Chromatin Immunoprecipitation of GFP-Tagged PML-Rara Coupled to High-Throughput Next Generation Sequencing.

Abstract 1276

Poster Board I-298

The PML-RARA fusion protein produced by the t(15;17) translocation is found in acute promyelocytic leukemia (APL) and acts as an aberrant transcription factor with oncogenic properties. To define the high affinity DNA binding sites of PML-RARA, we developed a novel system based upon chromatin immunoprecipitation of eGFP tagged PML-RARA (PR), coupled to next generation sequencing. Chromatin isolated from flow sorted, GFP+ PR9 cells (isolated 24 hours after electroporation with eGFP-PR) was immunoprecipitated with a highly specific anti-GFP monoclonal antibody, and massively parallel sequencing was performed using single-end read libraries generated from both input DNA and immunoprecipitated DNA. The sequenced reads were mapped to the reference human genome using the Burrows-Wheeler Alignment tool (BWA). Using the SamTools package, alignment files were then filtered to retain only reads with phred quality scores greater than 30. Finally, Model Based Analysis of ChIP-Seq (MACS) was used to obtain the predicted binding sites (approximately 13,000 for each replicate using a p value cutoff of 0.00001). To ensure the reproducibility of called peaks, we limited our analysis to 701 sites that occurred in replicate sequencing runs of the libraries, within a tolerance of ±50 bp at the peak. Visual inspection of graphically plotted peaks further filtered this list to 421 high quality sites. Using microarray expression data for 14 APL patients and 5 flow sorted, normal promyelocyte samples, we selected neighboring genes with at least a 3-fold difference between APL and promyelocyte expression. This step yielded a list of 82 neighboring genes whose expression may be altered by PML-RARA binding to adjacent DNA. While PML-RARA is often considered a transcriptional repressor, 68% of these genes were overexpressed in APL, suggesting that PML-RARA may also function as a transcriptional activator. Furthermore, 51 genes were dysregulated in the mCG-PML-RARA murine model of APL, 16 genes demonstrated expression changes following induction of PML-RARA expression in PR9 cells, and 12 genes had altered expression in both the PR9 and murine models. Collectively, these results demonstrate an association between the high confidence putative PR binding sites and gene expression changes. We next used the sequences found within the best 89 binding sites to define 6 potential in vivo consensus sites of PML-RARA using the CONSENSUS program. Three of these predicted sites were present in greater than 40% of the 421 high confidence sites. Two of these common motifs resemble the motifs found in retinoic acid response elements (RAREs), but have less stringent conservation at the 5' end while retaining the 3' TCA sequence. Taken together, our results suggest that PML-RARA has an extended repertoire of genomic DNA binding sites compared to wild-type RARA, reflecting novel gain-of-function properties of the fusion protein. Binding of some of these sites appears to have direct consequences for the expression of several tightly linked genes, which may themselves be involved in transcriptional regulatory networks that contribute to APL pathogenesis.