I read with great interest the recent article by Daskalakis et al1 on the potential use of the DNA methyltransferase inhibitor 5-Aza-2′-deoxycytidine (5adC) in patients with myelodysplastic syndromes (MDSs). The authors demonstrate that methylation-mediated silencing of the tumor suppressor genep15/INK4B can be partially overcome using 5adC. Bisulphite sequencing of the p15 promoter shows significant CpG demethylation after several courses of 5adC. Of the 23 patients studied, 15 were scored with p15 methylation status above 15%. Twelve patients showed p15 hypermethylation before 5adC treatment, with reversal in 9 patients. But questions underlying the precise consequence of methylation changes outside of thep15 allele warrants attention. In an attempt to generate discussion over the topic, I am compelled to raise the issue of methylation-dependent silencing of the multidrug resistance gene 1 (MDR1) in myeloproliferative disorders.2 Recent studies reveal MDR1 gene expression is controlled by epigenetic silencing mechanisms such as histone deacetylation3 and CpG methylation4,5 in various hematologic malignancies. Methylation of the MDR1promoter is inversely correlated with gene expression.6Given that MDR1 expression is a negative prognostic factor for response and survival in acute myeloid leukemia (AML),7 this would suggest that epigenetic modification by a DNA methyltransferase inhibitor may inadvertently aggravate theMDR1 phenotype.
The article presented by Daskalakis et al1 convincingly demonstrates that 5adC has profound epigenetic effects onp15 gene silencing. It is believed that incorporation of the cytidine analog is substituted into newly synthesized DNA strands, thus blocking methylation by covalently trapping the DNA methyltransferase enzyme while attempting methyl-group transfer.8,9Experimental evidence at present suggests that 5adC exposure invariably results in a more “global” demethylation profile instead of gene-specific demethylation, and microarray expression analysis reveals a number of different genes are activated.10 The methyl-CpG signal on MDR1 chromatin targets the methylation-specific transcriptional repressor MeCP2 for gene silencing.11 Methyl-CpG binding domain (MBD) proteins physically associate with histone deacetylase to remodel chromatin, and methylated DNA provides a dominant mechanism of repression. Consistent with this epigenetic model, azacytidine-induced demethylation causes the release of the transcriptional repressor and primes the promoter for gene activation. The results to these experiments suggest silent MDR1 chromatin can be efficiently remodeled through the use of DNA methylation inhibitors.
Given the evidence, it is likely that 5adC may activate other silent genes. The authors indeed appreciate that MBD proteins recognize methylated CpG sites to repress gene activity, and this may represent a pathway of transcriptional loss for p15. In a similar fashion, the methyl-CpG binding protein MBD2 cooperates with histone deacetylase to silence p14 and p16. The hypermethylated alleles can be reactivated by 5adC, with corresponding release of MBD2 from the promoter sequence.12
Therefore, it is worthwhile remembering that the benefit of activating proliferation-associated genes by DNA demethylation may also run the risk of activating a drug resistance pathway believed to be a major cause of treatment failure in patients with myeloproliferative disorders.
Acute myeloid leukemia of the elderly: results of induction after previous treatment of high-risk myelodysplastic syndrome with a demethylating agent
We have read with interest the thoughtful comments of Dr El-Osta regarding our observation of in vivo demethylating activity of 5-aza-2′-deoxycytidine (decitabine; DAC) upon the p15/INK4Bgene in myelodysplastic syndrome (MDS) patients.1-1Since many genes can be inactivated by DNA methylation,1-2reactivation of additional genes bearing a pathologic hypermethylation pattern in these patients is not unlikely and warrants sequential analyses of other putative target genes before and during treatment with demethylating agents. Additionally, many genes up-regulated by azanucleosides do not have CpG islands in their promoters or have not been shown to be methylated before treatment.1-3,1-4 Based on the concern that treatment with demethylating agents may also reactivate genes that could potentially interfere with anticancer treatment, studies have addressed the possibility that the multidrug resistance pump P-glycoprotein (Pgp), the product of the MDR1gene, may be up-regulated via demethylation. Indeed, this gene was shown to be up-regulated by both 5-azacytidine and 5-Aza-2′-deoxycytidine (as described by Dr El-Osta) in several cell line models. Other studies have challenged this view: fine-mapping of its CpG-rich promoter has suggested that the methylation status of this domain does not act as a switch to regulate MDR1expression,1-5 and a study using an anthracycline-resistant subclone of the myeloid leukemia cell line K562 in fact demonstrated down-regulation of MDR1 by a demethylating agent (possibly mediated by demethylation of a silencer element in the MDR1promoter.1-6
The concern that in vivo responsiveness of malignant cells to chemotherapy is mitigated by pretreatment with a demethylating agent can first be addressed by the fact that Pgp confers in vitro resistance to some but clearly not all cytostatic drugs in common use. Cytosine arabinoside, which is considered to be the most active drug against AML, is no substrate of P-glycoprotein. In fact, pretreatment with a demethylating agent may even sensitize previously drug-resistant xenograft tumors to both MDR1-dependent and -independent cytostatic drugs.1-7. Second, while it is well established that MDR1 expression is a negative prognostic factor in AML,1-8 it is also associated with an immature, stem cell–like phenotype (CD34+ blast cells) in AML and MDS, which is also a poor prognostic factor in AML.1-9Correspondingly, MDR1 as well as other drug transporters, such as the breast cancer resistance protein (BCRP, also known as ABCG2) are highly expressed in normal hematopoietic stem cells, and their expression is down-regulated during myeloid differentiation.1-10 1-11 Taken together, in some casesMDR1 expression may be an indicator of an early phenotype, rather than the primary cause for failure to antileukemic induction chemotherapy.
Among 101 MDS patients treated with DAC at 2 centers, 15 patients (median age, 63 years; range, 50-77 years) received standard induction chemotherapy at time of progression to AML. Eight of 15 had shown a previous response to DAC during a median of 7 treatment courses (range, 3-9 courses, with or without a trial of retreatment at relapse), whereas 7 of 15 had not responded to this drug (median, 1 course; range, 1-2 courses). Nine (60%) of 15 patients receiving induction treatment of secondary AML achieved a response, including 6 (40%) of 15 achieving complete remissions (Table1-1). This included both DAC responders and nonresponders. Thus, pretreatment with DAC (independent of its effectiveness) did not result in a leukemia phenotype with complete in vivo resistance to AraC/anthracycline-based induction treatment. In fact, the response rates are comparable to other published series of patients older than 55 to 60 years with AML from MDS receiving AraC-based induction treatment.1-9 But since we did not experimentally address the question of P glycoprotein (Pgp) up-regulation during decitabine treatment in MDS patients, we cannot formally exclude that in the cells belonging to the MDS clone, MDR1 was up-regulated during an early treatment phase in the responding patients. Whether or not a demethylation-induced up-regulation ofMDR1 may translate into clinical treatment failure is currently unknown and requires further study. In addition, one should keep in mind that in several tumor entities, including breast and ovarian cancer, overexpression of the MDR1 gene is not necessarily correlated with clinical drug resistance.1-13,1-14Finally, recent studies suggest that the AML subgroup of elderly patients with frequent myelodysplastic features, a higher rate of poor-risk cytogenetics, and p15 hypermethylation may already have increased MDR1 expression prior to any treatment.1-15-1-17 It will be of interest to see whether in comparative studies of normal hematopoietic progenitor cells versus leukemic myeloid cells treated with demethylating agents, preferential up-regulation of MDR1 might occur in the normal cells. This could be hypothesized based on the observation that untransformed murine fibroblasts were more susceptible to the differentiating activity of a demethylating agent than their transformed counterpart.1-18
Supported by Wilhelm Sander-Stiftung (grant 99.032.1)
