Large-Scale Identification and Mapping of Nuclear Matrix-Attachment Regions in the Distal Imprinted Domain of Mouse Chromosome 7

Abstract

Mammalian imprinted genes, which are expressed from only one of the parental alleles, have a tendency to form clusters and are regulated by long-range mechanisms. Nuclear matrix-attachment regions (MARs), the anchor points of loop domains, are involved in coordination of gene expression and could play a role in regulation of imprinted domains. We have identified and mapped a total of 52 MARs in a 1-Mb imprinted domain on mouse distal chromosome 7 using our cosmid contigs and an in vitro MAR assay. We find two MAR clusters (comprising 20 and 19 MARs), one of which is mapped in the Th-Ins2 intergenic region, coincident with the boundary between the two imprinted subdomains. However, the imprinted/non-imprinted boundaries are not associated with a MAR. Based on the sequence information, we find that many of the MARs are rich in long interspersed nuclear elements. In addition, comparisons of the results obtained with several MAR-prediction software programs reveal good performance of ChrClass in terms of both sensitivity and specificity. This study presents the first large-scale mapping of MARs in an imprinted domain and provides a platform for understanding the roles of MARs in imprinting.

Depletion of Kcnq1ot1 non-coding RNA does not affect imprinting maintenance in stem cells

“…In mouse, the region between Th and Ins2 is over 200 kb and contains an extraordinarily high concentration of tandem repeats, long interspersed nuclear element (LINE) retrotransposons and endogenous retroelements (Shirohzu et al, 2004). This repeat-rich region also exhibits asynchronous replication, similar to the rest of the domain, and contains 18 matrix attachment regions (Yatsuki et al, 2000;Purbowasito et al, 2004). In human, the distance between TH and INS is considerably shorter (~10 kb).…”

In-silico prediction and observations of nuclear matrix attachment

“…This exemplifies the complex nature of the nuclear matrix detection problem. Using these and other in-silico approaches and again tolerating a substantial over-prediction, Purbowasito aligned 95% of the experimentally validated MARs [79] with at least one of the predictions. These and other studies have consistently revealed a residual variance even after the substantial overprediction between the in-silico predictions and the in-vivo evidence.…”

“…The incorporation of a statistical distance correction term and a cross-interaction matrix based on observed coincidence or anti-coincidence of MAR sequences may extend these approaches. Several studies have been pursued to biologically evaluate nuclear matrix binding relative to in-silico predictions [31,79]. One of the first studies compared the ability of MARFinder, MARSCAN and SMARTest to accurately identify the MARs encompassing the human beta-globin domain [31].…”

Revisiting the function of nuclear scaffold/matrix binding proteins in X chromosome inactivation

“…Interestingly, gene loci that exhibit monoallelic expression have several correlations with S/MARs and LINEs. 17,29,30 Future works will evaluate the contribution of the nuclear scaffold/matrix proteins in epigenetic gene regulation and noncoding RNA-based chromatin modifications.…”

“…15 More strikingly, the LINE sequences are overrepresented by nearly two-fold among the human S/MARs, relative to their overall abundance in the human or mouse genome. 16,17 Because LINEs prefer AT-rich regions as their integration sites, their close correlation with the S/MARs that are innately AT-rich may be due to these preferences. 7 Alternatively, integrated LINE-1 sequences themselves serve as a part of the S/MARs.…”