27‐Hydroxycholesterol represses G9a expression via oestrogen receptor alpha in breast cancer

Abstract 27‐hydroxycholesterol (27‐HC) is a cholesterol metabolite and the first discovered endogenous selective estrogen receptor modulator (SERM) that has been shown to have proliferative and metastatic activity in breast cancer. However, whether 27‐HC metabolite modulates the epigenetic signatures in breast cancer and its progression remains unclear. The current study, reports that 27‐HC represses the expression of euchromatic histone lysine methyltransferase G9a, further reducing di‐methylation at H3K9 in a subset of genes. We also observed reduced occupancy of ERα at the G9a promoter, indicating that 27‐HC negatively regulates the ERα occupancy on the G9a promoter and functions as a transcriptional repressor. Further, ChIP‐sequencing for the H3K9me2 mark has demonstrated that 27‐HC treatment reduces the H3K9me2 mark on subset of genes linked to cancer progression, proliferation, and metastasis. We observed upregulation of these genes following 27‐HC treatment which further confirms the loss of methylation at these genes. Immunohistochemical analysis with breast cancer patient tissues indicated a positive correlation between G9a expression and CYP7B1, a key enzyme of 27‐HC catabolism. Overall, this study reports that 27‐HC represses G9a expression via ERα and reduces the levels of H3K9me2 on a subset of genes, including the genes that aid in breast tumorigenesis and invasion further, increasing its expression in the breast cancer cells.


| INTRODUC TI ON
is an oxysterol and the most abundant cholesterol metabolite in the bloodstream. It is the first identified endogenous selective oestrogen receptor modulator (SERM), implying that its actions are highly tissue-specific. 1 27-HC promotes atherosclerosis, 2 reduces bone density, 3 contributes to brain-related pathogenesis, 4 aids in thyroid cancer 5 and breast cancer proliferation. 6 27-HC promotes breast cancer proliferation via ERα and induces metastasis via LXRα. 6 The metabolite further modulates γδ-T cells, 7 impairs T cell expansion, 6 and promotes the secretion of extracellular vesicles, including exosomes 7 from polymorphonuclear neutrophils, which in turn promotes tumour growth and metastasis. At the clinical level, the role of 27-HC is yet to be elucidated. CYP7B1, the key enzyme responsible for the catabolism of 27-HC, is associated with increased recurrence-free survival, 8 whereas the expression of the 27-HC synthesising enzyme CYP27A1 is linked to unfavourable tumour characteristics. [9][10][11][12] A study has demonstrated that the serum 27-HC level does not correlate with its levels in breast tumours, unlike in normal breast tissue, suggesting a possible association of dysregulation of 27-HC metabolism in breast cancer cells. 12 A recent study has shown that elevated levels of circulating 27-HC are associated with a reduced risk of breast cancer in postmenopausal women. 10 However, it is important to note that it is not the level of circulating 27-HC but the level of intratumoral 27-HC that provides a better picture of its tumorigenicity.
Aberrant epigenomic signatures play a pivotal in breast cancer development, progression and therapeutic resistance. 13 15 In the later stages of cancer progression, additional epigenetic changes, along with subclonal mutations and signals from the microenvironment, dictate the cancer cell phenotype, affect the metastatic propensity of the tumour, and confer endocrine therapy resistance. 13,15 Other epigenetic modifiers, such as noncoding RNAs, especially micro RNAs and long noncoding RNAs, and changes in histone marks, are also known to contribute to breast cancer progression. 14 Attempts are being made to understand the intricate network of functional interactions among oestrogen receptors, oncogenic transcription and epigenetic machinery, with emphasis on how these factors are assembled upon acute oestrogen exposure, which is one of the known risk factors for breast cancer. 13 Although it is known to some extent that oestrogen drives breast cancer proliferation, the molecular mechanisms, hierarchical events upon oestrogen and/or ERα ligand exposure, and effects on gene regulation and chromatin organisation at the genomic and epigenomic levels are not well understood. 16 27-HC is known to induce cellular transition in breast cancer cells. Nevertheless, whether aberrant epigenome contributes to 27-HC-mediated cellular and morphological changes and breast cancer progression is unknown. We had earlier reported that 27-HC induces DNA methylation changes in subsets of genes in breast cancer cells. 17 In this study, 27-HC-mediated changes in histone methyltransferases in breast cancer cells were investigated. 27 The treatments were done in phenol-red free DMEM supplemented with 5% charcoal-stripped serum.

| RNA isolation, RT-PCR and qPCR
The MCF-7 cells were seeded at a cell density of 1 × 10 5 cells in 60 mm dishes. Total RNA was isolated using TriZol reagent (Sigma).
The quality and concentration of RNA was measured with Nanodrop (2000, ThermoFischer). First strand cDNA synthesis was carried out with One Step RT-PCR Kit Ver.2 (Cat. #RR055A/(TAKARA) using 1 μg of total RNA. The qPCR was carried out as per the manufactures instructions using SYBR Premix Ex Taq II (TAKARA) in a 7500 Real-Time PCR System (Applied Biosystems). Gene expression levels were calculated by the 2 −ΔΔCt method using GAPDH as control (Table S1). All the samples were assessed in triplicates and the experiment was done in triplicates.

| Western Blot
The cells were seeded at a cell density of 1×10 6 cells in 100 mm dishes. As the cells attained their morphology, they were treated with 27-HC at 1 μM for different time points, the cells were seeded in parallel. The cells were scraped and lysed using RIPA buffer NaCl (500 mM) 0.175 g, 1% NP 40,0.5% Sodium deoxycholate, 0.1% SDS, Tris buffer (pH 8.0). Total protein concentration was quantified by Bradford's assay. 40-60 μg of protein was loaded onto 7%-10% SDS-PAGE gel and was subjected to electrophoresis. Proteins were further transferred to PVDF membrane by western blot. The membranes were blocked with 5% BSA for an hour and washed with 1X TBST buffer. The blot was incubated with primary antibody overnight at 4°C. This was followed by wash with 1X TBST. The blots were further incubated with suitable secondary antibody for 90 min.

| Histone extraction
The cells were harvested and washed twice with cold 1X phosphate-

| Chromatin immunoprecipitation assay
The cells were seeded at a cell density of 1 × 10 6 cells in 100 mm dishes. As the cells attained their morphology, they were treated with 27-HC at 1 μM for 72 h. After the treatment period, the cells were cross-linked for 10 min at room temperature with 1% formaldehyde and then quenched with 0.125 M of glycine for 5 min.

| ChIP sequencing
ChIP DNA and the input DNA were subjected to end repair and tailing with dA-tail followed by ligation of adapter sequences. These adapter ligated fragments are then size selected using SPRI bead, followed by size selected fragments are indexed during limited cycle PCR to generate final libraries for paired-end sequencing. The resulting libraries are quantified and subjected to sequenced on Illumina HiSeq 2500/4000 system to generate 2 × 50 bp sequence reads.

| Data processing
The fastq files were subjected to quality and the following parameters from fastq file were checked to confirm the quality. The base quality score distribution, sequence quality score distribution, average base content per read, PCR amplification issues and analysis of over-represented sequences. The adapter trimming was performed using Trimmomatic (Ver-0.36). 18 The paired-end reads are aligned to the reference hg19/GRCh37 human genome built. Only uniquely mapped reads were taken further for analysis. To remove any false positive location of binding, the black listed region of the human genome was filtered out and proceed to peak calling using MACS2 (MACS 2.1.3 version). 19 These peaks are annotated using HOMER (annotatepeaks) and GREAT tool. 20 Peak annotation visualisation using ChipSeeker R package. 21 The motif discovery was performed using DREME 22 and MEME software. 23 The MEME discovers novel, ungapped motifs in the sequences. DREME discovers short, ungapped motifs that are relatively enriched in the sequences compared with shuffled sequences.

| Tissue microarray construction
Sixty-eight ER-positive and 19 ER-negative samples were selected to construct the tissue microarrays (TMAs). The microarray was con- and scored them by perceiving >5 high power fields with roughly 1000 cells. The ERα and PR IHC slides were scored based on Allred scoring system with samples having more than 1% nuclear expression considered as positive. The samples negative for ERα and PR by IHC were grouped as hormone receptor negative breast cancers. For other proteins studied such as EHMT2/G9A, CYP7B1 and CYP27A1 scoring was performed in semi-quantitative way by H-score method.

| Statistical analysis
Results were expressed as a mean ± standard deviation of three biological replicates. The p value was determined using paired two-

| 27-HC downregulates G9a expression in MCF-7 cells
Of the selected HKMTs, significant downregulation of G9a was observed in MCF-7 cells upon treatment with 27-HC. G9a is a well-known euchromatic histone lysine methyltransferase that introduces the H3K9me2 mark on the chromatin and is linked to gene repressor activity. 24 The expression of the G9a protein after treatment with 27-HC was assessed, which revealed that G9a levels decreased concurrently at transcript levels at all time points. The highest reduction was observed at 72 h (Figure 2A,B). As G9a is one of the key HKMTs that introduces the H3K9me2 mark on the chromatin, the levels of the H3K9me2 mark on the chromatin isolated from MCF-7 cells treated with 27-HC were analysed. In concordance with the downregulation of G9a, a reduction in the H3K9me2 mark was observed after 27-HC treatment ( Figure 2C,D). G9a is known to regulate the expressions of numerous genes. The expression of one of the well-known G9a targets, CDKN1A/p21 gene, was analysed.
Studies have shown that G9a can either activate 25 or repress p21 expression. 26,27 In this study, 27-HC was found to downregulate G9a levels, which led to reduced expression of p21 at the transcriptional and translational levels ( Figure 2E-G), suggesting that G9a functions as a transcriptional activator of CDKN1A/p21 gene expression in MCF-7 cells. p21 is a two-faced regulator depending on cell type, cellular localisation, p53 status and the type and level of genotoxic stress. 28,29 p21 can acquire oncosuppressive properties when it is in a p53-proficient environment and oncopromoting properties when it is in a p53-deficient environment. 29 As MCF-7 cells are known to contain wild-type p53, the observed reduction in p21 levels upon 27-HC treatment may also contribute to the tumorigenic function in cells.

| Mechanisms of 27-HC-mediated suppression of G9a expression
The mechanisms of 27-HC-mediated suppression of G9a were studied next. Whether G9a reduction was mediated by the ERα receptor was tested. 27-HC was observed to reduce ERα, which is in agreement with a previous report ( Figure 3A,B). 30    The graph represents the intensity of H3K9 dimethylation normalized with the intensity of corresponding Histone H3 levels. (E) p21gene (CDKN1A), a non-histone target of G9a was checked for its expression by qRT-PCR. p21 levels were seen to decrease. (F, G). The graph represents the intensity of p21 normalized with intensity of corresponding β-Actin levels. The western blot analysis of p21 was consistent with its transcriptional expression pattern. The western blot and qRT-PCR were performed in biological triplicate. The intensity of bands in western blot analysis were measured using ImageJ. The p values were calculated using two-way anova and the differences were considered significant at p values p < 0.05(*) <0.01(**) <0.001(***). Interestingly, all these genes were linked to cancer progression by either involving in, invasion and metastasis, or stemness. [31][32][33][34][35][36][37][38] As H3K9me2 is generally a gene repressor mark, the expressions of selected ChIP target genes were tested using qRT-PCR. The expression levels of BRIP1, BCAS1, TBX4, EMB and CYP24A1 were analysed, which showed that the expressions of BRIP1, BCAS1, TBX4 and CYP24A1 were upregulated 2.5, 3.4, 5.8 and 5.9 folds, respectively ( Figure 5B). Hence, it is evident that 27-HC regulates a subset of cancer progression genes via modulation of the euchromatic histone lysine methyltransferase G9a to deregulate the H3K9me2 mark in MCF-7 cells.

| Correlation of the 27-HC catabolising enzyme CYB7B1 and G9a expression in breast cancer tissues
The expressions of key enzymes involved in 27-HC metabolism, namely, CYP27A1 and CYP7B1, 9  to act either as a repressor or as an activator depending upon its interacting partners. 40 G9a has been observed to be mutated or amplified at a low frequency in various tumours and has been described as an oncogene or a tumour suppressor. 40,41 G9a repression has been reported to be associated with self-renewal 42 and highly tumorigenic tumour-propagating or cancer stem cells. 43 Furthermore, unlike their wild-type counterparts, G9a-depleted tumours develop after a prolonged latency. Nevertheless, these tumours are more aggressive and have an expanded cancer progenitor pool and pronounced genomic instability. 41 In this study, reduced levels of G9a and H3K9 dimethylation were observed, which might induce epigenetic reprogramming to stimulate the proliferative and invasive capacity of cancer cells.
ChIP assay demonstrated the reduced occupancy of ERα at the G9a promoter. Although oestrogen drives breast cancer proliferation, the molecular mechanisms of ligand-bound ERα and its effects on gene regulation and chromatin organisation remain poorly understood. There is limited understanding of the hierarchical events that occur at the genomic and epigenomic levels hours after exposure to oestrogen. 16 ERα-mediated transcriptional repression includes the physiological squelching of coactivator proteins and the involvement of components of the basal transcriptional machinery. 44 Moreover, polycomb complexes are recruited to enhance the inhibiting interaction with p300/CBP, thus preventing transcriptional progression. ERα has also been reported to repress genes by targeting DNMT3B. 45 We have previously shown that 27-HC can induce DNA hypermethylation on the promoters of a subset of genes via the de novo DNA methyltransferase DNMT3B in MCF7 cells. 17 ChIP results allude that 27-HC mediated the reduced ERα occupancy at the G9a promoter, which resulted in the transcriptional silencing of G9a.
Furthermore, G9a has been shown to act as a coactivator of ERα. 46 G9a reduction by 27-HC was found to decrease p21 at the transcriptional and translational levels, which agrees with the earlier findings that G9a activates p21 expression. 25 p21 is known to act as a doubleedged sword in cancer cells depending on cell type, cellular localisation, p53 status and the type and level of genotoxic stress. 28,29,47 27-HC reduced ERα, which is consistent with a previous report. 30 Moreover, MG132 pretreatment prevented 27-HC-mediated ERα degradation. Additionally, the ERα antagonists fulvestrant and the These genes were linked to cancer progression. [32][33][34][35][36][37][38]48,49 A moderate correlation was observed between G9a and CYP7B1 in patients belonging to the premenopausal age group. Statistically significant expression changes were not noted in CYP7B1 and CYP27A1 between menopausal status and tumour grade. A slightly higher expression of CYP27A1 was observed in premenopausal women than in postmenopausal women.
Additionally, when CYP27A1, the 27-HC synthesising enzyme was higher, 27-HC was likely to be higher, G9a showed a mild negative correlation (−0.3229), although not statistically significant. This observation agrees with our in vitro study in which 27-HC downregulated G9a expression. Also, in postmenopausal women, the expressions of CYP7B1 and CY27A1 tend to be lower than those in postmenopausal women. The activity of these enzymes and their expression ratios may decide the level of 27-HC. Elevated tumoral mRNA expression of CYP7B1 has also been reported to be associated with increased recurrence-free survival. 8 The Cancer Genome Atlas data analysis showed that CYP27A1 levels are more or less similar in breast tumours and normal breast tissues, whereas the CYP7B1 expression is significantly lower in breast tumours. 50 CYP7B1 has also been shown to be hypermethylated in breast tissues, and the recruitment of monocytes to breast tissues is likely to enhance the accumulation of 27-HC. 51 In this study, increased CYP7B1 expression was associated with increased G9a.
27-HC levels in the serum were not assessed in this study as these levels are not correlated with the levels in breast tumours, unlike normal breast tissues. 12 Assuming that the levels of 27-HC are lower in the breast tissues as CYP7B1 is higher, the higher G9a correlates well with our observed cellular results.
In this study, a statistically significant difference was not obtained between nuclear and cytoplasmic G9a. However, the median H-score of G9a in the cytoplasm tends to be higher than that in the nucleus. Although G9a was previously thought to be localized primarily in the nucleus, accumulating evidence indicates that the G9a protein is localized both in the nucleus and in the cytoplasm. The localisation of endogenous G9a in the cytoplasmic compartment requires the exclusion of exon 10 (E10) from the G9a mRNA. 52 G9a has been reported to shuttle between the nucleus and the cytoplasm). 53 Interestingly, G9a cytoplasmic intensity, but not nuclear intensity, in oral squamous carcinoma cells was associated with histological grade of differentiation (tumour grade). 54 Further studies and analysis using larger patient samples are re- writing -review and editing (equal).

ACK N O WLE D G E M ENTS
We acknowledge the RGCB core facility and the intramural support. VR is supported by ICMR Senior Research Fellowship, Govt of India.

CO N FLI C T O F I NTE R E S T S TATE M E NT
The authors declare that they have no conflict of interest with the contents of this article.

DATA AVA I L A B I L I T Y S TAT E M E N T
All the data generated in this study are included in the manuscript.
The NGS files related to H3K9me2 ChIP sequencing are generated in this study are available for download from NCBI GEO accession number: GSE232895.