DNA methylation in stem cell renewal and multipotency

Owing to their potential for differentiation into multiple cell types, multipotent stem cells extracted from many adult tissues are an attractive stem cell resource for the replacement of damaged tissues in regenerative medicine. The requirements for cellular differentiation of an adult stem cell are a loss of proliferation potential and a gain of cell-type identity. These processes could be restricted by epigenetic modifications that prevent the risks of lineage-unrelated gene expression or the undifferentiated features of stem cells in adult somatic cells. In this review, we focus on the role of DNA methylation in controlling the transcriptional activity of genes important for self-renewal, the dynamism of CpG methylation of tissue-specific genes during several differentiation programs, and whether the multilineage potential of adult stem cells could be imposed early in the original precursor stem cells through CpG methylation. Additionally, we draw attention to the role of DNA methylation in adult stem cell differentiation by reviewing the reports on spontaneous differentiation after treatment with demethylating agents and by considering the evidence provided by reprogramming of somatic cells into undifferentiated cells (that is, somatic nuclear transfer or generation of induced pluripotent cells). It is clear from the evidence that DNA methylation is necessary for controlling stem cell proliferation and differentiation, but their exact contribution in each lineage program is still unclear. As a consequence, in a clinical setting, caution should be exerted before employing adult stem cells or their derivatives in regenerative medicine and appropriate tests should be applied to ensure the integrity of the genome and epigenome.


Introduction
Multipotent stem cells extracted from many adult tissues are an attractive stem cell resource for the replacement of damaged tissues in regenerative medicine and have been identifi ed in many organs and tissues, including bone marrow, peripheral blood, fat, skeletal muscle, brain, skin, cornea, heart, gut, liver, ovarian epithelium, and testis. Multipotent stem cells are all defi ned as undiff eren tiated cells, are able to self-renew with a high proliferative rate, and have the potential to diff erentiate into specialized cells with specifi c functions [1]. Unlike pluripotent embryonic stem (ES) cells, multipotent stem cells are usually restricted to a particular lineage (mesodermal, endodermal, or ectodermal) but have the potential to diff erentiate into distinct somatic cell types with appropriate stimulation (Figure 1). Two main advantages for their use in clinical applications are that they avoid some ethical issues associated with pluripotent ES cells, resulting in a more timely approval for research and thera peutic use, and that adult stem cells and tissues derived from them are currently believed to be less likely to initiate rejection after transplantation.
Although human adult stem cells represent a promising tool for applying new clinical concepts in support of cellular therapy, many aspects remain to be explored in order to guarantee appropriate quality assurance and control of these cells, such as avoiding inappropriate gene expression in transplanted cells or the undesirable traits of tumorigenesis. Gene expression potential in stem cell renewal and diff erentiation could be regulated by epigenetic processes that confer a specifi c chromatin conformation of the genome, of which DNA methylation is the best characterized ( Figure 1) [2]. DNA methylation, the addition of a methyl group to the carbon 5 of the cytosine into CpG contexts, is known to be an essential process in development and cellular diff erentiation [3]. It is involved in gene regulation of housekeeping and tissue-type genes, silencing of one allele of imprinted genes, and compensation of the extra copy of the X chromosome in females. It acts as a defense mechanism, preventing genomic instability due to transposon movements or insertion of endoparasitic sequences in the genome [4]. It must be pointed out that DNA methylation does not work alone

Abstract
Owing to their potential for diff erentiation into multiple cell types, multipotent stem cells extracted from many adult tissues are an attractive stem cell resource for the replacement of damaged tissues in regenerative medicine. The requirements for cellular diff erentiation of an adult stem cell are a loss of proliferation potential and a gain of cell-type identity. These processes could be restricted by epigenetic modifi cations that prevent the risks of lineage-unrelated gene expression or the undiff erentiated features of stem cells in adult somatic cells. In this review, we focus on the role of DNA methylation in controlling the transcriptional activity of genes important for self-renewal, the dynamism of CpG methylation of tissue-specifi c genes during several diff erentiation programs, and whether the multilineage potential of adult stem cells could be imposed early in the original precursor stem cells through CpG methylation. Additionally, we draw attention to the role of DNA methylation in adult stem cell diff erentiation by reviewing the reports on spontaneous diff erentiation after treatment with demethylating agents and by considering the evidence provided by reprogramming of somatic cells into undiff erentiated cells (that is, somatic nuclear transfer or generation of induced pluripotent cells). It is clear from the evidence that DNA methylation is necessary for controlling stem cell proliferation and diff erentiation, but their exact contribution in each lineage program is still unclear. As a consequence, in a clinical setting, caution should be exerted before employing adult stem cells or their derivatives in regenerative medicine and appropriate tests should be applied to ensure the integrity of the genome and epigenome.
in controlling chromatin conformation since histone modifi cations and non-coding RNA regulation also collaborate in its control. So, we must consider the existence of an 'epigenetic code' in which several epigenetic factors act in a gradual and progressive manner for controlling chromatin structure.
Currently, much attention is being paid to the eff ects of CpG methylation on stemness and diff erentiation. Th e fi rst piece of evidence came from the observation that important genes for the maintenance of ES cells, such as Oct4 and Nanog genes, are usually hypomethylated when activated but became hypermethylated during diff erentiation [5,6]. Knowledge of the genome-wide contribution of CpG methylation to stem cell maintenance and diff erentiation has increased in recent years, mainly because of the development of technical approaches for assessing epigenetic factors. High-throughput strategies demonstrate that human ES cells have a unique CpG methy lation signature that, in combination with histone modi fi cations, drives stem cell diff erentiation through the restriction of the developmental potential of progenitor cells [7,8]. In comparison with the broadranging infor mation obtained from ES cells, the role of CpG methy lation in regulating diff erentiation of adult stem cells has been less extensively examined. In this review, we consider the reported evidence of how the developmental potential of adult stem cells could be restricted by the gain of DNA methylation of self-renewal genes (prevent ing the undiff erentiated features of stem cells in adult somatic cells) and the DNA methylationdependent control of tissue-specifi c genes (abolishing the risks of lineage-unrelated gene expression). Th e opportunities that this presents for manipulating the epigenome by means of pharmacological treatments and its consequences for stem cell diff erentiation and reprogram ming will be analyzed.

DNA methylation, global chromatin context, and stemness
It is important to point out that the relationship between promoter DNA methylation and promoter activity depends on the CpG content of the promoters: high CpG promoters (HCPs), intermediate CpG promoters, or low CpG promoters (LCPs). In ES cells and multipotent progenitor cells, HCP promoters are characterized by low DNA methylation levels, whereas LCP promoters are enriched in DNA methylation [6,8,9] (Figure 2). Furthermore, specifi c histone modifi cations (that is, H3K4me3 and H3K27me3) in HCPs appear to be more decisive for expression of the corresponding genes and suggest a degree of protection from DNA methylation [10] ( Figure 2). Conversely, methylated LCP promoters are depleted of bivalent histone marks and are mostly repressed in ES cells [6,8,9] ( Figure 2). It is suggested that silencing of pluripotency-related genes occurs by means of CpG promoter hypermethylation, whereas gain of diff erentiation features is defi ned by gene regulation of Polycomb targets [8]. Specifi c epigenetic features at a global level also underpin the pluripotency of ES cells. Recent studies have demonstrated that ES cell chromatin is in a highly dynamic state with global DNA hypomethylation and a general abundance of transcriptionally active chromatin marks such as H3K4me3 and acetylation of histone H4, which is refl ected in the relatively decondensed chromatin of ES cells [2,11]. Th is global lack of DNA methylation in stem cells could be associated with the ability of such cells to activate a wide range of cell typespecifi c genes during the diff erentiation pro grams [2]. It must not be forgotten that DNA methylation and histone modifi cations do not work alone and that the epigenetic inactivation of diff erentiation-specifi c genes in stem cells (that is, Hox and Pax family of genes) is usually repressed by alternative chromatin remodeling factors, such as Polycomb proteins [11,12]. Consequently, further study of the interplay of all of the chromatin regulators is essential for understanding the dynamism of trans criptional control during stem cell renewal and diff erentiation.

DNA methylation-dependent regulation of genes associated with self-renewal of stem cells
It has been widely reported that maintenance of the pluripotency state is conferred by a set of developmentassociated transcription factors -such as OCT4, NANOG, and SOX2 -that occupy promoters of active genes associated with self-renewal [13,14]. Expression of the aforementioned transcription regulators is usually controlled by CpG promoter methylation, and diff erentiation of ES cells is accomplished by partial or full methylation of pluripotency-associated genes, resulting in their downregulation [6,[15][16][17]. Th e opposite association has been found in the reprogramming of induced pluripotent stem (iPS) cells from diff erentiated cells, in which unmethylated active promoters of ES cell-specifi c genes were described [18] (Figure 2). Despite the consider able information about silencing of pluripotency ES genes during diff erentiation, very little is known about the epigenetic control of genes associated with selfrenewal and maintenance of multipotent adult stem cells. In adipose-derived stem cells (ASCs) and mesenchymal stem cells from bone marrow (BM-MSCs), OCT4 is silenced by promoter hypermethylation, whereas Nanog and Sox2 are unmethylated despite the repressed state of the genes [19]. Th e same patterns of methylation were found in diff erentiated fi broblasts and keratinocytes [19]. It seems that, whereas Oct4 regulation is strongly infl uenced by CpG promoter hypermethylation, the control of Nanog and Sox2 expression could be due to other repressive mechanisms such as histone modifi cation patterns [19]. Enrichment of H3K27me3 and H3K9me3 and reduction of H3K79me3 have been described in the Nanog and Sox2 promoters of ASCs and diff erentiated cells but not in pluripotent cells [20]. Th ese results demonstrate that the transcriptional repression mechanisms could vary depending on the gene and the state of cellular diff erentiation (that is, multipotency versus diff er en tiation) [19] and could constitute a mechanism for preventing aberrant reactivation of pluripotency and minimizing the risk of de-diff erentiation [21]. In line with this hypothesis, ES cells with genetic mutations of DNA methyltransferase result in rapid apoptosis-mediated cell death [22,23].
Th e promoter methylation status of additional stem cell-determining genes for self-renewal (not exclusively markers of pluripotency) has also been investigated [24]. Silencing of the mesodermal transcription factor Brachyury gene during diff erentiation from BM-MSCs to mesodermal lineages involves hypermethylation of its promoter but not changes in promoter hypermethylation of genes such as LIN28, NESTIN, or ZFP42. Th is could be associated with changes of expression during diff erentiation of BM-MSCs [24]. Currently, we have a limited understanding of how multipotency is established and maintained in adult stem cells, and it would be very interesting to study the CpG promoter methylation status of transcription factors that confer multipotency on adult stem cells beyond the traditional role of pluripotency genes such as Oct4, Nanog, and Sox2.   [3]. However, the role of specifi c promoter methylation in controlling gene diff erentiation remains a matter of controversy. On one hand, there are some clues in favor of the hypothesis that cell type-specifi c patterns of DNA methylation infl uence cell type-specifi c gene expression and, by extension, cellular diff erentiation. For example, promoter methylation of SERPINB5 is inversely correlated with the unique expression of SERPINB5 in epithelial cells [25], and the rSPHK1 and hSLC6A8 promoter hypermethylation associated with gene silencing in specifi c tissues allows expression in unmethylated brain tissue only [26,27]. On the other hand, genome-wide analysis of CpG methylation changes during the conversion of human pluripotent/multipotent stem cells into diff erentiated somatic cells reveals small changes in DNA methylation at promoter regions [8,9,[28][29][30]. For example, lineage commitment of neural progenitor cells into terminally diff erentiated neurons occurs with a very moderate number of promoter DNA hypermethylated genes as cells diff erentiate [8]. Further work is needed to test whether these weak associations between gene repression and CpG hypermethylation during diff erentiation are due to limitations of the analytical techniques or to the existence of additional methylation-independent regulatory mechanisms.

Does CpG methylation of multipotent stem cells restrict lineage specifi cation?
One of the main features of adult stem cells is their multipotency (that is, their ability to diff erentiate into a number of cell types), but, in contrast to pluripotent cells, they are restricted to those of a closely related family of cells. For example, BM-MSCs primarily form mesodermalspecifi c cell types such as chondrocytes, myocytes, adipo cytes, or osteoblasts [1]. However, we should remember that, given the information collected in recent years, this could be a very general statement, and there is some evidence to suggest that lineage restriction could be more permissive. For instance, BM-MSCs could be diff eren tiated into cells of all three germ layers and generate tissues such as osteocytes (mesoderm), hepatocytes (endoderm), or neurons (ectoderm) [31][32][33]. Multipotent cells isolated from diff erent tissues have common in vitro phenotypic and functional characteristics (for example, MSCs share fi broblast-like morphology, plastic adherence, proliferation ability, and clonogenicity) but diff er in the expression of specifi c lineage markers (for example, ASCs and BM-MSCs diff er in the expression of the surface markers CD90, CD105, CD106, and adhesion molecules [34,35] and in their diff erentiation potential). Since gene expression in adult stem cells is regulated by epigenetic processes, a question arises: is the diff er entiation potential in adult stem cells predicted by DNA methylation of specifi c lineage promoters? Th ere is some evidence in favor of a diff erentiation restriction imposed by promoter hypermethylation in progenitor stem cell states, whereas promoter hypomethylation does not have any predictive value with respect to diff erentiation potential [35,36]. Characterization of DNA methylation profi les of all human RefSeq promoters in mesenchymal adult stem cells from various origins, including adipose, hematopoietic, and neural progenitors and muscle tissue, shows that the majority of the lineage-specifi c genes are hypomethylated even if the progenitor is not able to diff erentiate into this specifi c lineage [10]. Th ere are some examples of epigenetic silencing associated with restriction to diff erentiation: endothelial markers such as CD31 and CD144 are strongly methylated in ASCs that show very limited capacity for endothelial diff erentiation [36] or osteogenic and adipogenic restriction of C2C12 myoblast cell line diff erentiation [37]. Furthermore, the restriction for diff erentiation in specifi c programs imposed by means of DNA methylation is established early in development, in the progenitor state, and persists after diff erentiation, as most of the hypermethylated promoters in undiff erentiated cells remain hypermethylated in somatic cells [10,37]. Th is is in agreement with the low level of de novo methylation described after diff erentiation of adult stem cells [8,9]. Results lead to the conclusion that the diff erentiation restriction associated with promoter hypermethylation clearly diff ers between pluripotent and multipotent cells: lineage-specifi c promoters are mostly hypermethylated in ES cells [6] in contrast to the low-percentage hypermethylation found in MSCs [35].

Treatment with demethylating agents results in spontaneous diff erentiation
Th e involvement of DNA methylation in controlling the diff erentiation potential of stem cells has been supported by several reports of spontaneous diff erentiation after treatment with demethylating agents (Table 1). For example, the use of 5-aza-2'-deoxycytidine (5-ADC) promotes diff erentiation of ASCs into cardiac myogenic cells [38]. Pretreatment with 5-ADC also drives the osteogenic diff erentiation of BM-MSCs by enhancing the expression of osteogenic genes (such as Dlx5) associated with demethylation of its CpG shore [39,40]. However, we must remember that DNA methylation is just one compo nent of the epigenetic machinery and that removing DNA methy lation is often insuffi cient to reactivate gene expression (Table 1). Treatments with the histone deacetylase (HDAC) inhibitor trichostatin A (TSA) enhance chon dro genic diff erentiation of BM-MSCs accomplished by increased expression of Sox9 [41]. Similarly, neural induction was achieved when multipotent stem cells were exposed to TSA simultaneously with 5-ADC [42]. Furthermore, the eff ects of treatments with HDAC inhibitors are known to alter DNA methylation levels [41,42]. Additional evidence for the role of epigenetic control in diff erentiation comes from the functional consequences of defects in enzymes of the epigenetic machinery. For example, recovery of the expression of a defect in a histone modifi er (NSD1) suppresses cell growth and increases the diff erentiation of neuro blastoma cell lines [43]. Recovery of epigenetic patterns, by treatment with epigenetic drugs or by genetic models, highlights the potential of epigenetic modifi ers, possibly in combination with other factors, to enhance the ability of multipotent stem cells to form functional diff erentiated cells and has signifi cant therapeutic implications. Some consistent lines of evidence support this therapeutic application since epigenetic drugs, among them demethy lating agents, have shown signifi cant antitumor activity and the US Food and Drug Administration has approved the use of some of them to treat patients with cancer [4]. Indeed, new HDAC inhibitors (that is, romidepsin, belinostat, or givinostat) that are currently being tested in clinical trials for anticancer therapy [44] must also be considered as candidate molecules for assessing stem cell diff eren tiation. Further understanding of the epigenetic regula tion of tissue-specifi c genes along with the development of additional specifi c epigenetic drugs may hold the key to our ability to reset the epigenome successfully during stem cell diff erentiation.

Concluding remarks
It is clear that cell diff erentiation of multipotent stem cells is a result of a complex and dynamic network of transcriptional regulators, among them epigenetic factors that play a central role through controlling the expression/ repression of tissue-specifi c genes and multipotencyrelated genes. However, it is not currently possible to manipulate cell diff erentiation even if we consider all of the genetic and epigenetic knowledge available for a specifi c lineage commitment. For example, epigenetic treatments may have a pleiotropic eff ect on the diff erentiation of stem cells, depending on multiple factors, mainly the origin of the precursor cell and environment conditions (presence of growth factors, transcriptional regulators, and so on) [39,41,45], suggesting that global epigenetic modifi cations, though necessary, are not sufficient to transdiff erentiate by themselves [46]. Th ese fi ndings underline the necessity of evaluating in more detail the importance of epigenetic chromatin remodel ing for establishing and maintaining stemness or, on the other hand, initiating a diff erentiation program. Th e reprogram ming of somatic cells provides a new oppor tunity to study the contribution of epigenetics to diff erentiation. A mature cell can be converted into a pluripotent state by three experimental approaches: somatic nuclear transfer into enucleated oocytes, the in vitro application of a defi ned set of transcription factors creating iPS cells, or fusing ES cells with somatic cells to generate heterokaryons and hybrids [47]. Epigenetic rearrange ments are observed independently of the technique [48,49]. In fact,

Multipotent stem cells
Adipose-derived stem cells 5-aza-2'-deoxycytidine Cardiomyocytes [38] Adipose-derived stem cells 5-aza-2'-deoxycytidine; trichostatin A Cardiomyocytes [57] Bone marrow mesenchymal stem cells 5-aza-2'-deoxycytidine Osteocytes [39,40] Bone marrow mesenchymal stem cells 5-aza-2'-deoxycytidine Cardiomyocytes [45,58] Bone marrow mesenchymal stem cells 5-aza-2'-deoxycytidine; trichostatin A Osteocytes; chondrocytes [41] Bone marrow mesenchymal stem cells 5-aza-2'-deoxycytidine; trichostatin A Neural-like cells [42] Bone marrow mesenchymal stem cells Sodium butyrate Osteocytes [59] Cardiac progenitor stem cells 5-aza-2'-deoxycytidine Cardiomyocytes [60] Neural progenitor stem cells Trichostatin A Neuronal cells [61] Neural progenitor stem cells Valproic acid Neuronal cells [62] Umbilical cord mesenchymal stem cells 5-aza-2'-deoxycytidine Cardiomyocytes [63] Pluripotent stem cells Embryonic stem cells 5-aza-2'-deoxycytidine Cardiomyocytes [64,65] Embryonic stem cells 5-aza-2'-deoxycytidine Endothelial cells [66] Embryonic stem cells Trichostatin A Cardiomyocytes [67,68] Berdasco there is evidence that HDAC inhibitors and DNA demethylating agents are useful for enhancing iPS reprogramming [50,51]. A prerequisite in reprogramming of iPS from somatic cells is that some stemness-related promoters become demethylated. How might this demethylation be achieved? It could be done through a DNA repair mechanism [52,53] or by the recent discovery of TET proteins, a group of enzymes that convert methylated 5-methylcytosine to 5-hydroxy methylcytosine [54]. Although experimental models for reprogram ming have generated a considerable amount of infor mation, many questions remain. How diff erent is the epigenetic regulation of pluripotent and multipotent cells? Does CpG methylation underpin self-renewal in adult stem cells, as it does in ES cells? Do epigenetic marks defi ne the lineage potential of an adult stem cell? Is it possible to revert the diff erentiation program by manipu lating the epigenome? How safe is this reversion? Th e recent discovery that nearly one quarter of all methylation identifi ed in ES cells was found in a non-CG context [55] suggests that the genomic context must also be addressed. Do ES cells use a diff erent methylation mechanism for gene regulation? Furthermore, long-term in vitro culture of adult stem cells, a prerequisite for large-scale expansion previous to implantation with thera peutic purposes, showed specifi c alterations of CpG island methylatyion [56]. As a consequence, it is necessary to optimize and standardize the experimental protocols used for in vitro expansion which minimize epigenetic-related instability. In conclusion, although manipulation of epigenetic activity might be an interesting means of generating populations of specifi c cell types, additional epigenetic research on the understanding of stem cell biology must be done before they can be used as diff erentiation agents in stem cell-based therapies.