The transcriptional regulation of the human epidermal growth factor receptor-2 (HER2) contributes to an enhanced HER2 expression in HER2-positive breast cancers with HER2 gene amplification and HER2-low or HER2-negative breast cancers following radiotherapy or endocrine therapy, and this drives tumorigenesis and the resistance to therapy. Epigenetic mechanisms are critical for transcription regulation, however, such mechanisms in the transcription regulation of HER2 are limited to the involvement of tri-methylated histone 3 lysine 4 (H3K4me3) and acetylated histone 3 lysine 9 (H3K9ac) at the HER2 promoter region. Here, we report the identification of a novel enhancer in the HER2 3’ gene body, which we have termed HER2 gene body enhancer (HGE). The HGE starts from the 3’ end of intron 19 and extends into intron 22, possesses enhancer histone modification marks in specific cells and enhances the transcriptional activity of the HER2 promoters. We also found that TFAP2C, a known regulator of HER2, binds to HGE and is required for its enhancer function and that DNA methylation in the HGE region inhibits the histone modifications characterizing enhancer and is inversely correlated with HER2 expression in breast cancer samples. The identification of this novel enhancer sheds a light on the roles of epigenetic mechanisms in HER2 transcription, in both HER2-positive breast cancer samples and individuals with HER2-low or HER2-negative breast cancers undergoing radiotherapy or endocrine therapy.
Human epidermal growth factor receptor-2 (HER2)/Erb-B2 receptor tyrosine kinase 2 is a member of the erbB-like oncogene family, its overexpression occurs in approximately 20–30% of breast cancers1 and is strongly associated with poor prognosis.2 HER2 has roles in the development of HER2-positive breast cancers3, 4 and resistance to therapy in HER2-low or HER2-negative breast cancers, in which HER2 is transcriptionally upregulated by radiotherapy or by endocrine therapy.5, 6, 7 HER2 gene amplification is a major mechanism for HER2 overexpression, however, higher transcription rate of HER2 per gene copy was also observed in HER2-amplified breast cancer cells,8, 9, 10 that is, HER2 mRNA levels are 4- to 8-fold and 64- to 128-fold higher in HER2-overexpressing and HER2-amplified breast cancer cells, respectively, than would be expected from HER2 gene copy numbers.8 A run-on assay showed SKBR3 cells displayed about two fold HER2 transcription rate higher than in BT474 cells.11 Transcription factors such as TFAP2,12, 13 Sp1,14 PBP,15 YY1,16 ETS,17 YB-118 and EGR219 have been shown to positively regulate HER2, whereas MYB,20 FOXP3,21 GATA4,22 PEA3,23 MBP-1,24 NOTCH and RBP-Jk25 have negative effects. Most of these studies focused on the regulation of the originally characterized HER2 promoter (promoter 2),14, 26, 27 which has the dominant role in the overexpression of HER2 in breast cancers28 despite of the identification of an alternative promoter (promoter 1)29 (Figure 1a). Moreover, intron 1 enhancer when bound by PAX2 mediates transcriptional repression of HER2 by activated ER.5 Although these transcriptional mechanisms can explain HER2 regulation in part, the molecular basis of the increase in HER2 transcription in certain cancers remains unexplained.
Chromatin modifications can greatly influence transcriptional regulation and contribute to cancer development.30 H3K4me3 and H3K9ac, two histone marks typically associated with gene activation, were reported to be critical for inducing HER2 transcription through promoter 2.10 WDR5, a key component of the H3K4me3 methyltransferase complex, is essentially involved in this process.10 However, these mechanisms are common to general transcriptional activation. Thus, additional mechanisms may exist and specifically contribute to HER2 overexpression. We discovered a novel enhancer HER2 gene body enhancer (HGE) in the 3’ gene body of HER2.
The HGE activates promoters 1 and 2 in trans., and hence the TFAP2C-mediated transcriptional induction of HER2 expression. This novel regulatory mechanism of HER2 transcription contributes to the understanding of increased expression of HER2.
Results and discussion
Identification of a novel enhancer in HER2 locus
Enhancer interacts with promoter to recruit RNA polymerase II and regulate transcription.31 Chromatin signatures such as DNase I hypersensitivity sites32 and histone modifications, that are, H3K4me1 and H3K27ac can be used to predict putative enhancers.33, 34, 35 We took advantage of data from the Encyclopedia of DNA Elements (ENCODE) Consortium36 to search for novel regulatory element(s) that may contribute to the increase of HER2 transcription. With the ENCODE Regulation Super-track Settings, DNase I hypersensitivity sites, H3K4me1 and H3K27ac were found in the previously identified intron 1 enhancer5 (using the annotation of NM_001005862); 5-kb upstream of promoter 2; and in a previously undiscovered region within the 3’ gene body (Figure 1a). The 3’ gene body region starts from intron 19 and ends within intron 22 (based on NM_001289936) (Figure 1a) and, in addition to the above-described features, it contains binding sites for many transcription factors (including POLII, TEAD4, c-MYC and PHF8) identified in K562 cells (Supplementary Figure 1). We named this region as the HGE.
TFAP2C is known to positively regulate HER2 expression as it is required for the HER2 expression in BT474 cells;37 binds to and regulates HER2 promoter 2;12, 38, 39 mediates the repression of HER2 by estrogen;38 its expression is positively correlated with HER2 expression in primary breast cancers.37 We analyzed the TFAP2C occupancy data from four cell lines (GSE36351)40 and revealed that TFAP2C is enriched at both promoter 2 and intron 1 enhancer in all cell lines, with stronger enrichments in HER2-amplified SKBR3 and BT474 lines than in HER2-low MCF7 cells (Figure 1b). Importantly, TFAP2C also binds to the HGE in both SKBR3 and BT474 cells, with the stronger occupancy in SKBR3 cells than in BT474 cells (Figure 1b), which supports HGE as a candidate enhancer. Chromatin immunoprecipitation of TFAP2C, H3K4me1 and H3K27ac were performed in SKBR3 cells and TFAP2C occupancies were confirmed at promoters 1 and 2, the intron 1 enhancer and HGE (Figure 1c). H3K27ac is enriched at all four regions, whereas H3K4me1 enrichment was only obvious at HGE (Figure 1c). Chromatin immunoprecipitation experiments in another HER2-amplified breast cancer cell line, HCC1954 (Figure 1d), revealed that TFAP2C was significantly enriched at all four regions, however, its enrichments at both enhancers were dramatically lower than at promoter 2. H3K27ac was significantly enriched at both promoters and intron 1 enhancer, whereas, H3K4me1 was more enriched at promoter 2 compared with a slight increase at promoter 1 and HGE (Figure 1d). These data support the enhancer feature of the HGE and its cell type dependency. Importantly, a chromosome 17-wide binding data of ERRα and PGC-1β in SKBR3 cells showed binding of ERRα to HGE in addition to intron 1 enhancer and both promoters,41 indicating that the HGE can recruit additional transcription factors.
The HGE enhances the transcriptional activity of HER2 promoters
We tested the ability of the HGE to regulate the transcriptional activities of the HER2 promoters by luciferase reporter assay in 293T, SKBR3 and BT474 cells (Figure 2a). Both promoters (placed upstream of the Luciferase gene), but not the HGE (placed downstream of Luciferase gene), had basal activities in all cases (Figure 2a). Notably, in all three cell lines, the transcriptional activity associated with promoter 2 was stronger than that associated with promoter 1 (Figure 2a), consistent with a previous study.28 Importantly, when HGE was placed at the enhancer position in pGL3-basic vector, that is, 5’ to 3’ downstream of the Luciferase gene, it enhanced the transcriptional activities of both HER2 promoters in all three cell lines (Figure 2a). As H3K4me3 presents at HGE in K562 cells and HER2 C-terminal fragment (CTF) 687 uses exon 21 to initiate its transcription42 (Figure 1a), we engineered HGE upstream of the Luciferase gene. Only a minor transcriptional activity was detected in 293T cells (Figure 2b), suggesting that the transcriptional initiation capacity of HGE is much weaker compared with that of HER2 promoters. Interestingly, HGE, when placed immediately upstream of the Luciferase gene significantly interrupted the transcriptional activities of both promoters (Figure 2b). Meanwhile, the HGE maintained its enhancer function for both HER2 promoters when it was inversely inserted into the enhancer position, that is, 3’ to 5’ downstream of the Luciferase gene (Figure 2c). Moreover, when compared with Intron 1 enhancer, the HGE possesses similar enhancer function for promoter 1 but slightly weaker for promoter 2 (Figure 2c). These data support that the HGE has enhancer function comparable with the intron 1 enhancer.
TFAP2C regulates the enhancer function of HGE
To determine the minimal enhancer element of HGE, we generated a series of deletions including the deletion of exon 20 and intron 20 (T20), the deletion of exon 21 and intron 21 (T21) and the deletion of exon 22 and intron 22 (T22) (Figure 3a, left panel). The transcriptional activities of these constructs were analyzed using the luciferase assay in 293T and SKBR3 cells. The T21 deletion abolished the enhancement of transcriptional activity from promoter 2 in both cell lines, whereas the T20 deletion had no significant effect in either case, and the T22 deletion increased the transcriptional activity of promoter 2 (Figure 3a). These data suggested that exon 21 and intron 21 contain sequences involved in transcriptional activation, and exon 22 and intron 22 contain sequences involved in transcriptional repression. We next searched for the TFAP2C consensus sequence (GCCTGAGGG)43 and identified three closest potential TFAP2C-binding sites (GCCCCAGAG, GCCCTAGGG, GCCCAGGGC) (Figure 3b) located within intron 21. An electrophoretic mobility shift assay using in vitro translated TFAP2C and three oligonucleotide probes corresponding to the three potential TFPA2C-binding sites showed that TFAP2C binds to all the three probes (Figure 3b). The specificity of TFAP2C binding was further confirmed by competition of non-labeled oligos and the supershifts when cultured with a specific TFAP2C antibody (Figure 3b). TFAP2C silencing attenuated luciferase activity of the HGE enhancer in both 293T and SKBR3 cells (Figure 3c) and downregulated HER2 protein levels in SKBR3 cells. (Figure 3d). These data support our hypothesis that TFAP2C has an important role in regulating the enhancer function of the HGE.
We next carried out genomic editing using CRISPR-Cas9 system to determine the role of the TFAP2C-binding sites at the HGE in the regulation of HER2 expression. Six single-guide RNAs (gRNA) were designed to mutate or truncate TFAP2C-binding sites (Figure 3b). SKBR3 stable cell lines with the single or combined gRNA(s) were attempted to be established, however, only one stable cell line with gRNA6 was achieved. It is likely that the cell viability of SKBR3 cells depends on HER2 as knockdown HER2 in SKBR3 cells results in growth arrest and apoptosis.44 HER2 protein in this cell line was dramatically downregulated while cleaved PARP was induced (Figure 3e). The genotyping of SKBR3-gRNA6 showed heterogenous genomic compositions, that is, normal HGE region and extended mutations from the gRNA6-targeting site (Supplementary Figure 2). This data suggest that CRISPR-Cas9 system with these gRNAs introduced additional mutations in HER2 gene and may interfere with HER2 mRNA splicing, resulting in decreased HER2 expression and apoptosis of SKBR3 cells. In fact, a recent study show that genomic editing of HER2 gene using CRISPR-Cas9 system produced short truncated HER2 caused by alternative splicing of HER2 gene and inhibits cell proliferation in both SKBR3 and BT474 cells,45 consistent with an recent observation of off-target mutations introduced by CRISPR-Cas9 system.46
Considering the HER2-dependent cell viability of SKBR3 cells, the six gRNAs were transiently transfected to SKBR3 cells for 48 h and downregulation of HER2, phospho-AKT levels at various extents was found without inducing obvious apoptosis (Figure 3f). These data suggest that the gRNAs targeting intron 21 containing three TFAP2C-binding sites and the junction of exon 21 and intron 21 interfere with HER2 expression. We also examined the mRNA levels of TFAP2C and HER2 from 12 unidentifiable HER2-positive breast cancer samples and found positive correlation (r2=0.6073) between them (Figure 3g).
DNA methylation within the HGE inhibits the enhancer histone modifications and is inversely correlated with HER2 gene expression in breast cancer samples
The cell-type-dependent enrichments of TFAP2C, H3K4me1 and H3K27ac in the HGE region (Figures 1c and d) prompted us to investigate the underlying mechanism. DNA methylation is known to prevent TFAP2C from accessing the target promoter,47 and DNA methylation and certain histone modifications such as H3K4me3 are mutually exclusive.48 Thus, we hypothesized that DNA methylation within the HGE affects TFAP2C binding and enrichment of the enhancer histone modifications. Analysis of both Methyl 450 K bead array data (ENCODE/HAIB) and Reduced Representation Bisulfite Seq data (ENCODE/HudsonAlpha)36 showed that DNA hypomethylation in the HGE region is coincident with enrichments of transcription factors and enhancer histone modifications in K562 cells (Supplementary Figure 1). We performed a bisulfite sequencing assay to determine the DNA methylation status of 28 CpG sites within the HGE. The HGE is extensively DNA-methylated in MCF7, BT474, HCC1954, MDA-MB-231, MCF10A and ZR-75-1 cells. It is less methylated in K562 and hypomethylated in SKBR3 cells (Figure 4a). These data support our hypothesis that DNA methylation status is critical for the enrichments of TFAP2C, H3K4me1 and H3K27ac in the HGE region. The minor enrichments of TFAP2C at HGE in both BT474 and HCC1954 cells (Figures 1b and d) likely reflect the existence of minor cell populations possessing hypomethylated HGE or the cells toward complete establishment of DNA methylation during cell cycle. We next performed in vitro methylation of the pGL3-promoter constructs with and without HGE, using the CpG methyltransferase M.SssI49, 50 and found a strong decrease of enhanced luciferase activity of the methylated pGL3-promoters vector with HGE compared with that without HGE (Supplementary Figure 3).
Furthermore, we performed CRISPR-dCas9-guided specific DNA methylation51 at the HGE region in SKBR3 cells. The six gRNAs (Figure 3b) were inserted into pdCas9-DNMT3A-EGFP vector and again, stable cell lines using SKBR3 cells were failed to establish. However, we were able to collect cells from gRNA 1 and 6 during the attempt and found that the cells undergo apoptosis (Figure 4b, left panel). Importantly, the HGE region became partially methylated in SKBR3 cells that transfected with pooled gRNAs, validating the target-specific DNA methylation by the dCas9-DNMT3A system and indicating that the apoptosis might be caused at least partially by loss of HER2 (Figure 4c). The dCas9-DNMT3A-gRNA 1 and 6 transient transfected into SKBR3 cells for 48 h markedly downregulated HER2 expression, phospho-AKT, but not that of TFAP2C and no apparent apoptosis was observed (Figure 4b, right panel). These data suggest that the DNA methylation at the HGE region is important for repression of HER2 expression in SKBR3 cells.
DNA methylation is in general strongly correlated with HER2 expression (r2=0.5055868) in breast cancer.52 Using the methylation database MethHC,53 we analyzed the correlation of the DNA methylation status of 47 CpG probes and HER2 expression in 839 breast invasive carcinoma samples cataloged in The Cancer Genome Atlas. The general inverse correlation of DNA methylation in HER2 gene body and HER2 mRNA is stronger (r=−0.48) compared with that of the promoters (r=−0.22 and −0.19 for promoters 1 and 2, respectively) (Supplementary Figure 4). The correlation of the DNA methylation status of all 47 CpG probes with HER2 mRNA expression shows strong inverse correlation (r values ranging from −0.291 to −0.408) between the DNA methylation of the four HGE CpG sites and HER2 mRNA expression (Figure 4d). Further analysis in HER2-positive breast cancers revealed that the DNA hypomethylation of all four HGE CpG sites is inversely correlated with HER2 mRNA (Figure 4e). In the samples with hypomethylated HGE region, TFAP2C mRNA is positively associated with HER2 mRNA (Figure 4f). As discussed earlier that ERRα also binds to HGE region and regulates HER2 expression in SKBR3 cells,41 we also found a positive, but weaker association between the mRNA of the coding gene ESRRA and HER2 expression. In contrast, the expression of PPARGC1B does not show any association. These data suggest that the hypomethylation of the HGE region in breast cancers contributes to HER2 expression by gaining the accessibility of transcription factors such as TFAP2C and ERRα.
In sum, this study unveiled a novel regulatory mechanism by a 3’ gene body enhancer contributing to the transcriptional regulation of HER2. Further studies are sought to determine the role of this enhancer in the transcriptional upregulation of HER2 in HER2-low or HER2-negative breast cancers that undergo radiotherapy or endocrine therapy.
We thank Dr Brad Amendt and his lab for helpful discussions, and the ENCODE Consortium and the ENCODE production laboratories for generating the relevant data sets. The results of the bioinformatical analysis are based, in whole or part, upon data generated by The Cancer Genome Atlas Research Network (http://cancergenome.nih.gov/). The K562 cell line was a kind gift from Dr Fenghuang Zhan. We also thank Drs Christine Blaumueller and Marie Gaine for editorial consultation. This work was supported by to HHQ start-up funds from the Department of Anatomy and Cell Biology, the Carver College of Medicine, University of Iowa; Carver Trust Young Investigator Award (01-224 to HHQ) from the Roy J Carver Charitable Trust; a Breast Cancer Research Award (to HHQ) by the Holden Comprehensive Cancer Center at University of Iowa; The NIH grant (P30 CA086862) to the Genomics and Flow Cytometry core facilities at the University of Iowa. NIH grants R01CA183702 (PI: RJW) and by a generous gift from the Kristen Olewine Milke Breast Cancer Research Fund (PI: RJW). NB was supported by NIH MD/PhD fellowship (F30 CA206255); WZ was supported by NIH grants CA200673, and CA203834, the V Scholar award, a Breast Cancer Research Award and an Oberley Award (National Cancer Institute Award P30 CA086862) from Holden Comprehensive Cancer Center at the University of Iowa.
About this article
Supplementary Information accompanies this paper on the Oncogene website (http://www.nature.com/onc)