Chemical footprinting and kinetic assays reveal dual functions for highly conserved eukaryotic tRNAHis guanylyltransferase residues

tRNAHis guanylyltransferase (Thg1) adds a single guanine to the −1 position of tRNAHis as part of its maturation. This seemingly modest addition of one nucleotide to tRNAHis ensures translational fidelity by providing a critical identity element for the histidyl aminoacyl tRNA synthetase (HisRS). Like HisRS, Thg1 utilizes the GUG anticodon for selective tRNAHis recognition, and Thg1-tRNA complex structures have revealed conserved residues that interact with anticodon nucleotides. Separately, kinetic analysis of alanine variants has demonstrated that many of these same residues are required for catalytic activity. A model in which loss of activity with the variants was attributed directly to loss of the critical anticodon interaction has been proposed to explain the combined biochemical and structural results. Here we used RNA chemical footprinting and binding assays to test this model and further probe the molecular basis for the requirement for two critical tRNA-interacting residues, His-152 and Lys-187, in the context of human Thg1 (hThg1). Surprisingly, we found that His-152 and Lys-187 alanine-substituted variants maintain a similar overall interaction with the anticodon region, arguing against the sufficiency of this interaction for driving catalysis. Instead, conservative mutagenesis revealed a new direct function for these residues in recognition of a non–Watson–Crick G−1:A73 bp, which had not been described previously. These results have important implications for the evolution of eukaryotic Thg1 from a family of ancestral promiscuous RNA repair enzymes to the highly selective enzymes needed for their essential function in tRNAHis maturation.

tRNA His guanylyltransferase (Thg1) adds a single guanine to the ؊1 position of tRNA His as part of its maturation. This seemingly modest addition of one nucleotide to tRNA His ensures translational fidelity by providing a critical identity element for the histidyl aminoacyl tRNA synthetase (HisRS). Like HisRS, Thg1 utilizes the GUG anticodon for selective tRNA His recognition, and Thg1-tRNA complex structures have revealed conserved residues that interact with anticodon nucleotides. Separately, kinetic analysis of alanine variants has demonstrated that many of these same residues are required for catalytic activity. A model in which loss of activity with the variants was attributed directly to loss of the critical anticodon interaction has been proposed to explain the combined biochemical and structural results. Here we used RNA chemical footprinting and binding assays to test this model and further probe the molecular basis for the requirement for two critical tRNA-interacting residues, His-152 and Lys-187, in the context of human Thg1 (hThg1). Surprisingly, we found that His-152 and Lys-187 alanine-substituted variants maintain a similar overall interaction with the anticodon region, arguing against the sufficiency of this interaction for driving catalysis. Instead, conservative mutagenesis revealed a new direct function for these residues in recognition of a non-Watson-Crick G ؊1 :A 73 bp, which had not been described previously . These results have important implications for the evolution of eukaryotic Thg1 from a family of ancestral promiscuous RNA repair enzymes to the highly selective enzymes needed for their essential function in tRNA His maturation.
tRNAs are nearly all transcribed as precursor-tRNA with additional 5Ј and 3Ј sequences. These excess sequences are subsequently removed by endonucleases and/or exonucleases to produce mature tRNA with uniform 5Ј-and 3Ј-ends for their use in translation (1). However, in the case of histidine tRNA (tRNA His ), 3 the presence of an additional 5Ј guanosine nucleotide (G Ϫ1 ) distinguishes it from all other tRNAs and is a critical identity element for efficient aminoacylation by histidyl-tRNA synthetase (HisRS) (2)(3)(4)(5). The HisRS requirement for G Ϫ1 is a conserved feature throughout all domains of life, and a relatively small number of species described so far deviate from this rule (6 -9).
Interestingly, bacteria and eukaryotes have evolved different mechanisms of incorporating the additional Ϫ1 nucleotide into tRNA His (10 -15). In almost all bacteria and some eukaryotic organelles, the G Ϫ1 nucleotide is encoded in the tRNA His gene, incorporated by transcription, and retained in the mature tRNA through aberrant cleavage by 5Ј-endonuclease RNase P (11,12,15). In contrast, most eukaryotes studied to date incorporate G Ϫ1 posttranscriptionally to the processed 5Ј end of tRNA His (10,16). This posttranscriptional addition of G Ϫ1 is performed by the tRNA His guanylyltransferase (Thg1), which was first described in Saccharomyces cerevisiae and produces a non-Watson-Crick (non-WC) G Ϫ1 :A 73 bp as a result of this reaction (Fig. 1A) (16).
Shortly after the gene encoding Thg1 in S. cerevisiae was identified, it was demonstrated that this enzyme also performed Watson-Crick-templated 3Ј to 5Ј polymerase activity, in addition to the non-WC-dependent activity described initially (17). Despite a lack of overall sequence identity to 5Ј to 3Ј polymerases, the crystal structure of human Thg1 (hThg1) revealed that the active sites of 3Ј to 5Ј and 5Ј to 3Ј polymerases were structurally analogous and both enzymes share a common two metal ion-mediated mechanism (18, 19).
Catalysis of G Ϫ1 addition to tRNA His by Thg1 occurs in three chemical steps (Fig. 1B) (19 -21). First, 5Ј monophosphorylated tRNA His , the product of RNase P cleavage, is activated via adenylylation using ATP. Next, the adenylylated intermediate is attacked by the 3Ј-hydroxyl of the incoming GTP, completing the nucleotidyl transfer step. Finally, pyrophosphate is released, returning the 5Ј-end to its original monophosphorylated status, and extended by one nucleotide (16,20). Kinetic and structural studies have revealed important roles for highly conserved Thg1 residues in catalyzing these chemical steps both in vitro and in vivo (18,19,(22)(23)(24)(25)(26)(27). These include residues that specifically facilitate adenylylation (Lys-44 and Asn-161 according to hThg1 numbering), and residues that selectively facilitate nucleotidyl transfer (Arg-27, Lys-95, and Arg-130) ( Fig. 1) (19,22,25). These important catalytic roles are underscored by the very high (at or near 100%) conservation of these residues among representative eukaryotic Thg1 enzymes studied to date (22). However, these biochemical studies also identified several catalytically essential residues that were similarly conserved, suggesting important roles, but for which specific functions in catalysis remained unknown. Residues His-152 and Lys-187 (hThg1 numbering) were among the most defective of these previously studied alanine variants, exhibiting no detectable ability to add G Ϫ1 to tRNA His in vitro or in vivo (18, 22).
Reflecting the important role of G Ϫ1 in specifying tRNA His identity, eukaryotic Thg1 enzymes are strictly specific for tRNA His , and this specificity is accomplished by recognition of the tRNA His GUG anticodon (28). A crystal structure of Candida albicans Thg1 (CaThg1) bound to tRNA His revealed a network of molecular interactions between the enzyme and the anticodon loop, consistent with the role of the anticodon nucleotides as the major identity element for eukaryotic Thg1. Interestingly, His-152 was featured in this network of Thg1 residues that directly interact with the anticodon, and residue Lys-187 was also located nearby ( Fig. 1B) (23,28). Thus, a model was proposed based on the location of these residues near the critical anticodon element. This model attributed the absence of G Ϫ1 addition activity with the alanine variants to disruption of Figure 1. Thg1 enzymes catalyze G ؊1 addition to tRNA His . A, overall reaction catalyzed by Thg1. 5Ј-monophosphorylated tRNA is first activated by transfer of AMP (orange) to the 5Ј-phosphate (yellow) from ATP. The incoming GTP nucleotide (red) attacks the activated 5Ј-phosphate generating the G Ϫ1 -containing tRNA His . The additional phosphates on the G Ϫ1 product are removed by a pyrophosphate removal step that is catalyzed by the same two-metal ion active site, generating 5Ј-monophosphorylated G Ϫ1 -containing tRNA His . B, Candida albicans tRNA His guanylyltransferase bound to tRNA His lacking the G Ϫ1 nucleotide. This depiction of the CaThg1-tRNA complex was created using PDB ID 3WC1. CaThg1 is a homotetramer that binds two tRNA substrates. For clarity, one of two bound tRNAs is shown (blue) and monomers interacting with the tRNA at the 5Ј-end and at the anticodon loop are highlighted in purple. Absolutely conserved Thg1 residues from 15 representative eukaryotic organisms including C. albicans include Lys-44, Asn-161, Arg-27, Lys-95, Arg-130, His-152, and Lys-187 (human Thg1 numbering). Side chains are colored according to their known functions (top left image), with green indicating roles in adenylylation, orange indicating function in nucleotidyl transfer, and with the His-152 and Lys-187 residues that are the focus of this work highlighted in yellow. His-152 and Lys-187 also appear in proximity to the anticodon stem loop, where a direct interaction between His-152 and tRNA anticodon nucleotides had been previously observed (bottomright image).

Dual functions for conserved Thg1 residues
the enzyme's interaction with the anticodon leading to an inability to correctly position the tRNA 5Ј-end in the active site (23). However, this hypothesis was not directly tested in any mechanistic assays.
We sought to further investigate the molecular basis for the inactivity of alanine variants of His-152 and Lys-187 using chemical footprinting and kinetic assays. Here, we show that removal of either of these residues does not significantly diminish chemical protection of the anticodon nucleotides by the variant enzymes. Moreover, instead of being catalytically inactive as previously thought, we discovered that the H152A and K187A hThg1 variants exhibit robust 3Ј to 5Ј addition activity with tRNA variant substrates that have not been tested previously, but which allow WC-dependent nucleotide addition. Finally, we demonstrated that the inability of the alanine variants to catalyze the non-WC reaction to create the G Ϫ1 :A 73 bp results primarily from a defect in adenylylation, the first step of the 3Ј to 5Ј addition reaction (Fig. 1B). Together, these results suggest dual independent roles for His-152 and Lys-187 residues in the context of Thg1 catalysis and tRNA recognition and indicate that Thg1 utilizes independent WC and non-WC bp recognition sites involving distinct sets of residues.

hThg1 variants and their interaction with the tRNA His anticodon
To probe the interaction of hThg1 with tRNA His at single nucleotide resolution, we performed chemical footprinting with 1-cyclohexyl-(2-morpholinoethyl)carbodiimide metho-ptoluene sulfonate (CMCT). CMCT was chosen because of its ability to modify the N-1 and N-3 positions of unpaired guanine and uracil residues, respectively, and G/U nucleotides make up six of the seven nucleotides within the tRNA His anticodon loop. Positions of CMCT modification were identified as blocks to primer extension using a primer complementary to the 3Ј-end of the tRNA ( Fig. 2A). Consistent with the expected tRNA structure, CMCT-dependent modification of the in vitro transcribed tRNA His was readily detected in the anticodon and D-loops of the tRNA in the absence of protein (Fig. 2B, compare lanes 1-3 versus 4 -6). The location of the primer at the 3Ј-end prevented visualization of T-loop nucleotides.
Modification by CMCT was then performed on the same tRNA His substrate in the presence of 1 M and 5 M hThg1 (concentrations chosen according to the ϳ1 M K D,app,tRNA measured previously using kinetic assays) (19). Protection from CMCT modification across the entire anticodon loop was

Dual functions for conserved Thg1 residues
observed with WT hThg1 in a concentration-dependent manner, consistent with previous structural and biochemical data suggesting the interaction between Thg1 and this part of the tRNA (Fig. 2B, lanes 10 -15; Fig. S1).
We then used this assay to test the role of His-152 and Lys-187 in maintaining the anticodon interaction that was previously hypothesized to be responsible for orienting the tRNA 5Ј-end in the active site for catalysis. Interestingly, both H152A and K187A variant enzymes display very similar patterns of CMCT protection to WT hThg1 ( Fig. 2, C and D), with quantification also revealing statistically significant protection of anticodon nucleotides compared with the no Thg1 controls (Fig. 3). Moreover, no statistically significant differences were identified between the levels of protection observed for either of the variants at each anticodon loop position compared with WT hThg1. Filter-binding assays were used to determine apparent affinity for tRNA His for each variant enzyme ( Fig. S2; Table 1). The K187A variant exhibited a modest binding defect (about 5-fold) compared with WT consistent with this residue playing some role in overall tRNA binding, possibly through electrostatic interaction with the tRNA backbone. For H152A, the observed binding defect was more severe (ϳ25-fold compared with WT) which was again consistent with the previously observed interaction of this residue with tRNA (Table 1). Nonetheless, the ability of the H152A variant to significantly protect the anticodon nucleotides from chemical modification suggests that sufficient binding occurs under the conditions of the assays performed here. Taken together, these data indicate that the severe catalytic defects associated with H152A and K187A vari-ants cannot be attributed simply to a loss of the overall anticodon interaction that causes subsequent misplacement of the tRNA 5Ј-end in the active site. Instead, the loss of activity may be because of other, possibly more direct, roles of His-152 and Lys-187 in catalysis.

Unexpected Watson-Crick-dependent addition activities of His-152 and Lys-187 variants
Thg1 is a homotetramer that binds to two tRNA substrates. In the complex, the two tRNA molecules are bound asymmetrically, with the anticodon end of each tRNA in one monomer, and the 5Ј-end (site of G Ϫ1 addition) located in another monomer ( Fig. 1B) (18, 23). Thus, the possibility of nonequivalent roles for residues derived from different monomers was considered. Indeed, His-152 and Lys-187 residues found in the two subunits that do not interact with the anticodon end of the tRNA are located relatively near the G Ϫ1 addition site, although the lack of interpretable electron density for tRNA nucleotides at the active site prevents observation of any direct contacts between these residues and the bound tRNA. Previous studies demonstrated that alteration of His-152 and Lys-187 to alanine causes complete loss of G Ϫ1 addition activity with WT tRNA His , yet no available structural information helps to rationalize whether either of these residues plays a direct role in the chemistry of nucleotide addition, possibly in addition to its interaction with the bound tRNA.
For all previous studies testing the ability of hThg1 variants to catalyze G Ϫ1 addition, the WT tRNA His that contains an A 73 discriminator nucleotide was used as the substrate in these assays (22,23). However, an alternative substrate that contains a C 73 discriminator nucleotide has also been utilized extensively in other studies, because of the ability of Thg1 enzymes to act as a "reverse polymerase" and incorporate multiple Watson-Crick base paired nucleotides into this substrate (17,21,29,32). We sought to revisit the effects of alterations at His-152 and Lys-187 by testing the WC-dependent addition activity with the C 73 -tRNA His substrate, which had not been tested previously. For these assays, 5Ј-end labeled substrate was incubated with varied concentrations of hThg1 enzyme in the presence of ATP and GTP. After quenching reactions with EDTA and RNase A (which cleaves at the site of the red arrow shown in Fig. 4), reactions were subsequently treated with phosphatase. The use of phosphatase exploits the difference in accessibility of the labeled 5Ј-phosphate in unreacted substrate versus in nucleotide addition products where the labeled phosphate is protected from phosphatase removal. Thus, labeled oligonucleotide fragments corresponding to reaction products can be resolved from labeled P i by TLC (Fig. 4A) (28). As expected based on previous results, WT hThg1 produced products corresponding to addition of multiple G-nucleotides (G Ϫ1 pGpC and G Ϫ2 pG Ϫ1 pGpC) to C 73 -tRNA His (Fig. 4A). Surprisingly, however, K187A catalyzed robust nucleotide addition to this substrate, and H152A was also active, although to a noticeably lesser extent (Fig. 4A). These results are in stark contrast to the complete lack of catalytic activity observed in at least three different studies for H152A and K187A variants with the WT tRNA His (18, 22, 23). Reactivity with CMCT at each indicated position in the tRNA His anticodon loop was quantified based on the intensity of the band corresponding to each primer extension product normalized to the band intensity observed for the full-length tRNA in the same reactions (see Fig. 2). Data represent the average reactivity from three experiments, with error bars corresponding to the S.E. of each measurement. Statistical tests for significance were measured using a two-tailed t test comparing the protein-treated samples (hThg1 ϭ gray; H152A ϭ striped; K187A ϭ dotted) with the CMCTonly control (no protein; black). All protein-treated samples exhibited differences from the CMCT-only control reactivity with p values Ͻ 0.05 (denoted by a single asterisk *).

Dual functions for conserved Thg1 residues
Because the specific activity of the H152A variant was quantifiably weaker than that of K187A hThg1 (Ͻ0.4% of WT compared with 5% for K187A; Fig. 4A), we further probed the nature of the requirement for histidine at this position by making conservative alterations to either phenylalanine, glutamine, or lysine. These replacements for His-152 were chosen to mimic different biochemical properties (aromatic, hydrogen-bonding capacity, and possible positive charge, respectively) of the native histidine side chain. The H152K alteration did not exhibit any activity in any assay, with any tested substrate ( Fig.  4A and data not shown) and was not pursued further. Interestingly, the H152F variant resulted in the most enhancement of WC-dependent G Ϫ1 addition activity compared with the original alanine alteration, whereas the H152Q alteration caused a modest but detectable increase in specific activity compared with H152A (Fig. 4A). Nonetheless, consistent with previous results for H152A hThg1 (18), none of these alternative His-152 replacements elicited any detectable non-WC addition of G Ϫ1 across from A 73 with WT tRNA His compared with the WT enzyme (Fig. 4B). Thus, we conclude that His-152 and Lys-187 play more important roles during the formation of a non-WC G Ϫ1 :A 73 bp than for the WC G Ϫ1 :C 73 nucleotide addition.
We considered two possible explanations for the different patterns of activity of the His-152 and Lys-187 variants with the C 73versus A 73 -containing tRNA. First, His-152 and/or Lys-187 may interact specifically with the discriminator A 73 , and loss of direct contacts to the A 73 nucleotide could be the cause of abrogated activity with A 73 , but not C 73 , variant tRNA. Alternatively, His-152 and/or Lys-187 could be involved in discriminating between the two types of bp (non-WC versus WC), and thus particularly required for efficiency of the non-WC interaction. We reasoned that testing the ability of each variant to perform U Ϫ1 addition with the WT (A 73 ) tRNA would allow us to distinguish between these possibilities, because this reaction would also measure a WC-dependent activity (formation of a U Ϫ1 :A 73 bp), but the substrate would also retain an A 73 nucle-  The diagram to the left shows the aminoacyl-acceptor stem of the labeled tRNA His substrate (the rest of the tRNA is omitted for clarity), with the labeled 5Ј-phosphate indicated in yellow, and the red arrow indicating cleavage site with RNase A to yield the G Ϫ1 pGpC product fragment that is visualized by TLC after G Ϫ1 addition. The presence of additional C 74 and C 75 nucleotides in the substrate 3Ј-end enables polymerization of additional WC-paired G-nucleotides to yield G Ϫ2 pG Ϫ1 pGpC product, as described previously (17,18). Lanes indicated by (Ϫ) correspond to no enzyme control reactions. A labeled species (indicated by *) that occurs as a result of the labeling procedure but is not a product of enzyme reaction because it is present in equal amounts to the no enzyme control, is indicated. Quantification of specific activity exhibited by each enzyme was performed as in "Experimental Procedures," and percent specific activity relative to WT for each active enzyme is indicated below the panel. B, assays testing non-WC addition of G Ϫ1 to WT tRNA His were performed as described in (A), but with the A 73 -tRNA His substrate, which was also processed as indicated by the diagram to the left of the panel to generate the indicated G Ϫ1 pGpC or P i products (and the same nonenzyme-dependent species labeled with *), as indicated. The lack of observable activity prevents comparison of actual specific activity, but based on the inactivity of reactions containing the highest (36 M) enzyme, it can be estimated that the variants are at least 10 4 -fold defective relative to the WT.

Dual functions for conserved Thg1 residues
otide in this case. Under identical experimental conditions used in experiments shown in Fig. 4B, we observed that each of the active His-152/Lys-187 variants for G Ϫ1 :C 73 addition also performed readily detectable U Ϫ1 addition to the A 73 -tRNA, albeit less efficiently than the WT hThg1 (Fig. 5). The H152F variant again exhibited slightly stronger activity than either A or Q alterations. Thus, the need for His-152 and Lys-187 residues for G Ϫ1 :A 73 addition cannot be attributed solely to a requirement for these side chains to specifically interact with the discriminator A 73 nucleotide, but rather to an interaction with the entire non-WC G Ϫ1 :A 73 bp during catalysis.

His-152 and Lys-187 play critical roles in activation of tRNA His for non-WC addition
To further define the role of His-152 and Lys-187 in the non-WC G Ϫ1 addition reaction, we took advantage of our previous kinetic framework that allowed us to separate the three chemical steps of the G Ϫ1 addition reaction to tRNA His (Fig.  1A). We previously demonstrated that use of a 5Ј triphosphorylated tRNA His (ppp-tRNA His ) substrate bypasses the requirement for activation of the tRNA with ATP by providing an alternative activated 5Ј-phosphate at the tRNA 5Ј-end, thus enabling measurement of the rate-limiting step for nucleotidyl transfer (Fig. 1, step 2) (19,28). In contrast to the complete lack of G Ϫ1 addition activity with any tested His-152 or Lys-187 variant with monophosphorylated A 73 -tRNA His (Fig. 4B), G Ϫ1 addition activity was readily observed for all of the variants with this same tRNA when presented in triphosphorylated form (ppp-tRNA His ). This is evident from the release of labeled pyrophosphate (which is also further hydrolyzed to P i during the assay) with all tested enzymes (Fig. 6).
To quantify these effects, single turnover rate constants (k obs ) were compared for WC versus non-WC base-paired additions to triphosphorylated tRNAs (Table 1; Figs. S3 and S4). All assays were performed at high concentrations of enzyme (15 M) and GTP (1 mM), which are saturating for this activity catalyzed by WT hThg1. Non-WC addition of G Ϫ1 to A 73 -tRNA His was significantly reduced for H152A by ϳ95% compared with WT. The stimulatory effect of the H152F variant was readily observed resulting in a 5-fold improvement in k obs compared with the alanine variant (Table 1; Fig. S3). However, none of these effects are as severe as the Ն10 4 -fold loss of activity that was evident in the assays with the monophosphorylated p-tR-NA His (Fig. 4B) (18, 19). Therefore, we conclude that the requirement for His-152 and Lys-187 during addition of the non-WC G Ϫ1 nucleotide to A 73 -tRNA His primarily reflects an important role for these residues in the first (activation) step of the reaction. Moreover, and once the 5Ј-end activation step is completed, the need for His-152 and Lys-187 is no longer as critical (Fig. 1).
Interestingly, for the WC G:C reaction, k obs were affected quite differently, with the variant enzymes exhibited slightly faster (4-to 5-fold) k obs for this nucleotidyl step compared with the WT enzyme (Table 1; Fig. S4). The molecular basis for this observation is not known, but again supports the model of an active site that utilizes distinct active site features for the addition of incoming nucleotides depending on the type of basepaired interaction they make with the substrate tRNA.

Discussion
The interaction of Thg1 enzymes with the tRNA His anticodon is a critical feature for selective tRNA recognition that is well-supported by both biochemical and structural data (23,28). Nonetheless, the mechanism by which this interaction enables catalysis to occur at the 5Ј-end of tRNA, ϳ70 Å away from the anticodon, has remained largely a mystery. Here we carried out a detailed biochemical characterization of the catalytic role of two key tRNA-interacting residues in the context of human Thg1. Our data reveal dual functions for the conserved His-152 and Lys-187 residues in both the tRNA-anticodon interaction and in bp recognition during catalysis. This previously unknown catalytic role for the His-152 and Lys-187 side chains is most significantly observed on the 5Ј-end activation step for non-WC G Ϫ1 addition to tRNA His , which is the essential physiological function of this enzyme in the majority of eukaryotic systems.
In the CaThg1-tRNA complex structure, a network of highly conserved residues, including the analogous residues to His-152 and Lys-187, are observed either interacting directly with or nearby the anticodon of the tRNA (23). Nonetheless, the relatively unchanged tRNA His anticodon footprint with either of the variant enzymes suggests that the overall interaction with the anticodon loop is sustained in H152A and K187A variants that were previously believed to be entirely catalytically defective. The protections observed across the whole anticodon loop with hThg1 suggest that the interaction with this part of the tRNA is multifaceted. Therefore it may not be a surprise that any single mutation of these residues fails to abolish all interactions with the anticodon loop. Instead, the biological fidelity of Thg1 is apparently ensured through the combined effects of the interactions with a network of residues.
The 4:2 (Thg1:tRNA) stoichiometry revealed by the C. albicans tRNA-bound structure suggested the possibility of nonidentical functions for individual monomeric subunits in the context of the tetrameric enzyme (23) (Fig. 1B). This study fur- Figure 5. Thg1 variants perform Watson-Crick U -1 addition to A 73 -tRNA His . WC-dependent U Ϫ1 addition across from the A 73 discriminator nucleotide was measured by phosphatase protection assay. Reactions contained 5Ј-32 P-labeled A 73 -tRNA His and 36 M hThg1 enzymes and were incubated at room temperature for 2 h. The diagram to the left shows the aminoacyl-acceptor stem of the labeled tRNA His substrate (the rest of the tRNA is omitted for clarity), with the labeled 5Ј-phosphate indicated in yellow, and the blue arrow indicating cleavage site with RNase T1 to yield the U Ϫ1 pG product fragment after U Ϫ1 addition, and P i from remaining unreacted substrate. The identity of the labeled reaction products is indicated on the TLC image, including the nonenzyme-dependent species marked with an asterisk (*) that is also observed in the no enzyme control reactions.

Dual functions for conserved Thg1 residues
ther supports this scenario, providing the first examples of Thg1 residues that appear to have distinct molecular functions in the context of different Thg1 monomers. Although in the CaThg1-tRNA co-crystal structure, His-152 (His-154 in CaThg1) is located on ␣ helix 5, which appears in close proximity to the tRNA 5Ј-end, the molecular nature of its catalytic function could not have been interpreted from the structure because of lack of electron density at the apparent site of G Ϫ1 addition. Here, however, kinetic studies imply a specific and unexpected role for His-152 and Lys-187 in the activation step for non-WC G Ϫ1 addition to tRNA His (Figs. 4 -6; Table 1) (19,22,23,28). Although the molecular interactions that enable His-152 and Lys-187 to participate in this specific reaction cannot be discerned by available structural models, these residues are located in close proximity to known adenylylation residues Lys-44 and Asn-161, consistent with this role (Fig. 1B) (19, 23).
Interestingly, the catalytic role for His-152 appears to depend substantially on the aromatic nature of this residue, as judged from the recovery of activity with H152F but not Gln or Lys variants (Figs. 4 -6). These biochemical data could be explained by a stabilizing interaction of His-152 with the incoming non-WC base-paired GTP nucleotide itself, which, interestingly, is analogous to the stacking interactions between the residue and G-nucleotides in the anticodon loop (23). The selective nature of the requirement for His-152/Lys-187 for the non-WC base-paired interaction could be explained by slightly different conformations of the active site when the incoming GTP forms a WC bp with the C 73 nucleotide on this tRNA His variant. If the WC bp is formed, apparently the optimal positioning of GTP for subsequent nucleotidyl transfer is achieved even in the absence of His-152 and Lys-187, and the reaction can proceed. The more significant kinetic effect on the activation step, which occurs prior to use of the incoming GTP (Fig.  1A), also suggests a picture of a preformed active site with both nucleotides (ATP for activation and GTP for nucleotidyl transfer) bound simultaneously.
This work provides further insight into two apparently distinct components of the active site that are involved in discriminating WC versus non-WC bp (17,29). Other members of the Thg1 superfamily known as Thg1-like proteins (TLPs), which are found in all three domains of life, are strictly WC-dependent 3Ј to 5Ј polymerases that do not catalyze efficient non-WC G Ϫ1 addition (21,29). Moreover, TLPs are not generally tRNA Hisselective enzymes that depend on the presence of the GUG anticodon (30 -32). Therefore, it is intriguing that His-152 and Lys-187 are absolutely conserved residues in eukaryotic Thg1 enzymes but do not have obvious counterparts in TLPs. This suggests an evolutionary scenario where both WC-dependent and non-tRNA His -specific activities of the ancestral 3Ј to 5Ј polymerases could have been simultaneously adapted in the earliest eukaryotic ancestor by the acquisition of a relatively limited number of amino acid changes. In this scenario, the emergence of dual function residues such as His-152 and Lys-187 would enable a single alteration to affect both catalytic activity and tRNA specificity of the enzyme.

Site-directed mutagenesis
Mutagenesis was performed using Phusion polymerase (Thermo Fisher) according to the manufacturer's instructions. Reactions contained 50 ng of template DNA encoding hThg1-His 6 and 0.5 M for each primer (T m Ͻ 80°C, 36 -49 nucleotides in length) (16). Following 25 rounds of temperature cycling, DNA was ligated, digested with DpnI (New England Biolabs), and transformed into Escherichia coli XLI-Blue cells. Plasmids from transformed colonies were isolated and sequenced to confirm mutation.

Human Thg1 expression and protein purification
Overexpression and purification by immobilized metal-ion affinity chromatography of C-terminal His 6 -tagged human Thg1 variants was performed as described previously (16). Purified variant and WT Thg1 proteins were dialyzed into 50% glycerol buffer for storage at Ϫ20°C. Purity of hThg1 proteins was qualitatively assessed by SDS-PAGE and protein concentrations were determined using the Bradford reagent (Bio-Rad).

In vitro transcription of tRNA substrates
Run-off transcription and labeling of tRNA His containing either A 73 or C 73 discriminator nucleotide was performed as described previously (16). Briefly, in vitro transcripts were gel purified, extracted with phenol:chloroform, and precipitated with ethanol for storage at Ϫ20°C in TE buffer, pH 7.5. To  (10,20,30,90,120,180, and 240 min). Addition of the unlabeled GTP nucleotide to produce the G Ϫ1 :A 73 bp results in the release of the labeled pyrophosphate from the tRNA substrate, which is visualized by TLC (some of the PPi can be further hydrolyzed to P i during the long time course of the assay). The diagram to the left shows the aminoacyl-acceptor stem of the labeled tRNA His substrate (the rest of the tRNA is omitted for clarity), with the labeled 5Ј triphosphate moiety indicated in orange.

Dual functions for conserved Thg1 residues
generate 5Ј-32 P monophosphorylated tRNA (p*tRNA His ) for phosphatase protection assays, transcribed tRNA was treated with phosphatase (New England Biolabs) followed by T4 polynucleotide kinase in the presence of [␥-32 P]ATP (Perkin Elmer, 6000 Ci/mmol). Labeled 5Ј triphosphorylated tRNA (p*pptRNA His ) used in nucleotidyl transfer experiments was transcribed in the presence of [␥-32 P]GTP (Perkin Elmer, 6000 Ci/mmol). Uniformly labeled tRNA His used in filter binding experiments was in vitro transcribed in the presence of [␣-32 P]GTP (Perkin Elmer, 3000 Ci/mmol). All labeled tRNAs were gel-purified and stored at Ϫ20°C.

Chemical footprinting and primer extension
Chemical footprinting with CMCT was performed using unlabeled in vitro transcribed tRNA His . Modification by CMCT was performed according to Ref. 33. Reactions contained 50 nM tRNA His and 0, 1, or 5 M hThg1 in Buffer B (10 mM potassium borate, pH 7.9, 100 mM KCl, 5 mM MgCl 2 ). Protein and tRNA were allowed to bind at room temperature for ϳ1 h prior to chemical treatment. After binding, 5 l of 25 mg/ml CMCT was added and reactions were placed at 4°C for 2 h. Following incubation, reactions were quenched with 200 l of Buffer B containing 20 g of yeast carrier RNA and precipitated with 1/10 volume of 3 M sodium acetate, pH 5.3, and 2.5 volumes of 100% ethanol. Samples were precipitated again, resuspended in 4 l of water, and stored at Ϫ20°C. Reverse transcription primer (5Ј-TGGTGCCATCTCCTAG) for S. cerevisiae tRNA His was 5Ј-end labeled in a kinase reaction with T4 PNK and [␥-32 P]ATP (6000 Ci/mmol, Perkin Elmer). Labeled primer (1 pmol) and CMCT treated/untreated RNA (2 pmol) were denatured at 95°C for 5 min, then slowly cooled to room temperature to anneal. After annealing, 0.2 mM dNTPs, 5ϫ reaction buffer, and reverse transcriptase (USB/Promega) was added to initiate reactions. Extension was performed for 5 min at 25°C, followed by 37°C for 1-2 h. An equal amount of 2ϫ RNA loading dye (formamide, 0.5 M EDTA, pH 8, 10% bromphenol blue, and 10% xylene cyanol) was added to all samples, which were then resolved on a 10% polyacrylamide/8 M urea gel. Dried gels were visualized using GE Typhoon Trio Imager Scanner (GE Healthcare Life Sciences) and quantified by ImageQuant. To measure normalized CMCT reactivity, primer stop intensity at each position within tRNA treated with CMCT and in the presence and absence of each hThg1 variant was divided by the sum of that intensity and the intensity of fully extended RNA.

Filter-binding assay
Double-filter binding assays were performed as described in (18) with uniformly labeled tRNA His . Briefly, varied concentrations of human Thg1 WT and variants (0.05-10 M) were incubated with limiting amounts of tRNA and bound versus free RNA were separated by filtration through nitrocellulose and Hybond filters, respectively. The fraction of enzyme-bound tRNA was plotted as a function of enzyme concentration and fit to a binding isotherm to yield the apparent K D,tRNAHis (Fig. S2).

Thg1 activity assays with 5-monophosphorylated tRNA His
Phosphatase protection assays were performed as described previously (28). In brief, limiting concentrations of 5Ј p * -tRNA His were incubated with 5-fold dilutions WT and variant hThg1 (from 36 M to 290 nM) in the presence of 1 mM each ATP and GTP at room temperature for 1-2 h, as indicated in each figure. Reactions testing for G Ϫ1 addition were treated with RNase A, whereas U Ϫ1 addition reactions were treated with RNase T1. Nuclease digested reactions were then treated with phosphatase (Invitrogen) producing the indicated oligonucleotide fragments corresponding to nucleotide addition products, or 32 P i from unreacted substrate. A nonenzyme-catalyzed labeled species (*) is observed in these reactions as a consequence of the labeling process and has been indicated on all assay images. All reaction products are resolved by silica TLC using 1-propanol-NH 4 OH-H 2 O (55:35:10, v/v/v) solvent system.
To compare specific activity of each variant enzyme to WT, total percent product formation was quantified at each enzyme concentration, and 1 unit of total activity was set to reflect 30% product formation during the 2-h reaction. The concentration of enzyme required to catalyze 1 unit of activity was calculated and reported as the fraction of WT specific activity for each active enzyme.

Thg1 activity assay with 5 triphosphorylated tRNA His
Nucleotidyl transfer assays were executed as described previously (19). In brief, triphosphorylated p*pptRNA His was incubated in the presence of excess (15 M) hThg1 WT and variant enzymes at room temperature. At various time points (10 -240 min), 2-l aliquots were taken from the reaction and quenched with 0.25 M EDTA and 5 mg/ml RNase A (Ambion) then incubated at 50°C for 10 min. Next, RNase-treated samples were treated with an equal volume of 10% trichloroacetic acid (TCA), incubated on ice for 5 min, and centrifuged for 5 min to pellet precipitated protein and improve clarity of TLC resolution. The supernatant after TCA precipitation was resolved on PEI cellulose TLC plates (EMD Millipore) in a 0.5 M KPO 4 Ϫ :CH 2 OH (80:20, v/v) solvent system. All experiments were imaged using GE Typhoon Trio Imager Scanner and quantified using ImageQuant.
Observed rates (k obs ) were obtained by fitting the data to a single exponential equation (Eq. 1), where P t is total product conversion at a time point, ⌬P is the maximum percent product formed over the time course, and t is time.
Rates for slow hThg1 variants that were not able to achieve maximal product formation during the time of the assays were determined using the method of initial rates (Eq. 2) where v 0 , the slope obtained from percent product plotted against time is divided by the maximum percent of product produced, ⌬P.