Characterization of randomly-obtained matrix attachment regions (MARs) from higher plants.

Matrix attachment regions (MARs) can be operationally defined as DNA fragments that bind to the nuclear matrix. We have created a library of randomly obtained MARs from tobacco (Nicotiana tobacum) by cloning DNA fragments that co-isolate with nuclear matrixes prepared by a method involving lithium diiodosalicylate. The interactions of several of the cloned MARs with nuclear matrixes were tested by an in vitro binding assay in which genomic DNA was used as competitor. Based on this assay, the MARs were classified as strong, medium, and weak binders. Examples of each of the binding classes were further studied by in vitro binding using self- and cross-competition. Estimates of dissociation constants for several MARs ranged from 6 to 11 nM and correlated inversely with binding strength. The number of binding sites per matrix for several MARs ranged from 4 x 10(5) to 9 x 10(5) and correlated directly with binding strength. We conclude that binding strength, as we have measured it, is a function of both numbers of binding sites and affinity for the sites. The tobacco MARs were sequenced and analyzed for overall AT content, for distribution of AT-rich regions, and for the abundance of several MAR-related motifs. Previously identified MAR motifs correlate to various degrees with binding strength. Notably, the Drosophila topoisomerase II motif does not correlate with binding strength of the tobacco MARs. A newly identified motif, the "90%AT Box," correlates better with binding strength than any of the previously identified motifs we investigated.