Article | Published:

Chemical trapping and crystal structure of a catalytic tRNA guanine transglycosylase covalent intermediate

Nature Structural Biology volume 10, pages 781788 (2003) | Download Citation


  • A Corrigendum to this article was published on 01 July 2004


Prokaryotic tRNA guanine transglycosylase (TGT) catalyzes replacement of guanine (G) by 7-aminomethyl-7-deazaguanine (PreQ1) at the wobble position of four specific tRNAs. Addition of 9-deazaguanine (9dzG) to a reaction mixture of Zymomonas mobilis TGT and an RNA substrate allowed us to trap, purify and crystallize a chemically competent covalent intermediate of the TGT-catalyzed reaction. The crystal structure of the TGT–RNA–9dzG ternary complex at a resolution of 2.9 Å reveals, unexpectedly, that RNA is tethered to TGT through the side chain of Asp280. Thus, Asp280, instead of the previously proposed Asp102, acts as the nucleophile for the reaction. The RNA substrate adopts an unusual conformation, with four out of seven nucleotides in the loop region flipped out. Interactions between TGT and RNA revealed by the structure provide the molecular basis of the RNA substrate requirements by TGT. Furthermore, reaction of PreQ1 with the crystallized covalent intermediate provides insight into the necessary structural changes required for the TGT-catalyzed reaction to occur.


Most functional RNAs, such as ribosomal RNA, transfer RNA (tRNA) and small nuclear RNA, are subject to post-transcriptional modification. Currently, 96 modified nucleosides have been discovered in RNAs, and 81 of them can be found in tRNA1. For a particular tRNA, the region surrounding the anticodon (positions 34 and 37) is often modified extensively. As biological consequences of these modifications emerge, it is not surprising to discover that they are related to the efficiency and fidelity of protein synthesis2,3,4,5.

One of these hypermodifications, the replacement of a guanine (G) by a queuine (Q) at position 34 in four tRNAs specific for asparagine, aspartate, histidine and tyrosine, is catalyzed by an RNA modification enzyme called tRNA guanine transglycosylase6,7. So far, the TGT reaction has been studied most extensively in eubacteria, where G34 is replaced by PreQ1 rather than Q (Fig. 1a; note that Q has an additional modified ribose attached to the PreQ1 through the amino group shown in red). In these organisms, two additional enzymatic reactions are required to convert PreQ1 to Q8,9. Eukaryotic TGT, however, catalyzes the replacement of G34 by Q in a single step10. Despite this difference, high sequence similarity exists between TGT from these two classes of organisms, implying that they share the same mechanism of catalysis.

Figure 1: Trapping of a TGT–RNA covalent intermediate.
Figure 1

(a) The chemical reaction carried out by a prokaryotic TGT. The N9 of G and PreQ1, which is a carbon atom in the base analog 9dzG, are blue. The change in chemical structure resulting from the reaction is red. (b) SDS-PAGE analysis of reaction mixtures of TGT and RNA substrates under conditions indicated. (c) SDS-PAGE analysis of purity, stability and chemical reactivity of the isolated covalent intermediate. SM, size markers; d, days; h, hours; m, minutes.

Biochemical studies of TGT from Escherichia coli defined the minimum requirement for an effective RNA substrate as a stem-loop RNA with a recognition sequence of UGU at positions 33–35 (refs 11,​12,​13). Crystal structures of TGT from Zymomonas mobilis alone and in complex with PreQ1 and mutational studies suggested a critical role of Asp102 in catalysis14,15. These studies led to a proposed mechanism of the TGT-catalyzed reaction: (i) nucleophilic attack of the side chain of Asp102 on the C1′ of the targeted nucleotide, G34, to assist the departure of the guanine base; (ii) replacement or exchange of guanine by PreQ1 in the guanine/PreQ1-binding pocket in TGT; and (iii) covalent linkage of PreQ1 to the RNA through a nucleophilic attack of the N9 of PreQ1 on the C1′ of the targeted ribose to form the product.

The proposed mechanism implies the formation of a TGT–RNA covalent intermediate during the reaction. This has been further supported by several studies, including identification of a covalent complex as detected by SDS gel analysis of a slowly denatured TGT–RNA reaction mixture15,16. By using the base analog 9dzG, which differs from guanine by only one atom—carbon instead of nitrogen at position 9—we have successfully trapped the covalent intermediate of the TGT-catalyzed reaction. Here we report the trapping process, the crystal structure of the covalent intermediate and the structure of TGT bound to the RNA product that resulted from chemical reaction of PreQ1 with the trapped covalent intermediate in a crystal.


Trapping of the TGT–RNA covalent intermediate

In the presence of excess 9dzG, incubation of TGT with a stem-loop RNA substrate resulted in the appearance of a retarded band in a SDS gel (Fig. 1b, lane 4). Efficient formation of this product required both the RNA substrate and 9dzG, as reactions lacking both (lane 1) or lacking RNA (lane 2) did not produce such a product. Formation of a small amount of this product was also observed when TGT was added to a reaction with the RNA substrate in the absence of 9dzG (lane 3), consistent with observations of previous studies15,16. The D102A (lane 5) and D102N (lane 6) mutations abolished the ability of these TGT mutants to form such a product. Alternatively, changing the RNA substrate at the recognition nucleotide positions 33, 34 and 35 also eliminated the ability of wild-type TGT to form such a product (lanes 7–9).

The product described above was isolated by MonoQ anion exchange chromatography in a FPLC system. MALDI-MS analysis of the isolated material confirmed its identity as the covalent intermediate (data not shown). The isolated covalent intermediate was at least 95% pure as judged by an SDS gel (Fig. 1c, lane 1). In aqueous solution, the intermediate was less stable in the denatured form (lane 4) than in the native form (lanes 2 and 3). Even in native form, some degradation occurred with prolonged incubation, especially at higher temperature (compare lane 3 with lane 2), but the degradation could be minimized by the addition of 9dzG immediately after purification (lane 5 compared with lane 3). Brief incubation of the isolated covalent intermediate with a molar excess of guanine or PreQ1 at 25 °C resulted in almost complete conversion of the covalent intermediate to TGT (lanes 6 and 7). This finding suggested that the purified covalent intermediate was chemically active—incubation with excess guanine resulted in the formation of the original substrate, and incubation with excess PreQ1 produced the product.

Structure of TGT–RNA–9dzG ternary complex

The purified covalent intermediate was crystallized and the structure was determined to a resolution of 2.9 Å. TGT forms a symmetric dimer, with two dimers occupying the asymmetric unit. Within each dimer, both monomers contain 9dzG, but only one monomer associates with the RNA substrate. Structural analysis indicates that the TGT monomer in the TGT–RNA–9dzG ternary complex is solely responsible for the recognition of the RNA substrate and for catalysis. Therefore, our discussions will focus on this TGT monomer and its interactions with RNA and 9dzG. We cannot rule out the possibility that the TGT monomer in the TGT–9dzG binary complex may have some role, such as contributing to binding when an entire tRNA is used as the substrate, in the TGT-catalyzed reaction.

The structure of TGT in the TGT–RNA–9dzG ternary complex is similar to the one in the binary complex, as well as the one reported previously14. In brief, TGT adopts a variation of an eight-fold β/α TIM barrel17, differing by additions at both N and C termini, as well as by insertions in the β3/α3 and the β8/α8 loops, when compared with the standard (β/α)8 TIM barrel proteins. Therefore, instead of sequential assignment of secondary structures from the N to the C terminus as described previously14, here we divide the elements of the secondary structure of TGT into two components. One component is the core of (β/α)8 TIM barrel (Figs. 2 and 3a, grayish blue). The other component is the collection of additions in the N and C termini (named βABC and αG, respectively) and insertions in the β3/α3 and β8/α8 loops (named Ins3 and Ins8, respectively) (Figs. 2 and 3a, orange). This new assignment may reflect a possible evolutionary origin of TGT, and the benefit of the assignment will become clear when the interaction of TGT with RNA substrate is discussed later.

Figure 2: Alignment of TGT sequences from different organisms.
Figure 2

The sequence of Z. mobilis TGT is compared with representatives of other prokaryotic (E. coli and Bacillus subtilis) and eukaryotic (Saccharomyces pombe and human) TGTs. Residue number over the alignment corresponds to Z. mobilis TGT. The conserved residues are boxed in color, with completely conserved residues in magenta, identical residues in yellow and similar residues in cyan. The secondary structure of TGT is depicted schematically above the primary sequence, with α-helices highlighted as cylinders and β-sheets as arrows. The assignment is based on structural comparison of TGT with the standard (β/α)8 TIM barrel proteins. The core of the (β/α)8 barrel is grayish blue, and the N-terminal and C-terminal additions (βABC and αG), as well as insertions in β3/α3 and β8/α8 loops (Ins3 and Ins8), are orange. The asterisk denotes residues acting as ligands for a zinc ion.

Figure 3: Overall structure of the trapped covalent intermediate.
Figure 3

(a) Ribbon representation of the TGT–RNA–9dzG ternary complex. TGT is colored according to the assignment of the secondary structures in Figure 2. The RNA is dark green. 9dzG is shown in ball-and-stick model in red, and a zinc ion is represented by a ball in magenta. (b) Summary of TGT–RNA, TGT–9dzG and RNA–RNA interactions. TGT, RNA and 9dzG are colored as in a except the phosphate groups of RNA are in magenta. The conserved RNA nucleotides 33–35, 9dzG and Asp280 are highlighted with color-filled boxes. Arrows in blue represent hydrogen bonds between RNA and TGT, RNA and water, or 9dzG and TGT. Arrows in red show stacking of amino acid side chains from TGT on bases of RNA or on 9dzG. Two arrows in green represent intramolecular interactions in RNA. 'W' in a circle represents well-positioned water molecules in the structure.

The RNA substrate sits on top of the C-terminal face of the (β/α)8 barrel core, flanked by Ins3 and Ins8 (Fig. 3a). As predicted from its nucleotide sequence, the RNA folds into a stem-loop structure, with six Watson-Crick base pairs as the stem and six unpaired nucleotides and a ribose (resulting from the detachment of base G34 by TGT) as the loop (Fig. 3). The overhanging 5′ adenine residue is involved in crystal packing, base pairing with the same residue of RNA in the second dimer within the asymmetric unit, thus resulting in extension of helix from one stem to the other (data not shown). The RNA stem interacts only with Ins8 of TGT (Fig. 3a). These interactions are not sequence specific, acting primarily through van der Waals contacts and hydrogen bonds between residues in TGT and the phosphate backbone (Fig. 3b). In contrast, the loop of RNA interacts extensively with several regions of TGT, including Ins8, Ins3 and the (β/α)8 barrel core (Fig. 3a). Many of these interactions are sequence specific (Fig. 3b). Although the stem maintains the conformation of a standard A-form RNA duplex seen in other RNA stem-loop structures, the conformation of the loop in the structure of the covalent intermediate is unusual. Specifically, the helical stacking of base pairs in the stem extends to the first base of the unpaired nucleotide in the loop, C32. The base of the conserved nucleotide, U33, has flipped out, resting in a cleft formed by the loops of β6/α6 and β7/α7 in TGT. The original nucleotide of G34 is no longer present. In its place is a ribose that is covalently linked to TGT through the side chain of Asp280, and a 9dzG molecule that occupies a nearby pocket and interacts with several strictly conserved residues in TGT (Fig. 3b). The conformations of the four remaining nucleotides in the loop, U35–U38, are also unusual. Although the bases of U35 and A37 remain inside the loop, the bases of A36 and A38 are flipped out, resulting in a zigzag conformation for this part of the loop (Fig. 3b). In addition to interactions of RNA with TGT, two sequence-independent intramolecular hydrogen bonds in RNA are at least partially responsible for this zigzag conformation: the 2′ OH group of U35 donates a hydrogen bond to the phosphate group of A37, and the N2 of A37, a position considered the universal hydrogen bonding acceptor in the minor groove of nuclei acids, accepts a hydrogen bond from the 2′ OH group of C39 (Fig. 3b, arrows in green).

Conformational changes

In addition to the local base flipping, important global conformational changes also occurred in the loop of RNA upon its association with TGT, as judged by structural comparison with a representative tRNA. Three-dimensional superimposition of the stem of our structure with the stem of tRNAPhe shows that the loop of RNA in our complex is pointed in the direction opposite to that of the tRNAPhe anticodon loop18 (Fig. 4a). Comparisons with other available tRNA structures from the Protein Data Bank, tRNAAsp, tRNALys and tRNAiMet, also produced similar results19,20,21. Some conformational changes in the loop of RNA were expected, as the loop is intrinsically flexible, but the extent of changes observed in our structure was unexpected. We have also compared our structure to the selected tRNAs in complex with their aminoacyl-tRNA synthetases, because conformational changes in the anticodon regions of tRNAs were also observed upon their association with the aminoacyl-tRNA synthetases22,23,24. Of particular interest was the tRNAAsp in complex with the aspartyl-tRNA synthetase22, because the coordinates of the tRNAAsp in the complex were used for building the docking model of the TGT–tRNA interaction in the previous TGT structure14. The comparisons indicate that the conformation in the loop region of RNA in our structure is markedly different from that of the previous structure (data not shown). The change in the loop region of the RNA upon its association with TGT is most likely imposed by TGT, because without the change, a steric clash between the loop of RNA and TGT would occur (for example, imagine placing RNA in red in Fig. 4a into the position of the RNA in Fig. 4b).

Figure 4: Conformational changes in RNA and TGT.
Figure 4

(a) Three-dimensional superposition of the RNA (green) with the corresponding nucleotides of tRNAPhe (red). Only the C1′ and P atoms of the nucleotides from the stem part of RNA, nucleotides 27–31 and 39–43, were used for the alignment. The r.m.s. deviation of the superposition is 1.4 Å. (b) Ca superposition of the TGT in the TGT–RNA–9dzG ternary complex (grayish blue, with RNA and 9dzG in green) with the one in the TGT–9dzG binary complex (red, with 9dzG in yellow). Only the coordinates of the core of (β/α)8 barrel were used for the alignment.

In contrast to RNA, the protein structure seems not to undergo any substantial reorganization. When the TGT in the TGT–RNA–9dzG ternary complex was aligned with the TGT in the TGT–9dzG binary complex, only small structural changes surrounding Ins3 and Ins8, presumably resulted from its association with RNA substrate, were observed (Fig. 4b). Comparison with the previous TGT structure14 also resulted in the same conclusion.

Active site and substrate specificity

Specific interactions revealed by the structure of TGT–RNA–9dzG ternary complex account for the requirements of an effective RNA substrate of TGT: an RNA with a stem-loop structure and the conserved nucleotide sequence of UGU at positions 33–35 (Fig. 5).

Figure 5: The active site and molecular basis of substrate specificity.
Figure 5

(a) Stereo view of TGT active site. The main chains of TGT are grayish blue, the side chains of TGT are orange and the RNA and 9dzG are green. The hetero-atoms are colored individually, with nitrogen in blue, oxygen in red and phosphate in magenta. The C9 of 9dzG is highlighted in black. A water molecule is shown as a small ball in red and labeled 'W'. Hydrogen bonds are indicated with dashed lines. The side chains of Asp106 and Cys158, which block 9dzG from the top, are omitted for clarity. (b,c) Recognition of U33 (b) and U35 (c) in RNA substrate by TGT. The color schemes are the same as in a. The structure of the phosphate group of A38 in c is omitted but labeled for clarity.

At position 34 in RNA, only a ribose, which was presumably generated by the detachment of guanine catalyzed by TGT, could be suitably fitted into the electron density. The extension of continuous electron density from the ribose to the TGT through the side chain of Asp280 points to the formation of a covalent bond between C1′ of the ribose and one of the two terminal oxygen atoms in the side chain of Asp280 (Fig. 7a). This clearly defines Asp280 as the nucleophile responsible for the TGT-catalyzed reaction. The second terminal oxygen atom in the side chain of Asp280 accepts a hydrogen bond from the hydroxyl group in the side chain of Tyr258, presumably orienting the side chain of Asp280 into a conformation suitable for the catalysis (Fig. 5a). The position of Tyr258 itself is secured by its hydrophobic interactions with surrounding residues, such as Met43, Leu100, Met153, Phe199, Met260 and Met278, all of which are strictly conserved (Fig. 2). The importance of Asp280 and Tyr258 was further conformed by mutational studies, as no enzymatic activity was detected with the D280N mutant (Fig. 6, lane 2), and the Y258F mutant had greatly reduced TGT enzymatic activity (Fig. 6, lane 4).

Figure 6: Enzymatic activities of TGT mutants.
Figure 6

(a) SDS-PAGE analysis of reaction mixture to determine the ability of TGT mutants to form the covalent intermediate with the stem-loop RNA substrate in the presence of 9dzG. (b) The relative activity of TGT mutants. The relative activity of each mutant was determined by comparing the amount of radioactivity incorporated into RNA substrate by the mutant to that incorporated by the wild-type TGT, with the relative activity of the wild-type enzyme normalized to 100%.

Figure 7: The chemical reaction carried out in crystal and the resulting structural changes.
Figure 7

(a,b) Soaking crystal of the trapped covalent intermediate in a PreQ1-containing solution results in structural changes near the active site of TGT from a to b. Structures are presented analogous to those in Figure 5, except that the simulated annealing omit electron maps with σA40-weighted mFoDFc coefficients for the Asp280, the ribose 34 and the bases are included. The contour level of the maps is 4.5 σ. (c) Superimposition of structures before and after the reaction. The backbones of the structure of the covalent intermediate are grayish blue, whereas the side chains, the RNA, and the 9dzG are green. The backbones of the structure of TGT bound to the RNA product are magenta, and the side chains and RNA are in red.

On the opposite side of Asp280, relative to the position of the ribose, 9dzG occupies a nearby pocket in TGT with the C9 facing the ribose (Fig. 5a). TGT binds 9dzG in a similar fashion whether or not a RNA substrate is bound (Fig. 4b), which is in good agreement with the binding of PreQ1 as reported14. Most specific interactions seen in the previous TGT–PreQ1 structure can also be observed here. Specifically, the side chain of Asp156 accepts bifurcate hydrogen bonds from the N1 imino group and the N2 amino group of 9dzG, respectively; the side chain of Gln203 and the main chain amide of Gly230 each donate a hydrogen bond to the O6 of 9dzG; and the base of 9dzG stacks on top of the side chain of Met260 (Fig. 5a). However, there is a critical difference as compared with the previous structure of the TGT–PreQ1 binary complex—the conformation of the side chain of Asp102. In our structure, the side chain of Asp102 accepts a hydrogen bond from the N2 amino group of 9dzG and a second one from a well-positioned water molecule, which in turn forms three additional hydrogen bonds with other residues in TGT (the side chain of Gln107 and the main chain amides of Gly104 and Tyr106). In the previously reported structure14, the side chain of Asp102 points away from the PreQ1. The origin of this difference is unclear, but it is unlikely that RNA in the crystals contributes to this difference, as the conformation of the Asp102 side chain in the TGT–RNA–9dzG ternary complex is similar to that in the TGT–9dzG binary complex. Recent structures of archaeosine tRNA-guanine transglycosylase, which catalyzes the exchange of guanine by PreQ0 at position 15 of a tRNA in archaebacteria, have a side chain conformation in the equivalent Asp102 that is similar to that observed in our structure25. Taken together with the observation that Asp280 is covalently linked to the ribose, our structure clearly shows that Asp102 is not the nucleophile for the TGT-catalyzed reaction as previously proposed. However, its role is revealed by additional insight from the structure of TGT bound to the RNA product (discussed in detail below).

The interaction of TGT with U33 is not extensive (Fig. 5b), probably reflecting the fact that uridine is a universal nucleotide at position 33 for all tRNAs (except initiator tRNAiMet of higher eukaryotes)26; therefore, there is probably not much evolutionary pressure for TGT to evolve strict recognition of uracil at this position. In our structure, U33 forms three hydrogen bonds with TGT: the 2′-OH of the ribose to the main chain carboxyl group of Gly261, N3 of the base to the side chain of Asp267, and O4 of the base to the side chain of Lys264 (Fig. 5b). Gly261 is strictly conserved in TGT, and its replacement with alanine results in loss of enzymatic activity (Fig. 6, lane 5). In contrast, Lys264 and Asp267 are not strictly conserved (Fig. 2). The D267N mutant is as active as the wild-type enzyme (data not shown), and even the D267A mutant possesses 3% of the enzymatic activity of the wild-type enzyme (Fig. 6, lane 6). Our structure may not reflect the state of initial recognition of RNA by TGT, in which the base U34 forms hydrogen bonds with the main chain carbonyl and amide groups of Gly263, which are only 4 Å away (Fig. 5b). This possibility is supported in part by mutational studies, as the enzymatic activity of the G263A mutant is 0.5% of that of the wild-type TGT (Fig. 6, lane 7). Perhaps a more likely scenario is that a uridine at position 33 plays an important role for the formation of specific conformation of tRNA required for initial recognition by TGT, consistent with the report that U33 interacts with sugar-phosphate backbones of the nucleotides 35 and 36, as well as with the aromatic ring of base 35 to expose the anticodon to the solvent27.

In contrast to U33, U35 makes extensive interactions with TGT (Fig. 5c), consistent with the requirement for strict recognition of a uracil at position 35 to ensure that only four tRNAs, with GUN anticodons (where N equals adenine, cytosine, guanine or uracil), are the targets of the modification by TGT. The position of base U35 is secured by stacking with the side chains of Lys52 (charge-π interaction) and Val282 (hydrophobic interaction). The specificity for U35 is achieved through extensive hydrogen bonding between the functional groups in the base U35 and side chains in TGT. Specifically, the O2 accepts two hydrogen bonds from the side chain of Arg286, which in turn forms three additional hydrogen bonds with other parts of RNA substrates. The O4 forms a hydrogen bond with the side chain of Arg289, which in turn forms three additional hydrogen bonds with the phosphate group of A38 in RNA and a well-positioned water molecule. O4 also forms an additional hydrogen bond either to a second well-positioned water molecule or to the side chain of Thr285. Arg286 and Arg289 in TGT are of particular interest in terms of the extent of their interactions. Mutational study indicated that Arg289 is critical for the function of TGT, as even the most conserved mutation, R289K, results in total loss of the enzymatic activity (Fig. 6, lane 9). The R286K mutant still retained 3% of the enzymatic activity of the wild-type enzyme (Fig. 6, lane 8), indicating its interaction with RNA is less specific.

Chemical reaction in crystals

Similar to the chemical reaction in solution (Fig. 1c, lane 7), we have now carried out the reaction in crystals by soaking crystals of the covalent intermediate in a solution containing PreQ1. In contrast to the rapid reaction in solution, the reaction in crystal was slow, requiring 3 d to complete with small crystals and 16 d for crystals of a size suitable for data collection. These reaction times probably reflect physical difficulties of carrying out reactions in crystals, especially for reactions that require the replacement of one substrate by another in the active site, such as that catalyzed by TGT.

The continuous electron density from the base to the ribose, and the discontinuous one from the ribose to the side chain of Asp280, indicated the formation of a covalent bond between the base and the ribose and the breakage of the bond between the ribose and the side chain of Asp280 (Fig. 7b). This contrasted completely with the structure of the covalent intermediate, in which the opposite was true (Fig. 7a). The additional electron density near position 7 of the base in the structure of TGT bound to the RNA product further supported the conclusion that PreQ1 was incorporated. All the specific interactions between TGT and 9dzG seen in the covalent structure, such as interactions of Asp102, Asp156 and Gly230 in TGT with the 9dzG, can also be observed with the PreQ1 in the structure of TGT bound to RNA product (Fig. 6a,b). Furthermore, the methylene amino group in PreQ1 contributed additional interactions. The terminal amino group donates two hydrogen bonds to the main chains in TGT, one to the carbonyl group of Leu231 that was also observed in the previous TGT–PreQ1 structure14, and the other to the carbonyl group of Met260 (Fig. 6b).

The structural changes required for the chemical reaction occurred primarily in the RNA and were localized to nucleotide 34 (Fig. 6c). The main body of the ribose rotated 40° away from the side chain of Asp280 and toward the base. The anchoring points of the rotation were the two phosphate groups directly attached to the ribose, whose positions were essentially unchanged. The rotation was accompanied by positional changes of 2.3 Å for the C1′ and 3.7 Å for the 2′ OH group of the ribose, respectively. If we assumed PreQ1 was in the same position as 9dzG after the 9dzG/PreQ1 exchange in the guanine/PreQ1-binding pocket but before the occurrence of the reaction, the position of PreQ1 also changed. The base had a slight rotation and translation in the direction toward the ribose, accompanied by a net positional change of 0.6 Å for the N9. Therefore, formation of a covalent bond between the N9 of PreQ1 and the C1′ of the ribose required movements of the base and, more so, of the ribose toward each other (Fig. 6c). The movement of the ribose also allowed the 2′ OH group to donate a hydrogen bond to the Asp102 side chain (Fig. 6b). The extent of structural change in TGT is much less pronounced than the changes in the RNA. The positions of amino acids that interact with the base, such as the side chain of Asp156 and the main chain of Gly230, appeared to shift slightly, although to a smaller extent and in the same direction as the movement of the base. The only exception was the side chain of Asp102, which appeared to move in the opposite direction to the base (Fig. 6c). The side chain of Asp102 is closer to the C9 of 9dzG (4.1 Å) than to the N9 of PreQ1 (4.9 Å), which has implications in the role of Asp102 in TGT-catalyzed reactions.


Insights obtained from the two structures described here, in combination with reported biochemical studies, allowed us to address two fundamental issues: the molecular recognition of tRNA substrates by TGT and the mechanism of the TGT-catalyzed reaction.

The requirements that an effective RNA substrate must possess a stem-loop structure, as well as the nucleotide sequence UGU at positions 33–35, imply sequence-specific recognition of UGU in the RNA substrate by TGT. Our structures indicate that the recognition of G34 and U35 by TGT is indeed sequence specific. The amino acids involved in the recognition are strictly conserved and mutations of these amino acids either abolish or greatly diminish the activity of the enzyme. However, the key amino acids observed interacting with U33 are not conserved, implying that its recognition may not be sequence specific. Our interpretation is that U33 probably has a structural role in the formation of a specific conformation of the stem-loop. Therefore, the likely scenario for substrate recognition is that TGT initially recognizes the global feature of a stem-loop RNA with a specific conformation, and then specifically recognizes G34 and U35. The recognition of the RNA substrate by TGT accompanies a substantial conformational reorganization in the loop region of RNA, resulting in maximum surface and charge complementation between the loop of RNA and the TGT.

As for catalysis, we now know that Asp280 is the nucleophile. However, Asp280 alone is not sufficient for the TGT-catalyzed reaction to proceed to completion, for the following reasons. After the detachment of G34 and the exchange of guanine by PreQ1 in the guanine/PreQ1-binding pocket, the N9 of the incoming preQ1 must be deprotonated before it is able to form a covalent bond with the C1′ of the ribose. This deprotonation cannot be accomplished without the assistance of a nearby external general base, either from TGT or a well-positioned water molecule, because of the high pKa of the N9 (estimated to be 18)28. Inspection of the surroundings of the TGT near the C9 of 9dzG or the N9 of PreQ1 pointed to Asp102 as the only logical candidate for that general base. One could imagine that a slight conformational change of the side chain of Asp102 from its current position (4.1 Å away from the C9 of 9dzG) would put it in an optimal position to serve as the general base to deprotonate the N9 of the incoming PreQ1. In addition, the well-positioned water molecule hydrogen-bonded with the side chain of Asp102 could shuttle the proton (Fig. 5a). This conclusion is consistent with earlier mutational studies15 as well as those presented here (Fig. 1b).

Having assigned the functional roles of the two critical amino acids, Asp280 and Asp102, involved in catalysis, we can summarize the probable mechanism of the TGT-catalyzed reaction as follows. First, TGT recognizes and associates with tRNA substrates that possess UGU nucleotide sequences at positions 33–35. The side chain of Asp280 then attacks the C1′ of nucleotide 34 to detach the guanine base from the RNA, forming a covalent intermediate between TGT and RNA. Next, PreQ1 replaces the detached guanine in the guanine/PreQ1-binding pocket; this is made possible by conformational changes of Tyr106 and Cys158 side chains in TGT that appear to be in positions to block the pathway for the guanine/PreQ1 exchange. The side chain of Asp102 then deprotonates the N9 of the PreQ1, allowing a nucleophilic attack on the C1′ of the ribose to form the product. Finally, the RNA product dissociates from TGT.

In addition to furthering understanding of the molecular recognition and the mechanism of the TGT-catalyzed reaction, the structures presented here may also shed light on the possible evolutionary origin of TGT. The precise functional assignment of each region of TGT, made possible by the structures presented here, may indicate that TGT acquired its function from a standard (β/α)8 TIM barrel protein by insertions in the β3/α3 and β8/α8 loops (Ins3 and Ins8) and additions of the N termini (βABC) and a C-terminal helix (αG) (Figs. 2 and 3a). Ins3 and Ins8 are mainly responsible for the interaction with the RNA substrate, and αG, which associates with Ins8, may reinforce the interaction (Fig. 4b). Finally, βABC may be required to serve as a lid to seal the N-terminal face of the (β/α)8 barrel core. Without this lid, the active site might be accessible to solvent or other small molecules that could hydrolyze or react with the covalent intermediate to introduce an error.

To our knowledge, TGT is unique as the only known enzyme that is capable of replacing an internal base by an external base in nucleic acids (RNA and DNA) in a single enzymatic reaction. Mechanistically, DNA glycosylases that are involved in DNA repair share some similarities with TGT29,30,31. Similar to TGT, a DNA glycosylase carries out a reaction that detaches a modified base from a deoxyribose. However, for TGT, the detachment of the base is only the first step of reaction, and is followed by exchange of the detached base with a new base and attachment of the new base to the ribose. In this regard, TGT seems to be a more sophisticated enzyme, able to carry out a more complicated and difficult reaction. Our studies here shed light on the structural and functional requirements necessary for an enzyme to carry out such a complicated reaction.

Note added in proof: While this manuscript was under review, the structure of archaeosine tRNA-guanine transglycosylase bound to a tRNA was reported41 (PDB entry 1J2B). Although the mechanism of catalysis was not discussed in the paper, inspection of that structure indicates that the equivalent Asp280 (Asp249) appears to be in a position to act as the nucleophile for the reaction.


SDS gel analysis of trapping reactions.

Construction of an expression vector and expression and purification of recombinant Z. mobilis TGT were analogous to published procedures32. The RNA was purchased from Dharmacon Research (Boulder, CO). PreQ1 and 9dzG were synthesized according to published procedures33,34. For a typical trapping reaction, 10 μM TGT was mixed with 20 μM stem-loop RNA in the presence of 50 μM 9dzG in 10 μl of reaction buffer containing 10 mM HEPES, pH 7.5, 100 mM NaCl, 10 mM MgCl2 and 1 mM EDTA. After the reaction mixture was incubated at 25 °C for 1 h, 5 μl of 2 × standard SDS loading buffer containing 4 mM DTT was added. The resulting mixture was further incubated at 25 °C for 5 min and 12 μl of each sample was loaded into a 12% SDS gel.

Purification and crystallization of the covalent intermediate.

The in situ trapping procedure was carried out on a preparative scale, thereby generating a covalent complex comprising TGT bound to a 20-nucleotide RNA oligomer. This complex was purified by anion-exchange chromatography on MonoQ (Amersham Pharmacia Biotech). Solution containing 9dzG was added to the purified covalent intermediate, and the sample was concentrated. Additional 9dzG solution was then added to the concentrated covalent intermediate, resulting in a crystallization sample containing 175 μM of the covalent intermediate and 200 μM of the free 9dzG. Crystals of the covalent intermediate were obtained using the vapor diffusion method by mixing the crystallization sample with an equal volume of a well solution containing 100 mM HEPES, pH 7.5, 28% (w/v) PEG 6000, 100 mM NaCl and 10 mM spermine. Tetragonal crystals grew as thin rods in 3–5 weeks at 4 °C.

Data collection, structure determination and refinement.

Crystals were briefly soaked (3–5 min) in a cryoprotective solution containing all the components of the well solution plus 20% (v/v) glycerol. The soaked crystals were then mounted in a nylon loop and flash-frozen in liquid nitrogen. Diffraction data were collected at the 14-BMC beamline at the Advanced Photon Source (APS, Argonne, Illinois, USA). Data were reduced with DENZO and SCALEPACK35 (Table 1). The published crystal structure of TGT14 was used for molecular replacement calculations with AMoRe36. Four solutions were obtained. After rigid body refinement, density for RNA substrate was evident, and the stem of RNA was then built with O37 and refined with CNS38. Additional nucleotides in the loop of RNA, as well as 9dzG, were added as refinement proceeded. Phases were improved by many rounds of rebuilding and refinement, with two-fold noncrystallographic symmetry (NCS) restraints in all refinement protocols.

Table 1: Data collection and refinement statistics

Chemical reaction in crystals and structural refinement of TGT bound to the RNA product.

Crystals of the covalent intermediate grown at 4 °C were gradually warmed to 25 °C and then transferred to a reaction solution containing 100 mM HEPES, pH 7.5, 30% (w/v) PEG 6000, 5% (v/v) DMSO, 100 mM NaCl and 10 μM PreQ1. After 16 d of incubation at 25 °C, crystals were harvested for data collection analogous to that of the covalent intermediate. Diffraction data were collected at the 19-ID beamline at APS. To avoid model bias, the structure of TGT bound to the RNA product was refined starting from the published structure of TGT14, independent of the structure of the covalent intermediate. The entire process of refinement was similar to the structural refinement of the covalent intermediate as described.

Assays of enzymatic activities of wild-type and mutant TGTs.

The protocol used for the enzymatic assays is analogous to the one used by Curnow and Garcia13. Specifically, 0.3 μM TGT were incubated with 10 μM stem-loop RNA and 15.6 μM [8-14C]guanine in a buffer containing 100 mM HEPES, pH 7.5, 20 mM MgCl2 and 10 mM DTT. After a 1-h incubation at 37 °C, 50 μl of the reaction mixture was taken out, and the resulting radioactive RNA was recovered by ethanol precipitation, followed by filtration with a glass fiber filter (Whatman GF/C). The recovered RNA was dried and the radioactivity was measured with a liquid scintillation counter. The relative activity of each mutant was determined as the percentage of radioactivity incorporated relative to that by the wild-type TGT. Figures 3a, 4, 5 and 7 were made with RIBBONS39.


The coordinates and structure factors have been deposited in the Protein Data Bank (accession codes 1Q2R and 1Q2S).



  1. 1.

    , & Summary: the modified nucleosides of RNA. Nucleic Acids Res. 22, 2183–2196 (1994).

  2. 2.

    , , , & Improvement of reading frame maintenance is a common function for several tRNA modifications. EMBO J. 20, 4863–4873 (2001).

  3. 3.

    et al. Accurate translation of the genetic code depends on tRNA modified nucleosides. J. Biol. Chem. 277, 16391–16395 (2002).

  4. 4.

    , , , & Wobble modification defect in tRNA disturbs codon-anticodon interaction in a mitochondrial disease. EMBO J. 20, 4794–4802 (2001).

  5. 5.

    et al. Defect in modification at the anticodon wobble nucleotide of mitochondrial tRNALys with the MERRF encephalomyopathy pathogenic mutation. FEBS Lett. 467, 175–178 (2000).

  6. 6.

    & Isolation and characterization of a guanine insertion enzyme, a specific tRNA transglycosylase, from Escherichia coli. J. Biol. Chem. 254, 3061–3066 (1979).

  7. 7.

    et al. Novel mechanism of post-transcriptional modification of tRNA. Insertion of bases of Q precursors into tRNA by a specific tRNA transglycosylase reaction. J. Biol. Chem. 254, 3067–3073 (1979).

  8. 8.

    , , & A new function of S-adenosylmethionine: the ribosyl moiety of AdoMet is the precursor of the cyclopentenediol moiety of the tRNA wobble base queuine. Biochemistry 32, 7811–7817 (1993).

  9. 9.

    , , & New function of vitamin B12: cobamide-dependent reduction of epoxyqueuosine to queuosine in tRNAs of Escherichia coli and Salmonella typhimurium. J. Bacteriol. 170, 2078–2082 (1988).

  10. 10.

    , , , & Evidence that the nucleic acid base queuine is incorporated intact into tRNA by animal cells. Biochemistry 23, 1171–1176 (1984).

  11. 11.

    , , & tRNA-guanine transglycosylase from Escherichia coli: gross tRNA structural requirements for recognition. Biochemistry 32, 5239–5246 (1993).

  12. 12.

    et al. A UGU sequence in the anticodon loop is a minimum requirement for recognition by Escherichia coli tRNA-guanine transglycosylase. J. Biol. Chem. 269, 32221–32225 (1994).

  13. 13.

    & tRNA-guanine transglycosylase from Escherichia coli. Minimal tRNA structure and sequence requirements for recognition. J. Biol. Chem. 270, 17264–17267 (1995).

  14. 14.

    , , & Crystal structure of tRNA-guanine transglycosylase: RNA modification by base exchange. EMBO J. 15, 2850–2857 (1996).

  15. 15.

    , , & Mutagenesis and crystallographic studies of Zymomonas mobilis tRNA-guanine transglycosylase reveal aspartate 102 as the active site nucleophile. Biochemistry 35, 15734–15739 (1996).

  16. 16.

    , & tRNA-guanine transglycosylase from Escherichia coli: recognition of noncognate-cognate chimeric tRNA and discovery of a novel recognition site within the TpsiC arm of tRNAPhe. RNA 6, 233–244 (2000).

  17. 17.

    , , , & Atomic coordinates for triose phosphate isomerase from chicken muscle. Biochem. Biophys. Res. Commun. 72, 146–155 (1976).

  18. 18.

    & The crystal structure of yeast phenylalanine tRNA at 1.93 Å resolution: a classic structure revisited. RNA 6, 1091–1105 (2000).

  19. 19.

    , & Restrained refinement of two crystalline forms of yeast aspartic acid and phenylalanine transfer RNA crystals. Acta Crystallogr. A 44, 112–123 (1988).

  20. 20.

    et al. The crystal structure of HIV reverse-transcription primer tRNA(Lys,3) shows a canonical anticodon loop. RNA 6, 1347–1355 (2000).

  21. 21.

    & The 3 Å crystal structure of yeast initiator tRNA: functional implications in initiator/elongator discrimination. EMBO J. 10, 3105–3111 (1991).

  22. 22.

    et al. Class II aminoacyl transfer RNA synthetases: crystal structure of yeast aspartyl-tRNA synthetase complexed with tRNAAsp. Science 252, 1682–1689 (1991).

  23. 23.

    et al. The crystal structure of phenylalanyl-tRNA synthetase from Thermus thermophilus complexed with cognate tRNAPhe. Structure 5, 59–68 (1997).

  24. 24.

    , & Structural basis of anticodon loop recognition by glutaminyl-tRNA synthetase. Nature 352, 213–218 (1991).

  25. 25.

    et al. Crystal structure of archaeosine tRNA-guanine transglycosylase. J. Mol. Biol. 318, 665–677 (2002).

  26. 26.

    & tRNomics: analysis of tRNA genes from 50 genomes of Eukarya, Archaea, and Bacteria reveals anticodon-sparing strategies and domain-specific features. RNA 8, 1189–1232 (2002).

  27. 27.

    & An extended structural signature for the tRNA anticodon loop. RNA 7, 334–341 (2001).

  28. 28.

    , , & Determination of acidic ionization constants of certain 5-substituted 2-aminopyrrolo[2,3-d]pyrimidin-4-ones. J. Heterocyclic Chem. 33, 767–781 (1996).

  29. 29.

    et al. A nucleotide-flipping mechanism from the structure of human uracil-DNA glycosylase bound to DNA. Nature 384, 87–92 (1996).

  30. 30.

    , , , & Crystal structure of a human alkylbase-DNA repair enzyme complexed to DNA: mechanisms for nucleotide flipping and base excision. Cell 95, 249–258 (1998).

  31. 31.

    , & Structural basis for recognition and repair of the endogenous mutagen 8-oxoguanine in DNA. Nature 403, 859–866 (2000).

  32. 32.

    , , & Purification, crystallization, and preliminary X-ray diffraction studies of tRNA-guanine transglycosylase from Zymomonas mobilis. Proteins 24, 516–519 (1996).

  33. 33.

    & Pyrrolo[3,2-d]pyrimidine folate analogues: “inverted” analogues of the cytotoxic agent LY231514. J. Org. Chem. 60, 7947–7952 (1995).

  34. 34.

    , , & Synthesis of queuine, the base of naturally occurring hypermodified nucleoside (queuosine) and its analogues. J. Chem. Perkin Trans. I, 1637–1644 (1988).

  35. 35.

    & Processing of X-ray diffraction data collected in oscillation mode. Methods Enzymol. 277, 307–326 (1997).

  36. 36.

    AMoRe: an automated package for molecular replacement. Acta Crystallogr. A 50, 157–163 (1994).

  37. 37.

    , , & Improved methods for building protein models in electron density maps and the location of errors in these models. Acta Crystallogr. A 47, 110–119 (1991).

  38. 38.

    et al. Crystallography & NMR System: a new software suite for macromolecular structure determination. Acta Crystallogr. D 54, 905–921 (1998).

  39. 39.

    Ribbon models of macromolecules. J. Mol. Graph. 5, 103–106 (1987).

  40. 40.

    Improved Fourier coefficients for maps using phases from partial structures with errors. Acta Crystallogr. A 42, 140–149 (1986).

  41. 41.

    et al. Alternative tertiary structure of tRNA for recognition by a posttranscriptional modification enzyme. Cell 113, 383–394 (2003).

Download references


We are grateful to T. Conway (University of Oklahoma) for a plasmid encoding the Z. mobilis TGT and to P. Tyler (Industrial Research Ltd., New Zealand) for providing the initial 9dzG. We thank the staffs of beamlines 14-BMC (K. Brister, T. Teng and R. Pahl) and 19-ID (S. Ginell, A. Joachimiak and Y. Kim) at APS for their support during data collections; Y. Elias, K. Phannach and other members of the Huang research group; and D. Kranz, R. Switzer and J. Gerlt for many helpful discussions and critical reading of the manuscript. The work was supported by start-up funds from the University of Illinois and a grant from the National Institutes of Health.

Author information


  1. Department of Biochemistry, School of Molecular and Cellular Biology, University of Illinois at Urbana-Champaign, 600 South Mathews Avenue, Urbana, Illinois 61801, USA.

    • Wei Xie
    • , Xianjun Liu
    •  & Raven H Huang


  1. Search for Wei Xie in:

  2. Search for Xianjun Liu in:

  3. Search for Raven H Huang in:

Competing interests

The authors declare no competing financial interests.

Corresponding author

Correspondence to Raven H Huang.

About this article

Publication history





Further reading