A Comprehensive Genomic Approach Using Gain of Function and Loss of Function Cell Models and ChIP-on-Chip Technology Identifies Novel Promyelocytic Zinc Finger Protein Target Genes.

The t(11;17)(q23;q21) form of APL involves the production of reciprocal fusion proteins, PLZF-RARα and RARα-PLZF, which mediate malignant transformation by binding to and dysregulating RARα and PLZF target genes. PLZF is expressed in hematopoietic stem cells and is downregulated as cells differentiate. The identification of PLZF target genes including cyclin A2 and MYC is consistent with the hypothesis that PLZF maintains stem cell quiescence by repressing cell cycle driving genes and provides insight into transcriptional pathways disrupted in leukemogenesis. In order to identify additional target genes of PLZF, we constructed a loss of function model in which we suppressed endogenous expression of PLZF using siRNA in KG1a leukemia cells. Our gain of function model consisted of the ectopic expression of PLZF in U937 leukemia cells which do not naturally express PLZF. Expression profiling using GeneChip™ Human Genome U133 Plus 2.0 arrays, which analyze the expression of more than 47,000 transcripts, was performed using both systems. Of the 346 genes identified in the loss of function model, 25% were also regulated by PLZF in the gain of function U937 cell line. Changes in expression of these genes could be direct (through PLZF) or indirect (through secondary effects). In order to determine which genes modulated by changes in PLZF expression are direct transcriptional targets, we immunoprecipitated chromatin using PLZF antibodies in KG1a cells, amplified the products by ligation-mediated PCR and co-hybridized these products with input chromatin to NimbleGen 1.5kB promoter arrays, which represent 24,275 human promoters. Genes bound by PLZF were identified by determining whether consecutively tiled probes were enriched in PLZF-precipitated chromatin as compared to chromatin precipitated with a non-specific antibody. Using a statistical algorithm designed to exclude those probes whose signals of PLZF enrichment might be spuriously identified, we identified 52 genes of the 24,275 on the array as potential PLZF target genes. Strikingly, correlation of these genes with expression analyses revealed that 44% of genes were also significantly regulated by PLZF in the gain of function model and 11% of genes were regulated in the loss of function model. Promoter analyses of a subset of these genes that were identified by ChIP-on-Chip and differentially expressed at least >1.3 fold in PLZF arrays (p<0.05), revealed the presence of a consensus PLZF binding site GTC(C/A)AG in 75% of genes. Analysis of gene ontology for those genes identified by ChIP-on Chip, revealed an enrichment of genes involved in RNA binding and processing as well as genes encoding small G proteins. One gene in particular, RECQL, was directly bound by PLZF in the ChIP-on-Chip assay and transcriptionally regulated by PLZF in both KG1a loss of function and U937 gain of function models. The RECQL protein is a member of the RecQ family of DNA helicases, a class of genes whose mutation is associated with genomic instability tumorigenesis and premature ageing. These data indicate a robust system for the identification of PLZF targets and suggest that PLZF may play a role in genome integrity.