KRAS-IRF2 Axis Drives Immune Suppression and Immune Therapy Resistance in Colorectal Cancer

Abstract

The biological functions and mechanisms of oncogenic KRASG12D (KRAS∗) in resistance to immune checkpoint blockade (ICB) therapy are not fully understood. We demonstrate that KRAS∗ represses the expression of interferon regulatory factor 2 (IRF2), which in turn directly represses CXCL3 expression. KRAS∗-mediated repression of IRF2 results in high expression of CXCL3, which binds to CXCR2 on myeloid-derived suppressor cells and promotes their migration to the tumor microenvironment. Anti-PD-1 resistance of KRAS∗-expressing tumors can be overcome by enforced IRF2 expression or by inhibition of CXCR2. Colorectal cancer (CRC) showing higher IRF2 expression exhibited increased responsiveness to anti-PD-1 therapy. The KRAS∗-IRF2-CXCL3-CXCR2 axis provides a framework for patient selection and combination therapies to enhance the effectiveness of ICB therapy in CRC.

Keywords: CXCL3; CXCR2; IRF2; KRAS; anti-PD-1; colorectal cancer (CRC); immune checkpoint blockade (ICB).

Copyright © 2019 Elsevier Inc. All rights reserved.

Figures

Figure 1.. KRAS* Promotes an Immune Suppressive Microenvironment in CRC Progression

(A) viSNE analysis of immune cells from iAP and iKAP tumors colored by relative expression of CyTOF markers, with populations indicated. (B) Quantification of tumor-infiltrating immune (CD45+) cells in iAP (n = 3) and iKAP (n = 3) primary CRC, assessed by CyTOF and analyzed by FlowJo. Cell populations were identified as T cells (CD45+CD3e+TCRβ+), CD4+ T cells (CD45+CD3e+TCRβ+CD8-CD4+), CD8+ T cells (CD45+CD3e+TCRβ+CD8+CD4-), MDSCs (CD45+CD11b+F4/80-Gr-1+), PMN-MDSC (CD45+CD11b+Gr-1+ Ly-6G+Ly-6C-), and M-MDSC (CD45+CD11b+Gr-1+ Ly-6G-Ly-6C+). (C) Quantification of T cells, CD4+ and CD8+ T cells, and MDSCs in iKAP tumors following withdrawal of Dox for 1 week (n = 3) as compared with iKAP tumors maintained on Dox (n = 3). (D) IHC analysis for CD4+ (CD4), CD8+ (CD8) and MDSC (Gr-1, S100A8, and S100A9) markers. Scale bars, 50 μm. Representative data of triplicate experiments are shown. (E and F) Representative CFSE flow cytometry histograms (E) showing the effect on in vitro T-cell proliferation by MDSCs isolated from iKAP tumors and summarized result (F). Unstimulated T cells were used as negative control. Position of CFSE peaks can be used to denote the T-cell division times. High and low proliferation were defined as T-cell division ≥ 2 and ≤ 1, respectively (n ≥ 3, biological replicates). (G) IFN-γ secretion by CD8+ T cells in the assay in (E), measured by ELISA (n = 3, biological replicates). In panels B-D and G, data represent mean ± s.d. * p < 0.05, ** p < 0.01, *** p < 0.001, **** p <0.0001, Student’s t-test. See also Figure S1 and Table S1.

Figure 2.. KRAS* Suppresses IFN Responses in CRC

(A) IPA analysis of RNAseq data (iAP vs iKAP) showing the top 10 pathways that are suppressed in iKAP tumors (n = 7) as compared with iAP tumors (n = 8). Dataset: SRP097890. Graph displays category scores as-log10 (p value) from Fisher’s exact test. (B) IPA analysis of RNAseq data (primary iKAP cell lines, Dox-off 24h vs Dox-on, n = 6 for each group) showing the top 10 pathways that are enriched in Dox-off cells as compared with Dox-on cells. Dataset: SRP097890. Graph displays category scores as -log10 (p value) from Fisher’s exact test. (C) Real-time qPCR validation of representative IFN response genes in a cultured primary iKAP cell line (iKAP-1) maintained on Dox or following withdrawal of Dox for 24 or 48 hr (right). Extinction of KRAS* was measured by western blotting using anti-p-ERK antibody (left). Data are shown as mean ± s.d. from each of three independent experiments. n = 3. (D) Venn diagram analysis cross three groups of genes. i: IFN-α and IFN-γ signature genes (n = 223); ii: Differentially expressed, invasive vs non-invasive (n = 5727); iii: Mutually exclusive with KRAS mutation (n = 56). (E) Genomic alterations of KRAS andIRF2 in TCGA CRC database (n = 633). The gene alteration percentages are shown. See also Figure S2, Table S2 and Table S3.

Figure 3.. IRF2 is a Key Downstream Target of KRAS* Mediating IFN Signaling Suppression

(A) IHC staining of IRF2 in areas with or without KRAS* expression in the same iKAP mouse tumor. GFP is used as a marker of KRAS* expression. The lower panels are amplified images of the boxed regions. Scale bars, 500 μm (upper panels) and 20 μm (middle and lower panels). The right graph shows quantification and Pearson Correlation of IRF2 and KRAS* expression in iKAP CRC (n = 20, biological replicates). (B) IHC staining of IRF2 in iKAP CRC maintained on Dox (Dox-on) or taken off Dox (Dox-off) for 48 hr (upper) or 1 week (lower). Scale bars, 50 μm. (C) The top 10 pathways changed in IRF2-overexpressing cells as compared with control cells revealed by microarray and IPA analysis. (D) Real-time qPCR validation of representative IFN response genes in iKAP cells overexpressing IRF2 as compared with vector control. Data are shown as mean ± s.d. n = 3. (E) ChIP-seq in iKAP cells revealed binding peaks for IRF2 on the promoters of IFN response genes. See also Figure S3.

Figure 4.. IRF2 Suppresses KRAS*-Driven MDSC Migration by Targeting CXCL3/CXCR2 Axis

(A) Migration of MDSCs toward conditioned medium from cultured primary iKAP cells (Dox-on vs Dox-off 24 hr), or iKAP cells overexpressing IRF2 compared with vector control, were evaluated using in vitro transwell migration assay in triplicate. (B) Real-time qPCR measurement of IRF2 expression in MC38 cells using control shRNA or shRNA specific to mouse IRF2 (shIRF2). (C) Migration of MDSCs toward conditioned medium from MC38 cells (control shRNA vs shIRF2) were analyzed by transwell MDSC migration assays in triplicate. (D) MDSC infiltration in the MC38 tumors (control shRNA vs shIRF2) were analyzed by CyTOF. (E) IRF2 binding peaks at the Cxcl3 locus in iKAP cells revealed by ChIP-seq. (F) ChIP-PCR validation of IRF2 binding to the Cxcl3promoter. (G and H) Real-time qPCR (G) and ELISA (H) analysis of CXCL3 in cultured primary iKAP CRC cell lines (Dox-off 24 hr vs Dox-on); or in iKAP cells overexpressing IRF2 or vector control constructs. (I) Expression and co-localization of CXCR2 and Gr-1 in iKAP CRC tissues by immunofluorescence (IF) staining. Scale bars, 20 μm. (J) Migration of MDSCs toward basal media supplemented with recombinant CXCL3. (K) Migration of MDSCs toward conditioned medium (CM) from iKAP tumor cells treated with Vehicle or IgG control, CXCL3-neutralizing antibody, SX-682, or CXCR2-neutralizing antibody. (L) The effect of SX-682 on MDSC and total T-cell infiltration in iKAP mice, measured by flow cytometry. n = 3 for each cohort. (M) IF staining and quantification for MDSC (Gr-1), CD4+ T cells (CD4) and CD8+ T cells (CD8) in the presence or absence of SX-682. Scale bars, 50 μm. (N) The effect of SX-682 on the tumor histopathology in iKAP CRC tumors. Hematoxylin and eosin (HE) staining shows the histology of tumor tissues. IHC against GFP shows tumor cells with KRAS* expression. The right graph shows the quantification of tumor cells (GFP) as analyzed by IHC. n = 3, biological replicates. Scale bars, 100 μm. In A-D, F-H and J-N, analyses were done in triplicate. Data represent mean ± s.d. from each of three independent experiments. * p

Figure 5.. Overexpression of IRF2 Increases the Sensitivity of KRAS* CRC Cells to ICB Therapy

(A) Expression of IRF2 and p-ERK in MC38, MC38K, and MC38KI cells by western blotting. (B) Real-time qPCR showed expression of representative IFN stimulated genes and CXCL3 gene in the indicated cell lines. (C) MDSCs (CD45+CD11b+Gr-1+) were examined in MC38, MC38K, and MC38KI tumors, analyzed by Flow cytometry. (D) Tumor growth of MC38, MC38K, and MC38KI tumors in C57BL/6J mice treated with anti-PD-1 or isotype control. (E) Survival of C57BL/6J mice with MC38, MC38K, and MC38KI tumors and treated with anti-PD-1 or isotype control. ** p

Figure 6.. Targeting CXCR2 Increases the Sensitivity of KRAS* CRC Cells to Anti-PD-1 Therapy

(A) Survival of iKAP mice treated with SX-682 or anti-PD-1 as a single agent, or SX-682 treatment in combination with anti-PD-1. IgG+Vehicle: n = 9. Anti-PD-1: n = 6. SX-682: n = 12. SX682 plus anti-PD-1: n = 8. n means biological replicates. * p + T cells to FOXP3+ regulatory T cells in iKAP tumors treated with SX-682 plus anti-PD-1 as compared to those treated with SX-682 or anti-PD-1 monotherapy. (C) Representative images and quantification of CD8 and Tregs analyzed by IF staining in iKAP CRC tumors. n = 3, biological replicates. Scale bars, 50 μm. The small boxed areas in the bottom row are amplified images of cells with FOXP3 and CD4 staining. Scale bars in the small boxed areas, 25 μm. (D) Tumor volumes of MC38K in C57BL/6J mice treated with anti-PD-1, SX-682, SX-682 plus anti-PD-1, or isotype control. (E) Survival of MC38K tumor bearing C57BL/6J mice treated with anti-PD-1 or SX-682 plus anti-PD-1. *** p

Figure 7.. Clinical Relevance of KRAS* and IRF2 Expression in CRC Patients

(A) GSEA analysis identified the IFN-γ and IFN-α signatures as the top suppressed pathways in total and MSS KRAS* CRC in TCGA CRC data. (B) GSEA analysis identified the IFN-γ signature as the top activated pathway in IRF2-high CRC patients (TCGA). (C) Representative IHC staining for IRF2 in human CRC TMA with wild-type KRAS (n = 42) and mutant KRAS (n = 40). The bar graph shows Pearson Correlation and two-tailed p value. Scale bars, 400 μm (left) and 40 μm (right). (D) Correlation analysis of IRF2 expression and response to anti-PD-1 therapy in MSI-H CRC (n = 14). We define CR (Complete Response), PR (Partial Response) and SD (Stable Disease) without subsequent PD (Progression Disease) as responders, while PD as non-responders. CR: n = 1; PR: n = 7; SD: n = 1; PD: n = 5. The graph shows Pearson Correlation and two-tailed p value. (E) Schematic representations of the role of KRAS/IRF2 axis in immune suppression and ICB resistance in CRC. See also Figure S7.