During our work on a nuclear complex involved in a disease, ATRX syndrome, we noticed a possibly important clue to its function: it contains a PHD finger motif that resembles the one in de novo DNA methyltransferases, DNMT3a and DNMT3b. This would imply a potential connection between ATRX and DNA methylation. Indeed, levels of DNA methylation are altered at several chromosomal loci in cell lines derived from ATRX patients compared to those from unaffected individuals. It is possible that ATRX could open up chromatin structure to facilitate methylation of DNA by methyltransferase, analogous to the role of CHD4 of the NURD complex in helping histone deacetylase to modify histones. However, there has been no biochemical evidence linking the two processes. This encouraged us to begin more targeted studies of DNA methylation, which is increasingly studied for its roles in mammalian development, differentiation, several disease conditions, and aging. In our initial studies, we were especially motivated to investigate whether the de novo DNA methyltransferases and ATRX are components of one complex. We established collaboration with Dr. En Li's group, which has cloned DNMT3a and DNMT3b, and has antibodies to both proteins, knockout mice and other valuable reagents. Using their antibodies, we found no detectable levels of DNMT3a or DNMT3b in the ATRX complex by either mass spectrometry or immunoblotting. To rule out the possibility that only a small fraction of ATRX may associate with methyltransferase, we immunoisolated DNA methyltransferase complexes. Despite the importance of these enzymes, complexes containing these proteins have not been extensively purified or characterized.] In repeated experiments, the immunoisolated DNMT1 does not appear to have any associated proteins. This is in contrast to previous reports that DNMT1 associates with several important proteins, including the retinoblastoma protein, histone deacetylase and E2F1, though we cannot rule out the possibility that our DNMT1 antibody has displaced these proteins from the DNMT1. In contrast to results with DNMT1, DNMT3b and DNMT3a were found to coimmunoprecipitate with several polypeptides, which could be components of potential methyltransferase complexes. Mass spectrometric analysis confirmed the presence of DNMT3b and DNMT3a in these polypeptides. Again, ATRX was not detected among these polypeptides, so our original question is answered negatively. However, some of the polypeptides identified can help to understand the action of methyltransferases. In genetic studies in plants, mutation in a SWI2/SNF2-like ATPase (DDM1) leads to sharp reduction of DNA methylation in the plant genome, again suggesting a connection between ATP-dependent chromatin remodeling and DNA methylation. We hypothesized that DDM1 might be part of a methyltransferase complex and attempted to purify a complex containing the mouse homolog of DDM1, LSH. We established a collaboration with Kathrin Muegge's lab, which has cloned LSH gene and developed many reagents, including antibodies and knockout mice. Using these reagents, we successfully purified the endogenous LSH. Unfortunately, we found that the complex containing LSH consists of only itself, but not other partners. We are currently studying whether LSH contains ATP-dependent chromatin-remodeling activity like other SWI/SNF proteins.
