Genome Wide Profiling of USF1 and USF2 Occupancy in Human Primary Erythroid Cells.

Abstract 1456

Poster Board I-479

USF1 and USF2 are ubiquitously expressed basic helix-loop-helix leucine zipper proteins that participate in a large number of biologic processes. USF1 and USF2 bind DNA as homodimers or heterodimers, typically binding E box consensus motifs. One role of USF proteins is functioning as transcription factors. Although ubiquitously expressed, they regulate expression of many cell-type and developmental-stage specific genes, such as hepcidin in hepatocytes and surfactant protein A in fetal lung cells. Another role of USF proteins is in the maintenance of chromatin architecture in barrier insulator elements, such as the well characterized 5'HS4 insulator element in the chicken beta-globin locus. In mammalian erythroid cells, USF1 and USF2 participate in the regulation of beta-globin transcription, interacting both at hypersensitive site 2 (HS2) of the beta-globin locus control region (LCR) and at the beta-globin promoter. Depletion of USF proteins leads to decreased beta-globin production. We hypothesize that in addition to beta-globin, USF proteins are important for regulation of many erythroid expressed genes. To address this hypothesis, chromatin immunoprecipitation with antibodies against USF1 and USF2 was coupled with ultra high throughput, massively parallel sequencing (Illumina Solexa sequencing, ChIP-seq) to generate a genome-wide map of USF1 and USF2 occupancy in primary erythroid cells. To generate cells for ChIP and mRNA expression profiling, human CD34+ cells isolated from peripheral blood were cultured in serum free media with erythropoietin to induce erythroid differentiation. After 14 days in culture, FACS analysis was used to confirm cells were positive for both CD 71 and glycophorin A (the R3/R4 stage of erythroid development). mRNA transcript analyses were performed using Illumina human V6-2 expression arrays and quantitative real time RT-PCR. ChIP-seq experiments for USF1 and USF2 were done in duplicate and only binding sites present in both ChIP-seq replicates were included in data analyses. A total of 20450 USF1 and 21128 USF2 sites of occupancy were identified. Co-localization of USF1 and USF2 was common, with 16739 sites binding both USF1 and USF2 (81.9% of USF1 sites and 79.2% of USF2 sites). In an analysis of a subset of erythroid expressed focus genes, USF binding was associated with active transcription. In agreement with previous studies, there was binding of USF proteins in the beta-globin LCR, and beta-globin promoter. USF binding most commonly occurred close to annotated genes, with 48.5% of USF1 sites, 44.6% of USF 2 sites and 53.0% of sites of USF1-USF2 co-localization located within 1 kb of a transcription start site (TSS), supporting the role of USF proteins as a transcription factor in these locations. A small, but significant, number of USF binding sites were located in intergenic regions > 100 kb from any annotated TSS. (1206 USF1, 1408 USF2, and 776 USF1-USF2). Interestingly, at sites of intergenic binding, USF1 and USF2 were much less likely to co-localize, (64% of USF1 and 55% of USF2 sites), implying that the USF proteins serve a different function at these remote binding sites than at sites of binding in close proximity to a TSS. USF proteins can bind DNA in an E-box dependent or independent manner. The Weeder Algorithm (Pavesi, Bioinformatics, 2001) was used to determine the most common binding motifs for USF1 and USF2. Over-represented motifs at sites of USF1 and USF2 binding were similar, with the most common sequences being a canonical E-box, CACGTG, as well as the related sequences ACGTGA and TCACGT. This genome-wide map of USF binding correlated with mRNA expression data indicates that USF proteins serve several different, important functions throughout the human genome and support the hypothesis that USF proteins participate in the regulation of many erythroid-expressed genes.