1.5 Epigenetic Regulation
HERV expression is regulated like every other gene within the human genome only with more stringency due to its repetitive nature. Beyond the role of repressors, cells regulate HERV expression through epigenetics. We will start this section with an overview of important epigenetic mechanisms before exploring their involvement in illness – focusing on cancer -, and HERV regulation.
Epigenetics received its name in 1942 by Conrad Waddington who was interested in the mechanism that connected genetic code and adult phenotype (73). Since then the field has evolved significantly from its definition to the inclusion of several mechanisms that collaborate to regulate genetic expression. The modern definition of epigenetics refers to inheritable differences from cellular or parental origin in cellular phenotype due to changes in gene regulation that do not affect Watson-Crick base pairing. Time has shown that epigenetics is critical in normal development – such as pluripotency and pregnancy – and plays a large role in diseases – such as depression and various cancers (66, 67, 71). Inheritance of epigenetic modifications – in particular, DNA methylation on CpG dinucleotides – are strongly conserved and are maintained over several generations. For example, inheritance of methylated and unmethylated regions of DNA can be propagated upwards of 100 cell divisions in vitro (67, 74).
Epigenetic modifications regulate gene expression by inducing conformational changes to chromatin structure that allow or restrict access of transcriptional machinery through many complex mechanisms. The most well studied mechanisms involve DNA methylation of CpG dinucleotides and histone post-translational modifications with nucleosome rearrangement and RNA-mediated gene silencing being relatively new to the field (66, 67, 69, Figure 1.5.1). It should be emphasized that none of these modifications act in isolation: each modification relies on other modifications to create a phenotype. Furthermore, while there are general understandings of how these modifications work there are some exceptions. Both exceptions and examples of epigenetic mechanisms working collaboratively will be indicated throughout this section.
DNA methylation is primarily used to silence genes by restricting transcription machinery accessibility. It results from the covalent addition of a methyl group to the 5-carbon position of cytosine in CpG dinucleotides by one of three DNA methyltransferases (DNMTs): DNMT1, DNMT3a, and DNMT3b (67). DNMT1 is responsible for “maintenance” methylation by targeting regions of hemimethylation to add methyl groups to the unmethylated daughter strand during replication while DNMT3a and 3b are responsible for de novo methylation during early development (67).
DNA methylation plays an important role in cellular functions, such as the silencing of transposable elements (71, 75-79), embryonic development (80), imprinting (81), and X-chromosome inactivation (82) (67, 69). Typically, regions that are heavily methylated are low in transcriptional activity while those that are low in methylated CpGs are high in transcriptional activity (71). As a result, methylated CpG dinucleotides are not found in promoter regions of genes active in healthy cells (67). This decrease in transcriptional activity is thought to be due to blockage of transcriptional machinery from binding to DNA promoters and due to recruitment of histone modifying enzymes that silence portions of chromatin (66, 67). However, there are instances where hypermethylation over a repressor binding site, hypermethylation on promoters for repressors, and transcription factors capable of binding to methylated sites can allow for increased expression of a nearby gene product (69, 70).
Histone post-translational modifications are more diverse and can have a wide range of results on chromatin structure (66, 68, Figure 1.5.2). Histone proteins – H2A, H2B, H3, and H4 – assemble into a dimeric octomer and wind DNA into a spool-like structure which is secured by the H1 histone protein to form a nucleosome – the basic unit of DNA packaging (Figure 1.5.2). Each protein within the dimeric octomer extends a
tail into the nuclear lumen which can be modified by a diverse set of enzymes (69, Figure 1.5.2). Known modifications consist of acetylation, methylation, phosphorylation, and ubiquitination which affect transcriptional machinery access to gene promoters through structural modifications of chromatin (Figure 1.5.2). For example, histone acetylation of H3K9 or H3K14 results in gene activation through relaxation of chromatin structure to allow transcriptional machinery access to DNA (67, 71).
Acetylation of histones are straight forward in that once acetylated the neighboring genomic landscape relaxes around the histone to allow for transcriptional machinery binding. In this respect, acetylation is commonly referred to as an activating modification. However, once you start to consider other histone modifications such as methylation things get more complicated (69). In the case of histone methylation, these modifications can be either repressive or permissive and the modified sites can be mono, di, or tri-methylated (69). For example, tri-methylation of H3K27 or H3K9 is associated with gene silencing while trimethylation of H3K4 and H3K36 are associated with gene activation (69). This brings up an interesting point where residues subject to more than one modification can have two different results. For example, acetylation of H3K9 results in activation while methylation of the same residue results in inactivation.
There are two other mechanisms of epigenetic modification that we would like to mention but to go into detail would go beyond the scope of this dissertation: non-coding RNAs and nucleosomal rearrangement. Some genes when transcribed produce short RNA transcripts that will bind to DNA or protein to prevent the expression of gene products (66). Non-coding RNAs can be segmented into small (under 200 nucleotides) and large (> 200 nucleotides), but are assigned into subgroups such as microRNAs (miRNAs), long non-coding RNAs (lncRNAS), small nucleolar RNAS (snoRNA), PIWI-interacting RNA (piRNA), and small interfering RNA (siRNA) (69, 71). Each of these non-coding RNAs have been increasingly recognized as being vital to normal cellular function and have been found to be aberrantly expressed in cancer (69, 71). For example, there is an inverse correlation between expression of the histone methyltransferase EZH2 and expression of miRNA-101. In breast cancer cell lines and prostate cancer tumors, and overexpression of EZH2 is correlated with increased histone methylation and a decrease in miRNA-101 expression (66, 83). Finally, nucleosomal rearrangement can block or permit access of transcriptional machinery to DNA depending on their placement throughout the genome.
To get effective regulation of the genome, all of these modifications work together (66, 67, Figure 1.5.3, Figure 1.5.4). For example, silencing a gene through CpG methylation of DNA can recruit binding of methylation binding proteins (MBPs) which can trigger the recruitment of histone deacetylases (HDACs) (66, 67, Figure 1.5.3). This recruitment will remove acetyl groups on histones, triggering a conformational change in chromatin, restricting transcriptional machinery access to DNA. Furthermore, DNMTs can be recruited to DNA by histone modifying proteins like heterochromatin protein 1 (HP1) to silence genes (Figure 1.5.3).
As already alluded to, when epigenetic mechanisms go awry disease immerges (66, 67). Epigenetic dysregulation has been reported in mood disorders (84, 85), arthritis (86), asthma (87), and especially in cancer (66). Epigenetic dysregulation in cancer typically focuses on aberrant global DNA methylation and histone acetylation, yet as previously mentioned other mechanisms such as non-coding RNAs also play a role (66, 67, Figure 1.5.5). For example, hypermethylation of MLH1 – which encodes a DNA mismatch repair enzyme – predisposes multiple generations to colon cancer due to its inactivation (66, 88). Furthermore, aberrant DNA methylation is characteristic of leukemias and lymphomas, and genome-wide screens of such cancers revealed several genes that exhibited differential DNA methylation compared to their non-cancerous counterparts (66, 89).
Global hypomethylation of cancer relative to their non-cancerous counterparts was first established in 1983 (90), and since then genome-wide DNA methylation dysregulation (i.e. genome-wide DNA hypomethylation with site-specific DNA hypermethylation) has been reported in several cancers (71, 91-96). Site-specific DNA hypermethylation typically involves CpG islands in promoter regions and results in transcription inactivation (71). This type of modification affects genes within major cellular pathways such as DNA repair, cell cycle control, Ras signaling, apoptosis, detoxification, hormone response, and p53 network, resulting in a greater growth advantage (71).
As for histone modifications in cancer, one of the most well explored is changes in acetylation. In various cancer types, such as prostate, colon, lung, liver, and stomach HDACs are over-expressed resulting in loss of acetylation and a decrease in local gene expression (71, 97-102). Furthermore, aberrant expression of histone methyltransferases (HMTs) and histone demethylases (HDMs) have been reported in various cancer types (71). For example, inactivating mutations in the histone methyltransferase SETD2 and the histone demethylase UTX and JARID1C were reported in clear cell renal carcinoma (71, 103). Finally, EZH2 over-expression in melanoma, prostate, breast, and endometrial cancer are correlated with malignancy, and knockdown of EZH2 in ovarian cancer induced apoptosis and suppressed invasion (71, 104, 105).
HERVs – receiving treatment akin to protein coding genes and being transposable elements – are naturally subject to epigenetic regulation. Through reinfection of the host organism, transposable elements can create genomic instability by spreading throughout the host genome (72). Interestingly, DNA hypomethylation in tumor cells also affects repetitive regions of the genome which results in genome instability (71). In the case of retroviruses, LTRs – which contain a surplus of transcription factor binding sites – are problematic for host cells because they can promote the expression of both ERVs and downstream sequences (72). The later can lead to neoplasia in infected cells due to expression of proto-oncogenes.
It is important to note that both proviral LTRs and solo LTRs can serve as promoters for downstream sequences (72). It is also important to note that these LTRs can still function as promoters when present in sense and antisense orientation to the rest of the genome (72). Therefore, placement of LTRs around proto-oncogenes or insertion of ERVs into tumor suppressor genes – intronic or exonic – can be detrimental for host health. Additionally, LTRs can affect malignancy by interfering with epigenetic regulation of non-coding RNAs. For example, one report displayed a correlation between the number of very long non-coding RNAs (vlincRNAs) expressed from ERV promoters and the severity of malignancy (106). Thus, HERVs are not only capable of affecting the expression of protein-coding genes but also the expression of lincRNAs to the detriment of the host (72, 106).
Typically, epigenetic regulation of HERVs can be achieved through DNA methylation and histone deacetylation (72). In healthy tissue, HERVs are heavily methylated which is lost in cancerous tissue (72, 107, 150). For example, HML-2 expression in melanoma cell lines is correlated with the amount of methylation present on their 5′ LTR, and their expression can be induced with treatment of 5-aza-2?-deoxycytidine (150). Furthermore, the lack of methylation of HML-2 5′ LTRs correlates with HML-2 expression in the germ cell tumor cell line Tera-1 and methylation of 5′ LTRs in vitro correlates with decreased expression (72, 108). This form of epigenetic regulation seems to be the most important in HERV regulation as treatment with HDAC inhibitors alone in HIV-infected cells and 293s did not significantly up-regulate their expression (72, 109).
Due to global methylation dysregulation in cancer, it’s possible that increased HERV expression in cancer is due to hypomethylation around HERV LTRs (72). For example, global hypomethylation of HERV-W and LINE-1 were reported in ovarian cancer and hypomethylation of HERV-K 5′ LTRs are seen in melanoma (72, 110, 111). However, upregulation of all HERVs does not seem to be the case for all cancers (72, 112). Treatment of neuroblastoma cell lines with 5′-azacytidine induced expression of HERV-W loci, indicating that there are some cancer cell lines where HERVs are suppressed by CpG methylation (72, 112).
Additionally, histone methylation or deacetylation in cancer can contribute to HERV activation (72). This was described recently in cancer cell lines where an increase in HERV-Fc1 expression was associated with active histone methylation (72, 113). Some HERV LTRs like that of ERV9 are easily induced by HDACi and are highly expressed in testes (72, 114, 115). This LTR regulates the expression of a tumor suppressor protein called GTAp63, which is absent in germ cell tumors but when cells are treated with HDACis GTAp63 expression is induced which triggers apoptosis (72, 114, 115). Interestingly, this effect was not seen with 5′-aza treatment so CpG methylation is not required for this LTR regulation (72).