Research ArticleOncology Open Access | 10.1172/jci.insight.154120

Yuchen Xie,1,2 Merve Sahin,3,4 Toru Wakamatsu,1 Akane Inoue-Yamauchi,1 Wanming Zhao,1 Song Han,1 Amrita M. Nargund,1 Shaoyuan Yang,1 Yang Lyu,5 James J. Hsieh,5 Christina S. Leslie,3 and Emily H. Cheng1,6,7

1Human Oncology and Pathogenesis Program, Memorial Sloan Kettering Cancer Center (MSKCC), New York, New York, USA.

2Gerstner Sloan Kettering Graduate School of Biomedical Sciences, New York, New York, USA.

3Computational and Systems Biology Program, MSKCC, New York, New York, USA.

4Tri-Institutional Training Program in Computational Biology and Medicine, New York, New York, USA.

5Molecular Oncology, Department of Medicine, Washington University, St. Louis, Missouri, USA.

6Department of Pathology and Laboratory Medicine, MSKCC, New York, New York, USA.

7Weill Cornell Medical College, New York, New York, USA.

Address correspondence to: Emily H. Cheng, Memorial Sloan Kettering Cancer Center, 1275 York Avenue, Box 20, New York, New York 10065, USA. Phone: 646.888.3258. Email: chenge1@mskcc.org.

Find articles by Xie, Y. in: JCI | PubMed | Google Scholar |

1Human Oncology and Pathogenesis Program, Memorial Sloan Kettering Cancer Center (MSKCC), New York, New York, USA.

2Gerstner Sloan Kettering Graduate School of Biomedical Sciences, New York, New York, USA.

3Computational and Systems Biology Program, MSKCC, New York, New York, USA.

4Tri-Institutional Training Program in Computational Biology and Medicine, New York, New York, USA.

5Molecular Oncology, Department of Medicine, Washington University, St. Louis, Missouri, USA.

6Department of Pathology and Laboratory Medicine, MSKCC, New York, New York, USA.

7Weill Cornell Medical College, New York, New York, USA.

Address correspondence to: Emily H. Cheng, Memorial Sloan Kettering Cancer Center, 1275 York Avenue, Box 20, New York, New York 10065, USA. Phone: 646.888.3258. Email: chenge1@mskcc.org.

Find articles by Sahin, M. in: JCI | PubMed | Google Scholar

1Human Oncology and Pathogenesis Program, Memorial Sloan Kettering Cancer Center (MSKCC), New York, New York, USA.

2Gerstner Sloan Kettering Graduate School of Biomedical Sciences, New York, New York, USA.

3Computational and Systems Biology Program, MSKCC, New York, New York, USA.

4Tri-Institutional Training Program in Computational Biology and Medicine, New York, New York, USA.

5Molecular Oncology, Department of Medicine, Washington University, St. Louis, Missouri, USA.

6Department of Pathology and Laboratory Medicine, MSKCC, New York, New York, USA.

7Weill Cornell Medical College, New York, New York, USA.

Address correspondence to: Emily H. Cheng, Memorial Sloan Kettering Cancer Center, 1275 York Avenue, Box 20, New York, New York 10065, USA. Phone: 646.888.3258. Email: chenge1@mskcc.org.

Find articles by Wakamatsu, T. in: JCI | PubMed | Google Scholar |

1Human Oncology and Pathogenesis Program, Memorial Sloan Kettering Cancer Center (MSKCC), New York, New York, USA.

2Gerstner Sloan Kettering Graduate School of Biomedical Sciences, New York, New York, USA.

3Computational and Systems Biology Program, MSKCC, New York, New York, USA.

4Tri-Institutional Training Program in Computational Biology and Medicine, New York, New York, USA.

5Molecular Oncology, Department of Medicine, Washington University, St. Louis, Missouri, USA.

6Department of Pathology and Laboratory Medicine, MSKCC, New York, New York, USA.

7Weill Cornell Medical College, New York, New York, USA.

Address correspondence to: Emily H. Cheng, Memorial Sloan Kettering Cancer Center, 1275 York Avenue, Box 20, New York, New York 10065, USA. Phone: 646.888.3258. Email: chenge1@mskcc.org.

Find articles by Inoue-Yamauchi, A. in: JCI | PubMed | Google Scholar

1Human Oncology and Pathogenesis Program, Memorial Sloan Kettering Cancer Center (MSKCC), New York, New York, USA.

2Gerstner Sloan Kettering Graduate School of Biomedical Sciences, New York, New York, USA.

3Computational and Systems Biology Program, MSKCC, New York, New York, USA.

4Tri-Institutional Training Program in Computational Biology and Medicine, New York, New York, USA.

5Molecular Oncology, Department of Medicine, Washington University, St. Louis, Missouri, USA.

6Department of Pathology and Laboratory Medicine, MSKCC, New York, New York, USA.

7Weill Cornell Medical College, New York, New York, USA.

Address correspondence to: Emily H. Cheng, Memorial Sloan Kettering Cancer Center, 1275 York Avenue, Box 20, New York, New York 10065, USA. Phone: 646.888.3258. Email: chenge1@mskcc.org.

Find articles by Zhao, W. in: JCI | PubMed | Google Scholar

1Human Oncology and Pathogenesis Program, Memorial Sloan Kettering Cancer Center (MSKCC), New York, New York, USA.

2Gerstner Sloan Kettering Graduate School of Biomedical Sciences, New York, New York, USA.

3Computational and Systems Biology Program, MSKCC, New York, New York, USA.

4Tri-Institutional Training Program in Computational Biology and Medicine, New York, New York, USA.

5Molecular Oncology, Department of Medicine, Washington University, St. Louis, Missouri, USA.

6Department of Pathology and Laboratory Medicine, MSKCC, New York, New York, USA.

7Weill Cornell Medical College, New York, New York, USA.

Address correspondence to: Emily H. Cheng, Memorial Sloan Kettering Cancer Center, 1275 York Avenue, Box 20, New York, New York 10065, USA. Phone: 646.888.3258. Email: chenge1@mskcc.org.

Find articles by Han, S. in: JCI | PubMed | Google Scholar

1Human Oncology and Pathogenesis Program, Memorial Sloan Kettering Cancer Center (MSKCC), New York, New York, USA.

2Gerstner Sloan Kettering Graduate School of Biomedical Sciences, New York, New York, USA.

3Computational and Systems Biology Program, MSKCC, New York, New York, USA.

4Tri-Institutional Training Program in Computational Biology and Medicine, New York, New York, USA.

5Molecular Oncology, Department of Medicine, Washington University, St. Louis, Missouri, USA.

6Department of Pathology and Laboratory Medicine, MSKCC, New York, New York, USA.

7Weill Cornell Medical College, New York, New York, USA.

Address correspondence to: Emily H. Cheng, Memorial Sloan Kettering Cancer Center, 1275 York Avenue, Box 20, New York, New York 10065, USA. Phone: 646.888.3258. Email: chenge1@mskcc.org.

Find articles by Nargund, A. in: JCI | PubMed | Google Scholar

1Human Oncology and Pathogenesis Program, Memorial Sloan Kettering Cancer Center (MSKCC), New York, New York, USA.

2Gerstner Sloan Kettering Graduate School of Biomedical Sciences, New York, New York, USA.

3Computational and Systems Biology Program, MSKCC, New York, New York, USA.

4Tri-Institutional Training Program in Computational Biology and Medicine, New York, New York, USA.

5Molecular Oncology, Department of Medicine, Washington University, St. Louis, Missouri, USA.

6Department of Pathology and Laboratory Medicine, MSKCC, New York, New York, USA.

7Weill Cornell Medical College, New York, New York, USA.

Address correspondence to: Emily H. Cheng, Memorial Sloan Kettering Cancer Center, 1275 York Avenue, Box 20, New York, New York 10065, USA. Phone: 646.888.3258. Email: chenge1@mskcc.org.

Find articles by Yang, S. in: JCI | PubMed | Google Scholar

1Human Oncology and Pathogenesis Program, Memorial Sloan Kettering Cancer Center (MSKCC), New York, New York, USA.

2Gerstner Sloan Kettering Graduate School of Biomedical Sciences, New York, New York, USA.

3Computational and Systems Biology Program, MSKCC, New York, New York, USA.

4Tri-Institutional Training Program in Computational Biology and Medicine, New York, New York, USA.

5Molecular Oncology, Department of Medicine, Washington University, St. Louis, Missouri, USA.

6Department of Pathology and Laboratory Medicine, MSKCC, New York, New York, USA.

7Weill Cornell Medical College, New York, New York, USA.

Address correspondence to: Emily H. Cheng, Memorial Sloan Kettering Cancer Center, 1275 York Avenue, Box 20, New York, New York 10065, USA. Phone: 646.888.3258. Email: chenge1@mskcc.org.

Find articles by Lyu, Y. in: JCI | PubMed | Google Scholar |

1Human Oncology and Pathogenesis Program, Memorial Sloan Kettering Cancer Center (MSKCC), New York, New York, USA.

2Gerstner Sloan Kettering Graduate School of Biomedical Sciences, New York, New York, USA.

3Computational and Systems Biology Program, MSKCC, New York, New York, USA.

4Tri-Institutional Training Program in Computational Biology and Medicine, New York, New York, USA.

5Molecular Oncology, Department of Medicine, Washington University, St. Louis, Missouri, USA.

6Department of Pathology and Laboratory Medicine, MSKCC, New York, New York, USA.

7Weill Cornell Medical College, New York, New York, USA.

Address correspondence to: Emily H. Cheng, Memorial Sloan Kettering Cancer Center, 1275 York Avenue, Box 20, New York, New York 10065, USA. Phone: 646.888.3258. Email: chenge1@mskcc.org.

Find articles by Hsieh, J. in: JCI | PubMed | Google Scholar |

1Human Oncology and Pathogenesis Program, Memorial Sloan Kettering Cancer Center (MSKCC), New York, New York, USA.

2Gerstner Sloan Kettering Graduate School of Biomedical Sciences, New York, New York, USA.

3Computational and Systems Biology Program, MSKCC, New York, New York, USA.

4Tri-Institutional Training Program in Computational Biology and Medicine, New York, New York, USA.

5Molecular Oncology, Department of Medicine, Washington University, St. Louis, Missouri, USA.

6Department of Pathology and Laboratory Medicine, MSKCC, New York, New York, USA.

7Weill Cornell Medical College, New York, New York, USA.

Address correspondence to: Emily H. Cheng, Memorial Sloan Kettering Cancer Center, 1275 York Avenue, Box 20, New York, New York 10065, USA. Phone: 646.888.3258. Email: chenge1@mskcc.org.

Find articles by Leslie, C. in: JCI | PubMed | Google Scholar |

1Human Oncology and Pathogenesis Program, Memorial Sloan Kettering Cancer Center (MSKCC), New York, New York, USA.

2Gerstner Sloan Kettering Graduate School of Biomedical Sciences, New York, New York, USA.

3Computational and Systems Biology Program, MSKCC, New York, New York, USA.

4Tri-Institutional Training Program in Computational Biology and Medicine, New York, New York, USA.

5Molecular Oncology, Department of Medicine, Washington University, St. Louis, Missouri, USA.

6Department of Pathology and Laboratory Medicine, MSKCC, New York, New York, USA.

7Weill Cornell Medical College, New York, New York, USA.

Address correspondence to: Emily H. Cheng, Memorial Sloan Kettering Cancer Center, 1275 York Avenue, Box 20, New York, New York 10065, USA. Phone: 646.888.3258. Email: chenge1@mskcc.org.

Find articles by Cheng, E. in: JCI | PubMed | Google Scholar |

Published February 22, 2023 - More info

Published in Volume 8, Issue 4 on February 22, 2023
JCI Insight. 2023;8(4):e154120. https://doi.org/10.1172/jci.insight.154120.
© 2023 Leslie, et al. This work is licensed under the Creative Commons Attribution 4.0 International License. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/. Published February 22, 2023 - Version history
Received: August 16, 2021; Accepted: January 18, 2023 View PDF Abstract

SETD2, a H3K36 trimethyltransferase, is the most frequently mutated epigenetic modifier in lung adenocarcinoma, with a mutation frequency of approximately 9%. However, how SETD2 loss of function promotes tumorigenesis remains unclear. Using conditional Setd2-KO mice, we demonstrated that Setd2 deficiency accelerated the initiation of KrasG12D-driven lung tumorigenesis, increased tumor burden, and significantly reduced mouse survival. An integrated chromatin accessibility and transcriptome analysis revealed a potentially novel tumor suppressor model of SETD2 in which SETD2 loss activates intronic enhancers to drive oncogenic transcriptional output, including the KRAS transcriptional signature and PRC2-repressed targets, through regulation of chromatin accessibility and histone chaperone recruitment. Importantly, SETD2 loss sensitized KRAS-mutant lung cancer to inhibition of histone chaperones, the FACT complex, or transcriptional elongation both in vitro and in vivo. Overall, our studies not only provide insight into how SETD2 loss shapes the epigenetic and transcriptional landscape to promote tumorigenesis, but they also identify potential therapeutic strategies for SETD2 mutant cancers.

Graphical Abstract Introduction

SETD2 is an RNA polymerase II–associated (Pol II–associated) histone methyltransferase involved in the cotranscriptional methylation of H3K36 to generate H3K36me3 in the bodies of actively transcribed genes (1–5), and this process is important for transcriptional elongation, repression of cryptic transcription initiation, cotranscriptional RNA processing, and alternative splicing (1–7). In addition, SETD2-mediated H3K36me3 has been shown to be involved in DNA mismatch repair (8), DNA double-strand break repair by homologous recombination, and the maintenance of genome stability (9, 10). SETD2 is one of the most frequently mutated chromatin-modifying genes across different cancer types, with the highest mutation rate in clear cell renal cell carcinoma (ccRCC, 13%) followed by lung adenocarcinoma (9%) (11, 12). Based on The Cancer Genome Atlas Lung Adenocarcinoma (TCGA-LUAD) data set, SETD2 is the eighth most commonly mutated gene and the most frequently mutated epigenetic modifier in lung adenocarcinoma (12). The majority of SETD2 mutations identified in lung adenocarcinoma are truncating mutations that result in the production of truncated proteins lacking either the histone methyltransferase Su(var)3-9, Enhancer-of-zeste and Trithorax (SET) domain or the Set2-Rpb1–interacting (SRI) domain that mediates the interaction of SETD2 with Pol II. Furthermore, loss of heterozygosity (LOH) at chromosome 3p, where SETD2 resides, is commonly detected in lung adenocarcinoma (12–15). Together, these data support that SETD2 is a tumor-suppressor gene in lung adenocarcinoma. Notably, SETD2 mutations often cooccur with other well-established driver mutations, such as KRAS, EGFR, and BRAF, that activate the RTK/RAS/RAF pathway in lung adenocarcinoma (12), suggesting that SETD2 inactivation probably cooperates with these driver mutations to promote lung tumorigenesis. Setd2 deficiency was recently reported to cooperate with KrasG12D or both KrasG12D and p53 deficiency to promote the initiation of mouse lung cancer using CRISPR/Cas9-mediated genome editing (16, 17). Furthermore, Setd2 loss was reported to promote Kras-induced acinar-to-ductal metaplasia and epithelia-mesenchymal transition during pancreatic carcinogenesis (18). Nonetheless, how SETD2 loss-of-function promotes tumorigenesis in lung remains unclear.

Here, we have generated conditional Setd2-KO mice to interrogate the molecular mechanisms by which Setd2 deficiency cooperates with KrasG12D to promote lung tumor initiation. Of note, homozygous deletion of Setd2 in mice results in embryonic lethality, vascular defects, and loss of H3K36me3 without alterations of H3K36me1 and H3K36me2 (4). Consistent with the reported findings (16), we showed that Setd2 deficiency accelerated the initiation of KrasG12D-driven lung tumorigenesis, increased tumor burden, and significantly reduced mouse survival. Mechanistically, we demonstrated that Setd2 deficiency resulted in a coordinated reprogramming of the epigenome and the transcriptome, which enables the amplification of specific oncogenic signatures that promote KrasG12D-driven lung tumorigenesis. SETD2 loss appears to create a permissive epigenetic landscape for the cooperating driver oncogenes to amplify their transcriptional output for tumor initiation in a context-dependent manner. Furthermore, we uncovered mechanism-based therapeutic strategies for SETD2-deficient cancers through inhibition of histone chaperones and transcription elongation.

Results

Setd2 deficiency cooperates with KrasG12D to promote lung tumorigenesis. To investigate the oncogenic cooperation between SETD2 loss and RAS activation in lung cancer pathogenesis in a whole-organism setting, we have generated conditional Setd2-KO mice (Supplemental Figure 1A; supplemental material available online with this article; https://doi.org/10.1172/jci.insight.154120DS1). Setd2fl/fl mice were crossed to mice carrying a conditionally activatable Lox-Stop-Lox KrasG12D allele (hereafter called KrasLSL–G12D) (19). Intranasal administration of Cre-expressing adenovirus (adeno-Cre) was performed to activate the expression of KrasG12D as well as to delete the floxed Setd2 alleles. As reported in ref. 19, KrasLSL–G12D/+ mice following adeno-Cre administration developed adenoma and nonmetastatic adenocarcinoma with a median survival of 201 days. Strikingly, homozygous deletion of Setd2 accelerated the initiation of KrasG12D lung tumors, increased tumor burden, and significantly reduced mouse survival (Figure 1, A and B, and Supplemental Figure 1, B and C). The majority of KrasLSL–G12D/+ Setd2fl/fl mice died from lung adenocarcinoma within 3 months following adeno-Cre infection (median survival, 79 days), whereas KrasLSL–G12D/+ Setd2fl/+ mice exhibited comparable survival to KrasLSL–G12D/+ mice (Figure 1A). Notably, all Setd2fl/fl mice remained healthy at 1 year after adeno-Cre infection (Figure 1B), indicating that Setd2 deficiency alone is insufficient for tumor initiation. PCR-based genotyping confirmed the efficient deletion of floxed Setd2 alleles in lung tumors, with greatly reduced Setd2 expression determined by quantitative PCR (qPCR) (Figure 1, C and F).

Figure 1

Setd2 deficiency cooperates with KrasG12D to promote lung tumorigenesis. (A) Kaplan-Meier survival curves of KrasLSL–G12D/+, KrasLSL–G12D/+ Setd2fl/+, and KrasLSL–G12D/+ Setd2fl/fl mice after adeno-Cre infection. P values denote the comparison of mice of the indicated genotypes with KrasLSL–G12D/+ mice (Mantel–Cox test). (B) Representative MRI, gross images, and histological sections stained with H&E of lungs from Setd2fl/fl mice at 1 year, KrasLSL–G12D/+ mice at 3 months, and KrasLSL–G12D/+ Setd2fl/fl mice at 3 months after adeno-Cre infection. Scale bars: 200 μm. (C) PCR-based genotyping of Setd2 alleles in lung tumors from a representative KrasLSL–G12D/+ mouse and in lung tumors and adjacent normal lung tissues from representative KrasLSL–G12D/+ Setd2fl/fl mice after adeno-Cre infection. (D) Representative H&E staining and IHC for phospho-ERK and H3K36me3 of KrasG12D and KrasG12DSetd2–/– lung tumors. Scale bars: 100 μm. (E) Representative IHC for Ki67, phospho-H3S10, cleaved caspase-3, and TUNEL assays of KrasG12D and KrasG12DSetd2–/– lung tumors. The percentage of positive cells for each staining was quantified (mean ± SD, n = 6). Scale bars: 100 μm. (F) The mRNA levels of Setd2, Ccnd1, Cdkn2a, and Cdkn2b were assessed by qPCR. Data were normalized against β-actin (mean ± SD, n = 4). *P < 0.05; **P < 0.01; ****P < 0.0001 by Student’s t test.

Histological examination of KrasG12DSetd2–/– lung tumors showed mostly well differentiated to moderately differentiated adenocarcinoma with focal invasion and juxtatumoral desmoplastic stromal reaction (Figure 1B and Supplemental Figure 1D). Although KrasLSL–G12D/+ mice only displayed small adenomas at 3 months, some mice developed extensive adenocarcinoma at 5–6 months following adeno-Cre infection, as reported (19). For the ensuing molecular characterization, we used KrasG12DSetd2–/– lung tumors at 3 months following adeno-Cre infection and KrasG12D lung tumors at 5–6 months following adeno-Cre infection that exhibited comparable tumor grades (Supplemental Figure 1E). IHC showed that KrasG12DSetd2–/– lung tumors exhibited reduced H3K36me3 and yet comparable phospho-ERK staining in comparison with KrasG12D lung tumors (Figure 1D). KrasG12DSetd2–/– lung tumors showed increased Ki67 and phospho-H3S10 staining, with no differences in cell death markers in comparison with KrasG12D lung tumors (Figure 1E). Consistent with increased proliferation markers in KrasG12DSetd2–/– lung tumors, Setd2 deficiency upregulated cyclin D1 and downregulated Cdkn2a and Cdkn2b (Figure 1F). Collectively, our studies presented compelling evidence of the oncogenic cooperation between Setd2 deficiency and KrasG12D in accelerating lung tumorigenesis.

Setd2 deficiency increases chromatin accessibility and oncogenic transcriptional output in KrasG12D lung tumors. To investigate the impact of SETD2 loss-induced transcriptome changes on the pathogenesis of KrasG12D-driven lung cancer, RNA-Seq was performed on KrasG12D and KrasG12DSetd2–/– lung tumors with comparable histopathological features and tumor grades (Supplemental Figure 1E). Setd2 deficiency in KrasG12D lung tumors led to a global alteration of transcriptome with 3,296 differentially expressed genes (FDR < 0.05; Supplemental Figure 2A). Gene set enrichment analysis (GSEA) revealed upregulation of several oncogenic pathways upon SETD2 loss, including the KRAS transcriptional signature, the PTEN-loss transcriptional signature, and PRC2-repressed targets identified in liver cancer (20), and malignant peripheral nerve sheath tumors (MPNST) (21) (Figure 2A and Supplemental Table 1). Although SETD2 loss does not affect the KRAS/RAF/MEK/ERK signal transduction pathway (Figure 1D), it enhances the transcriptional output downstream of KRAS signaling (Figure 2A). SETD2 loss also led to upregulation of G-protein–coupled receptor signaling, DNA packaging, and RNA catabolism (Supplemental Table 1).

Figure 2

Setd2-deficient KrasG12D lung tumors show increased chromatin accessibility and oncogenic transcriptional output compared with KrasG12D lung tumors. (A) GSEA plots of the differentially expressed genes (FDR < 0.05) comparing KrasG12DSetd2–/– with KrasG12D mouse lung tumors using the indicated gene sets. MPNST, malignant peripheral nerve sheath tumors. NES, normalized enrichment score. (B) PRC2 signature enrichment plots of the differentially expressed genes (FDR < 0.05) comparing KrasG12DSetd2–/– with KrasG12D mouse lung tumors and comparing SETD2MT with SETD2WT human lung adenocarcinomas (LUAD) from TCGA using the composite PRC2 signature. (C) Volcano plots of ATAC-Seq peaks comparing dissociated KrasG12DSetd2–/– with KrasG12D mouse lung tumor cells. The number of peaks with significant changes (FDR < 0.05 and log2FC > 1) upon Setd2 deletion is shown. (D) Pie chart showing the percentage of differentially accessible ATAC-Seq peaks (FDR < 0.05) at promoter, intronic, intergenic, and exonic regions comparing KrasG12DSetd2–/– with KrasG12D mouse lung tumors. (E) Heatmap of differentially accessible ATAC-Seq peaks described in C (FDR < 0.05 and log2FC > 1) in 5 kb window grouped by localization at promoter, intron, and intergenic regions. (F) The 20 most significantly enriched transcription factor binding motifs in open (red) and closed (blue) chromatin peaks comparing KrasG12DSetd2–/– with KrasG12D mouse lung tumors. (G) Distribution of chromatin accessibility changes associated with significantly upregulated (red) or downregulated (blue) genes comparing KrasG12DSetd2–/– with KrasG12D mouse lung tumors. P values calculated using 1-sided Kolmogorov-Smirnov (KS) test comparing peaks associated with differentially expressed genes to all genes. (H) Venn diagram showing overlap of differentially expressed genes detected by RNA-Seq (FDR < 0.05) and genes with differentially accessible ATAC-Seq peaks (FDR < 0.05) comparing KrasG12DSetd2–/– with KrasG12D mouse lung tumors. (I) GSEA plots of the 1,394 differentially expressed genes shown in H using the KRAS and PRC2 signatures.

In mammalian cells, 2 main polycomb-repressive complexes (PRCs) have been defined: PRC1 and PRC2, both of which repress gene expression (22, 23). The PRC2 complex catalyzes the methylation of histone H3 at lysine 27 through its enzymatic subunits EZH1 and EZH2; the enrichment of H3K27me3 correlates with gene silencing (22, 23). Of note, PRC2 exerts either oncogenic or tumor-suppressive function in a context-dependent manner. Loss of PRC2 complex has been reported to promote tumor aggressiveness in p53-deficient KrasG12D-driven mouse lung cancer (24). Our demonstration of upregulation of PRC2-suppressed target genes in response to Setd2 deletion is functionally equivalent to the inactivation of the PRC2 complex, which probably contributes to lung tumor progression. By merging various reported PRC2 modules (20, 21, 25–27), we generated a composite PRC2 signature detailed in the method section. Remarkably, both mouse lung tumors and human lung adenocarcinomas from TCGA data set showed highly enriched PRC2 signature in tumors with SETD2 loss (Figure 2B). Of note, no global difference in H3K27me3 levels was observed comparing KrasG12DSetd2–/– with KrasG12D tumors (Supplemental Figure 2B), suggesting that SETD2 loss–induced upregulation of PRC2 targets does not simply occur through a direct inactivation of the PRC2 complex. In addition, no global difference in H3K4me3, H3K4me1, or K3K27ac levels was observed comparing KrasG12DSetd2–/– with KrasG12D tumors (Supplemental Figure 2C).

We hypothesized that increased transcriptional output of oncogenic pathways in Setd2-deficient lung tumors may be caused by an altered epigenetic landscape upon the ablation of H3K36me3 marks. To interrogate this hypothesis, ATAC-Seq (an assay for transposase-accessible chromatin using sequencing) was performed on dissociated mouse lung tumor cells to assess genome-wide changes in chromatin accessibility (28). Setd2 deletion in KrasG12D lung tumors induced significant genome-wide chromatin accessibility changes at ~14,400 sites, among which 82.3% showed increased chromatin accessibility (Figure 2C and Supplemental Figure 2D). Among the differentially accessible ATAC-Seq peaks (FDR < 0.05), 43.1% were found at introns, 38.7% at intergenic regions, 16.6% at promoters, and 1.6% at exons (Figure 2, D and E). A genome-wide increase in chromatin accessibility was also observed in SETD2-deficient human ccRCC and primary mouse renal tubular epithelial cells compared with respective SETD2-proficient counterparts (29), indicating that increased chromatin accessibility is a primary phenotype caused by SETD2 loss across different tissue types. This is consistent with the reported association of SETD2 mutations with increased chromatin accessibility preferentially in gene bodies in human ccRCC tumors (30).

The genomic loci with open chromatin peaks induced by Setd2 deletion in KrasG12D lung tumors were highly enriched with Forkhead box (FOX) family transcription factor binding motifs (TFBM) (Figure 2F). The open chromatin peaks were also enriched with the binding motif of FOS, one of the key transcription factors downstream of ERK signaling that drives RAS-mediated transcription (31). This is consistent with the enhanced KRAS signature observed in KrasG12DSetd2–/– lung tumors (Figure 2A). Notably, Gene Ontology (GO) and KEGG pathway analysis of genes with open chromatin peaks upon Setd2 deletion also showed enrichment of the RAS signaling pathway (Supplemental Figure 2E). These findings prompted us to investigate the correlation between SETD2 loss-induced alterations in chromatin accessibility and transcriptional output by integrating the ATAC-Seq data and the RNA-Seq data. Overall, the upregulated genes upon Setd2 deletion in KrasG12D lung tumors exhibited more open chromatin, whereas the downregulated genes exhibited more closed chromatin (Figure 2G, P < 2.2 × 10–16). GSEA of the differentially expressed genes detected by RNA-Seq that also exhibited differentially accessible ATAC-Seq peaks showed upregulation of KRAS and PRC2 signatures upon SETD2 loss (Figure 2, H and I). Consistently, the genes that were most upregulated in response to SETD2 loss within the KRAS or PRC2 signature displayed mainly open chromatin status, whereas those that were most downregulated displayed closed chromatin status (Supplemental Figure 2F). Collectively, our data reveal that SETD2 loss increases chromatin accessibility to enhance the oncogenic transcriptional output.

SETD2 loss induces ETV1 expression through activation of an intronic enhancer to promote transformation. To understand how Setd2 deficiency increases chromatin accessibility to induce the expression of KRAS signature, we first focused on Etv1, one of the transcription factors downstream of ERK signaling and a well-defined oncogene in multiple cancer types (31, 32). Notably, Etv1 is among the 12 genes that are differentially expressed upon SETD2 loss in both mouse and human lung adenocarcinomas (Figure 3A). We first confirmed that both Etv1 mRNA and ETV1 protein were upregulated in KrasG12DSetd2–/– mouse lung tumors compared with KrasG12D tumors (Figure 3, B and C). Due to the large coding sequence of SETD2, efficient transduction of the full-length SETD2 using a retroviral or lentiviral vector was not possible. Nevertheless, it has been reported that the N-terminal truncated SETD2 (SETD2ΔN) is fully functional (33). Consistently, we demonstrated that retroviral transduction of the SETD2ΔN lacking the first 1,241 amino acids while retaining all the important functional domains was sufficient to fully restore H3K36me3 and reduce Etv1 mRNA and protein in primary cells derived from KrasG12DSetd2–/– mouse lung tumors (Figure 3D). Functionally, lentiviral transduction of Cas9 and the sgRNA targeting Etv1 significantly reduced the ETV1 protein expression and the ability of KrasG12DSetd2–/– mouse lung tumor cells to form colonies in soft agar (Figure 3E). Overall, these data suggest that ETV1 is one of the important downstream oncogenic targets induced upon SETD2 loss to promote oncogenic transformation.

Figure 3

Setd2 deficiency induces Etv1 through activation of an intronic enhancer to promote transformation. (A) Venn diagram showing overlap of differentially expressed genes (FDR < 0.05) comparing KrasG12DSetd2–/– with KrasG12D mouse lung tumors and comparing SETD2MT (n = 20) with SETD2WT (n = 210) human lung adenocarcinomas from TCGA. Heatmap showing these genes in mouse lung tumors. (B) Left, qPCR analysis of Etv1 in mouse lung tumors (mean ± SD, n = 4). Right, the normalized ETV1 expression comparing SETD2MT with SETD2WT human lung adenocarcinomas was obtained from cBioPortal. (C) Immunoblot analyses of the indicated mouse lung tumors. The number denotes the ETV1 expression normalized against β-actin (P = 0.0301, KrasG12D versus KrasG12DSetd2–/–). (D) A schematic of the domain structure of SETD2 and SETD2ΔN. Whole cell lysates (WCL) and histone fractions from primary KrasG12DSetd2–/– (KS) mouse lung tumor cells ± SETD2ΔN transduction or KrasG12Dp53–/– mouse lung tumor cells were analyzed by immunoblots. The Etv1 mRNA levels were assessed by qPCR (mean ± SD, n = 3). (E) Primary KS cells transduced with the indicated sgRNAs were analyzed by soft agar colony formation assays and immunoblots. (F) Representative ATAC-Seq tracks at the Etv1 locus in mouse lung tumors. (G) Primary mouse lung tumor cells were assessed by ChIP-qPCR at the indicated genomic regions (mean ± SD, n = 3). (H and I) KS cells transduced the indicated sgRNAs were analyzed by qPCR (mean ± SD, n = 3), immunoblots, or soft agar colony formation assays. (J) A549 cells were transiently transfected with pGL2-pro vector or pGL2-pro containing the putative intron 4 enhancer of Etv1 ± deletion of the FOS binding motif, together with the pRL-SV40 plasmid (Promega) as a normalization control (mean ± SD, n = 3). *P < 0.05; **P < 0.01; ***P < 0.001 by Student’s t test.

To interrogate how chromatin accessibility might affect Etv1 expression in lung cancer, we assessed the ATAC-Seq tracks at the Etv1 locus, which revealed significantly increased chromatin accessibility at both promoter and intron 4 upon SETD2 loss (Figure 3F). Notably, the distinct ATAC-Seq peak at the intron 4 of mouse Etv1 coincided with H3K4me1 ChIP-Seq peaks shown in the mouse lung tissue ENCODE data (Supplemental Figure 3A), and this distinct peak likely represents an intronic enhancer. H3K4me1 and H3K27ac are commonly used to annotate enhancers, and H3K27ac specifically marks active enhancers (34–36). Accordingly, we hypothesized that SETD2 loss may increase chromatin accessibility of oncogenic genes to transcription factors and chromatin modifiers — e.g., AP-1 (a dimeric complex composed of members from the JUN, FOS, or ATF protein families — which in turn increases H3K27ac levels and activates certain intronic enhancers to drive respective gene expression. To examine this hypothesis, ChIP-qPCR was performed on dissociated mouse lung tumor cells to assess the impact of Setd2 deletion in chromatin modifications within the Etv1 locus. Setd2 deficiency significantly increased H3K27ac at the intron 4 of Etv1 (Figure 3G), suggesting that SETD2 loss activates this intronic enhancer. To demonstrate direct regulation of Etv1 expression by this putative intronic enhancer, CRISPR/Cas9-mediated deletion of the ATAC-Seq peak region at the intron 4 of Etv1 was performed, which led to reduced Etv1 expression and soft agar colony formation of KrasG12DSetd2–/– mouse lung tumor cells (Figure 3, H and I, and Supplemental Figure 3, B and C). To further prove the presence of enhancer activity at mouse Etv1 intron 4, we cloned the DNA fragment from the ATAC-Seq peak region into a luciferase reporter construct that was subsequently transfected into A549, a human KRAS mutant lung cancer cell line. Indeed, this DNA fragment conferred a ~3-fold increase in luciferase activity (Figure 3J). Because motif analysis identified a FOS binding site within the intron 4 enhancer of Etv1 (Supplemental Figure 3D), we next assessed the potential contribution of FOS binding to enhancer activation by deleting the FOS binding motif in the luciferase reporter construct. Deletion of the FOS binding motif diminished the ability of the Etv1 intron 4 enhancer to induce luciferase activity (Figure 3J). Collectively, these data support that SETD2 loss creates an epigenetic landscape consisting of open chromatin, enabling transcription factors and chromatin modifiers to activate intronic enhancers and amplify the KRAS-driven transcriptional output.

We next investigated whether SETD2-mediated regulation of ETV1 is conserved in human ccRCC in which the highest mutation rate of SETD2 is observed (11). The distinct ATAC-Seq peak at the intron 4 of mouse Etv1 coincided with H3K4me1 ChIP-Seq peaks at the equivalent intron 5 of human ETV1 with significant sequence homology (Figure 4A and ENCODE data not shown). To determine whether the conserved sequence at the intron 5 of human ETV1 also contains an enhancer that is regulated by SETD2, ChIP-qPCR was performed on a patient-derived ccRCC cell line JHRCC12 that harbors a truncating mutation of SETD2 at the SRI domain (p.E2531*) (37). Retroviral transduction of SETD2ΔN in JHRCC12 restored H3K36me3 to levels comparable with 786-O cells carrying WT SETD2, and it reduced ETV1 expression (Figure 4B). ChIP-qPCR showed that transduction of SETD2ΔN in JHRCC12 cells greatly reduced H3K27ac at the intron 5 of ETV1 (Figure 4C), suggesting that SETD2 loss of function activates this intronic enhancer. Transduction of SETD2ΔN in JHRCC12 cells also reduced H3K27ac at the promoter of ETV1 but to a lesser extent than the intron 5 (Figure 4C). Importantly, CRISPR/Cas9-mediated deletion of the intron 5 enhancer significantly reduced ETV1 expression in JHRCC12 cells (Figure 4, D and E), supporting a direct activation of ETV1 transcription by this intronic enhancer. Furthermore, luciferase reporter assays confirmed the presence of enhancer activity within this putative enhancer in human ETV1, and the enhancer activity was abrogated by deletion of the FOS-binding motif (Figure 4F). Collectively, data obtained from both mouse lung tumors and human ccRCC revealed a conserved regulatory mechanism of ETV1 transcription upon SETD2 loss through increased chromatin accessibility and intronic enhancer activity.

Figure 4

SETD2 loss induces human ETV1 expression through activation of an intronic enhancer. (A) Sequence homology between ATAC-Seq peak regions at the intron 4 of mouse Etv1 and the intron 5 of human ETV1. (B) Whole cell lysates (WCL) and histone fractions from JHRCC12 cells infected with control retrovirus or retrovirus expressing SETD2ΔN or from 786-O cells were analyzed by immunoblots. The mRNA levels of ETV1 were assessed in the indicated JHRCC12 cells by qPCR and normal