Bubble-seq analysis of the human genome reveals distinct chromatin-mediated mechanisms for regulating early- and late-firing origins.

We have devised a method for isolating virtually pure and comprehensive libraries of restriction fragments that contained replication initiation sites (bubbles) in vivo. We have now sequenced and mapped the bubble-containing fragments from GM06990, a near-normal EBV-transformed lymphoblastoid cell line, and have compared origin distributions with a comprehensive replication timing study recently published for this cell line. We find that early-firing origins, which represent ∼32% of all origins, overwhelmingly represent zones, associate only marginally with active transcription units, are localized within large domains of open chromatin, and are significantly associated with DNase I hypersensitivity. Origin "density" falls from early- to mid-S-phase, but rises again in late S-phase to levels only 17% lower than in early S-phase. Unexpectedly, late origin density calculated on the 1-Mb scale increases as a function of increasing chromatin compaction. Furthermore, the median efficiency of origins in late-replicating, heterochromatic domains is only 25% lower than in early-replicating euchromatic loci. Thus, the activation of early- and late-firing origins must be regulated by quintessentially different mechanisms. The aggregate data can be unified into a model in which initiation site selection is driven almost entirely by epigenetic factors that fashion both the long-range and local chromatin environments, with underlying DNA sequence and local transcriptional activity playing only minor roles. Importantly, the comprehensive origin map we have prepared for GM06990 overlaps moderately well with origin maps recently reported for the genomes of four different human cell lines based on the distributions of small nascent strands.

The effect of in vitro dieldrin exposure on the rat paraoxonase 1 (pon1) promoter.

The legacy organochlorine insecticide, dieldrin, is still found in soil and accumulation in individuals is possible. Paraoxonase 1 hydrolyzes the oxon metabolites of organophosphorus insecticides, as well as other substrates. Putative binding sites for pregnane X receptor (PXR) exist in the paraoxonase promoter, and studies have indicated that dieldrin can activate PXR-regulated gene expression. We examined rat paraoxonase promoter activity in the presence of dieldrin alone or combined with nuclear receptors (NRs). In vitro dieldrin concentrations from 10 to 100 µM significantly increased (p < 0.05) promoter activity in the presence of Pxr or Rxrα alone and when Pxr plus Rxrα were on the same vector, indicating that dieldrin can increase paraoxonase promoter activity in the presence of NRs. To our knowledge, this is the first report of dieldrin increasing paraoxonase promoter activity. Since many organochlorine insecticides are in the same chemical class as dieldrin, these results could be typical of other bioaccumulative persistent pollutants.

Fine mapping and functional studies of risk variants for type 1 diabetes at chromosome 16p13.13.

Single nucleotide polymorphisms (SNPs) located in the chromosomal region 16p13.13 have been previously associated with risk for several autoimmune diseases, including type 1 diabetes. To identify and localize specific risk variants for type 1 diabetes in this region and understand the mechanism of their action, we resequenced a 455-kb region in type 1 diabetic patients and unaffected control subjects, identifying 93 novel variants. A panel of 939 SNPs that included 46 of these novel variants was genotyped in 3,070 multiplex families with type 1 diabetes. Forty-eight SNPs, all located in CLEC16A, provided a statistically significant association (P < 5.32 × 10(-5)) with disease, with rs34306440 being most significantly associated (P = 5.74 × 10(-6)). The panel of SNPs used for fine mapping was also tested for association with transcript levels for each of the four genes in the region in B lymphoblastoid cell lines. Significant associations were observed only for transcript levels of DEXI, a gene with unknown function. We examined the relationship between the odds ratio for type 1 diabetes and the magnitude of the effect of DEXI transcript levels for each SNP in the region. Among SNPs significantly associated with type 1 diabetes, the common allele conferred an increased risk for disease and corresponded to lower DEXI expression. Our results suggest that the primary mechanism by which genetic variation at CLEC16A contributes to the risk for type 1 diabetes is through reduced expression of DEXI.