Two substrates are better than one: dual specificities for Dnmt2 methyltransferases Albert Jeltsch 1 , Wolfgang Nellen 2 and Frank Lyko 3 1 School of Engineering and Science, International University Bremen, Campus Ring 1, 28759 Bremen, Germany 2 FB 18, Division of Genetics, University of Kassel, Heinrich-Plett-Strasse 40, 34132 Kassel, Germany 3 Division of Epigenetics, Deutsches Krebsforschungszentrum, Im Neuenheimer Feld 580, 69120 Heidelberg, Germany Dnmt2 enzymes have been widely conserved during evolution and contain all of the signature motifs of DNA (cytosine-5)-methyltransferases; however, the DNA methyltransferase activity of these proteins is compara- tively weak and their biochemical and functional proper- ties remain enigmatic. Recent evidence now shows that Dnmt2 has a novel tRNA methyltransferase activity, raising the possibility that the biological roles of these proteins might be broader than previously thought. This finding has important implications for understanding the evolutionary relationships among these enzymes. Eukaryotic DNA methyltransferases DNA methylation is important in organismal development and human disease [1,2]. In eukaryotes, there are three distinct families of DNA methyltransferases: Dnmt1, Dnmt2 and Dnmt3 [3]. Whereas Dnmt3 enzymes seem to be responsible for establishing DNA methylation patterns (de novo methyltransferases), Dnmt1 enzymes are involved in maintaining methylation patterns. By contrast, the biochemical activity and the biological function of Dnmt2 enzymes have remained enigmatic for a long time. Dnmt2 proteins contain the consensus catalytic motifs of DNA (cytosine-5)-methyltransferases and have been strongly conserved during evolution (Figure 1). In addition, the 3D structure of human DNMT2 has been obtained by X-ray crystallography and is highly similar to that of bacterial DNA methyltransferases [4]. In agree- ment with this structural conservation, Dnmt2 has been shown, by different methods in various systems, to have DNA methyltransferase activity [5–10]. Although some indications of the biological importance of Dnmt2 have been obtained (see later), the activity of these enzymes is comparatively weak; thus, it has remained possible that Dnmt2 might also methylate substrates other than DNA. Dnmt2 methylates tRNA Asp In a recent biochemical study, Goll et al. [11] generated Dnmt2 knockouts in diverse model systems (mouse, Drosophila and Arabidopsis) to investigate the function of this enzyme. Using RNA preparations from these Dnmt2 mutants, they found that tRNA Asp is methylated with high efficiency by Dnmt2. Intriguingly, this modification seems to be highly specific, because they did not detect Dnmt2-dependent methylation of other RNA molecules. Detailed mass spectrometric analyses indi- cated that the methylation takes place at cytosine 38, the last base of the tRNA anticodon loop. Goll et al. [11], however, did not check for differences in the function of the methylated and nonmethylated tRNA Asp molecules and, despite the strong evolutionary conservation of Dnmt2 enzymes, they did not detect phenotypic abnormalities in the mutants. Thus, the biological role of the methylation of tRNA Asp remains unclear. Of note, Dnmt2 activity has been associated with RNA interference in Dictyostelium [9] and with covalent histone modifications in Drosophila [7], which suggests that Dnmt2 has a role in epigenetic regulation. Whether this role is related to the DNA methyltransferase or the tRNA Asp methyltransferase activity of Dnmt2 remains to be determined. Functional characterization of Dnmt2 enzymes Drosophila and Dictyostelium are especially well suited to functional studies because both organisms contain a single Dnmt2-like methyltransferase candidate gene; thus, their use avoids the need to differentiate experi- mentally among related enzymatic activities with poten- tially overlapping functions. In both organisms, the gene encoding Dnmt2 is differentially expressed during development, suggesting that it might have a regulatory function. The mutation or knockout of Dnmt2 in either organism does not result in obvious phenotypes [9,11], even though a Dnmt2 mutant strain of Dictyostelium has been shown to exhibit minor morphological defects [10]. Microarray experiments and random sequencing have been used to identify several genomic loci, including the retroelements DIRS-1 and Skipper, that are methylated in wild type but not in Dnmt2 mutant Dictyostelium [9]. For the Skipper retroelement, loss of methylation results in enhanced expression and an increase in genomic copy number. Methylation thus seems to be required for transcriptional silencing of this DNA. By contrast, expression and transposition of DIRS-1 are not affected by the loss of methylation. Similar to the situation in Drosophila [12], some Dictyostelium genes are methylated at a few cytosine residues in their coding sequence, but the functional role of this modification remains unknown. Our understanding of Dnmt2 should benefit greatly from characterizing Dnmt2 proteins at the cellular level. Corresponding author: Lyko, F. (f.lyko@dkfz.de). Available online 6 May 2006 Update TRENDS in Biochemical Sciences Vol.31 No.6 June 2006 306 www.sciencedirect.com