Prolonged activity of the transposase helper may raise safety concerns during DNA transposon-based gene therapy.

DNA transposon-based gene delivery vectors represent a promising new branch of randomly integrating vector development for gene therapy. For the side-by-side evaluation of the piggyBac and Sleeping Beauty systems-the only DNA transposons currently employed in clinical trials-during therapeutic intervention, we treated the mouse model of tyrosinemia type I with liver-targeted gene delivery using both transposon vectors. For genome-wide mapping of transposon insertion sites we developed a new next-generation sequencing procedure called streptavidin-based enrichment sequencing, which allowed us to identify approximately one million integration sites for both systems. We revealed that a high proportion of piggyBac integrations are clustered in hot regions and found that they are frequently recurring at the same genomic positions among treated animals, indicating that the genome-wide distribution of Sleeping Beauty-generated integrations is closer to random. We also revealed that the piggyBac transposase protein exhibits prolonged activity, which predicts the risk of oncogenesis by generating chromosomal double-strand breaks. Safety concerns associated with prolonged transpositional activity draw attention to the importance of squeezing the active state of the transposase enzymes into a narrower time window.

ChIPSummitDB: a ChIP-seq-based database of human transcription factor binding sites and the topological arrangements of the proteins bound to them.

ChIP-seq reveals genomic regions where proteins, e.g. transcription factors (TFs) interact with DNA. A substantial fraction of these regions, however, do not contain the cognate binding site for the TF of interest. This phenomenon might be explained by protein-protein interactions and co-precipitation of interacting gene regulatory elements. We uniformly processed 3727 human ChIP-seq data sets and determined the cistrome of 292 TFs, as well as the distances between the TF binding motif centers and the ChIP-seq peak summits. ChIPSummitDB enables the analysis of ChIP-seq data using multiple approaches. The 292 cistromes and corresponding ChIP-seq peak sets can be browsed in GenomeView. Overlapping SNPs can be inspected in dbSNPView. Most importantly, the MotifView and PairShiftView pages show the average distance between motif centers and overlapping ChIP-seq peak summits and distance distributions thereof, respectively. In addition to providing a comprehensive human TF binding site collection, the ChIPSummitDB database and web interface allows for the examination of the topological arrangement of TF complexes genome-wide. ChIPSummitDB is freely accessible at http://summit.med.unideb.hu/summitdb/. The database will be regularly updated and extended with the newly available human and mouse ChIP-seq data sets.

Motif oriented high-resolution analysis of ChIP-seq data reveals the topological order of CTCF and cohesin proteins on DNA.

BACKGROUND: ChIP-seq provides a wealth of information on the approximate location of DNA-binding proteins genome-wide. It is known that the targeted motifs in most cases can be found at the peak centers. A high resolution mapping of ChIP-seq peaks could in principle allow the fine mapping of the protein constituents within protein complexes, but the current ChIP-seq analysis pipelines do not target the basepair resolution strand specific mapping of peak summits. RESULTS: The approach proposed here is based on i) locating regions that are bound by a sufficient number of proteins constituting a complex; ii) determining the position of the underlying motif using either a direct or a de novo motif search approach; and iii) determining the exact location of the peak summits with respect to the binding motif in a strand specific manner. We applied this method for analyzing the CTCF/cohesin complex, which holds together DNA loops. The relative positions of the constituents of the complex were determined with one-basepair estimated accuracy. Mapping the positions on a 3D model of DNA made it possible to deduce the approximate local topology of the complex that allowed us to predict how the CTCF/cohesin complex locks the DNA loops. As the positioning of the proteins was not compatible with previous models of loop closure, we proposed a plausible "double embrace" model in which the DNA loop is held together by two adjacent cohesin rings in such a way that the ring anchored by CTCF to one DNA duplex encircles the other DNA double helix and vice versa. CONCLUSIONS: A motif-centered, strand specific analysis of ChIP-seq data improves the accuracy of determining peak positions. If a genome contains a large number of binding sites for a given protein complex, such as transcription factor heterodimers or transcription factor/cofactor complexes, the relative position of the constituent proteins on the DNA can be established with an accuracy that allow one to deduce the local topology of the protein complex. The proposed high resolution mapping approach of ChIP-seq data is applicable for detecting the contact topology of DNA-binding protein complexes.

PRMT1 and PRMT8 regulate retinoic acid-dependent neuronal differentiation with implications to neuropathology.

Retinoids are morphogens and have been implicated in cell fate commitment of embryonic stem cells (ESCs) to neurons. Their effects are mediated by RAR and RXR nuclear receptors. However, transcriptional cofactors required for cell and gene-specific retinoid signaling are not known. Here we show that protein arginine methyl transferase (PRMT) 1 and 8 have key roles in determining retinoid regulated gene expression and cellular specification in a multistage neuronal differentiation model of murine ESCs. PRMT1 acts as a selective modulator, providing the cells with a mechanism to reduce the potency of retinoid signals on regulatory "hotspots." PRMT8 is a retinoid receptor target gene itself and acts as a cell type specific transcriptional coactivator of retinoid signaling at later stages of differentiation. Lack of either of them leads to reduced nuclear arginine methylation, dysregulated neuronal gene expression, and altered neuronal activity. Importantly, depletion of PRMT8 results in altered expression of a distinct set of genes, including markers of gliomagenesis. PRMT8 is almost entirely absent in human glioblastoma tissues. We propose that PRMT1 and PRMT8 serve as a rheostat of retinoid signaling to determine neuronal cell specification in a context-dependent manner and might also be relevant in the development of human brain malignancy.