Cell response to external stimuli requires the integration and activation of a cascade of interdependent signals that travel along the cytosol with the cell nuclei as the final destination. Cell–cell interactions, differential kinetics of ligands, and thus different signal transduction cascades converge in the translocation and integration of those fine-tuning signals for the activation or repression of gene expression. Such interplay of molecular signals will give the ability to the cell to start with exquisite precision developmental and differentiation programs. An aspect that has been less explored and that emerges as relevant is how these signals contribute to the establishment of specific transcriptional programs and, moreover, how these programs are perpetuated through cell divisions. On the basis of such kind of questions, and the expansion of epigenetics as a regulatory driver of processes like signal transduction, cell cycle control, and stress response, we are facing a totally new era that needs to be analyzed from a distinct perspective (1–3). Here we focus on some examples of the relationship between epigenetic processes and some signal transduction pathways with the aim to extend our vision from external stimuli integration to gene expression regulation through chromatin modifications.
The genome is organized into a sophisticated set of DNA–protein interactions known as chromatin. The chromatin structure represents the basal level of eukaryotic genome organization. Such kind of organization has profound effects on several nuclear processes such as DNA repair, DNA replication, recombination, and gene transcription (4). Surprisingly, very little is known concerning how extra- and intra-cellular signals are transduced and translated into an epigenetic response. Epigenetic regulation involves a set of processes that affect gene expression without any change in DNA sequence that can be segregated post-mitotically (5). Historically, the term epigenetics was proposed by Conrad Waddington in the early 1940s while trying to explain phenotypic changes without genetic variations and, at the same time, comprehend how the environment can yield a phenotype (6).
It is generally accepted that epigenetic regulation includes histone post-translational modifications and DNA methylation. But the epigenetic landscape is much more sophisticated than these two important processes. Actually, we should incorporate the ATP-dependent chromatin remodelers, the repressor and activator group of proteins Polycomb (PcG) and Trithorax (TrxG), and more recently, the noncoding RNAs and nuclear dynamics as epigenetic components (7–12).
Briefly, we will discuss histone modifications, DNA methylation, and the groups of proteins PcG and TrxG to better understand their roles in signal transduction cascades.
Histone Post-translational Modifications and Epigenetics
The basic component that allows the formation of the chromatin structure is the nucleosome which is composed by a histone octamer and 146 base pairs (bp) of DNA (13). To a certain point, it is expected that modulation of chromatin structure in response to regulatory instructions requires direct histone post-translational modifications with the concomitant relaxation or compaction of the chromatin structure (14). Such covalent modifications mainly occur along the N-terminal domains of histones and include a combination of modifications like acetylation/deacetylation, methylation, and more recently demethylation, phosphorylation, ubiquitination, poly(ADP-ribosyl)ation, among others (14). Notably, based on genome-wide scale studies, there is a well-defined idea of the distribution of histone marks representative of different areas of the genome. Such marks, with the collaboration of transcription factors and co-factors, affect transcription and many other processes like elongation or alternative splicing. For example, tri-methylation of lysine 4, 36, and 79 on histone H3 (H3K4me3, H3K36me3, and H3K79me3, respectively), and differential acetylation of lysine residues of histone H3 and H4 (H3ac and H4ac, respectively) are representative histone modifications constituting the so-called histone code that result in gene activation (15). In contrast, di- and tri-methylation of histone H3 lysine 9 (H3K9me2/3) and H3K27me3 are well established repressive histone marks. Importantly, the relative recent description of a family of histone demethylases permits a versatile reversibility of either active or repressive marks in a tightly regulated manner (16). Finally, based on the action of transcription factors and co-factors and their ability to respond to highly specific signals, there is the need for recruitment of enzymes that covalently modify histones to specific regulatory locations, affecting the expression of target genes.
Epigenetic Silencing Through DNA Methylation
Historically, DNA methylation has been considered as one of the most relevant epigenetic processes. Based on the variety of epigenetic processes and their interdependency, this assumption can be hardly supported. In any case, DNA methylation is synonymous of epigenetic silencing, and it is involved in organism development, X-chromosome inactivation, imprinting, genome stability, and cell differentiation, among others (5). DNA methylation occurs at the 5′ position of the cytosine ring mainly within the CpG dinucleotides. The incorporation of the methyl (CH3) group to the CpG is basically performed by two groups of DNA methyltransferases. The Dnmt3a and Dnmt3b DNA methyltransferases, known as the de novo enzymes, act in the early stages of development establishing genome-wide specific patterns of DNA methylation and contributing to cell lineage commitment. The second type of DNA methyltransferase is the Dnmt1, known as the maintenance enzyme, which reads and segregates DNA methylation patterns during replication (5). In addition, the molecular mechanisms responsible for epigenetic silencing by DNA methylation include a family of proteins with the capacity to bind methylated DNA, like the methyl-CpG-binding proteins MeCP2, MBD2, Kaiso, among others (17, 18). It is well known that aberrant DNA methylation participates in different pathologies including cancer, through a not well understood combination of hypo- and hyper-methylation of undesired regions of the genome (17). Which molecules target abnormal DNA methylation is a question that remains unanswered. Based on this question, it is relevant to investigate if there are particular signaling pathways linked to aberrant DNA methylation. Related to this, there is a growing list of evidences describing alternative mechanisms for DNA demethylation (19, 20). For years, it has been thought that DNA methylation was the most stable epigenetic mark and that it was almost erased during early stages of mammalian development (21). The isolation of the enzymes and the elucidation of the mechanisms for active DNA demethylation have been complicated tasks. In the last years, it has been proposed that active DNA demethylation involves the conversion of the 5-methylcytosine (5mC) into 5-hydroxymethylcytosine (5hmC) by the Tet group of proteins (22), deamination reactions by the activation-induced cytidine deaminase (AID), and base excision repair mechanisms (19, 23, 24). Moreover, one of the most exciting findings of this topics is that active DNA demethylation is not restricted just to early development (25, 26). Without no doubt, this is a relevant topic and our prediction is that DNA demethylation should be a highly regulated process dependent on specific signaling pathways.
The Repressive and Activating Group of Proteins Polycomb and Trithorax
Another epigenetic component that has acquired a growing relevance is the PcG and TrxG group of proteins (8, 11). The PcG and TrxG group of proteins were initially discovered and described in the fruit fly Drosophilamelanogaster as negative and positive regulators, respectively, of homeotic genes required for specifying cell identity along the antero-posterior axis of segmented animals. PcG and TrxG group of proteins are multimeric complexes that need to be attracted to their target locations by transcription factors, but more recently, such recruitment has been documented through the intervention of short and even, long noncoding RNAs (11, 27, 28). PcG and TrxG coordinate critical function for cell homeostasis, like embryonic stem cell (ESC) self-renewal capacity, epigenetic silencing of defined genomic regions, inactivation of X-chromosome, cell differentiation, regulation of cell signaling genes, among others. (see later). The PcG and TrxG complexes' subunits possess histone methyltransferase activities, incorporating methyl groups to specific lysine residues that are subsequently “read” by other combination of PcG and TrxG proteins.
PcG-mediated epigenetic silencing requires the formation of two interrelated repressive complexes named PRC2 and PRC1 (11). In the PRC2 complex, the histone methyltransferase EZH2 is responsible for the di- and tri-methylation of the histone H3 at lysine 27 (H3K27me2/3) in the context of a complex formed by SUZ12, EED, and RbAp46/48 proteins. The H3K27me3 represents the signal for the incorporation of the PRC1 complex formed by BMI1, RING1A, RING1B, MEL18, or NSOC1, complementing the formation of repressive chromatin configuration. This is reinforced by the action of RING1A/1B which ubiquitinates the lysine 119 of histone H2A (H2AK119ub1). Importantly, both complexes, PRC1 and PRC2, are formed with an additional combination of peptides that are proposed to increment its specificity and facilitate higher levels of activity (11). How the PRC2 complex is recruited to the PcG-regulated regions of the genome is an aspect that still remains unsolved in mammals. It is worth mentioning that none of the PRC2/PRC1 complexes possess DNA-binding domains' subunits. Thus, two main proposals have been recently suggested on the literature: (1) the action of specific transcription factors, like YY1, which have been involved in the regulation of genes through PcG of proteins and (2) an attractive and novel strategy has been documented on the basis of the action of noncoding RNAs, like in the case of the Kcnq1 imprinted domain or the action in trans of HOTAIR in the HOX genes loci, among others (29, 30).
The activation group of protein TrxG counteracts the repressive function of PcG as “anti-silencers,” not only of the homeotic genes but also of many others (31). TrxG forms multisubunit complexes with variable compositions. Such complexes are then able to mediate their activities through histone modifications and ATP-dependent chromatin remodelers. The histone modifications are dictated by a SET domain histone methyltransferase known in humans as the mixed lineage leukemia factor (MLL) which tri-methylates the lysine 4 of the histone H3 to establish an active chromatin configuration (32). As mentioned, the TrxG group of proteins contain ATP-dependent chromatin remodeling proteins with the capacity for repositioning nucleosomes. Among them we found subunits of the SWI/SNF complex, like BRM, BAF250, BAF170, and BAF47, and components of the NURF complex like SNF2L, RpAp46, and RpAp48 (8, 31). Furthermore, histone acetylation has also been associated to TrxG activity. Co-purified with MLL-containing complexes it has been identified the MYTS1/MOF histone acetyltransferase which specifically acetylates histone H4K16 which is a histone mark of actively transcribed genes (33). This histone mark favors an open chromatin structure that in addition prevents the incorporation of the repressive histone mark H3K27me3. Therefore, the interplay between PcG and TrxG complexes' activities associated with different covalent histone modifications and ATP-dependent chromatin remodeling complexes modulates different chromatin configurations and gene expression. An outstanding aspect of such complexes is their varied composition that has been suggested, in terms of their ability to respond to discrete signals in a cell-specific manner or during organism development probably through the activation of different signal transduction pathways.
In conclusion, today the PcG and TrxG proteins symbolize one of the most important epigenetic regulators. In the context of the present review, it is critical to mention that there are no substantial evidences of the connection between signal transduction pathways and regulated recruitment of PcG and TrxG complexes. Consequently, this is a topic that clearly deserves particular attention and should be investigated in the next future.
With this brief background in mind about some epigenetic regulatory processes, we address different signal pathways that are relevant in cell differentiation and organism development from the epigenetic perspective. Despite the intense research that has been done in both cell signaling and epigenetics, linking epigenetics and signaling pathways remains as a vast unstudied field. Although scarce, pioneer studies have begun to uncover some aspects of this signaling–epigenetic crosstalk.
In particular, we present an overview of the MAPK, Wnt, Notch, JAK-STAT, JNK, NF-κB, and PKA signaling pathways. Based on such overview, we further propose future research directions linking cell signaling and epigenetics.
THE MITOGEN-ACTIVATED PROTEIN KINASE (MAPK) PATHWAY TARGETING CHROMATIN
Mitogen-activated protein kinase (MAPK) pathways regulate eukaryotic gene expression at multiple levels in response to diverse extracellular signals. This occurs not only by regulating the activity of transcription factors but also by modulating the chromatin structure of regulatory elements among other cellular processes (34). MAPKs mediate their effects through specific phosphorylation, activating or inactivating a large range of protein substrates including MAPK-activated protein kinases and other types of kinases (Fig. 1A; 35). Among the downstream MAPKs pathway targets, the mitogen- and stress-activated protein kinases 1 and 2 (MSK1/2) and the p38 MAPK show particular interest since they are able to directly target several types of transcription factors, including the cAMP-response-element binding factor (CREB), the nuclear factor κB (NF-κB), and they can also induce phosphorylation of the histone H3 and the HMG-14 chromatin-associated protein (34, 36–38). In particular, MSK1/2 possesses at their C-terminal domain a docking configuration that is responsive to activating extracellular-signal-regulated kinase (ERK) and p38 MAPK (Fig. 1A).
As mentioned earlier, the induction of multiple transcriptional activation or repression programs by signal transduction cascades do not only have transcription factors as targets, integral components of the chromatin structure are also responsive to those signals. The differential phosphorylation of Ser10 and Ser28 of the histone H3 is a clear example. Historically, it has been assumed that histone H3 Ser10 phosphorylation is the major modification that has a vital role in maintaining condensed mitotic chromosomes allowing successful chromosome segregation (39). The activation of ERK and p38 MAPK allows an efficient phosphorylation of the histone H3 (36). Interestingly, there is a differential phosphorylation of histone H3 Ser10 and Ser28 (36, 37). Based on immunofluorescence and sequential immunoprecipitation experiments using phospho-Ser10- and phospho-Ser28-specific antibodies, it has been demonstrated that each type of covalent modification occurs in distinct subsets of histone H3, for example, that each modification is associated invivo to chromatin distributed in different genomic regions.
But, it is until recently, that the histone H3-Ser28 phosphorylation functional meaning has been elucidated (40). Based on the targeting of MSK1 to the α-globin gene promoter and the consequently induced phosphorylation of the histone H3-Ser28, Lau and Cheung demonstrated that Ser28 phosphorylation displaces the binding of the PcG repressive complex causing a significant reduction in the incorporation of the histone H3K27me3 repressive histone mark (40). Furthermore, this modification induces a dual code associated to gene activation through the phosphorylation of Ser28 and the acetylation of the Lys 27 (Fig. 1B). Thus, it has been suggested that in response to MSK1 pathway, histone H3 Ser28-acK27 marks antagonize PcG repressive function, facilitating the subsequent recruitment of the other chromatin modifier enzymes and/or specific transcription factors allowing gene expression activation.
We conclude that histone H3 phosphorylation through ERK and p38 MAPK activation of MSK kinases represents a critical step that with no doubt couples signal transduction pathways that are responsive to external stimuli to regulate gene expression in particular cell-types.
The p38 MAPK pathway participates in numerous biological processes like response to stress and inflammation, cell proliferation, differentiation, and survival of particular cell types (41). The p38 MAPK family members are approximately 60% identical in their amino acid composition and are known as p38α, p38β, p38γ, and p38δ. The p38α MAPK presents a constitutive expression in most cell types, in contrast to others which show a more tissue-specific expression profiles (41). As mentioned, for their activation MAPK pathways require phosphorylation of target proteins by MAPK kinases which induce conformational changes facilitating substrate recognition.
Myogenesis represent one of the most studied cell lineage differentiation system in which epigenetics and cell signaling give us a nice example of cross talk during this critical process. Myoblasts can be differentiated in multinucleated myotubes, mainly using the C2C12 murine cells as an in vitro differentiation system simply based on serum withdrawal. This system has provided a very useful way to study signaling pathways and chromatin associated events allowing myoblast cells to differentiate. MyoD is a transcription factor which has been referred to a master regulator of myogenesis. It is able to recognize E-boxes in the genome and can initiate the myogenic program in differentiating myoblasts (42). At the chromatin level, MyoD is associated with histone acetylation of its target regulatory elements associated with key myogenic genes (42). Interestingly, many of these genes are repressed by the PcG proteins. In myoblast, the histone methyltransferase EZH2 and PRC2 components are recruited to target genes by the transcription factor YY1 with the concomitant enrichment of the histone repressive mark H3K27me3 (43). During myoblast differentiation, several chromatin-modifying enzymes are coordinately recruited through transduction of external signals. First, the repressive histone mark H3K27me3 needs to be erased. Thus, in MyoD target genes the homeobox transcription factor Six4 recruits the histone demethylase specific of the H3K27me3, UTX, and removes this histone repressive mark at promoter regions of genes involved in myogenic differentiation (44). Once the chromatin structure of regulatory elements is more permissive, the MADS-box transcription factor Mef2d and MyoD can now start to positively regulate transcription of myogenic genes (45). p38 MAPK is activated upon cell-to-cell contact and Medf2d is phosphorylated. Then, the phosphorylated form of Mef2d can recruit the TrxG histone methyltransferase Ash2L/MLL2 which is able to incorporate the chromatin “open” histone mark, H3K4me3 to target genes and induce transcription initiation (45). The activation mechanism is then complemented by the pTEF-b-mediated phosphorylation of the RNA polymerase II which is recruited by MyoD (46). Interestingly, and as part of the myogenic differentiation program, EZH2 and YY1 are post-transcriptionally inhibited by specific microRNAs (47, 48).
In summary, myogenesis responds to an interdependent cascade of events that initiates with the reversal of the repressive chromatin context imposed by PcG proteins, the recruitment of myogenic transcription factors, and, importantly, the activation of the p38 MAPK pathway in response to extra-cellular signals, which further enhances the recruitment of chromatin modifying proteins that change chromatin structure to activate transcription.
WNT CELL SIGNALING PATHWAY AND ITS CROSSTALK WITH EPIGENETICS
Cell differentiation can be summarized as the process in which a pluripotent cell becomes a more specialized cell through the controlled establishment of specific gene expression patterns. This complex process is guided by both, intra- and extra-cellular signals that establish those patterns. It has been widely recognized that cell signaling plays a crucial role during the differentiation process. On one hand, a cell must be able to respond to the external stimuli that drives the differentiation process. On the other hand, the transcriptional changes guided by external stimuli must be inherited to the daughter cells after mitosis to maintain their lineage.
The epigenetic features of mammalian stem cells have been extensively studied, although the interplay between external stimuli and epigenetic effectors during self-renewal and differentiation processes is poorly understood. A recent publication by Gu et al. (49) described that Pygo2 is essential for mammary stem/progenitor cells self-renewal. Pygo2 contains a PHD domain, a common domain in epigenetic regulators (50). Deficient Pygo2 mice show a reduced proliferation of mammary epithelial cells (MEC). Moreover, Pygo2 deficiency rescues β-catenin-induced mammary overgrowth thus implying an interaction between Pygo2 and Wnt/β-catenin signaling pathway. Furthermore, it was proved that Pygo2 promotes the trimethylation of H3K4 in the promoter regions of Wnt/β-catenin target genes, a typical mark of transcriptional active chromatin, through the recruitment of histone methyltransferases (HMTs). Taken together, this evidence shows a clear interaction between the Wnt/β-catenin pathway, the PHD finger protein Pygo2, and HMTs during the maintenance of MEC self-renewal (49, 51).
Several interesting issues remain to be studied regarding the Pygo2-Wnt/β-catenin crosstalk. It is not clear if β-catenin and Pygo2 interact directly or if there are other factor involved in the recruitment of Pygo2 to the Wnt/β-catenin target gene promoters. It would also be interesting to determine if other epigenetic factors, such as histone demethylases, guided by Wnt or other signaling pathways influence the transcription of the Wnt/β-catenin target genes during MEC differentiation. Also, there is an interesting link between Pygo2 and breast cancer. Abnormal over-expression of Pygo2, caused by genetic or epigenetic or both processes, may induce uncontrolled proliferation. In this sense, it has been observed that Pygo2 is upregulated in cancer cell lines and breast cancer (49, 52). This example opens the possibility of a completely new field of study for therapeutic targets in cancer regarding the epigenetic–cell signaling crosstalk.
Histone deacetylases (HDACs) are involved in epigenetic silencing, a fundamental process for cell function, both during differentiation and in differentiated cells (for review, see ref. 53). In an interesting example, Ye et al. (54) proved recently that HDAC1 and HDAC2 are involved in oligodendrocyte development and interact with the Wnt/β-catenin signaling pathway (Fig. 2). To assess the role of HDAC during oligodendrocyte formation, these authors generated oligodendrocyte-specific HDAC1/2 double knock-out mice. HDAC1/2 mice are not able to generate mature oligodendrocytes. Furthermore, they showed that β-catenin is abnormally stabilized and translocated to the nucleus of oligodendrocyte precursor cells in HDAC1/2 mice, then inhibiting the expression of Olig2, an essential gene for oligodendrocyte differentiation. This evidence indicates that HDAC1/2 are essential for oligodendrocyte maturation and that they antagonize the Wnt/β-catenin pathway. β-catenin activates canonical Wnt signaling by forming a biparite transcriptional factor with a member of the TCF/LEF transcription factor family (55). They also found that TCF7L2 is the oligodendrocyte specific effector of the Wnt/β-catenin signaling pathway. They further proved that HDAC1/2 is able to interact with TCF7L2, and that this interaction is disrupted by β-catenin over-expression. Based on all their evidence, they finally propose that while β-catenin/TCF7L2 dimer act as a repressor of oligodendrocyte differentiation, HDAC1/2 competes with β-catenin for the binding to TCF7L2, promoting the function of TCF7L2 as an activator of oligodendrocyte differentiation. Together, these findings show an unexpected role for HDACs: not only they act as epigenetic repressors, but they can also inhibit Wnt signaling through directly disrupting β-catenin/TCF7L2 interaction (Fig. 2; 54). It remains to be solved if the HDAC/TCF7L2 dimer is involved directly in the transcriptional regulation of genes or if its only function is disrupting β-catenin/TCF7L2. Also, it will be interesting to explore if this kind of mechanism, in which epigenetic factors interact and modify directly the effector proteins of signaling pathways is a rule rather than an exception. Finding new functions for canonical epigenetic factors and their contribution in co-regulating signaling transduction pathways, besides their direct role in chromatin modification, will be an exciting area of research in the near future that remains unexplored.
NOTCH SIGNALING PATHWAY AND ITS EPIGENETIC REGULATION
Notch signaling is a highly conserved developmental pathway that is present in all metazoans and controls different processes like cell fate specification, self-renewal, differentiation, proliferation, and apoptosis throughout development and regeneration (56). Notch receptors are single-pass transmembrane receptors comprising one extracellular and one transmembrane segment (57), which are activated by transmembrane ligands of the DSL family on neighboring cells. Whereas Drosophila has only one Notch receptor that is bound by two different ligands (Delta and Serrate), there are four Notch receptors (1–4) and five ligands (Jagged 1 and 2, and Delta-like 1,3,4) in mammals (56).
Activation of Notch pathway results in the proteolytic cleavage of the Notch receptor and the subsequent release of the Notch Intracellular Domain (NICD), which then translocates into the nucleus to activate specific gene targets via the CSL family (CBF1/RB-Jk, Su(H), and Lag1) of sequence-specific DNA-binding proteins (56, 58). Once activated, the NICD associates with CSL and other complexes inducing expression of genes that are members of the E(spl) class in Drosophila or the Hes class in mammals. These genes are transcription factors, the so-called Notch-effector genes (57, 59, 60). In the absence of NICD, CSL recruits transcriptional co-repressor complexes to actively inhibit expression of most Notch target genes (56, 58, 61–66).
Different chromatin modifying proteins, like histone deacetylases, H3K9 methyltransferases, CtBP, NcoR/SMRT, Groucho, and, more recently, histone chaperones and histone H3K4 demethylases have been reported to associate with Notch repressor complexes to silence target genes (58, 61–64). Despite the existence of different co-repressor complexes, all appear to recruit HDACs (65) and histone H3K4 demethylases (58, 61, 66).
Recent reports suggest an important role for histone chaperones during gene silencing of Notch-regulated genes (58). Histone chaperones bind specific histones and include the highly conserved H3/H4 chaperones ASF1, CAF1, HIRA, and SPT6 and the H2A/H2B chaperones NAP1, nucleoplasmin, and FACT (67, 68). A proteomics survey of the protein interaction networks of the histone chaperones ASF1, CAF1, HIRA, and NAP1 in Drosophila embryos revealed that ASF1 and NAP1 interact with two related repressor complexes. ASF1 interacts with LAF (including the H3K4 demethylase LID/KDM5) whereas NAP1 interacts with RLAF, comprising LAF + RPD3 (histone deacetylase, 69). These two repressor complexes interact with Notch target genes via Su(H) and mediate gene selective silencing (58). Depletion of ASF1 or NAP1 resulted in de-repression of the E(spl) genes. Importantly, depletion of other histone chaperones (i.e., HIRA) had no effect on Notch target gene expression, suggesting a specific role for these histone chaperones in Notch gene silencing (58).
Both ASF1 and NAP1 are necessary for the specific removal of H3K4me3 by facilitating LID/KDM5 recruitment to chromatin. Further, depletion of NAP1, but not ASF1, caused a strong loss of nucleosomes at the E(spl) promoters and enhancers indicating that NAP1 mediates higher nucleosome density at regulatory elements of Notch target genes (Fig. 3B). Finally NAP1 cooperates with RLAF stimulating histone deacetylation by RPD3 (58).
Other HDACs have been reported to affect Notch target gene expression. For example, SIRT1, a H4K16 deacetylase (70), interacts with LSD1/KDM1A a histone H3K4 demethylase and CtBP1 a repressor adaptor protein making a complex that acts selectively as a co-repressor of the CSL/Notch target genes in the absence of Notch signaling via the establishment of repressive chromatin. In particular, SIRT1 is necessary for deacetylation of H4K16 during repression of target genes in absence of Notch signaling (Fig. 3B). Upon induction of Notch signaling, H4K16 is acetylated and H3K4 is methylated, inducing a chromatin state permissive for transcription (61).
Even that so much work has been done to identify those proteins that interact with CLS DNA-binding proteins to regulate Notch target gene expression, less is known about which chromatin-modifying proteins could be regulating Notch receptors and ligands.
In Drosophila, imaginal eye discs are groups of epithelial cells present in larvae that will differentiate into specific parts of the adult fly. The Notch signaling pathway can control growth of eye imaginal discs (71, 72). Eye imaginal discs bearing cells with mutations in ph (polihomeotic), a member of the Polycomb Repressive Complex 1 (PRC1), have defects in proliferation and differentiation. Moreover, mutant cells with Ras over-expression invade other tissues and trigger the formation of metastases. When ph mutant cells are transplanted into the abdomen of wild-type adult flies, they over-proliferate, killing the host, resembling the phenotype of mammalian epithelial cancers. Finally, mutations in ph induced loss of Notch, Ser and eyg repression, indicating that PRC1 is controlling gene expression of key genes along the entire pathway (Fig. 3A) (i.e., is affecting the expression of the receptor, ligand, and target genes) (71).
In summary, the Notch signaling pathway is regulated at all levels by important chromatin modifying proteins. Different co-repressor complexes associate specifically with certain histone demethylases, histone deacetylases, and histone chaperones that collectively repress target gene expression. Moreover, the receptors and ligands are regulated at the epigenetic level by PcG proteins. Most of these proteins are highly conserved among different organisms from Drosophila to humans, suggesting that the mechanisms of regulation may also be conserved.
JAK-STAT SIGNALING PATHWAY AND ITS EPIGENETIC REGULATION
JAK-STAT signaling is a highly conserved pathway that has been implicated in different processes like regulation of differentiation, apoptosis, proliferation and disease like cancer. (73). JAK-STAT signaling activation involves the binding of one of the Upd ligands to the hop receptor, which activates the Jak kinase resulting in the phosphorylation of STAT92E and its translocation to the nucleus, where it functions as a transcriptional activator, inducing expression of target genes (74). Different lines of evidence suggest that JAK-STAT signaling pathway can be regulated at the epigenetic level (74–77). Most of this work has been done using D. melanogaster as a model system, in part because the components of this pathway are encode by a single gene copy and also because of its powerful genetics.
In Drosophila, mutations in components of the PRC1 complex can reactivate JAK/STAT signaling in eye imaginal discs producing an enlarged eye phenotype in adults, characterized by an increase in cell number. It was demonstrated that all Upd ligands (Upd1, Upd2, and Upd3) are direct targets of PRC1-mediated repression (Fig. 4A) (76). This contrasts with Notch signaling, where just one of the two ligands are affected by PRC1 (71). Moreover, the Notch signaling pathway seems to be regulated by PRC1 at all levels (receptor, ligand, target genes). Recently, the ATP-dependent chromatin remodeler complex, nucleosome remodeling factor (NURF) was implicated in the regulation of JAK-STAT signaling in Drosophila testis (75). The NURF complex is composed by different subunits (ISWI, Nurf55, Nurf30) being Nurf301 the only component that is not shared by other ATP-dependent chromatin remodeler complexes (78). The JAK-STAT signaling pathways are responsible for the maintenance of the germ stem cells (GSC) and cyst progenitor cells (CPC) in Drosophila testis (79, 80). The Upd ligand is secreted by a group of somatic post mitotic cells called the niche that are adjacent to the GSC and the CPC, thus activating the JAK-STAT signaling and preventing differentiation. By using different genetic approaches it was demonstrated that Nurf301 promotes expression of STAT92E and prevents the expression of the differentiating factor Bam. Flies in which STAT92E was overexpressed in a mutant loss-of-function background for Nurf301 recover CPC cells; also, Nurf301 interacts genetically with SOCS36E (repressor of the pathway) in a manner consistent as NURF acting as a positive regulator (Fig. 4B). This finding is surprising because during hematopoiesis, NURF represses STAT target genes through interaction with the transcriptional repressors Ken and Barbie (81, 82). At this point, it is unknown if NURF promotes JAK-STAT signaling by transcriptional activation or indirectly by repression of JAK-STAT inhibitors (75).
So far, we have described some examples of epigenetic regulation of JAK-STAT pathway in eye imaginal disc and also in Drosophila testis during spermatogenesis. Recent findings in Drosophila have identified a noncanonical mode of JAK-STAT signaling which controls heterochromatin stability (77). An hyperactivated form of Jak kinase is associated with a high incidence of hematopoietic tumors, a leukemia-like phenotype. Loss-of-function mutations in the heterochromatin components, HP1 and Su(var)3-9, enhance tumor formation under hyperactive Jak kinase background. It was demonstrated that hyperactivation of Jak kinase induces loss of heterochromatin gene silencing. This disruption allows derepression of genes that are not direct targets of STAT92E as evidenced by loss of heterochromatin-mediated position effect variegation (PEV) silencing of white gene (Fig. 4C). Jak loss of function enhances heterochromatin silencing, while HP1 overexpression suppresses the tumorigenic phenotype associated with Jak kinase hyperactivation. This phenomena demonstrates a global effect of JAK-STAT signaling activation that can affect many genes irrespective of STAT93E binding to their regulatory regions.
In conclusion, in Drosophila, PRC1 controls JAK-STAT signaling through direct inhibition of expression of Upd genes, thus blocking activation of the pathway, even that the neighboring cells are expressing hop ligand. Moreover, the NURF complex seems to positively regulate JAK-STAT signaling in Drosophila testis, through positive regulation of STAT92E in GSC and CPC cells. Finally, this pathway can affect global heterochromatin conformation through disruption of HP1 binding. It could be possible that in spermatogenesis, JAK-STAT signaling maintains the stem cell phenotype not just by repression of Bam gene, but also through induction of an open chromatin configuration affecting genes that are not direct targets of STAT92E but are important for stem cell maintenance. During differentiation, those genes could be silenced because of heterochromatin formation as a consequence of silencing of JAK-STAT pathway.
We have given some examples about how signaling pathways use the epigenetic machinery to regulate gene expression, and also we have discussed some examples in which important chromatin modifying proteins affect critical components of the signaling pathways. However, it is extremely important to emphasize those cases in which signaling pathways can affect chromatin-modifying proteins. From this part, we present examples of direct regulation of critical epigenetic components by signaling pathways.
JNK SIGNALING PATHWAY AND THE REGULATION OF Polycomb GROUP OF PROTEINS
During regeneration of Drosophila imaginal discs, cells can change their identity during a process known as transdetermination. In 2005, it was reported that some components of PcG are downregulated during transdetermination (83). In particular, it was shown that the frequency of transdetermination from leg to wing is enhanced in PcG mutant flies. This downregulation is directly controlled by the JNK signaling pathway, which is activated in cells undergoing regeneration, repressing the expression of PcG genes like Pc, ph-h, and E(Pc), thereby losing the repression of target genes and allowing cell plasticity. Finally, they carried out the analogous experiment in mammalian cells, using mouse embryonic fibroblast in which JNK signaling was activated by exposition to UV light. They found that upon induction of JNK signaling, Mph2 (mouse polihomeotic) was also downregulated (83), suggesting that JNK signaling can control expression of PcG genes across species.
NF-kB SIGNALING PATHWAY AND THE REGULATION OF HISTONE DEMETHYLASES
Cells can recognize external harassment like microbial infection, tissue injury, and many others through a large variety of sensing mechanisms that mainly involve trans-membrane receptors activating different signaling transduction pathways (84). Thus, through a cascade of biochemical signals the stimuli is translated and transmitted to the cell nucleus with the aim to activate or repress the transcription of genes encoding a broad range of regulatory and effectors proteins. A less explored aspect of such response to external stimuli is the contribution of the chromatin structure and its modification enzymes to such variety of signals. In other words, even though signal transduction remains a major player, it is now relevant to understand how chromatin structure is targeted in a regulated manner to allow differential gene expression in response to an inflammatory stimulus.
An attractive example is shown by the regulated alteration of PcG-mediated silencing of genes involved in inflammation response, in particular, through the action of specific histone demethylating enzymes (1). Natoli et al. hypothesized that demethylation of specific histone repressive residues, in response to external stimuli, may regulate inflammatory response. Therefore, they assessed the expression profile of 30 genes containing the JumonjiC domain which is known to mainly catalyze site-specific histone demethylation of mono-, di-, and tri-methylated lysine residues (16, 85). Interestingly, they found that the Jmjd3 histone demethylase expression (specific for H3K27me3) is robustly induced in macrophage exposed to bacterial products and inflammatory cytokines. Furthermore, they demonstrated that Jmjd3 induction depends on direct binding of NF-κB to a set of three κB-binding sites in the Jmjd3 gene promoter (1). NF-κB was the first transcription factor described whose DNA-binding activity could be induced by an extracellular stimulus (86). NF-κB family of nuclear factors consists of five members: p56, p52, RelA, c-Rel, and Relb (87). A variety of dimeric and multimeric NF-κB-IκB complexes have been described in the cytoplasm of unstimulated cells (88). One of the most common activation pathways involves phosphorylation of the associated IκB, which leads to its ubiquitylation and proteasome-mediated degradation, thereby releasing the NF-κB dimer allowing it to translocate to the nucleus (87, 89). Then, among several mechanisms by which the NF-κB pathway can activate subsets of target genes in response to a particular stimulus, the activation of a specific histone demethylase is one of them, pointing out the relevance of chromatin structure in this particular pathway.
It has been documented that the Jmjd3 histone demethylase gene is highly expressed in differentiating bone marrow cells and downregulated in differentiated macrophages. In contrast, the Utx histone demethylase, which belongs to the Jmjd3 subfamily, that is also able to demethylate the histone H3K27me3, is expressed in constant levels and is considered a housekeeping gene (1). This implies that the Jmjd3 histone methylase is expressed in an inducible and cell-type restricted fashion and that is responsive to the NF-κB signal transduction pathway.
Furthermore, a novel series of evidences show that acetylation of NF-κB has an effect on the specificity, strength, and duration of the NF-κB-dependent signal pathway (90). It is actually known that in addition to histones, HATs and HDACs have other targets in particular, transcription factors favoring or decreasing their affinity to their DNA-binding sites with regulatory consequences. Serine and threonine acetylation of NF-κB residues has been shown to positively regulate NF-κB signaling cascade. This is further supported by the demonstration of the direct acetylation of NF-κB subunits like p52 and p65 (90). These findings have contributed to propose that acetylation of NF-κB subunits and the recruitment of HATs and HDACs by members of the same pathway are responsible of an intricate crosstalk between NF-κB signal pathway and the trans-activation or repression of genes that are associated to this signal transduction cascade.
Those are examples of mechanisms by which chromatin structure and specific transcription factors contribute to the selective regulation in response to external signals and outline the versatile relevance of histone demethylases, HATs and HDACs.
PKA SIGNALING-DEPENDENT REGULATION OF HISTONE DEMETHYTLASE PHF2.
The glucagon–PKA signaling pathway regulates glucose homeostasis by controlling glycogenesis and glycogenolysis (91, 92). After glucagon binds to its receptor in the cell surface, the protein kinase A (PKA) becomes activated and phosphorylates the glycogen phosphorylase kinase which activates other enzymes, resulting in an increased glycogen breakdown (glycogenolysis) and production of glucose, for hepatic output (92). The PKA can phosphorylate other proteins, like histone demethylases (91). PHF2 is a jmjC H3K9me2 demethylase that is a target of PKA. Inactive in its unphosphorylated form, this histone demethylase is catalytically activated after phosphorylation by PKA (91). Once PHF2 has been phosphorylated, it can bind and demethylate ARID5B, a DNA-binding protein, making a complex that associates with target promoters of genes like Pepck and G6Pase where it removes the H3K9me2 mark (91). This is the first example of a jmjC histone demethylase that in order to be active needs a post-transcriptional modification, like phosphorylation, mediated by a protein that is active only if a signaling pathway is induced.
These remarkable findings indicate that some signaling pathways can control directly the expression of PcG and histone demethylases genes during critical processes like development or regeneration. So far, just a couple of examples exist that link signaling pathways to regulation of critical components of the epigenetic machinery (93).
FUTURE DIRECTIONS ON CELL SIGNALING AND EPIGENETICS
Signal Transduction and MicroRNAs: The Epigenetic Connection
The crosstalk between the epigenetic machinery and cell signaling pathways may not be always direct. Diverse mechanisms could affect this relationship and have a great impact in the way that cell signaling and epigenetics interact. An attractive prediction of such mechanisms are noncoding RNAs. In particular, microRNAs could participate as an intermediary between cell signaling pathways and the epigenetic machinery. Briefly, microRNAs (miRNAs) are small RNAs of about 21 bases that recognize its target mRNAs by sequence complementarity and trigger their degradation or inhibit their translation. The sequence that miRNAs recognize, typically known as seed sequence, is composed by 7–8 bases and is critical for the miRNAs specificity (94). It has been widely recognized that miRNAs are fundamental post-transcriptional regulators and have been implicated in several cellular processes such as cell cycle control and progression, differentiation, apoptosis, and stress response, among others (94–96).
There are several examples in which it is clear that miRNAs are key players in the regulation of cell signaling pathways (for review, see ref. 97). Moreover, miRNAs are also involved in the regulation of several epigenetic factors, and are regulated by epigenetic mechanisms as well (98). Therefore, miRNAs are excellent candidates for a functional interphase between cell signaling pathways and epigenetic mechanisms. For example, it has been reported that miR-145 is essential for stem cell differentiation and participates in a complex regulatory feedback loop: it is negatively regulated by the pluripotency factor OCT4 while it targets the post-transcriptional inhibition of OCT4, SOX2, and KLF4, all essential factors for pluripotency maintenance (99). Interestingly, it has been shown that BMP signaling pathway can trigger human stem cell differentiation by inhibiting OCT4, therefore allowing miR-145 expression which in turn silences OCT4, SOX2, and KLF4 (97, 99–101). Furthermore, it has recently been shown that miR-145 can be epigenetically silenced by DNA hypermethylation in prostate cancer (102). Moreover, it has also been demonstrated that OCT4 transcription is downregulated through differentiation by DNA methylation (103).
Taken together, the previous findings can show us a complex regulatory mechanism that links cell signaling, epigenetics, and miRNAs. During stem cell maintenance, OCT4 promoter remains hypomethylated (103), and we may speculate that miR-145 is hypermethylated. During differentiation, BMP pathway inhibits OCT4 expression, maybe by recruiting DNA methylases and histone deacetylases to OCT4 promoter/enhancer region. On the other hand, miR-145 transcription is upregulated, although it remains to be studied which mechanisms are involved in such process. One attractive possibility is that BMP or other related pathway is also involved in miR-145 upregulation. Thereafter, miR-145 post-transcriptionally inhibits OCT4, SOX2, and KLF4 and finally, as differentiation goes on, their gene promoters are epigenetically silenced. It should also be interesting to study if signaling pathways, miRNAs, or a combination of both regulates such silencing.
Other interesting example is miR-7 in Drosophila. miR-7 has been implicated in the regulation of important signaling pathways, like Notch (104) and EGF (105) signaling pathways. Its function has been elucidated during eye development (Fig. 5). In early photoreceptors, miR-7 transcription is repressed by Yan, a transcription factor involved in the differentiation of retinal progenitor cells, by direct interaction with miR-7 cis-regulatory sequences. As photoreceptor differentiation goes on, EGF signaling is transiently induced and breaks this equilibrium by upregulating miR-7, while inhibiting Yan transcription. During EGF transient stimulation, miR-7 avoids Yan repression and in turn is able to inhibit Yan protein synthesis. After the stimuli are over, in mature photoreceptors, miR-7 transcription is sustained, and Yan is permanently inhibited by miR-7 (105). This example is illustrative of a noncanonical epigenetic process guided by a signaling pathway: a double-negative-feedback loop in which two negative regulators that are mutually exclusive change their expression patterns in response to a signal (Fig. 5). Before, and after the signal, the cell has a steady transcriptional memory that is maintained until it is perturbed by the signal. Such regulatory loops are common during processes that require robust and fine-tuned changes in expression patterns such as development (106). It is quite probable that classic epigenetic mechanisms, for example, DNA methylation, histone post-transcriptional modifications, are also important for the transcriptional switch between miR-7 and Yan expression, and for the memory maintenance once the change was done. It will be interesting to determine if such epigenetic processes are guided by signaling pathways and involved other miRNAs.
There is a vast and unexplored field of study regarding the interaction between miRNAs, epigenetic processes, and cell signaling pathways. However, we hypothesize that miRNAs as well as other noncoding RNAs are fundamental players in the crosstalk between epigenetics and cell signaling.
Transgenerational Epigenetic Inheritance: Integrating Information to the Epigenome
Epigenetic information as well as DNA have to be replicated during mitosis and been inherited to the next cell generations to maintain cell fate. During gametogenesis, germ cells have the capacity of erasing this epigenetic memory thus enabling reprogramming of epigenetic information. However, certain regions of the genome maintain this information, bypassing the reprogramming events and allowing the transmission of this information to the offspring (21). The transmission of epigenetic information through generations, that is, from F0 to F1 to F2…Fn, is defined as transgenerational epigenetic inheritance (TEI) (107). True gametic transmission requires that the epigenetic changes can be inherited during the generations independent of the environmental conditions that induced these modifications in the parental generation (107).
Transgenerational epigenetic inheritance can occur through the maternal and the paternal germ line (108). For a long time, it was believed that the paternal genome had almost no contribution to the epigenetic constitution of the new organism, because most of the histones were replaced during spermatogenesis by protamines. Recently, it has been reported that not all histones are replaced by protamines during spermatogenesis and that some histones with specific marks are retained specially within the regulatory regions of important developmental genes, potentially affecting the early development of the embryo (109, 110). Some very interesting examples have come out during the past years about TEI. The most important example supporting TEI comes from studying the agouti (A) locus in mice that controls coat color. Those mice carrying an IAP transposon inserted upstream of the agouti locus (Avy allele) have ectopic expression of the Agouti protein (111). The methylation status of IAP transposon can affect the transcription of the agouti locus; the methylated state of this transposon was shown to be inherited through the female germ line (111) and bypass the epigenetic reprogramming events during early development (112); however, the mechanism of protection toward the erasing of epigenetic marks is not fully understood. Other studies suggest that paternal diet can affect lipid gene expression in the offspring and that epigenetic information in the sperm respond to this environmental influence (113) affecting the expression of Pparα possibly by affecting the methylation status of a putative enhancer for this gene. In the offspring of male parents that were fed with a low-protein diet, the putative enhancer of Pparα shows higher methylation when compared with control offspring. This suggests that the change in methylation could happen during spermatogenesis and it can be potentially inherited to the next generations. However, no change in methylation for the Pparα enhancer could be detected when sperm of the parental generation where analyzed, suggesting that these change could possible arise because of other modification that may have happened during spermatogenesis. In fact, diet affected the RNA content and chromatin packaging of sperm, thus making histone marks and noncoding RNAs as potential candidates for transmission of epigenetic information to the next generation (113).
Several questions arise from these works. First, which signaling pathways the cells use to integrate specific environmental information and induce changes in gene expression that can be potentially inherited? Second, are all the cells capable of responding to the environmental influences? and finally, how do the germ cells integrate this environmental information? Is this information transduced to the cells during the early process of gametogenesis? Do they use different signaling pathways that those that control gametogenesis? Which are the target genes that these pathways could affect to potentially change the epigenome of the next generation? And which molecular mechanisms make possible to bypass the reprogramming events of early development to maintain specific epigenetic changes?
We must remember that epigenetic information is reversible and dynamic. This dynamic nature allows the epigenome to respond or to be affected by more environmental influences than the more stable DNA sequence (21). Owing to their dynamic nature, epigenetic marks seem an ideal way to ensure a short-term adaptation to environmental changes (107). Identifying the molecular mechanisms and the signaling pathways that can control this dynamic nature is an exciting and promising area of research that should be addressed and certainty will give us amazing answers and also will open up new challenges to our understanding of the relationship between the epigenome and cell signaling.
As we can appreciate from these examples, the ways in which cell signaling pathways can interact with epigenetic elements appear to be varied and complex. We are just beginning to understand such interactions, and is very likely that in the future, more and more examples like those presented earlier will show unexpected links between cell signaling networks and epigenetic mechanisms. Integrating both networks, cell signaling and epigenetics, is the next necessary step for the comprehension of complex processes such as development, cell differentiation, and cell plasticity. We can no longer see cell signaling and epigenetics phenomenon as separated issues. There is increasing evidence that shows they are not only complementary networks, but in most cases, they are just two sides of the same coin. We cannot understand differentiation without epigenetics, but we neither can explain the epigenetic processes without the input and specificity given by cell signaling pathways.
The relationship between signal transduction pathways and their target effects over diverse epigenetic processes is forming part of a new era for epigenetic regulation. Until recently, it is our opinion that epigenetics was situated in a more descriptive context. The accumulated epigenetic evidences are currently allowing the incorporation of novel processes like signal transduction. For example, microRNAs could be considered as candidates to better understand the relationships between signal transduction and epigenetics. Furthermore, attractive fields like cell differentiation and reprogramming, trans-differentiation or even the heritable aspects of epigenetics, like transgenerational phenomena, are emerging.
Studying both mechanisms will be complex and may require the development of new technologies, but integrative research will be the only way to understand, and eventually manipulate complex processes such as development, both in health and in disease.
We thank Ricardo Saldaña-Meyer and Alejandro Athie-Cuervo for critical reading of the manuscript. This work was supported by the Dirección General de Asuntos del Personal Académico-Universidad Nacional Autónoma de México (IN209403, IN214407, and IN203811) and Consejo Nacional de Ciencia y Tecnología (CONACyT: 42653-Q, 58767, and 128464). RGA-M and DV-G are supported by a Ph.D. fellowship from CONACyT (255707, 239663).