STAT3 signaling from RORγt⁺ lymphocytes controls C. rodentium infection

IL-17 and IL-22 are known to be regulated by RORγt (encoded by Rorc) and are mainly produced by RORγt⁺ ILCs and T cells (Ivanov et al., 2006; Qiu et al., 2011). To determine whether Sunitinib’s adverse effect was due to STAT3 inhibition on ILC3s and T cells, Stat3-floxed mice were crossed with Rorc-cre transgenic mice to achieve specific deletion of STAT3 in RORγt-expressing cells (Rorc-cre-Stat3^fl/fl). Because RORγt is transiently expressed at high levels at the double positive stage of T cell development, Rorc-cre-Stat3^fl/fl mice not only lack STAT3 expression by RORγt⁺ ILC3s, but also most αβ T cells (Figure S2A–S2C).

We next tested whether STAT3 signaling is necessary for controlling the C. rodentium infection. After infection, Rorc-cre-Stat3^fl/fl mice rapidly lost body weight and died around day 10, whereas no weight loss or death was observed in their littermate control wild-type (Stat3^fl/fl) mice (Figure 2A and 2B) or Rorc-cre-Stat3^fl/+ heterozygous mice (Figure S2D). To test whether the increased morbidity and mortality were associated with bacterial colitis, we checked bacterial titer and diarrhea status. Rorc-cre-Stat3^fl/fl mice had 10–100 times higher bacterial titers in the feces compared to WT mice at day 5 and 8 post-infection (Figure 2C and S2E). Rorc-cre-Stat3^fl/fl mice also showed increased bacterial titers in the blood, liver and spleen (Figure 2D and S2E), suggesting systemic dissemination of C. rodentium. The Rorc-cre-Stat3^fl/fl mice exhibited severe diarrhea, colonic shortening, severe inflammation, and tissue injury in the colon and severely disruption of the epithelial layer compared with WT mice (Figure 2E, 2F and S2F). Together, these data indicate that STAT3 signaling in RORγt⁺ cells is essential for host defense against a mucosal bacterial pathogen.

Open in a separate window

Figure 2

STAT3 signaling is essential for control of intestinal C. rodentium infection

(A–F) 7 weeks old Rorc-cre-Stat3^fl/fl (n = 7) and their littermate wild type Stat3^fl/fl (n = 9) mice were orally inoculated with C. rodentium (2 × 10⁹ CFU) and body weight was measured at the indicated time points. Body weight change (A) and survival rates (B) are shown.

(C) Bacterial titers from fecal homogenate cultures at indicated day post infection in the mice in A, B.

(D) Bacterial titers from blood and spleen, liver homogenate cultures at indicated day post infection are shown.

(E) Rorc-cre-Stat3^fl/fl mice show colon shortening at day 8 post infection.

(F) Rorc-cre-Stat3^fl/fl mice show more inflammatory monocytes or neutrophils infiltration in the colon. LPLs were isolated from colon and gated in CD45⁺ cells.

*P<0.05, **P<0.01, ***P<0.001 (Student’s t-test). Data are representative of three independent experiments (A–C; mean ± s.e.m.) or two independent experiments (D–E; mean ± s.e.m.). See also Figure S2.

STAT3 signaling from RORγt⁺ lymphocytes controls both IL-22 and IL-17 production during C. rodentium infection

We have shown that tyrosine kinase (including STAT3 pathway) inhibition suppressed both IL-22 and IL-17 expression after infection, which are required for the protection against C. rodentium infection (Ishigame et al., 2009; Ivanov et al., 2009; Wang et al., 2010; Zheng et al., 2008). We also measured IL-22 and IL-17 amounts in the colon of WT and Rorc-cre-Stat3^fl/fl mice early after C. rodentium infection. Compared to WT mice, expression of IL-22 and IL-17 was significantly reduced in the colon of Rorc-cre-Stat3^fl/fl mice on day 5 post infection (Figure 3A). RegIIIγ and RegIIIβ, two antimicrobial proteins dependent on IL-22 (Pickert et al., 2009; Zheng et al., 2008), were also reduced in the colon of Rorc-cre-Stat3^fl/fl mice after infection (Figure 3A). These results suggest that the STAT3 signaling is essential for IL-22 and IL-17 production in the gut after mucosal bacterial infection.

Open in a separate window

Figure 3

IL-22, but not IL-17, is an essential and sufficient pathway downstream of STAT3 signaling to protect mice against C. rodentium infection

(A) Rorc-cre-Stat3^fl/fl and wild type Stat3^fl/fl littermate mice (n = 5 per group) were infected with 2 × 10⁹ CFU of C. rodentium. The mRNA expression of IL-22, IL-17A, IL-17F and IL-22- dependent RegIIIγ and RegIIIβ antimicrobial proteins in the colon of naïve or 5 days infected mice were measured by real-time. *P<0.05, **P<0.01, ***P<0.001 (Student’s t-test). Data are representative of two independent experiments (A; mean ± s.e.m.).

(B–D) Stat3^fl/fl (n = 5) and Rorc-cre-Stat3^fl/fl mice were orally inoculated with 2 × 10⁹ CFU of C. rodentium. 10 μg control vector (n = 5), IL-17 (n = 5), IL-22 (n = 5), or both IL-17 and IL-22 (n = 4) expressing plasmids were hydrodynamic injected into Rorc-cre-Stat3^fl/fl mice through the tail vein 6 hr after C. rodentium infection. Body weight changes (B) and survival rates (C) were monitored at indicated time points. (D) CFU counts of C. rodentium in the fecal pellets and blood on day 5 after infection are shown. Data are representative of two independent experiments (mean ± s.e.m.). *P<0.05, ***P<0.001; ns, no significant difference (Student’s t-test). nd, nondetectable.

(E) STAT3 signaling is required for both innate and adaptive IL-22 production. IL-22 expression in CD3⁻ and CD3⁺ cells were analyzed by intracellular cytokine staining followed by flow cytometry. Rorc-cre-Stat3^fl/fl and wild type Stat3^fl/fl littermate mice were infected with 2 × 10⁹ CFU of C. rodentium. Intestinal LPLs were isolated from the colons of Stat3^fl/fl or Rorc-cre-Stat3^fl/fl mice at day 4 and day 8 post-infection. Cells were stimulated with IL-23 (25 ng/ml), PMA (50 ng/ml) and ionomycin (750 ng/ml) for 4 hours and gated in Thy1⁺ lymphocytes. Naïve mice were used as control. Data are representative of three independent experiments. See also Figure S3.

Exogenous IL-22 but not IL-17 rescues STAT3 deficiency mice during C. rodentium infection

Because both IL-17 and IL-22 are controlled by STAT3 signaling, we wondered whether administration of exogenous IL-17 or IL-22 is sufficient to rescue the death of Rorc-cre-Stat3^fl/fl mice from C. rodentium infection. To address this question, IL-17A and IL-22 expressing plasmids were hydrodynamically injected into Rorc-cre-Stat3^fl/fl mice at the same day of C. rodentium infection. All of the Rorc-cre-Stat3^fl/fl mice treated with IL-22 survived after infection and lost weight only slightly, whereas vector and IL-17 alone treated mice rapidly lost body weight and succumbed to C. rodentium infection (Figure 3B and 3C). Accordingly, IL-22 treated Rorc-cre-Stat3^fl/fl mice had reduced bacterial counts in their feces and blood compared to vector and IL-17 alone treated mice (Figure 3D). Collectively, these data demonstrate that exogenous IL-22 but not IL-17 is able to rescue Rorc-cre-Stat3^fl/fl mice from a lethal C. rodentium infection and IL-22 is an essential and sufficient pathway downstream of STAT3 signaling to protect mice against mucosal bacterial pathogens.

STAT3 signaling is required for both innate and adaptive IL-22 production

Previous studies have shown that C. rodentium infection induces early wave of IL-22 production by ILC3s and second wave by CD4⁺ T cells (Th22) (Basu et al., 2012; Qiu et al., 2011; Tumanov et al., 2011). To test whether STAT3 control innate or adaptive IL-22 production, we examined cytokine expression in the intestinal lamina propria lymphocytes (LPLs) isolated from both naïve and infected Rorc-cre-Stat3^fl/fl and wild type mice by intracellular cytokine staining. Consistent with the previous study (Basu et al., 2012; Qiu et al., 2011), CD3⁻ innate LPLs were the main IL-22 producers compared to CD3⁺ T cells in naïve and infected WT mice at an early stage of infection (day 4), while there were more IL-22 producing T cells than ILCs at a later stage of infection (day 8) in WT mice (Figure 3E). However, Rorc-cre-Stat3^fl/fl mice had severely decreased expression of IL-22 by intestinal LPLs, regardless of CD3⁻ innate cells or CD3⁺ T cells (Figure 3E, S3A and S3B). Moreover, Sunitinib treatment also suppressed IL-22 production in innate LPLs from wild type mice (Figure S3E). In a similar manner as IL-22, the IL-17 production was also significantly reduced in both innate and adaptive lymphocytes (Figure S3C and S3D). Since PMA and ionomycin are not physiological stimuli, the LPLs were also stimulated with anti-CD3 and anti-CD28 to mimic TCR signaling. Similar to PMA and ionomycin treatment, STAT3 deficient T cells also showed little IL-22 and IL-17 production after anti-CD3 and anti-CD28 stimulation (Figure S3F). Together, our data indicate that STAT3 signaling is essential for IL-22 and IL-17 expression in both ILC3s and T cells.

STAT3 signaling and IL-22 from ILC3s but not from T cells are essential for the control of C. rodentium infection

Because STAT3 signaling regulates IL-22 expression in both innate and adaptive lymphocytes, we wondered whether innate or adaptive IL-22 is more important for the protection from C. rodentium infection. To test whether the STAT3 signaling from ILC3s is sufficient for the protection against C. rodentium infection in the transient absence of T cells, CD4 mAb was administered to infected Rorc-cre-Stat3^fl/fl and WT mice to deplete the CD4⁺ T cells. To avoid depletion of CD4⁺ LTi cells, we used low dose of CD4 mAb to transiently remove CD4⁺ T cells but not CD4⁺ ILC3s. As shown in Figure S4A, 90% of peripheral CD4⁺ T cells were depleted, while there was no proportional reduction of ILC3s in the lamina propria after multiple CD4 mAb injection. Because CD4⁺ T cells are required for host adaptive immunity especially for antibody responses against C. rodentium (Bry et al., 2006), depletion of CD4⁺ T cells provided another model to study whether the innate immune response is sufficient against this pathogen in the absence of adaptive responses. Treatment with CD4 mAb resulted in reduction of C. rodentium specific IgG in WT mice and all of them died at 3 weeks after C. rodentium infection (Figure S4B, 4A and 4B), which is the same as Rag1^−/− mice, suggesting that adaptive immunity is responsible for the late phase of protection. Treatment of Rorc-cre-Stat3^fl/fl mice with additional CD4 mAb did not aggravate or ameliorate the C. rodentium infection compared with Rat IgG (control for CD4 mAb) treatment group. All of the Rorc-cre-Stat3^fl/fl mice with or without depletion of CD4⁺ T cells still died on day 8 post-infection (Figure 4A–4C). These data suggest that STAT3 signaling and IL-22 from CD4⁺ T cells are not essential for protecting mice in the early phase of C. rodentium infection.

Open in a separate window

Figure 4

STAT3 signaling and IL-22 from RORγt⁺ ILCs, but not T cells, are essential for the protection from C. rodentium infection

(A–C) Rorc-cre-Stat3^fl/fl and wild type Stat3^fl/fl mice were infected with 5 × 10⁶ CFU of C. rodentium and injected intraperitoneally with mAb GK1.5 or Rat IgG (50 μg per mouse each time) at day −5, 0, 5, 10 and 15 post infection for depletion of CD4⁺ T cells (4 to 6 mice per group). Average body weight change (A) and survival rates (B) at the indicated time points are shown. (C) Bacterial titers from blood and fecal homogenate cultures at day 5 post infection. Each dot represents one individual mouse.

(D, E) Cd4-cre-Stat3^fl/fl (n=6) and wild type Stat3^fl/fl mice (n=6) were orally infected with 2 × 10⁹ CFU of C. rodentium. (D) Body weight changes are shown at the indicated time points. (E) Bacterial titers from fecal homogenate cultures at indicated day post infection in the mice in D.

(F, G) IL-22 expression in CD3⁻ and CD3⁺ cells were analyzed by intracellular cytokine staining followed by flow cytometry. Cd4-cre-Stat3^fl/fl and wild type Stat3^fl/fl littermate mice were infected with 2 × 10⁹ CFU of C. rodentium. LPLs were isolated from the colons at day 8 post-infection. Cells were stimulated with IL-23 (25 ng/ml), PMA (50 ng/ml) and ionomycin (750 ng/ml) for 4 hours and gated in Thy1⁺ lymphocytes. Each dot represents one individual mouse.

*P<0.05, **P<0.01, ***P<0.001; ns, no significant difference (Student’s t-test). nd, nondetectable. Data are representative of two independent experiments (A–G; mean ± s.e.m.). See also Figure S4.

To further test whether STAT3 signaling in T cells is required for the later T and B cell response and prevention of chronic infection, we rescued Rorc-cre-Stat3^fl/fl mice by exogenous IL-22 and examined the antibody production. IL-22-rescued Rorc-cre-Stat3^fl/fl mice produced the same amount of C. rodentium specific IgG as WT mice (Figure S4C). Moreover, 60 days after first C. rodentium infection, IL-22-rescued Rorc-cre-Stat3^fl/fl mice were challenged with high dose of C. rodentium again. Interestingly, none of the IL-22-rescued Rorc-cre-Stat3^fl/fl mice showed diarrhea (data not shown) and weight loss (Figure S4D), suggesting IL-22-rescued Rorc-cre-Stat3^fl/fl mice developed normal protection. Together, these data demonstrate that STAT3 signaling from CD4⁺ T cells is not essential for normal IgG production and protecting mice after C. rodentium infection.

To further test whether the STAT3 signaling from RORγt⁺ ILCs is sufficient for the protection against C. rodentium infection, we crossed the Stat3-floxed mice with Cd4-cre mice to generate mice lacking STAT3 only in T cells (Cd4-cre-Stat3^fl/fl) but not in CD4⁺ RORγt⁺ ILCs (Eberl and Littman, 2004) (data not shown). Even after high dose of C. rodentium infection, the Cd4-cre-Stat3^fl/fl mice did not lose body weight and all survived (Figure 4D). Furthermore, the fecal titers of C. rodentium were not significant higher in Cd4-cre-Stat3^fl/fl mice than WT mice 5–11 days after infection (Figure 4E). Similar to the Rorc-cre-Stat3^fl/fl mice, Cd4-cre-Stat3^fl/fl mice also lacked IL-22 producing CD4⁺ T cells in the gut (Figure 4F and 4G). However, the IL-22 producing ILCs in Cd4-cre-Stat3^fl/fl mice were not reduced compared with WT mice, which were different with Rorc-cre-Stat3^fl/fl mice (Figure 3E, 4F and 4G). Furthermore, CD4 mAb was also administered to Cd4-cre-Stat3^fl/fl mice and Stat3^fl/fl mice. As shown in Figure S4E and S4F, both CD4 mAb treated Stat3^fl/fl or Cd4-cre-Stat3^fl/fl mice lost body weight and died around day 20 post infection, while untreated mice survived, suggesting that CD4⁺ T cells are important for the protection against C. rodentium infection, but STAT3 signaling from T cells is not essential. Together, these data demonstrate that STAT3 signaling and IL-22 from ILC3s, but not CD4⁺ T cells, is essential for protecting mice from death during C. rodentium infection.

CD90^hiCD45^lo ILC3s protect Rorc-cre-Stat3^fl/fl mice during C. rodentium infection

Because we have shown that STAT3 signaling from ILC3s but not T cells is essential for the protection from C. rodentium infection, we wonder whether adoptive transfer of ILC3s could rescue Rorc-cre-Stat3^fl/fl mice from infection. However, because of extremely low number of ILC3s and limited Rorc^gfp/+ mice, it was difficult to get enough ILC3s. To find better cell surface marker to identify ILC3s, we screened a large panel of surface markers and found ILC3s could be strictly identified by CD45 and CD90. As shown in Figure 5A, after gating with CD45 and CD90, CD90⁺ LPLs from the large intestine and small intestine of Rag1^−/− mice fell into three separate populations, CD90^hiCD45^lo, CD90^loCD45^int and CD90^loCD45^hi. Most of CD90^hiCD45^lo LPLs (96%) were RORγt⁺ ILC3s, and most of CD90^loCD45^int LPLs (95%) were NK1.1⁻RORγt⁻ ILCs, whereas nearly three fourths of CD90^loCD45^hi were NK1.1⁺ NK cells. Similar to Rag1^−/− mice, CD90⁺ LPLs from both large intestine and small intestine of C57BL/6 WT mice fell into two separate populations, CD90^hiCD45^lo and CD90^loCD45^hi. Most of CD90^hiCD45^lo LPLs (94%) were RORγt⁺ ILC3s, whereas CD90^loCD45^hi were CD3⁺ T cells and other RORγt⁻ ILCs (Figure S5). These data suggest that CD90^hiCD45^lo are specific cell surface markers for intestinal RORγt⁺ ILC3s.

Open in a separate window

Figure 5

CD90^hiCD45^lo ILC3s protect Rorc-cre-Stat3^fl/fl mice from C. rodentium infection

(A) Intestinal CD90^hiCD45^lo LPLs are RORγt⁺ ILCs. LPLs were isolated from both the large intestine and small intestine of Rag1^−/− mice and gated by CD45 and CD90 expression. Three separate populations, CD90^hiCD45^lo, CD90^loCD45^int and CD90^loCD45^hi were further analyzed with the expression of NK1.1 and RORγt. Data are representative of two independent experiments.

(B–D) Rag1^−/− mice were infected with 2 × 10⁹ CFU of C. rodentium and CD90^hiCD45^lo ILC3s were sorted from both the colon and small intestine LPLs at day 1 post infection. Sorted ILC3s were stimulated with IL-23 (25 ng/ml) for 1 hour, then were transferred by i.v. injection (2 × 10⁵ cells per mouse) into Rorc-cre-Stat3^fl/fl recipients at day 0 and day 3 post inoculated with 5 × 10⁶ CFU of C. rodentium. Untreated Rorc-cre-Stat3^fl/fl and wild type Stat3^fl/fl littermate mice were as control. Body weight changes (B) and survival rates (C) were monitored at indicated time points. (D) CFU counts of C. rodentium in the fecal pellets and blood on day 5 after infection are shown. Data are pooled from two independent experiments with nine to eleven mice per group (mean ± s.e.m.). ***P<0.001 (Student’s t-test). See also Figure S5.

Then we sorted CD90^hiCD45^lo ILC3s and transferred small number of them into Rorc-cre-Stat3^fl/fl mice with C. rodentium infection. Compared with untreated Rorc-cre-Stat3^fl/fl mice, ILC3s-transfered Rorc-cre-Stat3^fl/fl mice lost less body weight and about one third of the mice survived and recovered from C. rodentium infection (Figure 5B and 5C). Accordingly, ILC3s-transferred Rorc-cre-Stat3^fl/fl mice exhibited reduced bacterial counts in their feces and blood compared to untreated mice (Figure 5D). Collectively, these data demonstrate that CD90^hiCD45^lo ILC3s are sufficient to protect mice against C. rodentium infection.

The development of RORγt⁺ Th cells, but not RORγt⁺ ILC3s, requires STAT3 signaling

Because Rorc-cre-Stat3^fl/fl mice showed reduced IL-22 in both ILCs and T cells, we next determined how STAT3 regulates IL-22 production. IL-22 is mainly produced in RORγt⁺ cells, and RORγt directly regulates IL-22 production at the transcriptional level. Previous studies have shown that STAT3 regulates the development of RORγt⁺ Th17 cells (Ivanov et al., 2006; Laurence et al., 2007; Yang et al., 2007). To determine whether STAT3 signaling is shared by both innate and adaptive colonic RORγt⁺ cells for their development, we examined the RORγt expression in the intestinal LPLs isolated from Rorc-cre-Stat3^fl/fl mice and their littermate control wild-type mice. Consistent with Cd4-cre-Stat3^fl/fl mice (Figure S6A), both the percentage and number of RORγt⁺ Th cells (CD3⁺CD4⁺RORγt⁺) in the colon were much lower in Rorc-cre-Stat3^fl/fl mice compared with the WT mice (Figure 6A–6C and S6B). Unexpectedly, there were no differences in the percentages and numbers of the total RORγt⁺ ILC3s (Figure 6D–6F), the subsets of ILC3s (LTi, NCR⁺ ILC3 and NCR⁻ ILC3; Figure 6D and S6C) or other ILCs (ILC1, ILC2; Figure S6C) between Rorc-cre-Stat3^fl/fl and WT mice, suggesting that STAT3 was required for RORγt expression in T cells but not ILC3s or maintenance of ILC3s. Moreover, similar to naïve mice, there were no differences in percentages and total numbers of the ILC3s after C. rodentium infection between Rorc-cre-Stat3^fl/fl and WT mice (Figure 6E and 6F). Taken together, our data indicate that STAT3 signaling is essential for the development of RORγt⁺ Th cells, but not ILC3s, and that STAT3-controlled IL-22 production is not through regulating RORγt expression in innate cells, which also suggests that RORγt is not sufficient to induce IL-22 production in the absence of STAT3 signaling.

Open in a separate window

Figure 6

The development of RORγt⁺ T helper cells, but not RORγt⁺ ILCs is dependent on STAT3 signaling

(A–F) Colonic LPLs were isolated from Stat3^fl/fl and Rorc-cre-Stat3^fl/fl littermate mice before or 5 days after C. rodentium infection.

(A, D) RORγt and CD4 expression were analyzed in colonic LPLs from naïve mice by flow cytometry after gating on Thy1⁺ CD3⁺ adaptive lymphocytes and Thy1⁺ CD3⁻ innate lymphoid cells.

(B, C) Percentage of RORγt⁺ cells in the CD4⁺ CD3⁺ T cell population, as well as the absolute numbers of RORγt⁺ CD4⁺ T cells in the colons of naïve and C. rodentium infected Stat3^fl/fl and Rorc-cre-Stat3^fl/fl mice are shown.

(E, F) Percentage of RORγt⁺ cells in the CD3⁻ Thy1⁺ ILCs population, as well as the absolute numbers of RORγt⁺ ILCs in the colons of naïve and C. rodentium infected Stat3^fl/fl and Rorc-cre-Stat3^fl/fl mice are shown.

**P<0.01, ***P<0.001; ns, no significant difference (Student’s t-test). Data are representative of three independent experiments (mean ± s.e.m.). Each dot represents one individual mouse (B, C, E, F). See also Figure S6.

Infection with C. rodentium and Treatment

Mice were orally gavaged with C. rodentium strain DBS100 (ATCC 51459; American Type Culture Collection) and body weight, survival, colony-forming units (CFUs) counts, tissue histology were assessed as previously described (Tumanov et al., 2011; Wang et al., 2010). Colon culture was done as previously described (Zheng et al., 2008). Where indicated, Sunitinib (SU11248; SUTENT) or control was gavaged orally, once a day, at doses of 4 or 40 mg/Kg body weight after infection. Where indicated, mice were injected intraperitoneally with mouse IL-22-Ig fusion protein on 0, 2, 4 day after C. rodentium infection at a dose of 50 μg per mouse each time. Control groups received isotype control immunoglobulin. For depletion of CD4⁺ T lymphocytes, mice were injected intraperitoneally with mAb GK1.5 or Rat IgG on −5, 0, 5, 10, 15 day after C. rodentium infection at a dose of 50 μg per mouse each time. Where indicated, IL-17 and IL-22-expressing plasmids were introduced into mice using a hydrodynamic tail vein injection-based gene transfer technique (Tumanov et al., 2011).

Isolation of Intestinal LPLs

The isolation of intestinal lamina propria cells was done as previously described (Wang et al., 2010). In brief, mice were killed and colons were removed and placed in ice-cold PBS. The intestines were cut open longitudinally, thoroughly washed in ice-cold PBS, and cut into 1.5 cm pieces. Intestines were incubated with shaking in RPMI 1640 (Invitrogen) containing 3% FBS, 1mM DTT, 5 mM EDTA and 10 mM HEPES at 37°C for 30 min. After washing with RPMI 1640 containing 2 mM EDTA, the tissues were then digested in RPMI 1640 containing 0.05% DNase I (Sigma) and Liberase (Roche) (0.1 mg/ml) at 37°C for 30 min with slow rotation. The digested tissues were homogenized by vigorous shaking and passed through 100 μm cell strainer. Mononuclear cells were then harvested from the interphase of an 80% and 40% Percoll gradient after a spin at 2500 rpm for 20 min at room temperature.

Flow Cytometry, Antibodies and ELISA

Antibodies against CD3, CD4, CD8, CD45, CD90, Nkp46, Gr-1, CD11b, KLRG1, TCRβ, NK1.1 and Streptavidin-APC were purchased from BioLegend. Antibodies against RORγt and IL-17 were purchased from eBioscience. Antibodies against STAT3 and Rabbit IgG were purchased from Cell Signaling Technology. Anti-IL-22 antibody was a gift from Genentech. For nuclear staining, cells were fixed and permeabilized using a Transcription Factor Staining Buffer Set (eBioscience). For cytokine production, cells were stimulated ex vivo by IL-23 (25 ng/ml, R&D), PMA (50 ng/ml, Sigma) and ionomycin (750 ng/ml, Calbiochem) or anti-CD3 (plate coated, 1 μg/ml) and anti-CD28 (2 μg/ml) for 4 hr and Brefeldin A (10 μg/ml, eBioscience) was added 2 hr before cells were harvested for analysis. IC Fixation Buffer and Permeabilization Buffer were used according to the manufacturer’s instructions (eBioscience) for intracellular cytokine staining. Flow cytometry analysis was performed on FACSCanto (BD Biosciences) instruments and analyzed with FlowJo software (Tree Star Inc.). IL-22 (R&D Systems), IL-17, IL-23 (eBioscience) in supernatants were measured by ELISA according to the manufacturer’s recommendations.

Quantitative Real-Time RT-PCR

RNA from tissues was isolated with Trizol reagent (Invitrogen) and reverse transcribed with AMV Reverse Transcriptase (Promega). RNA from sorted cells was isolated with the RNeasy Micro Kit (Qiagen) and was reverse transcribed with Seniscript Reverse Transcription Kit (Qiagen). Real-time RT-PCR was performed using SSoFast EvaGreen supermix (Bio-Rad) and different primer sets (Table S1) on StepOne Plus (Applied Biosystems). Samples were normalized to HPRT and reported according to the 2^−ΔΔC_T method.

Adoptive Transfer Experiments

Rag1^−/− mice were infected with C. rodentium. CD90^hiCD45^lo ILC3s were sorted from both the colon and small intestine LPLs on a FACS Aria III instrument (BD Bioscience) at day 1 post infection. Sorted ILC3s were stimulated with IL-23 (25 ng/ml) for 1 hour, then were transferred by i.v. injection (2 × 10⁵ cells per mouse) into recipients inoculated with C. rodentium at day 0 and day 3.

Chromatin Immunoprecipitation

ChIP assays with EL4 cells were performed as previously described (Qiu et al., 2011). Briefly, 5×10⁷ EL4 cells were first fixed in 1% formaldehyde in RPMI1640 media for 10 min at room temperature. Then the reaction was stopped by adding glycine solution. Fixed cells were washed with PBS and resuspended in lysis buffer. Chromatin was sheared by sonication on ice, and the insoluble fraction was removed by centrifugation. The lysate was precleared with Protein A beads (Santa Cruz Biotechnology) and then incubated with anti-Flag beads (Sigma) at 4°C for 5 hrs. After washing, precipitated chromatin fragments were eluted with flag peptide (Sigma). Then the samples were reverse cross-linked at 65°C by adding Tris-EDTA with 1% SDS. After proteinase K digestion, DNA was extracted using Phenol-Chloroform, and precipitated for quantitative real-time PCR analyses using specific primers. CD90^hiCD45^lo ILC3s were sorted from Rag1^−/− mice and ChIP assays were performed with anti-STAT3 rabbit mAb (D3Z2G) or Rabbit IgG (Cell Signaling Technology). MegnaChIP protein A magnetic beads (Millipore) were used to isolate immune complexes. Crosslinks were reversed by heating at 65ting at 65rosslproteinase K treatment. DNA was purified with Qiaquick PCR purification kit (QIAGEN).

We are grateful to H. Yu (Beckman Research Institute of City of Hope, CA) for Stat3-floxed mice and Sunitinib (SU11248; SUTENT); D. Littman and I. Ivanov (New York University, NY) for Rorc-cre mice; W. Ouyang (Genentech, CA) for IL-22-expressing plasmid, IL-22-Ig and anti-IL-22 antibody; C. Dong (MD Anderson Cancer Center, TX) for IL-17 plasmid; V. Krishnamoorthy (University of Chicago, IL) for ChIP protocol; S. Kron and J. Cunningham (University of Chicago, IL) for sharing sonicator. The work was supported by the National Institutes of Health grants (AI 089954 and AI 091962 to L.Z., DK080736 and CA141975 to Y.X.F.), by a Pew Scholarship and a Cancer Research Institute Investigator Award (to L.Z).