ABSTRACT
A DNA affinity column containing a synthetic double-stranded nuclear matrix attachment region (MAR) was used to purify a 100-kDa protein from human erythroleukemia K562 cells. This protein was identified as nucleolin, the key nucleolar protein of dividing cells, which is thought to control rRNA gene transcription and ribosome assembly. Nucleolin is known to bind RNA and single-stranded DNA. We report here that nucleolin is also a MAR-binding protein. It binds double-stranded MARs from different species with high affinity. Nucleolin effectively distinguishes between a double-stranded wild-type synthetic MAR sequence with a high base-unpairing potential and its mutated version that has lost the unpairing capability but is still A+T rich. Thus, nucleolin is not merely an A+T-rich sequence-binding protein but specifically binds the base-unpairing region of MARs. This binding specificity is similar to that of the previously cloned tissue-specific MAR-binding protein SATB1. Unlike SATB1, which binds only double-stranded MARs, nucleolin binds the single-stranded T-rich strand of the synthetic MAR probe approximately 45-fold more efficiently than its complementary A-rich strand, which has an affinity comparable to that of the double-stranded form of the MAR. In contrast to the high selectivity of binding to double-stranded MARs, nucleolin shows only a small but distinct sequence preference for the T-rich strand of the wild-type synthetic MAR over the T-rich strand of its mutated version. The affinity to the T-rich synthetic MAR is severalfold higher than to its corresponding RNA and human telomere DNA. Quantitative cellular fractionation and extraction experiments indicate that nucleolin is present both as a soluble protein and tightly bound to the matrix, similar to other known MAR-binding proteins.
