Nod/Ripk2 signaling in dendritic cells activates IL-17A-secreting innate lymphoid cells and drives colitis in T-bet-/-.Rag2-/- (TRUC) mice

Abstract

T-bet(-/-).Rag2(-/-) (TRUC) mice spontaneously develop microbiota-driven, TNF-mediated large bowel inflammation that resembles human ulcerative colitis. We show here that IL-23 and IL-1-dependent secretion of IL-17A by innate lymphoid cells (ILCs; defined as CD45(+)lin(-)Thy1(hi)NKp46(-)) is a second critical pathway in this model. Using an in vitro coculture system of bone marrow-derived dendritic cells (DCs) and freshly isolated FACS-purified ILCs, we demonstrate that IL-23 and IL-1 secreted by DCs in response to microbial stimulation work together to induce IL-17A production by ILCs. TNF is not required for IL-17A secretion by ILCs in vitro but synergizes with IL-17A to induce the expression of neutrophil-attracting chemokines. Upstream, activation of the IL-23/IL-17A axis is regulated by nucleotide-binding oligomerization domain containing (Nod)/receptor-interacting serine-threonine kinase 2 (Ripk2) signals in DCs. Genetic ablation of the Nod/Ripk2 signaling pathway protects TRUC mice from developing colitis without affecting the colitogenicity of the intestinal microbiota. Our data provide insight into the complex network of interactions between IL-17A-secreting ILCs and other components of the innate immune system in the development of colitis.

Fig. 1.

TRUC disease is mediated by IL-23R–positive IL-17A–secreting ILCs. (A) TRUC.IL-23R^GFP/+ (IL-23R+/−) and TRUC.IL-23R^GFP/GFP (IL-23R−/−) mice (n = 30) were killed at 8 wk of age. Standard H&E sections of the colon were scored in a blinded fashion. Histopathology scores for individual mice are shown with horizontal lines representing the group mean. (B) BALB/c TRUC mice were treated from weaning with weekly i.p. injections of isotype control or antibodies blocking IL-23p19 (a-p19) or the IL-23 receptor (a-IL-23R). Histopathology was analyzed at 8 wk of age. Results were pooled from two experiments with n = 9–10 mice per group. (C) IL-23R expression on colon LP cells was analyzed by FACS in an 8-wk-old TRUC.IL-23R^GFP/+ reporter mouse. The majority of GFP⁺ cells were Thy1^hiSca1^hi and lineage-negative (CD11b⁻CD11c⁻). The Inset in the FACS plot on the right shows cells in the GFP⁻ gate for comparison. The experiment was performed three times with identical results. BALB/c TRUC mice were treated from weaning with weekly i.p. injections of a depleting anti-Thy1 antibody or isotype control (D) or a neutralizing antibody against IL-17A or isotype control (E). Histopathology was analyzed at 8 wk of age. Results were pooled from two experiments with n = 10 mice per group.

Fig. 2.

IL-17A secretion by ILCs requires IL-23 and IL-1 signals. (A) CD11c⁺ BM-DCs and FACS-sorted ILCs (CD45⁺lin⁻Thy1^hiNKp46⁻) pooled from mLN and colon LP of Rag2^−/− mice were stimulated either alone or in coculture with 1 µg/mL PGN. SN was harvested after 24 h, and the concentration of IL-17A and TNF was determined by ELISA. (B) BM-DCs and ILCs were stimulated with 1 µg/mL PGN for 24 h in the presence of blocking antibodies against IL-23R, IL-1R1, TNF, TNFR1, or isotype control (all 10 µg/mL). (C) ILCs were incubated alone for 24 h in the presence of recombinant murine IL-23, IL-1α, TNF, IL-23 + IL-1α, or IL-23 + TNF (all 10 ng/mL). (D) mLN cells from TRUC mice were stimulated for 24 h with IL-23, TNF, or IL-23 + TNF (all 10 ng/mL) together with anti–IL-1R1 or isotype control (10 µg/mL). In B–D, IL-17A was measured in the SN by ELISA. All experiments were performed at least twice with similar results. (E) TRUC.IL1R1^−/− mice and TRUC.IL1R^+/+ littermate controls were generated by mating TRUC.IL1R1^+/− parents. Colon histopathology (n = 9 per genotype) was scored at 8 wk of age.

Fig. 3.

TNF and IL-17A synergistically induce the expression of neutrophil-recruiting chemokines. (A) TRUC mice were treated from weaning with biweekly injections of a depleting anti-GR1 antibody or isotype control (n = 8 per group). Histopathology was analyzed at 8 wk of age. (B) MODE-K cells were stimulated with recombinant murine TNF, IL-17A, or TNF + IL-17A (all 10 ng/mL) for 6 h. The expression of Cxcl1, Cxcl2, and Cxcl5 was determined by real-time qPCR using Hprt for normalization. Data represent fold change relative to unstimulated cells. Mean and SD of biological duplicates are shown. The experiment was performed twice with similar results. (C) Four-week-old TRUC mice received a single i.p. injection of a blocking antibody against TNF (a-TNF) or the IL-23 receptor (a-IL-23R) or isotype control (n = 6–7 per group). Mice were killed after 3 d, and RNA was isolated from the distal third of the colon. Cxcl1, Cxcl2, and Cxcl5 mRNA levels were determined by real-time qPCR and normalized to Hprt. Values are expressed relative to isotype-treated TRUC mice.

Fig. 4.

TRUC.Ripk2^−/− mice are protected from colitis and Nod/Ripk2 stimulated DCs drive IL-17A secretion by ILCs. (A) TRUC.Ripk2^+/− mice backcrossed to BALB/c for at least seven generations were mated to generate TRUC.Ripk2^−/− and TRUC.Ripk2^+/+ littermates. (B) Crossfostering of TRUC.Ripk2^+/+ and TRUC.Ripk2^−/− pups. Breeders were set up in parallel and newborn mice were exchanged between cages within 24 h after birth. Colon histopathology in A and B was analyzed at 8 wk of age. Results are pooled from multiple litters (n = 8–13 animals per group). (C and D) BM-DCs and FACS sorted ILCs were prepared from TRUC.Ripk2^+/+ and TRUC.Ripk2^−/− mice and stimulated for 24 h with 1 ng/mL LPS alone or LPS plus 1 μg/mL L18-MDP. The concentration of TNF in the SN of DCs cultured alone (C) and IL-17A in the SN of cocultures of DCs and ILCs (D) was determined by ELISA. (E) BM-DCs from TRUC.Ripk^+/+ and TRUC.Ripk2^−/− mice were stimulated for 6 h with 1 ng/mL LPS alone or LPS plus 1 μg/mL L18-MDP. Total RNA was isolated and real-time qPCR analysis for IL-12p40 and IL-23p19 expression was performed. The data were first normalized to Hprt and then scaled by dividing over the mean of the unstimulated samples. All in vitro assays described in this figure were performed twice with similar results.

Fig. 5.

Nod1 and Nod2 signals are redundant for Ripk2-dependent induction of TRUC colitis. (A) TRUC.Nod1^+/− mice were mated to generate TRUC.Nod1^+/+, TRUC.Nod1^+/−, and TRUC.Nod1^−/− littermates. The same approach was used to generate TRUC.Nod2^−/− (B) and TRUC.Ripk2^−/− (C) animals. (D) After “fixing” the Nod2^−/− genotype, TRUC.Nod1^+/−.Nod2^−/− mice were mated to generate TRUC.Nod1^−/−.Nod2^−/− DKO mice and littermate controls as indicated in the figure. Histopathology (n = 8–12 per genotype) was scored at 8 wk of age.