KMT2C deficiency promotes small cell lung cancer metastasis through DNMT3A-mediated epigenetic reprogramming

Abstract

Small cell lung cancer (SCLC) is notorious for its early and frequent metastases, which contribute to it as a recalcitrant malignancy. To understand the molecular mechanisms underlying SCLC metastasis, we generated SCLC mouse models with orthotopically transplanted genome-edited lung organoids and performed multiomics analyses. We found that a deficiency of KMT2C, a histone H3 lysine 4 methyltransferase frequently mutated in extensive-stage SCLC, promoted multiple-organ metastases in mice. Metastatic and KMT2C-deficient SCLC displayed both histone and DNA hypomethylation. Mechanistically, KMT2C directly regulated the expression of DNMT3A, a de novo DNA methyltransferase, through histone methylation. Forced DNMT3A expression restrained metastasis of KMT2C-deficient SCLC through repressing metastasis-promoting MEIS/HOX genes. Further, S-(5'-adenosyl)-L-methionine, the common cofactor of histone and DNA methyltransferases, inhibited SCLC metastasis. Thus, our study revealed a concerted epigenetic reprogramming of KMT2C- and DNMT3A-mediated histone and DNA hypomethylation underlying SCLC metastasis, which suggested a potential epigenetic therapeutic vulnerability.

Extended Data Fig. 1 |. Generating primary and orthotopic SCLC with metastases with genome edited lung organoids in mice.

(a) Representative bright-field and fluorescent images of lung organoids transduced with V2TC-sgRNAs and Myc. Three independent repeats. Scale bar, 200 μm. (b) T7 endonuclease 1 (T7e1) assays on Rb1 or Trp53 using infected lung organoid. Cleaved bands were pointed by arrowheads. Three independent repeats. (c) Representative IHC staining of KI67 of lung sections of the PRM mice (n = 4 mice). Scale bar, 20 μm. (d) Representative bright-field and fluorescent images showing metastatic lesions in the lymph nodes of the PRM mice (n = 4 mice). Scale bar, 2 mm. (e) Representative H&e staining of the lymph nodes (n = 4 mice). Scale bar, 20 μm. (f) Representative TTF1 staining of the lymph nodes (n = 4 mice). Scale bar, 20 μm. (g) Representative pictures showing staining of ASCL1, CHGA, SYP and NeUROD1 of liver sections of the PRM mice (n = 4 mice). Scale bar, 40 μm. (h) Representative pictures showing staining of ASCL1, CHGA, SYP and NeUROD1 of the PRM mice’s lymph node (n = 4 mice). Scale bar, 40 μm.

Extended Data Fig. 2 |. Single-cell transcriptome analyses of primary and metastatic SCLC cells.

(a) The ATAC-Seq plot showing signals at the TSS of the PRM and PRM-Met SCLC cells. (b) The bar graph showing the numbers of differential chromatin accessible peaks in the PRM and PRM-Met tumor cells. (c) Genome-wide distribution of chromatin open peaks (left) and close peaks (right) in PRM-Met compared to PRM, measured by ATAC-seq analyses. (d) The dot plot showing the marker genes of each cell type. (e) The openTSNe maps showing the expression distributions of Chga, Ddc and Ncam1. (f) The Alluvial plot showing the composition variation of cell populations in PRM and PRM-Met. (g) The URD map showing the CDX metastasis score projected on the metastasis trajectory. (h) The Kaplan-Meier survival curves of SCLC patients with low and high mouse metastatic scores. p-value was calculated by log-rank test (n = 61, low; n = 16, high). (i) The dynamically expressed genes and enriched gene signatures on the metastasis trajectory. (j) The Kaplan-Meier survival curves of SCLC patients with low and high Module I gene signatures. p-value was calculated by log-rank test (n = 7, low; n = 70, high).

Extended Data Fig. 3 |. KMT2C deficiency in SCLC metastases and its function in lung organoids.

(a) The Venn plot showing the mutation frequency of epigenetic regulatory genes in Module I. (b) The Ridge plot showing gene ontology enrichment in the Ascl1 + primary SCLC cells, analyzed by GSeA. (c) The Kaplan-Meier survival curves of SCLC patients with low and high expressions of KMT2C. Calculated by log-rank test. (n = 42, low; n = 35, high) (d) Representative western blotting pictures of H3K4me1 and H3K4me2 in the PRM and PRM-Met SCLC cells. Three independent repeats. (e) Representative staining of KMT2C, H3K4me1and H3K4me2 of the PRM primary (n = 3) (left) and PRM-Met (n = 3) liver sections (right). Scale bar, 40 μm. (f) The levels of H3K4me1 bound at the enhancers in the PRM and PRM-Met tumor cells, measured by the CUT&Tag analyses. (g) Genome-wide distribution of the H3K4me1 up-regulated genes (top) and down-regulated genes (bottom) in PRM-Met compared to PRM, measured by CUT&Tag analyses. (h) T7 endonuclease 1 (T7e1) assays showing the mutations of Kmt2c in the premalignant PRM organoids. Three independent repeats. Cleaved bands were pointed by arrowheads. (i) Dynamics of relative expression levels of Cyp2f2, Notch1 and Mki67 on the normal development (black line) and malignant transformation (purple line) trajectories.

Extended Data Fig. 4 |. Pathological analysis of the metastases in the PrM and PrMK mice.

(a) Representative H&e (left) and TTF1 staining (right) of the lung sections of the PRMK mice (n = 4 mice). Scale bar, 200 μm. (b) Representative staining of ASCL1, CHGA, NeUROD1 and SYP of the lung (top) and liver (bottom) sections of the PRMK mice (n = 4 mice). Scale bar, 20 μm. (c) Representative images of the lymph nodes (left) and kidney (right) of the PRM (n = 4) (top) and PRMK (n = 4) (bottom) mice. Scale bar, 2 mm. (d) Representative H&e (top) and TTF1 staining (bottom) of the PRM (n = 4) and PRMK (n = 4) mice’s lymph nodes (left) and kidney (right) sections. Scale bar, 200 μm. (e) The −ΔΔCT value of SCLC marker genes Ascl1, Ncam1, Chga and Syp in the CTC of PRM mice, measured by RT-qPCR (n = 3 technical replicates). (f) The −ΔΔCT value of SCLC marker genes Ascl1, Ncam1, Chga and Syp in the CTC of PRMK mice, measured by RT-qPCR (n = 3 technical replicates).

Extended Data Fig. 5 |. The histone methylation in the PrM and PrMK SCLC.

(a) Representative staining of KMT2C, H3K4me1 and H3K4me2 of the lung sections of PRM (n = 3 mice) (top) and PRMK (n = 3 mice) (bottom). Scale bar, 40 μm. (b) Representative western blotting pictures showing the H3k4me1, H3k4me2, H3K4me3, H3K9me3, H3K27me3, H3K27ac and H3K36me3 levels in the PRM and PRMK SCLC cells. Three independent repeats. (c) The IGV plots showing the mutations and expression levels of Kmt2c in the PRM and PRMK SCLC.

Extended Data Fig. 6 |. The epigenetic reprogramming in SCLC with Kmt2c loss.

(a) The levels of H3K4me1 bound at the enhancer (left) and TSS (right) in the PRM and PRMK SCLC cells, measured by CUT&Tag analyses. (b) Genome-wide distribution of the H3K4me1 up-regulated genes (left) and down-regulated genes (right) in PRMK compared to PRM, measured by CUT&Tag analyses. (c) The levels of H3K4me2 bound at the TSS in the PRM and PRMK SCLC cells, measured by CUT&Tag analyses. (d) Genome-wide distribution of the H3K4me2 up-regulated genes (left) and down-regulated genes (right) in PRMK compared to PRM, measured by CUT&Tag analyses. (e) The levels of H3K4me3 bound at the TSS in the PRM and PRMK SCLC cells, measured by CUT&Tag analyses. (f) Genome-wide distribution of the H3K4me3 up-regulated genes (left) and down-regulated genes (right) in PRMK compared to PRM, measured by CUT&Tag analyses. (g) The levels of ATAC bound at the TSS in the PRM and PRMK SCLC cells, measured by CUT&Tag analyses. (h) Genome-wide distribution of the ATAC open genes (top) and close genes (bottom) in PRMK compared to PRM. (i) The levels of KMT2C bound at the peaks of gene body in the PRM, PRM-Met and PRMK tumor cells, measured by the CUT& Tag analyses. (j) Genome-wide distribution of the KMT2C binding peaks in PRM, measured by CUT&Tag analyses.

Extended Data Fig. 7 |. The effect of epigenetic reprogramming on gene expressions in SCLC with Kmt2c loss.

(a) The Venn diagram showing overlapping of the H3K4me2 down-regulated genes and the KMT2C down-regulated genes in PRMK compared to the PRM SCLC cells. p-value was calculated by a hypergeometric test. (b) Heatmap showing the differential expressed genes in the PRM and PRMK organoids. (c) The Venn diagram showing overlapping of the chromatin close genes and those downregulated in the PRMK cells compared to the PRM cells. p-value was calculated by a hypergeometric test. (d) The Venn diagram showing overlapping of the down-regulated genes and those with reduced H3K4me1 in the PRMK compared to the PRM cells. p-value was calculated by a hypergeometric test. (e) The Venn diagram showing overlapping of the H3K4me1 up-regulated genes in the PRM metastasis cells compared to the PRM primary cells and the H3K4me1 up-regulated genes in the PRMK cells compared to the PRM primary. p-value was calculated by a hypergeometric test. (f) The Venn diagram showing overlapping of the chromatin open genes in the PRM metastasis cells compared to the PRM primary cells and the chromatin open genes in the PRMK cells compared to the PRM primary. p-value was calculated by a hypergeometric test. (g) The Venn diagram showing overlapping of the chromatin open genes and those upregulated in the PRMK cells compared to the PRM cells. p-value was calculated by a hypergeometric test. (h) GSeA showing positive enrichment of the PRMK up-regulated gene set in the metastasis SCLC, compared to primary tumor cells in the CCLE cohort.

Extended Data Fig. 8 |. Identifying DNMT3A as a downstream target of KMT2C in SCLC.

(a) The Venn diagram showing overlapping of the down-regulated genes in RNA expression, chromatin accessibility, H3K4me1 in PRM-Met compared to PRM. p-value was calculated by a hypergeometric test. (b) The scatter plot showing the correlation between the relative expression levels of KMT2C and DNMT3A in multiple SCLC cohorts. (c) The Kaplan-Meier survival curves of SCLC patients with high or low expressions of DNMT3A. (n = 55, low; n = 22, high).

Extended Data Fig. 9 |. KMT2C loss gave rise to DNA hypomethylation in SCLC.

(a) The density plot showing all C sites’ methylation levels in the genome of the PRM and PRMK SCLC cells. (b) The 5mC levels of the CpG regions in the PRM and PRMK cells. (c) The density plot showing all C sites’ methylation levels in the genome of the PRM and PRM-Met SCLC cells. (d) The density plot showing the methylation levels of the CpG regions in the CCLE primary and metastasis SCLC cells. (e) The density plot showing the methylation levels of the CpG regions in the CCLE SCLC cells with or without KMT2C mutations. (f) The scatter plot showing the differentially methylated regions (DMRs) in the PRM and PRMK SCLC cells. (g) The scatter plot showing the differentially methylated sites (DMS) in the KMT2C-WT and KMT2C-Mut CCLE SCLC samples. (h) Pie charts showed the genomics region annotation (top) and CpGs subtypes (bottom) of hyper- and hypo- DMRs in PRMK compared to PRM. (i) The Venn diagram showed overlapping of the hypomethylated genes and open genes in PRMK compared to PRM (left); p-value was calculated by a hypergeometric test. The box plot displayed the normalized expression levels of 171 overlap genes of hypomethylated genes and open genes in PRM and PRMK (right), The box bounds the interquartile range divided by the median, with the whiskers extending to a maximum of 1.5 times the interquartile range beyond the box. p-value was calculated by Wilcoxon signed-rank test. (j) Dot blotting showed the expression of 5mC in PRMK organoids with vector or DNMT3A overexpression (left). The statistics of DNA 5mC levels in PRMK cells (right). (mean ± SeM, n = 3). Calculated by Student’s t-test, two-sided. All p-value, *, p < 0.05.

Extended Data Fig. 10 |. SAM treatment for KMT2C deficient SCLC.

(a) The dot blotting (left) and the relative levels (right) of 5mC in the PRMK organoids treated with vehicle or SAM. (mean ± SD, n = 3). *, p < 0.05, Calculated by Student’s t-test, two-sided. (b) The representative western blotting pictures showing the H3k4me1 and H3k4me2 levels in the PRMK organoids treated with vehicle or SAM. Three independent repeats. (c) The representative flow cytometry plots of CTCs in the peripheral blood of PRMK mice treated with vehicle or SAM. (d) Heatmap showing the differential pathways in SAM treated PRMK SCLC cells compared to those treated with vehicle, measured by RNA-seq analyses.

Fig. 1 |. Generating primary and orthotopic SCLC with metastases with genome-edited lung organoids in mice.

a, Top, schematic of the organoid-based strategy for generating primary and orthotopic SCLC in immunocompetent mice. Bottom, schematic of the constructs for expressing sgRNAs and Myc; MSCV, murine stem cell virus; IReS, internal ribosome entry site. b, Representative bioluminescent images of mice transplanted with PRM organoids at 3 months after transplantation. c, Representative micro-CT images of PRM-recipient mice. d, Left, representative brightfield (BF) image of the lungs of a PRM recipient. The lesion on the left lung is indicated, and a representative fluorescent image of the lungs of the PRM recipient is shown on the right; (n = 4 mice). e, Representative hematoxylin and eosin (H&e) staining of a lung section from a PRM mouse (n = 4 mice). f, Representative staining of TTF1 in a lung section from a PRM mouse (n = 4 mice). g, Representative IHC staining of ASCL1, SYP, CHGA and NeUROD1 in lung sections of PRM mice (n = 4 mice). h, Representative brightfield (left) and fluorescent (right) images showing metastatic lesions in the liver of a PRM mouse (n = 4 mice). i, Representative H&e staining of the liver (n = 4 mice). j, Representative TTF1 staining of the liver (n = 4 mice).

Fig. 2 |. A metastasis trajectory of SCLC revealed by single-cell transcriptome analyses.

a, Open t-distributed stochastic neighbor embedding (openTSNe) map of scRNA-seq analyses of SCLC from the lung and liver of the same PRM mouse; 7,925 cells and 3,513 cells were captured from the primary site and metastatic lesion (liver) of the same mouse, respectively. b, OpenTSNe plot showing the organ origins from PRM (lung) and PRM metastasis (PRM-Met; liver). c, OpenTSNe map showing expression levels of Ascl1 and Neurod1 in primary SCLC cells and metastatic SCLC cells. d, URD map showing the metastasis trajectory of Ascl1⁺ SCLC cells. e, URD map showing the CCLE metastasis score projected on the metastasis trajectory. f, Dynamically expressed gene modules on the metastasis trajectory. g, Alluvial plot showing the composition variation of three molecular subtypes defined by the dynamically expressed gene modules in limited-stage (LS) and extensive-stage (eS) SCLC in the Simpson cohort.

Fig. 3 |. KMT2C deficiency is associated with SCLC metastasis and gives rise to a premalignant population in lung organoids.

a, Bar graph showing the mutation frequencies of KMT2C in primary and metastatic SCLC from TCGA (left) and CCLE cohorts (right). b, Bar graph showing the relative expression levels of Kmt2c in mouse primary and metastatic (left) and human LS and eS (right) SCLC cells. Data are shown as mean ± s.e.; n = 2,365, primary; n = 13,798, metastatic; n = 10,000, LS SCLC; n = 2,000, eS SCLC. Significance was assessed by Wilcoxon signed-rank test. c, Top, Integrative Genomics Viewer (IGV) plot showing the distributions of ATAC-seq peaks in the Kmt2c locus in PRM and PRM-Met SCLC cells. Bottom, dynamics of relative expression levels of Kmt2c on metastasis trajectory. d, Scatter plot showing the correlation between the relative expression levels of KMT2C and the SCLC metastasis score in individuals with SCLC. e, Levels of H3K4me1 bound at enhancers in PRM and PRM-Met tumor cells, as measured by CUT&Tag analyses; kb, kilobases. f, Left, representative fluorescence images of sgScr and sgKmt2c lung organoid growth. Right, diameters of the lung organoids with sgScr or sgKmt2c. Data are shown as mean ± s.d.; n = 50, sgScr; n = 52, sgKmt2c. Significance was assessed by two-sided Student’s t-test. g, Uniform manifold approximation and projection (UMAP) plots showing populations and trajectories in the lung organoids with sgScr or sgKmt2c. The 2,794 cells and 2,794 cells were captured from the sgScr sample (n = 1) and sgKmt2c sample (n = 1), respectively. h, Dot plot showing the marker genes of each population. i, Stacked graph showing the percentages of cell populations in sgScr and sgKmt2c organoids. j, The dynamic expression levels of SCLC up (left) and metastasis up (right) gene signatures on the development (black line) and malignant differentiation (purple line) trajectories are shown; *P < 0.05, ****P < 0.0001.

Fig. 4 |. KMT2C deficiency promotes tumorigenesis and metastasis of SCLC.

a, Diameter quantification of premalignant PRM (n = 69), PRMK-shRNA (n = 64) and PRMK-sgRNA (n = 115) lung organoids. Data are shown as mean ± s.d. Significance was calculated by two-sided Student’s t-test. b, Number of premalignant PRM (n = 8), PRMK-shRNA (n = 6) and PRMK-sgRNA (n = 9) lung organoids. Data are shown as mean ± s.d. Significance was calculated by two-sided Student’s t-test. c, Representative bioluminescent images of mice transplanted with PRM, PRMK-shRNA or PRMK-sgRNA organoids at 3 months after transplantation. d, Luciferase fluorescence signal intensity of PRM (n = 6), PRMK-shRNA (n = 3) and PRMK-sgRNA (n = 5) mice. Data are shown as mean ± s.e.m. Significance was calculated by two-sided Student’s t-test. e, Kaplan–Meier survival curves of mice transplanted with PRM (n = 8), PRMK-shRNA (n = 7) and PRMK-sgRNA (n = 9) organoids. All curves were analyzed by log-rank test. f, Representative brightfield and fluorescence images of the lungs (top) and livers (bottom) of PRM, PRMK-shRNA and PRMK-sgRNA mice. g, Representative images of H&e staining of the livers of PRM (n = 4) and PRMK (n = 4) mice. h, Statistical graphs showing the lesion diameters (left) and number of metastases (right) in the livers of PRM and PRMK mice. Data are shown as mean ± s.d.; left: n = 66 PRM and n = 116 PRMK; right: n = 3). Significance was calculated by two-sided Student’s t-test. i, Numbers of metastatic lesions in the kidneys of PRM and PRMK mice (left) and percentages of PRM or PRMK mice with indicated numbers of organs with metastases (right). Data are shown as mean ± s.d.; left, n = 3. Significance was calculated by two-sided Student’s t-test. j, Representative flow cytometry plots showing the CD45⁻mCherry⁺ CTCs in the peripheral blood of PRM and PRMK mice. k, Percentage of CTCs in the peripheral blood of PRM and PRMK mice. P1 represents the CD45⁻ and mCherry⁺ CTC. Data are shown as mean ± s.e.m.; n = 3. Significance was calculated by two-sided Student’s t-test. l, Average number of axon-like protrusions per organoid. Data are shown as mean ± s.d.; n = 49. Significance was calculated by two-sided Student’s t-test. m, Representative images showing the morphology of the PRM (n = 4), PRMK-shRNA (n = 4) and PRMK-sgRNA (n = 4) SCLC organoids; *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

Fig. 5 |. epigenetic reprogramming in Kmt2c-mutant SCLC.

a, Levels of KMT2C bound at the center of peaks, levels of H3K4me1 bound at the enhancer and levels of H3K4me2, H3K4me3 and ATAC bound at the TSS in PRM and PRMK tumor cells, as measured by CUT&Tag analyses. b, Venn diagram showing overlapping of the H3K4me1 downregulated genes in PRM metastatic cells compared to in PRM primary cells and the H3K4me1 downregulated genes in PRMK cells compared to in PRM primary cells. The P value was calculated by a hypergeometric test. c, Venn diagram showing overlapping of the chromatin closed genes in PRM metastatic cells compared to in PRM primary cells and the chromatin close genes in PRMK cells compared to in PRM primary cells. The P value was calculated by a hypergeometric test. d, Levels of KMT2C bound at the center in PRM, PRM-Met and PRMK tumor cells, as measured by the CUT&Tag analyses. e, Venn diagram showing overlapping of H3K4me1 downregulated genes and KMT2C downregulated genes in PRMK cells compared to in PRM SCLC cells. The P value was calculated by a hypergeometric test. f, Venn diagram showing overlapping of the common genes with reduced H3K4me1 levels in both PRM-Met and PRMK cells and the KMT2C downregulated genes in PRMK compared to PRM SCLC cells. The P value was calculated by a hypergeometric test. g, Venn diagram showing overlapping of the common genes with closed chromatin accessibility in both PRM-Met and PRMK cells and the KMT2C downregulated genes in PRMK compared to PRM SCLC cells. The P value was calculated by a hypergeometric test.

Fig. 6 |. DNA hypomethylation mediates the functions of KMT2C deficiency in SCLC metastasis.

a, IGV plots showing H3K4me1, H3K4me2, H3K4me3 and KMT2C binding density on Dnmt3a in PRM, PRM-Met and PRMK cells. The significant variation regions are labeled. b, Dynamics of relative expression levels of Dnmt3a on metastasis trajectory. c, Relative expression levels of Dnmt3a in murine sgScr and sgKmt2c lung organoids (left; n = 2,794 cells sgScr and n = 2,794 cells sgKmt2c) and human primary and metastatic SCLC cells (right; n = 10,000 cells primary and n = 2,000 cells metastatic). Data are shown as mean ± s.e.m. The box bounds the interquartile range divided by the median, with the whiskers extending to a maximum of 1.5 times the interquartile range beyond the box. The P value was calculated by Wilcoxon signed-rank test. d, Dot blotting of DNA 5-methylcytosine (5mC) in PRM and PRMK SCLC cells (left). The statistics of DNA 5mC levels in SCLC cells is shown on the right. Data are shown as mean ± s.e.m.; n = 6. Significance was calculated by two-sided Student’s t-test. e, Density plot showing methylation levels at CpG sites in PRM and PRMK SCLC cells. f, Density plot showing methylation levels of CpG sites in PRM and PRM-Met SCLC cells. g, Pie chart showing the concordant percentage of upregulated genes with reduced 5mC levels in PRMK cells compared to in PRM cells. h, Numbers of organoids (left) and percentages of organoids (right) with axon-like protrusions with empty vector or Dnmt3a overexpression. Data are shown as mean ± s.d.; left, n = 4; right, n = 7. Significance was calculated by two-sided Student’s t-test. i, Representative bioluminescence images of mice with PRMK-vector or PRMK-Dnmt3a SCLC at days 7 and 14 after transplantation. j, Numbers of metastatic lesions in the livers of vector and Dnmt3a mice. Data are shown as mean ± s.d.; n = 3. Significance was calculated by two-sided Student’s t-test. k, Representative flow cytometry plots showing CD45⁻mCherry⁺ CTCs in mice with PRMK-vector SCLC or PRMK-Dnmt3a SCLC. l, Percentages of CTCs in the peripheral blood of mice with PRMK-vector or PRMK-Dnmt3a SCLC. Data are shown as mean ± s.d.; n = 4 mice. Significance was calculated by two-sided Student’s t-test; *P< 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

Fig. 7 |. Meis/Hox genes are downstream of the KMT2C–DNMT3A epigenetic reprogramming for SCLC metastasis.

a, Venn diagram showing overlapping of the hypomethylated genes in PRM-Met compared to PRM, PRMK compared to PRM and CCLE KMT2C mutant compared to CCLE KMT2C wild type. P values were calculated by a hypergeometric test. b, IGV plots showing the 5mC densities and RNA-seq peaks of the Meis2 gene bodies. c, IGV plots showing the 5mC densities and RNA-seq peaks of the Hoxb5 and Hoxb7 gene bodies. d, Relative expression levels of Meis2, Hoxb2, Hoxb3, Hoxb4, Hoxb5 and Hoxb7 in PRM and PRMK tumor organoids, as measured by real-time quantitative PCR (RT–qPCR). Data are shown as mean ± s.d.; n = 3 mice. Significance was calculated by two-sided Student’s t-test. e, Relative expression levels of Meis2, Hoxb2, Hoxb3, Hoxb5, Hoxb7 and Hoxb9 in PRM tumor organoids with sgScr and sgDnmt3a, as measured by RT–qPCR; n = 3 mice. f, Relative expression of Meis2, Hoxb5 and Hoxb7 in PRMK organoids with vector or Dnmt3a overexpression, as measured by RT–qPCR. Data are shown as mean ± s.d.; n = 3 organoids. Significance was calculated by two-sided Student’s t-test. g, Percentages of the sgScr and sgMeis2 PRMK SCLC organoids with axon-like protrusions (left) and the number of total sgScr and sgMeis2 PRMK organoids (right). Data are shown as mean ± s.d.; n = 4. Significance was calculated by two-sided Student’s t-test. h, Luciferase fluorescence signal intensities of mice with sgScr or sgMeis2 PRMK SCLC. Data are shown as mean ± s.e.m.; n = 4 mice. Significance was calculated by two-sided Student’s t-test. i, Percentages of sgScr and sgMeis2 PRMK mice with the indicated number of metastases. j, Gene set enrichment analysis showing negative enrichment of the epithelial–mesenchymal transition gene signature in sgMeis2 PRMK cells compared to those with sgScr; *P < 0.05, **P < 0.01, ***P < 0.001; NeS, normalized enrichment score; FDR, false discovery rate.

Fig. 8 |. SAM treatment reversed both H3K4 and DNA hypomethylation in KMT2C-deficient SCLC and restrained metastasis.

a, Growth curves of PRMK organoids treated with vehicle or SAM at the indicated concentrations over time. Data are shown as mean ± s.d.; n = 3 experiments. Significance was calculated by two-sided Student’s t-test. b, Percentages of PRMK SCLC organoids with axon-like protrusions after treatment with vehicle or 0.02 mM SAM. Data are shown as mean ± s.d.; n = 6 experiments. Significance was calculated by two-sided Student’s t-test. c, Representative brightfield images of PRMK SCLC organoids treated with vehicle (n = 5) or SAM (n = 5). d, Relative luciferase fluorescence signal intensities (SAM versus vehicle) of three pairs of SCLC mice at day 0 (D0) and day 7 (D7) after SAM or vehicle treatments. The box bounds the interquartile range divided by the median. e, Percentages of CTCs in SCLC treated with vehicle or SAM. Data are shown as mean ± s.d.; n = 4 mice. Significance was calculated by two-sided Student’s t-test. f, Representative fluorescence images of liver tissue from mice with PRMK SCLC treated with vehicle (left) or SAM (right); n = 3 mice. g, Number of metastatic liver lesions in the livers of three pairs of PRMK SCLC mice treated with vehicle or SAM at the end of treatments. h, Representative H&e staining of liver sections of PRMK SCLC mice treated with vehicle or SAM. i, Growth curves of SCLC organoids from two individuals treated with vehicle and SAM at the indicated concentrations over time. Data are shown as mean ± s.d.; three independent repeats were performed. Significance was calculated by two-sided Student’s t-test. j, Left, Venn diagram showing overlapping of the upregulated genes in SAM-treated PRMK SCLC cells compared to those treated with vehicle and H3K4me1 downregulated genes in PRMK cells compared to PRM cells. Right, Venn diagram showing overlapping of the downregulated genes in SAM-treated PRMK SCLC cells compared to those treated with vehicle and hypomethylated genes in PRMK cells compared to PRM cells. The P values were calculated by a hypergeometric test. k, Working model for KMT2C in SCLC tumorigenesis and metastasis; *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.