SPOUT: a class of methyltransferases that includes spoU and trmD RNA methylase superfamilies, and novel superfamilies of predicted prokaryotic RNA methylases.

Nucleotide modification is a major source of structural and biochemical diversity of RNAs, particularly of rRNAs and tRNAs (Rozenski et al., 1999). Basespecific methylation is one of the prevalent modes of post-transcriptional modification along with pseudouridylation and thio-uridylation. The catalytic domains of the majority of RNA methylases belong to the vast la rge superc lass o f Rossmanfo ld enzymes (Lo Conte et al., 2000). These Rossman-fold methylases (RFM) bind S-adenosyl-L-methionine (SAM) via the nucleotide-binding loop typical of this fold and catalyze the transfer of its methyl group to substrates such as DNA, RNA, proteins or small molecules. However, two families of RNA methylases, typified, respectively, by Escherichia coli proteins SpoU (TrmH) and TrmD, appear to be unrelated to the RFMs. SpoU family proteins have been characterized as tRNA (G) methyltransferases in E. coli and S. cerevisiae (Cavaille et al., 1999; Koonin et al., 1993; Persson et al., 1997), whereas TrmD is responsible for the bacteriaspecific tRNA methylation at the G position (Bjork et al., 1989; Li et al., 1999). Here, we provide evidence for a common evolutionary origin of the TrmD and SpoU methylase superfamilies that were previously considered unrelated. We show that the SpoU superfamily encompasses a greater diversity of (predicted) methylases than previously appreciated and contains two previously undetected families that are specific to the archaea and thermophilic bacteria. Additionally, we identify two previously unnoticed superfamilies of proteins from archaea and bacteria, respectively, that are predicted to be methylases related to both the TrmD and SpoU superfamilies. These four superfamilies together define a new structural scaffold for RNA methylase activity. As part of a systematic survey of the enzymes involved in RNA metabolism (VA, EVK and LA, unpublished), we investigated the SpoU and TrmD families by iterative searches of the Non-redundant database (National Center for Biotechnology Information, NIH, Bethesda) using the PSI-BLAST program (Altschul et al., 1997). Such a search (profile inclusion expectation (E) value threshold of 0.01) seeded with the E. coli SpoU sequence detected not only the obvious SpoU orthologs and previously described homologs (Koonin, 1996) from bacteria, archaea, and eukaryotes, but also several uncharacterized proteins from archaea (typified by MJ1385) and the thermophilic bacteria Aquifex and Thermotoga (typified by aq_054). These proteins showed significant E-values (10!3!10!7) on first detection and retained the conservation pattern typical of the SpoU proteins (Figure 1), suggesting that they were bona fide members of this superfamily. Additionally, in these searches we detected several uncharacterized archaeal proteins, typified by AF2226, and the TrmD methylases with marginally significant E-values (.06-.1). A reciprocal search started with the AF2226 sequence recovered the TrmD methylases and SpoU superfamily members with similar borderline E-values. Another confirmatory PSIBLAST search initiated with the TrmD ortholog from Aquifex aeolicus (2983865) detected not only the AF2226-like archaeal proteins, but also members of a highly conserved superfamily of predominantly bacterial proteins, typified by YbeA from E. coli, with significant E-values (<0.001). The TrmD methylases, the SpoU methylases, and the newly detected AF2226like, and YbeA-like proteins showed striking conservation in the region predicted to be the SAM-binding loop in the SpoU superfamily (Koonin, 1996) (Figure 1). Furthermore, in the archaeal AF2226-like proteins, the putative methylase domain is fused with a recently described RNA-binding domain, THUMP, that has also been found fused to classic, Rossmann-fold RNA methylases, thiouridine synthases and pseudouridine synthases (Aravind et al., 2001). These observations suggested that the SpoU superfamily, the TrmD superfamily, and the newly detected AF2226-like and YbeA-like superfamilies, belong to a class of methyltransferases with a common structural fold that does not include any structurally characterized proteins and is distinct from the classic Rossmann fold. To investigate this relationship further, we searched the entire set of members of the SpoU, TrmD, YbeA and AF2226 superfamilies for potential conserved motifs using the Gibbs sampling procedure (Neuwald et al., 1997). Two motifs were detected, one corresponding to the predicted SAM-binding loop and the other located directly C-terminal of it, with the *For correspondence. Email. aravind@ncbi.nlm.nih.gov J. Mol. Microbiol. Biotechnol. (2002) 4(1): 71–75. JMMB Communication