Abstract
Hypomethylating agents (HMA) such as azacytidine and decitabine are the mainstay of treatment for higher risk myelodysplastic syndromes (MDS) and are also used to treat older, unfit patients with acute myeloid leukemia (AML). Being cytidine analogues, both azacytidine and decitabine are incorporated into DNA of highly proliferating cells leading to genome-wide decrease of methylation levels (Stresemann & Lyko., 2008; Gnyszka et al., 2013), whereas azacytidine is additionally incorporated into RNA molecules. Although several putative modes of action have been suggested for HMA, the precise mechanism underlying treatment success or failure remains incompletely understood. One possible mechanism of HMA action is through 'viral mimicry' of transcriptionally repressed endogenous retroelements (EREs), which is thought to trigger innate immune pathways. EREs comprise nearly half of the human genome and their transcriptional activity is repressed by diverse mechanisms including DNA methylation. According to the 'viral mimicry' hypothesis, HMA induce unphysiological levels of ERE transcription in transformed cells, which in turn generated nucleic acid species, such as double-stranded RNAs from complementary ERE transcripts, activating innate immune sensors. Although support and a mechanistic basis for this hypothesis is provided from a number of in vitro studies, in vivo evidence from the clinical use of HMA is currently lacking.
To explore the possible involvement of EREs in the HMA mode of action, we have compared the transcriptional profiles of CD34+ HSCs isolated from bone marrow samples of healthy donors (n=9) and patients diagnosed with AML (n=9), chronic myelomonocytic leukemia - II (CMML-II, n=9) or high-risk MDS (n=11). For MDS and CMML, samples were obtained before, 15 days (D15) after the initiation of azacytidine and/or after cycle 6. Our analysis revealed that ERE transcription, measured as a proportion of the total polyA-selected transcriptome, is globally repressed in untreated MDS and CMML, in line with the proposed epigenetic repression that characterizes these conditions. Treatment with azacytidine had a measureable effect in overall ERE transcription in HSCs from MDS and CMML patients, which by the 6th cycle was raised to levels equivalent to those seen in HSCs healthy controls. Comparable results were also obtained following analysis of a publicly available dataset from CD34+ HSCs isolated from MDS and CMML patients prior to and after the 6th cycle of azacytidine treatment (GSE76203). However, despite noticeable upregulation of overall ERE transcription relative to gene transcription by azacytidine, the therapeutic response was not correlated with or predicted by ERE activity. Indeed, ERE transcriptional activation was frequently observed in azacytidine-treated patients who failed to respond to treatment, whereas it was frequently low or absent in patients who attained complete remission (figures 1a & b).
It remained theoretically possible that a therapeutic response to azacytidine depended on the transcriptional activation of a select few ERE loci with innate immune stimulatory properties, which might have been masked by the analysis of global ERE activity. However, few individual ERE loci differed in their activity between patients who responded or not to azacytidine treatment. Moreover, our analysis failed to detect induction of either interferon-inducible genes or interferon-inducible EREs, irrespective of treatment outcome(figures 2a & b). Together, our current results do not support a role for transcriptional activation of EREs or for innate sensing of their nucleic acid products in the therapeutic response of MDS and CMML patients to azacytidine. Investigation of alternative potential mechanisms of azacytidine is therefore warranted.
No relevant conflicts of interest to declare.
