m6 A RNA methyltransferases METTL3/14 regulate immune responses to anti-PD-1 therapy

Abstract

An impressive clinical success has been observed in treating a variety of cancers using immunotherapy with programmed cell death-1 (PD-1) checkpoint blockade. However, limited response in most patients treated with anti-PD-1 antibodies remains a challenge, requiring better understanding of molecular mechanisms limiting immunotherapy. In colorectal cancer (CRC) resistant to immunotherapy, mismatch-repair-proficient or microsatellite instability-low (pMMR-MSI-L) tumors have low mutation burden and constitute ~85% of patients. Here, we show that inhibition of N⁶ -methyladenosine (m⁶ A) mRNA modification by depletion of methyltransferases, Mettl3 and Mettl14, enhanced response to anti-PD-1 treatment in pMMR-MSI-L CRC and melanoma. Mettl3- or Mettl14-deficient tumors increased cytotoxic tumor-infiltrating CD8⁺ T cells and elevated secretion of IFN-γ, Cxcl9, and Cxcl10 in tumor microenvironment in vivo. Mechanistically, Mettl3 or Mettl14 loss promoted IFN-γ-Stat1-Irf1 signaling through stabilizing the Stat1 and Irf1 mRNA via Ythdf2. Finally, we found a negative correlation between METTL3 or METTL14 and STAT1 in 59 patients with pMMR-MSI-L CRC tumors. Altogether, our findings uncover a new awareness of the function of RNA methylation in adaptive immunity and provide METTL3 and METTL14 as potential therapeutic targets in anticancer immunotherapy.

Keywords: CD8+ T cells; colorectal carcinoma; immunotherapy; m6A methylation.

Figure 1. Depletion of Mettl3 or Mettl14 sensitizes CT26 and B16 tumors to immunotherapy

1. A, B

   Immunoblotting were performed to validate Mettl3 or Mettl14 expression levels in CT26 and B16 cells as indicated. Gapdh served as a loading control.

2. C, D

   Tumor volume was monitored for control and Mettl3‐ or Mettl14‐depleted tumors with treatment as indicated in CT26 colon cancer and B16 melanoma, respectively. Data are mean ± SEM of the indicated number of mice in each group. n, the numbers of mice. *P < 0.05; ***P < 0.001 by Student's t‐tests.

3. E, F

   Survival analysis of control tumors and those with depleted genes were recorded as indicated in CT26 colon cancer and B16 melanoma, respectively. Data are mean ± SEM of the indicated number of mice in each group. n, the numbers of mice. *P < 0.05; **P < 0.01; ***P < 0.001 by Student's t‐tests.

Source data are available online for this figure.

Figure EV1. Depletion of Mettl3 or Mettl14 enhanced the response to immunotherapy. (Related to Fig 1)

1. A

   CT26 tumor volume was measured from tumor‐bearing mice treated with IgG (IgG control), P (anti‐PD-1 antibody), and P plus C (anti‐CTLA-4 antibody) therapeutic modalities. n, the numbers of mice. Data are mean ± SEM of the indicated number of mice. *P < 0.05; ***P < 0.001 by Student's t‐tests.

2. B

   Survival analysis after various treatments of mice bearing CT26 tumors. n, the numbers of mice. Data are mean ± SEM of the indicated number of mice. ***P < 0.001 by Student's t‐tests.

3. C, D

   C57BL/6J or BALB/c mice bearing control and Mettl3‐ or Mettl14‐depleted tumors (C, CT26; D, B16) were treated with various therapeutic modalities as indicated. Tumor volume was recorded over time as indicated. Each line represents one mouse.

4. E, F

   Immunoblots of Mettl3 and Mettl14 were carried out in the indicated CT26 and B16 mouse tumors in triplicates with Gapdh as a loading control.

5. G

   Representative images of Ki‐67 were stained by IHC analysis. Tissue sections from BALB/c mice bearing the indicated knockout of genes with treatment of PD1 antibody. Scale bars, 50 μm.

Source data are available online for this figure.

Figure EV2. Loss of Mettl3 or Mettl14 has no effect to cell proliferation and tumor growth

1. A

   Cell proliferation was assessed in knockout of Mettl3, Mettl14, and non‐targeting control (NTC) CT26 and B16 cells using MTS assay in vitro. Mean ± SD of n = 3.

2. B, C

   Tumor growth of xenografts from CT26 and B16 cells with Mettl3‐ or Mettl14‐depleted genes and control as indicated. n, the numbers of mice. Data are mean ± SEM of the indicated number of mice.

3. D, E

   Tumor growth from C57BL/6J or BALB/c mice with control and Mettl3 or Mettl14‐depleted tumors (D, CT26; E, B16). Tumor volume was recorded over time as indicated. Each line represents one mouse.

Figure 2. Mettl3 or Mettl14 deficiency enhances tumor‐infiltrating CD8⁺ T cells and cytokine production

1. A

   Percentage of tumor‐infiltrating T cells, Treg, and NK cells were identified by flow cytometry from CT26 tumors as indicated. Each spot represents one mouse. *P < 0.05; **P < 0.01 by Student's t‐tests.

2. B

   Representative images of CD8 by IHC staining. Tissue sections from BALB/c mice bearing the indicated knockout of genes with treatment of PD1 antibody. Scale bars, 50 μm.

3. C

   Percentage of granzyme B‐expressing CD8⁺ T cells from control and Mettl3‐ or Mettl14‐deficient CT26 tumors. Each spot represents one mouse. *P < 0.05; **P < 0.01 by Student's t‐tests.

4. D, E

   Mice bearing control and Mettl3 or Mettl14 null tumors were treated with CD8‐depleting antibody and PD‐1 antibody or PD‐1/GVAX as indicated. Tumor volume was measured over time points. n, the numbers of mice. *P < 0.05; **P < 0.01; ***P < 0.001 by Student's t‐tests.

5. F, G

   IFN‐γ production in serum (F) and intratumor (G) from BALB/c mice by ELISA. The results are representatives of at least three independent experiments. n, the numbers of mice. Data are mean ± SEM. *P < 0.05 by Student's t‐tests.

Figure EV3. Tumor‐infiltrating CD8⁺ T cells and chemokines concentration were altered in Mettl3 or Mettl14 null tumors. (Related to Fig 2)

1. Representative examples for CD8⁺ T cells from FACS analyses in CT26 tumors.

2. Percentage of tumor‐infiltrating CD8⁺ T cells and NK cells were analyzed from control and Mettl3‐ or Mettl14‐deficient B16 tumors using flow cytometry. Each spot represents one mouse. Data are mean ± SEM of the indicated number of mice. *P < 0.05; **P < 0.01 by Student's t‐tests.

3. Percentage of granzyme B‐expressing CD8⁺ T cells from B16 tumors as indicated. Each spot represents one mouse. *P < 0.05; **P < 0.01 by Student's t‐tests.

4. Secretion of Cxcl9 in serum from the indicated BALB/c mice by ELISA. Each spot represents one mouse. *P < 0.05 by Student's t‐tests.

5. Intratumoral Cxcl9 concentration were determined by ELISA in the indicated CT26 tumor extracts and then calculated by the total protein concentration. Each spot represents one mouse. n, the numbers of mice. *P < 0.05 by Student's t‐tests.

6. Secretion of Cxcl10 in serum from the indicated BALB/c mice by ELISA. Each spot represents one mouse.

7. Intratumoral Cxcl10 concentration were determined by ELISA in the indicated CT26 tumor extracts and then calculated by the total protein concentration. Each spot represents one mouse. n, the numbers of mice. *P < 0.05; **P < 0.01 by Student's t‐tests.

Figure 3. Identification of target genes of Mettl3 and Mettl14 by RNA‐seq and m⁶A‐seq

1. Volcano plot of differentially expressed genes obtained by DESeq2 analysis in Mettl3 or Mettl14 null tumors compared to control tumors. Significantly upregulated or downregulated genes are plotted in red and blue points, respectively. n.s, non‐significant.

2. Venn diagrams showing 202 significantly co‐upregulated genes and 28 significantly co‐downregulated genes in the indicated tumors.

3. Meta‐enrichment analysis summary for 202 significantly co‐upregulated genes as indicated in (C).

4. Consensus m⁶A motifs and P value identified by HOMER from two biological replicates, Student's t‐tests.

5. Schematic workflow for analysis of Mettl3 and Mettl14 downstream genes and identified genes or peaks number.

6. Representative genes with m⁶A sites generated by integrative genomics viewer. Data are representative of duplicates with similar results. Red represents reads coverage of IP sample and blue represents reads coverage of input sample. Rectangular cyan shade represents the m⁶A peaks located on transcripts.

7. m⁶A enrichment of Stat1 and Irf1 was examined by m⁶A RIP‐qPCR in control, Mettl3‐, or Mettl14‐depleted CT26 tumors as indicated. Ctla4 functioned as a m⁶A negative control (Wang et al, 2019). Data are mean ± SD. **P < 0.01 by Student's t‐tests.

8. Immunoblots of p‐Stat1 (phosphorylated), Stat1, and Irf1 were carried out in the indicated tumors in triplicate with Gapdh as a loading control.

9. Tumor growth from CT26 cells with Mettl3‐, Mettl14‐, Mettl3/Stat1‐, Mettl3/Irf1-, Mettl14/Stat1‐, or Mettl14/Irf1‐depleted genes and control under treatment of PD‐1 antibody as indicated. n, the numbers of mice. Data are mean ± SEM of the indicated number of mice in each group. *P < 0.05; **P < 0.01 by Student's t‐tests.

Source data are available online for this figure.

Figure EV4. Gene expression changes and analysis of m⁶A modification in Mettl3‐ or Mettl14‐depleted tumors (Related to Fig 3)

1. Transcriptional analysis of the indicated genes identified from the RNA‐seq data using quantitative RT–PCR. mRNA levels in Mettl3‐ or Mettl14‐deficient tumors are presented as the relative fold change compared to control sgRNA tumor. The mean ± SD of five replicates is shown. *P < 0.05; **P < 0.01; ***P < 0.001 by Student's t‐tests.

2. Dot blot of the total m⁶A levels in mRNA extracted from Mettl3‐ or Mettl14‐depleted and control tumors.

3. Number of consensus m⁶A peaks identified from two biological replicates in the indicated tumors.

4. Venn diagram of upregulated m⁶A containing genes, downregulated m⁶A containing genes, and common m⁶A genes without expression level changes as indicated. Orange shade represents upregulated m⁶A containing genes from Mettl3‐depleted tumors compared to control, blue shade represents downregulated m⁶A containing genes from Mettl3‐depleted tumors compared to control. Green shade represents upregulated m⁶A containing genes from Mettl14‐depleted tumors compared to control, yellow shade represents downregulated m⁶A containing genes from Mettl14‐depleted tumors compared to control. Gray shade represents common m⁶A genes without any changes from Mettl3‐ and Mettl14‐depleted tumors compared to control.

5. GO analysis was performed on 64 co‐upregulated m⁶A containing genes from D as indicated.

6. Distribution of m⁶A peaks in the indicated tumors. Pie charts show the proportion of m⁶A peaks in the 5′‐UTR (orange), CDS (gray), 3′‐UTR (yellow), and promoter‐TSS (blue).

7. Representative genes with m⁶A sites generated by integrative genomics viewer. Blue represents reads coverage of input sample and red represents reads coverage of IP sample. Rectangular cyan shade represents the m⁶A peaks located on transcripts.

Source data are available online for this figure.

Figure EV5. Stat1 and Irf1 are targets regulated by Mettl3 and Mettl14 (Related to Fig 3)

1. A, B

   Immunoblot analysis of the protein levels of Mettl3, Mettl14, Irf1, and Stat1 in CT26 cells as indicated. Gapdh served as a control.

2. C–E

   Tumor growth in BALB/c mice bearing the lacking indicated genes treated with PD‐1 antibody. Each line represents one mouse.

3. F

   Survival analysis of tumors with the lacking indicated genes and control were observed in CT26 colon cancer. n, the numbers of mice. Data are mean ± SEM of the indicated number of mice in each group. **P < 0.01 by Student's t‐tests.

Source data are available online for this figure.

Figure 4. Tumor cells with knockout of Mettl3 or Mettl14 exhibit enhanced response to IFNγ

1. A

   Cellular proliferation analysis of Mettl3‐ or Mettl14‐depleted and control CT26 cells treated with indicated combinations of cytokines for 48 h. The mean ± SD of three replicates is shown. *P < 0.05; **P < 0.01 by Student's t‐tests.

2. B

   BALB/c mice bearing Mettl3‐ or Mettl14‐deficient and control tumors were treated with IFNγ‐blocking antibody and PD‐1 antibody as indicated. Tumor size was measured over time. n, the numbers of mice. Data are mean ± SEM of the indicated number of mice. *P < 0.05; **P < 0.01; ***P < 0.001 by Student's t‐tests.

3. C

   Quantitative RT–PCR was performed to identify transcriptional changes of the IFN‐γ response gene expression (n = 3). Data are shown as the relative fold change (RFC, color coded bar).

4. D, E

   mRNA stability of Stat1 and Irf1 were measured by qRT–PCR in tumor cells treated with IFN‐γ and actinomycin D. Mean ± SD of n = 3. **P < 0.01; ***P < 0.001 by Student's t‐tests.

5. F

   Validation the effect of knockout of Ythdf1‐3 using Western blotting, Gapdh served as a loading control.

6. G

   qPCR analysis of Stat1 and Irf1's expression in the indicated depletion of CT26 cells with/without stimulation of IFN‐γ. Mean ± SD of n = 3. *P < 0.05 by Student's t‐tests.

7. H

   Western blot analysis of Mettl3 and Mettl14 in overexpressed CT26 cells. Gapdh served as a loading control for each.

8. I, J

   qPCR analysis of the mRNA stability of Stat1 and Irf1 in the indicated CT26 cells treated with IFN‐γ and actinomycin D. Mean ± SD of n = 3. *P < 0.05; **P < 0.01; ***P < 0.001 by Student's t‐tests.

Source data are available online for this figure.

Figure 5. The negative correlation of METTL3, METTL14, and STAT1 in human pMMR‐MSI‐L CRC colon tissue

1. A, B

   The protein level of STAT1 was negatively correlated with METTL3 and METTL14 in human pMMR‐MSI-L CRC colon tissues (r ² = −3.2477 for METTL3, r ² = −2.7491 for METTL14). Each dot represents one tumor tissue.

2. C

   Schematic showing the functional and molecular mechanisms of Mettl3 and Mettl14 in antitumor immunotherapy.