As a response to dynamic environmental conditions, the interaction of numerous signaling pathways will eventually remodel the epigenetic regulatory network to fluctuate gene expression but leave genome physically unchanged. Fifty years ago, Waddington proposed an epigenetic landscape model to describe the cell differentiation from stem cells as the trajectory of a ball into branching valleys [
43]. It was updated recently to emphasize the self-renewal and plasticity of stem cells. Temporal oscillations in gene expressions restricted the cellular state to a certain region in a fixed state or a set of dynamically changing states [
44]. Cell–cell communication in addition to autocrine or paracrine cytokines was suggested to regulate the fate determination of stem cells through epigenetic regulatory network to control dynamic expression of some particular genes [
45]. The epigenetic network mainly consist of three major events, histone modifications especially histone methylation or acetylation, DNA modifications mainly DNA methylation and noncoding RNAs. Development of a probabilistic model to cluster genomic sequences based on the similarity of temporal changes of multiple epigenomic marks reveals a variety of rules of dynamic gene regulation during epidermal stem cell differentiation and proliferation [
46].
2.1. Histone Modifications and Epidermal Stem Cells
One molecular root of the counterpoise between stemness and differentiation of epidermal stem cells is the interplay of histone modifications with tissue-specific transcription factors and other epigenetic regulators. A group of epigenetic regulators such as Polycomb group (PcG) proteins are required for epigenetic silencing of lineage-defining genes to maintain stemness of epidermal stem cells [
47]. There are two main complexes, the Polycomb Repressive Complexes-2 (PRC2) and the Polycomb Repressive Complexes-1 (PRC1). The PRC2 complex is comprised of three subunits, Ezh2/Ezh1, Eed and Suz12. Ezh2 or its paralog Ezh1 modulates tri-methylation of histone H3 at lysine 27 (H3K27me3), which is a chromatin modification related to transcriptional repression [
48]. Meanwhile, the PRC1 complex consist of four subunits, Cbx (Cbx2-4-6-7-8), RING1 (RING1A/B), PHC (PHC1-2-3) and PCGF (PCGF1-6) [
49]. The antagonistic actions between Polycomb and Trithorax are responsible for appropriate cell fate determination in epidermis while the differentiation-associated transcription factor GRHL3/GET1 recruits the Trithorax complex to several differentiation-associated genes [
50]. The synergistic or antagonistic interactions between PRC1/2 and histone modifications like H3K27me3, H3K36me3, H2A.Z, H3K4me1/2/3, H3K9me3 or H3K27ac, are proven essential to preserve stemness and regulate differentiation in epidermal stem cells.
Though covalently modification of histone tails, the PcGs turn to be obstacles of transcriptional activity. After nonenzymatic sticking to the chromatin, Ring1 in PRC1 mono-ubiquitylates histone H2A at lysine 119 [
51]. This modification is dispensable for its target binding but indispensable for efficient repression of target genes to maintain stemness of epidermal stem cells [
52]. There are two categories of PRC1 complexes been identified. One contains the Cbx subunit and depends on H3K27me3 for chromatin adherence. The other has RYBP and binds to the chromatin independent of PRC2 and H3K27me3, overturning the orthodoxy that recruitment of PRC1 to chromatin depends on its concerted action with PRC2 [
53]. Cbx4, another PRC1-associated protein, maintains human epidermal stem cells as slow-cycling and undifferentiated to protect the epidermal stem cells from senescence [
54].
The other polycomb complex PRC2 catalyzes trimethylation of H3K27me3, and its subunit Ezh2 prevents the differentiation of basal epidermal cells by precluding the binding of the pro-differentiation AP-1 transcription factor to late differentiation genes [
55]. Knockout of Ezh2 in embryonic basal epidermal stem cells totally eliminates the presence of H3K27me3 mark but moderately affect the fate determination of epidermal stem cells. The changes induced by knockout of Ezh2 could be found during embryogenesis and in early postnatal epidermis rather than in epidermis of an adult. Moreover, Ezh2 expression is decreased in adult epidermis. Different epidermal stem cells within one Polycomb-dependent tissue respond differently to loss of H3K27me3. Despite striking phenotypic difference of HF suspended morphogeny and epidermis excessive hypermorphosis, similar genes are up-regulated in HF and epidermal Ezh1/2-null progenitors. The PcG proteins and Ezh2 can be induced by Twist-1, resulting in an increase of H3K27me3 on the Ink4A/Arf locus [
56]. The proliferation or survival of Ezh1/2-null HF progenitors was restored after transduction of Ink4b/Ink4a/Arf shRNAs, suggesting the relevance of this locus to the HF phenotypes [
57].
Jarid2 relates to all of the known canonical PRC2 components by a conserved physical interaction with PRC2 and has a major effect on PRC2 recruitment. Besides,
in vivo studies revealed that Jarid2 mutants affected only H3K27me3 but no other histone modifications [
58]. However, the details of PRC1 and PRC2 recruited to genes are not fully apprehended. There is an evidence for an interaction of the transcription factor REST with PRC1 and RC2. REST has context-dependent functions for PRC1- and PRC2- recruitment and also function as a limiting factor for PRC2 recruitment at CpG islands [
59].
More than the role in the preservation of stemness, PcGs were recently found to be involved in the regulation of cell differentiation. Release from Polycomb repression only partially explains the activation of differentiation genes. Stable knockdown of SUZ12, a cornerstone for PRC2 assembly and function, leads to a significant precocious expression of a subset of terminal differentiation markers in intestinal cell models. This identifies a mechanism whereby PcG proteins participate in slow down terminal differentiation in the TA cell population [
60]. Similarly, loss of polycomb-mediated silencing may enable the upregulation of repair-related genes and stimulate the epidermal stem cells to initiate terminal differentiation [
61].
Generally, transcriptionally upregulated genes are marked by H3K36me3 in gene bodies, H3K4me3 and H3K9ac on promoters and H3K27ac and H3K4me1 in enhancer regions [
62]. Recent studies show that H3K36me3 affiliating to polycomb-like (PCL) proteins PHF19 leads to the recruitment of PRC2 and subsequently de novo gene silencing. Coexistence of H3K36me3, H3K27me3, and PHF19/PCL3 at a subset of poised developmental genes is identified in murine mutipotent stem cells. PHF19/PCL3 Tudor motif is required for the recognition of H3K36me3 to promote the intrusion of PRC2 complexes into active chromatin regions and consequently gene silencing [
63]. The combined activities of KDM5a (and possibly KDM5b), plus NO66 and/or KDM2b may remove both marks of transcriptionally active genes, H3K4me3 and H3K36me3, facilitating PcG-mediated silencing of previously active genes [
64].
In addition, histone variant H2A.Z plays essential roles in mediating nucleosome depletion and recruiting transcription cofactors to cis-regulatory elements [
65]. In an
in vivo mouse hair follicle stem cell model, H2A.Z shows specific immunodetection on immortal DNA chromosomes, indicating H2A.Z as an asymmetric self-renewal-associated (ASRA) biomarker. Its mRNA is significantly downregulated during asymmetric self-renewal compared to symmetric self-renewal [
66]. H2A.Z is highly enriched at promoters or enhancers and is required for both self-renewal and differentiation. In self-renewing stem cells, knockdown of H2A.Z compromises OCT4 binding to its target genes and leads to decreased binding of MLL complex to active genes and of PRC2 complex to repressed genes. H2A.Z can also accumulations at developmentally silenced genes in a polycomb independent manner [
67]. In addition, inhibition of H2A.Z also compromises RA-induced RARalpha binding, activation of differentiation markers, and the repression of pluripotency genes during differentiation of stem cells. Therefore, H2A.Z acts as a ying-yang facilitator to regulate the access of both activating and repressive transcriptional factors [
68].
In addition to histone methylation, histone acetylation also exerts significant influences in stem cells. Either histone acetyltransferase (HAT) or histone deacetylase (HDAC) dedicates to the network of chromatin modification in epidermal stem cells. MOZ (monocytic leukemia zinc-finger protein) and MORF (MOZ related factor) are catalytic subunits of HAT complexes essential in stem cell developmental programs. The canonical HAT domain of MORF/MOZ follows a tandem of plant homeodomain (PHD) fingers. The tandem PHD fingers of MORF recognize the
N-terminal tail of histone H3 while acetylation of Lys9 (H3K9ac) or Lys14 (H3K14ac) enhances this binding [
69]. During differentiation and survival of epidermal keratinocytes, peroxisome proliferator-activated receptors (PPARs) play a key role as well. p65/RelA represses PPARdelta-dependent transactivation in a HDAC activity dependent manner [
70]. EGFR-ERK pathway is also implicated in advanced stages of epidermal differentiation at least in part through modulating HDAC activity [
71].
Interacting with chromatin remodeling factors, transcription factors such as p63 and Myc have some modulatory impacts on the differentiation and proliferation of epidermal stem cells. IFE and HF develop from surface ectodermal progenitor cells. Deletions of ectodermal HDAC1 and HDAC2 result in failure of HF specification and epidermal stratification, paralleling with the loss of the important ectodermal transcription factor p63 [
72]. P63 makes use of several chromatin-remodeling proteins to exert its function in epidermal embryogenesis and adult epidermal homeostasis. To repress expression of anti-proliferative genes, p63 requires HDAC1/2 and the meCpG stabilizer Lsh/HELLS [
73]. The exiting of epidermal stem cells from their niches and terminal differentiation is dependent on activation of Myc and subsequently chromatin modifications. Quiescent stem cells in the IFE and the hair follicle bulge have high levels of H3K9ac or H3K14ac. With the aid of HDAC activities, Myc can initiate the increased acetylation at histone H4 and mono-methylation at lysine 20 which then switches to chromatin silencing epigenetic modifications di-methylation at histone H3 lysine 9 and histone H4 lysine 20 [
74]. In addition, an androgen receptor/p53 axis is also involved in c-MYC-induced sebaceous gland differentiation [
75].
2.2. DNA Modification and Epidermal Stem Cells
Another category of epigenetic modification in epidermal stem cells takes place on the genomic DNA. 5-Methylcytosine (5-mC) is identified as a crucial and prevalent epigenetic DNA modification involved in the development [
76]. Genomic methylation patterns in somatic differentiated cells are generally stable and heritable, ensuring tissue-specific gene expression in a heritable manner throughout development. Therefore, methylation of DNA at CpG islands, which is associated with transcriptional repression, is regarded as hereditary imprints [
77]. De novo methylation of DNA is performed by DNA (cytosine-5-)-methyltransferase 3a (DNMT3a) and DNMT3b while DNMT1 is responsible for methylation maintenance [
78].
While early studies reveal that DNMT1 is dispensable for embryonic stem cell maintenance [
79], recent studies focused on whether DNMT1 makes a determinate action for maintaining the progenitor state in the epidermis [
80]. Paralleling with a role in the spatial-upward epidermis differentiation process, DNMT1 is mainly confined to cells in the basal layer of adult epidermis where epidermal stem cell located and is lost in outer differentiated layers. DNMT1 knockdown leads to premature differentiation within the progenitor-containing compartments and defects in tissue self-renewal [
80]. Genome-wide analysis revealed that a significant portion of epidermal differentiation gene promoters are methylated in self-renewing. All of these evidences point to a fact that DNMT1 is essential for the function of epidermal stem cell, mainly in preservation of progenitor state and trigger of exit from the niche [
81]. UHRF, a component of the DNA methylation machinery that targets DNMT1 to hemi-methylated DNA, is also necessary to suppress premature differentiation and sustain proliferation [
82]. Lsh, a chromatin remodeling protein, can interact directly with DNMT3a and DNMT3b or indirectly with DNMT1 via DNMT3b [
83] and participate in DNA methylation [
84]. Mice with a K14-Cre-mediated loss of DNMT1 show an uneven epidermal thickness, alterations in hair follicle size and shorter and thinner hair fibers [
85]. These results highlight the significance of DNA methylation in HF regeneration.
As 5-methylcytosine (5-mC) may be further modified by TET family [
86] to 5-hydroxy-methylcytosine (5-hmC) in embryonic stem cells [
87], it provides a new avenue to look into the epigenetic modulation of stem cells as a potential mechanism leading to active DNA demethylation. Enrichment of 5-hmC at enhancers marked with H3K4me1 and H3K27ac suggests 5-hmC is probable implicated in the regulation of some specific promoters and enhancers related to stem cell function [
88]. 5-hmC modification is prevalent in embryonic stem cells and neurons [
89], but the distribution of 5-hmC in epidermis has not been rigorously explored. High levels of 5-hmC are reported to be lost during differentiation, but reappear during the generation of induced pluripotent stem cells, suggesting that 5-hmC enrichment associates with a pluripotent cell state [
90]. As the neurons and epidermis have some common in the tissue origination and the above-mentioned remarkable dynamic change of 5-mC during the epidermal stem cell exiting from the niches, it is reasonable to make an audacious speculation that if some intriguing changes of 5-hmC would arise in epidermal stem cells process. In differentiated nonreplicating somatic cells, DNA methylation can also be reversible although it is still unknown whether active demethylation is involved [
91].
2.3. Noncoding RNAs and Epidermal Stem Cells
It is less than ten years when noncoding RNAs including micro-RNAs (miRNA) and long noncoding RNAs (lncRNAs) have been appreciated to regulate epidermal stem cell identity, proliferation and differentiation. In 2006, Yi
et al. cloned more than 100 miRNAs [
92] from skin and showed that epidermis and HF differentially express discrete miRNA families. MiRNAs orchestrate the formation of epidermis and skin appendages. The multifunctional enzyme Dicer is required in the processing of miRNAs and their assembly into the RNA-induced silencing (RISC) complex. Its deletion in embryonic skin progenitors has an impact on developing hair germs to evaginate layers rather than invaginate, thereby perturbing the epidermal organization. In the Dicer mutant, no normal hair shafts are produced, together with the degenerated follicles lacking stem cell markers although the epidermis is hyperproliferative. These results implicate the vital roles for miRNAs in epidermis and HF development [
93]. Several miRNAs, particularly
miR-203, have been confirmed to be crucial to the differentiation and proliferation in epidermis primarily through directly or indirectly targeting epigenetic regulators or transcription factors.
MiR-203 is the well-studied miRNA in epidermis biology. It is upregulated in cultured keratinocytes after differentiation inducers treatment [
94,
95].
MiR-203 promotes epidermal differentiation by restricting proliferative potential and inducing cell-cycle exit. The targets of
miR-203 include suppressor of cytokine signalling-3 (SOCS-3) and p63.
MiR-203 starts to express in human foetal skin at week 17 and is most prominent in the suprabasal layers of the epidermis while p63 and SOCS-3 are preferentially expressed in the basal layer [
96]. Differentiation-induced upregulation of
miR-203 expression is blocked by treatment of protein kinase C (PKC) inhibitor. The activator protein-1 (AP-1) proteins c-Jun and JunB regulate
miR-203 expression in keratinocytes.
Several studies are looking into the interaction between p63 and miRNAs in epidermis. In the absence of p63,
miR-34a and
miR-34c are increased in primary keratinocytes with concomitant G1-phase arrest, inhibition of cyclin D1 and cyclin-dependent kinase 4 (Cdk4). P63 inhibits the expression of
miR-34a and
miR-34c through directly binding to p53-consensus sites in regulatory regions of both genes [
97]. In addition, the miR-17 family including
miR-17,
miR-20b,
miR-30a,
miR-106a,
miR-143, and
miR-455-3p are regulated by p63 [
98].Multiple MAPKs, pRb and p21 are the direct target of these p63-regulated miRNAs. Inhibition of these miRNAs led to the defects in early keratinocyte differentiation. Thus, both p63-regulating miRNAs like
miR-203 and p63-regulated miRNAs constitute a feedback circuit in the regulation of epidermal differentiation.
As an inhibitory member of the ASPP (apoptosis stimulating protein of p53) family, iASPP comprise an autoregulatory feedback loop with p63 through microRNAs like
miR-574-3p and
miR-720. IASPP regulate epithelial integrity program by regulates the expression of genes essential for cell adhesion through affecting the expression of
miR-574-3p and
miR-720 that directly target p63. Silencing of iASPP in keratinocytes by RNA interference promotes and accelerates a differentiation pathway [
99]. In addition to p63, other epigenetic regulators like PcGs are also regulated by miRNAs to determine the proliferation and differentiation of epidermal stem cells. For instance,
miR-137 [
100] and
miR-26a [
101] repress the expression of EZH2 which is appreciated as a crucial role in epidermal homeostasis and differentiation.
An additional
let-7 target
mLin41 (mouse homologue of lin-41) presents in epidermal stem cell niches. It is a potential contributor to the mutipotency regulatory circuit of
let-7 miRNA and its target gene
Lin-28. MLin41 interferes with silencing of target mRNAs by
let-7 and cooperates with the mutipotency factor Lin-28 in suppressing let-7 activity, revealing a dual control mechanism regulating let-7 in epidermal stem cells [
102].
When most studies concentrate on regulation of keratinocytes [
103], some studies begin to check into other component of epidermis. One study shows that
miR-146a is constitutively expressed at higher levels in human Langerhans cells (LC). It is induced by the transcription factor PU.1 in response to TGF-beta1, a key microenvironmental signal for epidermal LC differentiation. Besides, high constitutive
miR-146a levels might render LCs less susceptible to inappropriate activation by commensal bacterial TLR2 triggers at body surfaces [
104].
In addition to miRNAs, lncRNAs are postulated to play important roles in epidermal stem cells. Using transcriptome sequencing and tiling arrays to compare lncRNA expression in epidermal progenitor populations
versus differentiating cells, researchers identified ANCR (anti-differentiation ncRNA) as an 855-base-pair lncRNA down-regulated during differentiation. Depletion of ANCR in epidermal stem cells leads to rapid induction of differentiation genes [
105], indicating that the ANCR lncRNA is responsible for the maintenance of undifferentiated state of epidermal stem cells.
Functionally contrary to ANCR, a human lncRNA terminal differentiation-induced ncRNA (TINCR) has been characterized recently. TINCR-deficient epidermis lacks terminal differentiation ultrastructure, including keratohyalin granules and intact lamellar bodies. Through a post-transcriptional mechanism to ensure differentiation mRNAs expression, TINCR controls human epidermal differentiation. TINCR interacts with a range of differentiation mRNAs including FLG, LOR, ALOXE3, ALOX12B, ABCA12, CASP14 and ELOVL by TINCR-mRNA interaction occurring through a 25-nucleotide “TINCR box” motif. Direct binding of TINCR to the staufen1 (STAU1) protein is required for skin differentiation [
106].
2.4. Epigenetic Regulations of Epidermal Stem Cells and Skin Disorders
As discussed above, dynamic epigenetic regulation has been proved crucial in epidermal stem cell differentiation and homeostasis maintaining. Recently, contribution of epigenetic changes and epidermal stem cells to skin disorders has attracted increasing attentions [
107]. The development of cancer is a multistage long process resulting from the accumulation of genetic and epigenetic changes in the stem or progenitor cells [
108]. Similarly, epidermal stem cells or melanocyte precursors undergo several epigenetic and genomic changes probably induced by UV (ultra violet) to eventually develop into melanoma cells. These changes provide stem cells with abilities to avoid apoptosis, replicate without limitation, sustain angiogenesis and invade and spread to other sites [
108,
109]. UV present in the sunlight exhibits deleterious effects on DNAs. Interestingly, methylated cytosines are much more susceptible to UV-induced cyclobutane pyrimidine dimers (CPDs) formation than unmethylated ones [
110]. As a consequence of epigenetic and genetic changes, many signaling pathways important to the function of epidermal cells are dysregulated in melanoma cells. For example, murine double minute 2 (MDM2)-p53 pathway is important to maintain the number and function of epidermal stem cell [
111]. The increase of p53 resulted from the deletion of MDM2 in the epidermis induced the loss of epidermal stem cell function and consequent aging phenotypes such as thin epidermis, retarded wound healing and a progressive loss of fur. During melanoma development, this pathway was inactivated by the loss of p53 function although the point mutation of p53 was relatively infrequent, indicating the potential implication of the epigenetic inactivation of p53 signaling. Indeed, MDM2-targeting
miR-18b was epigenetically downregulated in human melanoma specimens and cell lines [
112]. Stable overexpression of
miR-18b greatly suppressed melanoma cell viability, cell migration and invasiveness, reversed epithelial-to-mesenchymal transition (EMT), and inhibited tumor growth
in vivo. Meanwhile, some other well-known tumor suppressor genes such as PTEN and RARB are inactivated by DNA methylation in melanoma [
113].
Defects in the epigenetic regulation of epidermal stem cells also contribute to the development of non-malignant skin diseases. Some characteristic cutaneous features present in patients with disruption of gene imprinting. These diseases include Beckwith–Wiedmann, Silver–Russell, Prader–Willi, McCune–Albright and Angelman syndromes, Albright’s hereditary osteodystrophy, progressive osseous heteroplasia, Von Hippel–Lindau syndrome, hypomelanosis of Ito and dermatopathia pigmentosa reticularis [
114]. More common diseases include eczema and psoriasis may also have predominantly maternal or paternal modes of transmission. In addition to gene imprinting, other epigenetic changes are also relevant to the pathogenesis of skin diseases. For example, overexpression of
miR-31 contributes to skin inflammation in psoriasis lesions by regulating the production of inflammatory mediators and leukocyte chemotaxis to the skin [
115]. Psoriasis can be characterized by a specific microRNA expression profile and
miR-31 is highly upregulated microRNAs in psoriasis skin by TGF-β (Tumor growth factor-β) which was predominant in psoriasis epidermis. It directly targets serine/threonine kinase 40 (STK40) and inactivates NF-κB signaling. Inhibition of
miR-31 activates NF-κB signaling and promotes the expression of many inflammatory cytokine or chemokines such as interleukin-1β (IL-1β), IL-8 and epithelial-derived neutrophil-activating peptide 78. In addition, the mammalian homolog of Grainyhead in Drosophila, Grainyhead-like 2 (GRHL2) regulates telomerase (hTERT) expression in epidermal cells. Interestingly, GRHL2 was also upregulated in chronic skin lesions such as psoriasis. Exogenous GRHL2 expression inhibited the recruitment of histone demethylase Jmjd3 to enhance the level of histone 3 Lys27 trimethylation enrichment at promoters and suppress the expression of epidermal differentiation complex genes (EDC) such as IVL, KRT1, FLG, LCEs, and SPRRs [
116].