Dysregulation of ILC3s unleashes progression and immunotherapy resistance in colon cancer

Abstract

Group 3 innate lymphoid cells (ILC3s) regulate immunity and inflammation, yet their role in cancer remains elusive. Here, we identify that colorectal cancer (CRC) manifests with altered ILC3s that are characterized by reduced frequencies, increased plasticity, and an imbalance with T cells. We evaluated the consequences of these changes in mice and determined that a dialog between ILC3s and T cells via major histocompatibility complex class II (MHCII) is necessary to support colonization with microbiota that subsequently induce type-1 immunity in the intestine and tumor microenvironment. As a result, mice lacking ILC3-specific MHCII develop invasive CRC and resistance to anti-PD-1 immunotherapy. Finally, humans with dysregulated intestinal ILC3s harbor microbiota that fail to induce type-1 immunity and immunotherapy responsiveness when transferred to mice. Collectively, these data define a protective role for ILC3s in cancer and indicate that their inherent disruption in CRC drives dysfunctional adaptive immunity, tumor progression, and immunotherapy resistance.

Keywords: checkpoint blockade immunotherapy; colorectal cancer; innate lymphoid cells; intestinal inflammation; microbiota.

Figure 1:. ILC3s are dysregulated in human and mouse CRC.

Tumor-infiltrating and lamina propria cells were respectively isolated from resected tumors, adenoma, IBD lesions and adjacent tissues from CRC and IBD patients. (A) ILCs were gated as CD45⁺ and lineage (CD3, CD4, CD8, CD11c, CD14, CD19, CD34, CD94, CD123, FcεR1α) negative, CD127⁺ and further divided by expression of CRTH-2 (ILC2s; red) or CD117 (ILC3s; blue) or as lacking expression of both markers (ILC1s; black). After data analysis and filtering, 47 CRC and 7 adenoma paired-samples reached minimal threshold for ILC3 numbers and were validated for further analyses. (B-D) Frequencies of ILC subsets among total ILCs were compared between (B) tumors, (C) adenomas, (D) IBD lesions and non-malignant/non-inflamed adjacent tissues. (E,F) Age- and sex-matched CDX2^Cre-APCmin^+/F mice developing spontaneous colonic adenoma-carcinoma were examined for the frequency of ILC3s within adenoma and adjacent non-malignant colon tissues. (E) ILCs were gated as CD45⁺ and lineage (B220, CD3ε, CD5, CD8, CD11b, CD11c, Ly6G) negative, CD90.2⁺CD127⁺ and further divided by expression of GATA-3 (ILC2s; red) or RORγt (ILC3s; blue) or as lacking expression of both markers and expressing T-bet and NKp46 (ILC1s). (F) Frequencies of ILC subsets among total ILCs and CD45⁺ cells were compared between adenoma and non-malignant colon tissues. (G) Representative pictures of immunofluorescence staining of frozen-tissue sections from mouse colon adenomas isolated from APC^min/+ mice. Sections were stained for DAPI (grey), CD3 (red), RORγt(green), and IL-7Rα(blue). Scale bar = 200 μm. White stars indicate CD3^-IL-7Ro⁺RORy⁺ILC3s. Data include (A,B) n=47 CRC, (C) n=7 adenomas (D) n=16 IBD patients or (E,F) three independent experiments pooled. (B-D,F) Results are shown as box plots with 10/25/50/75/90 percentiles. Statistical analyses between groups are performed using a (B) paired Student’s t, (C) Wilcoxon or (D,F) Mann-Whitney U test. * p < 0.05, ** p < 0.01, *** p <0.001, **** p <0.0001. See also Figure S1 and Table S1.

Figure 2:. ILC3s in CRC exhibit increased plasticity towards ILC1s.

(A) Histogram demonstrating mean normalized counts from RNA-Seq of genes expressed in sort-purified ILC3s. Markers of highly expressed core ILC and ILC3-related transcripts are indicated. (B) PCA analysis of cell-sorted ILC3 RNA-Seq performed on 4 tumor and 4 non-malignant adjacent tissues from CRC patients. (C,D) Heatmap of normalized counts comparing gene expression for transcripts related to (C) ILC1 genes and (D) to ILC3-ILC1 transitional signature. Scale based on Z-score of Log₂(normalized counts). (E,F) Analysis of transitional ILC3a to ILC1a subsets on 20 CRC patients. (E) ILC3s, ILC1s and transitional populations were defined as Lineage (CD3, CD4, CD8, CD14, CD19, CD34, CD123, FcεRIa) negative and NKp44⁺CD56⁺. ILC3a (CD103^-CD300LF⁺CCR6⁺), ILC3b (CD103⁺CD300LF⁺CCR6⁺), ILC1b (CD103⁺CD300LF^-CCR6⁺) and ILC1a (CD103⁺CD300LF^-CCR6^-) frequencies were then compared between tumors and healthy adjacent tissues. (F) Mean Fluorescence Intensity (MFI) expression of markers related to ILC3-ILC1 transition among ILC3a and ILC1a subsets. (G) ILC3 subsets were gated according to the expression of CD45RA and NKp44 to identify NKp44⁺, CD45RA⁺ and NKp44^-CD45RA^- ILC3s and (H) frequencies of ILC3 subsets among total ILC3s were compared between tumor and non-malignant adjacent tissues from CRC patients. (I,J) Age- and sex-matched RORγt-eGFP mice were treated with the AOM/DSS CRC protocol and examined for the frequency of ILC3 subsets within adenoma and non-malignant colon tissues. (I) ILC3 subsets were divided by expression of NKp46 (red) and CCR6 (blue) and (J) compared between adenoma and non-malignant colon tissues. Data include (E,F) n=20, (H) n=47 CRC or (J) three independent experiments pooled. (E,F,H,J) Results are shown as (E) the mean ± SEM or (F,H,J) box plots with 10/25/50/75/90 percentiles. Statistical analyses between groups are performed using a (B) PERMANOVA, (C-D) the DESeq2 R package with FDR threshold of 0.1, (E,F) Wilcoxon, (H) paired Student’s t or (J) Mann-Whitney U test. * p < 0.05, ** p < 0.01, *** p <0.001, **** p <0.0001. See also Figure S2.

Figure 3:. ILC3s are imbalanced with T cells in CRC and disruption of these interactions impairs microbiota-dependent type-1 immunity.

(A-D) Tumor-infiltrating and lamina propria cells were respectively isolated from resected (A,D) CRC tumors, (B) adenomas, (C) IBD lesions and adjacent tissues from CRC and IBD patients. (A-F) Ratios and frequencies of ILC3s and T cells were determined in human (A,D) CRC tumors, (B) adenomas, (C) IBD lesions and adenomas from (E) CDX2^Cre-APCmin^+/F and (F) RORγt-eGFP mice treated with the AOM-DSS chemically-induced CRC protocol. (G) Proportion of colonic T_H1 CD4 and T-bet⁺ CD8 T cells were compared in age- and sex-matched MHCII^∆ILC3 and MHCII^F/F littermate control mice. (H) Principal Coordinates Analysis (PCoA) analysis of 16S microbiota composition, determined by 16S sequencing, from Rag1^−/− mice before (day 0) and after (day 46) adoptive transfer with CD4 T cell from MHCII^∆ILC3 and MHCII^F/F mice. (I) Proportion of colonic T_H1 CD4 and T-bet⁺ CD8 T cells were compared in C57BL/6J mice after FMT with luminal microbiota from MHCII^∆ILC3 and MHCII^F/F mice. Data include (A,D) n=47 CRC, (B) n=7 adenomas, (C) n=16 IBD patients or (E,F) three or (G-I) two independent experiments. Results are shown as box plots with 10/25/50/75/90 percentiles. Statistical analyses between patient groups are performed using a (A,D) paired Student’s t, (B) Wilcoxon, (C,E-G,I) Mann-Whitney U or (H) a PERMANOVA test. * p < 0.05, ** p < 0.01, *** p <0.001, **** p <0.0001. See also Figure S3.

Figure 4:. MHCII⁺ ILC3s protect from experimental CRC progression and invasion.

(A-F) Age- and sex-matched MHCII^∆ILC3 mice and MHCII^F/F littermate controls were treated with the AOM-DSS chemically-induced CRC protocol. (A) Proportion of colonic T_H1 CD4 and T-bet⁺ CD8 T cells and (B) macroscopic visualization of colon tumor at the end-point of the protocol with (C) polyp counts, size and load. (D) Swiss-rolled colons were fixed in formalin, paraffin-embedded, cut for 5 μm sections and H&E (top) or β-catenin (bottom) stained for histology. Scale bar: 200 μm, black arrows indicate malignant areas. (E) Score for tumor dysplasia and (F) dysplasia and invasion involvement. (G-I) APC^min/+-MHCII^ΔILC3 crossed-mice and APC^min/+-MHCII^F/F littermate controls were scored for (G) tumor dysplasia, (H) tumor number and (I) were followed for survival. (A-I) Data include (A) two, (C-F) five or (G-I) three independent experiments pooled. (A,C,F,H) Results are shown as the mean ± SEM. Statistical analyses were evaluated using a (A,C,E,F,H) Mann-Whitney U or (I) Logrank Mantel-Cox test. * p < 0.05, ** p < 0.01, *** p <0.001, **** p <0.0001. See also Figure S4.

Figure 5:. MHCII⁺ ILC3s preserve type-1 immunity in tumors and prevent resistance to immunotherapy.

(A-F) MHCII^ΔILC3 mice and MHCII^F/F littermate controls, (G,H) germ-free mice transplanted with luminal fecal microbiota from MHCII^ΔILC3 mice or littermate controls, or (I,J) C57BL/6J mice treated with DSS were injected subcutaneously with MC38 and treated with either an anti-PD-1 or a control mAb at the indicated days. (A,G,I) Tumor growth curve and (B,H,J) tumor weight with mean ± SEM pooled from (A,B,I,J) three or (G,H) two independent experiments are shown. (C) Proportion of tumor-infiltrating T_H1 CD4 and T-bet⁺ CD8 T cells and (D) PD-1⁺ CD8 T cells were compared between MHCII^∆ILC3 and littermate control mice. (E,F) Weighted UniFrac PCoA analysis of 16S sequencing luminal colonic microbiota from MC38 tumor-bearing MHCII^∆ILC3 and littermate control mice treated with (E) control or (F) anti-PD-1 mAbs. Results are shown as the mean ± SEM. Statistical analyses were evaluated using a (A,G,I) two-way ANOVA, (B-D,H,J) Mann-Whitney U or (E,F) PERMANOVA test. * p < 0.05, ** p < 0.01, *** p <0.001, **** p <0.0001. See also Figure S5 and Table S2.

Figure 6:. IBD patients harbor microbiota that cause resistance to immunotherapy.

(A) Frequencies of ILC3s among CD45⁺ cells and T cell/ILC3 ratio comparison between biopsies from healthy and IBD donors. (B) Weighted UniFrac PCoA analysis of 16S sequencing of microbiota from 3 IBD and 4 healthy donors. (C) Weighted UniFrac distances between pairs of mouse fecal samples, showing donor effect. (D-H) ABX-treated mice were transplanted with microbiota from IBD and healthy donors, injected subcutaneously with MC38, and treated with either anti-PD-1 or an isotype control mAb at the indicated days. (D) Proportion of T-bet⁺ colonic CD4 T cell, (E) tumor growth curve and (F) tumor weight. (G-H) Correlation analyses between the response score to anti-PD-1 mAb and the relative bacterial abundance in animals treated with anti-PD-1 mAb. Results are shown as (A) box plots with 25/50/75 quartiles or (D-F) the mean ± SEM. (D-H) Data include (D) three or (E-H) four independent experiments pooled. Statistical analyses were evaluated using a (B) PERMANOVA, (C) permutation testing (D,F) Mann-Whitney U, (E) two-way ANOVA and (G,H) Spearman correlation test. * p < 0.05, ** p < 0.01, *** p <0.001, **** p <0.0001. See also Figure S6 and Table S3 and S4.