EHMT2 Inhibition Induces Cell Death in Human Non-Small Cell Lung Cancer by Altering the Cholesterol Biosynthesis Pathway

Non-small cell lung cancer (NSCLC) is a major subtype of lung cancer. Besides genetic and environmental factors, epigenetic alterations contribute to the tumorigenesis of NSCLC. Epigenetic changes are considered key drivers of cancer initiation and progression, and altered expression and activity of epigenetic modifiers reshape the epigenetic landscape in cancer cells. Euchromatic histone-lysine N-methyltransferase 2 (EHMT2) is a histone methyltransferase and catalyzes mono- and di-methylation at histone H3 lysine 9 (H3K9me1 and H3K9me2, respectively), leading to gene silencing. EHMT2 overexpression has been reported in various types of cancer, including ovarian cancer and neuroblastoma, in relation to cell proliferation and metastasis. However, its role in NSCLC is not fully understood. In this study, we showed that EHMT2 gene expression was higher in NSCLC than normal lung tissue based on publicly available data. Inhibition of EHMT2 by BIX01294 (BIX) reduced cell viability of NSCLC cell lines via induction of autophagy. Through RNA sequencing analysis, we found that EHMT2 inhibition significantly affected the cholesterol biosynthesis pathway. BIX treatment directly induced the expression of SREBF2, which is a master regulator of cholesterol biosynthesis, by lowering H3K9me1 and H3K9me2 at the promoter. Treatment of a cholesterol biosynthesis inhibitor, 25-hydroxycholesterol (25-HC), partially recovered BIX-induced cell death by attenuating autophagy. Our data demonstrated that EHMT2 inhibition effectively induced cell death in NSCLC cells through altering cholesterol metabolism-dependent autophagy.


Introduction
Lung cancer is globally the leading cause of cancer death according to World Health Organization (WHO) [1]. Non-small cell lung cancer (NSCLC), a major subtype of lung cancer, accounts for about 85% of total diagnosed lung cancers [2]. Diverse risk factors, including smoking, exposure to asbestos and radon, and genetic mutations have been well identified. Recently, epigenetic alteration has been highlighted as a driving force of tumorigenesis [3,4]. Epigenetic changes are considered a hallmark of cancer and a key driver of cancer initiation and progression [5,6]. Aberrant expression and activity of epigenetic modifying enzymes change the epigenetic landscape in the cancer cell.
Euchromatic histone-lysine N-methyltransferase 2 (EHMT2), also known as G9A, is a nuclear histone methyltransferase which mainly catalyzes mono-and di-methylation at histone H3 lysine 9 (H3K9me1 and H3K9me2, respectively). Increased H3K9me1 and H3K9me2 are generally associated with gene silencing in euchromatin [7]. Overexpression of EHMT2 has been reported in various cancers including ovarian cancer and lung cancer [8,9]. In lung cancer, overexpressed EHMT2 was reported to contribute to rapid proliferation and invasion [10,11], suggesting that it can be a potential therapeutic target. Nevertheless, the underlying mechanism behind EHMT2 overexpression is not fully understood.
One possible mechanism depends on a link between EHMT2 and cell metabolism. A previous study has reported that EHMT2 promoted proliferation of neuroblastoma by modulating cellular amino acid metabolism [12], suggesting a role of EHMT2 in cancer metabolism. The importance of metabolism in cancer cells has been recently revisited and emphasized [13]. Highly proliferative cancer cells tend to have limited levels of nutrient and oxygen availability [14,15], promoting nutrient uptake to support energetic and biosynthetic pathways and facilitating absorption of macromolecules [16]. Among macromolecule metabolisms, alterations of lipid metabolism have been reported in cancers of the breast, prostate, lung, and colon [17] and contribute to growth, energy and redox homeostasis, and metastases (reviewed in [18]). In particular, cholesterol has important roles in the composition of cellular membranes, hormone synthesis, and signal transduction pathways [19]. Upregulation of its biosynthesis has been reported in cancer [20] and gene expression involved in the biosynthesis is correlated with patient survival in specific types of cancer [21]. Fatty-acid and cholesterol biosynthesis is regulated by master regulators such as sterol regulatory element-binding proteins (SREBPs) [22]. There are three isoforms: SREBP1a, SREBP1c, and SREBP2. SREBP1a mainly controls fatty acid, phospholipid, and triacylglycerol synthesis, while SREBP2 does cholesterol synthesis [22]. SREBP2, encoded by the SREBF2 gene, has been demonstrated to support cell survival in prostate cancer through accumulation of cholesterol, and its inhibition is suggested as a potential cancer therapy [23].
In this study, we investigated a novel link between epigenetic alteration and cancer metabolism, targeting NSCLC. We found that inhibition of EHMT2 activity induced cell death through autophagy and the cell death was mediated by activating cholesterol biosynthesis pathway. Our data suggest that epigenetic control of EHMT2 could be an important regulator of cancer metabolism in NSCLC cells.

Overexpression of EHMT2 in NSCLC
To examine the expression levels of EHMT2 in different types of lung cancers, two datasets publicly available from Oncomine database (http://www.oncomine.com/) [24] were analyzed ( Figure 1A,B): the Hou lung data set [25] and the Bhattacharjee lung dataset [26]. EHMT2 expression was significantly higher in NSCLC, including adenocarcinoma (AD), squamous cell carcinoma (SCC), and large cell lung cancer (LCLC), compared to normal tissue, while it did not show a significant difference in small cell lung cancer (SCLC). On the contrary, EHMT1, another lysine methyltransferase, did not show differential expression in NSCLC (data not shown). In addition, high EHMT2 protein expression was significantly correlated with poor prognosis (Supplementary Figure S1). Together, this suggests that EHMT2 overexpression is a relevant cancer characteristic with possible ties to tumorigenesis.

Effects of EHMT2 Regulation on Cell Viability
Next, we investigated the effects of EHMT2 in lung cancer cell lines H1299 and A549 to understand the role of EHMT2 in cancer cell proliferation. The cells were treated with the EHMT2-specific inhibitor BIX01294 (BIX), and cell viability was measured using the MTT assay. Cell viability decreased after BIX treatment as compared to non-treatment in both cell lines (Figure 2A,B). Also, we examined cell growth in the IncuCyte Zoom system and found that EHMT2 inhibition by BIX treatment hindered cell proliferation in a dose-dependent manner ( Figure 2C). Since 7.5 µM of BIX treatment was the minimal concentration that showed a significant difference in cell viability, this concentration was used for further experiments. In addition, transcriptional repression of EHMT2 mediated by specific targeting siRNA significantly decreased cell viability ( Figure 2D). To elucidate the mechanism of suppressing cell proliferation by EHMT2 inhibition, we tested whether BIX-induced cell death was mediated by autophagy. The autophagy-related genes Autophagy Related gene 5 (ATG5), Autophagy Related gene 12 (ATG12), and Microtubule-associated proteins 1A/1B light chain 3B (MAP1LC3B), were measured upon BIX-treatment. The mRNA level of ATG5 and MAP1LC3B, not ATG12, significantly increased in a dose-dependent manner ( Figure 2E). Western blot analysis using an antibody against LC3B, encoded by the MAP1LC3B gene, confirmed the autophagy induction by BIX-treatment ( Figure 2F). These results suggested that EHMT2 inhibition induced cell death through autophagy.

Distinct Gene Expression Profiles with EHMT2 Inhibition in H1299 Cells
To understand the effects of EHMT2 inhibition on global gene expression, RNA sequencing analysis was conducted on cells treated with or without BIX. EHMT2 inhibition exhibited distinct gene expression profiles ( Figure 3A). In total, 569 genes out of 23,912 genes passed the cutoff (p value < 0.05 and log 2 FC ≥ |0.6|) and among them, 147 genes (26%) were downregulated and 422 genes (74%) were upregulated. The major biological function of differentially expressed genes (DEGs) was analyzed by biological process (BP) of Gene Ontology (GO) and Reactome using Enrichr. Interestingly, metabolism-related terms were overrepresented in upregulated DEGs, where the cholesterol biosynthesis pathway was a top ranked biological term from both the Reactome and GO BP pathways ( Figure 3B,C). The majority of downregulated DEGs were, in contrast, involved in cell cycle-related processes, including mitotic chromosome condensation and DNA repair pathways ( Figure 3D,E). Taken together, BIX-mediated EHMT2 inhibition induced the cholesterol biosynthesis pathway and repressed cell cycle and DNA repair pathways.

Induction of Cholesterol Biosynthesis Pathway by BIX Treatment
Next, we focused on cholesterol biosynthesis pathway, which was identified as a top candidate of BIX-induced transcriptional changes, according to Figure 3B,C. We checked expression of individual genes related to cholesterol biosynthesis pathway ( Figure 4A,B). Most of cholesterol biosynthesis related genes, except CYP51 and TM7SF2, were significantly upregulated by EHMT2 inhibition, indicating that BIX-treatment promoted overall cholesterol biosynthesis pathway. Moreover, the expression of HMGCR and HMGCS1, which are the rate-limiting step of cholesterol biosynthesis [27], was verified by qRT-PCR. Both mRNA level of HMGCR and HMGCS1 increased significantly after BIX treatment ( Figure 4C). (E) Expression of cholesterol synthesis-related genes in BIX-treated A549 was analyzed. (F) Results from chromatin immunoprecipitation using antibodies against histone H3 lysine 9 mono-methylation (H3K9me1) and histone H3 lysine 9 di-methylation (H3K9me2) were presented as % enrichment of input at the promoter of SREBF2. IgG is a negative control for immunoprecipitation. * p < 0.05 against the 0 µM BIX treatment group.
To identify how integral cholesterol biosynthesis was regulated by BIX-treatment, we investigated the alteration of the cholesterol biosynthesis master regulator, SREBF2. We found that SREBF2 expression significantly increased upon EHMT2 inhibition ( Figure 4D). In addition, expression of SREBF2, along with HMGCR and HMGCS1, was also increased after BIX treatment in A549 ( Figure 4E). To further characterize SREBF2 regulation by EHMT2, chromatin immunoprecipitation (ChIP) assay was performed with H3K9me1 and H3K9me2 antibodies. Inhibition of EHMT2 significantly reduced the enrichment of H3K9me1 and H3K9me2 at the promoter of the SREBF2 locus ( Figure 4F). This suggested that EHMT2 inhibition directly upregulated SREBF2 transcription by lowering the enriched level of H3K9me1 and H3K9me2, leading to induction of cholesterol biosynthesis.

Involvement of Cholesterol Biosynthesis in BIX-Induced Cell Death
To confirm that cholesterol biosynthesis was involved in BIX-induced cell death, cells were co-treated with BIX and 25-hydroxycholesterol (25-HC), an inhibitor of cholesterol biosynthesis [28]. BIX increased the expression of SREBF2, HMGCR, and HMGCS1, while in contrast, their expressions were significantly reduced by the 25-HC treatment ( Figure 5A-C). Interestingly, 25-HC did not change the expression of genes related to cholesterol biosynthesis pathway in the absence of BIX treatment ( Figure 5A-C), suggesting that the induction of cholesterol biosynthesis pathway occurred upstream in response to BIX treatment. Based on the MTT assay, cell viability decreased upon EHMT2 inhibition, but 25-HC treatment recovered the decreased cell viability, at least in part (about 13%) ( Figure 5D). Since we discovered that BIX-induced cell death was involved in autophagy, we hypothesized that cholesterol biosynthesis pathway may be related to autophagy. Thus, we investigated whether inhibition of cholesterol synthesis pathway attenuated the BIX-induced autophagy. The expression of autophagy-related genes was measured at mRNA and protein level. We found that expression of ATG5 and MAP1LC3B was significantly upregulated by EHMT2 inhibition, and the upregulation was attenuated by 25-HC treatment at the mRNA level of ATG5, but not MAP1LC3B ( Figure 5E,F). Furthermore, 25-HC treatment also decreased the level of the LC3B protein upon EHMT2 inhibition, although the basal level in the absence of BIX was higher in 25-HC-treated group. The results indicated that cholesterol biosynthesis pathway was involved in BIX-induced cell death via autophagy.

Discussion
In this study, we showed that the inhibition of the histone methyltransferase EHMT2 using BIX induced cell death via autophagy. Most cholesterol biosynthesis-related genes were upregulated upon BIX-treatment. The link between EHMT2 and cholesterol biosynthesis was based on the finding that EHMT2 inhibition led to reduced H3K9me1 and H3K9me2 at the promoter of the SREBF2 locus, resulting in transcriptional upregulation of SREBF2 and its target genes. The 25-HC-mediated inhibition of cholesterol biosynthesis recovered BIX-induced cell death via attenuating autophagy. Therefore, inhibition of EHMT2 could be a potential approach to induce cell death via alteration of expression of SREBF2 and its downstream cholesterol synthesis.
NSCLC has higher EHMT2 expression compared to the normal tissues according to the Oncomine datasets ( Figure 1). The anti-cancer effect of EHMT2 inhibition has been studied in several cancer cell lines in primary cells of breast cancer, neuroblastoma, colon, and bladder cancer [12,29,30], suggesting possible therapeutic options for cancer treatment. Our results showed that EHMT2 inhibition led to cell death in NSCLC cells, associated with BIX-mediated induction of autophagy. Autophagy has been demonstrated to contribute to cell death in the absence of the apoptotic pathway [31]. ATG5 is constitutively conjugated to ATG12 and forms the ATG5-ATG12 complex, which is necessary for LC3B lipidation [32]. This process is required for autophagosome formation [33]. BIX treatment resulted in transcriptional induction of MAP1LC3B and ATG5 genes and an increase of LC3B-II production. Consistent with our result, previous reports also showed that EHMT2 inhibition induced autophagic cell death since suppression of autophagy by an inhibitor or siRNA knockdown reduced BIX-mediated cell death in other cell types including breast, pancreatic, and colon cancer cell lines [30,34,35]. In addition, we also observed apoptotic cell death upon the treatment although the extent was limited (Supplementary Figure S2).
Genome-wide RNA sequencing upon BIX-treatment revealed that the most prominent change in gene expression was induction of metabolic pathways, in particular, cholesterol biosynthesis ( Figure 2B,C). There have been a few studies about the roles of EHMT2 and its association with metabolism [12]. EHMT2 epigenetically activates the serine-glycine biosynthetic pathway, which is critical for cancer cell survival and proliferation through ribosome synthesis and cell cycle progression [12]. It also influences insulin regulation in hepatic cells by activating HMGA1 independently of its enzymatic activity [36]. Based on our findings, EHMT2 seems to directly regulate transcriptional expression of SREBF2, a master regulator for cholesterol biosynthesis, in effect lowering cholesterol in lung cancer. We proved this experimentally by blocking EHMT2, derepressing SREBF2 resulting from decreased H3K9me1 and H3K9me2, leading to transcriptional induction of canonical downstream target genes. Similar to our results, it has been reported that BIX treatment selectively induced upregulation of cholesterol biosynthesis in pancreatic cancer cells [35]. However, the BIX-induced upregulation did not take place in the liver cancer cells [35], indicating that it is cell-type specific.
Moreover, an SREBP2-mediated induction of the cholesterol biosynthesis pathway was involved in cell death in part through the induction of autophagy. We also proved this experimentally by showing that 5-HC-mediated inhibition of cholesterol synthesis attenuated expression of ATG5 ( Figure 4E). While it did not influence MAP1LC3B mRNA expression, it did reduce LC3B-II, likely as a result of the absence of ATG5, a key component of the LC3B-II binary complex. To corroborate our findings, previous studies have shown the relationship between cholesterol synthesis or SREBP2 and autophagy [37,38]. SREBP2 directly binds to the promoter of several murine autophagy genes, such as Map1lc3b, Atb4b, and Atg4d, and SREBF2 knockout disrupts autophagosome formation in nutrient starvation in mouse models [37]. Gastric cancer cells also represent a case where cholesterol supplement in the culture media decreased cell viability and clonogenicity via both autophagy and apoptosis [38]. In our results, it is not clear whether autophagy was induced by direct activation via SREBF2 binding to the promoters of autophagy genes or by cholesterol supplement for autophagosome formation. There have also been conflicting results suggesting cholesterol depletion-induced autophagy [39] and an inverse link between cholesterol biosynthesis and high cell proliferation rate in other cancers [40]. Further investigation is needed to elucidate why EHMT2 inhibition-induced autophagy required the cholesterol synthesis pathway. It nevertheless remains clear that cholesterol biosynthesis, EHMT2 transcriptional control, and autophagy are linked in ways that relate to proliferative activity and cell viability, likely essential to lung cancer-related tumorigenesis.
In conclusion, EHMT2 inhibition using the specific inhibitor BIX induced cell death via autophagy in NSCLC. It led to upregulation of cholesterol biosynthesis by derepressing SREBF2 expression.

MTT Assay for Cell Viability
Cells were seeded in 96-well plates 24 h prior to treatment. BIX were treated for 48 h. Thiazolyl blue tetrazolium bromide (MTT) (Sigma-Aldrich, Saint Louis, MO, USA; M2128) solution was added as one-tenth the original culture volume and incubated for 3 h. Media were removed and 100 µL DMSO (Sigma-Aldrich, Saint Louis, MO, USA; D4540) per well were added to convert MTT to purple formazan in mitochondria of viable cells. The optical density (OD) of each culture well was measured using an ELISA reader at 562 nm. The OD 562 in control cells was taken as 100%.

IncuCyte Zoom Assays
IncuCyte ZOOM System real-time instrumentation (Essen Bioscience, Ann Arbor, MI, USA) captures phase contrast images and measures cell proliferation. Cells were seeded in six-well plates and incubated for 24 h. After each treatment, plates were placed in IncuCyte. IncuCyte captured cell images every 2 h and the obtained images were analyzed by IncuCyte ZOOM™ 2015A software (Essen Bioscience, Ann Arbor, MI, USA).

RNA Isolation and Quantitative Reverse Transcription RT-PCR
Total RNA was isolated with Trizol (Ambion, Carlsbad, CA, UCA; 15596018) according to the manufacturer's protocol. RevertAid Reverse Transcriptase (Thermo Fisher Scientific, Darmstadt, Germany) was used for cDNA synthesis. Quantitative reverse transcription PCR (qRT-PCR) was performed using the SYBR green PCR master mix (Qiagen, Hilden, Germany) by Rotor Gene Q (Qiagen, Hilden, Germany). All samples were normalized to TATA-binding protein (TBP) mRNA levels. Primers used for RT-PCR were: ATG5 forward

RNA-Sequencing Analysis
Total RNA from H1299, treated with or without BIX 7.5 µM, was isolated using RNeasy Mini Kit (Qiagen). RNA-seq was performed on the RNA with TrueSeq RNA kit for library preparation and paired-end sequencings on an Illumina HiSeq 2500 instrument by Macrogen Inc (Seoul, Korea). 5.8 Gbp read bases per sample on average were obtained with a Q30 phred quality score of 94-96%. The received data were processed with Tuxedo pipeline, resulting in average 31,573 peaks counted from cufflinks. Overlapped genes from data sheets of four samples were merged and removed unannotated genes. Finally, 23,912 peaks were selected and reads counts for each gene were converted into base 2 logarithm. For further analysis, the cutoff standard to display differentially expressed genes (DEGs) was set at p-value < 0.05 and log 2 FC ≥ |0.6| (fold change 1.5), resulting in 569 genes. The DEGs were assessed by Gene Ontology (GO) biological process (BP) terms [43] and Reactome [44] using Enrichr, a comprehensive gene set enrichment analysis web server (http://amp.pharm.mssm.edu/Enrichr) [45].

Statistics
Statistical analysis was performed with Microsoft Excel and R studio using student's t-test with two-tailed and equal variances. Data were visualized with means ± standard deviations.

Conflicts of Interest:
The authors declare no conflict of interest.