Epigenetic chromatin remodeling plays a pivotal role in normal mammalian development and post-natal tissue homeostasis. Indeed, lineage specification and cellular differentiation, which underlie embryo development and morphogenesis from a single pluripotent stem cell, are epigenetically regulated processes. The final result is the “plasticity” of an individual genotype that, through the activation of molecular cascades, timely and sequentially controlled, produces different phenotypes in response to different microenvironments. In the last 10 years, special attention has been paid to the non-protein coding portion of the genome such as non-coding small RNAs, among which are microRNAs (miRNAs), considered to be major regulators of developmental pathways [1–8]. Of note, chromatin remodeling and miRNA pathways have been shown to be interconnected and able to regulate each other. To date, it is recognized that the deregulation of the epigenetic- and miRNA-dependent control of gene expression underlies tumorigenesis.

Rhabdomyosarcoma (RMS) and neuroblastoma (NB) are pediatric cancers derived from cells sharing molecular features of skeletal muscle and neuronal progenitors, respectively, blocked at different stages of differentiation. It has been clearly demonstrated that deregulation of developmental pathways plays a major role in both tumors.

The present review will focus on current knowledge about miRNAs deregulated in RMS and NB by epigenetic modifications. Their role in developmental pathways and RMS and NB tumorigenesis will be highlighted. Moreover, the translational implications and challenges of miRNAs modulation in these pediatric tumors will be discussed.

The term “epigenetics” describes cellular modifications caused by mechanisms other than DNA sequence variations that can be heritable and modified by environmental stimuli [9]. Therefore, in the present review, we will refer to epigenetics to describe the chemical reactions of chromatin that affect its accessibility in order to switch “on” and “off” gene expression.

To fit within the nucleus of eukaryotic cells, short lengths of DNA are wrapped around an octamer of histone proteins (two copies of core histones H2A, H2B, H3 and H4) forming the basic units of chromatin termed “nucleosomes” (Figure 1).

Higher-order organizations are formed by further compaction of chromatin structure. The degree of packaging influences the chromatin accessibility to transcriptional-regulatory complexes switching a gene “on” or “off.” Therefore, the “epigenome” encompasses genetic information that is the result of multiple processes involving chromatin modifications such as DNA methylation, histone proteins modification, histone variants replacement, nucleosome repositioning and mechanisms involving non-coding RNAs function. All these epigenetic modifications can be rapidly resumed and, therefore, fit well for the purpose of fine tuning the timely controlled developmental processes.

DNA methylation is catalyzed by DNA methyltransferases (DNMTs) and results in de novo methylation of unmethylated DNA and/or methylation maintenance of hemimethylated sequences. In normal tissues, DNA methylation is typically present and stable in the intergenic regions. In cancer cells, DNA methylation is exclusively found at the level of cytosines within CpG-rich regions of gene promoters leading to gene silencing. On the contrary, DNA hypomethylation of these CpG islands, also often aberrant in cancer, can increase gene expression [10]. The final result is either the silencing of tumor suppressor genes or the transcriptional activation of proto-oncogenes. In addition, DNA hypomethylation can lead to DNA helix breakpoints and, ultimately, to loss of heterozygosity (LOH) or aberrant chromosomal rearrangements. In strict conjunction with these mechanisms, the histone code generated by covalent modifications on histone tails, regulates chromatin remodeling for the accessibility of the transcription machinery to genes up to DNA repair, replication and segregation [11]. The major modifications of histone tails are controlled by histone acetyltransferases (HAT) and histone demethylases (HDM) in a competitive manner with histone deacetylases (HDAC) and histone methyltransferases (HMT), respectively. Among HMTs, the Polycomb group (PcG) protein EZH2, a component of the Polycomb repressor complex 2 (PRC2), is responsible for the trimethylation of Lysine 27 on histone H3 (H3K27me3) on gene promoters (Figure 2).

This modification allows the recruitment of the PRC1 complex which inhibits gene transcription through histone H2A ubiquitination. HDACs participate to the PRC complexes reinforcing the inhibition of gene expression by deacetylating H3K27, thus favoring H3K27 trimethylation. Finally, PcG complexes recruit DNMTs to specific gene loci to induce transcriptional silencing through DNA methylation (Figure 2). The final result is the maintenance of cell stemness and the support of self-renewal and pluripotency. Consistently, components of the PRC complexes are often aberrantly over-expressed in tumors (reviewed in [12]). Of note, epigenetic modifications that are maintained through mitosis and inherited during cell development and differentiation can be reversed by treatment with appropriate drugs. Therefore, compounds acting on “epimutations” can be used in association with conventional chemotherapy to induce growth arrest, differentiation and tumor cell death.

MiRNAs are a class of non-coding ~19–22 nucleotides (nt) single-strand RNAs transcribed in a developmental and tissue-specific manner during embryogenesis [13]. More than 1500 mature miRNAs have been identified in humans (http://www.mirbase.org), which are highly conserved across species. miRNAs are involved in post-translational gene silencing by binding complementary sequences in the 3′-untraslated regions (UTRs) of a target mRNA through their “seed” sequence leading to translational repression or mRNA degradation [14,15].

The expression of more than 60% of human protein-coding genes is controlled by miRNAs (reviewed in [13]). Several miRNAs can target the same mRNA and, conversely, each miRNA can target several mRNAs leading to additional layers of post-transcriptional control of gene expression. To further expand the scenario, specific miRNAs have been recently described to upregulate gene transcription also in quiescent cells [16] and others to target sequences within the 5′UTR [17], exonic regions [18] or gene promoter regions [19].

Mature miRNAs are the result of sequential processing of longer RNA precursors transcribed by RNA polymerase II and/or III and called pri-miRNA (Figure 3) [20,21]. This long transcript is processed by Drosha, a ribonuclease-III protein, in collaboration with DGCR8, a protein responsible for anchoring the pri-miRNA into the complex. The resulting 70-nt-long molecule termed pre-miRNA, is exported to the cytoplasm by the nuclear export protein Exportin-5 [13,22]. In the cytoplasm, the pre-miRNA is released and processed by Dicer to produce the functional ~19–22-nt double-strand RNAs. The less stable single strand is loaded by the Dicer-TRPB complex into the RNA-induced silencing complex (RISC) which carries it to complementary 3′UTR mRNA sequences.

MiRNAs regulate fundamental physiologic processes such as embryonic development, lineage/tissue identity specification and homeostasis [23–27]. Therefore, it is not surprising that miRNAs display an aberrant expression profile in a wide range of human diseases [13]. In tumor cells, as reported for protein-coding gene, miRNA promoters can be aberrantly modified by the deregulated epigenetic machinery leading to changes in their expression profile [28–31]. Increasing evidence suggesting tumor suppressor or oncogenic role for miRNAs has been obtained in adult cancers [1,13,32–34]. Consistently, the re-expression of tumor suppressor miRNAs in tumor cells has been suggested as a potential anti-cancer therapeutic approach [35–40].

Recently, miRNAs have been shown, by us and other groups, to play a role also in the pathogenesis of embryonal pediatric cancers such as RMS and NB [41–43]. So far, the majority of tumor suppressor miRNAs involved in RMS and NB regulates the differentiation of skeletal muscle and neuronal compartments, respectively. These miRNAs are often silenced by aberrant DNA or histone methylation or located in unstable chromosomal regions often involved in LOH, ie miR-34a [13,44].

Consistently, their re-expression after treatment with HDAC inhibitors, de-methylating compounds or differentiating agents is sufficient to allow tumor cell differentiation and to impair tumorigenesis. Therefore, an epigenetic approach aimed at re-expressing these miRNAs could have a therapeutic value.

RMS is a skeletal muscle-derived tumor and the most common soft-tissue sarcoma of childhood [45]. RMS cells express transcription factors specific to skeletal muscle progenitors, such as Myogenic Differentiation (MyoD) and myogenin [46,47], but their intrinsic myogenic program is disrupted at different stages of differentiation. In line with these observations, restoring terminal differentiation in RMS cells leads to inhibition of cell proliferation in vitro and tumor growth in vivo.

Pediatric RMS includes two major histological subtypes, namely alveolar (~25% of cases) and embryonal (~75% of cases) RMS [45,48]. The five-year overall survival rate of non-metastatic embryonal RMS patients is around 70%, while that of alveolar RMS and metastatic patients is still in the order of 25%, supporting the urgent need of novel therapeutic approaches.

The majority of alveolar RMS are characterized by chromosomal translocations, such as t (2; 13) or t (1; 13), leading to the expression of PAX3/FOXO1 or PAX7/FOXO1 fusion proteins, respectively [49–51]. These fusion products, especially PAX3/FOXO1, are a hallmark of high-risk tumors and correlate with poor prognosis [52]. Approximately 20% of alveolar RMS do not present known chromosomal translocations and are molecularly and clinically indistinguishable from embryonal ones [53–55]. However, novel chromosomal translocations involving PAX3 have been recently discovered in a subset of alveolar RMS previously diagnosed as fusion negative [51,56].

During skeletal muscle tissue development, PAX3 expression is detected in pluripotent progenitors and is followed by that of PAX7, which characterizes myogenic committed cells [57,58]. These transcription factors, on one hand, induce the expression of MyoD that marks myoblasts, and on the other hand, stimulate self-renewal and proliferation of myogenic cells. Thus, their expression is fundamental to assure the balance between cell proliferation and differentiation during embryogenesis leading to the right number of committed progenitors in order to form multinucleated myofibers. During differentiation, pro-myogenic miRNAs downregulate PAX3 and PAX7 to achieve complete myogenesis (reviewed in [59]).

An altered expression of miRNAs acting as pro-myogenic regulators has been reported in RMS, [43,59–61]. These miRNAs behave as tumor suppressors when re-expressed, halting tumorigenic processes working in negative feedback circuitries with epigenetic regulators. These findings demonstrate that aberrant epigenetic control of miRNA expression concurs to RMS pathogenesis.

NB is a neuroectodermal tumor that originates from precursor cells of the sympathetic nervous system and represents the third leading cause of cancer-related death in childhood [80]. The heterogeneous clinical behavior, ranging from spontaneous regression to rapid progression, is attributable to biological and genetic characteristics of the tumor.

NB diagnosed in patients under one year of age usually has a favorable prognosis since tumor cells undergo spontaneous differentiation or regression, whereas NB occurring in patients over one year of age tend to grow aggressively resulting in a fatal outcome. The prognosis of stage I-III NB, with a tumor confined to the originating organ or surrounding tissue, is quite favorable, whereas that of stage IV NB, where the tumor is metastatic, is dismal. Stage IV-S NB is a metastatic disease seen exclusively in infants, which is associated with high survival rate due to the spontaneous maturation and regression of tumor cells. Nevertheless, since disease and risk staging are not comprehensive and fully precise, they should be considered as surrogate markers of the underlying tumor biology.

NB cells correspond to adrenal neuroblasts arrested at different stages of sympathoadrenal development, thus representing multipotent progenitor cells with specific tumorigenic potential [81,82]. The degree of differentiation of NB is another important factor for establishing prognosis. Based on this, NB tumors are classified in different risk categories spanning from more mature and benign ganglioneuromas (GN), to intermediate and potentially malignant ganglioneuroblastomas (GNB), to undifferentiated NB, always malignant with worst prognosis. The factors responsible for malignant transformation from sympathetic neuroblasts into neuroblastoma cells are not well understood. In several NB cell lines, all-trans retinoic acid (ATRA)-induced differentiation is associated with increased expression of neurotrophic receptors, including Trk family receptors and glial cell-derived neurotrophic factor receptor, Ret. Of note, ATRA is a compound related to 13-cis-retinoic acid that is used as differentiating agent in children with high-risk NB.

From a clinical point of view, the presence of any segmental chromosomal imbalances correlates with a more aggressive phenotype, whereas tumors containing mostly whole chromosome gains or losses are associated with a benign phenotype and high propensity for spontaneous regression or differentiation. The most widely characterized cytogenetic alterations in NB tumors include amplification of the MYCN oncogenic transcription factor at chromosome band 2p24, LOH or rearrangements of the distal portion of the short arm of chromosome 1 (1p31-term), chromosome 3 (3p22) and chromosome 11 (11q23), or gains of chromosome arm 1q or 17q. Besides these chromosomal/molecular abnormalities, gains of chromosomes 4q, 6p, 7q, 11q and 18q, amplification of MDM2 and MYC genes and LOH at 14q, 10q and 19q13 have also been described [83]. In particular, loss of chromosome 1p region, occurring mainly through an unbalanced translocation that results in the gain of 17q [84], arises preferentially in tumors with MYCN amplification [85]. Some of these genetic alterations were proven to be independently associated with a poor clinical outcome. The region of 1p, lost in NB tumors, is quite large and contains multiple genes that have been shown to contribute to NB pathogenesis [86,87]. Similarly, the minimal common region of gain at distal 17q23 has been shown to contain at least 15 genes that remain to be investigated [88]. Of note, MYCN is one of the major players in normal neural crest differentiation, suggesting that its amplification/deregulation is the result of a developmental defect that probably has occurred during embryonic development [89]. However, although MYCN amplification is considered the most important prognostic marker of highly aggressive tumors; it is present in less than 30% of NB tumors [90].

More recently, activating mutations in the anaplastic lymphoma kinase (ALK) gene have been associated with the majority of hereditary NB and a 10% of sporadic NB cases [91–94]. ALK is a receptor tyrosine kinase involved in several other human cancers, including anaplastic large-cell lymphoma and non-small cell lung cancer [95–97]. ALK signaling drives cell transformation in vitro and in vivo through several pathways such as cell-cycle, survival and cell migration [98,99]. Nevertheless, additional molecular alterations concur to NB pathogenesis and their identification is essential to improve the prognostic classification of NB patients.

In the attempt to improve the risk stratification of NB subtypes, several groups have characterized and correlated methylation profiling of biologically and clinically different subgroups of NB patients with clinical risk factors and survival (for a detailed review see [100]). The first DNA methylation studies in NB have revealed that silencing of caspase 8 and RAS-association domain family 1 isoform A (RASSF1A) are important in the development and progression of disease [101,102]. Both genes are often methylated in primary NB cells and their methylation status negatively correlates with survival.

Recent advances in genome-wide methylation screening methodologies, such as re-expression analysis after treatment with 5-aza-2′-deoxycytidine (DAC), DNA methylation promoter assay after affinity-based capture, methylation microarray after bisulfate treatment and next-generation sequencing techniques, have allowed the identification of 75 different DNA methylation biomarkers involved in fundamental biological processes operative in NB ([103–105] and reviewed in [100]).

Epigenetic inactivation of miRNAs with tumor-suppressor activities is also recognized as a major hallmark of NB tumors. A recent study from Chavali group [106] showed that the majority of the miRNA genes are flanked by scaffold/matrix-attachment region (S/MAR) elements that regulate their tissue and cell type-specific expression by binding to epigenetic regulators MAR binding proteins. In this study, the authors showed that the miR-17-92 cluster, a well-known example of oncomiR that is over-expressed in high-risk NB [41], has two conserved MARs elements, one upstream and one downstream the cluster, being the former strongly bound by HMG I/Y protein, a chromatin modulator that promotes an open conformation facilitating mRNA transcription. Consistent with this finding, the miR-17-92 cluster expression was down-regulated in cells interfered for HMG I/Y family expression [106].

The majority of miRNAs aberrantly downregulated in RMS and NB behave, when re-expressed, as tumor suppressors re-establishing a proper differentiation program. Therefore, they represent promising therapeutic targets. Moreover, miRNAs detection in cancer patients could represent a powerful prognostic and diagnostic tool. A miRNA signature has been recently reported for RMS primary samples by two groups [77,135]. Intriguingly, the first group suggested that an embryonal sample had been previously misdiagnosed allowing a re-evaluation of the tumor followed by a diagnosis as alveolar [77]. The second group showed that miRNA expression clusterings in RMS samples are correlated with PAX3/FOXO1, PAX7/FOXO1 and embryonal subgroups, and, thus, are potentially useful for risk stratification [135].

Recently, miRNA signatures have been shown to be related also to NB prognosis. Lin and colleagues found that, using a combination of 15 biomarkers that consist of 12 miRNAs’ signature, expression levels of Dicer and Drosha and age at diagnosis, it was possible to segregate 66 NB patients into four distinct patterns that were highly predictive of clinical outcome [136]. Subsequently, a signature of 25 miRNAs was built and validated on an independent set of 304 frozen tumor samples and 75 archived formalin-fixed paraffin-embedded samples. This signature was significantly able to discriminate the patients with respect to progression-free and overall survival [137].

Furthermore, the possibility to detect and quantify miRNAs in body fluids whose profile mirrors physiological and pathological condition could also represent a reliable method to improve prognosis and diagnosis [52,138,139].

MiRNAs re-expression in tumors rises doubts regarding delivery, efficacy and safety of these molecules in vivo. Viral and non-viral vectors have shown pre-clinical feasibility and efficacy in animal models including primates [36,140–145]. However, their application on human patients could induce side effects linked to their immunogenicity or could be ineffective due to lack of specificity (reviewed in [146]). The use of nanoparticles represents an efficient system to deliver cDNA to cells and, therefore, it could be potentially useful for the delivery of miRNAs in vivo. Indeed, a first promising study using nanoparticles encapsulating miR-34a and conjugated to a GD2 antibody demonstrate the feasibility of a targeted delivery miRNA-based therapy for NB [147].

Based on these observations, therapeutic approaches able to modulate miRNA gene expression targeting epigenetic regulators present several advantages. Conversely to DNA mutations, epigenetic modifications are reversible and responsive to specific compounds, which gave promising results in clinical trials for adult tumors. Indeed, low doses of DNA methyltransferases inhibitors azacitidine (AZA) and DAC have shown therapeutic effects in myelodysplasia/leukemia as well as in lung cancer patients [148–153]. It is noteworthy that older trials with high doses of epigenetic inhibitors have highlighted extreme toxicities and no effect on epigenome [154,155]. Very recently, an elegant work of Tsai et al., investigated the mechanisms involved in the efficacy of low nanomolar doses of DAC and AZA in preclinical models of acute myeloid leukemia, breast, colon and lung cancers [156]. The authors showed that high-dose treatments resulted in high cytotoxic effects on tumor cells. Nevertheless, residual cells engrafted in immunodeficient mice were still able to grow and give tumors. Conversely, low concentrations of AZA or DAC reduced self-renewal of tumor cells both in vitro and in vivo, sustaining global DNA demethylation and re-expression of anti-tumor master gene pathways [156]. These results strongly confirm the potential efficacy of an epigenetic therapy that could re-express tumor suppressor genes and miRNAs.

Moreover, epigenetic targeted therapy with inhibitors of both DNA methyltransferases and HDACs, has shown anti-tumor effectiveness combined with a restored expression of tumor suppressor genes in preclinical models [157–159]. More important, well-tolerated combination regimens for low-dose AZA and HDAC inhibitors have been found effective in patients with myelodysplasia/leukemia [160,161]. Finally, very encouraging results have been reported in a phase I/II clinical trial on the use of a combined epigenetic low-dose therapy in recurrent metastatic non-small cell lung (NSCL) cancer refractory to conventional therapy [153]. In brief, AZA and entinostat, an inhibitor that blocks HDAC1-3 and HDAC9 function, were used on 45 NSCL patients, resulting in limited side effects while having efficacy on the aberrant epigenome of tumor cells. These findings suggest the potential use of these compounds also in solid cancers.

Novel agents that inhibit histone methyltransferases among which EZH2 are of particular interest for future applications since they do not require cell division to target cancer cells [162,163]. These compounds have been shown to synergize with other epigenetic agents in preclinical studies, representing promising drugs in order to de-repress EZH2 targeted miRNAs [164,165].

However, as reported for other targeted therapies, escape strategies for the miRNA re-expression approach could select resistant cells expressing mRNAs with mutated miRNA binding sites in their 3′UTRs. Moreover, PAX3/or PAX7/FOXO1 mRNAs expressed in fusion-positive alveolar RMS, being devoid of 3′UTR regions of PAX genes, are resistant to miRNAs’ post-transcriptional regulation. In addition, the de-repression of oncogenes could be a potential side effect of an epigenetic therapy [166]. Therefore, although the field of investigation in the epigenetic treatment of cancer appears very promising, further studies are needed to enhance the knowledge on the epigenome and its regulation both in normal and tumor contexts.