Substrate Recognition of tRNA (Guanosine-2′-)-methyltransferase from Thermus thermophilus HB27*

Transfer RNA (guanosine-2′-)-methyltransferase (Gm-methylase, EC 2.1.1.32) from Thermus thermophilus HB27 is one of the tRNA ribose modification enzymes. The broad substrate specificity of Gm-methylase has so far been elucidated using various species of tRNAs from native sources, suggesting that the common structures in tRNAs are recognized by the enzyme. In this study, by using 28 yeast tRNAPhe variants obtained by transcription with T7 RNA polymerase, it was revealed that the nucleotide residues G18 and G19 and the D-stem structure are essentially required for Gm-methylase recognition, and that the key sequence for the substrate is pyrimidine (Py)17G18G19. The other conserved sequences were found not to be essential, but U8, G15, G26, G46, U54, U55, and C56 considerably affected the methylation efficiency. These residues are located within a limited space embedded in the L-shaped three-dimensional structure of tRNA. Therefore, disruption of the three-dimensional structure of the substrate tRNA is necessary for the catalytic center of Gm-methylase to be able to access the target site in the tRNA, suggesting that the interaction of Gm-methylase with tRNA consists of multiple steps. This postulation was confirmed by inhibition experiments using nonsubstrate tRNA variants which functioned as competitive inhibitors against usual substrate tRNAs.

To date, more than 80 modified nucleosides in tRNAs have been isolated and characterized (1). These nucleosides are posttranscriptionally formed at specific positions of tRNA by specific tRNA modification enzymes and are presumed to play important roles in the structure and function of tRNA (2)(3)(4)(5)(6)(7).
Among tRNA modification enzymes, tRNA (guanosine-2Ј-)methyltransferase (Gm-methylase, 1 EC 2.1.1.34), one of the ribose modification enzymes, specifically catalyzes the transfer of a methyl group from S-adenosyl-L-methionine to the 2Ј-OH of the ribose ring of guanosine at position 18 (G18) in the D-loop (8,14). G18 is one of the hyperconserved residues located in the so-called three-dimensional core of tRNA (9,10) and is responsible for the formation of the L-shaped three-dimensional structure by D-loop/T-loop interaction through the tertiary base pair G18-⌿55 and G19-C56 (11,12). Although 2Ј-O-methylguanosine at position 18 (Gm18) is distributed widely in tRNAs of prokaryotes, eukaryotes, archaea, and plant mitochondria (13), purification of the enzyme has been reported solely from Thermus thermophilus (7). Recently, the Escherichia coli spoU gene has been reported to be essential for Gm18 modification, suggesting that spoU encodes E. coli Gm-methylase (14). With respect to the physiological role of the methylation of the ribose of the G18 residue, it is known that the resistance of tRNA against RNases is increased by this modification, thus probably prolonging the half-life of the tRNA (15). In higher plants, a relationship between Gm18 methylation and the transport of tRNA Leu into mitochondria has also been reported (16).
The mechanisms of interactions between tRNAs and modification enzymes are of interest not only physiologically but also biochemically as typical examples of RNA-protein interaction. However, there are only a few reports on tRNA recognition by modification enzymes. The following purified enzymes have been studied: tRNA-(m 1 G37)-methyltransferase (17), tRNAguanine transglycosylase for producing Q34 (18,19), and tRNA-(m 5 U54)-methyltransferase (20,21) from E. coli; tRNA-(m 5 C48)-methyltransferase (22) from HeLa cell line; and Gm18-methylase (23)(24)(25) and tRNA-(m 1 A58)-methyltransferase (26) from T. thermophilus. A crude extract, tRNA-(⌿35)synthase from a higher plant (27), has also been investigated. In addition, several tRNA modification enzymes from Xenopus laevis (28 -33) and yeast (34) have been reported using in vivo assay systems. Two technical difficulties have hindered investigations using purified enzymes. First, most tRNA modification enzymes are labile and only very scanty amounts are able to be purified. Second, special RNAs are usually required as substrates because the enzymes are highly specific for a particular nucleoside(s), sequence(s) and/or three-dimensional structure, and such tRNAs are not easy to prepare. Fortunately, the modification enzymes from T. thermophilus, one of which we used in this work, are relatively stable compared with those from other species. To overcome the second problem, we employed a T7 RNA polymerase system. A synthetic gene of yeast tRNA Phe was chosen as the template DNA of T7 RNA polymerase, because yeast tRNA Phe is one of the best substrate tRNAs for Gm-methylase, and its three-dimensional structure is well established (11,12). In this report, the essential regions in the tRNA for recognition by Gm-methylase and its recognition mechanism are discussed.

EXPERIMENTAL PROCEDURES
Materials-The methyl-14 C-labeled S-adenosyl-L-methionine (55-60 Ci/mol) was purchased from Amersham Pharmacia Biotech. DNA oligomers were synthesized by an Applied Biosystems model 381 DNA synthesizer. DNA modifying enzymes and human placenta RNase inhibitor were obtained from Takara (Ohtsu, Japan). A T7 RNA polymerase expression system (E. coli BL21/pAR1219) was kindly provided by Dr. F. W. Studier (Brookhaven National Laboratory) (35). T7 RNA polymerase was purified by the method of Grodberg and Dunn (36). Other chemical reagents were of analytical grade.
Transfer RNAs from Native Sources-Purified E. colitRNA 2 Glu , tRNA 3 Ser , and tRNA 2 Tyr were kindly provided by Dr. N. Hayashi (Tokyo Institute of Technology); Hallobacterium volcanii tRNA i Met tRNA was a gift of Dr. Y. Kuchino (National Cancer Research); Bacillus subtilis tRNA Gly was supplied by Dr. K. Murao (Jichi Medical School). Yeast tRNA Phe was purchased from Boehringer Mannheim.
Preparation of Yeast tRNA Phe Wild-type Transcript and Its Variants-A synthetic wild-type yeast tRNA Phe gene with a T7 promoter were constructed between the EcoRI and BamHI sites in the multicloning linker of pUC18, and the insert was then subcloned into the SalI and HindIII sites in the multicloning linker of pUC118 for site-directed mutagenesis. In the resultant plasmid, the transcriptional initiation site was designed to be G at the 5Ј terminal, the first position of yeast tRNA Phe . Yeast tRNA Phe gene variants were produced by site-directed mutagenesis using a Muta-Gene phagemid mutation kit (Bio-Rad). The sequences of all the tRNA genes were analyzed using a Sequenase version 2.0 DNA sequencing kit (U.S. Biochemical Corp.). Read-through transcription was carried out with 18 g of T7 RNA polymerase at 37°C for 3 h using 20 g of BstNI-digested plasmids encoding the yeast tRNA Phe gene variants as the template in a buffer containing 40 mM Tris-HCl (pH 8.0), 10 mM MgCl 2 , 5 mM dithiothreitol, 2 mM ATP, 2 mM GTP, 2 mM CTP, 2 mM UTP, 20 mM GMP, 1 mM spermidine, 5 g of bovine serum albumin, 50 units of human placenta RNase inhibitor, and 3% glycerol in a total volume of 100 l. The reaction mixture was extracted with phenol-chloroform (1:1, w/w) and then with chloroform-isoamylalcohol(25:1, v/v). Transcripts were recovered from the aqueous phase by ethanol precipitation. The dried pellet was dissolved in 50 l of buffer containing 10 mM Tris-HCl (pH 8.0) and 1 mM EDTA, and the transcript was purified by 10% polyacrylamide (7 M urea) gel electrophoresis.
Measurement of Melting Temperature--The melting profiles of yeast tRNA Phe and the wild-type transcript were measured by monitoring the change in the absorbance at 260 nm at a heating rate of 0.5°C/min with a Gilford Response II spectrophotometer using 0.32 A 260 unit RNA in 400 l of buffer containing 50 mM Tris-HCl (pH 7.5), 5 mM MgCl 2 , and 100 mM NaCl. The melting temperatures were determined by the first derivative of the melting curve.
Purification and Assay of Gm-methylase-Gm-methylase was purified by the method previously reported (8). The homogeneity of the purified enzyme was confirmed by SDS-polyacrylamide gel electrophoresis (37). The quantity of the protein was measured with a Bio-Rad protein assay kit using bovine serum albumin as the standard. The standard assay for Gm-methylase activity was used according to our previous report (8), except that the assays with transcripts were carried out at 60°C. The apparent kinetic parameters, K m and V max , were determined by Lineweaver-Burk plots of the methylation reaction in which incorporations of the 14 C methyl group into the transcripts was measured for 20 min.

RESULTS AND DISCUSSION
Transfer RNAs from Native Sources-In tRNA of the extreme thermophile T. thermophilus, Gm18 is one of the generally existing modified nucleosides, and is produced by Gmmethylase (8,38), which catalyzes the transfer of methyl groups to various kinds of tRNAs in vitro as well as in vivo. The tRNAs that have so far been identified as substrates of Gmmethylase in an in vitro methylation reaction are listed in Table I. The only tRNA in the table that cannot be methylated is E. coli tRNA 2 Tyr , because it already contains the Gm18 residue in the E. coli cells. This indicates that Gm-methylase does not catalyze the exchange reaction of the methyl group, which is in line with the reaction mechanisms of E. coli tRNA-(m 5 U54)-methyltransferase (39) and T. thermophilus tRNA-(m 1 A58)-methyltransferase (26). Since not only Class I but also Class II tRNAs (E. coli tRNA 3 Ser and B. subtilis tRNA Leu ) are methylated by Gm-methylase, the structural diversities derived from the sizes of the D-arm and the variable arm do not affect recognition by the enzyme. Moreover, the tRNA from an archaean, H. volcanii tRNA i Met , was a good substrate for Gmmethylase, suggesting that the recognition sites of Gm-methylase are common for tRNAs from three kingdoms, eukarya, prokarya, and archaea. This work a -, not determined. b "Not methylated" means that the relative initial velocity was below 0.5%.

Variants Substituted in Conserved Sequence and D-loop-
The nucleotide residues conserved and semiconserved in the tRNAs of the three kingdoms are shown in Fig. 1A. To clarify the recognition sites of Gm-methylase, 28 variants of yeast tRNA Phe transcribed by T7 RNA polymerase were employed. Nucleotide substitutions were mainly introduced into the conserved residues in the three-dimensional core region in the tRNA (Fig. 1B), because these residues interact directly or indirectly with the D-arm, which includes the methylation target site, the 2Ј-OH group of the G18 ribose. Previous results from foot printing and experiments using half fragments indicated that the essential region for the recognition by Gmmethylase was limited within the sequence G10 -G26 in E. coli tRNA f Met (24,25). Therefore, all the residues of the D-loop were individually substituted by other nucleotides, irrespective of whether they were conserved or nonconserved (Fig. 1B), by introducing point mutations into the synthetic yeast tRNA Phe gene. Each variant is designated by the original and mutated residues connected by arrows in Fig. 1B Table II. Fig. 2A shows the relationship between substrate concentration and the initial velocity of methyl group incorporation by Gm-methylase when the wild-type tRNA Phe transcript and some tRNA variants (U17A, U17G, and G18A) were used as the substrate. Fig. 2B shows the Lineweaver-Burke plots for wild-type transcript (left) and for U17G and U17A variants, from which the kinetic parameters (K m and V max ) were estimated as shown in Table II.
The methyl group acceptance activity of the wild-type transcript was about 1/20 that of the native yeast tRNA Phe through increase of K m (Tables I and II and Fig. 2B). The structural flexibility of a transcript can be estimated by its melting curve; the melting temperature of the wild-type transcript was determined to be 69.0°C, while that of the native yeast tRNA Phe was 76.0°C. Some of the modified nucleosides in the native tRNA affects the methylation efficiency, probably through the structural stability of the substrate RNA and/or the direct association with the enzyme. Since a suitable temperature for the methylation of the transcript was estimated to be 60°C (data not shown), all the experiments with transcripts were carried out at that temperature.
The difference in methyl group acceptance between native tRNA and its transcripts is more clearly observed with shorter length transcripts corresponding to a part of yeast tRNA Phe ; methyl group acceptance activity of the transcript corresponding to the 5Ј-half fragment (positions 1-33 in yeast tRNA Phe ) was hardly detectable at 37-60°C (data not shown), while that of the native 5Ј-half fragment could be detected, the initial velocity being about 20% of that of the full-length tRNA (24,25). In the case of the thermophile Gm-methylase, it was difficult to discern the so-called minimalist substrate by using a totally unmodified short fragment, since the initial velocities for such fragments could not be measured under the standard conditions; at least, the methylation of the chemically synthesized 18-mer corresponding to positions 9 -26 was not able to be detected at either 37 or 50°C in 24-h incubation under the standard conditions (data not shown). Thus, it is assumed that modified nucleosides such as m 2 G10, D16, D17, m 2 2 G26, and Cm32 present in the native 5Ј-half fragment strongly affect the methylation efficiency, probably through stabilizing the D-loop stem structure or making the enzyme recognition toward the substrate easier.
In the point-substituted full-length variants, the only residues essential for Gm-methylase recognition among those conserved or semiconserved were determined to be G18 and G19 ( Fig. 2A and Table II). Substitution of the nonconserved residue U17 by purine (Pu) (U17A and U17G) resulted in a drastic decrease in the V max /K m value. In contrast, no effect was observed when U17 was substituted by C. Analysis of the kinetic parameters indicated that the V max values for U17A and U17G were very small ( Fig. 2 and Table II), suggesting that the substitution of U17 by Pu changes the environment of the catalytic center in the Gm-methylase -tRNA complex. Thus, the most appropriate minimal sequence for Gm-methylase was deduced to be Py17G18G19 (Py ϭ pyrimidine). This is supported by the result for E. coli tRNA 3 Ser possessing an A17G18G19 sequence (Table I), which was the worst substrate for Gm-methylase among the native tRNAs tested.
In E. coli tRNAs, the G18 residue in almost all class II tRNAs is modified to Gm18, an exception being tRNA 3 Ser , which has an unmodified G18 (13). Moreover, no prokaryote tRNA possessing a Pu17Gm18 sequence has been reported (13). Thus, the optimum sequence deduced for the thermophile Gm-methylase, Py17G18G19, is likely to be applicable to most prokaryotic Gm-methylases. It is also clear that positions 17-19 were distinctly recognized by the enzyme, because the G18C variant with a C18G19G20 sequence was not methylated at all (Table  II). Judging from the results with the D-stem variants (discussed in the next section; see Table III), it is likely that the recognition of these positions depends on the steric distance and the angle from the phosphate-ribose backbone of the Dstem structure. On the other hand, it has been reported that the 2Ј-O-methylation of G34 (the anticodon first letter) in X. laevis is not affected by the nucleotide sequence around position 34 (30). The recognition mechanism of Gm34-methylase in eukaryotes is thus apparently different from that of Gm18methylase in prokaryotes, although the same modified nucleoside is produced.
The tertiary base pairs connecting the D-and T-loops were FIG. 1. A, conserved and semiconserved residues in eukaryotic, prokaryotic, and archaeae tRNAs. Upper-and lowercase letters represent conserved and semiconserved residues, respectively. Symbols: q, nonconserved residue; *, usually modified nucleoside; R/r, purine; Y/y, pyrimidine. This figure is based on Steinberg et al. (13) and Grosjean et al. (43). B, point mutations introduced into wild-type transcript of tRNA Phe . The mutations were individually introduced into the residues as indicated by arrows. The substituted nucleotides are indicated by the arrowheads.
disrupted in the variants U55A and C56G. As shown in Table  II, the K m values for both variants were much larger than that of wild-type transcript. This is in good agreement with our previous experimental results using a half fragment; the 5Јhalf fragment of yeast tRNA Phe showed methyl group acceptance activity of about 1 ⁄5 in terms of the initial velocity, but the K m value was much larger, as compared with the values for the full-length tRNA Phe (24,25). These data suggest that the formation of the tertiary base pair G18-U55 and G19-C56 enhances the methylation efficiency, although these base pairs or the existence of the U55 and C56 residues by themselves are not essential for methylation.
We also demonstrated in our previous study that chemical modification of the s 4 U8 residue in E. coli tRNA f Met decreases the efficiency of methylation by Gm-methylase (24). The variants U8A, U8C, and U8G, in which the U8 residue is substituted by A, C, and G, respectively, lack the original tertiary base pair U8-A14, although an alternative reverse Hoogsteen pairing C8-A14 may be recovered in the U8C variant. As shown in Table II, the methylation efficiency of these variants was reduced due to increased K m values, which is similar to what happened with the chemically modified E. coli tRNA f Met . Con-

TABLE II Effect of point mutations in tRNA Phe transcripts on their methyl group acceptance activity
The wild type is totally unmodified tRNA Phe transcribed by T7 RNA polymerase. The designation given to each variant indicates the position of the point mutation in the tRNA and the original and mutated residues. For example, U8A means the variant in which U at position 8 was substituted by A. Symbols: no, nonconserved; none, the residue is not responsible for the tertiary base pair; ND, methylation was not detected (the initial velocity was below 0.5% of that of the wild-type transcript). When an alternative tertiary base pair was created by point mutation, the disrupted and alternative base pairs are connected by 3. For example, G18-U55 3 A18-U55 means the base pair G18-U55 was disrupted to form A18-U55. The relative Vmax: Km is expressed in relation to that of the wild type which is taken as 100%. The kinetic parameters indicated in this table are the averages of data obtained in at least two independent experiments. sidering the result with the variant A14C, in which the same U8-A14 base pair was disrupted, it appears that disruption of the U8-A14 base pair slightly affected the methylation efficiency, but the substitution of U8 gave rise to a pronounced decrease in affinity with Gm-methylase. These findings suggest that the decrease in methylation efficiency following chemical modification observed in the previous study (24) arose from steric hindrance due to the s 4 U8 adduct resulting from the chemical modification.
Other tertiary base pairs, G15-C48, U54-A58, G26-A44, and G21-G46, were also disrupted individually by nucleotide substitutions (G15U, C48A, U54A, A58G, G26U, G26A, and G46C; see Fig. 1B and Table II). Although these tertiary base pairs are not essential, the variants G15U, G26U, G46C, and U54A clearly showed reduced methylation efficiency. In G15U this was due to a decrease in V max , in G46C and U54A to an increase in K m , and in G26U to both a decrease in V max and an increase in K m .
The tertiary Levitt base pair G15-C48 did not affect the methylation efficiency, since the variant C48A in which this tertiary base pair was disrupted, showed no marked reduction in methylation efficiency. Therefore, the G15 residue itself and/or the D-loop structure influenced by G15 probably affected the environment around the catalytic center of Gm-methylase.
The mechanism in the case of the variant G46C is less clear. In the native tRNA, G46 is usually modified to m 7 G46 and the tertiary base pair G21-m 7 G46 is formed; the variable loop region is located nearby the D-arm in the three-dimensional structure of the tRNA. Since a relatively large K m value was also observed when the neighboring nonconserved G45 residue was substituted by C, it is considered that the G45-G46 region in the variable loop may be located near the enzyme surface, thus affecting the affinity of the enzyme.
The G26 residue is conserved as purine and in yeast tRNA Phe its modified to m 2 2 G26 resulting in formation of the tertiary base pair m 2 2 G26-A44 (11,12). When G26 was substituted by A, no appreciable effect was observed. In contrast, substitution of G26 by U reduced the methylation efficiency by 70%. In the variant G26A, the tertiary base pair G26-A44 was disrupted and an alternative base pair A26-A44 was probably formed. Similarly, in the variant G26U, the tertiary base pair G26-A44 was disrupted and an alternative Watson-Crick base pair U26-A44 was formed. This Watson-Crick base pair newly formed in the variant G26U probably affected not only the connection between the D-stem and the anticoden-stem, but also the local structure of the variable loop. Since, as described above, the G45-G46 region in the variable loop has an influence on the methylation efficiency, the substitution of the G26 residue by U may affect the methylation efficiency through the newly formed U26-A44 base pair.
The variant U54A showed decreased methylation efficiency (to approximately half of that of the wild type in initial velocity) through an increase of the K m value. Since the U54 residue is usually modified to m 5 U54 (13) and affects the stability of the L-shaped tRNA structure (40), the effect of substituting U54 by A might arise from structural destabilization. The decrease in the methylation efficiency of U54A was smaller than that of the variant U55A and C56G (Table II).
The D-stem Variants-The methylation activity of the Dstem variants (Fig. 3) was also tested, because our previous foot printing data demonstrated that Gm-methylase was bound to the D-arm region in the enzyme-yeast tRNA Phe complex (25). The results are shown in Table III. Since there is no conserved sequence in the D-stem region of all tRNAs (Fig. 1A), the D-stem variants II and III were designed so as to form an artificial stem structure as shown in Fig. 3. In the D-stem variant II, the base pair Py11-Pu24 conserved in all elongator tRNAs was substituted by an alternative base pair A11-U24, and the tertiary base pair A9-A23 was disrupted; in the D-stem variant III, the semiconserved base pairs (Pu10-Py25 is conserved in almost all tRNAs and Py11-Pu24 is conserved in elongator tRNAs) were substituted by alternative base pairs, and the tertiary base pairs (G22-G46 and A9-A23) were disrupted.
As shown in Table III, the D-stem variant I lacking the D-stem structure was not methylated at all, but the D-stem variants II and III were methylated at about 50% of the methylation of the wild-type transcript. These results clearly show that the structure of the D-stem is one of the essential factors for tRNA recognition by Gm-methylase. Further, it is suggested that the phosphate-ribose backbone of the D-stem structure was recognized by the enzyme, because the nucleotide sequence of the D-stem seemed to have nothing to do with the methylation activity. Moreover, the effect of the disruption of the tertiary base pairs (G22-G46 and A9-A23) on the methyla-

TABLE III Kinetic parameters of D-stem variants
The nucleotide sequences of the variants are shown in Fig. 2. ND means that methylation was not detected (the initial velocity was below 0.5% of the wild-type transcript). The relative V max /K m is expressed in relation to that of the wild-type which is taken as 100%. The results in this tion efficiency was smaller than that of the substitution of the G46 residue by C (see the variant G46C in Table II).

Multistep Recognition Mechanism Inferred from Inhibition Experiments-
The regions essential for recognition by Gmmethylase so far elucidated are depicted on the L-shaped tRNA structure shown in Fig. 4, in which the most significant residues are circled. These features are in agreement with our previous foot printing data (25). However, since the residues involved in substrate recognition by Gm-methylase appear to be embedded inside the L-shaped tRNA molecule, disruption of the tertiary structure of the tRNA may be necessary for Gmmethylase to gain access to these residues. This is supported by the previous foot printing data, which showed that the sensitivity of the aminoacyl stem and T-loop toward RNases is enhanced in the enzyme-tRNA complex (25). This assumption suggests the existence of at least two steps in the interaction of substrate tRNA with Gm-methylase; the first step would involve association of the tRNA with Gm-methylase, while in the second step the structural change of the tRNA, involving interaction between the enzyme and the essential regions of the tRNA, would take place. Such a kind of multistep reaction mechanism has recently been proposed for some cases of DNAprotein interaction (41).
To clarify the above postulation, inhibition experiments were carried out using two variants different in type but both lacking methyl group acceptance activity (Fig. 5A). In the first variant, G18A, the tertiary base pairs connecting the T-loop with the D-loop were not likely to be disrupted because an alternative A18-U55 and the conserved G19-C56 base pairs were probably formed. The other variant, named ⌬D-arm, lacked the entire D-arm structure. Surprisingly, as shown in Fig. 5B, both variants inhibited methyl group incorporation of the wild-type transcript. The K i of G18A indicates that this variant possessed comparable affinity to that of the variant C56G. This finding is in line with an earlier report on tRNA methylation enzymes such as m5C(48)-and m7G(46)-methylases in rat liver (42). The other variant used in the inhibition experiments, ⌬D-arm, also showed unexpected inhibition toward the methylation reaction despite the deletion of all the regions in the tRNA essential for methylation. These results clearly show that Gm-methylase has the potential to form a complex with all tRNAs in the first binding step, irrespective of whether they are substrates for Gm-methylase or not. The L-shaped tRNA structure is probably desirable but not essen- FIG. 4. Essential regions for recognition by Gm-methylase in the L-shaped tRNA structure. The residues are depicted in the same manner as in Fig. 1A. The essential regions (G18, G19, and the D-stem) are circled by a boldface line. Residues whose substitution decreased the initial velocity of methylation by more than a half of that of the wild-type transcript are circled in thin lines. Dotted lines denote expected tertiary base pairs in the wild-type transcript. tial in the first step, since the tertiary base pair G18-U55 and G19-C56 enhance the methylation efficiency and the 5Ј-half fragment of the native yeast tRNA Phe has methyl group acceptance activity. Thus, it seems that the regions essential for the methylation are not required in the first step. As the methylation reaction proceeds, however, these essential regions probably become necessary. The residues U8, G15, Py17, and G46 might also be required in the latter step, because the tertiary base pairs formed by these residues did not affect the methylation efficiencies, and the substitution of the G15 and the Py17 decreased the V max . The low efficiencies observed for all the transcripts suggest the importance of the modified nucleoside(s) for the methylation reaction. Judging from the kinetic parameters, it is likely that most of the modified nucleosides are primarily involved in the first step, because both native tRNA Phe and its wild-type transcript have almost the same K m , but quite different V max values (Tables I and II). The same tendency is also observed for some variants which lose a certain modified nucleoside in base substitution (Table II). Further study is necessary for clarifying which step requires each of the modified nucleosides.
In this study, we have demonstrated the broad substrate specificity of Gm-methylase and clarified the regions essential for substrate recognition by the enzyme. Recently, Grosjean et al. (33) proposed an important framework for tRNA modification enzymes, in which they classified modification enzymes into two major groups depending on the requirement of the three-dimensional structure of the substrate tRNA and the position of the target site. In terms of this framework, Gmmethylase would be classified as a "group 1" enzyme because the target site, G18, is included in the three-dimensional core of tRNA. The results for Gm-methylase in our previous (24,25) and current studies are in good agreement with the properties of group 1 enzymes in the framework, because the whole tRNA structure is not necessary for methylation to occur, but a special sequence in the limited structure is essential. Since the enzymes classified into group 1 probably require disruption of tRNA three-dimensional structure (33), the multistep recognition mechanism revealed in this work is likely to be applicable to many of the group 1 enzymes.