Epigenetic memory in development and disease: unraveling the mechanism

Abstract

Epigenetic memory is an essential process of life that governs the inheritance of predestined functional characteristics of normal cells and the newly acquired properties of cells affected by cancer and other diseases from parental to progeny cells. Unraveling the molecular basis of epigenetic memory dictated by protein and RNA factors in conjunction with epigenetic marks that are erased and re-established during embryogenesis/development during the formation of somatic, stem and disease cells will have far reaching implications to our understanding of embryogenesis/development and various diseases including cancer. While there has been enormous progress made, there are still gaps in knowledge which includes, the identity of unique epigenetic memory factors (EMFs) and epigenome coding enzymes/co-factors/scaffolding proteins involved in the assembly of defined “epigenetic memorysomes” and the epigenome marks that constitute collections of gene specific epigenetic memories corresponding to specific cell types and physiological conditions. A better understanding of the molecular basis for epigenetic memory will play a central role in improving diagnostics and prognostics of disease states and aid the development of targeted therapeutics of complex diseases.

Introduction:

Conrad Waddington coined the term epigenetics derived from the Greek word “epigenesis”, long before genetic inheritance through the DNA was established, in an effort to find a relationship between genetics and the determinants of phenotype responsible for the developmental process from an undifferentiated state of the early embryo [1,2]. Epigenetics refers to inheritance and maintenance of gene expression patterns through cell divisions in the absence of alterations in the DNA sequence. The ability of the cells to stably retain and transmit the unique gene expression patterns to the daughter cells, referred to as epigenetic memory, is encoded by the epigenetic marks and the associated epigenetic memory factors. Overall, epigenetic memory defines not only the characteristics of germ as well as normal differentiated cells, but also the features responsible for the establishment of inherited properties in embryonic cells and their progeny as well as cells defining a disease condition. Despite the realization of the importance of epigenetic memory in defining phenotypes in a multitude of phenomenon such as germline inheritance, paramutations, genomic imprinting, X-chromosome inactivation and position effect variegation during development and disease, the molecular basis has remained elusive [3,4,5].

Epigenome defines cell function

Various epigenetic marks consisting of a collection of gene specific DNA methylation patterns, unique combinations of post-translational modifications of histones, polycomb group (PcG) proteins, transcription factors, non-coding RNAs (ncRNAs), chromatin remodeling factors, and other epigenetic memory factors form the epigenome. The distinctive epigenome defining the (epi)genomic code associated with each individual gene dictates the expression status of that gene. The epigenomic code for each gene could be unique to specific cell type or disease condition and therefore the same genome can acquire different collections of epigenomic codes to form the characteristic epigenome, which defines the overall cellular properties.

A nucleosome core particle, the basic structural unit of chromatin consists of ~146 base pairs of DNA wrapped around an octamer of histones composed of a central tetramer of H3 and H4 histones with two peripheral heterodimers of H2A and H2B histones (or specialized natural variants of these proteins). Covalent modification of DNA due to methylation of cytosine to form 5-methylcytosine (5-meC) predominantly in the context of CpG dinucleotides by DNA methyl transferases (DNMTs) is associated with transcriptional silencing [6]. Despite its importance in gene expression, approximately 40% of the genes are devoid of CpG dinucleotides and the majority of them are unmethylated under most of the differentiated states [7]. While the epigenetic changes of DNA occurs due to methylation of cytosine residues, histones become modified by acetylation and methylation of lysine (K) and arginine (R) residues, ubiquitylation and sumoylation of lysines and phosphorylation of serines (S) and threonines (T) and poly-ADP-ribosylation of glutamic acid (E) on their tails [8]. Additionally, the lysine residues are amenable to mono-, di- and tri- methylations, and the arginines can be either mono- or di-methylated, adding to the complexity of histone modifications. Up to three methyl groups per lysine residue can be progressively added to distinct states of the chromatin, resulting in both short-term and long-term imprints. Histone acetylation of H3 and H4 as well as di- or trimethylations of K4 of H3 (H3K4me2, H3K4me3), H3K36 and H3K79 methylations are associated with a transcriptionally-active state with H3K4 tri-methylation (H3K4me3) marking promoters and H3K36 and H3K79 methylations occurring primarily over gene bodies and mono-methylation of H3K4 is unique to enhancers. On the other hand, trimethylations of H3K9 (H3K9me3) or H4K20 (H4K20me3) form heterochromatin and dimethylation of H3K9 (H3K9me2) and trimethylation of H3K27 (H3K27me3) in the euchromatic region correspond to a repressed state. H3K27me3 in general marks dynamically regulated genes and easily reversible enabling switched on and off states in response to developmental signals. In addition to histone modifications, specific histone variants could also define functional status of chromatin and be enriched in transcriptionally active (e.g., H3.3) and inactive (e.g., H3.1) chromatin, while others may be neutral to the transcriptional status (e.g., H3.2) but indicate other roles [9]. Overall, these histone marks have been found to correspond to the functional status of the chromatin/cells [10–13] (Table1). Furthermore, an oxidized form of 5-mC, 5-hydroxymethylcytosine (5-hmC) has also gained lot of attention as an epigenetic mark which regulates chromatin modifications and gene transcription during developmental stages and cellular differentiation as well as in several cancers [14].

Table 1.

Major histone marks and their association with transcription or other roles.

Histone mark	Position	Transcription	Enzymes/Factors*
macroH2A	Compact chromatin, X chromosome	Repressed	ATRX
H2A.X	Double-strand DNA breaks		ATM, ATR, DNA-PK
H2A.Z (H2AZ1)	Promoters/transcription start sites	Active	INO80, SWR1
H2BK5Ac	Promoters
H2BK120Ac	CpG island promoters
cenH3 (CENP-A)	Centromere		HJURP
H3.3	Promoters	Active	HIRA,DAXX/ATRX
H3K4Ac	Enhancers
H3K4Me1	Promoters and enhancers		MLL, SETD1A, SETD1B, SETD7
H3K4Me2	Transcription start sites, CpG islands, promoters and enhancers		MLL, SETD1A, SETD1B, SETD7
H3K4Me3	CpG islands, promoters and enhancers	Active	MLL, SETD1A, SETD1B, SETD7
H3K9Ac	Coding regions	Active	KAT2A, KAT2B, p300, CBP
H3K9Me3	Promoters and enhancers, heterochromatin	Repressed	SUV39H1, G9A, GLP, SEDB1
H3K14Ac	CpG islands, promoters and enhancers	Active	KAT2A, KAT2B, p300, CBP
H3K27Ac	Coding regions	Active	KAT2A, KAT2B, p300, CBP
H3K27Me3	Coding regions, heterochromatin	Repressed	PRC2-EZH1/EZH2
H3K36Me3	Coding regions	Active	SETD2, ASH1L, NSD1
H3K79Me1	Coding regions	Active	DOT1L
H3K79Me2	Coding regions	Active	DOT1L
H4K16Ac	Euchromatin	Active	MOF
H4K20Me3	Heterochromatin	Repressed	SUV4-2OH1, SUV4-2OH2

^*- Catalytic enzymes/factors associated with the indicated histone marks

PcG proteins are grouped into multi-protein complexes known as the Polycomb repressive complexes (PRCs) such as the PRC2. Enhancer of zeste [E(Z)] homolog 2 (EZH2), suppressor of zeste 12 [SU(Z)12] and embryonic ectoderm development (EED) form the core of PRC2 and trimethylated H3 lysine 27 (H3K27me3) and to a lesser extent lysine 9 (H3K9me3) at target gene promoters, apparently initiates silencing by blocking access to chromatin remodeling proteins such as the SWI/SNF complex [15]. EZH2 is also found to recruit DNMTs to promote DNA methylation leading to the repression of transcription. There is also evidence for recruitment of histone methyl transferases such as SUV39H1 by MeCP2 to target methylated DNA leading to H3K9 methylation and gene silencing [16]. These studies suggest that DNA methylation can occur either downstream or upstream of histone methylation in mediating gene silencing.

Active transcription of target genes through the recruitment of transcription factors (TFs) to active promoters could specify the functional status of the cell. For example, in pluripotent ES cells, TFs Oct3/4, Nanog and Sox2 are associated with transcription activating marks including, H3 and H4 acetylation and H3K4me3. On the other hand, in these ES cells, several lineage specific genes such as Sox1, Nkx2-2, Pax3 and Msx1 are marked by “bivalent” histone modifications consisting of both H3K4me3 and H3K27me3 and remain repressed in potentially active promoters and may become fully activated upon receiving the signals for the differentiation program or continue to remain silenced upon the removal of activating marks [17–19].

The role of histone variants in the establishment of epigenetic memory

Our knowledge is very limited as to how multiple regulatory mechanisms including transcriptional factors and other epigenetic memory factors in conjunction with unique histone modifications, non-coding RNAs and DNA methylation patterns are coordinated to control the “on” and “off’ states of pluripotent, developmental or the disease conditions. Here are some possibilities that are worthy of consideration while we wait to learn more from the future studies.

Histone variants known as replacement histones differ from the canonical histones by one or more amino acids in sequence, are associated with specific chromosomal loci defining their functional status and tissue specificity [20,21]. Two canonical histone 3 variants [i.e., H3.1 and H3.2 which only differ by one amino acid at position 96 (a Cys-Ser substitution) and H3.3 which differs from H3.1 at five amino acid positions (i.e., 31, 87, 89, 90 and 96 with Ala-Ser, Ser-Ala, Val-Ile, Met-Gly and Cys-Ser substitutions respectively)] may define a different functional status of the chromatin [9]. H3.3 is also known to have additional variants such as the centromere specific, CenH3s and testis specific, H3t and may play specific roles in programming and reprogramming the functional status of the associated genome. While the histone variant, H3.3 is generally recruited to the highly expressed loci, H3.1 and H3.2 could be associated with both active and silenced genes. It is also likely that replacement of protamines and other basic nuclear proteins preferentially by H3.3 in an organized manner upon the entry of sperm into the oocyte, may determine the establishment of epigenetic marks on the paternal chromatin. On the other hand, macroH2A present in the mature oocyte, which is known to inhibit chromatin remodeling and associated with resistance to reprogramming, is lost upon fertilization and may be the obligatory step permitting epigenetic reprogramming of the maternal chromatin. Thus, upon receiving internal and external signals, the differential association of histone variants may mark the silenced imprinted genes in the maternal and paternal chromatin. Furthermore, similar to the histone chaperones such as Chd1 and HIRA that interact with H3.3 and Chtz1/SWR1 that complex with H2A.Z and deposit them at the appropriate chromosomal loci, there may be other factors involved in specifying imprinted regions of the chromosome [22,23]. There is more work required to establish the associations between histone variants and chaperones under the influence of specific signaling molecules/events to understand their contribution to epigenetic memory [9].

It is also noteworthy that despite the predominant epigenetic modifications of histones or histone variants associated with a specific loci of the genome may determine if a gene is active or repressed, the factors required to mediate the opposite of the apparent functional status may already be in place for the gene of interest enabling the reversion of the functional state. In other words, the genome remains poised for either of the states and requires the internal/external signal to initiate the reversion of state. For example, the Polycomb group (PcG) and Trithorax group (TrxG) protein complexes either repress or promote transcription respectively by acting as epigenetic regulators in a physiological status dependent manner [24].

Maintenance of imprinted status of specific loci

Genomic imprinting is a process that defines gene expression according to the parental origin of alleles in plants and mammals [25,26]. This asymmetry associated with specific loci is regulated by epigenetic marks such as differentially methylated regions (DMR) of DNA, chromatin modifications and other factors that differ on the two parental chromosomes. While histone modifications associated with repressed loci in the imprinted regions consisted predominantly of H3K9me3 and H4K20me3, it is often substituted with H3K4me3 and H3K27me3 in the developmentally regulated regions [27]. During differentiation of primodial germ cells (PGC), genomic demethylation first erases the methylation marks and after an interval of few days, a wave of DNA methylation marks allele-specific and imprinted regions during oogenesis and spermatogenesis on the maternal and paternal chromosomes in the germline known as the germline imprints. The majority of the affected genes are localized in clusters regulated by a cis-acting imprinting control region (ICR) where cycles of differential DNA methylation and establishment of chromatin marks occur. These imprints are stably inherited from the germline to offspring resulting in monoallelic expression which continues throughout the lifetime of the organism [28,29].

While the mechanism for active demethylation has been believed to occur mostly as a passive process due the failure of maintenance methylation following DNA replication for a long time, recent studies suggested that it could be an active replication-independent process [30,31]. Active demethylation is mediated by 5-methylcytosine hydroxylases TET1, TET2 and TET3 (TET refers to Ten-Eleven-Translocation); which convert 5-methylcytosine (5-mC) into 5-hydroxymethylcytosine (5-hmC) via oxidation [31]. Other evidence point to fluctuating levels of DNMT1 and/or de novo methyltransferases in different locations that aid the epigenetic reprogramming of the developing embryo [32,33].

While it is generally believed that demethylation to re-set the epigenetic marks occurs soon after fertilization, it is apparently not evenly distributed throughout the genome. Interestingly, in general a delayed demethylation of the maternal genome has been reported by several studies. It is also known that there may be retention of imprints in a male and female specific manner during this process. H3K4 demethylase (KDM1B) was shown to be involved in establishing DNA methylation imprints for maternal imprinting control regions (ICRs) but not the paternally methylated ICRs [34]. Furthermore, the maintenance of methylated ICRs seem to depend on additional factors. For example, PGC7/Stella is known to protect the maternal genome from demethylation to maintain methylation of several imprinted genes, thus playing a role in epigenetic reprogramming after fertilization [35,36]. Other studies indicate that Zfp57, a presumed KRAB zinc finger protein, maintains both paternal and maternal methylation imprints after fertilization at multiple imprinted regions [31]. KRAB zinc finger proteins belong to one of the largest transcription factor families and functions to recruit factors associated with repressive chromatin, including DNA methyltransferases, histone deacetylases and histone methyltransferases [37].

X-chromatin inactivation

X chromosome inactivation (XCI) ensures gene expression occurs from only one X chromosome in females [38,39]. Although the maternal X chromosome (Xm) remains active throughout preimplantation development, the paternal X chromosome (Xp) is epigenetically silenced during preimplantation to morula stage and maintained in trophoectoderm derivatives at the blastocyst stage but reactivated again in the inner cell mass (ICM). However, following implantation, either the Xp or Xm in the ICM can be stochastically chosen for inactivation [40,41]. In conjunction with DNA methylation, repressive chromatin modifications such as H3K27me3 and incorporation of macrohistones (i.e., MacroH2A) resulting in heterochromatin formation, a large cis-acting non-coding RNA corresponding to the Xist (X inactive specific transcript) gene expressed from the X inactivation center (XIC) only from the inactive X chromosome (Xi) are involved in silencing transcription. On the other hand, the Xist gene on the active X chromosome is silenced by DNA methylation. The antisense transcript expressed from XIC, Tsix (reverse spelling of Xist), acts as a repressor of Xist [39]. While the maintenance of XCI is also believed to depend on the association with PcG proteins, SmcHD1 and RNF12, their roles remain unclear [39,42]. On the other hand, the pluripotency factors, OCT4/Pou5f1, SOX2 and Nanog apparently act as the XCI-inhibitors [43,44]. Additionally, other XCI-activators and XCI-inhibitors are likely to be involved and the elucidation of their identity and interplay among these factors will be an important goal for future studies.

Epigenetic landscape associated with stem cell properties

Pluripotency and stem cell identity are derived from a complex network that consists primarily of the master regulatory transcription factors Oct4, Sox2, and Nanog, which is thought to have a feedback mechanism linked to several cofactor protein interaction networks [45]. Stem cells can self-renew and either exhibit the ability to differentiate into any cell type of the body (i.e., pluripotent) such as the embryonic stem cells (ESCs) derived from the ICM of the blastocyst stage of the embryo or harbor restricted potential to become cell types of only the parental organ (i.e., multipotent) such as the adult stem cells [e.g., mesenchymal stem cells (MSCs) that could form adipocytes, chondrocytes, osteocytes etc.]. Thus, these stem cells would display distinct epigenetic signatures consisting of DMRs, histone modifications and association with transcription factors. The Polycomb group (PcG) and Trithorax group (TrxG) protein complexes either silence or promote transcription respectively by regulating post-translational modification of histones [46–48]. The evidence for the existence of epigenetic cellular memory could be derived from bivalency due to PcG mediated silencing associated with H3K27me3 and TrxG-dependent activation marked by the presence of H3K4me3 in the same loci and cooccupation by the regulatory transcription factors such as Oct4 and Sox2. In addition to the landscape and the cooccupying transcription factors, the differential DNA methylation patterns could also be involved in this process. These modifications and associations could act as the epigenetic gatekeepers that are in a standby mode for regulation of genes involved in pluripotency and development as they could remain poised for either repression or induction based on the stochastic extrinsic and intrinsic signals. Several factors involved in chromatin modifications and remodeling such as the PcG or the TrxG proteins and Brg1 as well as transcription factors involved in pluripotency such as Oct3/4 and Sox2 are maternally inherited and likely to be associated with specific DNA sequences and therefore could enable the program of events to initiate and progress in an instantaneous manner [17,49].

The successful reprogramming of differentiated cells to generate induced pluripotent stem cells (iPSCs) has been achieved with a minimum of four virally transduced transcription factors, Oct4, Sox2, c-Myc and Klf4 in mouse fibroblasts as well as with Oct4, Sox2, Nanog, and Lin28 in human somatic cells [50–52]. In addition to transcription factors, other epigenome modulators including the large intergenic non-coding RNAs (lincRNAs) such as lincRNA-RoR or miRNAs such as miR-294 have also been implicated promoting pluripotency [53–55].

Recent progress made in converting differentiated somatic cells to pluripotent cells through dedifferentiation has opened up the potential for replacing one’s affected cells with his/her own cells by reprogramming the epigenetic memories that have become fixed through the development and differentiation process [56–58] (Figure 1). In the same vein, it is not difficult to envision, how the dedifferentiation process resulting from loss of epigenetic memories could generate the notorious cancer stem cells that can serve as reservoirs for the recurrence of the disease (Figure 1). While there is data supporting similarities in the profiles of epigenetic marks such as DNA methylation patterns and histone modifications along with gene expression between ESCs and iPSCs, which supports that enforcing the loss of epigenetic memories can revert somatic cells to embryonic state, there is overwhelming data to suggest that residual epigenetic memories remain in the iPSCs [59–63]. The degrees of these remaining residual epigenetic memories are apparently dependent upon their cells of origin and the experimental approaches used to induce pluripotency. Therefore, addressing these residual epigenetic memories will be a major challenge for the researchers to make progress towards developing stem cell based therapeutics.

Figure 1. Dedifferentiation and reprogramming of epigenetic memory defines cellular properties.

One differentiated somatic cell (e.g somatic cell I) could lose their epigenetic memories responsible for the differentiated state to become pluripotent which could be subsequently reprogrammed into a different somatic cell (e.g somatic cell II) and vice versa. On the other hand, it may also be possible to directly convert one differentiated cell into the other in a process of transdifferentiation without going through a pluripotent state [50,51]. These processes if they can be experimentally achieved, would serve as the basis for stem cell therapeutics by replacing one’s affected cells with his or her own reprogrammed somatic cells. In a related point, the loss of epigenetic memory and dedifferentiation process of somatic cells and reprogramming of new epigenetic memories occur during cancer progression and generate the notorious cancer stem cells that serve as reservoirs for the recurrence of the disease and/or the cancer cells that go on to metastasize at distal sites.

Establishment and maintenance of epigenetic memory by signaling events

The current state of knowledge suggest that epigenetic marks are heritable and could consist of DMRs, specific DNA sequence fingerprints determined chromatin landscapes and associated epigenetic memory factors (EMFs) which include transcription activating and repressive protein factors and ncRNAs that mediate the conversion and maintenance of epigenetic marks that are poised for becoming the functional unit. The known and still to be recognized/identified epigenetic memory factors (EMFs) involved in the various steps of the processes may play different roles as epigenetic memory erasing factors (EMEFs), epigenetic memory protecting factors (EMPFs), epigenetic memory initiating factors (EMIFs), epigenetic memory decoding factors (EMDFs), epigenetic memory mediating factors (EMMFs) and epigenetic memory fixing factors (EMFFs). It is conceivable that these EMFs are essential factors which constitute unique “epigenome memorysomes” that may consist of epigenome coding enzymes/co-factors/ncRNAs/scaffolding proteins which are unique to performing only a specific function at the specific differentiated/dedifferentiated stage during the establishment of the epigenetic memory and may play multiple roles including the establishment of epigenetic marks and elicit their functionality depending upon the physiological context of the cell. In summary, the intrinsic and extrinsic signals that are generated within the cell or in its microenvironment are likely to serve as the triggers to initiate a cascade of events that will convert the “ready state” epigenetic marks into the active epigenetic marks defining the properties and functional status of the cell at the various stages of development (Figure 2). While the epigenetic memory initiating factors (EMIFs), epigenetic memory decoding factors (EMDFs), epigenetic memory mediating factors (EMMFs) and epigenetic memory fixing factors (EMFFs) may become active in a sequential manner at the different stages of development (Figure 2), the epigenetic memory erasing factors (EMEFs) and epigenetic memory protecting factors (EMPFs) are likely to prepare the genome whenever there is a need for removing and preserving the epigenetic marks, respectively to suit the developmental process. These latter classes of factors (i.e., EMEFs and EMPFs) are likely to be highly active especially during gametogenesis, fertilization and the early embryonic stages.

Figure 2. Establishment of epigenetic memory during development and differentiation.

The process of establishing a collection of gene specific epigenetic memories that define a unique differentiated somatic cell could occur in multiple steps. Step I: The trigger of the process could manifest in the form of extrinsic and intrinsic signals such as the cytokines or their downstream effectors that act as epigenetic memory initiating factors (EMIFs); Step II: The EMIFs could activate the epigenetic memory decoding factors (EMDFs) that mediate a cascade of events that would decode the active or inactive status of the epigenetic marks that are poised for initiating the developmental/ differentiation process; Step III: Activation of an epigenetic program that creates the landscape of epigenetic marks required for establishing the unique associations. This may involve epigenetic memory mediating factors (EMMFs); Step IV: Establishment of gene specific epigenetic marks which may involve epigenetic memory fixing factors (EMFFs). Step V: With the repetition of steps I-IV occurring in a combinatorial manner, the unique epigenome of the differentiating cell will be established. While most of the epigenetic marks are removed by epigenetic memory erasing factors (EMEFs) to prepare the genome for re-programming during gametogenesis, fertilization and the early embryonic stages, some epigenetic marks continue to remain protected throughout these processes by the epigenetic memory protecting factors (EMPFs).

Because the plasticity of epigenome can be readily assessed during cancer progression, it can be effectively exploited as an experimental system to obtain a greater understanding of the various mechanisms involved in epigenetic dysregulation of genes involved in cellular integrity as well as the processes involved in the conversion of a normal cell to a cancer cell (Figure 3). While the exact signals that trigger the specific epigenetic changes remain elusive, studies have shown that signaling mediators may be involved in regulating epigenetically silenced state of genes in cell line model systems [64–66]. While, these and other studies have implicated that oncogenic Ras could turn on multiple pathways and mediate repression of tumor suppressor genes due to DNA methylation through regulation of DNMTs, the molecular details remain elusive [67]. One study showed that disabling Ras signaling could cause loss of promoter DNA methylation silencing of downstream mediators such as Fas and other candidate genes (i.e., Sfrp1, Par4, Plagl1 and H2-K1) in the ras - transformed NIH 3T3 mouse embryo fibroblast cell line. This study also predicted that association of DNMT1 to the respective promoters in these cells is likely to be responsible for promoter methylation of the target genes [64]. A different study that examined the role of oncogenic Ras found that it is involved in activating H3K27 demethylase JMJD3 and in down-regulating the methylatransferase EZH2 to activate p16^INK4a in both human and mouse fibroblast cell lines [59]. There is also accumulating evidence for Myc, a known downstream target of Wnt signaling, to be involved in widespread chromatin remodeling to regulate gene expression [68–71]. Furthermore, there is also evidence for coupling of cell proliferation to post-translational modification of EZH2 due to phosphorylation at Thr 350 by cyclin dependent kinase 1 (CDK1) and cyclin dependent kinase 2 (CDK2) which activates EZH2 recruitment to promote H3 lysine 27 trimethylation (H3K27me3) leading to silencing of target genes [72].

Figure 3. The role of epigenetic memory in regulating gene expression patterns in cancer.

Differential gene expression patterns define the functionality of the cancer cells which is dependent upon the abundance and assembly of transcription factors (TFs) in a highly orchestrated manner. Dysregulation of various epigenetic regulators (e.g., miRNAs, DNMTs, PcG proteins etc.) could affect the levels of TFs and in turn cause an imbalance in the essential gene products (e.g., adhesion molecules, polarity factors, cytoskeletal proteins, angiogenic factors, metastatic factors etc.) that help to maintain integrity and structure of the cellular components leading to the conversion of a normal cell to a cancer cell.

In an effort to find signals that may initiate and maintain epigenetic memory during cancer progression, we serendipitously observed that disruption of Smad signaling due to Smad7 overexpression or depletion of Smad2, resulted in DNA demethylation associated re-expression of a subset of corresponding genes [73]. With the use of a ras - transformed breast cancer cell line model system (MCF10A), we found that sustained TGFβ signaling is required to maintain the DNA methylation mediated silencing of a subset of genes (e.g., ABCG2, CLDN4, CLDN7, CDH1(E-CAD), CGN, DEFB, KLK10 and MUC1) that promote epithelial to mesenchymal transition (EMT). Interestingly, disruption of the TGFβ-Smad signaling pathway resulted in the reversal of the DNA methylation mediated silenced state of these genes and hence EMT suggesting that it directly regulated epigenetic alteration of genes required for epigenetic memory to favor breast cancer progression [73]. Furthermore, deregulation of signaling pathways has been shown to affect the development of many diseases including cancer due to regulation of histone proteins, and histone- or DNA-modifying enzymes [74]. By extension, it is likely that the presence or emergence of signaling molecules in a stage specific manner during development as well as in disease progression may regulate downstream pathways involved in establishing unique epigenetic programs that become fixed to define epigenetic memory to elicit the phenotypic and functional properties of the cell. Despite it remains as a challenge to identify specific signaling pathway mediators to epigenetic memory, it is noteworthy that there has been some success in identifying chemical cocktails responsible for maintaining features of specific cell types such as the CD34 positive cells by activating HOXA9, GATA2 and AKT-cAMP signaling pathways [75].

Epigenetic memory in genetic and epigenetic diseases

It has become increasingly clear that defects in epigenetic factors and epigenetic changes could be either causative or act as modifiers in various diseases including cancer (Figure 4). The number of diseases that can be somehow attributable to epigenetic deregulation has been climbing steadily in recent years due to high throughput tools available for the epigenetics research. Several of the epigenetic diseases present early in life and are associated with mental disorders due to aberrations in epigenetic marks which are often associated with defects in genetic integrity. For example, Fragile X syndrome resulting from silencing of FMR1 due to hypermethylation of CGG trinucleotide repeat in the 5’ untranslated region and those resulting from either paternal imprinting [e.g., Angelman syndrome (AS) arising from loss of maternally expressed ubiquitin protein ligase E3A (UBE3A) due to deletion of 15q11-q13 locus where the normal allele is paternally inactive due to imprinting] or maternal imprinting [e.g., Prader-Willie syndrome (PWS) arising from loss of paternally expressed genes due to microdeletions of the 3’ end of ICR localized to 15q11-q13 locus consisting of SNURF/SNRPN or uniparental disomy (UPD)of a maternal chromosome that harbors inactive 3’ end of ICR localized to 15q11-q13 locus due to imprinting] [76].

Figure 4. Molecular basis of disease due to aberrant genetic/epigenetic alterations.

The current state of the knowledge is consistent with the notion that despite an identical genomic content in the form of DNA sequence in every cell of the body, the various epigenetic modifications that are established, and in many cases fixed as epigenetic memory at each stage of development are essential for proper development and differentiation allowing the cells to assume unique properties and functionalities in the fully formed normal human body. It is also becoming increasingly clear that epigenetics provide an explanation to not only non-genetic diseases but also to genetic defects that do not always result in identical symptoms due to variations in penetrance. On a different note, the reversal of the differentiation process known as dedifferentiation may not involve all of the accumulated epigenetic memories and in some cases epigenetic reprogramming could cause various diseases, generate cancer cells or even pluripotent stem cells or cancer stem cells. While the goal of stem cell therapeutics is to reprogram the adult somatic cells into the cells of our choice to replace dysfunctional cells that have accrued over a period of time, it is also noteworthy that during cancer progression, there is not only dedifferentiation but also reprogramming to evolve through the barriers of immune suppression to ultimately destroy the normally functioning body.

Furthermore, some of the epigenetic diseases arise from defects in the machinery that encodes epigenetic marks. For example, Rett syndrome occurs due to mutations of the methyl-CpG-binding protein 2 gene (MECP2), systemic lupus erythematosus (SLS) results from lower DNMT1 level associated DNA hypomethylation in T cells and the immunodeficiency, centromic instability and facial anomalies (ICF) syndrome patients harbor mutation of the DNMT3B [76,77]. In addition to epigenetic diseases that can be traced to specific defects in the epigenetic marks or the specific components of the machineries, most of the complex diseases such as schizophrenia, bipolar disorder and others exhibit a range of symptoms with varying degrees of severity due to epigenetic alterations [78–80] (Figure 4).

Epigenetic aberrations in cancer have been extensively studied and remains as the leading experimental system to model the various associations between alterations in epigenetic memory and defects in the various regulators and components of the epigenetic machinery and disease progression [6] (Figure 4). Although the cancer genomic DNA is overall hypomethylated, hypermethylation and silencing of critical genes including cell cycle regulators, such as p14/ARF and CDKN2A (Ink4A/p16), DNA repair genes, such as BRCA1 and MGMT, pro-apoptotic genes, such as DAPK and RUNX3 and those involved in the maintainance of epithelial cell polarity and tight junction formation such as CDH1, CDH13, CLDN7 and LKB1 have been extensively documented in the literature. In addition to differential DNA methylation associated dysregulation of gene expression, recent studies show that non-coding RNAs such as miRNAs and lncRNAs and chromatin modifications also contribute to cancer progression in a stage specific manner [81–83]. As we have discussed in the previous sections, cancer progression can be explained through various stages represented by distinct functional properties of cancer cells based on dedifferentiation and reprogramming of epigenetic memory [73,84,85] (Figure 4). A general utility of epigenetic alterations is their use as biomarkers for prognosis as well as especially for early diagnosis of cancer as the symptoms often do not present themselves until the primary tumor has progressed to invade the surrounding tissue and as such it may provide avenues for early detection and cancer prevention [86–92].

Dysregulation of enzymes involved in establishing the epigenetic changes (writers; e.g., DNMT, PMT (DOT1L) and HAT (p300), removing these changes (erasers; e.g., HDACs, LSD1), and interacting factors (readers; e.g., BET proteins) that translate the presence/absence of unique epigenetic changes into specific activity or overactivity could be responsible for various diseases such as cancer. Therefore, targeting the disease causing functionalities resulting from these epigenomes and their associated factors have become an effective strategy to manage these indications [93–98] (Table 2).

Table 2.

Examples of epigenome specific therapeutic targets, drugs and diseases.

Target	Drug	Mode of action	Indication
DNMT	5-azacytidine (Vidaza)	Nucleoside inhibitors	MDS, CML
	Decitabine (Dacogen)		MDS, CTCL, PTCL
HDAC	Vorinostat (SAHA), Belinostat (Beleodaq)	HDAC inhibitors (Hydroxamates)	CTCL, PTCL
	Panobinostat (LBH589)		MM
	Depsipeptide (Romidepsin)	HDAC inhibitor (Cyclic peptide)	CTCL, PTCL
	Valporate, Phenyl butyrate	HDAC inhibitor (Fatty acids)	MDS, AML, CLL
DOT1L	Pinometostat (EPZ-5676)	Small molecule inhibitor	MLL-rearranged leukemia
EZH2	GSK126, EPZ6428	Small molecule inhibitors	Lymphoma
LSD1	ORY-1001	Selective covalent small molecule inhibitor	AML
p300	C646	Small molecule competitive inhibitor	Hematological malignancies, AR-positive prostate cancer
BET family of BRD proteins	JQ1, I-BET, I-BET151	Small molecule competitive inhibitors	MLL-rearranged leukemia, NUT midline carcinoma

AML-acute myeloblastic leukemia, AR-androgen receptor, BET- bromodomain (BRD) and extra-terminal family of proteins, CML-chronic myelomonocytic lymphoma, CTCL-cutaneous T-cell lymphoma, DOT1L- Disruptor of telomeric silencing-1-like, an exclusive H3K79 methyltransferase, DNMT- DNA methyltransferase, Enhancer of zeste homolog 2 (EZH2), HDAC-histone deacetylase, LSD1- Lysine-specific demethylase 1, MDS- Myelodysplastic syndromes, MLL- Mixed Lineage Leukemia, MM-multiple myeloma, NUT-nuclear protein in testis, p300 - a nuclear histone acetyltransferase/ a transcriptional co-factor, PTCL-peripheral T-cell lymphoma, SAHA-suberoylanilide hydroxamic acid.

Concluding remarks and future perspectives

Epigenetics have become well established as an important mechanism for all forms of human diseases including cancer [99]. Cancer cells continue to serve as excellent models to infer the molecular mechanisms of various cellular processes such as the cell cycle, DNA repair, recombination, genomic instability and epigenetic modifications along with the associated functional consequences and stem like properties due to the ability to accelerate these processes in experimental models enabling measurements that can be made in real time [100–108]. Therefore by accessing model systems used in cancer research and by deciphering the molecular basis of epigenetic memory, one can generate the framework of mechanisms that ultimately define how the genetic program encoded in the DNA unfolds into specific developmental outcomes and disease conditions. There is still much work to be done in deciphering the nature of intrinsic and extrinsic signals and the molecular basis of epigenetic machineries involved in programming epigenetic memories and the various factors involved in these processes, including the EMFs, to ultimately utilize this knowledge in diagnosis, prognosis and therapy of various diseases. The presence or absence as well as the changes in the pool of factors responsible for epigenetic memory, could also be manipulated for programming and re-programming of cellular properties, as in the case of iPS cells, to develop novel therapeutics for intransient complex diseases as well as increasing lifespan related illnesses that challenge the daily life of the aging population worldwide.

While the benefits of this line of research have enormous implications to improving the quality of life, some of the major obstacles that have to be overcome are as follows: (i) finding ways to address the residual epigenetic memories of iPS cells; (ii) discovery of the correct combinations of epigenomic memory modifiers (e.g., transcription factors, chemical activators/inhibitors etc.) to generate pluripotent cells from the somatic cells representing different degrees of differentiation; (iii) establish the conditions/microenvironments for reprogramming of pluripotent cells to desired somatic cells; and (iv) develop strategies to effectively induce transdifferentiation of somatic cells [109,110].

SUMMARY BOX.

Epigenetic memory in development, health and therapeutics.

• Cellular epigenetic memory, is encoded by the epigenetic marks and the associated epigenetic memory factors, and defines the ability of cells to retain and transmit unique gene expression patterns to the progeny cells.

• Epigenetic marks consist of a collection of gene specific DNA methylations, histone modifications and variations (histone marks), transcription factors, non-coding RNAs (ncRNAs), chromatin remodeling factors, and other epigenetic memory factors (EMFs).

• Genome imprinting dictates gene expression based on parental origin of alleles.

• X chromosome inactivation (XCI) enables the selection of only one X chromosome in females for active status.

• Different levels/layers of fixing of epigenetic memories define the progressive complexities in the developmental stages of a cell.

• Induced pluripotent cells (iPSCs) which lack engrained residual epigenetic memories are the ideal starting points for reprogramming with desired epigenetic memories to achieve successful stem cell based therapies.

• Loss of engrained epigenetic memories of differentiated cells results in acquiring stem cell properties of cancer cells and provides the chromatin templates for reprogramming with novel epigenetic memories that drive cancer progression.

• Altered epigenetic memories could be not only responsible for driving non-genetic diseases but it could also act in conjunction with genetic alterations as well as define the severity of genetic and complex diseases.

• Altered epigenome may serve as diagnostic and prognostic biomarkers for various diseases including cancer.

• Protein factors, enzymes and non-coding RNAs (ncRNAs) involved in encoding the epigenome as well as the readers responsible for eliciting their functional outcome responsible for various diseases including cancer could serve as therapeutic targets for improving human health.