent studies demonstrated that STAT5, but not STAT3, is required for the development and survival of ILC3s (11, 22). Although STAT3 activation promotes IL-22 production by ILC3s and protects from lethal C. rodentium infection (11), the role of STAT5 in ILC3-mediated host-protective responses remains unexplored. Because STAT3 and STAT5 have opposing roles in T cell differentiation and function, we hypothesized that this could also be true in ILC3s. To test this hypothesis, we generated mice lacking either Stat3 (STAT3 ΔT,ILC3) or Stat5a/b (STAT5 ΔT,ILC3) (Fig. 1A) specifically in both T cells and ILC3s by crossing Stat3-floxed or Stat5a/b-floxed mice with Rorc-cre–expressing mice. Moreover, we crossed Stat5a/b-floxed mice with Cd4-cre–expressing mice (STAT5 ΔT) to abrogate Stat5 expression only in T cells. Then, we orally infected these mice with C. rodentium, an attaching and effacing bacterial pathogen capable of causing lethal colitis in susceptible mouse strains. As described previously, 100% of STAT3 ΔT,ILC3 mice cannot clear infection and succumbed by day 15 (fig. S1 and Fig. 1B) (11). Unexpectedly, 50% of STAT5 ΔT,ILC3 mice died from infection and the surviving mice developed severe colitis compared with wild-type (WT) littermate controls (Fig. 1, B and C) with greater bacterial dissemination into the liver (Fig. 1D), suggesting that STAT5, like STAT3, has a critical role in resistance to C. rodentium–induced colitis. STAT5ΔT mice did not lose weight and survived the infection, suggesting a protective role of STAT5 in ILC3s rather than T cells (Fig. 1B). Extensive literature has demonstrated the protective role of IL-22 in C. rodentium infection (23). Six days after infection, STAT5ΔT,ILC3 mice showed a decreased concentration of IL-22 in the serum compared with WT controls (Fig. 1E). During intestinal inflammation, IL-22 is mainly expressed by ILC3s and TH17 cells. Analysis of the production of IL-22 by ILC3s and TH17 cells revealed that Stat5a/b deficiency impaired IL-22 production in colon lamina propria ILC3s but not in TH17 (Fig. 1, F to I, and fig. S2). Our data suggest that STAT5 is required to maintain intestinal ILC3 function in vivo.
(A) Phosphorylation of STAT5 in T cells after IL-7 stimulation in Rorc-CreStat5a/bfl/fl (Stat5 ΔT,ILC3) and WT littermate control. Data are representative of two independent experiments (n = 3 mice per group). (B) Rorc-CreStat3fl/fl (Stat3 ΔT,ILC3), Rorct-CreStat5a/bfl/fl (Stat5 ΔT,ILC3), Cd4-Cre Stat5a/bfl/fl (Stat5 ΔT), and their WT littermate control were orally infected with 109 C. rodentium. Survival was monitored over a 21-day period. (C) Photomicrographs of H&E-stained histological sections of colon and pathology score of Rorc-CreStat5a/bfl/fl (Stat5 ΔT,ILC3) and WT littermate control at day 13 after infection. Scale bars, 200 μm. Data are representative of two independent experiments (n = 6 mice per group). Mean ± SEM. *P < 0.05 (unpaired t test). (D to G) On day 6 after infection, bacterial dissemination in the liver (D) was assessed on MacConkey agar plates and the level of serum IL-22 (E) was measured by ELISA. Proportion (F) and absolute number (G) of colon lamina propria IL-22–producing ILC3s were determined by flow cytometry. Data are representative of two independent experiments (n = 6 mice per group). Mean ± SEM. *P < 0.05 (unpaired t test). Proportion (H) and absolute number (I) of colon lamina propria IL-22–producing CD4+ T cells were determined by flow cytometry. Data are representative of two independent experiments (n = 6 mice per group). Mean ± SEM. *P < 0.05 (unpaired t test). ns, not significant.
STAT5a and STAT5b are required for optimal production of IL-22 in intestinal ILC3s
STAT5 has two isoforms, STAT5a and STAT5b. Although often coactivated, STAT5a and STAT5b can have different roles in vivo (24–26). To study the function of STAT5, we transferred bone marrow (BM) cells from Stat5a−/−, Stat5b−/−, or WT littermate control mice into irradiated Il2ry−/−Rag2−/− mice as previously described (22) (these BM reconstituted mice are referred to as Stat5a−/−, Stat5b−/−, or WT mice, respectively). Compared with mice reconstituted with WT marrow, mice receiving either Stat5a−/− or Stat5b−/− BM were more susceptible to C. rodentium infection as demonstrated by an extensive bacterial dissemination into the liver (Fig. 2, A and B). Six days after infection, the IL-22 level was significantly lower in the serum and colon of both Stat5a−/− and Stat5b−/− mice (Fig. 2, C and D), suggesting that activation of both STAT5 isoforms enhanced IL-22 secretion. To study whether IL-22 reconstitution in Stat5a−/−and Stat5b−/− mice could reduce severity of infection, we injected IL-22 minicircle DNA via hydrodynamic tail vein delivery to induce systemic IL-22 expression. We show that exogenous recombinant IL-22 can delay lethal infection in Stat5-deficient BM-transferred mice (Fig. 2A). We observed a decreased proportion of colon lamina propria IL-22–producing ILC3s, but not IL-22–producing TH17 cells, in both Stat5a−/− and Stat5b−/− mice (Fig. 2, E and F, and fig. S3A). Accordingly, Stat5 deficiency enhances IL-17 production by TH17 cells without altering RORγt expression (fig. S3, B and C) (7). Stat5b−/− mice were more sensitive to C. rodentium infection than Stat5a−/− mice, suggesting nonoverlapping roles for these two isoforms. In our model of BM transfer, all hematopoietic subsets lacked STAT5 including ILC3s. To ensure that the effect observed was intrinsic to ILC3s, we generated a mixed BM chimera with a 1:1 ratio of Thy1.1 WT to Thy1.2 STAT5−/− cells and measured IL-22 production by colonic lamina propria ILC3s. Consistent with the BM transfer results, Stat5a- and Stat5b-deficient ILC3s produced less IL-22 after infection (Fig. 2G). Similar to a recent report (22), we found that Stat5b, but not Stat5a, controls the development of NKp46+ ILC3s (fig. S4). CD4+ ILC3 development was not affected in STAT5-deficient mice (fig. S4), suggesting that different subsets of ILCs may rely more on STAT5b than STAT5a. Together, our data suggest that STAT5 activation promotes IL-22 production by intestinal ILC3s.
(A) BM-transferred mice (Stat5a−/−, Stat5b−/−, or Stat5a+/+Stat5b+/+ control into Rag2−/−Il2rγ−/−) received 8 μg of IL-22 minicircle (mc) (or control minicircle) followed by oral infection with 109 C. rodentium. Survival was monitored over a 15-day period. Data are representative of two to four independent experiments (n = 8 to 17 mice per group). *P < 0.05, **P < 0.01, and ***P < 0.001 (Mantel-Cox test). (B to F) On day 6 after infection, bacterial loads (B) in the feces (left) and the liver (right) were assessed on MacConkey agar plate, serum level of IL-22 (C) was measured by ELISA, gene expression profile of the proximal colon (D) was assessed by real-time qPCR, and proportion (E and F) of colon lamina propria IL-22–producing ILC3s was determined by flow cytometry. Data are representative of two to five independent experiments (n = 7 to 13 mice per group). Mean ± SEM. *P < 0.05, **P < 0.01, and ***P < 0.001 (unpaired t test). (G) Percentage of IL-22–producing ILC3s in mixed (ratio 1:1) bone chimeras was measured by flow cytometry 6 days after infection. Data are representative of three independent experiments (n = 5 to 8 mice per group). Mean ± SEM. *P < 0.05 (paired t test).
IL-2 and IL-23 drive IL-22 production via STAT5 in ILC3s
In lymphoid cells, STAT5 can be activated by common gamma chain (γc) family cytokines including IL-2, IL-7, and IL-15 [(27) and fig. S5]. Therefore, we next determined which STAT5-activating cytokines could promote IL-22 production using lamina propria ILC3s. We found that IL-2, but not IL-7 or IL-15, has a similar capacity as IL-23 to drive IL-22 production in ILC3s isolated from Rag2−/− Rorc-eGFP or Rorc-eGFP mice (Fig. 3, A and B, and fig. S6A). We additionally observed increased IL-22 secretion in a dose-dependent manner after ex vivo stimulation of colonic ILC3s with IL-2 (fig. S7A). Moreover, we found that IL-2 directly stimulated Il22 expression but had no regulatory effects on the genes for other cytokines typically produced by ILC3s such as Il17a, Csf2, or Ifng (fig. S7B). To confirm the in vivo physiological relevance of IL-2 stimulating innate immunity, we infected Il2−/−Rag2−/− mice with C. rodentium and observed that Il2−/−Rag2−/− mice died earlier than Rag2−/− controls after infection (Fig. 3C). Moreover, we found that Il23r−/−Rag2−/− mice were even more susceptible to infection than Il2−/−Rag2−/− mice, suggesting that IL-2 might have a protective role in the late phase of infection when compared with the early protective role of IL-23 in this infection model.
(A) Colon lamina propria cells from Rag2−/− Rorc-eGFP were stimulated with various cytokines as indicated for 18 hours. IL-22 level was measured by ELISA from supernatants. (B) Cells were restimulated for 4 hours with PMA/ionomycin, and IL-22 production by ILC3s was assessed by flow cytometry. Data are representative of two independent experiments (n = 6 mice per group). Mean ± SEM. *P < 0.05, **P < 0.01, and ***P < 0.001 (unpaired t test). (C) Il2−/− Rag2−/−, Rag2−/−, and Il23r−/−Rag2−/− mice were orally infected with 109 C. rodentium. Survival was monitored over a 21-day period. Data are representative of two independent experiments (n = 7 to 8 mice per group). **P < 0.01 and ***P < 0.001 (Mantel-Cox test). (D and E) BM from Stat5a−/−, Stat5b−/−, or Stat5a+/+Stat5b+/+ control mice was transferred into Rag2−/−Il2ry−/− mice and parked for 6 to 10 weeks. Colon lamina propria cells were stimulated with recombinant IL-2 or IL-23 as indicated. (D and E) Representative dot plots (D) and proportion (E) of ILC3s producing IL-22. Data are representative of two to five independent experiments (n = 7 to 13 mice per group). Mean ± SEM. *P < 0.05, **P < 0.01, and ***P < 0.001 (paired t test). (F and G) Human normal colon cells were stimulated with recombinant IL-2 or IL-23 in the presence of STAT5 inhibitor as indicated. Representative dot plots (F) and proportion (G) of ILC3s producing IL-22. DMSO, dimethyl sulfoxide. Data are representative of five independent experiments (n = 6 samples). Mean ± SEM. *P < 0.05, **P < 0.01, and ***P < 0.001 (paired t test).
We next compared IL-2 and IL-23 activation of STAT5 to promote IL-22 production in ILC3s. With the expectation that IL-2, but not IL-23, would activate STAT5 to promote IL-22 secretion in ILC3s, we stimulated colonic lamina propria cells from Stat5a−/−, Stat5b−/−, or WT controls with recombinant IL-2 and IL-23 and observed that, in the absence of STAT5a or STAT5b, IL-2 cannot drive IL-22 production in ILC3s (Fig. 3, D and E). Unexpectedly, IL-23–stimulated ILC3s produce significantly less IL-22 in the absence of either STAT5a or STAT5b (Fig. 3, D and E). These findings provide additional evidence that Stat5a−/−, but not Stat5b−/−, ILC3s significantly up-regulate IL-22 in the presence of IL-23, suggesting once again that the functions of STAT5 in ILC3s are predominantly mediated by the STAT5b isoform. Although cytokine-mediated activation and regulation of murine ILC3s is well documented (28), regulation of IL-22 expression by human colonic ILC3s still remains unclear. Therefore, we used fresh normal human colon tissue to elucidate the regulation of IL-22 by ILC3s, and we found that IL-2 and IL-23 can promote IL-22 production by ILC3s in a STAT5-dependent manner (Fig. 3, F and G). Of note, ILC3s isolated from the blood of healthy donors do not produce IL-22 when stimulated with IL-23, suggesting that the circulating population is relatively unresponsive compared with gut-resident ILC3s (fig. S8). Together, our data suggest that both IL-2 and IL-23 use STAT5 to drive IL-22 production in mouse and human intestinal ILC3s.
STAT3 and STAT5 interact in IL-23–stimulated ILC3s
Because STAT5 contributes to IL-22 production after IL-23 stimulation, we next examined whether IL-23 can activate STAT5. We stimulated intestinal ILC3s with either IL-2 or IL-23 and measured phosphorylation of STAT3 and STAT5 by flow cytometry and immunofluorescence. IL-23 activates not only STAT3 as previously reported (11) but also STAT5 (Fig. 4, A and B), although IL-23 is a less potent STAT5 activator than IL-2 (Fig. 4A). Unlike IL-23, the other STAT3-inducing cytokines such as IL-6 and IL-21 do not activate STAT5 in intestinal ILC3 (fig. S5). Moreover, IL-23 does not induce STAT5 phosphorylation in intestinal RORγt+CD4+ T cells (fig. S5), suggesting a unique role of IL-23 in ILC3 function. In addition, STAT3 is activated in only 10% of ILC3s upon IL-2 stimulation (Fig. 4A). However, we show that IL-2 can still drive IL-22 production in the absence of STAT3 (fig. S9), suggesting that this weak activation of STAT3 by IL-2 does not impair the function of ILC3s. STAT protein complexes have been shown in various subsets of cells. Therefore, to investigate whether IL-23 induces a STAT3-STAT5 complex, we performed a proximity ligation assay on sorted intestinal ILC3s. Upon IL-23 stimulation, we identified a significantly increased number of STAT3-STAT5 interactions, as observed and quantified by fluorescent imaging, in the nucleus of ILC3s (Fig. 4C), revealing an unexpected mechanism of action for IL-23.
(A and B) Colon lamina propria cells from Rorc-eGFP Rag2−/− were stimulated with IL-2 or IL-23 as indicated. Phosphorylation of STAT3 and STAT5 in ILC3s was measured by flow cytometry (A) and immunofluorescence (B). Data are representative of (A) three independent experiments (n = 6 mice per group) or (B) two independent experiments (n = 3 mice). Mean ± SEM. *P < 0.05 and **P < 0.01 (paired t test). (C) Interaction between STAT3 and STAT5 after cytokine stimulation was determined by proximity ligation assay on sorted intestinal ILC3s (lineage−CD45intCD90high). Each row represents one cell from one independent experiment. Data are representative of (B) two independent experiments (n = 3 mice) or (A and C) five independent experiments (n = 6 to 7 mice). Mean ± SEM. *P < 0.05 and **P < 0.01 (paired t test).
STAT5 binds to IL-22 promoter in ILC3s
Il22 expression in ILC3s is driven by cytokines (e.g., IL-23 and IL-2), cytokine and chemokine receptors (e.g., IL-23R and CXCR6), and transcription factors (e.g., RORγt, AhR, and Notch2) (20, 29, 30). Mice deficient for any of those molecules have impaired IL-22 production by ILC3s. We measured the expression of various genes from Stat5a−/−, Stat5b−/−, or WT intestinal ILC3s. We found that STAT5 deficiency does not significantly down-regulate Ahr, Il23r, Rorc, and Notch2 expression (Fig. 5A), suggesting that STAT5 directly contributes to IL-22 production upon IL-2 and IL-23 stimulation. STAT5a deficiency resulted in increased Il23r and Notch2 expression in ILC3s, suggesting that STAT5 might directly control IL-22 expression (Fig. 5A). Analysis of the IL-22 promoter (accession number NC_000076.6) revealed that it contains three STAT-response elements (SREs; sequence TTCNNNGAA), three xenobiotic-responsive elements (XREs; AhR binding site; sequence GCGTG), and one RORγt-binding site (RORE; sequence AGGTCA) (Fig. 5B). To determine whether STAT5 binds to the IL-22 promoter of sorted intestinal ILC3s, we performed a chromatin immunoprecipitation assay followed by quantitative polymerase chain reaction (ChIP-qPCR) targeting specific regions of the IL-22 promoter (Fig. 5B). We show that IL-2– and IL-23–induced STAT5 preferentially binds to SRE1 but not the other SRE, XRE, or RORE sites (Fig. 5, C and D). In addition, upon IL-23 stimulation, both STAT5 and STAT3 proteins bind to the same region on the IL-22 promoter, suggesting that STAT3 and STAT5 cooperate to drive IL-22 production in ILC3s. In summary, we found that STAT3 and STAT5 can each induce IL-22 in ILC3s, but maximal induction is only achieved when they are simultaneously activated under conditions favoring STAT3-STAT5 complex formation.
(A) BM from Stat5a−/−, Stat5b−/−, or Stat5a+/+Stat5b+/+ control was transferred into Rag2−/−Il2rγ−/− mice and parked for 6 to 10 weeks. Gene expression by sorted intestinal ILC3s (lineage−CD45intCD90high) was measured by qPCR. Data are representative of two independent experiments (n = 4 to 5 mice). Mean ± SEM. **P < 0.01 and ***P < 0.001 (unpaired t test). a.u., arbitrary units. (B) SRE, RORE, and XRE in mouse IL-22 promoter. Numbers represent the distances from the transcription start site. (C and D) STAT3 and STAT5 binding to IL-22 promoter upon IL-2 or IL-23 stimulation was determined by ChIP-qPCR on sorted intestinal ILC3s. Data are representative of five independent experiments (ILC3s from five Rag2−/− Rorc-eGFP mice were pooled for each independent experiment).
DISCUSSION
Tissue-resident RORγt-expressing cells play an important role in maintaining intestinal barrier integrity through secretion of IL-17 and IL-22. Mice lacking these cytokines are highly susceptible to experimental colitis due to defective secretion of defensins and impaired tight junctions at the epithelial surface (11, 18, 19). RORγt-expressing cells are tightly regulated by the balance of STAT3 and STAT5 proteins. Whereas STAT3 drives Rorc, Il17, and Il22 expression, STAT5 suppresses Il17 expression by preventing STAT3 from binding to the IL-17 promoter of TH17 cells (7). Colon lamina propria Foxp3+ Tregs express high levels of RORγt compared with other organs (31). It has been previously reported that specific deletion of STAT3 or STAT5 in Tregs can lead to colitis (32), suggesting that although these two STATs are antagonistic in T cells, both STAT3 and STAT5 are required to maintain gut homeostasis. In this study, we demonstrate that STAT3 and STAT5 cooperate to maintain function of intestinal ILC3s. We show that STAT5 signaling in ILC3s is required to protect from C. rodentium lethal infection. Moreover, we reveal an unappreciated role for IL-2– and IL-23–induced STAT5 in intestinal IL-22–producing ILC3s.
Whereas the role of IL-23 as a potent inducer of IL-22 production in ILC3s is well known (33), we identify a previously unappreciated function of IL-2 as an IL-22–inducing cytokine. We show that IL-2, but not IL-7 or IL-15—twoSTAT5-activating cytokines—drives IL-22 production in both mouse and human intestinal ILC3s ex vivo. IL-15 and IL-2 both signal through IL-2Rβ and the γc chain but have distinct immunological functions (34). Ring et al. (35) demonstrated that although IL-2 and IL-15 induce the same JAK-STAT pathway, the signaling specificity of IL-2Rα and IL-15Rα as well as cis-trans signaling can determine cellular responsiveness. Both CCR6− and CCR6+ ILC3s respond to IL-2 stimulation in a dose-dependent manner, suggesting that IL-2 responsiveness is not restricted to one subset of ILC3s. Because IL-2–deficient mice spontaneously develop lethal autoimmune disease including colitis, we bred these mice onto the Rag2−/− background. Il2−/−Rag2−/− mice are healthy, suggesting that IL-2 does not spontaneously drive an innate cell-mediated colitis. In T and B cell–deficient mice, IL-2 is expressed by myeloid cells (36) and natural killer (NK) cells but not ILC3s (fig. S10), suggesting that IL-2 is not produced as an autocrine factor. NK cells are recruited late to the colon of C. rodentium–infected mice, and depletion of NK cells has been reported to be protective against lethal infection (37), suggesting that IL-2–producing NK cells could drive IL-22 production in ILC3s during a late phase of infection. To study the physiological relevance of IL-2 in ILCs, we infected Il2−/−Rag2−/− mice with C. rodentium and monitored their survival. Unlike IL-23R– or ILC3-deficient mice that die by day 10 after infection in our facilities, Il2−/−xRag2−/− mice survived significantly longer, suggesting that IL-2 is not required for differentiation of the ILC3 lineage but instead promotes the function of ILC3s during infection. We found that IL-23 activates both STAT3 and STAT5 in ILC3s, but not in RORγt+CD4+ T cells. Although the role of STAT5-inducing cytokines (e.g., IL-2, IL-7, and IL-15) for the development, proliferation, and survival of ILC subsets has been demonstrated (22), the functional role of IL-2 had yet to be determined. We found that, unlike cultured tonsil NK cells (38), IL-2 induces IL-22 production in intestinal ILC3s, suggesting a different requirement of STAT5 in the function of ILC subsets.
ILC3s isolated from T and B cell–deficient mice have a more activated phenotype and produce more IL-22 in vivo and ex vivo in response to IL-23 compared with ILC3s isolated from T and B cell–sufficient mice (19). We and others recently demonstrated that Tregs suppress the production of IL-23 by gut-resident macrophages and indirectly IL-22 production in ILC3s in vitro and in vivo (39, 40). In accordance with these studies, we showed that IL-2 and IL-23 equally activate STAT5 and STAT3, respectively, ex vivo in rested colon ILC3s isolated from C57BL/6 or Rag2−/− mice (fig. S9B). Moreover, ILC3s have a similar gene expression pattern (fig. S9C), suggesting that ILC3 responsiveness is not altered by the adaptive immune cells.
In addition to the expression of a T cell receptor (TCR), CD4+ T cells also express high levels of CD45 compared with ILC3s (fig. S1). CD45 is a tyrosine phosphatase that negatively regulates STAT signaling in hematopoietic cells without any particular specificity for individual STAT proteins (41). Because IL-23 is capable of activating both STAT3 and STAT5 signaling, we hypothesize that CD45 could affect IL-23R–mediated STAT activation. In accordance with this hypothesis, a recent study suggested that IL-23 induces the activation of STAT4 only in NKp46+ ILC3s. Contrary to our finding, IL-23–induced STAT4 was not required for IL-22 production by ILC3s, but instead drives interferon-γ (IFN-γ) production (42).
STAT protein complex formation after cytokine stimulation has been reported in various subsets of immune and nonimmune cells (2, 43). The IL-23 receptor is a heterodimer composed of IL-23R (which binds IL-23p19) and IL-12Rβ1 (which binds IL-23p40). Whereas the IL-23R subunit is unique to the IL-23 receptor, the IL-12Rβ1 chain can also heterodimerize with IL-12Rβ2 to form the functional IL-12 receptor. Because of its structural characteristics, IL-23 binding to the IL-23R has been reported to activate a STAT3-STAT4 heterodimer in pathogenic T cells (43). STAT5 activation by IL-23 has been reported in the Kit225 cell line (2), but the physiological relevance of IL-23–induced STAT5 in T cells remains unexplored. In our study, we demonstrated that IL-23–induced STAT3-STAT5 complexes sustain IL-22 production by intestinal ILC3s and protect the host from C. rodentium lethal infection. Our findings demonstrating collaborative function between STAT3 and STAT5 in innate immune cells could open up new avenues of research for lymphocyte STAT biology.
Analysis of the IL-22 promoter revealed three distinct STAT binding sites (identified by the sequence TTCNNNGAA) located near AhR DNA binding elements. SRE2 and SRE3 are less than 10 nucleotides away from the XRE2 and XRE3, respectively. This close proximity and the fact that AhR has been shown to transactivate IL-22 promoter activity in ILC3s suggest that STATs and AhR could bind to the IL-22 promoter as a single complex. We show that STAT3 and STAT5 preferentially bind to the first STAT binding site, which is separated from the AhR binding site by at least 130 nucleotides. This suggests that either STAT3 and STAT5 binding is independent of the presence of AhR or STAT3 and STAT5 binding to the SRE1 could recruit AhR to the IL-22 promoter as demonstrated in T cells (44). Further studies are required to determine the composition of the complex binding the IL-22 promoter upon cytokine stimulation.
JAK-STAT signaling is often up-regulated in IBD and cancer. Overactivation of JAK-STAT signaling is associated with disease progression, making the JAK-STAT pathway an attractive therapeutic target. However, tyrosine kinase inhibitors such as tofacinitib (a JAK2 inhibitor) and sunitinib (an inhibitor of multiple receptor tyrosine kinases) have shown mixed results in clinical trials. Whereas tofacinitib might be a good treatment strategy when used in addition to anti–tumor necrosis factor-α (TNF-α) therapy in patients with ulcerative colitis (45), it lacks therapeutic efficacy for Crohn’s disease (46). Common side effects of JAK-STAT inhibitors are diarrhea and occasional exacerbation of Crohn’s disease (47). Because of the differential roles of STAT3 and STAT5 that we have demonstrated, and the detrimental effects seen with inhibitors of multiple receptor tyrosine kinases, our work suggests that the development of more specific inhibitors for specific JAK or STAT proteins could help ameliorate unfavorable side effects in patients with IBD.
MATERIALS AND METHODS
Study design
The objective of this study was to determine the role of STAT5a/b in RORγt-expressing cells—mainly T cells and ILC3s—during C. rodentium–induced colitis. We designed and performed experiments using multiple techniques including cellular immunology, microbiology, protein biochemistry, molecular biology, and flow cytometry. The sample size and number of independent experiments are indicated in each of the figure legends.
Mice
C57BL/6, Il2−/−, and STAT3fl/fl mice were obtained from The Jackson Laboratory. Cd4-Cre STAT5fl/fl mice and BM from Stat5a−/− or Stat5b−/− mice (48) were obtained from the laboratory of J. O’Shea. Rag2−/− and Rag2−/−Il2ry−/− mice were obtained from Taconic. RORγtgfp/wt (also referred to as Rorc-eGFP), Rorc-CreStat3fl/fl, and Rorc-Cre Stat5a/bfl/fl mice were maintained under specific pathogen–free conditions and kept in microisolators with filtered air at the Merck Research Laboratories animal facility in Palo Alto. Mice used in this study (Table 1) were maintained on a C57BL/6 background. All animal procedures were approved by the Institutional Animal Care and Use Committee of Merck Research Laboratories in accordance with guidelines of the Association for Assessment and Accreditation of Laboratory Animal Care.
| Formal nomenclature | Name used in this study | Utility | Description |
| Rorc-CreStat5fl/fl | Stat5 ΔT,ILC3 | Conditional gene KO | Deletion of STAT5a/b in T cells and ILC3s |
| Rorc-CreStat3fl/fl | Stat3 ΔT,ILC3 | Conditional gene KO | Deletion of STAT3 in T cells and ILC3s |
| CD4-CreStat5fl/fl | Stat5 ΔT | Conditional gene KO | Deletion of STAT5a/b only in T cells |
| Transfer BM STAT5a−/− into Il2rγ−/−Rag2−/− mice | STAT5a−/− | BM cells transferred are gene KO | STAT5a−/−-deficient immune cells |
| Transfer BM STAT5b−/− into Il2rγ−/−Rag2−/− mice | STAT5b−/− | BM cells transferred are gene KO | STAT5b−/−-deficient immune cells |
| Il2−/−Rag2−/− | Il2−/−Rag2−/− | Global gene KO | IL-2–deficient mice with no T/B cells |
| Rorc-eGFP | Rorc-eGFP | Gene expression reporter | RORγt reporter mice |
| Rorc-eGFP Rag2−/− | Rorc-eGFP Rag2−/− | Gene expression reporter | RORγt reporter mice with no T/B cells |
Human samples
Human normal adjacent colon biopsies were obtained from Tissue Solutions. Human blood samples were obtained from normal healthy human volunteers participating in the Stanford Medical School Blood Center blood donation program. Permission to perform this investigation was granted by the Ethics Committee, and written informed donor consent was obtained.
BM transfer
BM cells were isolated from Stat5a−/−, Stat5b−/−, or Stat5a+/+Stat5b+/+ control mice. Cells (2 × 106) were injected intravenously into irradiated (8 Gy) Rag2−/−Il2ry−/− mice. For mixed BM chimera experiments, 2 × 106 BM cells from Thy1.2 Stat5a−/− or Thy1.2 Stat5b−/− and Thy1.1 C57BL/6 mice were injected intravenously into irradiated (8 Gy) Rag2−/−Il2ry−/− mice. Mice were analyzed 5 to 10 weeks after BM transfer.
C. rodentium infection
Mice were infected by oral gavage with 109 colony-forming units (CFUs) C. rodentium strain DBS100 (American Type Culture Collection). Six days after infection, feces and livers were harvested, weighed, and plated on a MacConkey agar (Sigma-Aldrich). One day before infection, mice were injected intravenously with 8 μg of IL-22 plasmid or plasmid control (System Biosciences) as previously described (49).
Colon and small intestine lamina propria cell isolation
Colon and small intestine lamina propria cells were isolated by first removing epithelial cells through the incubation of 0.5-cm-long gut tissue segments in Hank’s buffered salt solution containing 5 mM EDTA and 10 mM Hepes for 20 min at 37°C and then repeating this incubation one additional time. The remaining tissue was cut into small fragments and then digested with complete RPMI medium containing Liberase (0.250 mg/ml) (Roche), deoxyribonuclease (DNase) I (30 U/ml) (Sigma-Aldrich), and dispase (Corning) at the same conditions. The resulting cell suspension was layered onto a 40/80% Percoll gradient and centrifuged for 10 min at 600g; lamina propria cells were recovered at the interface.
Flow cytometry and antibodies
Cells were resuspended in phosphate-buffered saline and stained on ice for 30 min in the dark with a fixable viability stain (BD Biosciences). Then, cells were resuspended into Stain Buffer (FBS) (BD Biosciences) or Annexin V Binding Buffer (BioLegend; used only for Annexin V staining) and stained on ice for 30 min with various combinations of directly fluorochrome-conjugated antibodies. For intracellular antigens, surface-stained cells were permeabilized, fixed with Foxp3 Staining Buffer Set for 30 min on ice, and then stained with specific antibodies. The target antigens and clone designation(s) for the antibodies used to stain mouse cells were as follows: CD45 (30-F11), CD90.1 (OX-7), CD90.2 (53-2.1 or 30-H12), CD3 (17A2 or 145-2C11), TCRβ (H57-597), TCRγδ (GL3), CD4 (RM4-5 or GK1.5), CD25 (PC61), NK1.1 (PK136), CD11c (HL3), CD11b (M1/70), CD19 (1D3), RORγt (B2D), IL-22 (118PWSR or IL22JOP), NKp46 (29A1.4), and CCR6 (140706). Human cells were stained with the following antibodies: CD3 (UCHT1), CD5 (UCHT2), CD19 (HIB19), CD11c (BU15), CD11b (ICRF44), CD14 (M5E2), CD56 (HCD56), IL-22 (2G12A41), IL-17A (SCPL1362), CD4 (RPA-T4), CD25 (2A3), CD127 (A019D5), c-kit (104D2), and CRTH2 (BM16). All of the antibodies were purchased from BD Biosciences, BioLegend, or eBioscience. For all samples, acquisition was performed on an LSR II flow cytometer (BD Biosciences). Data were analyzed using FlowJo software.
When cytokine production was measured by flow cytometry, cells were stimulated with ionomycin (500 ng/ml), PMA (phorbol 12-myristate 13-acetate) (50 ng/ml) (Sigma-Aldrich), and IL-2 (50 ng/ml) (R&D Systems) or IL-23 (50 ng/ml) (R&D Systems and Merck) as indicated. After 1 hour, brefeldin A (BD Biosciences) was added for another 3 hours before staining. For STAT5 inhibition, cells were treated with 10 μM of STAT5 inhibitor II IQDMA (Sigma-Aldrich) for 1 hour before cytokine stimulation.
For STAT3 (4/P-STAT3) or STAT5 [47/Stat5(pY694)] phosphorylation, cells were rested for 2 hours and then stimulated with cytokines (50 ng/ml) (R&D Systems and Merck) for 15 min, then fixed with prewarmed BD Fix Buffer I (BD Biosciences) for 10 min, permeabilized with prechilled BD Perm Buffer III (BD Biosciences) for 30 min, and finally stained for 30 min on ice. Acquisition was performed on either an LSR II flow cytometer (BD Biosciences) or an Amnis imaging flow cytometer (Millipore). Data were analyzed using FlowJo software (BD Biosciences) or Ideas software (only for the Amnis imaging data).
Histology and pathologic scoring of colon tissue
Colon strips were fixed in 10% neutral buffered formalin overnight, transferred to 70% ethanol, processed routinely, embedded in paraffin, sectioned at 4 to 5 μm, and then stained with hematoxylin and eosin (H&E). Tissues were scored for the severity of disease by a pathologist in a blinded fashion according to three criteria. Inflammation: When present was characterized by infiltration with large numbers (60 to 70%) of mononuclear cells (macrophages and lymphocytes) and 30 to 40% of neutrophils and band cells. The scoring of inflammation includes severity of infiltration, loss of glands, erosion, dilatation of glandular lumina, presence of crypt abscesses, and degeneration of epithelial cells. Inflammation was scored on a scale of 0 to 4: 0, negative; 1, minimum; 2, mild; 3, moderate; 4, severe. Apoptosis: The prevalence of apoptotic bodies was scored on a scale of 0 to 3: 0, negative; 1, low; 2, moderate; 3, high. Regeneration: Regenerative changes assessed include scoring of the prevalence of mitotic figures in the upper one-third of the mucosa, nuclear density (nuclear crowding) within individual glandular structures, and regularity of the surface epithelium.
Proximity ligation assay
Sorted intestinal ILC3s (lineage−CD45intCD90high) were stimulated with recombinant IL-2 or IL-23 (50 ng/ml) (R&D Systems or Merck) for 15 min. Cells were fixed for 10 min (BD Phosflow Fix buffer I, BD Biosciences) and then permeabilized with cold BD Phosflow Fix buffer III (BD Biosciences) for 30 min. Cells were centrifuged onto slides with Cytospin (Thermo Fisher Scientific). Rabbit anti-STAT3 (Sigma-Aldrich) and mouse anti-STAT5 (clone 89/Stat5, BD Biosciences) antibodies were used for the proximity ligation assay according to the manufacturer’s protocol (Sigma-Aldrich). Stained slides were counterstained with DAPI (4′,6-diamidino-2-phenylindole) and coverslipped for analysis. Fluorescence was visualized on the EVOS FL Cell Imaging System (Life Technologies). Average number of STAT3/STAT5 interactions per nucleus was calculated on at least 20 cells per treatment per experiment.
Chromatin immunoprecipitation–quantitative polymerase chain reaction
Sorted intestinal ILC3s (lineage−CD45intCD90highRORγt-eGFP+) were stimulated with recombinant IL-2 or IL-23 (50 ng/ml) (R&D Systems or Merck) for 60 min. ChIP assay was performed using MAGnify Chromatin Immunoprecipitation System (Life Technologies) according to the manufacturer’s protocol. The sonication settings were as follows: 30% amplitude; two cycles of 8 min with 30 s on and 30 s off; 2-min break in between cycles. ChIP assays were performed with 5 μg of rabbit anti-STAT3 (Cell Signaling), rabbit anti-STAT5 (R&D Systems), or rabbit immunoglobulin G (IgG) control (R&D Systems) antibodies. After DNA purification, protein binding to the IL-22 promoter (accession number NC_000076.6) was measured by qPCR using the following specific primers: SRE 1, 5′-ACGGGAGATCAAAGGCTGCT-3′ (forward) and 5′-CCCTAAAACGTCACGGTGAGG-3′ (reverse); SRE 2, 5′-AAGGTGGGAAGGCTTGGAGG-3′ (forward) and 5′-ACACCGGGTTTTTTTGGTATTGTG-3′ (reverse); SRE 3, 5′-CAAGGAGGTACAGCTGCATC-3′ (forward) and 5′-AGACCTGCGGAGATAAAGCG-3′ (reverse); RORE, 5′-CAAGACTCCCCACACTGC-3′ (forward) and 5′-GAGGGACACTGGAGAAAGTG-3′ (reverse); XRE 1, 5′-ACAGTGATTTTCATGACTTCGCGTTCT-3′ (forward) and 5′-TCCCAGATAGCACCTGACAACTAGACT-3′ (reverse); LTbR Pro, 5′-CAGTGGCTCCAAGTGGCTTG-3′ (forward) and 5′-GCAAACCGTGTCTTGGCTGC-3′ (reverse). The primers for lymphotoxin β receptor (LTbR) were used as a negative control.
Enzyme-linked immunosorbent assay
Mouse IL-22 levels in cell culture supernatants or serum were measured by enzyme-linked immunosorbent assay (ELISA) (Quantikine kits) from R&D Systems according to the manufacturer’s protocol. Cleared supernatant was collected and stored at −80°C until ELISA analysis.
Total RNA isolation and subsequent gene expression analysis
For real-time PCR analysis, total RNA was isolated by either of two methods. Organs were homogenized in RNA STAT-60 (Tel-Test Inc.) with a Polytron homogenizer, and then RNA extraction was performed with the MagMAX-96 for Microarrays Kit (Thermo Fisher Scientific) per the manufacturer’s instructions. For cellular samples, RNA was isolated using the Arcturus PicoPure RNA Isolation Kit per the manufacturer’s instructions (Thermo Fisher Scientific).
DNase-treated total RNA was reverse-transcribed using QuantiTect Reverse Transcription (Qiagen) per the manufacturer’s instructions. Primers were obtained commercially from Thermo Fisher Scientific. Gene-specific preamplification was done on at least 2 ng of complementary DNA (cDNA) per Fluidigm Biomark manufacturer’s instructions (Fluidigm). Real-time qPCR was then done on the Fluidigm Biomark platform using two unlabeled primers at 900 nM each and 250 nM of FAM-labeled probe (Thermo Fisher Scientific) with TaqMan Universal PCR Master Mix containing UNG. Samples and primers were run on either a 48.48 array or a 96.96 array per the manufacturer’s instructions (Fluidigm). Ubiquitin levels were measured in a separate reaction and used to normalize the data by the ΔCt method. Using the mean cycle threshold value for ubiquitin and the gene of interest for each sample, the equation 1.8(Ct ubiquitin − Ct gene of interest) × 104 was used to obtain the normalized values. Primer reference sequences are available upon request.
Statistical analysis
The Mantel-Cox test was used to calculate statistical significance for survival studies. Two-tailed paired and unpaired t tests were used to calculate statistical significance in the rest of the study. *P < 0.05, **P < 0.01, and ***P < 0.001. Statistical analysis was performed using GraphPad Prism 7 software.
SUPPLEMENTARY MATERIALS
immunology.sciencemag.org/cgi/content/full/5/46/eaav1080/DC1
Fig. S1. STAT3 signaling in RORγt-expressing cells required for resistance to C. rodentium–induced colitis.
Fig. S2. Gating strategy of both mouse and human intestinal ILC3s.
Fig. S3. STAT5a and STAT5b deficiency enhances IL-17A, but not IL-22, production in T cells upon C. rodentium infection.
Fig. S4. Role of STAT5a and STAT5b in the development of intestinal ILC3s.
Fig. S5. STAT3 and STAT5 phosphorylation in ILC3s and RORγt+ CD4+ T cells.
Fig. S6. Similar activation of STAT3 and STAT5 in ILC3s isolated from WT or Rag2−/− mice.
Fig. S7. IL-2 drives IL-22 production in intestinal ILC3s.
Fig. S8. IL-22–producing ILC3s from human blood.
Fig. S9. IL-22–producing ILC3s in the absence of STAT3.
Fig. S10. NK cells are a potent source of IL-2 in the colon.
Table S1. Raw data (in Excel spreadsheet).
This is an article distributed under the terms of the Science Journals Default License.
REFERENCES AND NOTES
- Copyright © 2020 The Authors, some rights reserved; exclusive licensee American Association for the Advancement of Science. No claim to original U.S. Government Works
