EHMT1/EHMT2 in EMT, cancer stemness and drug resistance: emerging evidence and mechanisms

Metastasis, therapy failure and tumour recurrence are major clinical challenges in cancer. The interplay between tumour‐initiating cells (TICs) and epithelial–mesenchymal transition (EMT) drives tumour progression and spread. Recent advances have highlighted the involvement of epigenetic deregulation in these processes. The euchromatin histone lysine methyltransferase 1 (EHMT1) and euchromatin histone lysine methyltransferase 2 (EHMT2) that primarily mediate histone 3 lysine 9 di‐methylation (H3K9me2), as well as methylation of non‐histone proteins, are now recognised to be aberrantly expressed in many cancers. Their deregulated expression is associated with EMT, cellular plasticity and therapy resistance. In this review, we summarise evidence of their myriad roles in cancer metastasis, stemness and drug resistance. We discuss cancer‐type specific molecular targets, context‐dependent mechanisms and future directions of research in targeting EHMT1/EHMT2 for the treatment of cancer.

Deregulated EHMT1 and EHMT2 expression is associated with advanced tumour stage, metastasis and poor prognosis and is thus of clinical relevance. The aberrant epigenetic landscape mediated by EHMT1/ EHMT2 (EHMT1/2) is linked with cancer-promoting events including cellular proliferation, metabolic adaptation, EMT, metastasis, stress tolerance/response and stem cell maintenance [10][11][12]. We discuss EHMT1/2dependent mechanisms involved in EMT, stemness, drug resistance and metabolism.

Cancer metastasis
Cancer cells adapt to multiple constraints posed by the tumour microenvironment by altering the genetic and epigenetic landscape. Metastasis is associated with acquisition of mesenchymal characteristics and loss of epithelial traits. EMT is a dynamic, reversible process which involves transition of epithelial cells to mesenchymal-like cells by gaining characteristics including motility, invasiveness and resistance to apoptosis, while losing cell-cell adhesion and apical-basal polarity [13]. These changes are apparent by the downregulation of epithelial markers such as E-cadherin, and upregulation of mesenchymal markers such as N-cadherin and vimentin. EMT is associated with invasion, migration, stroma formation, metastasis, cellular plasticity and drug resistance [14]. In this section, we discuss the targets and mechanisms underlying EHMT1/2 in driving EMT and metastasis ( Fig. 1 and Table 1) in distinct cancers.

Breast cancer
Breast cancer is highly heterogeneous and is classified into multiple subtypes based on molecular differences (hormone receptor and gene expression pattern) [15]. Interestingly, transforming growth factor beta (TGFb)-induced EMT in normal breast epithelial cell lines is associated with EHMT1/2-mediated H3K9me2 repression mark at the E-cadherin promoter. EMT was apparent by the induction of Snail, the downregulation of epithelial markers E-cadherin and claudin-3 and -7, and acquisition of mesenchymal markers vimentin and N-cadherin [11].
EHMT2 is overexpressed in breast cancer. High EHMT2 expression is associated with uncontrolled proliferation, metastasis and recurrence [16]. Triple-negative breast cancers (TNBCs) are deficient in oestrogen receptor (ER), progesterone receptor (PR) and human epidermal growth factor receptor 2 (HER-2) and are characteristically highly aggressive and recurrent [17]. Treatment with the EHMT2 inhibitors (UNC0638 and BIX-01294) or siRNA-mediated inhibition of EHMT2 reduced metastasis in TNBC cell lines. Suppression of metastasis was accompanied by reduced expression of EMT transcription factors (EMT-TFs), Snail-1 and Snail-2/Slug and mesenchymal markers (N-cadherin and vimentin), and increased expression of epithelial markers (E-cadherin and claudin) and E-cadherin promoter activity. Furthermore, EHMT2 knockdown in TNBC cell lines resulted in altered cell adhesion, microtubule formation and depolymerisation of gene signatures Mechanistically, EHMT2 suppressed E-cadherin by regulating the expression and activity of MSK1, which transcriptionally upregulates Snail-1 and promotes metastasis in TNBC [18,19]. Apart from EMT markers, EHMT2-associated H3K9me2 and DNA methyltransferases (DNMT)-mediated DNA methylation were pivotal in the transcriptional repression of desmocolin 3 (DSC3) and mammary serine protease inhibitor (MAS-PIN), which are anti-metastatic and tumour suppressor genes respectively [20]. EHMT2 knockdown exhibited similar changes in EMT and metastasis in claudin-low breast cancer (CLBC) subtype. In addition, activation of signalling pathways including TGF-b and b-catenin suppressed tumour growth and lung colonisation. Mechanistically, Snail-mediated recruitment of EHMT2 and subsequent association with DNMTs were responsible for E-cadherin promoter methylation and its transcriptional silencing [11].
Hypoxia influences tumour growth and metastasis. Under hypoxic conditions, EHMT2 was stabilised through reduced proline hydroxylation, which impairs its proteasomal degradation. The increased EHMT2 levels under hypoxic conditions downregulated genes involved in tumour suppression, leading to increased motility and survival of breast cancer cells. In addition, EHMT2 transcriptionally silenced CDH10, a type II cadherin, leading to breast cancer metastasis [21,22].
Leptin is associated with obesity and tumour progression in various cancers [23,24] and contributes to breast cancer progression by activating signal transducer and activator of transcription 3 (STAT3). STAT3 was found to recruit EHMT2, resulting in suppression of miR-200c-3p and leading to EMT in MCF ER + breast cancer cells and tumour aggressiveness in vivo. Interestingly, enrichment of STAT3-EHMT2 at the miR-200c-3p promoter was higher in TNBC as compared to the luminal subtype [25]. The recruitment and interaction of EHMT2 with Chromodomain on Y-like 2 (CDYL) transcriptionally repressed miR124. The downregulation of miR124 is associated with invasive, migratory and stemness of breast cancer cells through activation of NF-jB and STAT3 signalling [26].
EHMT2 is a mediator of hypoxic response in breast cancer and it is apparent that EHMT2 is overexpressed in different breast cancer subtypes and exerts epigenetic The schematic depicts multiple EMT-related genes which are associated with and controlled by EHMT2. Although EHMT2 regulation of E-cadherin is well studied across various cancers, mechanisms through which EHMT2 regulates other EMT-related genes such as Snail, vimentin, fibronectin and claudin are unclear. It is also unknown how EHMT2 contributes to various states of EMT. regulation of multiple targets culminating in the induction of EMT. Hence, it may be a therapeutic target or a potential biomarker to stratify breast cancer patients on the basis of response to EHMT2 inhibitors. The role of EHMT1 in breast cancer appears distinct from EHMT2. Analysis of TCGA database revealed lower expression of EHMT1 in the highly metastatic basal-like breast cancer (BLBC) when compared to non-basal subtypes [27,28]. The expression of EHMT1 circular RNA (circEHMT1) is supressed in breast cancer. circEHMT1 reduced matrix metallopeptidase-2 (MMP-2) expression and metastasis by downregulating miR-1233-3p [29]. In addition, miR-210 correlated with poor prognosis and the transition from ductal carcinoma in situ (DCIS) to invasive ductal carcinoma (IDC). EHMT1 was one among the major genes [breast cancer type 1 susceptibility protein (BRCA1), RB1, CDH1 and PARP1] that was downregulated in IDC when compared to DCIS [30]. These studies suggest a tumour suppressor role for EHMT1 which requires further investigations.

Head and neck squamous cell carcinoma and endometrial cancer
In head and neck squamous cell carcinoma (HNSCC), elevated EHMT2 expression is correlated with poor prognosis [31]. EHMT2, in association with Snail-1, transcriptionally repressed E-cadherin via deposition of H3K9me2 marks at its promoter. In addition, EHMT2facilitated EMT in the presence of TGF-b. Depletion of EHMT2 activity with BIX-01294 prevented metastasis, reduced EMT markers (N-cadherin and vimentin) and reversed EMT in HNSCC [32]. In endometrial cancer cell lines, elevated EHMT2 expression was correlated with deep myometrial invasion and poor prognosis alongside negative correlation with E-cadherin. EHMT2-mediated recruitment of DNMT1 repressed E-cadherin, leading to invasion and migration. Although Snail was associated with the repressor complex, inhibition of EHMT2 did not affect the recruitment of Snail to E-cadherin promoter but resulted in a drastic reduction in DNMT1 and H3K9me2 marks [33].
Gastric cancer EHMT1 and EHMT2 overexpression in gastric cancer is associated with late tumour stage, advanced tumour growth (lymph node and peritoneal metastasis) and poor prognosis [34][35][36]. EHMT1 knockdown induced E-cadherin expression, resulting in reduced peritoneal metastasis in vitro and in vivo [34]. Regenerating protein IV (Reg IV) was found to upregulate EHMT2 expression, leading to activation of integrin-b3 (ITGB3), a member of the integrin family involved in metastasis by promoting the adherence of gastric cancer cells to the peritoneum [37]. EHMT2 served as a scaffolding protein to form an activator complex with glucocorticoid receptor (GR) and p300, leading to expression of the GR target gene ITGB3 in methyltransferase activity-independent manner.

Cervical cancer
Human cervical cancer cell lines (CaloGR) exhibit high levels of EHMT2 as compared to cervical epithelial cells. Loss-of-function studies indicated a reduction in cell adhesion and invasive capacities through downregulation of E-cadherin [38]. Pharmacological and genetic inhibition of EHMT2 impeded angiogenesis, invasion and migration both in vitro and in vivo, with a reduction in the expression of angiogenic genes  CXCL16, IL-8, VEGF, MMP9, MCP-1,  Pentraxin-3, GM-CSF, Serpin E1,TIMP-1, TIMP-4 and Thrombospondin-1 [39]. O-GlcNac transferase (OGT), an enzyme responsible for the addition of single O-linked N-acetylglucosamine (O-GlcNAc) to serine and threonine residues of proteins, is upregulated and correlated with poor prognosis in cervical cancer. EHMT1 and other histone modifiers were identified in the OGT interactome. Notably, Snail1 and ING were identified as targets of OGT, suggesting that OGT and histone modifiers may cooperate in regulation of EMT-TFs. As EHMT2 is involved in EMT through association with multiprotein repressor complexes involving Snail-1 and DNMT, it is possible that in cervical cancer EHMT1 and OGT act as a complex to regulate EMT-related genes [40].

Pancreatic cancer
EHMT2 overexpression in pancreatic cancer cell lines and pancreatic ductal epithelial cells promotes EMT. Similar effects were seen in a gemcitabine (GEM)resistant pancreatic cancer cell line. The EHMT2 inhibitor UNC0638 prevented invasion of tumours in orthotopic xenograft mouse models. EHMT2 mediated H3K9me2 marks at the E-cadherin promoter and also upregulated the expression of PCL3 (component of PRC2 complex) to promote enhancer of zeste homolog 2 (EZH2) recruitment, leading to E-cadherin silencing. Furthermore, EHMT2 downregulated lysine demethylase 7A (KDM7A) expression which mediates H3K27 demethylation [41,42]. However, no correlation was noted between EHMT2 expression and tumour metastasis in patient tissues.
Ovarian cancer EHMT2 expression is elevated in high-grade seroustype ovarian cancer (HGSOC) and metastatic lesions compared to primary tumours. Inhibition of EHMT2 reduced invasion, migration and anoikis resistance in cell lines, and attenuated metastasis to abdominal organs and ascites in a peritoneal metastasis model. EHMT2 regulates many metastasis genes, including Ecadherin, by promoter methylation. However, overexpression studies did not reveal a marked increase in metastasis suggesting that other epigenetic regulators may control metastasis [43].
EHMT1/2 levels are induced by hypoxia and exert tumorigenic effects via methylation of a wide range of histone and non-histone substrates [44]. Consistently, recruitment of EHMT1/2 and concomitant H3K9me2/ me3 at the E-cadherin promoter were drastically increased during hypoxia. Gene set enrichment analysis (GSEA) of 478 ovarian cancer tissues revealed a negative correlation between the expression of EHMT2 and anti-metastatic gene sets. Specially, 'aigner-ZEB1target' gene set which included genes involved in epithelial differentiation and cell adhesion (CDH1, CDH11, TSPAN15, CLDN7, MASPIN and DSC3) was observed [45].

Hepatocellular carcinoma (HCC)
Activation of TGF-b and Hedgehog signalling induced EMT in hepatocytes [46,47]. Initially, although interaction between Snail and EHMT2 was detected, there was no association with TGF-b-induced EMT in HCC [48]. However, much like other cancers, overexpression of EHMT2 in HCC was associated with aggressiveness and poor prognosis [49]. EHMT2 was upregulated due to gene amplification and loss of miR-1 (a negative regulator of EHMT2). Depletion of EHMT2 through RNAi/CRISPR/CAS9 and pharmacological inhibition reduced proliferation and metastasis both in vitro and in vivo through repression of RARRES3, a tumour suppressor gene [50]. Transcriptomic analysis in HCC cell line (HepG2) after treatment with CM-272 indicated a difference in genes involved in metastasis. Furthermore, treatment with CM-272 restored E-cadherin expression by mechanisms that are unclear [49]. Translation regulatory long non-coding RNA 1 (TRERNA1), which promotes cancer metastasis in several cancers [51][52][53], induced metastasis in HCC by recruiting EHMT2 to a Snail-1 repressor complex and/or by increasing H3K9me2 at the E-cadherin promoter. Surprisingly, other EMT markers (except vimentin and E-cadherin) were significantly unaltered upon TRERNA1 suppression [54]. In various cancers, the association of Snail-1 and EHMT2 is well documented in controlling Ecadherin expression but in HCC, in addition to Snail-1, Snail-2 was also associated with EMT. Furthermore, Snail-2 mediated recruitment of histone deacetylases (HDAC 1, 2 and 3), while EHMT2 was essential for the removal and deposition of H3K56ac/H3K4ac and H3K9me2 marks, respectively, to negatively regulate E-cadherin. Indeed, pharmacological inhibition of EHMT2 and HDAC attenuated metastasis by reactivating E-cadherin in HCC cell lines [55]. Thus, EHMT2 mostly works in association with other epigenetic regulators to alter EMT in HCC.

Lung cancer
Overexpression of EHMT2 was identified in non-small cell lung cancer (NSCLC) as compared to normal lung tissue. EHMT2 contributes to various stages of malignancy by regulating different signalling pathways. For instance, EHMT2 together with heterochromatin protein 1a (HP-1a) and DNMT1 transcriptionally supress the wingless-related integration site (Wnt) inhibitor adenomatous polyposis coli protein 2 (APC2), thereby promoting Wnt signalling and tumour growth [56]. Recently, EHMT2 was reported to promote invasiveness through upregulation of the focal adhesion kinase (FAK) pathway, wherein EHMT2 represses IjBa through H3K9me2, leading to activation of NF-jB and its downstream target FAK [57]. In addition, EHMT2 knockdown modulated the expression of HP-1a, APC2, ANGPTL4, u-PA, jPH3, TLN1, ABCG2 and TERT, which are involved in growth, adhesion, angiogenesis, apoptosis and hypoxia. Similar to HCC, in lung cancer cells, a repressor complex of Snail-2/ EHMT2/HDAC promoted EMT by repressing Ecadherin expression [58]. In NSCLC, EHMT2 silences caspase 1 (CASP1), a component of the inflammasome complex, thereby facilitating invasion and migration, colony formation and tumour growth [59]. The tumour suppressor SATB2 prevented invasion/EMT in NSCLC via downregulation of EHMT2 and associated histone marks (H3K9me3, H3K20me3 and H3K27me3). Furthermore, SATB2 upregulated expression of epithelial markers (E-cadherin and b catenin) and downregulated expression of EMT-TFs Slug and Snail and mesenchymal markers (vimentin). However, it is not clear whether EHMT2 is indispensable for SATB2-dependent effects [60].
EHMT2 is overexpressed in highly metastatic lung cancer cell lines [61]. As in other cancers, high EHMT2 expression is correlated with advanced tumour stage, lymph node metastases and poor prognosis. Tissue microarray of lung cancer patients indicated an inverse correlation between EHMT2 and epithelial cell adhesion molecule (EpCAM). Depletion of EHMT1 and EHMT2 prevented invasion, which is an early phase of metastasis, in lung cancer both in vitro and in vivo. EHMT1 was shown to be essential for EHMT2-mediated promotion of cell migration through its ability to stabilise the latter. Mechanistically, EHMT2 along with HP1, DNMT and HDAC prevented transactivation of Ep-CAM by SP-1 [62].

Prostate cancer
EHMT2 in association with Runx2 positively regulated the expression of genes (MMP9, MMP13, PGC, CSF2, SDF-1 and CST7) involved in prostate cancer growth, invasion and metastasis [63]. In contrast, interaction between EHMT1 and metallothionein 1h (MT1h) was necessary for histone methylation and tumour suppressor activity. Knockdown of EHMT1 or disruption of the binding with MT1h reduced tumour suppressor function. This study indicates a tumour suppressor role of EHMT1, with MT1h as a cofactor, in suppressing metastasis in PCa [64].
Osteosarcoma NEAT1, a long non-coding RNA (LncRNA) which is overexpressed in osteosarcoma, is associated with distant metastasis and poor prognosis. In osteosarcoma cell lines, NEAT1 is present in a repressor complex with EHMT2-Snail-DNMT1, resulting in repression of E-cadherin. Furthermore, NEAT1 reduced expression of epithelial markers (E-cadherin and a-catenin) and increased expression of mesenchymal markers (vimentin and N-cadherin) [65].

Other cancers
In oesophageal squamous cell carcinoma (ESCC), EHMT1 and EHMT2 were shown to be overexpressed compared to normal oesophageal tissue. The expression of EHMT1 and EHMT2 correlated with the magnitude of invasion, lymph node metastasis and tumour stage [66,67]. Similarly, EHMT2 expression was associated with distant metastasis in colorectal cancer (CRC) [68]. EHMT2 correlated with the expression of integrin subunit alpha-4 (VLA-4) in acute lymphoblastic leukaemia (ALL) which facilitated ALL cells to migrate. However, depletion or pharmacological inhibition of EHMT2 with BIX-01294 did not affect VLA-4 expression but modulated cell migration, suggesting a distinct mechanism by which EHMT2 regulates ALL cell motility.

Cancer stemness
In concordance with their role in maintenance of pluripotency in ES cells [69][70][71], EHMT1/2 have been shown to be critical regulators of stem-like transcriptional signatures in cancer cells ( Fig. 2 and Table 2).
In acute myeloid leukaemia (AML) mouse models, EHMT1/2 was shown to induce rapid progression towards end-stage AML. EHMT2-deficient mice showed reduced self-renewing leukaemia stem cell (LSC) frequency, with no adverse effect on haematopoietic stem cells (HSCs). Moreover, ablation of EHMT2 and treatment with EHMT1/2 inhibitor (UNC0638) both in vivo and in vitro showed a significant increase in differentiated myeloid cells, emphasising the importance of EHMT1/2-dependent methyltransferase activity in LSCmediated AML progression [72]. While elevated EHMT2-mediated H3K9me2 mark was correlated with increased cellular differentiation, EHMT2 overexpression in AML appears to augment LSC frequency.
In contrast, EHMT2 exhibited an inhibitory role in self-renewal of glioma cancer stem cells (CSC). CD133 + glioma CSCs were H3K9me2 negative in contrast to CD133cells, which immunostained positively for H3K9me2. Moreover, enhanced tumorsphere formation in EHMT2 inhibitor (BIX-01294)-treated glioma CSCs confirmed that EHMT2 obstructs self-renewal. Consistent with its role as a transcriptional repressor, EHMT2 directly repressed CD133 and SOX2 transcriptionally by methylating the promoter and enhancer. In glioma CSCs, EHMT2-mediated H3K9me2 serves as a repressive switch for self-renewal that is seen in ES cells as well [73]. Similarly, tumour repopulating cells (TRC) in melanoma exhibited low levels of H3K9me2 and elevated expression of SOX-2. EHMT2, FAK or CDC42 silencing increased proliferation/self-renewal capacity of melanoma cells through elevated SOX-2 expression [74,75].
Likewise, in lung adenocarcinoma, EHMT2 inhibited tumour progression by functioning as a suppressor of tumour propagating cells (TPC). EHMT2 inhibition in lung adenocarcinoma cells promoted stemness and accelerated tumour progression through upregulation of KRAS-and ECM-associated genes including MMP10. Indeed, depletion of Kdm3a suppressed CSCs, suggesting it as a potential target for lung adenocarcinoma [76]. Although this study was limited to analysis of TPCs expressing stem cell antigen-1 (Sca-1) and CD24, it nonetheless highlighted the importance of tumour heterogeneity, the cancer-type-dependent role of EHMT2 and the detrimental effects of inhibiting its activity in lung cancer. In contrast, knockdown of EHMT1, which is significantly elevated in lung cancers, reduced formation and aggregation of 3D spheroids via upregulation of CDKN1A [77].
Inhibition of EHMT2 in GEM-resistant pancreatic cancer cell line PANC-1-R reduced expression of CD133, Nestin and Lrg5, decreased tumorsphere In contrast, in breast cancer, CDYL2 recruits EHMT2 to microRNA genes to regulate stemness and metastasis via p65/NFjB and STAT3 pathways. In pancreatic cancer, EHMT2 regulates IL-8/CXCR1/2-mediated autocrine and paracrine signalling resulting in the induction of stemness and drug resistance. These studies highlight the context-dependent specific role of EHMT2 in various cancers.
formation and increased sensitivity to GEM. Furthermore, EHMT2-overexpressing PANC-1-R stimulated pancreatic stellate cells through activation of IL-8/ CXCR1/2 autocrine and paracrine signalling [42]. The study reveals a distinct mechanism of EHMT2dependent regulation of stemness in pancreatic cancer. In addition to IL-8, several other cytokines, namely CXCL5 and CXCL1, were modulated; the role of these cytokines needs further exploration.
HCC cell lines exhibited high expression of EHMT2 and its inhibition markedly reduced sphere-forming ability [49]. Interestingly, repression of miR124 by CDYL2-EHMT2 activated NF-jB and STAT3 signalling, resulting in stemness in addition to invasion and migration [26]. In addition, inhibition of EHMT2, which attenuated EMT in HNSCC, also reduced tumorspheres and the expression of CD44 [32], establishing a link between EHMT2-dependent regulation of EMT and stemness.
In colon cancer, EHMT2 regulates the CD133 + / ESA + (epithelial-specific antigen) CSCs. Indeed, EHMT2 knockdown reduced CD133 + cells, hindered migration and sphere formation [78]. Furthermore, the reduction in stem cell phenotype and metastatic properties was associated with its ability to alter DNA damage response through protein phosphatase 2Areplication protein A (PP2A-RPA) axis [79].
In non-small cell lung cancer (NSCLC), EHMT2 expressed in TICs was shown to induce CSC markers, sphere formation and growth both in vitro and in vivo. Genome-wide analysis of methylation changes upon EHMT2 knockdown in patient-derived TICs revealed upregulation of a set of tumour suppressor genes (CDYL2, DPP4, SP5, FOXP1, STAMBPL1 and ROBO1). This study established the role of EHMT2 in altering DNA methylation patterns in these target genes, leading to maintenance of TICs which facilitate metastasis and drug resistance [80].
In BCR-ABL + CML, leukaemia stem cells (LSCs) lie at the core of relapse and drug resistance. Tyrosine kinase inhibitors (TKIs) fail to eliminate LSCs. Similar to other cancers such as HNSCC [32], colon cancer [79] and pancreatic cancer [42], where EHMT2 has been reported to drive stemness and confer chemoresistance, EHMT2 is overexpressed in LSCs. Both in vitro and in vivo, EHMT2 inhibition resulted in significantly impaired self-renewal, increased apoptosis and prolonged survival. RNA sequencing analysis identified the tumour suppressor SOX6 as target of EHMT2 in CML LSCs [81]. As EHMT2 has a limited effect on normal haematopoiesis, its selective role in LSC sustenance suggests it is a promising therapeutic target for CML.

Drug resistance
CSCs contribute to drug resistance [82]. Given the importance of EHMT2 in maintenance of CSCs, it is not surprising that EHMT2 is involved in drug resistance by sustaining stemness. High EHMT2 expression is associated with drug resistance through its ability to modulate transcriptional regulation of tumour suppressor genes, drug metabolism, DNA repair and cell survival pathways ( Fig. 3 and Table 3).

DNA damage
EHMT1/2 protects and supports the growth of cancer cells by preventing chromosome disruption, recruiting DNA repair factors [p53-Binding protein 1 (53BP1), BRCA1 and RNF168] at double-stranded breaks (DSBs) and inducing DNA repair mechanisms (homologous recombination (HR) and non-homologous end joining (NHEJ)). Indeed, EHMT2 knockdown disrupted chromatin organisation at centromeres leading to instability [83], impaired DNA repair pathways in a p53-independent manner and prevented HR and NHEJ through reduced H3K9me1/2. Furthermore, inhibition of EHMT2 in osteosarcoma sensitised cells to irradiation [84]. EHMT2 inhibitors in combination with chemotherapeutic drugs sensitised cancer cells to DNA DSB and enhanced cell death in U2OS cells  [85]. Similarly, in glioma cell lines, loss of EHMT2 led to impaired DNA DSB repair signalling (HR and NHEJ pathways) and increased sensitivity to ionising radiation. EHMT2 inhibition reduced activation of ataxia-telangiectasia mutated (serine/threonine kinase) (ATM) kinase and tat interactive protein (Tip60), consequently decreasing the phosphorylation of KRAB domain-associated protein 1 (KAP-1, repressor protein) and increasing radiosensitivity [86]. Recently, EHMT2 expression was shown to initiate cancer in DNA damaged hepatocytes both in vitro and in vivo.
Mechanistically, EHMT2 negatively regulated the expression of Bcl-G (proapoptotic member of Bcl-2 family) by blocking p53 in these cells. Treatment of hepatoma cell lines with EHMT2 inhibitor sensitised them to DNA damage inducers (irradiation and hydrogen peroxide) and enhanced cytotoxicity [87]. While most studies have shown the involvement of EHMT2 in DNA repair, depletion of EHMT2 induced DNA DSB, chromosomal aberrations and senescence in CRC. EHMT2 inhibitors in combination with topoisomerase I inhibitor elevated c-H2A histone family member X (c-H2AX) expression and led to CRC cell death [88]. High expression of EHMT2 conferred cancer stemness to colon cancer cells, eventually causing resistance to radiation treatment. Knockdown of EHMT2 increased chemo-radio sensitivity and promoted DNA damage through transcriptionally upregulating PP2A in HT-29 colon cancer cells. PP2A impaired the phosphorylation of checkpoint kinase 1 (CHK-1) and RPA, thereby enhancing replicationlinked DSB [79]. Interestingly, in HGSOC, PARP inhibitor (PARPi) resistance in patient-derived xenografts (PDX) and PARPi-resistant ovarian cancer cell lines exhibited high levels of H3K9me2 and EHMT1/2. Inhibition of EHMT1/2 (using UNC0642) restored sensitivity to PARP inhibitors through induction of DNA damage and altered cell cycle regulation. Furthermore, RNA-seq analysis revealed alterations in survival pathway including mTOR, PI3K and Ak strain transforming, protein kinase B (AKT) [89].
Survival signalling pathways EHMT1/2 imparts drug resistance by upregulating survival signalling pathways in tumour cells. Pharmacological inhibition of EHMT2 augmented anti-tumour efficacy of temozolomide (TMZ), a therapeutic agent used as first line of treatment in glioblastoma (GBM). Pre-and post-treatment with the EHMT2 inhibitor BIX-01294 sensitised GBM cells to TMZ and restored apoptosis. In addition, EHMT2 inhibitors induced autophagic cell death and enhanced the anti-tumour effect of TMZ. However, the mechanisms underlying apoptosis and autophagy-related cell death need further investigation [90]. In multiple myeloma, combinatorial treatment with EHMT1/2 inhibitors and proteosome inhibitors induced autophagy cell death and tumor reduction in-vitro and in-vivo respectively. Mechanistically, the combinatorial treatment upregulated SAPK/JNK and p38 signalling while inhibiting mTOR signalling and cMyc levels in MM cells [91]. In the pancreatic ductal adenocarcinoma (PDAC) cell line (PANC-1), suppression of EHMT2 upregulated p27 and G1 cell cycle arrest, leading to sensitisation to PI3K/mTOR inhibitors BEZ235 and Cay10626 [92]. In fact, overexpression of EHMT2 and associated H3K9me2 led to the repression of coiled-coil domaincontaining protein 8 (CCDC8), a tumour suppressor that is correlated with aggressiveness in radio-resistant lung cancer cell lines. Treatment with EHMT2 inhibitors sensitised resistant cells A549/IR and XWLC-05/ IR to radiotherapy [93]. Epidermal growth factor receptor (EGFR)-induced expression of HER3 and activation of the PI3K/AKT signalling pathway contributes to therapy resistance in lung cancer. To identify potential targets against EGFR + lung cancer cells, 172 therapeutic agents that target stemness were screened. Among these, STAT3 inhibitor BBI608 was found to reduce viability of EGFR + cells. The inhibition of STAT3 in EGFR-TKI-resistant lung cancer cell lines downregulated EHMT2 and suppressed HER3 through miR-145-59, thereby preventing HER3mediated EGFR-TKI resistance. In addition, the EHMT2 inhibitor UNC0642 cotreated with afatinib significantly reduced tumorspheres in TKI-sensitive EGFR + cell lines [94].

Immunotherapies
In melanoma patients, EHMT2 is inversely correlated with T-cell signatures. Administration of EHMT1/2 inhibitor UNC0642 in combination with antiprogrammed cell death protein 1 (anti-PD1) and anti-cytotoxic T-lymphocyte-associated protein 4 (anti-CTLA-4) significantly regressed the tumours in a syngeneic melanoma mouse model [95]. Similarly, in bladder cancer, EHMT2 was associated with tumour recurrence and poor outcome. A combination of EHMT2 inhibitor with cisplatin and/or immune checkpoint inhibitor anti-programmed cell death ligand 1 (PD-L1) regressed tumours and prevented metastasis in aggressive transgenic mouse models. The inhibition of EHMT2 sensitised bladder cancer cells to PD-L1 and converted cold tumours (non-inflamed) to hot tumours (inflamed) through its ability to enhance Fig. 3. EHMT2-associated drug resistance and mode of EHMT2 inhibitors in drug sensitisation. In a tumour population, cells overexpressing EHMT2 (indicated by a green asterisk) remain unaffected by standard treatment regimens (radiotherapy, chemotherapy, targeted therapy and immune therapy) and increase response to treatment. On the other hand, sensitive cells on exposure to drug acquire resistance through expression of EHMT2. EHMT2 regulates various functional properties in these cells eventually conferring drug resistance. Treatment with EHMT2/EHMT1 inhibitors sensitise cancer cells to standard drug therapies. immune response regulation (IFNa, IFNb and TNFc) [96]. Altogether, these findings suggest an immunoprotective role of EHMT2 in the tumour microenvironment, correlate its overexpression with immune therapy resistance and highlight the potential of EHMT2 inhibitors in augmenting the efficacy of immune checkpoint inhibitors.

Gemcitabine sensitivity
The anti-tumour activity of GEM and resistance development is related to the expression of genes involved in membrane transport (human equilibrate nucleoside transporter 1, hENT1) and drug metabolism (deoxycytidine kinase, dCK). In cervical cancer, EHMT2-mediated H3K9me was shown to be responsible for the inactivation of genes hENT and hcdk, which was associated with drug resistance. Depletion of EHMT2 restored GEM sensitivity in a resistant CaloGR. Further treatment with hydralazine reduced the expression and activity of EHMT2, consequently modulating the expression of hENT and hcdk in CaloGR [97]. Additionally, EHMT2 overexpression enhanced proliferation, metastasis and GEM resistance in pancreatic cancer through increased production of IL-8, which promotes stromal-cancer cell interaction and ECM deposition associated with drug resistance. Inhibition of EHMT2 reduced stemness and sensitised GEMresistant cell lines (PANC-1-R) both in vitro and in vivo [42]. Cisplatin-resistant HNSCC cells exhibited and apoptotic cell death [104] high levels of EHMT2. In addition, overexpression of EHMT2 was responsible for multidrug resistance (oxaliplatin, 5-FU and bleomycin) in HNSCC. EHMT2mediated H3K9me1 protected the HNSCC cells by transcriptionally upregulating the expression of glutamate cysteine ligase catalytic subunit (GCLC), one of the enzymes involved in glutathione (GSH) biosynthesis. GSH is involved in cancer progression and drug resistance in various cancers [98]. Depletion of EHMT2 expression and activity sensitised cells to cisplatin by reducing GSH content and promoting apoptosis [99].

Acquired drug resistance
The evidence summarised so far correlate EHMT2 overexpression in tumour cells with their intrinsic drug resistance ability. On the other hand, overexpression of EHMT2 is also indicated in acquired drug resistance following therapy administration. For instance, overexpression of EHMT2 was associated with erlotinib (EGFR-TKI) resistance in NSCLC. The inhibition of EHMT2 led to the transcriptional activation of PTEN and reduced AKT signalling, which attenuated aggressiveness and sensitised NSCLC cells to TKI. Indeed, a combination of erlotinib and UNC0638 significantly reduced cancer growth and induced apoptosis in resistant cells in vitro and in vivo [100]. In PDAC, treatment with the MEK inhibitor (MEKi) resulted in resistance and SETD5 was identified as a major driver of acquired targeted therapy resistance. SETD5 lacks enzymatic activity and forms a corepressor complex either with EHMT2 or HDAC3 to exhibit therapy resistance. Transcriptomic analysis of resistant cell lines revealed that the loss of SETD5 or EHMT2 downregulated genes involved in GSH metabolism and the cytochrome P450 pathway, leading to cellular reprograming and improved response to MEKi. Therefore, combined pharmacological inhibition of MEK1, EHMT2 and HDAC3 reduced tumour growth in a relapse PDX model [101]. EHMT2 overexpression in GEM-resistant cell line was reported to promote EMT through its ability to negatively regulate E-cadherin [41]. In B-cell ALL (B-ALL), glucocorticoid-induced cell death depends on the coactivator complex chromobox family protein, CBX-3 and EHMT1/2. Upregulation of aurora kinase B (AURKB) resulted in phosphorylation of EHMT1/2, thereby preventing expression of genes responsible for glucocorticoid-induced cell death, conferring drug resistance. Inhibition of AURKB potentiated cytotoxic effects of glucocorticoid in NALM6 cells [102]. On the other hand, automethylation of EHMT1/ 2 enhances recruitment of HP-1 and coactivator complex formation to induce GR target genes. In line with the role of EHMT1/2 and its methylation in glucocorticoid-mediated regulation, treatment with JmjC family of lysine demethylase inhibitor (JIB-04) prevented EHMT2 demethylation and resorted sensitivity to glucocorticoid causing cell death in Nalm-6 cells [103]. Therefore, contrary to other cancers, methylation of EHMT1/2 is shown to be important for drug sensitisation in B-ALL cancer.
Nonetheless, cancer cells also develop resistance to EHMT2 inhibitors. In AML, treatment with BIX-01294 led to the upregulation of phosphorylated PERK and eIF2a, causing endoplasmic reticulum stress. Genetic suppression of PERK sensitised ALL cells to BIX-01294-induced apoptotic cell death. However, the mechanism involved in PERK inhibition and sensitivity of AML cells to BIX-01294 is unclear [104].

Metabolic alterations
Cellular metabolism yields metabolites which function as cofactors and substrates for various epigenetic regulators. Abnormal levels of metabolites (oncometabolites) lead to epigenetic dysregulation, and conversely, deregulation of the epigenetic landscape affects expression of metabolic enzymes that contribute to cancer progression [105][106][107]. Upregulation of glycolysis is observed both in primary and metastatic tumours. Increased uptake of glucose and altered glucose metabolism (glycolytic enzyme activation, impaired OXPHOS and suppressed gluconeogenic enzymes) are correlated with cancer metastasis and poor prognosis [108]. Consistent with this notion, in BLBC cells, EHMT2 associated with Snail and DNMT to suppress the expression of fructose-1,6-biphosphatase (FBP1). The decrease in FBP1 had pleotropic effects, including increased glycolysis, reduced mitochondrial complex I activity contributing to lower oxygen consumption, increased lactate production and reactive oxygen species generation that collectively contributed to breast tumorigenicity [109]. Similarly in HCC, dual inhibition of EHMT2 and DNMT by CM-272 impaired metabolic adaptation of cancer cells to hypoxic conditions by rewiring glucose metabolism (increased gluconeogenesis and decreased glycolysis) and reactivating the expression of FBP1 through reduced promoter methylation [49].
The serine-glycine biosynthesis pathway is involved in cancer cell proliferation and survival [110,111]. In response to serine levels, EHMT2-mediated H3K9me1 upregulated the expression of multiple enzymes involved in serine synthesis pathway including phosphoglycerate dehydrogenase (PHGDH), phosphoserine aminotransferase 1 (PSAT1), phosphoserine phosphatase (PSPH) and serine hydroxy methyltransferase 2 (SHMT2). Inhibition of the same caused nutrient deprivation leading to autophagy-mediated cell death in various cancer cell lines (neuroblastoma, colorectal, cervical, sympathetic nervous system, breast and bone cancer) [112]. Furthermore, EHMT2-dependent overexpression of PSAT1 upregulated the citric acid cycle (TCA) through increased production of a-ketoglutarate (a-KG) in CRC [113].
Abnormal levels of TCA intermediates influence pluripotency and metastasis leading to cancer progression. In nasopharyngeal carcinoma, lymphocytespecific helicase (LSH) in combination with EHMT2 suppressed fumarate hydratase (FH) expression. Reduced FH levels increased the ratio of a-KG to fumarate, resulting in metastasis by upregulation of NF-jB signalling and EMT gene expression [117].
Reprogramming of iron metabolism facilitates cancer cell survival and metastasis [118,119]. EHMT2 in complex with HDAC1 and YY1 can alter iron metabolism by transcriptionally repressing the expression of iron homeostasis regulator ferroxidase hephasetin (HEPH), thereby leading to accumulation of cellular iron in breast cancer cells [120]. Also, suppression of EHMT2-induced expression of TP53induced glycolysis and apoptosis regulator (TIGAR) and pyruvate kinase M2 (PKM2) alters ROS and autophagy [121].

Conclusion
In the last few decades, major progress has been made in the management of cancers, which includes early detection and treatment of disease. Thus, mortality has declined for some common epithelial malignancies such as breast cancer and prostate cancer. Despite such advances, metastasis, drug resistance and recurrence pose a major challenge in the effective treatment of high-risk cancers. There is growing evidence for epigenetic dysregulation contributing to EMT and drug resistance. Given that these alterations are reversible, targeting epigenetic modulators that mediate these changes, in combination with standard treatment of care, is gaining momentum.
Manifestation of EMT or a partial EMT programme (expression of both epithelial and mesenchymal markers) is different in distinct cancer types. Indeed, modulation of any EMT-TF can affect the process as they are functionally non-redundant. Given the plasticity of cells undergoing EMT, in addition to alterations associated with morphology, EMT is also associated with stem cell properties, as well as proliferation and survival of cancer cells.
Collectively, studies suggest a strong association between EHMT1/2 with EMT dynamics that has an impact on all the above-mentioned phenotypes (Fig. 4). However, there are many questions which remain to be addressed. For instance, EHMT2 either directly associates with EMT-TFs or modulates their expression to facilitate EMT. However, it remains unclear how EHMT2 is recruited and chooses its interacting partners to regulate different signalling pathways during multiple stages of cancer development. Moreover, apart from EMT-related genes, transcriptomic analysis in several cancers reveal that EHMT2 regulates a wide spectrum of genes associated with metastasis (Table 1). Further investigations are needed to delineate these regulatory networks and their functions in metastasis. Second, the mechanisms/molecular pathways through which EHMT1/2 promote cellular plasticity and drug resistance are not fully understood. Moreover, not all cancers which metastasise exhibit EMT. Therefore, understanding other pathways that lead to metastasis are also of importance. Third, while the association of EHMT1/2 with EMT is established, their role in mesenchymal-to-epithelial transition (MET) which is needed for distant organ colonisation of tumours remains to be defined. Fourth, EHMT2 provides a survival advantage to cancer cells by meeting energy demands and resisting cell death. The advancements in high-throughput sequencing have shed light on gene signatures and pathways affected by EHMT2 in tissue and cancer subtype-specific manner. Further mechanistic studies would shed insights into mechanisms by which EHMT1/2 link metabolism with CSC maintenance. While some insights have been gleaned into mechanisms by which EHMT1 and EHMT2 are overexpressed in distinct cancers, the upstream regulators are largely uncharacterised. Identifying these regulatory pathways may provide relevant therapeutic targets. Moreover, as EHMT1 exhibits functions independent of EHMT2, further characterisation of its targets in cancer progression would provide avenues for selectively targeting its network. Depending upon the cancer type, EHMT2 can either positively or negatively regulate metastasis in a methylation-dependent or -independent manner. These differences in its function have to be considered while using EHMT2 inhibitors to prevent metastasis. In this context, the development of highly selective EHMT1 and EHMT2 inhibitors and degraders is important given their distinct cancer-type-dependent functions. Fig. 4. Association of EHMT2 in metabolic rewiring, EMT induction, cancer stemness and drug resistance. EHMT2 influences metabolic reprogramming and affects the expression of EMT inducing transcription factors/markers. Cells undergoing EMT exhibit plasticity leading to metastasis and stemness. Cancer stemness attenuates the efficacy of therapies.