Recent advances in stem cell biology have shed light on how normal stem and progenitor cells can evolve to acquire malignant characteristics during tumorigenesis. The cancer counterparts of normal stem and progenitor cells might be occurred through alterations of stem cell fates including an increase in self-renewal capability and a decrease in differentiation and/or apoptosis. This oncogenic evolution of cancer stem and progenitor cells, which often associates with aggressive phenotypes of the tumorigenic cells, is controlled in part by dysregulated epigenetic mechanisms including aberrant DNA methylation leading to abnormal epigenetic memory. Epigenetic therapy by targeting DNA methyltransferases (DNMT) 1, DNMT3A and DNMT3B via 5-Azacytidine (Aza) and 5-Aza-2’-deoxycytidine (Aza-dC) has proved to be successful toward treatment of hematologic neoplasms especially for patients with myelodysplastic syndrome. In this review, I summarize the current knowledge of mechanisms underlying the inhibition of DNA methylation by Aza and Aza-dC, and of their apoptotic- and differentiation-inducing effects on cancer stem and progenitor cells in leukemia, medulloblastoma, glioblastoma, neuroblastoma, prostate cancer, pancreatic cancer and testicular germ cell tumors. Since cancer stem and progenitor cells are implicated in cancer aggressiveness such as tumor formation, progression, metastasis and recurrence, I propose that effective therapeutic strategies might be achieved through eradication of cancer stem and progenitor cells by targeting the DNA methylation machineries to interfere their “malignant memory”.

The study of stem cell biology, especially that of hematopoiesis, has provided an insight into the biology of cancers[1,2]. In particular, normal stem cells and cancer cells share one of their characteristics, which is an ability to proliferate without differentiation, the property known as self-renewal. Three concepts of cancer biology are proposed based on the study of hematopoietic stem cells and their tumorigenic counterparts as the paradigm: (1) normal stem cells and cancer cells might possess similar mechanisms to control their proliferation or self-renewal; (2) cancer cells can be derived by differentiation of normal stem cells via mutation of transit-amplifying cells or progenitors; and (3) there might be cancer stem cells inside tumors where the tumorigenesis occurred with normal stem cells[3]. Leukemia, which is the cancers of hematopoietic system, has provided the best evidence that normal stem cells are the target of oncogenic evolution, and that leukemogenesis is driven by cancer stem cells. For example, the tumor initiating cells of myelodysplastic syndrome (MDS), which can develop into acute myeloid leukemia (AML), are cancer stem cells, at least those with genetic deletion of chromosome 5q[4]. Cancer stem and progenitor cells have been suggested to involve with relapse in many types of cancers[5-13]. This tumorigenic recurrence of cancer stem and progenitor cells, which often associates with more aggressive phenotypes compared with differentiated cancer cells, can occur via accumulation of genetic mutations and/or epigenetic defects leading to selective advantage and hence the oncogenic evolution of cancer stem and progenitor cells.

Gene expression must be tightly regulated by both genetic and epigenetic mechanisms in non-cancerous normal cells, thereby avoiding malignant transformation. Both genetic- and epigenetic-dependent mechanisms ensure that expression of genes, whose function is involved in cell proliferation, apoptosis, self-renewal and differentiation, is normally controlled in temporal- and spatial-dependent manners. Deregulated epigenetic mechanisms have been shown to implicate in tumorigenesis. One epigenetic mechanism, which has been prominently focused in cancer research, is DNA methylation which also represents one of the most active fields of epigenetic research.

DNA methylation is one of the most common defects in epigenetic regulation observed in tumorigenesis. Earlier studies have suggested that cancer cells possess global DNA hypomethylation, which results in chromosomal instability[14]. However, in the past decade most cancer epigenetics studies have paid attention to regulatory region-associated DNA hypermethylation[15]. DNA hypermethylation of CpG dinucleotides has been well-documented in many types of cancer such as leukemia and colon cancer, and has been proposed as a hallmark of tumorigenesis[16]. Specifically, DNA hypermethylation at promoters of tumor suppressor- and differentiation-associated genes has been frequently reported. The DNA hypermethylation of those genes then leads to a reduced gene expression, which in turn provides a selective advantage to cancer cells. Thus, down-regulation of the genes by DNA hypermethylation might result in the emergence of cancer stem and progenitor cells[17,18].

DNA methylation has been widely accepted to mediate malignant memory of cancer cells due to its stable inheritance during cell division. Gene regulatory network specific to cancer stem and progenitor cells might play a direct role in dysregulation of DNA methylation. For example, the pluripotency-associated transcription factors OCT4 and NANOG have been shown to directly up-regulate expression of the maintenance DNA methyltransferase (DNMT) DNMT1 in mesenchymal stem cells[19]. In fact, maintenance of DNA methylation has been reported to be critical for cancer stem cell properties in leukemic, colon and lung cancer stem cells[20-22], which might be required to suppress expression of differentiation- and apoptosis-related genes. For instance, p16 and MLH1 are frequently hypermethylated in certain types of cancer. Of interesting is MLH1, which is a member of DNA mismatch repair pathway. Thus the epigenetic instability of MLH1 might be a cause of genetic instability of cancer stem and progenitor cells. The epigenetic defects of other genes involved in different pathways related to cell migration, hormone receptors, and cell cycle control, have also been described in tumorigenesis[18]. Different patterns of DNA hypermethylation have also been found in different types of cancers, which might be explained by cellular function of the cells in respective tissues[16]. In addition, the pattern of DNA hypermethylation might be influenced by an interaction between DNMTs and polycomb repressive complex 2 (PRC2) protein subunits[23], since different tumors also have different patterns of PRC2 binding[17,24,25]. Epigenetic therapy based on DNA methylation inhibitors including 5-Azacytidine (Aza) and 5-Aza-2’-deoxycytidine (Aza-dC) has promised a therapeutic regimen to patients with DNA hypermethylation-induced cancer diseases[15]. Therefore, understanding how DNMTs transfer a methyl group to genomic DNA, how they function in cancer cells, and how the drugs inhibit DNMT activity is crucial for development of superior anti-cancer therapy.

DNA methylation is responsible for multiple chromatin-related processes including regulation gene expression by adding a methyl group from S-adenosyl methionine (SAM) to the C5 position of a cytosine base mostly within CpG dinucleotide context at gene regulatory regions such as promoters and CpG islands. The CpG methylation is catalyzed by DNMTs. Five DNMT family proteins, i.e., DNMT1, DNMT2, DNMT3A, DNMT3B and DNMT3L, have been identified in mammalian genomes[26]. Among these, DNMT2 does not function as a methyltransferase for DNA but as a RNA methyltransferase[27,28]. In addition, DNMT3L does not possess the methyltransferase catalytic domain[29], leaving only DNMT1, DNMT3A and DNMT3B that are functional DNA methyltransferases[30]. Two mechanisms underlying DNA methylation have been reported. The first mechanism is maintenance DNA methylation pattern, of which all three DNMTs have the ability to copy DNA methylation pattern from DNA template to daughter DNA strand[31-37]. In one study, Sharma et al[34] have proposed a homeostatic inheritance system for maintenance of DNA methylation, whereby high levels of 5-methylcytosine stabilize DNMT3A and DNMT3B and prevent them from proteolysis[34]. The second mechanism is de novo DNA methylation which can only be performed by DNMT3A and DNMT3B. Apart from repressing gene expression via methylation at gene promoters, DNA methylation also plays important roles in many epigenetic phenomena involved in development for example genomic imprinting, X chromosome inactivation and higher-order chromatin organization[38,39]. Hence, apart from cancers, dysregulation of DNA methylation are also implicated in pathogenesis of many diseases.

Several DNMT domains play important role in the methylation catalysis. Region IV harbors a prolyl-cysteinyl dipeptide that functions as the active site (Figure 1). Regions I and X share amino acid residues, which are critical for binding of SAM. Region V contains glutamyl/asparaginyl residue, which functions to protonate nitrogen 3 (N3) of targeted cytosine during the methyltransferase reaction. The enzymatic mechanism of DNA methylation begins with flipping cytosine out of the DNA helix upon binding with a DNMT. The conformational change of the cytosine nitrogenous base ensures that carbon 5 (C5) of the base aligns at the catalytic pocket with a bound SAM, and that C6 is at close proximity to a cysteine residue at the active site of the enzymes. The base flipping leads to formation of a covalent bond between C6 of the cytosine in DNA and the cysteine residue mediated by nucleophilic attack of sulfhydryl group of the cysteine to the C6 resulting in a change from C5 = C6 double bond into C5-C6 single bond (Figure 2A). The chemical bonding between C6 and cysteine then leads to an increase in electron flow toward C5 position of cytosine. At the same time, a methyl group from SAM is transferred to C5 generating C5-CH₃ of the cytosine and S-adenosyl-homocysteine as the products. The proton at C5 position is then abstracted due to its weak bonding with C5 resulting in β-elimination at the C5-C6 single bond and in the reformation of the C5 = C6 double bond, which return a single electron to the cysteine of DNMT and discharge to the enzyme[30]. The C5-cytosine is now become methylated giving rise to methylcytosine.

Open in New Tab   Full Size Figure   Download Figure

Figure 1 High similarity of critical amino acid residues at the catalytic domain of DNA methyltransferases. Multiple alignment of amino acids within the catalytic domain of three catalytically active human DNA methyltransferases DNMT1, DNMT3A and DNMT3B is shown. Highlighted amino acids with roman letters shown above indicate critical residues of the catalytic domain. DNMT: DNA methyltransferases.

Open in New Tab   Full Size Figure   Download Figure

Figure 2 Mechanism of DNA methylation and its inhibition by 5-Aza-2’-deoxycytidine via covalent trapping. A: Cytosine methylation by a DNMT is initiated by the formation of a covalent bond between the cytosine base and a cysteine residue at the active site of the methyltransferase. SAM then transfers a methyl group to the cytosine generating a methylcytosine. B: A DNMT is covalently trapped by the DNA-incorporated 5-Aza-2’-deoxycytidine (Aza-dC) adduct. The sulfhydryl group of the cysteine residue at the active site of DNMT forms a covalent bond with C6 of the Aza-dC in DNA. However, the active methyl group from SAM cannot be transferred to Aza-dC by DNMT. This covalent trapping mechanism then leads to a gradual loss of free DNMTs, because they are covalently linked with Aza-dC within DNA. Therefore, the loss of free DNMTs then precedes passive DNA demethylation. Thus cells expressing high levels of DNMTs will respond to Aza-dC at a higher extent compared with cells expressing low levels of DNMTs (see Figure 3). R: Methionine/cysteine; R’: Adenosine; Dr: Deoxyribose-5-phosphate; DNMT: DNA methyltransferases.

Open in New Tab   Full Size Figure   Download Figure

Figure 3 5-Aza-2’-deoxycytidine induces apoptosis of human teratocarcinoma stem cells but not their differentiated counterparts. Human nullipotent embryonal carcinoma (EC) cells, the stem cells of teratocarcinoma, are highly sensitive to 5-Aza-2’-deoxycytidine (Aza-dC), which induces apoptosis of the cancer stem cells. On the other hand, OCT4-knockdown differentiated nullipotent EC cells are resistant toward Aza-dC treatment. In human pluripotent EC cells NTERA2, DNMT3B mediates apoptosis induced by Aza-dC, whereas the enzyme mediates differentiation of human ES cells induced by the inhibitor. DNMT: DNA methyltransferases; ES: Embryonicstem; OCT4: Octamer-binding transcription factor 4.

The ability of Aza and Aza-dC to induce not only cell death but also differentiation had encouraged the exploitation of the compounds for treatment of cancers associated with stem/progenitor cells such as leukemia[40,41]. Unlike Aza which can be incorporated into both DNA and RNA, Aza-dC is only incorporated into DNA and is therefore a specific inhibitor for DNA but not RNA methyltransferases[42,43]. Consequently, the difference between the two epigenetic drugs results in differential cellular responses and causes distinct changes in the gene expression patterns of cells treated with Aza-dC and Aza[44,45].

Intracellular uptake of Aza-dC is mediated by the human concentrative nucleoside transporter 1 HCNT1, also known as SLC28A1[46]. Once inside cells, Aza-dC is then phosphorylated by deoxycytidine kinase yielding Aza-dCMP. Aza-dCMP is further phosphorylated to Aza-dCDP and subsequent Aza-dCTP, the active substrate form for DNA replication, by dCMP kinase and diphosphokinase, respectively. Hence the irreversible inhibition of Aza-dC on DNA methylation takes place only if Aza-dCTP is used as a substrate by DNA polymerases during DNA replication at S phase of the cell cycle, but not from free Aza-dC itself. Within DNA and with the presence of DNMTs, which bind to genomic DNA, the Aza-dC base can make a covalent bond with a DNMT enzyme (Figure 2B), disrupting the function of the enzyme. Mechanistically, because the nitrogen atom at position 5 (N5) of Aza-dC has a higher electronegativity than C5 of cytosine, the incorporated Aza-dC can covalently trap the DNMT enzyme. A single bond between N5-C6 is formed upon the nucleophilic attack of sulfhydryl group to the C6. However, unlike C5 in cytosine, the N5 can abstract a proton, and hence stably form a single bond to the proton with a strong affinity due to its high electronegativity. The DNMT is therefore trapped with the incorporated Aza-dC through a covalent bond between C6 and the cysteine residue leading to a DNA-DNMT adduct. Thus only in the presence of DNMTs can Aza-dC function to inhibit DNA methylation activity[47]. At chromatin level, gene reactivation activity of Aza-dC has been mechanistically demonstrated by Jones et al[31]. The authors devised a novel technique called AcceSssIble[48]. They find that approximately 2% of chromatin regions, which are demethylated after Aza-dC treatment in a colon cancer cell line, govern a feature of open chromatin[48] possibly associated with deposition of the histone variant H2A.Z[49]. Importantly, they show that of all genes reactivated by Aza-dC treatment, 90% of them possess DNA demethylation and chromatin accessibility supporting the role of the DNA methylation inhibitor as a gene reactivator. Interestingly, many reactivated genes are also tumor suppressor genes[48].

Several reports have demonstrated that depletion of DNMT1, DNMT3A or DNMT3B can counteract the cytotoxic and apoptotic effect of Aza-dC on treated cancer cells, confirming the essential role of DNMTs in mediating the Aza-dC response[47,50-52]. In contrast to the apoptotic response mediated by DNMTs, whether DNMTs also mediate stem cell differentiation induced by Aza-dC is not well understood. Epigenetic therapy by induced differentiation offers an option for treating malignancies, in which cancer tissues contain both cancer stem/progenitor cells and differentiated cancer cells, such as chronic leukemia, teratocarcinoma, and other solid tumors. Thus elucidating how cell differentiation can be triggered by Aza-dC or Aza could have an enormous impact on epigenetic therapy targeting cancer stem and progenitor cells.

Progresses in an understanding of normal stem cell biology have revolutionized the concept of cancer stem and progenitor cells, which might be the key toward success in treatment of a number of cancer diseases by targeting at cancer stem and progenitor cells. Aza and Aza-dC, although being the drugs currently used for treatment of hematologic malignancy, have held a great promise for treatment of solid tumors[65]. Future studies on mechanisms underlying differentiation and/or apoptosis induced by Aza and Aza-dC on cancer stem and progenitor cells are required for effective treatment of the epigenetic drugs. Several questions emerge including; (1) Do gene regulatory networks, which are specifically associated with cancer counterparts of stem and progenitor cells, up-regulate expression of DNMTs? (2) Which signaling pathways and their downstream target genes are functionally perturbed by Aza and Aza-dC treatment? (3) Are cell death pathways other than apoptosis involved in the anti-proliferation effect? (4) Are Aza and Aza-dC effective to eradicate cancer stem and progenitor cells in xenograft tumors? (5) In case of relapse, do recurrent cancer stem and progenitor cells evolve a distinct “malignant memory” at epigenomic levels compared with the cancer cells isolated before admission? and, if so, (6) Does the oncogenic evolution lead to more aggressive phenotypes such as a reduced differentiation potential of cancer stem and progenitor cells ? Treatments with Aza and Aza-dC can lead to certain toxicity and side effects to hematopoietic, nervous as well as metabolic systems[104-107]. Nevertheless, they have been recognized as having lower toxicity compared to traditional chemotherapy. In fact, low-dose treatment of Aza-dC in patients with MDS[108] and solid tumors[109] resulted in complete response of the patients. Thus using low-dose DNA methylation inhibitors may provide a safe therapeutic choice[65] especially for elderly patients[107,110]. Whether long-termed treatments of low-dose Aza and Aza-dC would perturb fates of stem and progenitor cells without side effects and toxicity in patients have yet to be elucidated. Addressing these questions will not only provide an understanding of cancer stem cell biology but also shed light on development of novel clinical regimes.

The author is grateful to the Thai Commission on Higher Education for the Strategic Frontier Research scholarship on stem cell biology.