Increased Global Gene Promoter Methylation after Relapse (Rel) of Acute Promyelocytic Leukemia (APL) from All-trans Retinoic Acid (ATRA)-Containing Treatment Is Dissociated from Concurrent Gene Expression Changes.

Increased methylation of gene-regulatory DNA sequences has been associated with the progression of malignancies, and, in APL, the pathogenetic PML-RARα fusion gene has been associated with DNA methyltransferase activity, which might contribute to this process. We, thus, hypothesized that promoter methylation levels would be increased in APL samples at disease Rel and that this might be contributory to gene expression changes related to Rel. We investigated this hypothesis using matched pretreatment (PTx) and Rel DNA and RNA samples from 5 patients of whom 3 had received combined chemotherapy(CT)-ATRA (A-set) and 2 had received only CT (C-set); one additional RNA matched pair was available from another CT-only patient. The specimens used for DNA/RNA extraction had ≥ 85% blasts. The HELP assay (Khulan, et al, Genome Res. 16, 1046, 2006), which is based on comparative hybridization of HpaII (methylation-sensitive) versus MspI (methylation-insensitive) digested genomic DNA to a microarray representing the promoters of ∼24,000 human genes, was used to distinguish hypermethylated loci (HR) from hypomethylated loci (HO). Gene expression analysis utilized the Affymetrix Human Genome U-133 Plus 2.0 chip. Unsupervised cluster analysis, performed on subsets of HELP gene probes (hgp), selected for variation of ≥1 sample by ≥2-fold from the mean methylation value of all samples, demonstrated segregation by patient in 4/5 of the paired DNA sets but by treatment type (A-set vs C-set) at Rel. A subset of 6,220 Affymetrix gene probe sets (gps), selected for a mean expression level of ≥100 and a mean absolute change from PTx to Rel of ≥100 among the 3 pairs of A-set and C-set RNA samples, also segregated by treatment type at Rel. In an overall analysis of methylation based on >23,000 hgp, the average HR increased by 5% in A-set and decreased by 6% in C-set patients. Conversely, the average HO decreased by 4.3% in the A-set and increased by 7% in the C-set patients. In order to more critically evaluate the methylation changes and their possible relationship to gene expression changes, we selected a subset of hgp within −500 to +200nt of the transcription start site (TSS) in each gene promoter. Further selecting these hgp for an ≥1.5-fold change from PTx to Rel for each matched sample pair from the A- or C-sets and performing a scatter plot analysis of this selected hgp subset versus the change in expression of all 6,220 Affymetrix gps, there was a predominant increase in the HR hgp in the A-set and of HO hgp in the C-set, paralleling the overall hgp results. However, the distribution of methylation changes was random with respect to changes in gene expression. Similar results were observed by a reverse selection analysis starting from Affymetrix gps that were changed by ≥1.5-fold in all samples from each set (559, A-set; 361, C-set) and plotting versus all near-TSS hgp (15,782 hgp). Within the limits of the small sample size analyzed, we conclude that treatment type affects both gene promoter methylation and gene expression at Rel, and, most confidently, that Rel from ATRA-containing treatment was associated with increased gene promoter methylation that was not broadly correlated with gene expression changes. Further studies are required to assess whether coordinated methylation/ expression changes occur in key genes with pathogenetic potential.