T cell-derived interleukin-22 drives the expression of CD155 by cancer cells to suppress NK cell function and promote metastasis

Abstract

Although T cells can exert potent anti-tumor immunity, a subset of T helper (Th) cells producing interleukin-22 (IL-22) in breast and lung tumors is linked to dismal patient outcome. Here, we examined the mechanisms whereby these T cells contribute to disease. In murine models of lung and breast cancer, constitutional and T cell-specific deletion of Il22 reduced metastases without affecting primary tumor growth. Deletion of the IL-22 receptor on cancer cells decreases metastasis to a degree similar to that seen in IL-22-deficient mice. IL-22 induced high expression of CD155, which bound to the activating receptor CD226 on NK cells. Excessive activation led to decreased amounts of CD226 and functionally impaired NK cells, which elevated the metastatic burden. IL-22 signaling was also associated with CD155 expression in human datasets and with poor patient outcomes. Taken together, our findings reveal an immunosuppressive circuit activated by T cell-derived IL-22 that promotes lung metastasis.

Keywords: CD155; CD226; NK cells; PVR; T helper; breast carcinoma; interleukin-22; lung adenocarcinoma; metastasis; poliovirus receptor.

Graphical abstract

Figure 1

IL-22-knockout reduces the number of lung metastases but does not affect tumor growth in syngeneic mouse models of lung and breast carcinoma (A) Subcutaneous (s.c.) mouse models of 4T1 breast and Line-1 lung metastasis in wild-type (WT) and Il22^−/− animals. (B and C) Subcutaneous tumor growth, macroscopic metastases in the lungs, and colonies in clonogenic metastasis assay of (B) 4T1 (n = 47 and 48 for tumor growth and metastasis number; n = 27 and 28 for clonogenic assay) (data of six experiments for tumor growth and four independent experiments for metastasis assay in each cell line were pooled) and (C) Line-1 cells (n = 26 and 25 for tumor growth and metastasis number; n = 12 and 11 for clonogenic assay) (data of four experiments for tumor growth and two experiments for metastasis assay were pooled). (D) Intravenous (i.v.) models of lung cancer metastasis. (E and F) Numbers of macroscopic metastases in the lungs and colonies in metastasis assay of (E) 4T1 (n = 24 and 29) and (F) Line-1 cells (n = 14 and 15). Data from six experiments for 4T1 and three for Line-1 were pooled. p values are calculated by the mixed-effect two-way analysis for tumor growth and by the Mann-Whitney U test for the number of metastases and clonogenic assay. (G) Intravenous model of E0771-GFP breast cancer metastasis. (H) Number of metastases in the lungs (n = 22 and 23). Representative flow cytometry plots of the lung cells. The numbers are the frequency of the parent gate. Frequency of E0771-GFP⁺ cells among live cells (n = 19 and 21). Data from four experiments were pooled. p values are calculated by the Mann-Whitney U test. (I) Representative reconstructed 3D images of the left lung from E0771-GFP-injected mice cleared using iDISCO protocol and imaged using light-sheet microscopy. The GFP signal is in green. (J) Numbers per left lung and the average surface area of metastases (μm²) per mouse (n = 8 and 9). Data from two experiments were pooled. p values are calculated by Welch’s t test. All data are presented as mean ± SEM; p values <0.05 are considered significant. See also Figure S1.

Figure 2

T cells are the primary source of IL-22 in the lung of tumor-bearing mice (A) Intravenous model of E0771 lung metastasis in Foxp3^mRFPIl17a^GFPIl22^sgBFP reporter mice. (B) Gating strategy to identify CD4⁺, CD8⁺, and double-negative (DN) T cells, γδ T cells, CD3⁺NK1.1⁺, and CD3⁻NK1.1⁺ cells in the lungs. Numbers represent the frequency of the parent gate. (C) Breakdown of CD45⁺IL-22⁺ cells by cell type as defined by the mean frequency of reported experiments. (D) IL-17A and IL-22 production by CD4⁺ T cells (n = 5). Data are presented as mean ± SEM, and the representative experiment of two is shown. p values by Welch’s t-test. (E) Representative 3D render of the z stack confocal images of metastatic foci and normal tissue of precision-cut lung slices from Foxp3^mRFPIl17a^GFPIl22^sgBFP mice injected with E0771 cells i.v. Fifteen fields of view of each type were acquired per mouse from 3 individual mice. The field of view is 160 × 160 μm; 20 slices with an interval of 1 μm were acquired; displayed grid step is 20 μm. Il17a^GFP is depicted in green, Il22^sgBFP in blue, CD4 PerCP-Cy5.5 staining in magenta, and TO-PRO-3 nuclear staining in red. Correlation between Il17a^GFP and Il22^sgBFP median fluorescence intensity (MFI) in the reporter cells in the metastatic foci. The number of reporter cells per field of view in the metastatic foci and normal tissue. Normalized CD4 PerCP-Cy5.5 MFI in Il22^sgBFP+ cells gated from the previous graph. Representative data are presented as means ± SEM of one animal of thee. p values <0.05 are considered significant as calculated by the unpaired t-test. (F) Intravenous mouse model of E0771-GFP lung metastasis in Il22^floxCd4^cre mice. Representative dot plots, numbers of macroscopic metastases, and frequency of E0771-GFP cells in the lungs of WT and Il22^floxCd4^cre mice (n = 18). Data are presented as means ± SEM; three independent experiments were pooled. p values <0.05 are considered significant as calculated by the Mann-Whitney U test. (G) Model of metastasis in Rag1^−/−Il22^−/− that received WT or Il22^−/− CD4⁺ T cells i.p. (2 × 10⁶ per mouse) 28 days before i.v. tumor injection. Representative dot plots, numbers of macroscopic metastases, and frequency of E0771-GFP cells in the lungs (n = 5 and 4). Data are presented as means ± SEM of one experiment. p values <0.05 are considered significant as calculated by the Mann-Whitney U test. See also Figure S2.

Figure 3

IL-22RA1 expression on tumor cells is indispensable for metastasis formation (A) Subcutaneous and intravenous models of 4T1 Il22ra1⁻ lung metastasis. (B) Tumor size, number of metastases, and clonogenic colonies in s.c. model per animal (n = 20 and 21). Data from three independent experiments for tumor size and metastasis count and two for metastasis assay were pooled. (C) Number of metastases and clonogenic colonies in i.v. model (n = 14). Data from three independent experiments were pooled. (D) Intravenous model of Line-1 Il22ra1⁻ lung metastasis. (E) Number of metastases and colonies of the Line-1 control (#1, #2, #3) or Il22ra1⁻ (#4, #5, #6) clones (n = 6 per clone). Data from two independent experiments were pooled and are presented as means ± SEM; p values <0.05 were considered significant by mixed-effect two-way analysis for tumor growth and by the Mann-Whitney U test for the number of metastases and colonies. See also Figures S3 and S4.

Figure 4

IL-22 signaling increases the expression of CD155 on the surface of tumor cells and confers resistance to metastasis control (A) 4T1 cells were stimulated with IL-22 (100 ng/mL) for 24 h in vitro before bulk mRNA sequencing. (B) The volcano plot depicts the fold change and adjusted p values of differentially regulated genes. One hundred forty-seven genes were discovered as defined by an adjusted p value threshold of 0.05 and Log₂ fold change of ±1. (C) CD155 expression in 4T1, Line-1, and E0771 cells after stimulation with rmIL-22 (100 ng/mL) for 72 h in vitro. (D) Representative histograms of CD155 expression at 24, 48, and 72 h. (E) MFI of CD155-PE signal normalized to control (n = 3). Data are presented as means ± SEM, and p values <0.05 are considered significant as calculated by the mixed model two-way analysis. (F) Intravenous model of E0771-GFP lung metastases in WT and Il22^−/− mice. (G) Numbers of metastases in the lungs, representative plots, and frequency of E0771-GFP⁺ cells (n = 6 and 10). The numbers are the frequency of the parent gate. (H) Representative histogram and MFI of CD155 staining of GFP + tumor cells. Linear regression analysis of CD155 MFI and frequency of E0771-GFP + tumor cells. Data from one experiment. p values are calculated by the Mann-Whitney U test. (I) Intravenous models of Line-1 Pvr⁻ and Pvr⁺ metastasis in WT and Il22^−/−mice. (J and K) Numbers of metastases and colonies in metastasis assay of (J) Line-1 Pvr⁻ and (K) Pvr⁺ cells. Data from three independent experiments for Pvr⁻ and two experiments for Pvr⁺ were pooled and are presented as means ± SEM; p values <0.05 were considered significant by the Mann-Whitney U test. See also Figure S5.

Figure 5

CD226 expression is higher on NK cells in Il22^−/− mice (A) Intravenous model of 4T1-luciferase⁺ (4T1-Luc) metastasis. (B) Representative IVIS images and average radiance from one experiment of two (n = 5). (C) Ex vivo imaging of the lung by IVIS. Linear regression analysis of clonogenic colonies vs. average radiance. (D) Representative dot plots and gating strategy of IFNγ-producing cells in the lungs. Numbers represent the frequency of the parent gate. Frequency of IFNγ⁺ cells in the lung and composition by cell type including CD8⁺ and CD4⁺ T cells, NKT cells, and NK cells. (E) Representative dot plots and frequency of IFNγ⁺ NK cells (n = 9 and 10). Data from two independent experiments were pooled. (F) Representative fluorescent microscopy images of lungs from 4T1-injected mice. NKp46 FITC signal is depicted in red, CD8 AF555 in yellow, CD155 PE in green, and Hoechst DNA staining in blue. Intralesional CD155 PE MFI per lesion in WT and Il22^−/− mice. Chip cytometry spatial distribution of NKp46⁺ (red), CD8⁺ (green), and CD155⁺ (gray) cells in the microscopy samples. Frequency of CD8 and NKp46 cells per lesion from the previous graph. Data presented as means ± SEM; p values <0.05 were considered significant by paired t test of one chip of 2 for each condition (n = 3 and 4). (G) Gating strategy and representative plots of CD226 staining on CD8⁺ T, CD4⁺ T, NKT, and NK cells. Isotype control is in black. Frequency of CD226⁺ cells in the lungs (n = 6). Data of one experiment of two. Data presented as means ± SEM; p values <0.05 were considered significant by the Mann-Whitney U test or multiple unpaired t tests with Holm-Sidak correction. See also Figure S6.

Figure 6

CD226 signaling is indispensable for IFNγ production from NK cells (A) Intravenous mouse model of 4T1 control or Pvr⁺ metastasis in Il22^−/− mice. Animals received injections of anti-CD226 blocking antibody (420.1, 200 μg per mouse) or control i.p. on days 0, 3, and 14. (B) Numbers of macroscopic lung metastases and colonies in metastasis assay. (C) IFNγ-producing NK cells in the lungs evaluated by flow cytometry. Data presented as means ± SEM; p values <0.05 were considered significant by the Mann-Whitney U test. (D) Representative dot plots depict CD226 low, medium, and high NK cells. Frequency of IFNγ+ cells among NK cell populations (n = 5–6). Numbers represent the frequency of the parent gate. Data presented as means ± SEM; p values <0.05 were considered significant by the two-way ANOVA. (E) Intravenous mouse model of 4T1 control or Pvr⁺ lung metastasis in Il22^−/− mice. Animals received injections of anti-TIGIT agonist antibody (1G9, 250 μg per mouse), anti-CD96 blocking antibody (3.3, 250 μg per mouse), or control i.p. on days 0, 3, and 14. (F) Numbers of macroscopic lung metastases. (G) IFNγ-producing NK and CD8⁺ T cells in the lungs were evaluated by flow cytometry. Numbers represent the frequency of the parent gate. Data presented as means ± SEM; p values <0.05 were considered significant by unpaired t- test (n = 5–6).

Figure 7

Cluster stratification and clinical outcome of LUAD and BRCA patient samples based on IL-22-CD155 signature genes (A) Dendrogram of agglomerative clustering of TCGA LUAD and BRCA HER2+ datasets based on IL22RA1, IL22RA2, IL10RB, and PVR expression. (B) t-distributed Stochastic Neighbor embedding (t-SNE) dimensionally reduced scaled gene expression values. Each dot represents one patient sample, and colors indicate cluster affiliation. (C) Heatmap of the mean scaled expression of IL22RA1, IL22RA2, IL10RB, and PVR per cluster. (D) Relative proportions of the clusters. (E) Kaplan-Meier survival plots of all clusters for each dataset respectively. Log rank (Mantel-Cox) test was used to compare curves. (F) Restricted mean survival time difference (δRMST) of clusters 0 and 1. Line indicates a cutoff of 1,825 days. (G) Frequency of summarized pathologic stages per cluster. A chi-squared test was used to compare groups; p values <0.05 are considered significant. All shown results were generated in one representative run of the analysis script.