|Home | About | Journals | Submit | Contact Us | Français|
While the multiple endocrine neoplasia type 1 (MEN1) gene functions as a tumor suppressor in a variety of cancer types, we explored its oncogenic role in breast tumorigenesis. The MEN1 gene product menin is involved in H3K4 trimethylation and co-activates transcription. We integrated ChIP-seq and RNA-seq data to identify menin target genes. Our analysis revealed that menin-dependent target gene promoters display looping to distal enhancers that are bound by menin, FOXA1 and GATA3. In this fashion, MEN1 co-regulates a proliferative breast cancer-specific gene expression program in ER+ cells. In primary mammary cells MEN1 exerts an anti-proliferative function by regulating a distinct expression signature. Our findings clarify the cell type-specific functions of MEN1 and inform the development of menin-directed treatments for breast cancer.
Multiple endocrine neoplasia type 1 (MEN1) is caused by inactivating germline mutations of the MEN1 gene and is predominantly characterized by parathyroid adenomas, pituitary adenomas and pancreatic and duodenal neuroendocrine tumors (Chandrasekharappa et al., 1997; Thakker et al., 2012). Previous studies have suggested however that the MEN1 gene has a dual role in breast tumorigenesis. Female MEN1 patients are at increased risk for developing breast cancer suggesting a tumor suppressive role. Consistent with this, breast tumors in MEN1 patients show complete loss of the MEN1 gene (Dreijerink et al., 2014). Moreover, genetic loss of function MEN1 mouse models show increased incidence of both in situ and invasive mammary cancer (Seigne et al., 2013). In contrast, in sporadic breast cancers the MEN1 gene appears to exert a proliferative function. MEN1 gene mutations are uncommon and expression of the MEN1 gene product menin has been reported to be involved in resistance to endocrine therapy (Imachi et al., 2010; TCGA, 2012). Menin is able to interact with and co-activate the estrogen receptor alpha (ERα), a critical driver in approximately 70% of sporadic breast cancer cases (Dreijerink et al., 2006; Imachi et al., 2010). A similar proliferative function of menin has recently been shown in sporadic androgen receptor (AR) expressing prostate cancer (Malik et al., 2015).
Menin is a ubiquitously expressed nuclear protein that has no intrinsic enzymatic activity. Over the years, many menin-interacting proteins have been reported. Most of the interacting proteins indicate a role for menin in transcriptional regulation, either as a co-activator or a co-repressor (Matkar et al., 2013). Menin was found to be an integral part of mixed-lineage leukemia MLL1/MLL2 (lysine methyltransferase [KMT2A/B]) containing protein complexes that have methyltransferase activity directed at trimethylation of lysine 4 of Histone H3 (H3K4me3) (Huang et al., 2012; Hughes et al., 2004; Yokoyama et al., 2004). Aberrant H3K4me3 is considered to contribute to MEN1 tumorigenesis as simultaneous knock out of the H3K4me3 demethylase Rbp2/Kdm5a resulted in longer survival in a MEN1 mouse model in which mice develop insulinomas (Lin et al., 2011). H3K4me3 is an epigenetic mark of active transcription and is localized primarily to transcription start sites (TSS) (Santos-Rosa et al., 2002). Menin has also been found to be predominantly present at TSS (Agarwal et al., 2007; Cheng et al., 2014; Scacheri et al., 2006). Reports addressing the genome-wide function of menin have yielded cell-specific results in terms of regulation of H3K4me3 and target gene expression (Agarwal and Jothi, 2012; Li et al., 2013; Lin et al., 2011; Lin et al., 2015).
A similar dual role in oncogenesis has been reported for other epigenetic regulators, such as the enhancer of zeste homolog protein 2 (EZH2 [KMT6]). EZH2 is the catalytic subunit of the polycomb repressive complex 2 that methylates H3K27 (Cao et al., 2002): Overexpression of EZH2 has been observed in breast and prostate cancer (Xu et al., 2012). Gain-of-function mutations are present in lymphomas. In contrast, loss-of-function mutations are found in myelodysplastic syndrome and leukemia (reviewed in (Lund et al., 2014)). These dualities likely reflect differential epigenetic regulation of predefined cell type-specific transcriptional programs.
In this study, by integrating chromatin immunoprecipitation combined with next-generation sequencing (ChIP-seq) and RNA sequencing (RNA-seq) we aimed to investigate the genome-wide function of menin in breast cancer. In addition, we combined our data with publicly available ChIP-seq and chromatin interaction data sets. We show that menin-H3K4me3 target gene preference is associated with the presence of menin at enhancer sites that are found to be involved in looping with their target gene TSS. In this fashion, menin controls a highly luminal breast cancer-specific proliferative gene expression program in breast cancer cells. In contrast, in primary luminal progenitor (LP) cells, menin regulates a different gene signature that is in line with its role as a tumor suppressor. Our results clarify the proliferative role of the MEN1 gene in sporadic ER+ breast cancer and provide a potential explanation for the cell type-specific actions of menin.
We chose the MCF-7 breast cancer cell line to study the role of menin in transcriptional regulation. Menin has been shown to co-activate ERα in a ligand-dependent fashion at the canonical ERα target gene TFF1 in these cells (Dreijerink et al., 2006). MCF-7 cell lines were established stably expressing doxycycline(dox)-inducible small hairpin shRNA targeting MEN1 or a scrambled shRNA construct (Figure 1A). After synchronization in phenol red-free medium containing 10% charcoal dextran-treated fetal bovine serum (CDT medium), cells were treated with either estradiol (E2) or epidermal growth factor (EGF) for 96 hours. Both in the E2 and EGF-stimulated cells, growth was severely reduced after MEN1 gene silencing (Figure 1B). To assess whether this effect could be extrapolated to other cell lines we studied the ERα positive (ER+) breast cancer cell line MDA-MB-361, and the ESR1 expressing αT3 and GH3 pituitary cell lines. In these cell lines E2-stimulated growth was likewise attenuated after silencing of MEN1 (Figure 1C).
Fluorescence activated cell sorting (FACS) showed that in the absence of E2, equal numbers of MCF-7 cells are in the G1 phase of the cell cycle both in control cells and MEN1 silenced cells. After four days of treatment with E2, fewer control cells were in G0/G1 compared with shMEN1 cells (Figure 1D). These observations support an essential role for the MEN1 gene in the ability of E2 to promote cell cycle progression in ER+ breast cancer cells.
In addition, we assessed the proliferative role of the MEN1 gene in ER+ breast cancer cells using data from recently performed Genome-scale CRISPR knock out (GECKO) screens in MCF-7 cells and also the ER+ T47D breast cancer cell line. In both cell lines the MEN1 gene was found to be a high-ranking proliferative gene (manuscript submitted).
In order to define the presence of the menin-H3K4me3 methyltransferase complex and its relevance for gene expression in ER+ breast cancer cells, we performed ChIP-seq for menin, MLL1, MLL2 and H3K4me3 and RNA-seq in cells cultured under hormone-free conditions.
Consistent with the published literature, we found menin, MLL1 and MLL2 to be present predominantly at TSS (TSS: defined as the transcription start site +/- 5kb) (menin 83% of peaks; MLL1 82%; MLL2 92%; Figure 2A). The distribution of MLL1, MLL2 and H3K4me3 peaks clustered with the menin-bound TSS (Figure 2B). Transcripts that have menin present at their TSS were found to be expressed at significantly higher levels compared with transcripts from TSS that lacked menin binding (Figure 2C). We expanded the menin ChIP-seq analysis to the T47D ER+ and MCF-10A ER- cell lines. We found that menin-promoter binding was comparable in MCF-7 and T47D cells. However, menin was not found to be present at these ER+ specific TSS in the MCF-10A cell line (Figure 2D). Thus, in ER+ breast cancer cells, the menin cistrome appears to be localized primarily at a set of transcriptionally active TSS that are not bound by menin in ER- breast cancer cells, suggesting distinct roles for menin in ER+ and ER- breast cancer.
Global H3K4me3 was assessed by immunoblotting of histone extracts and found not to be affected by MEN1 silencing, as previously reported (Figure 3A)(Dreijerink et al., 2006). We performed ChIP-seq of H3K4me3 in shCtrl cells versus the shMEN1 cells that showed the most thorough silencing in Figure 1A. There were no major differences in the numbers (20,083 in shCtrl vs. 20,066 in shMEN1) or locations of H3K4me3 peaks (Figure 3B). Comparison of peak heights after MEN1 silencing showed a modest decrease of the average H3K4me3 ChIP-seq signal at menin-bound TSS (Figure 3C). Reduced MEN1 expression did not affect global mRNA levels (Figure 3D). Assessment of H3K4me3 peaks at individual menin-bound sites identified a limited number of TSS that showed differential H3K4me3 (Figure 3E). Although global gene expression was not affected after silencing of MEN1, a number of genes were expressed at lower levels following MEN1 silencing (Figure 3F, Figure S1A). Thus, although menin is present at the TSS of many active genes, it regulates the level of H3K4me3 and subsequent transcription at a limited number of genes. Menin inhibitors that block the interaction of menin and MLL1 are being developed for the treatment of MLL-driven acute myeloid leukemia (Grembecka et al., 2012). In order to verify the changes in gene expression regulated by MEN1, we compared gene expression by RNA-seq following MEN1 silencing with gene expression in cells treated with the menin inhibitor MI-2. We found a set of 57 genes down regulated under both conditions (Figure 3G,H). Gene ontology (GO) analysis revealed that this gene list was most significantly enriched for oncogenic genes involved in estrogen and steroid hormone response such as the ERα gene (ESR1), carbonic anhydrase (CA2), neuropeptide y receptor Y1 (NPY1R) and the sonic hedgehog homolog (SHH) genes (Figure S1B).
To identify the direct menin-H3K4me3 target genes responsible for the MEN1-dependent growth differences observed in breast cancer cells, the results of RNA-seq following MEN1 silencing were integrated with the menin and H3K4me3 ChIP-seq data (Figure 4A). The 6403 TSS with menin ChIP-seq peaks were filtered for those showing a significant reduction in H3K4me3 (>0.5 log2 fold down in figure 3E) following MEN1 silencing resulting in 104 potential target genes. When these genes were further refined for those also showing a significant decrease in mRNA levels (>0.5 log2 fold down, p<0.05) four differentially expressed menin-H3K4me3 target genes were identified: ESR1, AGR3, AGR2 and VAV3 (Figure 4A). These genes have been implicated in breast tumorigenesis and are typically expressed in ER+ breast cancer (Figure 4B) (Curtis et al., 2012).
A detailed examination of the ESR1 gene locus revealed that menin is present at multiple sites at the ESR1 gene locus, along with MLL1 and MLL2 (Figure 4C). Following MEN1 silencing, H3K4me3 is reduced, especially at an upstream site (labeled 1, Figure 4C/D) and at the TSS of the transcription variant 4 (labeled 2, Figure 4C/D). By immunoblotting we could demonstrate that at the protein level total ERα was reduced after MEN1 silencing (Figure 4E). ERα was also reduced by MEN1 silencing in the other cell lines that exhibited similar E2-dependent growth differences (Figure S2). Inducible re-expression of an shMEN1#1 resistant MEN1 cDNA interfered with down regulation of ERα and could also rescue the attenuated growth of the shMEN1#1 cells, supporting the specificity of the observed effects (Figure 4E). Several ESR1 mRNA transcripts were also expressed at reduced levels after MI-2 treatment (Figure 3H). MI-2 treatment caused down regulation of ERα protein levels (Figure 4F). The results substantiate the finding that menin directly regulates ESR1 gene expression. Mammary cells from MEN1 patients lack one functional MEN1 allele from the initiation of cell differentiation. Therefore, in order to mimic the loss of MEN1 in MEN1 patients, we studied the effect of MEN1 silencing in primary luminal progenitor (LP) cells derived from healthy individuals undergoing reduction mammoplasty. LP cells represent a differentiation stage between the mammary stem cell and differentiated mature luminal cells and express ERα only at very low levels. Differential gene expression was studied in FACS-sorted CD49f+, EpCAM+ mammary LP cells (Lim et al., 2009). After harvesting and cell sorting, cells were treated with lentiviral particles containing sh constructs directed against MEN1 or a control sequence and subjected to puromycin selection. Treatment of the cells with shMEN1 viruses resulted in lower menin protein levels (Figure 4G). Total RNA from these cells was subjected to gene expression analysis and clusters of up and down regulated genes were identified. GO and gene set enrichment analysis were performed on the differentially expressed genes. The most significantly enriched expression profiles included genes that were up regulated after MEN1 silencing, in particular genes that are involved in extracellular matrix formation such as the matrix metalloproteinase genes 9 (MMP9) and also MMP3 (Figure S1C). Accordingly, in addition to the extracellular matrix GO term, gene set enrichment analysis also gave a significant match with an invasive breast cancer signature (Figure 4H) (Schuetz et al., 2006). There was no significant overlap of the differentially regulated genes after MEN1 silencing in the MCF-7 breast cancer cells compared with the LP cells (Figure 4I). The expression of the ESR1 gene, which is transcribed at low levels in LP cells, was not affected by MEN1 silencing. However, in the ZR-75-1 ER+ cell line that lacks menin protein expression, reintroduction of a MEN1 cDNA construct resulted in increased ESR1 mRNA levels, further supporting the relevance of this mechanism in ER+ cells (Figure 4J)(Barretina et al., 2012). These results indicate that MEN1 suppresses the expression of genes associated with invasion in normal LP cells consistent with its tumor suppressive properties while it stimulates the expression of growth promoting genes in ER+ breast cancer cells consistent with an oncogenic activity.
Using candidate gene approaches, we have previously shown that menin can be recruited to nuclear receptor binding sites after ligand treatment (Dreijerink et al., 2006; Dreijerink et al., 2009). We used E2 stimulation of MCF-7 cells to study genome-wide menin dynamics. Incubation of shCtrl cells with 10nM E2 for 45 minutes resulted in a shift of the menin cistrome (Figure 5A). Motif analysis at 1464 menin ChIP peaks gained after E2 treatment showed enrichment for the ERα binding motif (Figure 5B). ERα typically binds to enhancer regions and although the most common sites of menin binding after E2 stimulation were at TSS, overlapping the gained menin peaks with known ERα binding sites indeed showed recruitment to non-TSS sites (only 14% of menin and ERα bound sites after E2 treatment are at TSS)(Figure 5C,D)(Carroll et al., 2006). Recruitment of menin to the regulatory sites of the canonical ERα target gene TFF1 has been shown using ChIP-PCR previously (Dreijerink et al., 2006). In the ChIP-seq experiment shown, stimulation of the cells with E2 demonstrated recruitment of menin to the TFF1 TSS and a TFF1 upstream enhancer site (Figure 5E)(Kong et al., 2011).
These changes in the menin cistrome following E2 treatment suggest that recruitment to enhancer sites plays a role in menin’s co-activator function.
We revisited the menin ChIP-seq results at the MEN1-regulated genes ESR1, AGR3, AGR2 and VAV3 in order to further identify determinants of differentially expressed menin-H3K4me3 target genes. At these loci, we found that in addition to the TSS, menin was also present at non-TSS sites, as it was upstream of the ESR1 TSS (Figure 4C). In the menin ChIP-seq experiment, 1428 peaks were called outside TSS +/- 5kb regions. The average amplitude of these non-TSS or enhancer menin peaks was similar to menin peak heights at TSS (Figure 6A). Clustered motif analysis of the 1428 menin enhancer peaks showed enrichment of forkhead and GATA transcription factor binding sites (Figure 6B). In order to assess true enhancer TSS interactions we intersected the menin ChIP-seq data with available looping data from ChIA-PET experiments at CTCF and RNA polII-bound loci in MCF-7 cells (Encode Consortium, 2012; Li et al., 2012).
Three of the top candidate menin-H3K4me3 target genes showed presence of menin at sites that could also be bound by FOXA1 and GATA3 and displayed looping by ChIA-PET (Figure 6C) (Hua et al., 2009; Schmidt et al., 2010). By overlapping menin peaks outside TSS regions with publicly available FOXA1 and GATA3 ChIP-seq data and ChIA-PET looping data, 180 potential enhancers could be identified that are linked to 146 unique TSS (Figure 6D). GO analysis revealed that among the 146 transcripts from these TSS there is a significant enrichment for genes encoding transcriptional regulatory proteins (Figure S1D). The loops at the ESR1 and AGR gene loci identified by ChIA-PET could be confirmed by 4C-seq chromatin conformation capture (Figure 6E, Figure S3). The looped enhancers that are bound by menin, FOXA1 and GATA3 are also bound by MLL1 and MLL2 as indicated by an enrichment of MLL1 and MLL2 ChIP-seq signals at these sites compared with all menin peaks outside TSS (Figure 6F). The 180 menin/FOXA1/GATA3 sites that are associated with gene looping show clear enrichment of the canonical enhancer histone mark H3K27Ac, supporting that these are bona fide enhancer sites (Theodorou et al., 2013)(Figure 6G).
We assessed the effect of MEN1 silencing on H3K4me3 and gene expression of the 146 enhancer-linked TSS. Average H3K4me3 at the 146 TSS was clearly reduced in the shMEN1 expressing cells (Figure 7A). By integrating the 146 menin/FOXA1/GATA3 looping TSS with the MEN1-dependent H3K4me3 ChIP-seq data, we could identify six transcripts from this group that showed >0.50 log2 fold MEN1-dependent H3K4me3: in addition to the ESR1, AGR2 and AGR3 transcripts, we also identified the GREB1 and NR2F2 TSS as direct menin-H3K4me3 targets. Out of the nine TSS showing the largest overall H3K4me3 decrease (>0.75 log2 fold) after MEN1 silencing four TSS displayed looping to menin/FOXA1/GATA3 sites (Figure 7B). This represents a significant enrichment of menin/FOXA1/GATA3 enhancer-linked TSS versus all other menin-bound TSS (p<0.0001 using a Fisher’s exact test). To verify whether looping at the ESR1 and AGR loci was affected by MEN1 silencing, we compared the 4C-seq signals after induction of shCtrl or shMEN1 at the ESR1 and AGR loci and found that looping at these sites is not MEN1-dependent (Figure 6E).
Although MEN1 silencing did lead to lower average H3K4me3 at the 146 TSS, average mRNA levels were not affected (Figure 7C, left panel). However, the six TSS that had the most substantial MEN1-dependent H3K4me3 did show significantly reduced mRNA levels after MEN1 depletion (Figure 7C, right panel). To further investigate the mechanism of menin-H3K4me3 target gene regulation and the role of FOXA1, we interrogated gene expression data after siRNA-mediated knock down of FOXA1 in MCF-7 cells and compared the results with our RNA-seq data (Hurtado et al., 2011). We found a common subset of genes showing reduced expression both after MEN1 and FOXA1 silencing, which includes the ESR1, AGR2 and AGR3 genes (Figure S4). In addition, we performed FOXA1 ChIP-seq in shCtrl and shMEN1 expressing MCF-7 cells. FOXA1 mRNA levels were not affected by MEN1-silencing. We found that FOXA1 presence at the menin-bound enhancer sites was severely impaired after menin depletion, suggesting that FOXA1 DNA-binding is connected to menin’s function in MCF-7 cells (Figure 7D).
Thus, the combination of the presence of menin, FOXA1 and GATA3 at non-TSS sites that display looping marks genes that are specifically co-activated by menin as part of a H3K4 methyltransferase complex (Figure 7E).
The MEN1 gene has a dual role in breast cancer. In the normal mammary epithelium it exerts an anti-proliferative and tumor suppressive role consistent with the increased risk of breast cancer in women with the MEN1 syndrome. In contrast, in sporadic ER+ breast cancer MEN1 is oncogenic and promotes proliferation. This study provides important insights into this duality. The proliferative role of MEN1 in sporadic breast cancer is consistent with the function of menin in ER+ breast cancer cells both as a regulator of ESR1 gene expression and as a co-activator of ERα. MEN1 silencing also inhibited EGF-dependent growth, which has previously been demonstrated to rely on the presence of ERα (Lupien et al., 2010). In addition, the AGR2, AGR3 and VAV3 genes that were identified as menin-H3K4me3 targets are also involved in breast tumorigenesis. The anterior grade homology 2 protein (AGR2) gene was reported in a regulatory pathway with ERα and FOXA1 and has been associated with metastatic breast cancer (Liu et al., 2005; Wright et al., 2014). AGR3 may serve as a biomarker in luminal breast cancer (Garczyk et al., 2015). The guanidine nucleotide exchange factor gene VAV3 was recently found to be involved in resistance to endocrine therapy in breast cancer (Aguilar et al., 2014). In addition to ESR1, AGR2 and AGR3, the GREB1 and NR2F2 genes were identified as menin/FOXA1/GATA3 enhancer linked target genes. Interestingly, the GREB1 gene has recently been reported as an ERα interacting transcriptional co-activator (Mohammed et al., 2013). The NR2F2 gene encodes the COUP-TFII nuclear receptor that has been shown to have a role in breast cancer progression as well (Zhang et al., 2014).
In contrast to the findings in sporadic breast cancer, female MEN1 patients have an increased risk to develop mostly luminal type breast cancer. Most breast tumors in MEN1 patients express ERα or the progesterone receptor (Dreijerink et al., 2014). We analyzed gene expression after MEN1 silencing in primary mammary LP cells from non-MEN1 patients. In the LP cells, a distinct gene expression program was regulated by MEN1. The differentially expressed genes were mostly up regulated and mainly included genes involved in maintenance of the extracellular matrix. Both MMP9 and MMP3 have been reported to be important for breast tumorigenesis (Kessenbrock et al., 2013; Qin et al., 2008). In addition, the metastasis promoting periostin (POSTN) and the platelet-derived growth factor receptor (PDGFRA) genes that have been implicated in endocrine resistance were among the up regulated genes (Malanchi et al., 2012; van Agthoven et al., 2010). Both the MMP3 and POSTN genes are involved in Wnt/B-catenin signaling (Kessenbrock et al., 2013; Malanchi et al., 2012). Interestingly, menin has been reported to be able to interact with B-catenin and suppress Wnt signaling (Cao et al., 2009). The up-regulated genes identified could be direct menin target genes since menin, besides its H3K4me3 function, is also known to play a role in transcriptional repression (Kim et al., 2003). In LP cells, ESR1 is expressed at very low levels, suggesting that different enhancer signatures and transcriptional programs are important in these cells (Lim et al., 2009). MEN1 may thus exert its breast tumor suppressive function in LP cells independent of ERα.
We found that menin is present at TSS of a large proportion of active genes in MCF-7 cells and T47D cells. However, the functional relevance of its presence appears to be restricted to a small number of target genes that are linked to distant regulatory enhancer sites. These results reconcile the findings in published reports that have addressed the function of menin. In the ER-expressing cell lines MCF-7 and T47D we found that menin is present mostly at TSS, as has been reported in other cell types (Agarwal et al., 2007; Cheng et al., 2014; Scacheri et al., 2006). In accordance with previous findings, MEN1 silencing leads to down regulation of H3K4me3 and mRNA of a selected number of genes in a tissue-specific manner (Agarwal and Jothi, 2012; Li et al., 2013; Lin et al., 2011; Lin et al., 2015; Scacheri et al., 2006). Using the well-established MCF-7 breast cancer cell line enabled us to combine our results with published data sets and to take a comprehensive look at the genes regulated by MEN1. This analysis shows that a small number of menin-bound sites at specific enhancer regions mark MEN1-dependent target gene H3K4me3 and expression.
A group of regulatory DNA regions with high enhancer density termed super enhancers appear to include especially critical regulatory hot spots (Loven et al., 2013). On average, cells harbor 300-500 of such super enhancer regions. We used the combination of binding of menin, FOXA1 and GATA3 and looping to identify 180 enhancer sites. This number indicates that the use of these criteria is a valid approach to select active enhancers. The menin/FOXA1/GATA3 enhancers carry the H3K27Ac mark. Indeed, about 50% of the menin/FOXA1/GATA3-bound enhancers that displayed looping, including the sites near the ESR1, AGR2 and AGR3 genes, are within predicted super enhancer regions, indicating functional overlap of these menin-bound sites with super enhancers (data not shown) (Hnisz et al., 2013). Recently, a core transcriptional regulatory circuitry has been proposed in MCF-7 cells, consisting of eight transcriptional activating genes (Saint-Andre et al., 2016). Five of the eight reported transcriptional regulators: ESR1, GATA3, NR2F2, TBX2 and ZNF217, are among the 146 genes potentially connected to menin/FOXA1/GATA3 enhancers. Although we found the highest correlation of menin-dependent H3K4me3 and expression at genes looped to enhancers bound by both the FOXA1 and GATA3 DNA binding transcription factors, it is very likely that there are additional combinations of factors that could confer a comparable level of regulation. For example, the menin-bound putative enhancer looped to the VAV3 gene is bound by FOXA1 but not GATA3. Conversely, the ELF1 locus that showed menin-dependent H3K4me3 is connected to a looped menin-bound enhancer that is bound by GATA3 but not FOXA1.
Menin does not bind to DNA directly and is recruited to the DNA through interactions with numerous DNA-binding transcription factors (Matkar et al., 2013). For technical reasons, we used relatively mild lysis conditions in the menin, MLL1 and MLL2 ChIP-seq experiments. Therefore we cannot exclude the possibility that the menin signals detected at enhancers are in fact derived in trans from TSS-bound menin that is looped to distal sites. However, a model in which menin plays an important role at enhancer-looped TSS fits with previous reports that menin is able to interact with DNA binding transcription factors that are mostly present at enhancers but also the MLL1/MLL2 histone methyltransferase complexes that are predominantly located at TSS.
We demonstrate that menin has an important role at FOXA1 and GATA3-bound enhancers, as supported by the observation that MEN1 silencing leads to reduced FOXA1 binding at menin-bound enhancers. Re-evaluation of FOXA1-dependent gene expression identified ESR1 as a FOXA1-dependent gene. ESR1 gene expression has previously been reported to be down regulated after GATA3 silencing (Eeckhoute et al., 2007). In immunoprecipitation studies in Th2 cells, menin has been shown to co-precipitate with GATA3, suggesting that the presence of menin on GATA3 binding sites is mediated through a direct protein interaction (Nakata et al., 2010). Menin is also able to bind to the forkhead transcription factors CHES1/FOXN3 and FOXO1 (Busygina and Bale, 2006; Wuescher et al., 2011).
The presence of menin at FOXA1/GATA3 bound enhancers that by analysis of ChIA-PET data are looped to nearby promoters is not E2-dependent (data not shown). Overlap of these menin/FOXA1/GATA3 bound enhancer sites with ERα ChIP-seq data did however indicate the presence of ERα at 89% of these sites (data not shown) (Kong et al., 2011). Menin-bound enhancer sites containing FOXA1 and GATA3 were also enriched for MLL1 and MLL2. Regulation of enhancer-mediated transcription by MLL1 or MLL2 has been reported (Kaikkonen et al., 2013; Won Jeong et al., 2012). Since MLL1 has been shown to be a pioneer factor for FOXA1 and ERα DNA binding in MCF-7 cells, the ERα binding may occur following recruitment of the menin methyltransferase complexes (Won Jeong et al., 2012).
The insight gained from these experiments into the proliferative role of MEN1 may be useful for sporadic breast cancer from a therapeutic point of view. Activating ESR1 gene mutations have regained attention as these were found to be present in a significant proportion of ERα expressing metastatic breast tumors (Jeselsohn et al., 2014; Robinson et al., 2013; Toy et al., 2013). The MI-2 menin inhibitor tested is an early generation compound designed to disrupt the interaction between menin and MLL1 (Grembecka et al., 2012). Toxicity of this compound did not allow its use in cell growth assays. However, more specific and potent menin-MLL inhibitors have been developed (Cierpicki and Grembecka, 2014). Such compounds have been shown to suppress growth in AR expressing prostate cancer models (Malik et al., 2015). Next-generation menin inhibitors or related compounds could be used to down regulate the expression of ESR1 and may hold promise for the treatment of advanced ER+ breast cancer.
Detailed experimental procedures can be found in the supplemental information.
MCF-7, MDA-MB-361, T47D, ZR-75-1 (human ESR1 gene expressing breast cancer cell lines), αT3 (mouse Esr1 expressing pituitary), GH3 (rat Esr1 expressing pituitary) were grown in phenol red-free medium containing 10% charcoal dextran-treated fetal bovine serum (CDT medium) containing 10nM E2 or 100 ng/mL EGF. Cells were counted at the indicated time points. Experiments were repeated multiple times in triplicate.
For cell cycle analysis cells were resuspended in PBS containing 100 μg/mL RNase, 40 μg/mL propidium iodide (Sigma). The PI profile was measured using an LSR Fortessa machine (BD Biosciences, San Jose, CA) at the DFCI Flow Cytometry Facility. For data analysis FlowJo version 7.6.5 was used.
For chromatin immunoprecipitation (ChIP) with H3K4me3 and FOXA1 antibodies a standard SDS-based protocol was used, as has been described (Carroll et al., 2005).
For menin, MLL1 and MLL2 ChIP experiments a sarkosyl-based protocol was used, essentially as described (Lee et al., 2006). For sequencing we used the NextSeq 500 sequencing platform (Illumina, San Diego, CA).
Starting with total RNA, poly-A selected, non-stranded RNA-Seq libraries were constructed on the Sciclone liquid handler (Perkin Elmer). Forty bp single end reads were obtained using Illumina HiSeq machine (Illumina, San Diego, CA). For microarray analysis, total LP RNA was applied to Human Gene 1.0 ST Arrays (Affymetrix, Santa Clara, CA).
Two independent 4C experiments were performed in both shMEN1 and shCtrl expressing MCF-7 cells, as previously described (Splinter et al., 2012).
We thank the members of the Brown lab and the Center for Functional Cancer Epigenetics for useful discussion and feedback, especially Gilles Buchwalter, Laura Cato, Tom Westerling, Shengen Shawn Hu, Xintao Qiu and Fugen Li and also Marc Timmers at the UMC Utrecht. We thank John Daley III from the DFCI Flow Cytometry Facility for technical advice. KMD is supported by the Dutch Cancer Society (UU 2012-5370). ACG is funded by an Advanced Postdoc. Mobility fellowship from the Swiss National Science Foundation (P300P3_151145). EL is supported through Love your sister and an NBCF practitioner fellowship.
AUTHOR CONTRIBUTIONSKMD, ACG, EV, EL, WL, PR, MB conceived and designed the experiments. KMD, ACG, ESV, AF, DC, PR performed the experiments. KMD, ACG, ESV, LG, JR, CYL, WL analyzed the data. LG, EL, CYL, PR, HL contributed reagents, materials and analysis tools. KMD, MB wrote the paper.
The accession numbers for the data generated in this paper are in GEO: GSE85099, GSE85315, GSE86316, GSE85317, GSE94001 and GSE94009.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.