Protein-coding mutations in clear cell renal cell carcinoma (ccRCC) have been extensively characterized, frequently involving inactivation of the von Hippel–Lindau (VHL) tumor suppressor. Roles for noncoding cis-regulatory aberrations in ccRCC tumorigenesis, however, remain unclear. Analyzing 10 primary tumor/normal pairs and 9 cell lines across 79 chromatin profiles, we observed pervasive enhancer malfunction in ccRCC, with cognate enhancer-target genes associated with tissue-specific aspects of malignancy. Superenhancer profiling identified ZNF395 as a ccRCC-specific and VHL-regulated master regulator whose depletion causes near-complete tumor elimination in vitro and in vivo. VHL loss predominantly drives enhancer/superenhancer deregulation more so than promoters, with acquisition of active enhancer marks (H3K27ac, H3K4me1) near ccRCC hallmark genes. Mechanistically, VHL loss stabilizes HIF2α–HIF1β heterodimer binding at enhancers, subsequently recruiting histone acetyltransferase p300 without overtly affecting preexisting promoter–enhancer interactions. Subtype-specific driver mutations such as VHL may thus propagate unique pathogenic dependencies in ccRCC by modulating epigenomic landscapes and cancer gene expression.

Significance: Comprehensive epigenomic profiling of ccRCC establishes a compendium of somatically altered cis-regulatory elements, uncovering new potential targets including ZNF395, a ccRCC master regulator. Loss of VHL, a ccRCC signature event, causes pervasive enhancer malfunction, with binding of enhancer-centric HIF2α and recruitment of histone acetyltransferase p300 at preexisting lineage-specific promoter–enhancer complexes. Cancer Discov; 7(11); 1284–305. ©2017 AACR.

See related commentary by Ricketts and Linehan, p. 1221.

This article is highlighted in the In This Issue feature, p. 1201

Clear cell renal cell carcinoma (ccRCC) is the most common subtype of kidney cancer, with 338,000 new cases in 2012 worldwide (1). With most ccRCCs being radiochemoresistant, patients with metastatic ccRCC exhibit dismal 8% five-year overall survival (2). Although targeted therapies inhibiting angiogenesis and mTOR pathways can lead to initial tumor control, most patients develop resistance in less than a year (3, 4). A better understanding of ccRCC molecular dependencies and vulnerabilities is thus needed to develop new therapeutic strategies for patients who fail standard-of-care treatment.

Loss of the von Hippel–Lindau (VHL) tumor suppressor is a defining feature of ccRCC (5, 6). When partnered with inactivation of additional tumor suppressors (Pbrm1, Bap1, Trp53, Rb1, and/or Cdkn2a) and/or activation of oncogenes (Myc), Vhl loss drives spontaneous ccRCC formation in mouse models (7–11). VHL encodes an E3 ubiquitin ligase (12, 13) that targets HIF1α (HIF1A) and HIF2α (EPAS1) for degradation (14, 15). VHL loss in ccRCC results in constitutive activation of HIF1/2α and subsequent transactivation (16, 17) of downstream genes regulating angiogenesis, glycolysis (18), lipogenesis (19, 20), cell cycle (21), and antiapoptosis (22).

Most reports studying VHL/HIF transcriptional activation have focused on HIF-bound promoters (23–28). However, recent evidence suggests an emerging role for distal enhancer elements in VHL/HIF transcriptional control (29, 30). For example, HIF2α-bound distal enhancers activate the proto-oncogenes MYC (31) and CCND1 (21) and coincide with ccRCC genetic risk alleles. Nevertheless, such studies focused on individual enhancers, and the majority of distal elements in ccRCC remain largely unexplored.

Delineating the global ccRCC cis-regulatory landscape may also identify master regulators involved in tissue-specific disease processes. Compared with promoters that are largely cell-type invariant, distal enhancers integrate multiple lineage- and context-dependent signals, catering to the specialized needs of diverse cell types and diseases (32, 33). In cancer, such master regulators are frequently located near “superenhancers” or “stretch-enhancers” marked by long stretches of H3K27ac signals (34, 35). For example, subtype-specific genomic alterations such as EGFRvIII in glioblastoma (36) and EWS–FLI in Ewing sarcoma (37) induce de novo enhancers, causing reactivation of developmental master regulators required for self-renewal and lineage specification (36). Although VHL inactivation has been shown to modulate protein levels of different histone modifiers (e.g., KDM5C/JARID1C, ref. 38; HDAC1, ref. 39; JMJD1A, ref. 40; JMJD2B, ref. 40; JMJD2C, ref. 41), the impact of these protein alterations at specific epigenomic loci remains unclear. Moreover, previous studies profiling histone modifications in ccRCCs have also been limited by small sample sizes (2 cases; ref. 42), reliance on in vitro systems, and the lack of long-range interactome data and functional enhancer testing to accurately assign cognate enhancer targets.

In this study, we establish the most comprehensive collection of ccRCC histone profiles to date, annotating the precise genomic locations of altered promoters, enhancers, and superenhancers in ccRCC. Using isogenic cell lines with or without wild-type VHL, we further demonstrate that besides its well-defined role in oxygen sensing, VHL also safeguards the chromatin landscape; its loss induces tumor-specific enhancer gains around ccRCC hallmark genes such as angiogenic and metabolic targets through the stabilization of HIF2α/HIF1β (ARNT) heterodimers and recruitment of p300 histone acetyltransferase (EP300). One important target of epigenetic activation is ZNF395, a master regulator of ccRCC tumorigenesis. Taken collectively, our results reveal an epigenetic framework by which the major ccRCC-specific driver mutation, VHL, induces de novo enhancers, contributing to oncogenic transcription.

Cis-Regulatory Landscapes in ccRCC Tumors Are Aberrant

To explore whether ccRCC tumors display alterations in their cis-regulatory landscapes in vivo, we generated histone chromatin immunoprecipitation sequencing (ChIP-seq) profiles (3 marks: H3K27ac, H3K4me3, and H3K4me1) in 10 primary tumor/normal pairs, 5 patient-matched tumor-derived cell lines, 2 commercially available ccRCC lines (786-O and A-498), and 2 normal kidney cell lines (HK2 and PCS-400; refer to Supplementary Table S1 for patient clinical information). Of the original 87 samples, 79 samples passed pre-sequencing quality-control filters and were subjected to ChIP-seq processing and downstream analysis. In total, we generated 2,363,904,778 uniquely mapped reads (refer to Supplementary Table S2 for sequencing statistics). On average, 89% of H3K27ac peaks, 98% of H3K4me3 peaks, and 76% of H3K4me1 peaks obtained in our normal kidney tissues overlapped with peaks from adult kidney tissues in the Epigenomics Roadmap dataset (Supplementary Fig. S1A). Among the 10 primary ccRCCs, 9 harbored VHL mutations, detected by targeted sequencing and confirmed by Sanger sequencing (Supplementary Table S3). Cell lines 786-O and A-498 also harbor VHL truncating mutations (Supplementary Table S3). The VHL mutations co-occurred with somatic mutations of other chromatin modifiers commonly found in ccRCC, including PBRM1 (7/10), SETD2 (1/10), KDM5A (1/10), KDM5C (1/10), ARID1A (1/10), and KMT2C (1/10).

Specific histone modifications can distinguish different categories of functional regulatory elements—H3K4me3 is generally associated with promoters, H3K4me1 with enhancers, and H3K27ac with active elements (33, 43). Integrating signals from three histone marks and GENCODE v19 annotated transcription start sites (TSS), we defined active promoters as H3K27ac+/H3K4me3+/±2.0 kb TSS regions, and distal enhancers as H3K27ac+/H3K4me1+ regions not overlapping with promoters. Focusing on epigenomic events specific to somatic cancer cells, we derived cell lines from five primary tumors and, combined with the commercial lines, excluded peaks not found in any of the cell lines to reduce confounding effects from stromal cells. On average, we observed 80% overlap of ChIP-seq peaks between primary tumors and matched lines (Supplementary Fig. S1B). Using these criteria, we identified 17,497 putative promoters and 66,448 putative enhancers (Fig. 1A), numbers comparable with previous studies in other tumor types (43–45). The numbers of defined promoters and enhancers reached saturation after 4 and 16 samples, respectively, suggesting that a sample size of 20 (10 tumor/normal pairs) is sufficiently powered to discover the majority of cis-regulatory elements in ccRCC (Supplementary Fig. S1C and S1D). Principal components analysis (PCA) using the first two components of global H3K27ac intensities at promoters or enhancers (representing 83% and 64% of total variance, respectively; Supplementary Fig. S1E and S1F) successfully separated normal and tumor samples, indicating that genome-wide pervasive alterations in cis-regulatory elements are a salient feature of ccRCC (Fig. 1B).

Figure 1.

VHL-deficient ccRCC tumors exhibit an aberrant cis-regulatory landscape. A, Putative active promoters are defined by co-occurrence of H3K4me3, H3K27ac, and proximity to TSS within 2 kb. Putative active enhancers are defined by the presence of H3K4me1, H3K27ac, and exclusivity with promoters. B, PCA analysis using all 17,497 promoters and 66,448 enhancers classifies normal and tumors into distinct clusters. Patient IDs: 1-12364284; 2-17621953; 3-20431713; 4-40911432; 5-57398667; 6-70528835; 7-74575859; 8-77972083; 9-86049102; 10-75416923. C, Heat maps show H3K27ac levels of altered promoters and enhancers in a paired patient tissue (patient 40911432; yellow high, black low). D, The H3K27ac levels, chromatin accessibility (FAIRE-seq), DNA methylation of gained promoters and enhancers, and gene expression of the nearest ccRCC lncRNA are compared between normal and tumor tissues. ***, P < 0.001, two-sided t test. E, Shown are tracks of histone ChIP-seq (H3K27ac, H3K4me1, and H3K4me3) and RNA sequencing (RNA-seq) at the CCND1 locus in a tumor–normal pair of patient 40911432. The histone ChIP-seq profiles of normal adult kidney tissue from the Epigenome Roadmap are displayed above the normal tissue generated by Nano-ChIP-seq for comparison. A cell line was derived from the tumor tissue, and its histone profile is displayed below its matching primary tissue. This enhancer is known to interact with the CCND1 promoter from a previous study (21) and is situated close to an RCC susceptibility SNP rs7105934 (49).

Figure 1.

VHL-deficient ccRCC tumors exhibit an aberrant cis-regulatory landscape. A, Putative active promoters are defined by co-occurrence of H3K4me3, H3K27ac, and proximity to TSS within 2 kb. Putative active enhancers are defined by the presence of H3K4me1, H3K27ac, and exclusivity with promoters. B, PCA analysis using all 17,497 promoters and 66,448 enhancers classifies normal and tumors into distinct clusters. Patient IDs: 1-12364284; 2-17621953; 3-20431713; 4-40911432; 5-57398667; 6-70528835; 7-74575859; 8-77972083; 9-86049102; 10-75416923. C, Heat maps show H3K27ac levels of altered promoters and enhancers in a paired patient tissue (patient 40911432; yellow high, black low). D, The H3K27ac levels, chromatin accessibility (FAIRE-seq), DNA methylation of gained promoters and enhancers, and gene expression of the nearest ccRCC lncRNA are compared between normal and tumor tissues. ***, P < 0.001, two-sided t test. E, Shown are tracks of histone ChIP-seq (H3K27ac, H3K4me1, and H3K4me3) and RNA sequencing (RNA-seq) at the CCND1 locus in a tumor–normal pair of patient 40911432. The histone ChIP-seq profiles of normal adult kidney tissue from the Epigenome Roadmap are displayed above the normal tissue generated by Nano-ChIP-seq for comparison. A cell line was derived from the tumor tissue, and its histone profile is displayed below its matching primary tissue. This enhancer is known to interact with the CCND1 promoter from a previous study (21) and is situated close to an RCC susceptibility SNP rs7105934 (49).

Close modal

We performed differential analysis to identify altered promoters and enhancers. To define gained or lost regions, we applied a fold difference of H3K27ac RPKM ≥ 2, an absolute difference ≥ 0.5, and for greater stringency no alterations in the reverse direction in the remaining tumor/normal pairs (see Methods and Supplementary Fig. S1G for distribution of altered elements by number of patients). At the threshold of ≥5/10 patients, 80% of the altered regions achieved statistical significance (q-value < 0.1, paired t test, with Benjamini–Hochberg correction; Supplementary Fig. S1H), and at this same threshold, the increase in the fraction of samples meeting statistical significance reached a saddle point (Supplementary Fig. S1I). Applying these criteria, we obtained a high-confidence and comprehensive set of 4,719 gained promoters, 592 lost promoters, 4,906 gained enhancers, and 5,654 lost enhancers (Fig. 1A and C; Supplementary Table S4). Representative regions are presented in Supplementary Fig. S2.

Supporting the reliability of these data, gained promoters and enhancers exhibited increased chromatin accessibility measured by higher FAIRE-seq signals (46) in tumor tissues than normal tissues, respectively (P < 0.0001) and also decreased DNA methylation based on data from The Cancer Genome Atlas (TCGA; ref. 47), consistent with reciprocal relationships between active regulatory regions and DNA methylation (Fig. 1D). Interestingly, we also noted elevated expression of long noncoding RNAs (48) adjacent to gained promoters and enhancers in tumor tissues compared with normal tissues (P < 0.0001, respectively). Lastly, we confirmed that many of our cis-regulatory elements involved regions previously implicated in ccRCC; for example, we observed gains of H3K27ac signals and enrichment of H3K4me1 at a distal enhancer of CCND1 overlapping with an RCC susceptibility locus (rs7105934; refs. 21 and 49; Fig. 1E). Our ability to rediscover this important enhancer in our unbiased profiling supports our data reliability.

Tumor-Specific Enhancers Are Associated with Hallmarks of ccRCC

To identify genes modulated by the tumor-specific regulatory elements, we assigned enhancers using three approaches. The first approach utilized predefined linear proximity rules involving a set of highly confident genes (GREAT algorithm; ref. 50; Supplementary Table S5). MSigDB pathway analysis using GREAT-assigned genes revealed that gained enhancers exhibit a highly significant RCC-specific signature compared with gained promoters (enhancer q-value = 3.2 × 10−26; promoter q-value = 1.5 × 10−1, binomial FDR; Fig. 2A). Although gained promoters were involved in general cancer processes (e.g., cell cycle, transcription, and RNA metabolism; see Supplementary Table S6 for a complete list of promoter pathways), gained enhancers were enriched in disease-specific features of ccRCC, including HIF1α network activity, proangiogenic pathways (platelet activation and PDGFRβ signaling), and SLC-mediated transmembrane transport (Fig. 2A; Supplementary Table S7 for a complete list of enhancer pathways). Notably, HIF1α network activity consistently emerged as one of the top five pathways, even with perturbations in the patient thresholds used to define gained enhancers (≥3–8 patients; Supplementary Table S8).

Figure 2.

Enhancer aberration is a signature of ccRCC. A, Enriched pathways associated with gained promoters and enhancers revealed by GREAT (ranked by binomial FDR q-value). Red bars refer to ccRCC-specific pathways. B,De novo enhancers are acquired in a ccRCC tumor tissue upstream of VEGFA. Capture-C confirmed interactions of this VEGFA enhancer (E) with its promoter (P) in 786-O cells. The arcs represent significant interactions detected by r3Cseq (q < 0.05). The input-subtracted H3K27ac signals of this enhancer are highly correlated with VEGFA gene expression (Spearman correlation). C, Similar to (B), a de novo tumor enhancer interacts with the SLC2A1/GLUT1 promoter.

Figure 2.

Enhancer aberration is a signature of ccRCC. A, Enriched pathways associated with gained promoters and enhancers revealed by GREAT (ranked by binomial FDR q-value). Red bars refer to ccRCC-specific pathways. B,De novo enhancers are acquired in a ccRCC tumor tissue upstream of VEGFA. Capture-C confirmed interactions of this VEGFA enhancer (E) with its promoter (P) in 786-O cells. The arcs represent significant interactions detected by r3Cseq (q < 0.05). The input-subtracted H3K27ac signals of this enhancer are highly correlated with VEGFA gene expression (Spearman correlation). C, Similar to (B), a de novo tumor enhancer interacts with the SLC2A1/GLUT1 promoter.

Close modal

Individual genes associated with gained enhancers included well-known hypoxic targets (VEGFA, Fig. 2B; CXCR4) and metabolic genes involved in glycolysis, glutamine intake, and lipid storage (GLUT1/SLC2A1, Fig. 2C; HK2, PFKFB3, PLIN2, Supplementary Fig. S3A) and SLC38A1 (Supplementary Fig. S3B; ref. 51). The presence of enhancers around metabolic enzymes and transporters is largely consistent with the metabolic contexture of ccRCC, which involves increased glycolysis and glutaminolysis (19, 52–55). Indeed, gene ontology (GO) analysis of gained enhancers strongly reflected hallmark metabolic changes associated with ccRCC, including monocarboxylic acid transmembrane transporter activity (binomial FDR q-value = 1.6 × 10−10; Supplementary Fig. S3C).

We also used a second method of enhancer–gene assignment based on correlations between H3K27ac signals and expression of genes within the same topologic associated domain (TAD; ref. 34). Using a q-value of <0.05 based on Spearman correlation, we assigned 2,311 gained enhancers to 2,186 protein-coding targets (Supplementary Table S9). Reassuringly, H3K27ac signals of many gained enhancers were highly correlated with gene expression of their putative target genes. For example, H3K27ac levels of a VEGFA enhancer exhibited high correlation with VEGFA gene expression (r = 0.83, Spearman correlation), whereas H3K27ac signals of an SLC2A1 enhancer were highly correlated with SLC2A1 gene expression (r = 0.72, Spearman correlation; Fig. 2B; Supplementary Fig. S3D). Similar to the GREAT approach, the TAD correlation approach also highlighted hypoxia (Krieg_Hypoxia_not_via_KDM3A, FDR q-value = 7 × 10−120) and metabolism (Chen_Metabolic_Syndrome_Network, FDR q-value = 2 × 10−91) as highly enriched pathways (Supplementary Table S10).

Third, to independent validate the GREAT and TAD approaches in the specific context of ccRCC, we experimentally explored the interactome of ccRCC tumor-specific enhancers by performing Capture-C assays (56). Compared with other chromatin capture techniques, Capture-C offers both high-resolution (down to single Kb resolution) and high-throughput interrogation of user-defined regions (a usual working range of 10–500 regions). We designed probes against a subset of 56 gained enhancers and examined their interactions with protein-coding genes in 786-O cells. Each gene–enhancer pair revealed by Capture-C was further filtered by correlations between gene expression and H3K27ac levels (q-value < 0.05). The 56 gained enhancers were paired with 36 protein-coding genes (Supplementary Table S11)—of these, 58% were predicted by GREAT, and 80% by gene correlations within TADs. The median distance of interactions detected by Capture-C was 16 kb, and 83% of the interactions fell within a 100-kb window (Supplementary Fig. S3E). As a visual example, Capture-C confirmed interactions between VEGFA enhancer and the VEGFA TSS, spanning a distance of ∼100 kb (Fig. 2B), and interactions between the SLC2A1 enhancer and its promoter (Fig. 2C). Taken collectively, these findings highlight the disease-specific nature of enhancer elements (33) and an important role for enhancer malfunction in modulating ccRCC pathology.

Tumor Superenhancers Identify ZNF395 as a Master Regulator of ccRCC Tumorigenesis

The importance of enhancers in ccRCC led us to examine the landscape of “superenhancers” or “stretch-enhancers”—dense clusters of enhancers located near master regulators of cell identity and disease (35, 57). Using ROSE (35), we identified 1,451 superenhancers in the ccRCC cohort, of which 1,157 were gained in tumors and 294 were lost in tumors (Supplementary Table S12).

Putative targets of top gained superenhancers validated well-known oncogenes including MYC/PVT1, VEGFA, and HIF2A (Fig. 3A; Supplementary Fig. S4A and S4B). In addition, we found several less-known genes including ERGIC1, ZNF395, SLC28A1, and SMPDL3A (Fig. 3B). These genes were highly overexpressed in tumors compared with their matched normal tissues (Fig. 3B). Furthermore, they were unique to ccRCC and were not overexpressed in papillary and chromophobe RCCs, two other distinct ccRCC subtypes (Fig. 3B). For instance, ZNF395 exhibited a tumor–normal ratio of ∼7 in ccRCC (P = 1 × 10−22, paired t test) but experienced little overexpression in papillary and chromophobe RCC with tumor–normal ratios of 1.2 and 1.3, respectively (P = 0.02 in papillary and P = 0.06 in chromophobe, paired t test).

Figure 3.

Tumor superenhancers (SE) identify key oncogenic drivers. A, A total of 1,451 superenhancers are identified by ROSE and ranked by their differential H3K27ac intensity between normal and tumor tissues. Genes associated with the top gained and lost superenhancers are listed. B, TCGA RNA-seq data indicate that genes associated with top 10 gained enhancers are upregulated in tumors whereas genes associated with top 10 lost enhancers are downregulated. This tumor specificity is restricted to ccRCC, but not the other two RCC subtypes, papillary and chromophobe. C, Expression of ZNF395 and SMPDL3A are measured in a panel of normal kidney cell lines (black) and ccRCC cell lines (red) by RT-qPCR. D, Pooled siRNA against ZNF395 inhibits colony formation of A-498 and 786-O but not HK2 normal immortalized kidney cells. Pooled siRNA against SMPLD3A inhibits colony formation of A-498 but not 786-O. E, H3K27ac ChIP-seq shows an active ZNF395 superenhancer only in ccRCC cells (A-498 and 786-O) but not normal kidney cells (PCS-400, HK2). F, TCGA RNA-seq data show exclusive overexpression of ZNF395 among 12 cancer types. G–I,ZNF395 inhibition by two shRNA clones decreases colony formation (G) and in vitro proliferation (H) and increases apoptosis measured by cleavage of caspase 3/7 substrate (I). *, P < 0.05, two-sided t test. J,ZNF395 inhibition by shRNA leads to total elimination of A-498 tumors in vivo and delayed 786-O tumor growth. NC: n = 7; shZNF395-1: n = 7; shZNF395-2: n = 6.

Figure 3.

Tumor superenhancers (SE) identify key oncogenic drivers. A, A total of 1,451 superenhancers are identified by ROSE and ranked by their differential H3K27ac intensity between normal and tumor tissues. Genes associated with the top gained and lost superenhancers are listed. B, TCGA RNA-seq data indicate that genes associated with top 10 gained enhancers are upregulated in tumors whereas genes associated with top 10 lost enhancers are downregulated. This tumor specificity is restricted to ccRCC, but not the other two RCC subtypes, papillary and chromophobe. C, Expression of ZNF395 and SMPDL3A are measured in a panel of normal kidney cell lines (black) and ccRCC cell lines (red) by RT-qPCR. D, Pooled siRNA against ZNF395 inhibits colony formation of A-498 and 786-O but not HK2 normal immortalized kidney cells. Pooled siRNA against SMPLD3A inhibits colony formation of A-498 but not 786-O. E, H3K27ac ChIP-seq shows an active ZNF395 superenhancer only in ccRCC cells (A-498 and 786-O) but not normal kidney cells (PCS-400, HK2). F, TCGA RNA-seq data show exclusive overexpression of ZNF395 among 12 cancer types. G–I,ZNF395 inhibition by two shRNA clones decreases colony formation (G) and in vitro proliferation (H) and increases apoptosis measured by cleavage of caspase 3/7 substrate (I). *, P < 0.05, two-sided t test. J,ZNF395 inhibition by shRNA leads to total elimination of A-498 tumors in vivo and delayed 786-O tumor growth. NC: n = 7; shZNF395-1: n = 7; shZNF395-2: n = 6.

Close modal

Conversely, genes associated with lost superenhancers were recurrently suppressed in ccRCC and included EFHD1, EHF, MAL, GCOM1, and HOXB9 (Fig. 3B). In contrast to the lineage-specific nature of tumor superenhancers, genes associated with lost superenhancers were common between ccRCC and papillary RCC, implying a more universal function of tumor suppressor genes. For example, EHF/ESE2, a tumor suppressor previously found in prostate cancer (58, 59), exhibited reduced expression across all three RCC subtypes (ccRCC tumor/normal = 0.05, P = 3 × 10−15; papillary tumor/normal = 0.1, P 2 × 10−6; chromophobe tumor/normal = 0.1, P = 2 × 10−6).

Because current therapeutic targets in kidney cancer are limited to angiogenesis and mTOR pathways (3), we sought to examine these less-understood genes uncovered by superenhancer profiling. We chose ZNF395 and SMPDL3Afor their differential tumor expression (6–7 tumor–normal ratio; Fig. 3B) and high abundance (average RPKM of ZNF395 ∼112; average RPKM of SMPDL3A ∼58). Even though ZNF395 was previously identified as a potential ccRCC biomarker (60), its functional role in ccRCC malignancy remains unexplored. SMPDL3A shares 31% amino acid identity with the acid sphingomyelinase SMPD1 and is a target of a master regulator of cholesterol metabolism, liver X receptors (LXR; ref. 61).

Quantitative PCR (Fig. 3C) and immunoblotting (Supplementary Fig. S4C) confirmed that A-498 and 786-O ccRCC cells exhibited high expression of ZNF395 and SMPDL3A, whereas normal kidney proximal tubule cells, PCS-400 and HK2, exhibited low expression of both genes. siRNA mediated knockdown of SMPDL3A had a cell line–dependent effect on colony formation, inhibiting the growth of A-498 cells but having no observable effect on 786-O cells (Fig. 3D). On the other hand, ZNF395 consistently inhibited colony formation in both 786-O and A-498 cells but had minimal effect on normal kidney cells (Fig. 3D; Supplementary Fig. S4D). Consistent with this phenotypic observation, the ZNF395 superenhancer was active only in ccRCC cells (786-O and A-498) but silent in normal kidney cells (HK2 and PCS-400; Fig. 3E). Furthermore, among 12 types of cancer profiled by TCGA, ZNF395 was exclusively overexpressed in ccRCC tumors, consistent with the proposed lineage- and disease-specific nature of superenhancers (Fig. 3F).

No study to date has functionally tested the tumorigenic requirement of ZNF395 in ccRCC or any other cancer type. We validated ZNF395's tumor-promoting effect using individual shRNA clones (Supplementary Fig. S4E and S4F). Two independent ZNF395 shRNA clones drastically decreased in vitro colony formation (Fig. 3G) and cell viability (Fig. 3H) in both A-498 and 786-O cells. ZNF395 knockdown also resulted in increased apoptosis measured by cleavage of caspase 3/7 substrates (Fig. 3I) and Annexin V staining (Supplementary Fig. S4G). In vivo, tumor formation studies in mouse xenograft models revealed marked tumor suppression by ZNF395 depletion (Fig. 3J). Knockdown of ZNF395 led to elimination of A-498 tumors up to day 74, when tumors in the control group began to exceed the size limits imposed by institutional animal protocols. Similarly, ZNF395 depletion significantly slowed in vivo tumor growth of 786-O cells (Fig. 3J). Taken together, we showed the indispensable role ZNF395 plays in ccRCC tumorigenesis.

VHL Deficiency Remodels ccRCC Enhancer Landscapes

To explore the extent to which epigenetic changes observed in primary ccRCCs (Fig. 1) are directly driven by VHL loss, we examined chromatin changes in isogenic cell lines with and without VHL restoration. Consistent with earlier functional studies of VHL (62–64), VHL restoration in 786-O, A-498, and 12364284 cells had negligible effects on proliferation, colony formation, and apoptosis in vitro, but profoundly delayed tumor growth in vivo (Supplementary Fig. S5A–S5D), suggesting the importance of VHL in modulating processes required for in vivo tumorigenesis, including tumor–stroma cross-talk, angiogenesis, cell–matrix interactions, or tumor metabolism.

Focusing on the same regions defined in the primary tumors (4,719 gained promoters, 4,906 gained enhancers, and 1,157 gained superenhancers; Fig. 1A), we examined VHL-driven H3K27ac changes in four different cell lines (two commercial cell lines: 786-O and A-498; and two patient-derived cell lines: 12364284 and 40911432). Consistently across all four cell lines, VHL restoration induced more pronounced changes on enhancers and superenhancers than on promoters (Fig. 4A; Supplementary Fig. S6A–S6C). For example, in 786-O cells, after VHL restoration 12% of enhancers (549 enhancers) were significantly depleted, compared with 6.5% of promoters (321 promoters; Fig. 4A). We confirmed that a greater fraction of enhancers were significantly altered by VHL restoration than promoters (P < 2.2 × 10−16, proportions test), and an even higher proportion involved gained superenhancers (P < 2.2 × 10−16, proportions test).

Figure 4.

VHL deficiency remodels ccRCC enhancers. A, Log fold changes of H3K27ac ChIP-seq signals at gained promoters, enhancers, and superenhancers as defined in the primary ccRCC dataset after VHL restoration in 786-O cells. Red dots represent cis-regulatory elements with significant changes (P < 0.05, negative binomial) in H3K27ac levels after VHL restoration. The number and percentage of altered regions (P < 0.05, negative binomial) are shown at the upper and lower right corners. EV, empty vector. B, Read coverage of H3K27ac ChIP-seq at VHL-responsive enhancers in VHL-mutant ccRCC cell lines (red) compared with VHL–wild-type ccRCC (gray) and normal kidney cell lines (green) and 31 other cancer cell lines (black). C, Changes in expression of genes linked to VHL-responsive tumor enhancers after VHL restoration in 786-O cells. *, P < 0.05, two-sided t test. D–G, Examples of lost VHL-responsive enhancers/superenhancers are associated with EGFR (D), CCND1 (E), ITGB3 (F), and VEGFA (G) in 786-O cells. H, Log fold changes of H3K27ac (red), H3K4me1 (blue), and H3K27me3 (brown) signals at gained enhancers showing H3K27ac depletion after VHL restoration in 786-O cells.

Figure 4.

VHL deficiency remodels ccRCC enhancers. A, Log fold changes of H3K27ac ChIP-seq signals at gained promoters, enhancers, and superenhancers as defined in the primary ccRCC dataset after VHL restoration in 786-O cells. Red dots represent cis-regulatory elements with significant changes (P < 0.05, negative binomial) in H3K27ac levels after VHL restoration. The number and percentage of altered regions (P < 0.05, negative binomial) are shown at the upper and lower right corners. EV, empty vector. B, Read coverage of H3K27ac ChIP-seq at VHL-responsive enhancers in VHL-mutant ccRCC cell lines (red) compared with VHL–wild-type ccRCC (gray) and normal kidney cell lines (green) and 31 other cancer cell lines (black). C, Changes in expression of genes linked to VHL-responsive tumor enhancers after VHL restoration in 786-O cells. *, P < 0.05, two-sided t test. D–G, Examples of lost VHL-responsive enhancers/superenhancers are associated with EGFR (D), CCND1 (E), ITGB3 (F), and VEGFA (G) in 786-O cells. H, Log fold changes of H3K27ac (red), H3K4me1 (blue), and H3K27me3 (brown) signals at gained enhancers showing H3K27ac depletion after VHL restoration in 786-O cells.

Close modal

Even though gained enhancers were expected to show only depletion after VHL restoration, changes in H3K27ac levels were bidirectional (Fig. 4A). However, only gained enhancers with H3K27ac depletion were uniquely active in VHL-mutated ccRCC cell lines (786-O, A-498, and 12364284) compared with VHL–wild-type ccRCC cells (86049102L), normal kidney cell lines (PCS-400, HK2, and HKC-8), and 31 other cell lines of various cancer types (Fig. 4B). The lack of H3K27ac signals in normal kidney cell lines argues against tissue lineage as the dominant contributor to the high H3K27ac ChIP-seq signals seen in ccRCC cell lines. On the other hand, gained enhancers with H3K27ac enrichment after VHL restoration showed high activity across multiple cancer types, suggesting that these enhancers are not unique to ccRCC (Fig. 4B).

Furthermore, only gained enhancers showing H3K27ac depletion after VHL restoration were significantly associated with a concomitant downregulation of gene expression of their putative targets in both 786-O and 12364284 cells, whereas enhancers gained in primary ccRCCs and further H3K27ac-enriched after VHL restoration did not lead to significant gene upregulation on a global level (Fig. 4C; Supplementary Fig. S6D). These results suggest that the former enhancers (H3K27ac depletion) are likely to represent ccRCC- and VHL-specific epigenomic alterations, whereas the latter enhancers (H3K27ac enrichment) are likely to represent signify generic, compensatory mechanisms in response to VHL restoration.

Combining data from multiple lines, a total of 1,564 enhancers were depleted by VHL restoration in ≥1 cell line, representing almost a third (32%) of all gained enhancers identified in primary ccRCC tumors (Supplementary Table S13). The proportion of VHL-responsive enhancers increased with the level of patient recurrence—only 7.8% of nonrecurrent gained enhancers (1/10 patients) showed VHL-mediated H3K27ac depletion, whereas 18% of enhancers recurrently gained in 9 of 10 patients and 20% of enhancers gained in 10 of 10 patients showed H3K27ac depletion in 786-O cells (Supplementary Fig. S7A, P = 0.0001, proportions test), consistent with the high prevalence of VHL mutations (9/10 patients) in our discovery set. Interestingly, unsupervised clustering using the 1,564 VHL-responsive gained enhancers segregated the single VHL–wild-type tumor (ID 75416923) away from the remaining 9 VHL-mutant tumors (Supplementary Fig. S7B), with the VHL–wild-type tumor showing low H3K27ac signals at the ZNF395 superenhancer comparable with its patient-matched normal (Supplementary Fig. S7C). Collectively, pathway analysis of enhancers depleted in ≥2 cell lines highlighted direct p53 effectors, integrin-linked kinase signaling, and HIF1α transcription factor networks as the top five pathways (Supplementary Table S14), covering genes such as EGFR (Fig. 4D), CCND1 (Fig. 4E), ITGB3 (Fig. 4F), VEGFA (Fig. 4G), SLC2A1 (Supplementary Fig. S7D), and HK2 (Supplementary Fig. S7E). These results support a major role for VHL loss in ccRCC enhancer malfunction, even in the presence of other driver mutations.

We also examined whether other histone marks were concomitantly altered with H3K27ac marks. We found a surprisingly high degree of correlation between H3K27ac and H3K4me1 in response to VHL restoration in both 786-O cells (r = 0.77, Pearson correlation, Supplementary Fig. S7F) and 12364284 cells (r = 0.61, Pearson correlation, Supplementary Fig. S7G). Globally, enhancers exhibiting H3K27ac depletion also experienced concomitant H3K4me1 depletion (Fig. 4H). We next examined whether VHL restoration led to acquisition of the H3K27me3 repressive mark. Despite a moderate anticorrelation of H3K27ac and H3K27me3 (786-O cells: r = −0.28, Pearson correlation, Supplementary Fig. S7H; 12364284 cells: r = −0.22, Pearson correlation, Supplementary Fig. S7I), H3K27me3 levels remained low at gained enhancers even after VHL restoration (Fig. 4H). These findings suggest that VHL restoration may result in a loss of enhancer identity by codepletion of H3K27ac and H3K4me1, but not a formal transition to a poised enhancer state that would have retained H3K4me1 but acquired H3K27me3.

HIF2α–HIF1β Heterodimers Are Enriched at VHL-Responsive Enhancers

We sought to investigate which transcription factors might mediate VHL-dependent chromatin remodeling at gained enhancers. Beginning with the primary ccRCC dataset, we looked for enrichment of trans-regulators in gained enhancers over lost enhancers. Using HOMER (65), we found that the top enriched motifs were the AP1 family, ETS family, and NF-κB–p65–Rel and HIF1α/2α motifs (Fig. 5A, full list of motifs in Supplementary Table S15). For subsequent in vitro validation, we chose c-Jun as a representative AP1 family member because of its activation in ccRCC (66) and ETS1 as an ETS family representative because of its known interaction with HIF2α (67), but acknowledge that other family AP1 and ETS family members may play a role in ccRCC. Immunoblotting of c-Jun, ETS1, and NF-κB–p65 showed variable protein expression in both normal and tumor cell lines, but expression of HIF1α and HIF2α restricted to tumor cells only (Fig. 5B). HIF2α was expressed in a higher proportion of ccRCC cell lines than HIF1α (Fig. 5B). We further examined gene expression of these transcription factors in the TCGA cohort and found that ETS1, RELA (subunit of NF-κB–p65), and HIF2α were significantly overexpressed in tumors compared with normal tissues, with a range of tumor-association expression patterns similar to variations in ccRCC lines (Supplementary Fig. S8A).

Figure 5.

HIF2α is enriched at enhancers of VHL-responsive tumors. A, Motif analysis of gained enhancers using HOMER reveals significant enrichment of AP1 family, ETS family, NF-κB, and HIF1α/2α (hypergeometric test). Lost enhancers were used as background in the motif search to identify tumor-specific transcription factors. B, Protein expressions of putative transcription factors enriched at gained enhancers in 9 tumor cell lines (T; 4 commercial cell lines and 5 patient-derived cell lines) and 2 normal cell lines (N). ACHN is a papillary RCC cell line. C, ChIP-seq validated the enrichment of transcription factors (TF) in gained enhancers over lost enhancers. D, Protein expression of transcription factors is shown in 786-O and 12364284 cells with and without wild-type VHL. E, Transcription factor binding at VHL-responsive gained enhancers shows enrichment of HIF2α and HIF1β at enhancers with H3K27ac depletion (red) over regions with H3K27ac enrichment (black) after VHL restoration. F, ChIP-seq data show distribution of exogenous HIF1α and endogenous HIF2α binding at altered promoters and enhancers in 786-O cells that have been genetically engineered to overexpress HIF1α. G, ChIP-seq shows distribution of endogenous HIF1α and HIF2α binding at altered promoters and enhancers in 40911432 cells. H, Transcription factor binding at VHL-responsive enhancers shows higher enrichment of HIF2α than HIF1α at enhancers with H3K27ac depletion after VHL restoration (red) over regions with H3K27ac enrichment after VHL restoration (black). I, Example of a VHL-responsive enhancer near UBR4 with only HIF2α binding but not HIF1α binding. J, Example of a VHL-responsive superenhancer near CMIP with only HIF2α binding but not HIF1α binding.

Figure 5.

HIF2α is enriched at enhancers of VHL-responsive tumors. A, Motif analysis of gained enhancers using HOMER reveals significant enrichment of AP1 family, ETS family, NF-κB, and HIF1α/2α (hypergeometric test). Lost enhancers were used as background in the motif search to identify tumor-specific transcription factors. B, Protein expressions of putative transcription factors enriched at gained enhancers in 9 tumor cell lines (T; 4 commercial cell lines and 5 patient-derived cell lines) and 2 normal cell lines (N). ACHN is a papillary RCC cell line. C, ChIP-seq validated the enrichment of transcription factors (TF) in gained enhancers over lost enhancers. D, Protein expression of transcription factors is shown in 786-O and 12364284 cells with and without wild-type VHL. E, Transcription factor binding at VHL-responsive gained enhancers shows enrichment of HIF2α and HIF1β at enhancers with H3K27ac depletion (red) over regions with H3K27ac enrichment (black) after VHL restoration. F, ChIP-seq data show distribution of exogenous HIF1α and endogenous HIF2α binding at altered promoters and enhancers in 786-O cells that have been genetically engineered to overexpress HIF1α. G, ChIP-seq shows distribution of endogenous HIF1α and HIF2α binding at altered promoters and enhancers in 40911432 cells. H, Transcription factor binding at VHL-responsive enhancers shows higher enrichment of HIF2α than HIF1α at enhancers with H3K27ac depletion after VHL restoration (red) over regions with H3K27ac enrichment after VHL restoration (black). I, Example of a VHL-responsive enhancer near UBR4 with only HIF2α binding but not HIF1α binding. J, Example of a VHL-responsive superenhancer near CMIP with only HIF2α binding but not HIF1α binding.

Close modal

To further investigate chromatin occupancy of these factors, we generated ChIP-seq binding profiles of c-Jun, ETS1, and NF-κB cells and reexamined HIF2α, HIF1α, and HIF1β binding profiles from the previous literature (21, 30), all performed in 786-O cells. Of note, because 786-O cells have lost endogenous HIF1α expression through genomic deletion, the HIF1α ChIP-seq was performed on 786-O cells genetically manipulated to reexpress HIF1α protein (30). ChIP-seq results showed that all six transcription factors exhibited increased occupancy at gained enhancers compared with lost enhancers, validating the HOMER predictions (Fig. 5C).

To determine which of these transcription factors might be directly dependent on VHL, we then compared their protein expression in VHL-mutated isogenic cell lines with and without wild-type–VHL restoration. As shown in Fig. 5D, VHL restoration consistently downregulated HIF2α expression in both 786-O and 12364284 cell lines, but protein levels of other factors displayed contrasting trends between the two cell lines, implying that among the six factors examined, HIF2α protein expression was the most VHL dependent. Indeed, supporting an important role for HIF2α in VHL-dependent enhancer remodeling, only HIF2α and HIF1β were significantly enriched at enhancers showing VHL-dependent H3K27ac depletion (Fig. 5E). Moreover, among all known motifs in the HOMER database, HIF2α was the most enriched motif at VHL-responsive enhancers exhibiting H3K27ac depletion (P = 1 × 10−11; Supplementary Table S16).

In contrast, HIF1α was not enriched at enhancers showing H3K27ac depletion (Fig. 5E). Despite sharing many binding sites with HIF2α, HIF1α predominantly localized to promoter-proximal regions, whereas HIF2α frequently occupied introns and intergenic regions in 786-O cells (Supplementary Fig. S8B), consistent with a promoter-centric occupancy of HIF1α and an enhancer-centric occupancy of HIF2α (Fig. 5F). Gained enhancers displayed a HIF2α occupancy twice that of tumor-specific promoters (P < 1 × 10−16, proportions test) in 786-O cells, suggesting that HIF2α may play a greater role in regulating enhancers than promoters.

To extend these HIF1α and HIF2α occupancy-pattern findings to a system that expresses endogenous levels of both factors, we then performed HIF1α and HIF2α ChIP-seq in 40911432 ccRCC cells, which abundantly coexpress both HIFα subunits (Fig. 5B). Similar to 786-O, in 40911432 cells, HIF1α showed a preferential occupancy at promoter-proximal regions, whereas a large proportion of HIF2α was found in distal regions (introns and distal intergenic regions; Supplementary Fig. S8C). A higher proportion of HIF1α binding sites overlapped with gained promoters than HIF2α (68% of HIF1α vs. 41% of HIF2α, P = 0.002, proportions test; Fig. 5G). Conversely, a higher proportion of HIF2α binding sites overlapped with gained enhancers than HIF1α (29% of HIF1α vs. 51% of HIF2α, P < 2.2 × 10−16, proportions test). HIF2α's preferential occupancy at enhancers was further substantiated by its higher enrichment at enhancers showing H3K27ac depletion after VHL restoration than HIF1α (Fig. 5H). Specific examples of VHL-responsive enhancers bound exclusively by HIF2α but not HIF1α included an enhancer near UBR4 (Fig. 5I) and a superenhancer near CMIP (Fig. 5J). Therefore, even in HIF1α/HIF2α coexpressing ccRCC cells, these results suggest that HIF2α plays a greater role in VHL-mediated enhancer remodeling than HIF1α.

HIF2α–HIF1β-Bound Enhancers Modulate Gene Expression

To investigate the extent to which HIF2α silencing is sufficient to recapitulate the effects of VHL restoration, we performed H3K27ac ChIP-seq and RNA sequencing (RNA-seq) in 786-O cells with HIF2α siRNA-mediated knockdown and analyzed correlations between HIF2α siRNA knockdown and VHL restoration. When assessed against all genes, there was a low correlation (r = 0.1, P = 5.2 × 10−31) between HIF2α knockdown and VHL restoration. Importantly, however, this correlation increased to 0.23 (P = 5.8 × 10−14) for genes near HIF2α binding sites (Fig. 6A). Similar results were obtained at the epigenomic level, where for gained enhancers the correlation was low at 0.06 across all gained enhancers (P = 1.9 × 10−5) but increased substantially to 0.37 (P = 9.5 × 10−8) at HIF2α-bound enhancers (Fig. 6B) and at superenhancers increased from 0.089 (P = 0.0025) to 0.25 (P = 0.00054) at HIF2α-bound superenhancers (Fig. 6C). As a visual example, H3K27ac signals at the ZNF395 superenhancer were diminished after VHL restoration or HIF2α knockdown, concomitant with decreased ZNF395 gene expression (Fig. 6D). Validation by RT-qPCR showed that HIF2α siRNA knockdown downregulated VEGFA, SLC2A1, and ZNF395 expression to a comparable degree as VHL restoration (Fig. 6E). Decreases in luciferase reporter activity of enhancer elements were also consistent between HIF2α siRNA knockdown and VHL restoration (Fig. 6F).

Figure 6.

HIF2α–HIF1β bound enhancers modulate gene expression. A, Pearson correlation of gene expression changes after either VHL restoration or HIF2α siRNA knockdown at all genes or genes adjacent to HIF2α binding sites. EV, empty vector. B, Pearson correlation of H3K27ac changes after VHL restoration and HIF2α siRNA knockdown at either all gained enhancers or HIF2α-bound enhancers adjacent to binding sites. C, Pearson correlation of H3K27ac changes after VHL restoration and HIF2α siRNA knockdown at either all gained superenhancers or HIF2α-bound superenhancers adjacent to binding sites. D, Changes in RNA-seq and H3K27ac ChIP-seq signals after VHL restoration or HIF2α siRNA knockdown at ZNF395 superenhancer (SE), together with binding profiles of transcription factors enriched at enhancers. E, Both VHL restoration and HIF2α siRNA knockdown decrease expression of genes with HIF2α-bound enhancers in 786-O cells. *, P < 0.05, two-sided t test. F, Both VHL restoration and HIF2α siRNA knockdown decrease enhancer activities measured by luciferase reporter assay in 786-O cells. *, P < 0.05, two-sided t test. G, RT-qPCR measurement of ZNF395 expression in four wild-type clones (black) and four clones with ZNF395 enhancer depleted by CRISPR (red). Depleted region has the highest HIF2α binding at ZNF395 superenhancer in 786-O cells (deleted region indicated in D). *, P < 0.05, two-sided t test.

Figure 6.

HIF2α–HIF1β bound enhancers modulate gene expression. A, Pearson correlation of gene expression changes after either VHL restoration or HIF2α siRNA knockdown at all genes or genes adjacent to HIF2α binding sites. EV, empty vector. B, Pearson correlation of H3K27ac changes after VHL restoration and HIF2α siRNA knockdown at either all gained enhancers or HIF2α-bound enhancers adjacent to binding sites. C, Pearson correlation of H3K27ac changes after VHL restoration and HIF2α siRNA knockdown at either all gained superenhancers or HIF2α-bound superenhancers adjacent to binding sites. D, Changes in RNA-seq and H3K27ac ChIP-seq signals after VHL restoration or HIF2α siRNA knockdown at ZNF395 superenhancer (SE), together with binding profiles of transcription factors enriched at enhancers. E, Both VHL restoration and HIF2α siRNA knockdown decrease expression of genes with HIF2α-bound enhancers in 786-O cells. *, P < 0.05, two-sided t test. F, Both VHL restoration and HIF2α siRNA knockdown decrease enhancer activities measured by luciferase reporter assay in 786-O cells. *, P < 0.05, two-sided t test. G, RT-qPCR measurement of ZNF395 expression in four wild-type clones (black) and four clones with ZNF395 enhancer depleted by CRISPR (red). Depleted region has the highest HIF2α binding at ZNF395 superenhancer in 786-O cells (deleted region indicated in D). *, P < 0.05, two-sided t test.

Close modal

We sought to establish a causal link between HIF2α-bound enhancers and control of gene expression. We performed CRISPR-mediated genomic depletion of the ZNF395 enhancer region with the highest HIF2α peak (Fig. 6G). All four clones with the homozygous deleted ZNF395 enhancer consistently downregulated their ZNF395 expression compared with clones with the intact enhancer (P < 0.05), providing evidence that ZNF395 expression is epigenetically controlled by this HIF2α–HIF1β-bound enhancer (Fig. 6G). Taken together, these results indicate that that HIF2α is likely an important mediator of VHL-driven enhancer remodeling.

VHL Restoration Reduces p300 Recruitment but Preserves Promoter–Enhancer Interactions

Finally, we sought to investigate why VHL restoration caused a decrease in H3K27ac levels. Previous pulldown assays have reported that both HIF2α and HIF1β can interact with histone acetyltransferase p300 (68–71). Indeed, p300 frequently marks enhancers (43) and is thought to be recruited by tissue-specific transcription factors (72). However, chromatin profiles of p300 have not been previously established in kidney cancer cell lines, so the contribution of p300 in shaping enhancers in ccRCC remains unclear. Therefore, we performed p300 ChIP-seq in 786-O cells and confirmed its enrichment at gained enhancers over lost enhancers (Fig. 7A). Comparing p300 ChIP-seq with HIF2α ChIP-seq yielded a surprisingly high degree of overlap between HIF2α and p300 (96%), even more than that of HIF2α and HIF1β (89%; Fig. 7B and C). In contrast, other transcription factors such as c-Jun, ETS1, and NF-κB did not exhibit such a high degree of overlap (≤60%; Fig. 7B).

Figure 7.

VHL restoration reduces p300 recruitment but preserves promoter–enhancer interactions. A, Enrichment of p300 binding at gained and lost enhancers based on ChIP-seq. B, Percentage of overlap between HIF2α and other transcription factors. C, ChIP-seq binding profiles of HIF2α and p300. D, Protein expression of p300 with and without VHL in 786-O and 12364284 cells measured by immunoblotting. EV, empty vector. E, ChIP-qPCR of p300 binding at enhancers with and without VHL restoration in 786-O cells. NC, negative control regions. F, ChIP-qPCR of p300 binding at enhancers with and without HIF2α siRNA knockdown in 786-O cells. NC, negative control regions. G, Correlation of enhancer interactions measured by Capture-C (RPM, reads per million) between 786-O cells with and without VHL restoration at both VHL-responsive and non–VHL-responsive enhancers. H, Capture-C shows that VEGFA enhancer–promoter interactions are maintained even after VHL restoration. E, enhancer; P, promoter. I, Schematics of VHL-driven enhancer aberration in ccRCC.

Figure 7.

VHL restoration reduces p300 recruitment but preserves promoter–enhancer interactions. A, Enrichment of p300 binding at gained and lost enhancers based on ChIP-seq. B, Percentage of overlap between HIF2α and other transcription factors. C, ChIP-seq binding profiles of HIF2α and p300. D, Protein expression of p300 with and without VHL in 786-O and 12364284 cells measured by immunoblotting. EV, empty vector. E, ChIP-qPCR of p300 binding at enhancers with and without VHL restoration in 786-O cells. NC, negative control regions. F, ChIP-qPCR of p300 binding at enhancers with and without HIF2α siRNA knockdown in 786-O cells. NC, negative control regions. G, Correlation of enhancer interactions measured by Capture-C (RPM, reads per million) between 786-O cells with and without VHL restoration at both VHL-responsive and non–VHL-responsive enhancers. H, Capture-C shows that VEGFA enhancer–promoter interactions are maintained even after VHL restoration. E, enhancer; P, promoter. I, Schematics of VHL-driven enhancer aberration in ccRCC.

Close modal

We compared p300 binding at tumor enhancers with and without VHL. Despite increased p300 protein levels in 786-O cells after VHL restoration (Fig. 7D), binding of p300 decreased across all four enhancers examined (Fig. 7E). HIF2α depletion by siRNA knockdown also decreased p300 recruitment (Fig. 7F), suggesting that loss of HIF2α may interfere with p300 recruitment.

We investigated whether VHL restoration and the subsequent loss of p300 binding disrupted promoter–enhancer interactions. We performed Capture-C of enhancer regions in paired 786-O cell lines with and without VHL restoration. Surprisingly, Capture-C interactions showed a relatively high correlation between VHL-deficient and VHL-restored 786-O cells at VHL-responsive regions (r = 0.74, Pearson correlation), even higher than correlations observed at non–VHL-responsive regions (r = 0.57, Pearson correlation; Fig. 7G). As a visual example, interactions between the VEGFA promoter and enhancer were intact even after VHL restoration (Fig. 7H), indicating that loss of enhancer activity is likely insufficient to dissociate promoter–enhancer interactions. Furthermore, many of these promoter–enhancers were lineage specific; for example, the interaction between SLC2A1 enhancer with its promoter was not detected in KATOIII, a gastric cancer cell line (Supplementary Fig. S9). Therefore, promoter–enhancer interactions often preexist in kidney cells, frequently in a tissue-specific manner.

Understanding epigenomic alterations and their genetic origin can identify new disease mechanisms (34), vulnerabilities (73, 74), and therapeutic strategies (75–77). Through comprehensive profiling of histone modifications in primary normal–tumor pairs and cell lines, we generated a compendium of ccRCC-associated promoters and enhancers. Our study demonstrates that the most frequent ccRCC mutational event—VHL inactivation—leads to genome-wide enhancer and superenhancer remodeling, which directly imparts ccRCC hallmarks including angiogenesis and metabolic reprogramming. ZNF395, epigenetically controlled by a VHL-responsive superenhancer, emerged as a crucial regulator of ccRCC tumorigenesis.

Our work has three main advances. First, to our knowledge, this is the most comprehensive atlas of histone profiles in ccRCC and will likely provide an invaluable resource to the ccRCC field. Using high-resolution multiplexed interactome data (Capture-C, ref. 56) and H3K27ac-expression correlation, we minimized ambiguity in enhancer assignment and further confirmed the dependency of enhancers on VHL/HIF status by reporter assays. Second, using isogenic cell lines, we show that VHL loss contributes significantly to enhancer remodeling. Even though another mutation in ccRCC, SETD2, can mediate widespread increases in chromatin accessibility (46) and DNA hypomethylation (78), its relatively low mutation frequency at ∼10% in all ccRCC tumors (78) cannot explain epigenetic abnormalities in the vast majority of SETD2–wild-type tumors. Lastly, an examination of somatically altered superenhancers enabled us to identify a master regulator crucial to the pathogenesis of ccRCC, ZNF395. Even though ZNF395 overexpression in ccRCC has been previously reported (79–81) and its proximity to a superenhancer was independently noted (42), our study is the first to pinpoint the specific VHL-dependent enhancer required for ZNF395 expression and to show ZNF395's indispensable functional role for ccRCC tumorigenesis in vitro and in vivo.

Our data suggest that mechanistically, loss of VHL stabilizes HIF2α occupancy at tumor-specific gained enhancers, which in turn recruits histone acetyltransferase p300 (28, 82) to maintain H3K27 acetylation, upregulating expression of ccRCC-specific genes such as ZNF395 (Fig. 7I). Restoration of wild-type VHL resulted in codepletion of H3K27ac and H3K4me1 marks and thus abrogation of active enhancer identity at tumor-associated enhancers. Compared with the promoter-centric occupancy of HIF1α, HIF2α is predominantly found at enhancers, pointing toward a key difference between HIF1α and HIF2α. We also found that HIF2α siRNA knockdown specifically attenuates the activity of HIF2α-bound enhancers/superenhancers. Interestingly, the majority of promoter–enhancer interactions remained largely unaltered by VHL status, suggesting that these promoter–enhancer couplings are largely stable and preformed. This is consistent with a recent report demonstrating that promoter–enhancer interactions remain largely unchanged between normoxia and hypoxia (29). Our study demonstrating VHL's impact on chromatin remodeling also suggests that other cancer genes with high tumor-type–specific mutational penetrance, such as BRAF in melanoma (83) and APC in colon cancer (84), may also act to modify cellular epigenomes to effect broad yet disease-specific cellular changes, despite these genes not being classic chromatin modifiers.

Besides VHL, mutations in other genes such as PBRM1, SETD2, ARID1A, SMARCA4, JARID1C, and KDM6A/UTX have been reported in ccRCC (47, 85), and these are likely to augment VHL's core transcriptional effects (86), contributing to heterogeneity in disease phenotypes (87) and progression patterns (88). Using the example of 786-O cells, at least two other mutations in these cells may have a direct impact on chromatin—MLL3 (p.A3902G) and gain-of-function TP53 mutations (p.R248W). MLL3, a histone 3 lysine 4 methyltransferase, is directly responsible for formation of the H3K4me1 enhancer mark (89, 90) and plays a critical role for enhancer regulation (91). Gain-of-function TP53 mutants also bind aberrantly to chromatin, especially near methyltransferases MLL1 and MLL2, potentially contributing to tumor growth via chromatin deregulation (92). Besides mutations, structural variants are also known to alter enhancers via enhancer hijacking (93) or copy-number gains (94) in other cancers. Given the multitude of driver and bystander mutations in ccRCC, it is thus unlikely that VHL alone can account for all epigenomic changes observed in this tumor type. Nevertheless, by integrating data across multiple ccRCC cell lines, our data suggest that VHL inactivation is likely to account for almost a third (32%) of all gained enhancer regions, supporting its role as a dominant driver of epigenetic abnormalities in ccRCC despite the presence of other genetic changes.

Our epigenetic maps contain a wellspring of both well-validated and uncharacterized targets that may contribute to ccRCC tumorigenesis. We found extensive enhancer gains around well-characterized hypoxia-related targets (ref. 95; VEGFA, CXCR4, ref. 96; HK2), SLC-mediated membrane transporters (SLC2A1, SLC2A2, SLC38A1, ref. 97), SLC16A family (98), and adipogenesis (PLIN2, refs. 51, 99). New targets revealed in this study include SMPDL3A, which could be another important ccRCC-specific oncogene given its role in lipid and cholesterol metabolism (61, 100). Genes associated with lost superenhancers, which could only be identified with normal–tumor pairs, implicated potential tumor suppressors (EHF, MAL, GCOM1, and HOXB9) that warrant further investigation.

One notable finding from this epigenomic study is the tumorigenic requirement of ZNF395 in ccRCC. ZNF395 is also known as HDBP2 (101) or papillomavirus binding factor (PBF; ref. 102). ZNF395 is required for the differentiation of mesenchymal stem cells to adipocytes, by partnering with PPARγ2 to promote adipogenesis (103). ZNF395 has been shown to bind to the promoters of Huntington gene (101) and interferon-induced genes, and to cause upregulation of cancer-related genes (MACC1, PEG10, CALCOCO1, and MEF2C; ref. 104) and proangiogenic chemokines including IL6 and IL8 under hypoxia (105). It remains to be elucidated in future studies the precise mechanism contributing to ZNF395's tumorigenic role.

The enhancer landscapes profiled in this study have implications beyond ccRCC. The poorly perfused tumor core makes hypoxia a feature of virtually all solid tumors (106). MCF7 cells under hypoxia (but not normoxia) share similar H3K27ac profiles as 786-O (29). Although ZNF395 is highly expressed in ccRCC, its low basal expression can be upregulated upon hypoxia in other cancer types, including glioblastoma and skin cancer (104, 105). Targeting ZNF395 or its downstream effectors in future studies may be therapeutically relevant to both ccRCC and other hypoxic solid malignancies. Direct targeting of ZNF395 using a peptide-based cancer vaccine is undergoing phase I trials in patients with sarcoma (107–109), opening up the possibility of using immunotherapy to target the extracellular fragments of nuclear master regulators. Our study suggests that initiating a similar trial in ccRCC may be worthwhile. Moreover, given the recent progress in targeting transcription factors using various modalities, including small molecules and stapled peptides (110–112), inhibitors of ZNF395 may provide an important therapeutic inroad for ccRCC treatment.

Patient Information

Fresh–frozen normal–tumor tissues were obtained from nephrectomy cases under approvals from institutional research ethics review committees and patient consent under institutional review board protocol 2010/735/B. Normal tissues were harvested from sites distant from the tumor. Refer to Supplementary Table S1 for detailed patient information.

Cell Lines

Commercial cell lines (786-O, A-498, HK2, and PCS-400) were purchased from ATCC. Cell lines were maintained in RPMI (Invitrogen) with 10% FBS with the exception of primary renal proximal tubule epithelial cells, PCS-400, which were maintained in Renal Epithelial Cell Basal Medium (ATCC). Cell line authentication was performed by short tandem repeat (STR) analysis (Cancer Science Institute of Singapore) in 2015 against publicly available STR profiles. Mycoplasma testing was performed using the MycoSensor PCR assay kit (Stratagene).

Establishment of Tumor-Derived Cell Lines from Primary Tumors

Tumor cells were disassociated from primary tumors by collagenase, seeded, and maintained in RPMI with 10% FBS. At 80% to 90% confluency, the cells were passaged at a 1:3 ratio. Cultured cells were considered to be successfully immortalized after 60 passages. Correct pairing of tumor tissues and cell lines was achieved by comparing the percentage identity of single nucleotide polymorphisms (SNP) based on targeted sequencing. All tumor–cell line pairs showed identities of > 90% whereas shuffling of pairing showed identities < 80%. Tumors and cell lines from 12364284 and 40911432 showed the same VHL mutations, but 86049102 tissue (named 86049102T) is VHL-mutant, whereas the cognate 86049102 cell line (named 86049102L) is VHL–wild-type.

Stable VHL Restoration in ccRCC Lines

786-O cells (WT2, VHL+) and 786-O cells (RC3, VHL) were kindly provided by Dr. Michael Ohh (University of Toronto). Stable transduction of VHL was performed in A-498, 12364284, and 40911432 cells as follows: HA-VHL wt-pBabe-puro plasmid (a gift from Dr. William Kaelin, Dana Farber Cancer Institute; Addgene plasmid # 19234) was transfected into PlatA cells (RV-102, Cell Biolabs) at 2 μg DNA/well of a 6-well plate using Lipofectamine 3000 (LifeTechnologies). A medium change was performed 10 to 16 hours after transfection. The supernatant from PlatA cells containing retroviruses was harvested 48 hours later and added to ccRCC cells, which were then selected with puromycin for 3 days after transduction.

Histone Nano-ChIP-seq

Nano-ChIP-seq was performed as previously described (113) with slight modifications. Fresh–frozen cancer and normal tissues were dissected using a razor blade to obtain ∼5 mg of tissue. The tissues were fixed in 1% formaldehyde for 10 minutes at room temperature. Fixation was stopped by addition of glycine to a final concentration of 125 nmol/L. Tissue pieces were washed 3 times with TBSE buffer. Pulverized tissues were lysed in 100 μL lysis buffer and sonicated for 16 cycles (30s on, 30s off) using a Bioruptor (Diagenode). The following antibodies were used: H3K27ac (ab4729, Abcam), H3K4me3 (07-473, Millipore), H3K4me1 (ab8895, Abcam), and H3K27me3 (07-449, Millipore). The total volume of immunoprecipitation was 1 mL and the amount of antibody used was 2 μg. The input DNA was precleared with protein G Dynabeads (Life Technologies) for 1 hour at 4°C and then incubated with antibodies conjugated protein G beads overnight at 4°C. The beads were washed 3 times with cold wash buffer. After recovery of ChIP and input DNA, whole-genome amplification was performed using the WGA4 kit (Sigma-Aldrich) and BpmI-WGA primers. Amplified DNA was digested with BpmI [New England Biolabs (NEB)]. After that, 30 ng of the amplified DNA was used with the NEBNext ChIP-seq library prep reagent set (NEB). ChIP-seq in cell lines was performed using the same Nano-ChIP-seq protocol described above but with 1 × 106 cells. Each library was sequenced to an average depth of 20 to 30 million raw reads on HiSeq2500 using 101-bp single end reads.

Histone ChIP-seq Analysis

Sequencing tags were mapped against the human reference genome (hg19) using Burrows-Wheeler Aligner (BWA-mem; ref. 114; version 0.7.10). Reads were trimmed 10 bp from the front and the back to produce 81 bp. Only reads with mapQ >10 and with duplicates removed by rmdup were used for subsequent analysis. Significant peaks were called using CCAT (P < 0.05; ref. 115). The strength and quality of immunoprecipitation were assessed using CHANCE (116).

Transcription Factor ChIP-seq

For each transcription factor, 3 × 107 cells were cross-linked with 1% formaldehyde for 10 minutes at room temperature and stopped by adding glycine to a final concentration of 125 nmol/L. Chromatin was extracted and sonicated to ∼500 bp (Vibra cell, SONICS). The following antibodies were used for chromatin immunoprecipitation: c-Jun (sc-1694, Santa Cruz Biotechnology), NF-κB p65 (sc-372, Santa Cruz Biotechnology), ETS1 (sc-350, Santa Cruz Biotechnology), HIF1α (610959, BD Biosciences), HIF2α (NB100-122, Novus Bio), and p300 (sc-585, Santa Cruz Biotechnology). The total volume of immunoprecipitation was 1.5 mL and the amount of antibody used was 15 μg. Input DNAs were precleared with protein G Dynabeads (LifeTechnologies) for 2 hours at 4°C and then incubated with antibody-conjugated protein G beads overnight at 4°C. The beads were washed 6 times with wash buffer at room temperature. At least 10 ng of the DNA was used with the NEBNext ChIP-seq library prep reagent set (NEB). Each library was sequenced to an average depth of 30 to 50 million reads on a HiSeq2500 using 101-bp single-end reads.

Capture-C

Capture-C was performed as previously described (56). Briefly, 1 × 107 cells were cross-linked by 2% formaldehyde, followed by lysis, homogenization, DpnII digestion, ligation, and de–cross-linking. DNA was sonicated using a Covaris to 150 to 200 bp to produce DNA suitable for oligo capture. A total of 3 μg of sheared DNA was used for sequencing library preparation (NEB). Enhancer sequences were double captured by hybridization to customized biotinylated oligos (IDT) and enriched with Dynabeads (Life Technologies). Captured DNA was sequenced to an average depth of 2 million reads per probe on the HiSeq Illumina platform using 150-bp paired-end reads.

Capture-C Analysis and Gene Assignment

Preprocessing of raw reads was performed to remove adaptor sequences (trim_galore, http://www.bioinformatics.babraham.ac.uk/projects/trim_galore/), and overlapping reads were merged using FLASH (117). In order to achieve short read mapping to the hg19 reference genome, the resulting preprocessed reads were then in silico digested with DpnII and aligned using Bowtie (using p1, m2, best, and strata settings). Aligned reads were processed using Capture-C analyzer (118) to (i) remove PCR duplicates; (ii) classify subfragments as “capture” if they were contained within the capture fragment, “proximity exclusion” if they were within 1 Kb on either side of the capture fragment, or “reporter” if they were outside of the “capture” and “proximity exclusion” regions; and (iii) normalize read counts per 100,000 interactions in bigwig format. We additionally used the r3Cseq package (119) on the capture and reporter fragments to identify significant interactions of the viewpoint against a scaled background (q-value < 0.05). Gene assignment is defined by the overlap of significant Capture-C peaks with genes with start and end defined by GENCODE v19. Interactions were plotted using Epigenome Gateway v40.0.

Identification of Differentially Enriched Regions

Significant H3K27ac peaks called by CCAT were merged across all normal–tumor samples. The same was performed with H3K4me1 and H3K4me3 ChIP-seq data. TSS were based on GENCODE v19. Promoters were defined as regions of overlap between H3K27ac and H3K4me3 and also overlapping with ±2.0 Kb around the TSS. Enhancers were defined as regions of overlap between H3K27ac and H3K4me1 but not overlapping with promoters. To minimize stromal contamination, we performed further filtering using cell line data, where enhancers and promoters not overlapping with H3K27ac peaks in any of the cell lines were discarded. Wiggle files of window size 50 bp were generated using MEDIPs (120) from bam files. The input-subtracted signal for each promoter or enhancer region was computed using bigWigAverageOverBed to yield reads per kilobase per million (RPKM). The RPKM of H3K27ac, H3K4me1, and H3K4me3 ChIP-seq from promoters and enhancers were corrected for batch effects using Combat. Tumor-specific regions were defined as regions that have a fold difference of ≥2 and a difference of 0.5 RPKM from patient-matched normal tissue. Normal regions were defined as regions that have a fold difference of ≤0.5, and a difference of −0.5 RPKM from the corresponding regions in patient-matched tumors. Recurrently gained regions were defined as gain in ≥5/10 patients and no loss in any patients. Recurrently lost regions were defined as loss in ≥5/10 patients and no gain in any patients. Statistical testing for each cis-regulatory region was performed using paired t tests with Benjamini–Hochberg correction. The differential regions were visualized using NGSplot (121).

Identification of Superenhancer Regions

Superenhancer regions were identified using ROSE (ref. 35; with promoter excluded), using H3K27ac peak regions merged from all patients (both normal and tumor tissue). Wiggle files of window size 50 bp were generated using MEDIPs (120) from bam files. The input-subtracted signal for each superenhancer was computed using bigWigAverageOverBed (sum of reads over covered bases). The superenhancer regions were ranked by the average difference of normal–tumor H3K27ac ChIP-seq signals. Gained superenhancers were defined as regions that have average differential H3K27ac ChIP-seq signals >0. Lost superenhancers were defined as regions that have average differential H3K27ac ChIP-seq signals <0.

Additional methods can be found in Supplementary Methods.

Data Accession

ChIP-seq and RNA-seq data are available at Gene Expression Omnibus (GSE86095). https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE86095

No potential conflicts of interest were disclosed.

Conception and design: X. Yao, J. Tan, B.C. Tan, B.T. Teh, P. Tan

Development of methodology: X. Yao, J. Tan, K.J. Lim, Z. Li, M. Xing, G. Wijaya, J.O.J. Davies, G. Loh, B.C. Tan, S.G. Rozen, S. Li, D.L. Silver, J.R. Hughes, B.T. Teh, P. Tan

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): X. Yao, J. Tan, Z. Li, D. Huang, M. Xing, Y.S. Chan, J.Z. Qu, S.T. Tay, Y.N. Lam, J.H. Hong, A.P. Lee-Lim, C.Z. He, C. Xu, S.S. Myint, B.C. Tan, S.G. Rozen, C.W.S. Cheng, K.T.E. Chang, P.H. Tan, A. Lezhava, B.T. Teh

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): X. Yao, J. Tan, K.J. Lim, W.F. Ooi, P. Guan, C.Z. He, T. Nandi, A. Qamra, S.G. Rozen, B.T. Teh, P. Tan

Writing, review, and/or revision of the manuscript: X. Yao, J. Tan, K.J. Lim, T. Nandi, B.T. Teh, P. Tan

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): X. Yao, J. Tan, K.J. Lim, J. Koh, Z. Li, Y.S. Chan, S.T. Tay, M.S.W. Ng, J.S. Lin, T. Nandi, G. Loh, B.C. Tan, S.G. Rozen, Q. Yu, I.B.H. Tan, A. Lezhava, G. Steger, B.T. Teh, P. Tan

Study supervision: X. Yao, S.G. Rozen, B.T. Teh, P. Tan

Other (provided important technical advice and reagents): J.Y. Goh

We thank Hui Hoon Chua, Angie Tan, and Liang Kai Koh for their administrative support. We thank the Sequencing and Scientific Computing teams at the Genome Institute of Singapore for sequencing services and data management capabilities, and the Duke-NUS Genome Biology Facility for sequencing services. We thank the Advanced Molecular Pathology Laboratory of the Institute of Molecular and Cell Biology for pathology evaluation. Xiaosai Yao dedicates this paper to the memory of her mother, Tan Limin, for her unconditional love and support.

This work was supported by Biomedical Research Council Young Investigator Grant 1510851024, National Natural Science Foundation of China 61661146004, National Research Foundation grant NRF2016NRF-NSFC001-057, and National Medical Research Council grants NMRC/STaR/0026/2015, NMRC/STAR/0024/2014, NMRC/CIRG/1402/2014, and NMRC/CBRG/069/2014. Core funding was provided by the Genome Institute of Singapore and Institute of Molecular and Cell Biology under the Agency for Science, Technology and Research, and Duke-NUS Medical School. Other sources of support include the Cancer Science Institute of Singapore, NUS, funded by the National Research Foundation Singapore and the Singapore Ministry of Education under its Research Centres of Excellence initiative, and Biomedical Research Council Strategic Positioning Fund 2012/001. J.R. Hughes was supported by a Wellcome Trust Strategic Award (reference 106130/Z/14/Z) and Medical Research Council (MRC) Core funding. J.O.J. Davies was funded by a Wellcome Trust Clinical Research Training Fellowship (ref 098931/Z/12/Z).

1.
Ferlay J
SI
,
Ervik
M
,
Dikshit
R
,
Eser
S
,
Mathers
C
,
Rebelo
M
, et al
GLOBOCAN
.
Cancer Incidence and Mortality Worldwide: IARC CancerBase No. 11
. 
2014
.
2.
Society AC
.
Survival rates for kidney cancer by stage
.
Atlanta, GA
:
Society AC
.
3.
Rini
BI
,
Flaherty
K
. 
Clinical effect and future considerations for molecularly-targeted therapy in renal cell carcinoma
.
Urol Oncol
2008
;
26
:
543
9
.
4.
Rini
BI
,
Atkins
MB
. 
Resistance to targeted therapy in renal-cell carcinoma
.
Lancet Oncol
2009
;
10
:
992
1000
.
5.
Gerlinger
M
,
Horswell
S
,
Larkin
J
,
Rowan
AJ
,
Salm
MP
,
Varela
I
, et al
Genomic architecture and evolution of clear cell renal cell carcinomas defined by multiregion sequencing
.
Nat Genet
2014
;
46
:
225
33
.
6.
Gerlinger
M
,
Rowan
AJ
,
Horswell
S
,
Larkin
J
,
Endesfelder
D
,
Gronroos
E
, et al
Intratumor heterogeneity and branched evolution revealed by multiregion sequencing
.
N Engl J Med
2012
;
366
:
883
92
.
7.
Gu
YF
,
Cohn
S
,
Christie
A
,
McKenzie
T
,
Wolff
NC
,
Do
QN
, et al
Modeling Renal cell carcinoma in mice: Bap1 and Pbrm1 inactivation drive tumor grade
.
Cancer Discov
2017
;
7
:
900
17
.
8.
Nargund
AM
,
Pham
CG
,
Dong
Y
,
Wang
PI
,
Osmangeyoglu
HU
,
Xie
Y
, et al
The SWI/SNF protein PBRM1 restrains VHL-loss-driven clear cell renal cell carcinoma
.
Cell Rep
2017
;
18
:
2893
906
.
9.
Harlander
S
,
Schonenberger
D
,
Toussaint
NC
,
Prummer
M
,
Catalano
A
,
Brandt
L
, et al
Combined mutation in Vhl, Trp53 and Rb1 causes clear cell renal cell carcinoma in mice
.
Nat Med
2017
;
23
:
869
77
.
10.
Wang
SS
,
Gu
YF
,
Wolff
N
,
Stefanius
K
,
Christie
A
,
Dey
A
, et al
Bap1 is essential for kidney function and cooperates with Vhl in renal tumorigenesis
.
Proc Natl Acad Sci U S A
2014
;
111
:
16538
43
.
11.
Bailey
ST
,
Smith
AM
,
Kardos
J
,
Wobker
SE
,
Wilson
HL
,
Krishnan
B
, et al
MYC activation cooperates with Vhl and Ink4a/Arf loss to induce clear cell renal cell carcinoma
.
Nat Commun
2017
;
8
:
15770
.
12.
Iwai
K
,
Yamanaka
K
,
Kamura
T
,
Minato
N
,
Conaway
RC
,
Conaway
JW
, et al
Identification of the von Hippel–Lindau tumor-suppressor protein as part of an active E3 ubiquitin ligase complex
.
Proc Natl Acad Sci U S A
1999
;
96
:
12436
41
.
13.
Lisztwan
J
,
Imbert
G
,
Wirbelauer
C
,
Gstaiger
M
,
Krek
W
. 
The von Hippel–Lindau tumor suppressor protein is a component of an E3 ubiquitin-protein ligase activity
.
Genes Dev
1999
;
13
:
1822
33
.
14.
Maxwell
PH
,
Wiesener
MS
,
Chang
GW
,
Clifford
SC
,
Vaux
EC
,
Cockman
ME
, et al
The tumour suppressor protein VHL targets hypoxia-inducible factors for oxygen-dependent proteolysis
.
Nature
1999
;
399
:
271
5
.
15.
Jaakkola
P
,
Mole
DR
,
Tian
YM
,
Wilson
MI
,
Gielbert
J
,
Gaskell
SJ
, et al
Targeting of HIF-alpha to the von Hippel–Lindau ubiquitylation complex by O2-regulated prolyl hydroxylation
.
Science
2001
;
292
:
468
72
.
16.
Ebert
BL
,
Bunn
HF
. 
Regulation of transcription by hypoxia requires a multiprotein complex that includes hypoxia-inducible factor 1, an adjacent transcription factor, and p300/CREB binding protein
.
Mol Cell Biol
1998
;
18
:
4089
96
.
17.
Ema
M
,
Hirota
K
,
Mimura
J
,
Abe
H
,
Yodoi
J
,
Sogawa
K
, et al
Molecular mechanisms of transcription activation by HLF and HIF1alpha in response to hypoxia: their stabilization and redox signal-induced interaction with CBP/p300
.
EMBO J
1999
;
18
:
1905
14
.
18.
Biswas
S
,
Troy
H
,
Leek
R
,
Chung
YL
,
Li
JL
,
Raval
RR
, et al
Effects of HIF-1alpha and HIF2alpha on growth and metabolism of clear-cell renal cell carcinoma 786-0 xenografts
.
J Oncol
2010
;
2010
:
757908
.
19.
Metallo
CM
,
Gameiro
PA
,
Bell
EL
,
Mattaini
KR
,
Yang
J
,
Hiller
K
, et al
Reductive glutamine metabolism by IDH1 mediates lipogenesis under hypoxia
.
Nature
2012
;
481
:
380
4
.
20.
Rankin
EB
,
Rha
J
,
Selak
MA
,
Unger
TL
,
Keith
B
,
Liu
Q
, et al
Hypoxia-inducible factor 2 regulates hepatic lipid metabolism
.
Mol Cell Biol
2009
;
29
:
4527
38
.
21.
Schodel
J
,
Bardella
C
,
Sciesielski
LK
,
Brown
JM
,
Pugh
CW
,
Buckle
V
, et al
Common genetic variants at the 11q13.3 renal cancer susceptibility locus influence binding of HIF to an enhancer of cyclin D1 expression
.
Nat Genet
2012
;
44
:
420
5
,
S1–2
.
22.
Roe
JS
,
Kim
H
,
Lee
SM
,
Kim
ST
,
Cho
EJ
,
Youn
HD
. 
p53 stabilization and transactivation by a von Hippel–Lindau protein
.
Mol Cell
2006
;
22
:
395
405
.
23.
Tamiya
H
,
Ikeda
T
,
Jeong
JH
,
Saito
T
,
Yano
F
,
Jung
YK
, et al
Analysis of the Runx2 promoter in osseous and non-osseous cells and identification of HIF2A as a potent transcription activator
.
Gene
2008
;
416
:
53
60
.
24.
Ko
CY
,
Tsai
MY
,
Tseng
WF
,
Cheng
CH
,
Huang
CR
,
Wu
JS
, et al
Integration of CNS survival and differentiation by HIF2alpha
.
Cell Death Differ
2011
;
18
:
1757
70
.
25.
Imtiyaz
HZ
,
Williams
EP
,
Hickey
MM
,
Patel
SA
,
Durham
AC
,
Yuan
LJ
, et al
Hypoxia-inducible factor 2alpha regulates macrophage function in mouse models of acute and tumor inflammation
.
J Clin Invest
2010
;
120
:
2699
714
.
26.
Xie
L
,
Collins
JF
. 
Transcription factors Sp1 and Hif2alpha mediate induction of the copper-transporting ATPase (Atp7a) gene in intestinal epithelial cells during hypoxia
.
J Biol Chem
2013
;
288
:
23943
52
.
27.
Ruas
JL
,
Berchner-Pfannschmidt
U
,
Malik
S
,
Gradin
K
,
Fandrey
J
,
Roeder
RG
, et al
Complex regulation of the transactivation function of hypoxia-inducible factor-1 alpha by direct interaction with two distinct domains of the CREB-binding protein/p300
.
J Biol Chem
2010
;
285
:
2601
9
.
28.
Freedman
SJ
,
Sun
ZY
,
Poy
F
,
Kung
AL
,
Livingston
DM
,
Wagner
G
, et al
Structural basis for recruitment of CBP/p300 by hypoxia-inducible factor-1 alpha
.
Proc Natl Acad Sci U S A
2002
;
99
:
5367
72
.
29.
Platt
JL
,
Salama
R
,
Smythies
J
,
Choudhry
H
,
Davies
JO
,
Hughes
JR
, et al
Capture-C reveals preformed chromatin interactions between HIF-binding sites and distant promoters
.
EMBO Rep
2016
;
17
:
1410
21
.
30.
Salama
R
,
Masson
N
,
Simpson
P
,
Sciesielski
LK
,
Sun
M
,
Tian
YM
, et al
Heterogeneous effects of direct hypoxia pathway activation in kidney cancer
.
PLoS One
2015
;
10
:
e0134645
.
31.
Grampp
S
,
Platt
JL
,
Lauer
V
,
Salama
R
,
Kranz
F
,
Neumann
VK
, et al
Genetic variation at the 8q24.21 renal cancer susceptibility locus affects HIF binding to a MYC enhancer
.
Nat Commun
2016
;
7
:
13183
.
32.
Heinz
S
,
Romanoski
CE
,
Benner
C
,
Glass
CK
. 
The selection and function of cell type-specific enhancers
.
Nat Rev Mol Cell Biol
2015
;
16
:
144
54
.
33.
Heintzman
ND
,
Hon
GC
,
Hawkins
RD
,
Kheradpour
P
,
Stark
A
,
Harp
LF
, et al
Histone modifications at human enhancers reflect global cell-type-specific gene expression
.
Nature
2009
;
459
:
108
12
.
34.
Lin
CY
,
Erkek
S
,
Tong
Y
,
Yin
L
,
Federation
AJ
,
Zapatka
M
, et al
Active medulloblastoma enhancers reveal subgroup-specific cellular origins
.
Nature
2016
;
530
:
57
62
.
35.
Whyte
WA
,
Orlando
DA
,
Hnisz
D
,
Abraham
BJ
,
Lin
CY
,
Kagey
MH
, et al
Master transcription factors and mediator establish super-enhancers at key cell identity genes
.
Cell
2013
;
153
:
307
19
.
36.
Liu
F
,
Hon
GC
,
Villa
GR
,
Turner
KM
,
Ikegami
S
,
Yang
H
, et al
EGFR mutation promotes glioblastoma through epigenome and transcription factor network remodeling
.
Mol Cell
2015
;
60
:
307
18
.
37.
Riggi
N
,
Knoechel
B
,
Gillespie
SM
,
Rheinbay
E
,
Boulay
G
,
Suva
ML
, et al
EWS-FLI1 utilizes divergent chromatin remodeling mechanisms to directly activate or repress enhancer elements in Ewing sarcoma
.
Cancer Cell
2014
;
26
:
668
81
.
38.
Niu
X
,
Zhang
T
,
Liao
L
,
Zhou
L
,
Lindner
DJ
,
Zhou
M
, et al
The von Hippel–Lindau tumor suppressor protein regulates gene expression and tumor growth through histone demethylase JARID1C
.
Oncogene
2012
;
31
:
776
86
.
39.
Ramakrishnan
S
,
Ku
S
,
Ciamporcero
E
,
Miles
KM
,
Attwood
K
,
Chintala
S
, et al
HDAC 1 and 6 modulate cell invasion and migration in clear cell renal cell carcinoma
.
BMC Cancer
2016
;
16
:
617
.
40.
Beyer
S
,
Kristensen
MM
,
Jensen
KS
,
Johansen
JV
,
Staller
P
. 
The histone demethylases JMJD1A and JMJD2B are transcriptional targets of hypoxia-inducible factor HIF
.
J Biol Chem
2008
;
283
:
36542
52
.
41.
Luo
W
,
Chang
R
,
Zhong
J
,
Pandey
A
,
Semenza
GL
. 
Histone demethylase JMJD2C is a coactivator for hypoxia-inducible factor 1 that is required for breast cancer progression
.
Proc Natl Acad Sci U S A
2012
;
109
:
E3367
76
.
42.
Rhie
SK
,
Guo
Y
,
Tak
YG
,
Yao
L
,
Shen
H
,
Coetzee
GA
, et al
Identification of activated enhancers and linked transcription factors in breast, prostate, and kidney tumors by tracing enhancer networks using epigenetic traits
.
Epigenetics Chromatin
2016
;
9
:
50
.
43.
Heintzman
ND
,
Stuart
RK
,
Hon
G
,
Fu
Y
,
Ching
CW
,
Hawkins
RD
, et al
Distinct and predictive chromatin signatures of transcriptional promoters and enhancers in the human genome
.
Nat Genet
2007
;
39
:
311
8
.
44.
Creyghton
MP
,
Cheng
AW
,
Welstead
GG
,
Kooistra
T
,
Carey
BW
,
Steine
EJ
, et al
Histone H3K27ac separates active from poised enhancers and predicts developmental state
.
Proc Natl Acad Sci U S A
2010
;
107
:
21931
6
.
45.
Akhtar-Zaidi
B
,
Cowper-Sal-lari
R
,
Corradin
O
,
Saiakhova
A
,
Bartels
CF
,
Balasubramanian
D
, et al
Epigenomic enhancer profiling defines a signature of colon cancer
.
Science
2012
;
336
:
736
9
.
46.
Simon
JM
,
Hacker
KE
,
Singh
D
,
Brannon
AR
,
Parker
JS
,
Weiser
M
, et al
Variation in chromatin accessibility in human kidney cancer links H3K36 methyltransferase loss with widespread RNA processing defects
.
Genome Res
2014
;
24
:
241
50
.
47.
Cancer Genome Atlas Research N
. 
Comprehensive molecular characterization of clear cell renal cell carcinoma
.
Nature
2013
;
499
:
43
9
.
48.
Iyer
MK
,
Niknafs
YS
,
Malik
R
,
Singhal
U
,
Sahu
A
,
Hosono
Y
, et al
The landscape of long noncoding RNAs in the human transcriptome
.
Nat Genet
2015
;
47
:
199
208
.
49.
Purdue
MP
,
Johansson
M
,
Zelenika
D
,
Toro
JR
,
Scelo
G
,
Moore
LE
, et al
Genome-wide association study of renal cell carcinoma identifies two susceptibility loci on 2p21 and 11q13.3
.
Nat Genet
2011
;
43
:
60
5
.
50.
McLean
CY
,
Bristor
D
,
Hiller
M
,
Clarke
SL
,
Schaar
BT
,
Lowe
CB
, et al
GREAT improves functional interpretation of cis-regulatory regions
.
Nat Biotechnol
2010
;
28
:
495
501
.
51.
Qiu
B
,
Ackerman
D
,
Sanchez
DJ
,
Li
B
,
Ochocki
JD
,
Grazioli
A
, et al
HIF2alpha-dependent lipid storage promotes endoplasmic reticulum homeostasis in clear-cell renal cell carcinoma
.
Cancer Discov
2015
;
5
:
652
67
.
52.
Kim
JW
,
Tchernyshyov
I
,
Semenza
GL
,
Dang
CV
. 
HIF-1-mediated expression of pyruvate dehydrogenase kinase: a metabolic switch required for cellular adaptation to hypoxia
.
Cell Metab
2006
;
3
:
177
85
.
53.
Papandreou
I
,
Cairns
RA
,
Fontana
L
,
Lim
AL
,
Denko
NC
. 
HIF-1 mediates adaptation to hypoxia by actively downregulating mitochondrial oxygen consumption
.
Cell Metab
2006
;
3
:
187
97
.
54.
Wise
DR
,
DeBerardinis
RJ
,
Mancuso
A
,
Sayed
N
,
Zhang
XY
,
Pfeiffer
HK
, et al
Myc regulates a transcriptional program that stimulates mitochondrial glutaminolysis and leads to glutamine addiction
.
Proc Natl Acad Sci U S A
2008
;
105
:
18782
7
.
55.
Gameiro
PA
,
Yang
J
,
Metelo
AM
,
Perez-Carro
R
,
Baker
R
,
Wang
Z
, et al
In vivo HIF-mediated reductive carboxylation is regulated by citrate levels and sensitizes VHL-deficient cells to glutamine deprivation
.
Cell Metab
2013
;
17
:
372
85
.
56.
Davies
JO
,
Telenius
JM
,
McGowan
SJ
,
Roberts
NA
,
Taylor
S
,
Higgs
DR
, et al
Multiplexed analysis of chromosome conformation at vastly improved sensitivity
.
Nat Methods
2016
;
13
:
74
80
.
57.
Parker
SC
,
Stitzel
ML
,
Taylor
DL
,
Orozco
JM
,
Erdos
MR
,
Akiyama
JA
, et al
Chromatin stretch enhancer states drive cell-specific gene regulation and harbor human disease risk variants
.
Proc Natl Acad Sci U S A
2013
;
110
:
17921
6
.
58.
Albino
D
,
Longoni
N
,
Curti
L
,
Mello-Grand
M
,
Pinton
S
,
Civenni
G
, et al
ESE3/EHF controls epithelial cell differentiation and its loss leads to prostate tumors with mesenchymal and stem-like features
.
Cancer Res
2012
;
72
:
2889
900
.
59.
Cangemi
R
,
Mensah
A
,
Albertini
V
,
Jain
A
,
Mello-Grand
M
,
Chiorino
G
, et al
Reduced expression and tumor suppressor function of the ETS transcription factor ESE-3 in prostate cancer
.
Oncogene
2008
;
27
:
2877
85
.
60.
Dalgin
GS
,
Holloway
DT
,
Liou
LS
,
DeLisi
C
. 
Identification and characterization of renal cell carcinoma gene markers
.
Cancer Inform
2007
;
3
:
65
92
.
61.
Noto
PB
,
Bukhtiyarov
Y
,
Shi
M
,
McKeever
BM
,
McGeehan
GM
,
Lala
DS
. 
Regulation of sphingomyelin phosphodiesterase acid-like 3A gene (SMPDL3A) by liver X receptors
.
Mol Pharmacol
2012
;
82
:
719
27
.
62.
Iliopoulos
O
,
Kibel
A
,
Gray
S
,
Kaelin
WG
 Jr
. 
Tumour suppression by the human von Hippel–Lindau gene product
.
Nat Med
1995
;
1
:
822
6
.
63.
Gnarra
JR
,
Zhou
S
,
Merrill
MJ
,
Wagner
JR
,
Krumm
A
,
Papavassiliou
E
, et al
Post-transcriptional regulation of vascular endothelial growth factor mRNA by the product of the VHL tumor suppressor gene
.
Proc Natl Acad Sci U S A
1996
;
93
:
10589
94
.
64.
Schoenfeld
A
,
Davidowitz
EJ
,
Burk
RD
. 
A second major native von Hippel–Lindau gene product, initiated from an internal translation start site, functions as a tumor suppressor
.
Proc Natl Acad Sci U S A
1998
;
95
:
8817
22
.
65.
Heinz
S
,
Benner
C
,
Spann
N
,
Bertolino
E
,
Lin
YC
,
Laslo
P
, et al
Simple combinations of lineage-determining transcription factors prime cis-regulatory elements required for macrophage and B cell identities
.
Mol Cell
2010
;
38
:
576
89
.
66.
An
J
,
Liu
H
,
Magyar
CE
,
Guo
Y
,
Veena
MS
,
Srivatsan
ES
, et al
Hyperactivated JNK is a therapeutic target in pVHL-deficient renal cell carcinoma
.
Cancer Res
2013
;
73
:
1374
85
.
67.
Elvert
G
,
Kappel
A
,
Heidenreich
R
,
Englmeier
U
,
Lanz
S
,
Acker
T
, et al
Cooperative interaction of hypoxia-inducible factor-2alpha (HIF-2alpha) and Ets-1 in the transcriptional activation of vascular endothelial growth factor receptor-2 (Flk-1)
.
J Biol Chem
2003
;
278
:
7520
30
.
68.
Steunou
AL
,
Ducoux-Petit
M
,
Lazar
I
,
Monsarrat
B
,
Erard
M
,
Muller
C
, et al
Identification of the hypoxia-inducible factor 2alpha nuclear interactome in melanoma cells reveals master proteins involved in melanoma development
.
Mol Cell Proteomics
2013
;
12
:
736
48
.
69.
Kung
AL
,
Zabludoff
SD
,
France
DS
,
Freedman
SJ
,
Tanner
EA
,
Vieira
A
, et al
Small molecule blockade of transcriptional coactivation of the hypoxia-inducible factor pathway
.
Cancer Cell
2004
;
6
:
33
43
.
70.
Tohkin
M
,
Fukuhara
M
,
Elizondo
G
,
Tomita
S
,
Gonzalez
FJ
. 
Aryl hydrocarbon receptor is required for p300-mediated induction of DNA synthesis by adenovirus E1A
.
Mol Pharmacol
2000
;
58
:
845
51
.
71.
Kung
AL
,
Wang
S
,
Klco
JM
,
Kaelin
WG
,
Livingston
DM
. 
Suppression of tumor growth through disruption of hypoxia-inducible transcription
.
Nat Med
2000
;
6
:
1335
40
.
72.
Visel
A
,
Blow
MJ
,
Li
Z
,
Zhang
T
,
Akiyama
JA
,
Holt
A
, et al
ChIP-seq accurately predicts tissue-specific activity of enhancers
.
Nature
2009
;
457
:
854
8
.
73.
Bitler
BG
,
Aird
KM
,
Garipov
A
,
Li
H
,
Amatangelo
M
,
Kossenkov
AV
, et al
Synthetic lethality by targeting EZH2 methyltransferase activity in ARID1A-mutated cancers
.
Nat Med
2015
;
21
:
231
8
.
74.
Bernt
KM
,
Zhu
N
,
Sinha
AU
,
Vempati
S
,
Faber
J
,
Krivtsov
AV
, et al
MLL-rearranged leukemia is dependent on aberrant H3K79 methylation by DOT1L
.
Cancer Cell
2011
;
20
:
66
78
.
75.
Loven
J
,
Hoke
HA
,
Lin
CY
,
Lau
A
,
Orlando
DA
,
Vakoc
CR
, et al
Selective inhibition of tumor oncogenes by disruption of super-enhancers
.
Cell
2013
;
153
:
320
34
.
76.
McCabe
MT
,
Ott
HM
,
Ganji
G
,
Korenchuk
S
,
Thompson
C
,
Van Aller
GS
, et al
EZH2 inhibition as a therapeutic strategy for lymphoma with EZH2-activating mutations
.
Nature
2012
;
492
:
108
12
.
77.
Daigle
SR
,
Olhava
EJ
,
Therkelsen
CA
,
Majer
CR
,
Sneeringer
CJ
,
Song
J
, et al
Selective killing of mixed lineage leukemia cells by a potent small-molecule DOT1L inhibitor
.
Cancer Cell
2011
;
20
:
53
65
.
78.
Cancer Genome Atlas Research Network
. 
Comprehensive molecular characterization of clear cell renal cell carcinoma
.
Nature
2013
;
499
:
43
9
.
79.
Kim
Y
,
Choi
JW
,
Lee
JH
,
Kim
YS
. 
Loss of CDC14B expression in clear cell renal cell carcinoma: meta-analysis of microarray data sets
.
Am J Clin Pathol
2014
;
141
:
551
8
.
80.
Skubitz
KM
,
Zimmermann
W
,
Kammerer
R
,
Pambuccian
S
,
Skubitz
AP
. 
Differential gene expression identifies subgroups of renal cell carcinoma
.
J Lab Clin Med
2006
;
147
:
250
67
.
81.
Wozniak
MB
,
Le Calvez-Kelm
F
,
Abedi-Ardekani
B
,
Byrnes
G
,
Durand
G
,
Carreira
C
, et al
Integrative genome-wide gene expression profiling of clear cell renal cell carcinoma in Czech Republic and in the United States
.
PLoS One
2013
;
8
:
e57886
.
82.
Gu
J
,
Milligan
J
,
Huang
LE
. 
Molecular mechanism of hypoxia-inducible factor 1alpha-p300 interaction. A leucine-rich interface regulated by a single cysteine
.
J Biol Chem
2001
;
276
:
3550
4
.
83.
Hodis
E
,
Watson
IR
,
Kryukov
GV
,
Arold
ST
,
Imielinski
M
,
Theurillat
JP
, et al
A landscape of driver mutations in melanoma
.
Cell
2012
;
150
:
251
63
.
84.
Cancer Genome Atlas Network
. 
Comprehensive molecular characterization of human colon and rectal cancer
.
Nature
2012
;
487
:
330
7
.
85.
Varela
I
,
Tarpey
P
,
Raine
K
,
Huang
D
,
Ong
CK
,
Stephens
P
, et al
Exome sequencing identifies frequent mutation of the SWI/SNF complex gene PBRM1 in renal carcinoma
.
Nature
2011
;
469
:
539
42
.
86.
Gao
W
,
Li
W
,
Xiao
T
,
Liu
XS
,
Kaelin
WG
 Jr
. 
Inactivation of the PBRM1 tumor suppressor gene amplifies the HIF-response in VHL−/− clear cell renal carcinoma
.
Proc Natl Acad Sci U S A
2017
;
114
:
1027
32
.
87.
Bi
M
,
Zhao
S
,
Said
JW
,
Merino
MJ
,
Adeniran
AJ
,
Xie
Z
, et al
Genomic characterization of sarcomatoid transformation in clear cell renal cell carcinoma
.
Proc Natl Acad Sci U S A
2016
;
113
:
2170
5
.
88.
Ho
TH
,
Park
IY
,
Zhao
H
,
Tong
P
,
Champion
MD
,
Yan
H
, et al
High-resolution profiling of histone h3 lysine 36 trimethylation in metastatic renal cell carcinoma
.
Oncogene
2016
;
35
:
1565
74
.
89.
Herz
HM
,
Mohan
M
,
Garruss
AS
,
Liang
K
,
Takahashi
YH
,
Mickey
K
, et al
Enhancer-associated H3K4 monomethylation by Trithorax-related, the Drosophila homolog of mammalian Mll3/Mll4
.
Genes Dev
2012
;
26
:
2604
20
.
90.
Hu
D
,
Gao
X
,
Morgan
MA
,
Herz
HM
,
Smith
ER
,
Shilatifard
A
. 
The MLL3/MLL4 branches of the COMPASS family function as major histone H3K4 monomethylases at enhancers
.
Mol Cell Biol
2013
;
33
:
4745
54
.
91.
Dorighi
KM
,
Swigut
T
,
Henriques
T
,
Bhanu
NV
,
Scruggs
BS
,
Nady
N
, et al
Mll3 and Mll4 facilitate enhancer RNA synthesis and transcription from promoters independently of H3K4 monomethylation
.
Mol Cell
2017
;
66
:
568
76
e4
.
92.
Zhu
J
,
Sammons
MA
,
Donahue
G
,
Dou
Z
,
Vedadi
M
,
Getlik
M
, et al
Gain-of-function p53 mutants co-opt chromatin pathways to drive cancer growth
.
Nature
2015
;
525
:
206
11
.
93.
Northcott
PA
,
Lee
C
,
Zichner
T
,
Stutz
AM
,
Erkek
S
,
Kawauchi
D
, et al
Enhancer hijacking activates GFI1 family oncogenes in medulloblastoma
.
Nature
2014
;
511
:
428
34
.
94.
Zhang
X
,
Choi
PS
,
Francis
JM
,
Imielinski
M
,
Watanabe
H
,
Cherniack
AD
, et al
Identification of focally amplified lineage-specific super-enhancers in human epithelial cancers
.
Nat Genet
2016
;
48
:
176
82
.
95.
Schodel
J
,
Grampp
S
,
Maher
ER
,
Moch
H
,
Ratcliffe
PJ
,
Russo
P
, et al
Hypoxia, hypoxia-inducible transcription factors, and renal cancer
.
Eur Urol
2016
;
69
:
646
57
.
96.
Vanharanta
S
,
Shu
W
,
Brenet
F
,
Hakimi
AA
,
Heguy
A
,
Viale
A
, et al
Epigenetic expansion of VHL-HIF signal output drives multiorgan metastasis in renal cancer
.
Nat Med
2013
;
19
:
50
6
.
97.
Broer
A
,
Rahimi
F
,
Broer
S
. 
Deletion of amino acid transporter ASCT2 (SLC1A5) reveals an essential role for transporters SNAT1 (SLC38A1) and SNAT2 (SLC38A2) to sustain glutaminolysis in cancer cells
.
J Biol Chem
2016
;
291
:
13194
205
.
98.
Adijanto
J
,
Philp
NJ
. 
The SLC16A family of monocarboxylate transporters (MCTs)–physiology and function in cellular metabolism, pH homeostasis, and fluid transport
.
Curr Top Membr
2012
;
70
:
275
311
.
99.
Morrissey
JJ
,
Mellnick
VM
,
Luo
J
,
Siegel
MJ
,
Figenshau
RS
,
Bhayani
S
, et al
Evaluation of urine aquaporin-1 and perilipin-2 concentrations as biomarkers to screen for renal cell carcinoma: a prospective cohort study
.
JAMA Oncol
2015
;
1
:
204
12
.
100.
Traini
M
,
Quinn
CM
,
Sandoval
C
,
Johansson
E
,
Schroder
K
,
Kockx
M
, et al
Sphingomyelin phosphodiesterase acid-like 3A (SMPDL3A) is a novel nucleotide phosphodiesterase regulated by cholesterol in human macrophages
.
J Biol Chem
2014
;
289
:
32895
913
.
101.
Tanaka
K
,
Shouguchi-Miyata
J
,
Miyamoto
N
,
Ikeda
JE
. 
Novel nuclear shuttle proteins, HDBP1 and HDBP2, bind to neuronal cell-specific cis-regulatory element in the promoter for the human Huntington's disease gene
.
J Biol Chem
2004
;
279
:
7275
86
.
102.
Sichtig
N
,
Korfer
N
,
Steger
G
. 
Papillomavirus binding factor binds to SAP30 and represses transcription via recruitment of the HDAC1 co-repressor complex
.
Arch Biochem Biophys
2007
;
467
:
67
75
.
103.
Hasegawa
R
,
Tomaru
Y
,
de Hoon
M
,
Suzuki
H
,
Hayashizaki
Y
,
Shin
JW
. 
Identification of ZNF395 as a novel modulator of adipogenesis
.
Exp Cell Res
2013
;
319
:
68
76
.
104.
Jordanovski
D
,
Herwartz
C
,
Pawlowski
A
,
Taute
S
,
Frommolt
P
,
Steger
G
. 
The hypoxia-inducible transcription factor ZNF395 is controlled by IkB kinase-signaling and activates genes involved in the innate immune response and cancer
.
PLoS One
2013
;
8
:
e74911
.
105.
Herwartz
C
,
Castillo-Juarez
P
,
Schroder
L
,
Barron
BL
,
Steger
G
. 
The transcription factor ZNF395 is required for the maximal hypoxic induction of proinflammatory cytokines in U87-MG cells
.
Mediators Inflamm
2015
;
2015
:
804264
.
106.
Vaupel
P
,
Thews
O
,
Kelleher
DK
,
Hoeckel
M
. 
Current status of knowledge and critical issues in tumor oxygenation. Results from 25 years research in tumor pathophysiology
.
Adv Exp Med Biol
1998
;
454
:
591
602
.
107.
Tsukahara
T
,
Kawaguchi
S
,
Torigoe
T
,
Takahashi
A
,
Murase
M
,
Kano
M
, et al
HLA-A*0201-restricted CTL epitope of a novel osteosarcoma antigen, papillomavirus binding factor
.
J Transl Med
2009
;
7
:
44
.
108.
Tsukahara
T
,
Nabeta
Y
,
Kawaguchi
S
,
Ikeda
H
,
Sato
Y
,
Shimozawa
K
, et al
Identification of human autologous cytotoxic T-lymphocyte-defined osteosarcoma gene that encodes a transcriptional regulator, papillomavirus binding factor
.
Cancer Res
2004
;
64
:
5442
8
.
109.
Tsukahara
T
,
Emori
M
,
Murata
K
,
Hirano
T
,
Muroi
N
,
Kyono
M
, et al
Specific targeting of a naturally presented osteosarcoma antigen, papillomavirus binding factor peptide, using an artificial monoclonal antibody
.
J Biol Chem
2014
;
289
:
22035
47
.
110.
Chen
W
,
Hill
H
,
Christie
A
,
Kim
MS
,
Holloman
E
,
Pavia-Jimenez
A
, et al
Targeting renal cell carcinoma with a HIF-2 antagonist
.
Nature
2016
;
539
:
112
7
.
111.
Cho
H
,
Du
X
,
Rizzi
JP
,
Liberzon
E
,
Chakraborty
AA
,
Gao
W
, et al
On-target efficacy of a HIF-2alpha antagonist in preclinical kidney cancer models
.
Nature
2016
;
539
:
107
11
.
112.
Bernal
F
,
Wade
M
,
Godes
M
,
Davis
TN
,
Whitehead
DG
,
Kung
AL
, et al
A stapled p53 helix overcomes HDMX-mediated suppression of p53
.
Cancer Cell
2010
;
18
:
411
22
.
113.
Ng
JH
,
Kumar
V
,
Muratani
M
,
Kraus
P
,
Yeo
JC
,
Yaw
LP
, et al
In vivo epigenomic profiling of germ cells reveals germ cell molecular signatures
.
Dev Cell
2013
;
24
:
324
33
.
114.
Li
H
. 
Aligning sequence reads, clone sequences and assembly contigs with BWA-MEM
.
q-bioGN
2013
.
115.
Xu
H
,
Handoko
L
,
Wei
X
,
Ye
C
,
Sheng
J
,
Wei
CL
, et al
A signal-noise model for significance analysis of ChIP-seq with negative control
.
Bioinformatics
2010
;
26
:
1199
204
.
116.
Diaz
A
,
Nellore
A
,
Song
JS
. 
CHANCE: comprehensive software for quality control and validation of ChIP-seq data
.
Genome Biol
2012
;
13
:
R98
.
117.
Magoc
T
,
Salzberg
SL
. 
FLASH: fast length adjustment of short reads to improve genome assemblies
.
Bioinformatics
2011
;
27
:
2957
63
.
118.
Hughes
JR
,
Roberts
N
,
McGowan
S
,
Hay
D
,
Giannoulatou
E
,
Lynch
M
, et al
Analysis of hundreds of cis-regulatory landscapes at high resolution in a single, high-throughput experiment
.
Nat Genet
2014
;
46
:
205
12
.
119.
Thongjuea
S
,
Stadhouders
R
,
Grosveld
FG
,
Soler
E
,
Lenhard
B
. 
r3Cseq: an R/bioconductor package for the discovery of long-range genomic interactions from chromosome conformation capture and next-generation sequencing data
.
Nucleic Acids Res
2013
;
41
:
e132
.
120.
Lienhard
M
,
Grimm
C
,
Morkel
M
,
Herwig
R
,
Chavez
L
. 
MEDIPS: genome-wide differential coverage analysis of sequencing data derived from DNA enrichment experiments
.
Bioinformatics
2014
;
30
:
284
6
.
121.
Shen
L
,
Shao
N
,
Liu
X
,
Nestler
E
. 
ngs.plot: Quick mining and visualization of next-generation sequencing data by integrating genomic databases
.
BMC Genomics
2014
;
15
:
284
.