The MLL3/GRHL2 complex regulates malignant transformation and anti-tumor immunity in squamous cancer

Abstract

Upper aerodigestive squamous cell carcinoma (UASCC) presents significant challenges in clinical management due to its aggressive nature. Here, we elucidate the role of MLL3 mutations as early, clonal genomic events in UASCC tumorigenesis, highlighting their role as foundational drivers of cancer development. Utilizing CRISPR-edited, cross-species organoid modeling, we demonstrate that loss of MLL3 contributes to early squamous neoplastic evolution. Furthermore, we identify an MLL3/GRHL2 protein complex that regulates the UASCC epigenome, particularly impacting immune response pathways. Notably, a novel MLL3/GRHL2-IRF1 axis promotes the expression of Th1 chemokines, enhancing anti-tumor immunity by facilitating T cell infiltration into the tumor microenvironment. Consequently, MLL3 regulates the in vivo efficacy of immune checkpoint blockade (ICB) therapy, corroborated by the strong association between MLL3 expression and human patients' clinical response to ICB therapy. Our work underscores the significance of MLL3 in UASCC pathogenesis and highlights the interplay between MLL3/GRHL2 and immune response pathways as potential therapeutic targets for UASCC treatment.

Graphical abstract

Figure 1.

MLL3 mutations occur early during UASCC tumorigenesis. (A and B) Phylogenetic trees of two representative UASCC cases, reproduced using published data from us (Hao et al., 2016) and others (Chen et al., 2017; Yan et al., 2019). T1/2/3/5 were 4 intratumoral samples of the primary tumor while L1/2/3 were 3 lymph node metastasis lesions. (C) Column charts summarizing the mutational status of all detected MLL3 mutations. (D) Kaplan–Meier curves of overall survival and disease-free survival upon MLL3 mRNA expression with TCGA HNSCC patient survival. (E) Co-occurrence of mutations/deletions of MLL3, TP53, and CDKN2A in the TCGA ESCC and HNSCC datasets, plotted by the cBioportal tool (Cerami et al., 2012).

Figure S1.

Knockdown of MLL3 accelerates tumorigenesis. (A) Kaplan–Meier plots using the mean expression level of MLL3 as the statistical cutoff. (B and C) Sanger sequencing showing genomic mutations introduced to either human or murine TP53 (upper) and CDKN2A (lower) genes. (D) qRT-PCR data of MLL3 mRNA expression level following knockdown using shRNA in murine TP53/CDKN2A^DKO organoids (n = 3 biological replicates). (E) Representative images of murine TP53/CDKN2A^DKO organoids with and without MLL3 knockdown. Scale bar = 20 μm. (F) Quantification of organoid size, as measured by ImageJ. (G) Proliferation rate of organoids at the indicated time points (n = 3 biological replicates). (H and I) Representative images, scale bar = 50 μm, and (I) quantification of IF staining for Ki-67 in murine TP53/CDKN2A^DKO organoids with and without MLL3 knockdown (n = 3 biological replicates). *P < 0.05; **P < 0.01; and ***P < 0.001.

Figure 2.

Loss of MLL3 drives early neoplastic evolution in a TP53/CDKN2A DKO oral organoid model. (A) A schematic plot, generated using BioRender, showing the development of CRISPR-edited, cross-species organoid models for oral neoplastic evolution (TP53/CDKN2A^DKO) and subsequent knockdown of MLL3 expression. (B) qRT-PCR verification of the knockdown of MLL3 mRNA expression (n = 3 biological replicates). (C) Representative images of organoids from indicated groups at 40× magnification, scale bar = 20 μm. (D) Quantification of organoid size, as measured by ImageJ. (E) WST-1 assay to detect cell proliferation rate of organoids (n = 3 biological replicates). (F and G) Representative images, scale bar = 50 μm (F) and quantification of IF staining for Ki-67 in human organoids (n = 3 biological replicates) (G). (H) Individual tumor growth curves of control (TP53/CDKN2A^DKO + scramble sgRNA) versus MLL3-knockout organoids (TP53/CDKN2A^DKO + MLL3 knockout) (n = 5 biological replicates). (I) Kaplan–Meier plot showing the survival of mice from the two groups (n = 5 biological replicates). *P < 0.05; **P < 0.01; ***P < 0.001.

Figure 3.

MLL3 inhibits cancer cell proliferation and interacts with GRHL2. (A) Protein levels of MLL3 following siRNA transfection for 48 h in KYSE510 and KYSE180 cell lines. (B) Quantification of colonies measured by absorbance rate of 480 nm (n = 5 biological replicates). (C) Protein levels of MLL3 after viral infection of shRNA into KYSE510 and KYSE180 cell lines. (D) Quantification of colonies in MLL3-silenced versus control KYSE510 and KYSE180 cell lines (n = 5 biological replicates). (E and F) Relative mRNA and protein levels of MLL3 following its knockdown in MOC1 and MOC22 cell lines, measured by qRT-PCR and western blot, respectively. (G) Quantification of colonies in MLL3-silenced MOC1 and MOC22 cell lines (n = 5 biological replicates). (H) Table of top enriched transcription factor–binding motifs in MLL3 ChIP-seq peak regions. (I) Venn diagram displaying the number of overlapped peaks between GRHL2 and MLL3 ChIP-seq in KYSE510 cell line. (J) Co-IP assay by pulling down using the anti-MLL3 antibody followed by detection of GRHL2 protein by western blot. *P < 0.05; **P < 0.01; ***P < 0.001. Co-IP: co-immunoprecipitation. Source data are available for this figure: SourceData F3.

Figure S2.

MLL3 and GRHL2 co-bind large numbers of genomic loci and regulate immune response pathways. (A) Quantification of colonies after knockdown of MLL3 (n = 5 biological replicates) in A253 and FADU cell lines. (B) Pie charts showing genomic distribution of MLL3- and GRHL2-binding region in KYSE510 cell line. (C) Heatmap of ChIP-seq peaks of MLL3, GRHL2, H3K27ac, and H3K4me1. (D and E) Gene set enrichment analysis (GSEA) of either MLL3 KD or GRHL2 KD RNA-seq. *P < 0.05; **P < 0.01; and ***P < 0.001.

Figure 4.

MLL3/GRHL2 co-regulate immune response genes. (A) A volcano plot of up- and downregulated genes from RNA-seq in MLL3 knockdown cells. Genes having binding peaks of MLL3 are marked as red dots, otherwise black dots. A similar volcano plot for GRHL2 knockdown is at the lower panel. (B) Bar graphs of the fraction of different groups of genes with binding peaks of MLL3 (left) or GRHL2 (right). (C) A Venn diagram illustrating shared downregulated genes between MLL3 knockdown and GRHL2 knockdown. A pie chart showing MLL3- and/or GRHL2-binding peaks on the 505 overlapped downregulated genes. (D) Top enriched signaling pathway of the 505 genes. (E) Gene set enrichment analysis (GSEA) plots showing the enrichment of IFN signaling in the RNA-seq upon knockdown of either MLL3 or GRHL2. P values were adjusted for multiple comparisons.

Figure S3.

MLL3/GRHL2 knockdown accelerates tumor growth and reduces T cell infiltration. (A) Kaplan–Meier survival plots of control and MLL3-knockdown groups in MOC22 xenografts models (n = 3 biological replicates). Mice were sacrificed when tumor reached 1.5 cm. (B) Gating strategy for flow cytometry of Fig. 5, B and D. The cell population was first gated on SSC-A/FSC-A and FSC-H/FSC-A parameters, and then Zombie Violet dye was used to exclude dead cells. Next, CD45⁺CD3⁺-positive T cells were gated to isolate the T cell population. Finally, we measured the proportions of CD8⁺ and CD4⁺ T cells using the following equation: $\text{CD}{8}^{+}\mathrm{T}\hspace{0.17em}\text{cell}(\%)=\frac{\text{Number}\hspace{0.17em}\text{of}\hspace{0.17em}\text{CD}{8}^{+}\text{CD}{4}^{-}\hspace{0.17em}\mathrm{T}\hspace{0.17em}\text{cells}}{\text{Number}\hspace{0.17em}\text{of}\hspace{0.17em}\text{live}\hspace{0.17em}\text{cells}}\times 100$ and $\text{CD}{4}^{+}\hspace{0.17em}\mathrm{T}\hspace{0.17em}\text{cell}\hspace{0.17em}(\%)=\frac{\text{Number}\hspace{0.17em}\text{of}\hspace{0.17em}\text{CD}{8}^{-}\text{CD}{4}^{+}\mathrm{T}\hspace{0.17em}\text{cells}}{\text{Number}\hspace{0.17em}\text{of}\hspace{0.17em}\text{live}\hspace{0.17em}\text{cells}}\times 100$.

Figure 5.

Loss of MLL3/GRHL2 attenuates T cell infiltration in the tumor microenvironment. (A) Schematic illustration of the workflow for the investigation of infiltrated T cell infiltration in syngeneic mouse models. (B and D) FACS plots of Zombie⁻CD45⁺CD3⁺CD4⁻CD8⁺T cells and Zombie⁻CD45⁺CD3⁺CD4⁺CD8⁻ T cells in indicated samples. The left-most panel of B was also presented as the last gating plot (CD8 versus CD4) in Fig. S3 B. (C and E) Bar graph of the percentage of CD8⁺ T cells and CD4⁺ T cells among live cells (n = 5 for each group). (F) IF staining of infiltrated CD8⁺ T cells in the xenograft tumors and bar graph of quantification of CD8⁺ T cell (n = 3 biological replicates). Scale bar = 100 μm. (G) Schematic illustration of the workflow of T cell migration assay using transwell system. Sorted CD8⁺ T cells were cultured for 48 h using conditioned medium harvested from either MLL3 KD or GRHL2 KD cell line. (H and I) The number of migrated CD8⁺ T cells into the bottom counted by hemocytometer (n = 3 biological replicates). *P < 0.05; **P < 0.01; ***P < 0.001.

Figure 6.

MLL3/GRHL2 regulate IFN signaling through IRF1. (A) Schematic illustration of IFN signaling pathways. (B) Venn diagram of enriched genes between IFN signaling and IFNγ signaling in RNA-seq data following knockdown of either MLL3 or GRHL2. (C) Heatmap of mRNA expression of 12 overlapped genes upon knockdown of either MLL3 or GRHL2. (D) Protein levels of IRF1 and pSTAT1 in MLL3 or GRHL2 knockdown cells treated with IFNγ (10 ng/ml). (E) Relative mRNA levels of IRF1 upon knockdown of either MLL3 or GRHL2. (F) Integrative genomics viewer (IGV) tracks and line plots showing respective ChIP-seq profiles and H3K27ac high-throughput (HI)-ChIP loops at the locus of IRF1 gene. (G) Luciferase reporter assay after knockdown of either MLL3 or GRHL2. NC-B, negative control pGL3-basic vector; Pro-B, pGL3-basic with IRF1 promoter; NC-P, negative control pGL3-promoter vector; E1-P, pGL3-promoter with enhancer-1; and E2-P, pGL3-promoter with enhancer-2. (H) H3K27ac ChIP-qPCR of E1 and E2 regions in control and MLL3 knockout cells. (I) Heatmap of mRNA expression of 12 genes after silencing IRF1 in the presence of IFNγ (10 ng/ml). (J) Protein levels of IRF1 and pSTAT1 upon knockdown of IRF1 in the absence or presence of IFNγ (10 ng/ml). (K) Protein levels of pSTAT1 after overexpression of IRF1. n = 3 biological replicates for panels C, E, and G–I. *P < 0.05; **P < 0.01; and ***P < 0.001. Source data are available for this figure: SourceData F6.

Figure S4.

MLL3/GRHL2 directly regulates IRF1 expression levels. (A) Heatmap of 12 IFNγ-related genes after knockdown of either MLL3 or GRHL2 (n = 3 biological replicates). (B) Protein levels of IRF1 and p-STAT1 after silencing of either MLL3 or GRHL2 with treatment of IFNγ (10 ng/ml) for 48 h in FADU cell line. (C) Expression levels of IRF1 following dCas9-KRAB–mediated inhibition of either E1 or E2 in A253 and KYSE510 cell lines. Two independent sgRNAs were tested for each enhancer (n = 3 biological replicates). (D and E) Secretion levels of CXCL9 (D) and CXCL10 (E) were measured by the ELISA assay. NC, negative control; E1-1, sgRNA1 for enhancer-1; E1-2, sgRNA2 for enhancer-1; E2-1, sgRNA1 for enhancer-2; and E2-2, sgRNA2 for enhancer-2 (n = 3 biological replicates). (F) Heatmap of 12 IFN-related genes after silencing of IRF1 and treating IFNγ (10 ng/ml) in FADU cell lines (n = 3 biological replicates). *P < 0.05; **P < 0.01; and ***P < 0.001. Source data are available for this figure: SourceData FS4.

Figure 7.

The MLL3/GRHL2-IRF1 axis regulates the expression of chemokines CXCL9 and CXCL10. (A–C) Relative mRNA expression levels of CXCL9/10 in indicated cell line and organoid samples upon silencing of either MLL3 or GRHL2 in the presence of IFNγ (10 ng/ml). (D) Protein levels of IRF1 after overexpression of IRF1. (E) mRNA expression levels of CXCL9 and CXCL10 upon overexpression of IRF1. (F) ELISA assay of CXCL9/10 upon knockdown of either MLL3 or GRHL2 in the absence or presence of IFNγ (10 ng/ml). (G) IRF1 ChIP-qPCR in the absence or presence of IFNγ (10 ng/ml). n = 3 biological replicates for A–C and E–G. *P < 0.05; **P < 0.01; and ***P < 0.001. Source data are available for this figure: SourceData F7.

Figure S5.

Downregulation of MLL3/GRHL2 decreases CXCL9/10 levels through regulation of IRF1 expression. (A) Relative mRNA expression levels of CXCL9/10 after knockdown of either MLL3 or GRHL2 in the presence of IFNγ (10 ng/ml) in murine cell lines (n = 3 biological replicates). (B) Relative mRNA expression of IRF1 and CXCL9/10 upon knockdown of IRF1 in A253, FADU, KYSE510, and KYSE180 cell lines (n = 3 biological replicates). All samples were treated with IFNγ (10 ng/ml). (C) Expression levels of IRF1, CXCL9, and CXCL10 in scramble and MLL3-knockdown xenograft samples. (n = 5 for each group). (D and E) ELISA assay of CXCL9 and CXCL10 levels in either MLL3 or GRHL2 silenced FADU, KYSE180, MOC1, and MOC22 cell lines (n = 3 biological replicates). (F and G) CXCL9 and CXCL10 protein levels were detected by ELISA assay (n = 3 biological replicates). (H) Quantification of migrated CD8⁺ T cells in the bottom well in a transwell assay (n = 3 biological replicates). (I) Cell proliferation rates upon IRF1 knockdown in murine oral organoids (TP53/CDKN2A^DKO) or IRF1 overexpression in MOC1 cells (n = 3 biological replicates). (J) Quantification of migrated CD8⁺ T cell into bottom well in a transwell assay following anti-CXCR3 treatment (100 ng/ml) (n = 3 biological replicates). (K) Prediction of transcription factor (TF)-binding sites in CXCL9 and CXCL10 using JASPER. *P < 0.05; **P < 0.01; and ***P < 0.001.

Figure 8.

MLL3 regulates the efficacy of ICB therapy in HNSCC murine models. (A) Schematic illustration of the workflow of anti-PD1 treatment using a syngeneic orthotopic HNSCC model. (B) Kaplan–Meier survival plot of orthotopic models. Mice were euthanized due to either maximum tumor volume (a diameter of 15 mm in any direction) or tumor-associated morbidity. (C) Tumor growth curves for each group of mice (n = 5 mice in each group). (D) Flow cytometry measuring the fraction of tumor-infiltrated CD8⁺ T cells. (E) Box plots of the percentage of tumor-infiltrated CD8⁺ T cells and CD8⁺ IFNγ⁺ T cells among live cells in each group (n = 5 mice in each group). ***P < 0.001. (F) Kaplan–Meier survival plot of a subcutaneous syngeneic model with a similar treatment plan. (G) Individual tumor growth curves (n = 6 tumors in each group). (H and I) Survival analyses of the OAK/POPLAR cohorts of lung SCC patients treated with either ICB or chemotherapy. Patients were classified based on the mean value of MLL3 expression.