Functional Categorization of the Conserved Basic Amino Acid Residues in TrmH (tRNA (Gm18) Methyltansferase) Enzymes*

Transfer RNA (Gm18) methyltransferase (TrmH) catalyzes the methyl transfer from S-adenosyl-l-methionine (AdoMet) to the 2′-OH group of the G18 ribose in tRNA. To identify amino acid residues responsible for the tRNA recognition, we have carried out the alanine substitution mutagenesis of the basic amino acid residues that are conserved only in TrmH enzymes and not in the other SpoU proteins. We analyzed the mutant proteins by S-adenosyl-l-homocysteine affinity column chromatography, gel mobility shift assay, and kinetic assay of the methyl transfer reaction. Based on these biochemical studies and the crystal structure of TrmH, we found that the conserved residues can be categorized according to their role (i) in the catalytic center (Arg-41), (ii) in the initial site of tRNA binding (Lys-90, Arg-166, Arg-168, and Arg-176), (iii) in the tRNA binding site required for continuation the catalytic cycle (Arg-8, Arg-19, and Lys-32), (iv) in the structural element involved in release of S-adenosyl-l-homocysteine (Arg-11—His-71—Met-147 interaction), (v) in the assisted phosphate binding site (His-34), or (vi) in an unknown function (Arg-109). Furthermore, the difference between the Kd and Km values for tRNA suggests that the affinity for tRNA is enhanced in the presence of AdoMet. To confirm this idea, we carried out the kinetic studies, a gel mobility shift assay with a mutant protein disrupted in the catalytic center, and the analytical gel-filtration chromatography. Our experimental results clearly show that the enzyme has a semi-ordered sequential mechanism in which AdoMet both enhances the affinity for tRNA and induces formation of the tetramer structure.

Transfer RNA (Gm18) methyltransferase (TrmH (classical name, SpoU); tRNA (guanosine-2Ј-)-methyltransferase, EC 2.1.1.34) catalyzes the methyl transfer from S-adenosyl-L-me-thionine (AdoMet) 3 to the 2Ј-OH group of the ribose of guanosine at position 18 (G18) in tRNA (1)(2)(3)(4)(5)(6)(7)(8)(9)(10)(11). In the reaction, AdoMet is converted to S-adenosyl-L-homocysteine (AdoHcy). G18 is a highly conserved residue in the D-loop of tRNA (12,13) and is responsible for the formation of the L-shaped three-dimensional structure of the tRNA by the D-loop/T-loop interaction mediated through the G18-⌿55 tertiary base pair (14,15). Because the 2Ј-O-methylation of ribose at position 34 stabilizes the C3Ј-endo form (16), the 2Ј-O-methyl-G18 (Gm18) modification probably imposes conformational rigidity on the local structure of tRNA. Although the absence of the Gm18 modification in Escherichia coli tRNA has been shown to have no effect on either the activity of the supF amber suppressor tRNA or on the growth rate of cells (9), it has been reported that the growth rate is decreased and that the frequency of the frameshift errors is increased in an E. coli mutant lacking both Gm18 and ⌿55 (17). Thus, the Gm18 modification probably works in conjunction with the other modified nucleosides in the three-dimensional core such as ⌿55. The genes responsible for the Gm18 modification have been experimentally identified as trmH (spoU) in E. coli (9,10), Thermus thermophilus (5), and Aquifex aeolicus (6) and as trm3 in Saccharomyces cerevisiae (11). The Gm18 modification in tRNA is found in eubacteria and eukaryotes, consistent with the distribution of the trmH and trm3 genes (12,13). The Gm18 modification is also found in plant organelle tRNA, although the responsible gene has not been identified (12,13). Furthermore, it has been recently proposed that Sulfolobus solfataricus tRNA Gln undergoes Gm18 modification by the C/D box small RNA (sRNA) guide methylation system (18).
T. thermophilus TrmH has many merits as a model protein for structural and functional studies of the SpoU family, because it is a small, stable, and well characterized member of the family (1)(2)(3)(4)(5). In recent studies we have determined the crystal structure of the TrmH (7), clarified the residues involved in the AdoMet-binding (7,8), proposed a catalytic mechanism (7,8), and identified the roles of the conserved residues in the motifs (8). However, a very important question remains outstanding. The amino acid residues responsible for the contact with the substrate tRNA have not been identified. In the current study, therefore, we have focused on the basic amino acid residues that are conserved only in the TrmH enzymes and not in the other SpoU family proteins to clarify the tRNA binding site. Beyond our expectations, the initial site of tRNA recognition and the subsequent tRNA binding site required for the catalytic cycle to proceed could be distinguished by biochemical experiments. Furthermore, the basic amino acid residues that are involved not only in tRNA binding but also in AdoMet binding (AdoHcy releasing) have been identified. Together with the information from the crystal structures of the TrmH alone and the TrmH-AdoMet complex, the recognition sites of tRNA in TrmH are discussed in comparison to other SpoU family members.

EXPERIMENTAL PROCEDURES
Materials-[Methyl-14 C]AdoMet (1.95 GBq/mmol) and [methyl-3 H]AdoMet (2.47 TBq/mmol) were purchased from ICN, cold AdoMet, and AdoHcy from Sigma, and DE52 was from Whatman. The AdoHcy affinity column was synthesized from a HiTrap NHS-activated HP column (Amersham Biosciences) according to the manufacturer's manual. CM-Toyopearl 650M was from Tosoh. DNA oligomers were bought from Invitrogen, and T7 RNA polymerase was from Toyobo. All other chemical reagents were of analytical grade.
Site-directed Mutagenesis and Purification of the Proteins-The tRNA (Gm18) methyltransferase expression vector (pET30aGm) has been previously reported (5). We carried out the site-directed mutagenesis by using a QuikChange mutagenesis kit (Stratagene). The DNA sequences were analyzed with ABI PRISM 310 DNA sequencers.
The wild-type enzyme was expressed in E. coli and purified as reported previously (5). Ten mutant proteins (R8A, R19A, K32A, H34A, H71A, K90A, R109A, R168A, R176A, and L179Stop) were purified by the same method. Two mutant proteins (R11A and R166A) were purified by repeated ion-exchange chromatography, because AdoHcy affinity column chromatography did not work efficiently on the purification of them. In brief, fractions collected from the DE52 column chromatography step in the standard purification procedure were combined, dialyzed against buffer A (50 mM HEPES-NaOH (pH 6.8), 5 mM MgCl 2 , and 6 mM 2-mercaptoethanol), and separated by CM-Toyopearl 650M column chromatography (liner gradient; 0 -250 mM KCl in the buffer A). The mutant protein fractions were assessed by 15% SDS-PAGE. If the purity of the protein was not satisfactory, we repeated the CM-Toyopearl 650M column chromatography after dialysis against the buffer A. The fractions were combined, dialyzed against buffer B (50 mM Tris-HCl (pH 7.5), 5 mM MgCl 2 , 2 mM mercaptoethanol, and 50 mM KCl), and concentrated with Centriprep YM-10 centrifugal filters (Millipore). The quantity of protein was measured with a Bio-Rad protein assay kit using bovine serum albumin as the standard. The purified proteins were mixed with glycerol (final concentration 50%) and stored at Ϫ30°C.
Measurements of the Enzymatic Activity-A standard assay for enzyme purification was performed according to a previously reported method (4,5). To determine the kinetic parameters for AdoMet, the concentrations of each mutant protein and the yeast tRNA Phe transcript were fixed at 0.1 and 8.5 M, respectively. However, the range of the incubation time was varied (2-30 min) according to the methyl transfer activity of the protein. In a typical assay, [methyl-3 H]AdoMet concentrations of 0, 0.5, 1, 2.5, 5, 10, 15, 20, 40, 60, and 80 M were used. The kinetic parameters for tRNA were determined as previously reported (4). In experiments to assess the enzyme mechanism (see Fig. 4 Analytical Ado-Hcy Affinity Column Chromatography-The affinity of the purified proteins for AdoHcy was qualitatively analyzed by the Ado-Hcy column chromatography as described in our previous report (7,8).
Gel Mobility Shift Assay-Standard gel mobility shift assays were carried out as previously reported (8). In brief, the purified protein and 0.05 A 260 unit of yeast tRNA Phe transcript were incubated in 20 l of buffer B at 37°C for 20 min. 6% polyacrylamide gel (width, 90 mm; length, 90 mm; thickness, 1 mm) was prepared with buffer C (50 mM Tris base, 50 mM acetic acid, and 5 mM magnesium acetate). 4 l of loading solution (0.25% bromphenol blue and 30% glycerol) was added to each sample and loaded onto the gel immediately. The electrophoresis was carried out at room temperature by constant voltage (100 V) for 1 h. The gel was stained with Coomassie brilliant blue for detection of protein and then with methylene blue for detection of RNA. The quantities of RNA and protein were measured with a Fuji Photo Film BAS2000 imaging analyzer.
To analyze the effect of AdoMet (see Fig. 5, B and C), the experimental procedure was changed as follows. In the assay shown in Fig. 5B, the R41M protein (0 -11 M) and tRNA (5 M) were preincubated at 37°C for 10 min in the buffer B, AdoMet was then added to a final concentration of 50 M, and the mixture was incubated at 37°C for 10 min. In the assay shown in Fig. 5C, the R41M protein (0 -11 M) and AdoMet (50 M) were preincubated at 37°C for 10 min in the buffer B, tRNA was then added to a final concentration of 5 M, and the mixture was incubated at 37°C for 10 min. Loading solution (0.25% bromphenol blue and 30% glycerol) was added to each sample, which was then loaded onto the gel immediately.
Analytical Gel Filtration-Analytical gel filtration was performed with an Á KTAprime chromatography system (Amersham Biosciences) equipped with a Superdex 75 column (10/30; column volume, 23.6 ml) at room temperature. The column was first equilibrated with the buffer D (50 mM Tris-HCl (pH7.5), 5 mM MgCl 2 , 6 mM 2-mercaptoethanol, 200 mM KCl), and the sample was then injected. The flow rate was 0.5 ml/min. Elution profiles were monitored by the absorption of UV at 280 nm. Formation of an enzyme-AdoMet complex was analyzed as follows. The enzyme (14 M) and AdoMet (50 M) were incubated at 37°C for 10 min in 100 l of the buffer D and then injected onto the column. Fig. 1A shows a multiple amino acid sequence alignment of the typical SpoU family methyltransferases reported thus far (5,6,9,22,24,26,27,30,31). We used the ClustalW program to generate the alignment and then modified it according to the publications of Gustafsson (10), Anantharaman (35), and Renalier (24). In Fig. 1A, TrmH and aTrm56 are tRNA (Gm18) methyltransferase from eubacteria (5,6,9) and tRNA (Cm56) methyltransferase from Archaea (24,25), respectively; AviRb (22,28) and RlmB (21,31) are rRNA methyltransferases, which have a ribosomal protein-like domain in their N termini; RrmA (26) and YibK (30) are function-unknown proteins, although the amino acid sequence of RrmA strongly suggests that this protein is T. thermophilus ortholog of the E. coli RlmB. The three conserved motifs (motifs 1, 2, 3) in the SpoU family are boxed in red. The conserved arginine residue in motif 1 (Arg-41 in T. thermophilus TrmH) is the catalytic center proposed in our previous studies (7,8). The 11 blue boxes highlight the basic amino acid residues that are conserved only in the TrmH enzymes. TrmH catalyzes the methyl transfer from AdoMet to the 2Ј-OH group of the G18 ribose in tRNA (Fig. 1B). In the tRNA docking model of the TrmH dimer ( Fig. 1C), these 11 residues are located on the enzyme surface and are accessible to bound tRNA with slight structural changes in the protein or tRNA. We, therefore, expected that some of these residues might be involved in the recognition of tRNA. In this study we substituted these basic amino acid residues with alanine. In addition, the Leu-179 residue was exchanged by stop codon (L179Stop), because the limited proteolysis by trypsin in our previous study showed that the Arg-178 residue was one of the main cleavage sites in TrmH (5). Thus, the L179Stop mutant corresponds to the N-terminal region and the catalytic core of the TrmH. The other 14 mutant proteins used in the tables have been previously reported (7,8); they each contain the substitution of an amino acid residue in one of the three conserved motifs, and we confirmed the decreased affinity of these mutants for tRNA (see Table 2). All of the proteins were purified to homogeneity as judged by 15% SDS-PAGE (data not shown, see Fig. 2). In this study the yeast tRNA Phe transcript was used as tRNA because this is known to be a good substrate RNA for this enzyme (4).

Selection of the Target Sites for the Site-directed Mutagenesis-
Affinity of the Mutant Proteins for AdoMet-First, we assayed the affinity of the mutant proteins for AdoMet. A convenient method to detect the affinity for AdoMet is the AdoHcy affinity column chromatography (Fig. 2). Because AdoHcy is produced from AdoMet by the methyl transfer reaction, this experiment shows the affinity of the enzyme not only for an AdoMet analogue but also for the direct product of the reaction. As shown in Fig. 2, the wild-type enzyme (Control) bound tightly to the column and was not eluted in the flow-through fraction (Flowthrough) and was not eluted by the buffer containing 2 M KCl (Wash). The bound enzyme was eluted from the column by the buffer containing 2 M KCl and 6 M urea (Elution). Ten mutant proteins (R8A, R19A, K32A, H34A, K90A, R109A, R166A, R168A, R176A, and L179Stop) showed the similar elution profiles, demonstrating that the wild-type enzyme and these proteins have substantial affinity for AdoMet. On the basis of these results, we analyzed the kinetic parameters of these proteins for AdoMet ( Table 1). The 10 mutant proteins had K m values for AdoMet comparable with that of the wild-type enzyme, consistent with the results of the AdoHcy affinity column chromatography. Remarkably, however, the V max values of these mutant proteins were considerably lower than that of the wildtype enzyme ( Table 2). These decreases in the V max values can be explained by the lower affinity of the mutants for tRNA, even though they bind to AdoMet sufficiently.
Two mutant proteins (R11A and H71A) showed differences in affinity for AdoHcy as compared with the wild-type enzyme (Fig. 2). The R11A mutant did not absorb to the AdoHcy affinity column and eluted in the flow-through fraction. Thus, the R11A mutant had lost its affinity for AdoMet. Indeed, the methyl transfer activity of the R11A mutant could not be detected (Table 1). By contrast, the H71A mutant was strongly absorbed to the column and was not eluted by 6 M urea under the standard conditions (Fig. 2). The protein was eluted slowly by the repeated treatment with 6 M urea (data not shown). These results indicated that the H71A mutant had increased affinity for AdoHcy as compared with the wild-type enzyme.
The kinetic parameters of the H71A mutant for AdoMet could be measured. The K m value of the H71A protein was not changed significantly as compared with that of the wild-type enzyme; however, the initial velocity of the methyl transfer was decreased to 0.068 that of the wild-type enzyme (Table 1). Thus, the release of the AdoHcy seems to be slower for the H71A mutant than for the wild-type enzyme. To address these changes in the affinity for AdoHcy, we mapped the Arg-11 and His-71 residues on the crystal structure of the TrmH-AdoMet complex (see Fig. 7). The Arg-11 and His-71 residues are located near the dimer interface and contact with the Met-147 residue in the ␤6-␣5 connection loop in the other subunit. The ␤6-␣5 connection loop corresponds to the conserved motif 3 region (Fig. 1A), and it contains amino acid residues such as Met-144 and Leu-151 that form the AdoMet binding site (7). Substitution of Arg-11 or His-71 seems to affect the affinity for AdoHcy by causing the structural change in the ␤6-␣5 connection loop. Thus hydrophobic interaction among the Arg-11, His-71, and Met-147 residues is required for the local structure of the enzyme. Furthermore, it should be mentioned that all three residues, Arg-11, His-71, and Met-147 are conserved only  NOVEMBER 10, 2006 • VOLUME 281 • NUMBER 45

Basic Amino Acid Residues in TrmH Enzymes
in TrmH enzymes and not in other proteins in the SpoU family (Fig. 1A). Therefore, the interaction of the Arg-11, His-71, and Met-147 residues may work in conjunction with the local structure of tRNA in release of the product (AdoHcy).
Taking these experimental results together we confirmed that the substitution of each of the nine basic amino acid residues (Arg-11, Arg-19, Lys-32, His-34, Lys-90, Arg-109, Arg-166, Arg-168, and Arg-176) with alanine did not change the affinity for AdoMet but did decrease the initial velocity of the methyl transfer, and two basic amino acid residues (Arg-11 and His-71) are required for the stability of the local structure of the enzyme. Furthermore, we also confirmed that the deletion of the C-terminal region (L179Stop) did not reflect on the AdoMet binding, consistent with our previous study (5).
Affinity of the Mutant Proteins for tRNA-Next, we performed a gel mobility shift assay to detect the affinity of the mutant proteins for tRNA (Fig. 3). In our previous study (8), we found that a complex containing TrmH tetramer and one tRNA transcript was formed at a high concentration of the protein (Fig. 3A). Therefore, the precise K d value of the wild-type enzyme could not be measured, although it was estimated at around 2.8 M (Fig. 3A). In the current study we assayed 12 mutant proteins by the same method and expressed the results in terms of five grades (Ϫ, Ϯ, ϩ, ϩϩ, and ϩϩϩ) (Fig. 3B). We quantified the intensities of the band with a Fuji Photo Film BAS2000 imaging analyzer and calculated the K d values. The distinction between the grades Ϫ (Ͼ8.0 M) and Ϯ (5.0 -8.0 M) was considerably difficult because of the weak intensities of the bands; for example, the difference between R11A and R19A mutant proteins was apparently small. However, it is clear that these mutants classified in the grades Ϫ and Ϯ severely decreased the affinity for tRNA. As expected, 10 mutants (R8A, R11A, R19A, K32A, H71A, K90A, R166A, R168A, R176A, and L179Stop) showed decreased affinity for tRNA (Fig. 3B). Among these, the lower affinity of the R11A and H71A mutants can be explained by changes in the local structure of the enzyme as described above. According to the tRNA binding, the Arg-11, His-71, and Met-147 residues may contact with tRNA directly because these residues are located near the elbow region of the bound tRNA in the docking model (Fig. 1C).
Unexpectedly, two mutant proteins (H34A and R109A) did not show decreased affinity for tRNA under the tested conditions (Fig. 3B). Our previous structural study showed that the His-34 together with Lys-32 could contact the phosphate group of the G18 ribose (7). Measurement of the kinetic parameters in the current study demonstrated that the substitution of the His-34 residue did not have a significant effect on the enzyme activity as compared with substitution of Lys-32 (Table 2). Thus, these data show that the Lys-32 residue is more important for the phosphate binding than the His-34 residue.
Unexpectedly, however, the R109A mutant showed slightly increased affinity for tRNA as judged by the gel shift assay, and the kinetic assay showed that this mutation did not have a significant effect on the K m and V max values. In the crystal structure of the TrmH-AdoMet complex, Arg-109 of one subunit forms a hydrogen bond with the Glu-165 residue in the other, but this hydrogen bond is missing in the structure of the TrmH without AdoMet (see Fig. 7B). This structural change in the absence of AdoMet may increase the affinity for tRNA as observed in the gel shift assay. The Arg-109 residue is located near the phosphate group of the C63 in the T-stem of tRNA in the docking model (Fig. 1C). In our previous study we confirmed that the T-stem structure is not essential for tRNA recognition by TrmH (3). Therefore, the substitution of the Arg-109 residue does not have a significant effect on the methyl transfer activity. Furthermore, the L179Stop mutant markedly decreased both the tRNA affinity and the methyl transfer activity of the enzyme (Fig. 3B and Table 2), demonstrating the importance of the C-terminal region for tRNA binding, although no conserved amino acid residues were present in this region (Fig. 1A).
Taking these results together, we conclude that eight basic amino acid residues (Arg-8, Arg-19, Lys-32, His-71, Lys-90, Arg-166, Arg-168, and Arg-176) are involved in tRNA binding and that alanine substitution of the His-34 or Arg-109 residue does not have a significant effect on tRNA recognition. Furthermore, we also confirmed that the C-terminal region, which comprises only 19 amino acid residues in T. thermophilus TrmH, is required for efficient tRNA binding despite the low sequence conservation.
Residues Required for Initial tRNA Contact and for Continuation of the Catalytic Cycle-Beyond our expectation, the measurement of the K d and K m values for tRNA binding provided us important information. We found that the conserved basic amino acid residues could be categorized to two classes, (i) residues related only to the K d value and (ii) residues related to both the K d and the K m values (Fig. 3 and Table 2). His-71, Lys-90, Arg-166, Arg-168, and Arg-176 affected only the K d value. In contrast, Arg-8, Arg-11, Arg-19, and Lys-32 affected both the K d and the K m value. Because the His-71 and Arg-11 residues were already shown to be structurally important, as described above, the results with these residues should be ruled out. But why do Lys-90, Arg-166, Arg-168, and Arg-176 influence only the K d value and not the K m value? In general, the K m value incorporates not only binding of substrate tRNA but also releasing of the methylated tRNA, whereas the K d value incorporates only the affinity for substrate tRNA. In the TrmH reaction the release of the methylated tRNA is probably slow, which would significantly affect the K m value. Indeed, the methylated tRNA transcript inhibits the methyl transfer reaction as a competitive inhibitor (4), suggesting that the TrmH does not recognize the Gm18 modification of the bound tRNA at least in the initial binding process. Thus, our experimental results suggest that Lys-90, Arg-166, Arg-168, and Arg-176 are required for the initial binding event but are not involved in the subsequent stage of the catalytic cycle, including the release of the methylated tRNA. In contrast, Arg-8, Arg-11, Arg-19, and Lys-32 are required for both initial binding and continuation of the catalytic cycle.
The categorized residues are assembled in local regions of the TrmH structure (see Fig. 7A). For example, Arg-8, Arg-11, Arg-19, and Lys-32 residues are assembled around the catalytic center (Arg-41) and the conserved motifs (Asn-35 in motif 1, Glu-124 in motif 2, and Asn-152 in motif 3). These residues are located around the G18 in the elbow region of tRNA (Fig. 1C). As the methyl transfer reaction proceeds, these residues and   NOVEMBER 10, 2006 • VOLUME 281 • NUMBER 45 tRNA seem to change the structure of the enzyme. These structural changes, including release of the tRNA, are reflected in both the K m and the K d values. In contrast, Lys-90, Arg-166, Arg-168, and Arg-176 are located in the region of the enzyme that contacts the three-dimensional core of tRNA ( Fig. 1C and see Fig. 7A). Because these residues did not influence the K m value, they seem to function only in the initial tRNA recognition event.

Basic Amino Acid Residues in TrmH Enzymes
Semi-ordered Sequential Mechanism-In this study we used two methods, determination of the K d value by the gel shift assay and measurement of the K m value by the methyl transfer reaction, to analyze the affinity of the mutants for tRNA. However, there is a considerable difference between two values; the K d and K m values of the wild-type enzyme are around 2.8 and 0.1 M, respectively. The K d value was estimated in the absence of AdoMet, whereas the K m value was measured in the presence of AdoMet. Therefore, this discrepancy can be explained if the enzyme has enhanced affinity for tRNA in the presence of AdoMet.
To test this possibility we carried out a kinetic study in which we changed both the AdoMet and the tRNA concentrations (Fig. 4). In general, a two-substrate reaction assay requires that one substrate is fixed at an excess concentration. In this case, however, we intentionally carried out the kinetic assays at various concentrations around the K m values for AdoMet and tRNA. As expected, in accordance with an increase in the AdoMet concentration, the K m value for tRNA was decreased (Fig. 4A), demonstrating that the enzyme shows enhanced affinity for tRNA in the presence of AdoMet. By contrast, when the concentration of tRNA was varied, the K m value for AdoMet changed but not significantly (Fig. 4B). These experiments clearly show that the catalytic mechanism of the TrmH is a semi-ordered sequential mechanism. Based on this finding, we compared the crystal structures of TrmH alone and TrmH-AdoMet complex (see Fig. 7B). Although we could not conclude the mechanism structurally, the structural differences could be observed as described in the discussion section.
AdoMet-dependent Change of the Subunit Structure-The above results prompted us to carry out the additional experiments. In our previous study we found that a methionine substitution in one mutant protein disrupted the catalytic center (Arg-41) by the substitution by Met (8). It should be mentioned that this mutant (R41M) did not lose all activity; very slow methyl transfer was detected (the initial velocity at 50°C was 0.04% of that of the wild-type enzyme) (8). This weak methyl transfer activity is probably brought about the deprotonation of the OH group of the G18 ribose by the other amino acid residues and/or a water molecule instead of the Arg-41 residue. Thus, the complex between the R41M mutant protein, AdoMet, and tRNA dissociates very slowly with the progression of the methyl transfer reaction. We considered, however, that the AdoMet-dependent changes in the affinity of TrmH for tRNA might be monitored by carrying out the gel shift assay with this R41M mutant protein.
In the absence of the AdoMet, the migration pattern of the R41M mutant is similar to that of the wild-type protein, although the K d value is slightly larger (Fig. 5A). In the experiment shown in Fig. 5B, the R41M protein was preincubated with tRNA, and then AdoMet was added. The gel shift pattern of this sample resembles that obtained in the absence of AdoMet (Fig. 5, A and B). However, the preincubation with AdoMet caused a marked change in the gel-shift pattern (Fig.   FIGURE 3. Gel mobility shift assay. A, affinity of the wild-type TrmH for the yeast tRNA Phe transcript was analyzed by the gel mobility shift assay using conditions described under "Experimental Procedures." The gel was doublestained with Coomassie Brilliant Blue and methylene blue. As the protein concentration increased, a tetramer-tRNA complex was formed in addition to the dimer-tRNA complex. Thus, it was difficult to determine the precise K d value, although it was estimated to be around 2.  5C). The majority of the protein formed a tetramer-tRNA-AdoMet complex, and the K d value of the dimer protein for tRNA could not be estimated because the dimer-tRNA-AdoMet complex was hardly detectable (Fig. 5C). Thus, the change in the affinity of TrmH caused by the presence of AdoMet was difficult to demonstrate by the gel shift assay.
Importantly, however, the results showed that AdoMet has an effect on the subunit structure of TrmH. To address this idea, we carried out the gel-filtration chromatography of the wild-type enzyme. In the absence of AdoMet the wild-type enzyme was eluted as a single peak at the location of the dimer (Fig. 6A). In the presence of AdoMet, the elution profile of TrmH showed a more complicated pattern (Fig. 6B). Two peaks eluted at 8.9 and 12.2 ml, corresponding to the elution points of the tetramer and dimer, respectively. The two further peaks at 19.9 and 26.1 ml were derived from AdoMet (data not shown). Thus, these experiments demonstrate that AdoMet causes the change of the subunit structure of TrmH from a dimer to a tetramer. It should be noted, however, that the formation of the tetramer by AdoMet is observed only at high concentrations of the enzyme. The tetramer band on the gel shift assay was detected at an enzyme concentration of more than 2 M (Fig.  5C), and the gel-filtration chromatography was carried out at 14 M (Fig. 6B). Thus, even at a concentration of 14 M, not all of the protein formed the tetramer structure. On the other hand, the kinetic enzyme assay, which demonstrated the semi-ordered sequential mechanism, was carried out with an enzyme concentration of 2.3 nM.
Recently, we used Western blotting analysis to quantify the amount of TrmH in living T. thermophilus cells. 4 Our preliminary results showed that the TrmH concentration in late log phase cells cultured at 75°C is 0.1-1.0 M, although estimation of the precise volume of the cells was difficult. Thus, in living cells the concentration of the TrmH protein seems to be too low to form the tetramer structure observed in vitro study. In living cells, the concentration of AdoMet is expected to be much higher than the K m value for AdoMet. Therefore, almost all TrmH proteins probably exist in vivo as the AdoMet-bound form, and the majority of them seem to behave as a dimer. If the TrmH protein is localized to one area in a living cell (for example, as part of the tRNA-processing-modification complex), however, it may form a tetramer structure.

DISCUSSION
Here we have demonstrated the roles of the conserved basic amino acid residues in the TrmH enzymes. As a result, these residues can be categorized as having a role (i) in the catalytic center (Arg-41), (ii) in the initial site for tRNA binding (threedimensional core recognition) (Lys-90, Arg-166, Arg-168, and Arg-176), (iii) in the tRNA binding site required for continuation of the catalytic cycle (elbow region recognition) (Arg-8, Arg-19, and Lys-32), (iv) in the structural element involved in release of AdoHcy (Arg-11-His-71-Met-147 interaction), (v) in the assisted phosphate binding site (His-34), or (vi) in unknown function (Arg-109). The positions of these residues are highlighted on the TrmH structure (Fig. 7A). Of these residues it is not clear whether those in category (ii) maintain contact with the tRNA through the whole catalytic cycle; however, it is clear that the substitution of these residues with alanine has no obvious effect on the release of the methylated tRNA. The gel mobility shift assay showed that the individual substitution of these residues had a weak effect on the affinity of the enzyme for tRNA, suggesting that they support the correct binding of 4 C. Iwashita and H. Hori, unpublished information. To investigate the effect of AdoMet on the subunit structure of TrmH, the gel mobility shift assays were carried out using the R41M mutant protein. The gel was doubly stained with Coomassie Brilliant Blue for protein detection and methylene blue for RNA detection. A, the gel mobility shift assay was carried out in the absence of AdoMet. B, the gel mobility shift assay was carried out after preincubation of the R41M protein with tRNA followed by addition of AdoMet as described under "Experimental Procedures." C, the gel mobility shift assay was carried out after preincubation of the R41M protein with AdoMet followed by the addition of tRNA as described under "Experimental Procedures." FIGURE 6. Analytical gel filtration chromatography to analyze the effect of AdoMet on the subunit structure of TrmH. A, the wild-type enzyme alone was analyzed by Superdex 75 gel-filtration chromatography. Arrows show the location of the molecular standard markers. B, the wild-type enzyme was analyzed as in panel A after preincubation with AdoMet, as described under "Experimental Procedures" and then injected into the column. mAu, milliabsorbance units.
tRNA to the catalytic pocket around the Arg-41 residue. In contrast, the substitution of the residues in category (iii) had a severe effect on both tRNA binding and the methyl transfer activity, demonstrating that these residues are important for the induced fit of the tRNA to the catalytic pocket as well as the release of the product. Until now, we have not been able to visualize the mechanism by which the TrmH-methylated tRNA-AdoHcy complex dissociates after the reaction. However, our current results provided a clue to elucidating this mechanism. In the previous study we showed that the substitution of the Asn-35 residue by Gln enhances the affinity for AdoHcy, and the catalytic cycle was halted (8). In the current study we found that the residues in category (iv) are involved at least in the release of AdoHcy. Taking these results together, the subunit interaction around bound AdoHcy seems to be involved in the release of the AdoHcy. This idea is in line with the fact that all SpoU family members characterized to date (Fig. 1A) have a dimer structure (7, 8, 26 -31). In the current study we demonstrated that the C-terminal region of the protein significantly affects tRNA recognition. Recently, the crystal structure of A. aeolicus TrmH was solved (27). Our previous study showed that A. aeolicus TrmH effectively methylates specific tRNA species (6), whereas T. thermophilus TrmH methylates all tRNAs. The obvious difference between these TrmH proteins is the length of the C terminus. Taking these results together with the tRNA modification pattern (12,13) and the amino acid sequence of E. coli TrmH (9), the structure of the C terminus seems to be involved in the specificity of TrmH toward tRNA species despite the low conservation of the amino acid sequence in this region.
Among the SpoU members characterized, Haemophilus influenzae YibK protein is the smallest protein (Fig. 1). The crystal structure of this protein was previously solved (30), and more recently, the mechanism underlying the formation of the H. influenzae YibK dimers has been studied in detail (38). The striking homology between TrmH and YibK suggests that YibK is an RNA ribose 2Ј-O-methyltransferases; amino acid residues involved in the catalytic center and AdoMet binding site are conserved in the YibK protein. Furthermore, our computational analysis of the H. influenzae genome suggests that the YibK protein is the best candidate for an enzyme of the Gm18 modification system (data not shown), although whether the Gm18 modification occurs in H. influenzae tRNA has not been confirmed. However, the N-terminal region is missing in the YibK protein. In the current study, we demonstrated that the basic amino acid residues in the N terminus of the TrmH (Arg-8, Arg-11, and Arg-19) are required both for continuation of the catalytic cycle and for the structural element involved in release of AdoHcy. Lim et al. (30) have predicted that the YibK may require another partner protein (subunit) for its enzyme activity. The experimental results demonstrated in this study support their idea.
The crystal structure of Bacillus subtilis TrmB (tRNA (m 7 G46) methyltransferase (EC 2.1.1.33)) (39 -42), which catalyzes the methyl transfer from AdoMet to the N 7 atom of the semiconserved G46 in the extra-loop of tRNA, has been reported recently (37). The study revealed that TrmB is a unique variant of the Rossmann-fold methyltransferase (class I fold). In the study Zegers et al. (37) showed that sinefungin, an analogue of AdoMet, induced the aggregation of the TrmB and tRNA complex. In the current study, we demonstrated that AdoMet enhanced the affinity for tRNA and induced the formation of the tetramer structure of TrmH. Therefore, these phenomena, in which AdoMet and its analogue change the affinity for tRNA and a subunit structure, may be common in tRNA methyltransferases despite the difference of the protein fold. Fig. 7B shows the differences of the crystal structures of TrmH alone and Trm-AdoMet complex. The FIGURE 7. Summary of the functions of the conserved basic amino acid residues and structural changes caused by AdoMet. A, summary of the functions of the conserved basic amino acid residues. The basic amino acid residues categorized under "Discussion" are shown on the TrmH dimer structure. One subunit of the dimer is indicated in green, and the other is indicated in cyan. The residues in the cyan subunit are indicated with a prime (e.g. Arg-8Ј). The bound AdoMet molecules are indicated in red. Arg-41Ј (dark blue) forms the catalytic center, which functions in the transfer of methyl from AdoMet bound in the green subunit. The residues involved in the initial binding of tRNA (Lys-90Ј, Arg-166Ј, Arg-168Ј, and Arg-176Ј) are indicated in purple. The residues required for continuation of the catalytic cycle (Arg-8Ј, Arg-19Ј, and Lys-32) are indicated in blue. The structural elements involved in release of AdoHcy (Arg-11Ј, His-71Ј, and Met-147) are indicated in orange. The assisted phosphate binding site (His-34) is indicated in gray. The residue whose function remains unknown (Arg-109) is indicated in magenta. B, comparison of the structure of TrmH alone and TrmH-AdoMet complex. The structure of TrmH alone (orange) is superimposed onto that of TrmH-AdoMet complex (cyan). The amino acid residues analyzed in the current study are indicated by stick models. The residues in the TrmH alone and TrmH-AdoMet complex are indicated in magenta and dark blue, respectively. The N-terminal amino acid residues (1-8) were disordered in the structure of TrmH alone. marked difference between two structures is the direction of the basic amino acid residues in category (iii), which are required for continuation of the catalytic cycle and located near the elbow region of tRNA in the docking model; these residues in the TrmH-AdoMet complex are opened to the solvent. These structural differences seem to change the affinity for tRNA and affect the subunit structure.
Recently, a new member of the SpoU family, aTrm56 was found in the Archaea domain (24,25). The amino acid sequence of this archaeal enzyme is considerably different from that of bacterial SpoU members (Fig. 1). Indeed, several amino acid residues in motifs 1-3 are replaced by the other amino acid residues; asparagines in motif 1 (Asn-35 in TrmH) and in motif 3 (Asn-152 in TrmH) are replaced by histidine and glutamic acid, respectively, in aTrm56. In TrmH, these residues form a hydrogen-bond network (Asn-35-Glu-124 -Asn-152) and function in tRNA binding (phosphate binding) and protein stability (8). Furthermore, basic amino acid residues investigated in the current study are not found in the aTrm56 sequence. Therefore, the tRNA recognition mechanism of the aTrm56 protein may differ considerably from that of bacterial TrmH. To resolve these issues further studies will be necessary.