Once there were twenty.

Aminoacyl-tRNA synthetases play a central role in protein synthesis by covalently linking the correct amino acid to the correct tRNA (1). Each aminoacyl-tRNA is then carried to the ribosome where it interacts with the cognate trinucleotide codon on the mRNA and transfers the amino acid onto a growing polypeptide chain. Work on tRNAs and aminoacyltRNA synthetases from bacteria, fungi, plants, and mammals led to the general notion of the occurrence of 20 aminoacyltRNA synthetases in all organisms, one for each of the 20 amino acids. Early indication of an exception came from the finding that three Gram-positive bacteria lacked the enzyme glutaminyl-tRNA synthetase (GlnRS),† which covalently links glutamine to the glutamine tRNA (tRNAGln) (2). Instead the glutamine tRNA is first aminoacylated with glutamic acid using glutamyl-tRNA synthetase (GluRS) to form Glu-tRNAGln. In a second step, the Glu-tRNAGln is converted to Gln-tRNAGln by an amidotransferase now called Glu-AdT (Fig. 1). The latter reaction requires ATP, which activates the side chain carboxyl group of glutamic acid on the tRNA to form a carboxyphosphate anhydride and glutamine or asparagine, which provides the amino group for transamidation (3). Thus, depending on the organism, there are two pathways for the synthesis of Gln-tRNAGln, a direct pathway involving GlnRS and an indirect pathway involving GluRS and Glu-AdT (Fig. 1). Further work showed that also in archaebacteria (4, 5), cyanobacteria, mitochondria, and chloroplasts (6, 7), Gln-tRNAGln is formed not directly but in a two-step process as outlined above. In archaebacteria, such as halobacteria, Asn-tRNAAsn is also synthesized in an analogous two-step process involving Asp-tRNAAsn as an intermediate (8). The widespread utilization of the Glu-AdT pathway for incorporation of glutamine into proteins and its possible similarity to other glutamine dependent amidotransferase reactions involved in a variety of biochemical reactions (9) has generated much interest in the nature of enzymes involved in this process. The paper by Curnow et al. (10) in this issue of Proceedings describes the cloning, sequence analysis, and molecular characterization of the Glu-AdT gene from Bacillus subtilis. The enzyme is shown to have three subunits, A, B, and C. Homologues to these subunits have been identified in other organisms. The A subunit may have a P-loop-type ATP binding motif, suggesting that this subunit might be involved in activation of the side chain carboxyl group of glutamic acid. The A subunit also has substantial homology to a class of amidases and can convert glutamine to glutamic acid and ammonia in vitro. This reaction provides enzyme bound ammonia as a source of amino group for the amidation of glutamic acid on the tRNA to glutamine. The B subunit might be used to select the correct tRNA substrate. Curnow et al. (10) have identified two different types of B subunits in some archaebacteria, one more closely related to the B. subtilis B subunit than the other. They suggest that one is used for transamidation of GlutRNAGln whereas the other is used for transamidation of Asp-tRNAAsn. The role of the C subunit is not clear. It may not be present in all other organisms that use the Glu-AdT pathway or it could have diverged enough not to be recognized as a homologue. The C subunit is necessary for the expression of the B. subtilis A subunit in Escherichia coli suggesting that it could be involved in the modification, folding andyor stabilization of the A subunit. The genes for the three subunits are organized into an operon, which is transcribed to produce a polycistronic mRNA. The polycistronic nature of the transcript and the close proximity of the stop codons of the preceding ORFs to the start codons of the downstream ORFs raises the possibility of coupled translation to produce stoichiometric amounts of the three subunits and possibility of translational regulation (11). The gene for Glu-AdT is shown to be essential in B. subtilis. Thus, this enzyme provides the only pathway for the formation of Gln-tRNAGln in this organism. The Glu-AdT belongs to an important class of enzymes that modify amino acids while they are linked to tRNAs. Examples of these are enzymes that convert methionine to formyl methionine (used for initiation of protein synthesis in eubacteria and in mitochondria and chloroplasts; reviewed in ref. 12), serine to selenocysteine (used for insertion of selenocysteine at very specific sites in proteins; reviewed in refs. 13 and 14), and glutamic acid to glutamic a-semialdehyde (used for synthesis of d-amino levulinic acid on the pathway to chlorophyll biosynthesis in plants, archaebacteria and eubacteria; ref. 15). These enzymes are specific for the amino acid and for the tRNA to which the amino acid is attached (16–18). Like these other enzymes, the Glu-AdT must also act specifically on only one tRNA, Glu-tRNAGln, the glutamine tRNA that is aminoacylated with glutamic acid. The successful reconstitution of the Glu-AdT subunits expressed in E. coli described in this paper and the availability, thereby, of the purified enzyme should now expedite studies on the mechanism of transamidation and the molecular basis of tRNA specificity. A surprising aspect of the B. subtilis Glu-AdT gene is the lack of any sequence homology to aminoacyl-tRNA synthetases (10). This implies that these two enzymes had quite different origins. It is, therefore, intriguing that the B. subtilis tRNAGln is a substrate for E. coli GlnRS (10). This implies that most of the determinants necessary for recognition by the E. coli GlnRS (19) are present on the B. subtilis tRNAGln. An inspection of the tRNA sequence indicates that this is indeed the case. Because B. subtilis does not have GlnRS, why does the tRNAGln contain determinants for a bacterial GlnRS? One possibility is that these determinants, particularly in the acceptor stem of the tRNA, are needed for aminoacylation of the tRNA by the B. subtilis GluRS (Fig. 1). This is probably not the