Template-dependent 3^′–5^′ nucleotide addition is a shared feature of tRNA^His guanylyltransferase enzymes from multiple domains of life

Abstract

The presence of an additional 5^′ guanosine residue (G_-1) is a unique feature of tRNA^His. G_-1 is incorporated posttranscriptionally in eukarya via an unusual 3^′–5^′ nucleotide addition reaction catalyzed by the tRNA^His guanylyltransferase (Thg1). Yeast Thg1 catalyzes an unexpected second activity: Watson–Crick-dependent 3^′–5^′ nucleotide addition that occurs in the opposite direction to nucleotide addition by all known DNA and RNA polymerases. This discovery led to the hypothesis that there are alternative roles for Thg1 family members that take advantage of this unusual enzymatic activity. Here we show that archaeal homologs of Thg1 catalyze G_-1 addition, in vitro and in vivo in yeast, but only in a templated reaction, i.e. with tRNA^His substrates that contain a C₇₃ discriminator nucleotide. Because tRNA^His from archaea contains C₇₃, these findings are consistent with a physiological function for templated nucleotide addition in archaeal tRNA^His maturation. Moreover, unlike yeast Thg1, archaeal Thg1 enzymes also exhibit a preference for template-dependent U_-1 addition to A₇₃-containing tRNA^His. Taken together, these results demonstrate that Watson–Crick template-dependent 3^′–5^′ nucleotide addition is a shared catalytic activity exhibited by Thg1 family members from multiple domains of life, and therefore, that this unusual reaction may constitute an ancestral activity present in the earliest members of the Thg1 enzyme family.

An additional G-residue (G_-1) at the 5^′ end of tRNA^His is a nearly universal feature, whose loss results in decreased aminoacylation of tRNA^His both in vitro and in vivo, as observed in organisms from E. coli to yeast (1–6). The only organisms known to lack G_-1 on tRNA^His are 20 α-proteobacteria (7) in which there is a simultaneous alteration in the histidyl-tRNA synthetase (HisRS) that is suggested to compensate for loss of this universal tRNA feature (8, 9).

Although the presence of G_-1 on tRNA^His is highly conserved, the mechanism by which G_-1 is incorporated into tRNA varies in different domains of life. In bacteria, G_-1 is genomically encoded and present in the precursor-tRNA transcript (10). The presence of G_-1 and a universally conserved C₇₃ discriminator nucleotide results in formation of an unusual 8-base pair acceptor stem in bacterial tRNA^His that directs the tRNA 5^′-end maturation enzyme, RNase P, to cleave at an alternate site, leaving G_-1 in the mature tRNA (Fig. 1A) (11). In eukarya, the presence of G_-1 is the result of a posttranscriptional nucleotide addition reaction, catalyzed in yeast by the essential enzyme tRNA^His guanylyltransferase (Thg1) (Fig. 1B) (12, 13). The widespread conservation of Thg1 throughout eukarya, including humans, suggests a similar mode of G_-1 addition.

Fig. 1.

Domain-dependent incorporation of G_-1 into tRNA^His. (A) In bacteria, G_-1 is genomically encoded and retained by RNase P during 5^′-processing. (B) Thg1 adds G_-1 to eukaryal tRNA^His, and catalyzes template-dependent 3^′–5^′ addition to tRNA^His variant substrates. Nucleotides added posttranscriptionally by Thg1 are indicated by parentheses. For clarity, only the altered 3^′ end sequences of tRNA^His variant substrates are shown.

Thg1 adds G_-1 by an unusual 3^′–5^′ nucleotide addition reaction, via an unknown molecular mechanism that can not be predicted, due to the lack of sequence similarity between Thg1 and any known enzyme family. In yeast, G_-1 is added opposite an A₇₃ discriminator nucleotide that is universally conserved throughout eukarya (Fig. 1B) (14). However, in vitro, wild-type yeast Thg1 exhibits a second biochemical activity with variant tRNA^His substrates that contain alterations at A₇₃. Use of these substrates revealed that wild-type yeast Thg1 adds multiple G or C residues to the 5^′ end of the tRNA in a reaction that uses the 3^′ acceptor stem nucleotides as a template for 3^′–5^′ addition (Fig  1B) (15). Although nontemplated (G opposite A) nucleotide addition plays a well-documented physiological role in the maturation of tRNA^His, the biological significance of templated 3^′–5^′ nucleotide polymerization, in yeast or any other organism, has not been demonstrated. Yet the ability of yeast Thg1 to recognize and add Watson–Crick base-paired nucleotides was completely unexpected for an enzyme that seemingly only functions to add G_-1 opposite A₇₃ in tRNA^His. Moreover, the ability of yeast Thg1 to add deoxynucleotides as well as ribonucleotides to tRNA substrates, as well as the observation of templated 3^′–5^′ addition with other tRNA substrates in vitro suggest that alternative roles for templated addition are a possibility (15).

Both bacteria-like and eukarya-like systems for incorporation of G_-1 into tRNA^His appear to be present in archaea. Of 55 annotated archaeal tRNA^His genes (16), 40 contain a genomically encoded G_-1 residue and a C₇₃ discriminator nucleotide, suggesting that G_-1 could be incorporated into tRNA^His using the RNase P-dependent pathway. However, the remaining 15 annotated archaeal tRNA^His genes lack a G residue at the -1 position in their coding sequence. Notably, homologs to yeast Thg1 are present in all of these species with the exception of N. equitans (17), consistent with a requirement for enzymatic G_-1 addition in these organisms (Fig. 2). Intriguingly, among the 40 archaea that already contain a genomically encoded G_-1 residue, 7 of these also contain a Thg1 homolog (Fig. 2), thus these organisms may use either, or both, pathways for addition of G_-1 to tRNA^His.

Fig. 2.

Phylogeny of archaeal Thg1 homologs. An unrooted phylogenetic tree of the archaeal Thg1 homologs was constructed with Geneious Pro software (v.4.7.5) using the neighbor-joining method with a Jukes–Cantor genetic distance model and no outgroup. The tree was built upon a ClustalW alignment (36) of Thg1 sequences from 20 of 21 Thg1-containing archaea (P. aerophilum Thg1 is similar to the other three Pyrobaculum homologs and was omitted for clarity). Organism names are colored according to the presence of G_-1 on the annotated tRNA^His; blue, G_-1 is genomically encoded; red, G_-1 is not present in the genomic sequence. The four archaea investigated in this work are indicated by boxes. The dashed line indicates the separation between crenarchaeal (Upper Left) and euryarchaeal (Lower Right) clades.

The discriminator nucleotide in archaeal tRNA^His is universally a C₇₃, as it is in bacteria (14), raising the possibility that enzymatic G_-1 addition in archaea could require a templated 3^′–5^′ nucleotide addition reaction similar to that previously associated with yeast Thg1 reverse (3^′–5^′) polymerization activity. Although templated 3^′–5^′ nucleotide addition activity appears to be required for an unusual form of 5^′-tRNA editing in certain protozoan mitochondria (18, 19), the enzyme(s) that catalyze this reaction remain to be identified. Thus, archaeal Thg1 has the potential to be a unique example of a purified enzyme that catalyzes templated 3^′–5^′ nucleotide addition with a demonstrated biological relevance.

In this study, we investigated the biochemical function of Thg1 homologs from four archaea. The results show that both yeast and archaeal Thg1 enzymes catalyze templated 3^′–5^′ nucleotide addition, therefore this property appears to be a shared feature of Thg1 enzymes from multiple domains of life. In contrast, the nontemplated reaction catalyzed by eukaryal Thg1 during maturation of tRNA^His may be a derived evolutionary trait unique to eukarya.

Results

Archaeal Thg1 Homologs Do Not Efficiently Add G_-1 to A₇₃-Containing tRNA^His.

To test the activities of archaeal Thg1 enzymes, we cloned, expressed, and purified homologs from M. thermoautotrophicus (MtThg1), M. kandleri (MkThg1), M. acetivorans (MaThg1), and M. barkeri (MbThg1) (see SI Text). These four archaea include two that are predicted to require enzymatic addition of G_-1 to tRNA^His (Mt and Mk) and two in which G_-1 is already encoded in the genome (Ma and Mb) (Fig. 2). The MaThg1 coding sequence is interrupted by an in-frame amber stop codon, and appears to be a previously undocumented case of nonsense suppression by insertion of pyrrolysine at this position, as is seen with other proteins from methanobacteria, although the presence of pyrrolysine is not required for activity (see SI Text) (20).

We first tested the ability of each purified archaeal Thg1 to add G_-1 to yeast tRNA^His that contains the universally conserved eukaryal A₇₃ discriminator nucleotide. G_-1 addition activity was tested using a previously described assay with 5^′-³²P labeled monophosphorylated tRNA^His (p-tRNA^His) (21). Briefly, following incubation of purified Thg1 with the labeled tRNA, the reactions are treated with RNase A, followed by phosphatase, finally yielding inorganic phosphate () from unreacted substrate, and a labeled G_-1p^∗GpC trimeric oligonucleotide if the G_-1 addition product is formed during the reaction (Fig. 3A). No products were observed in reactions with MtThg1 and this A₇₃-containing tRNA, whereas reaction products were observed with the other three archaeal enzymes (MkThg1, MaThg1, and MbThg1) (Fig. 3A; compare lane 2 with lanes 3–5). However, the primary reaction product of these enzymes’ activity migrates slightly lower in the TLC solvent system used for this assay than the G_-1-containing product produced by yeast Thg1. This slower migrating product corresponds to the adenylylated-tRNA^His reaction intermediate (App^∗G₊₁pC) produced by yeast Thg1 (13). The major reaction product observed with all of the archaeal enzymes (a representative analysis with MaThg1 is shown) is only dependent on addition of ATP (Fig. 3B), and migrates identically to the adenylylated tRNA^His species produced by yeast Thg1, as demonstrated by nuclease analysis of isolated reaction products (13).

Fig. 3.

Archaeal Thg1 homologs do not efficiently add G_-1 to wild-type yeast tRNA^His. (A) Assays using 5^′-³²P-labeled tRNA^His and 1 µL each Thg1, as indicated, were performed with 0.1 mM ATP and 1.0 mM GTP as described in Methods. Lane-, buffer only. Reaction products are labeled to the right of each panel, and indicated by arrows. (B) 5^′-³²P-tRNA^Hisassay performed with 1 µL yeast or MaThg1. The presence or absence of ATP or GTP (1 mM) is denoted by (+) and (-). (C) Assays of 5^′-³²P-tRNA^His with titrations (5-fold serial dilutions, starting with approximately ~50 M enzyme) of purified yeast or MaThg1; Lane-, buffer only control reaction. (D) Scheme of steps for G_-1 addition to tRNA^His showing first, adenylylation of p-tRNA^His, followed by nucleotidyltransfer to yield G_-1-containing tRNA. The apparent block in the 2nd step of the reaction observed with archaeal Thg1 is as indicated.

Adenylylated-tRNA^His is the terminal product of archaeal Thg1 activity with A₇₃-tRNA^His. Measurements of total product formation as a function of decreasing Thg1 concentration reveal similar amounts of App-tRNA^His and G_-1-tRNA^His formed at the same low concentrations (0.4 µM) of MaThg1 or yeast Thg1, respectively (Fig. 3C), indicating that the overall activity of MaThg1 is not significantly lower than the yeast enzyme. Instead, the persistent accumulation of the App-tRNA^His intermediate even at the highest concentrations tested (46 µM) suggests that archaeal Thg1 enzymes can not catalyze the second, guanylyltransfer step of the three-step Thg1 reaction with A₇₃-tRNA^His (Fig. 3D) (13).

Archaeal Thg1homologs Efficiently Add G_-1 to C₇₃-Containing tRNA^His In Vitro.

We next tested the ability of archaeal Thg1 to add G_-1 to a yeast tRNA^His variant containing C₇₃ instead of A₇₃; this tRNA is a substrate for template-dependent 3^′–5^′ polymerization by yeast Thg1 in vitro (Fig. 1B). In contrast to the inefficient addition of G_-1 to A₇₃-containing tRNA described above, all four archaeal Thg1 enzymes catalyzed efficient addition of G_-1 to this C₇₃-containing tRNA substrate (Fig. 4). Whereas yeast and MkThg1 catalyze the addition of multiple G-residues (G_-1, G_-2, and G_-3), the single G_-1 addition product is the predominant product observed with Ma, Mb, and MtThg1, and G_-2/G_-3 are added inefficiently, if at all (Fig. 4). No accumulation of adenylylated tRNA intermediate was observed with any of the archaeal Thg1 enzymes and C₇₃-tRNA^His.

Fig. 4.

Archaeal Thg1 homologs efficiently catalyze G-addition opposite C₇₃ with a tRNA^His variant substrate. Assay of purified yeast and archaeal Thg1 homologs (1 µL each) with 5^′-³²P labeled C₇₃-tRNA^His variant. Lane-, buffer only. Positions of G_-1, G_-2, and G_-3 reaction products are indicated to the right of the figure.

To quantify the relative catalytic abilities of archaeal Thg1 enzymes, we measured the steady-state kinetic parameters for addition of G_-1 to C₇₃-containing tRNA^His catalyzed by MaThg1 and MtThg1, as representative enzymes from archaea that contain or lack, respectively, a genomically encoded G_-1 residue. For comparison, kinetic parameters for the same reaction with yeast Thg1 were also determined. Steady-state kinetic assays were performed with triphosphorylated (ppp - tRNA^His) substrates as previously described (21).

With yeast Thg1, k_cat/K_M for C₇₃-tRNA^His is essentially the same as for wild-type (A₇₃) tRNA^His (Table 1, SI Text). Despite the relatively slow k_cat exhibited by MaThg1 for C₇₃-tRNA^His, the significantly lower K_M results in only a 10-fold decreased k_cat/K_M for MaThg1 compared to the yeast enzyme (Table 1, SI Text). Initial determination of MtThg1 kinetic parameters similarly performed at 25 °C indicated very low activity, with a significantly decreased k_cat (SI Text). However, because M. thermoautotrophicum is a thermophile that grows optimally at 65 °C (22), MtThg1 activity was measured at 55 °C, where the enzyme exhibits maximal activity in the in vitro assay (SI Text). At this temperature, initial rates for MtThg1 activity at 2 µM and 4 µM tRNA were essentially unchanged, consistent with the K_M measured at 25 °C (SI Text). Thus, the k_cat/K_M calculated for MtThg1 under optimal temperature conditions is within 20-fold of that exhibited for yeast Thg1 (Table 1).

Table 1.

Steady-state kinetic parameters for G_-1 addition to pppC₇₃-tRNA^His.

Substrate	Enzyme	kcat (hr-1)	KM (µM)	kcat/KM (M-1 s-1)	%kcat/KM (relative to yeast Thg1)
A73-tRNAHis	yeast Thg1	8.4*± 0.9	0.42*± 0.13	5500*± 1200	—
C73-tRNAHis	yeast Thg1	20.4 ± 2.4	0.99 ± 0.29	5670 ± 1200	100
C73-tRNAHis	MaThg1	0.30 ± 0.03	0.12 ± 0.05	700 ± 230	12
C73-tRNAHis	MtThg1	1.1†	1.1 ±0.85‡	300	5

^*Values taken from ref (21), errors are ± SEM.

^†k_cat taken from the [tRNA]-independent v_o/[E] at the peak temperature for MtThg1 activity (55 °C) (SI Text).

^‡K_M measured at 25 °C (SI Text).

Archaeal Thg1 Homologs Complement the Essential Function of Yeast Thg1 in Yeast Cells When Provided with an Appropriate tRNA Substrate for Templated 3^′–5^′ Addition.

To test in vivo function of the archaeal enzymes, we performed a variation of a previously described plasmid shuffle assay (13, 23). Archaeal Thg1 homologs (CEN LEU2) were introduced into the yeast thg1Δ strain JJY20 [relevant genotype: thg1Δ leu2Δ ura3Δ (CEN URA3 PTHG1-THG1)], and subsequently grown on 5-fluoroorotic acid (FOA)-containing media that selects against the yeast THG1 URA3 covering plasmid, resulting in a strain in which the archaeal Thg1 enzymes are the only potential source of Thg1 activity in the cells. Consistent with the lack of G_-1 addition activity exhibited by archaeal Thg1 in vitro with wild-type yeast tRNA^His (Fig. 5), each archaeal Thg1-expressing strain was unable to grow on FOA (Fig. 5, compare growth of strains spotted in the first column for each of the three media shown), indicating that these archaeal Thg1 enzymes are not able to replace the essential function of yeast Thg1 in vivo in yeast.

Fig. 5.

Archaeal Thg1 homologs support growth of the yeast thg1Δ strain in the presence of C₇₃-tRNA^His. Replica plating results with yeast thg1Δ strains transformed with LEU2 plasmids containing various Thg1 homologs (indicated to the right of the figure) and HIS3 plasmids containing various tRNA^His genes (indicated on top of each panel). Growth media are indicated below each panel; images were taken after 3–4 days of growth at 30 °C.

We then tested whether the ability of the archaeal homologs to add G_-1 opposite C₇₃ could be exploited to support growth of the yeast deletion mutant. To the thg1Δ strains created above, we simultaneously introduced a third plasmid, containing either an additional copy of wild-type (A₇₃-containing) or variant (C₇₃-containing) tRNA^His, and again tested the ability of the resulting strains to grow on media containing FOA (Fig. 5, second panel). In each strain containing archaeal Thg1, significant growth was observed only in strains that also expressed C₇₃-tRNA^His, and not in those expressing wild-type tRNA^His or a vector control. The ability to complement the thg1Δ growth phenotype was dependent on expression of archaeal Thg1, because the complementation was only observed on galactose-containing media (inducing conditions for expression from the P_GAL thg1 promoter) and not in dextrose-containing media (repressing conditions) (Fig. 5, compare second and third panels). Because the presence of G_-1 is required for histidylation and thus growth in yeast (4), these results indicate that archaeal Thg1 homologs add G_-1 to C₇₃-containing tRNA^His in vivo. The complementation observed in the presence of C₇₃-tRNA^His also demonstrates that the lack of growth observed in strains containing only wild-type (A₇₃) tRNA^His is not due to limiting expression of the archaeal enzymes in yeast.

Archaeal Thg1 Homologs Catalyze Template-Dependent 3^′–5^′ Addition with Wild-Type tRNA^His.

In light of the strong preference for addition of G_-1 opposite C₇₃ exhibited by the archaeal Thg1 enzymes, we revisited the lack of G_-1 addition previously observed with the wild-type (A₇₃-containing) tRNA^His substrate. We tested the ability of archaeal Thg1 homologs to add U_-1 to wild-type tRNA^His, to test the hypothesis that these enzymes might require the ability to form a Watson–Crick base pair for nucleotide addition to occur. Unlike G_-1 addition to A₇₃-yeast tRNA^His that is catalyzed by yeast Thg1 but not the archaeal homologs, the addition of U_-1 opposite A₇₃ is observed with both yeast and archaeal enzymes (Fig. 6A, representative assay for MaThg1 activity is shown). Therefore, the inability of the archaeal enzymes to add G_-1 to wild-type tRNA^His does not reflect an inherent inability to recognize and use this tRNA substrate, but instead reflects an apparent preference of archaeal Thg1 for template-dependent 3^′–5^′ addition.

Fig. 6.

Archaeal Thg1 catalyzes template-dependent U_-1 addition activity with wild-type tRNA^His. (A) U_-1 addition assay with 5^′-³²P labeled tRNA^His and titrations (5-fold serial dilutions) of yeast or MaThg1. Lane-, buffer only. Identity of observed reaction products indicated to the left of the panel. (B) Initial rates of N_-1 addition to wild-type tRNA^His (2 µM) with yeast Thg1 (0.2 µM) (solid bars) or MaThg1 (0.15 µM) (hatched bars), and 1 mM of each NTP, as indicated. Note separate scale on Y-axis for each enzyme's activity.

To further probe the template-dependent nature of the reaction catalyzed by MaThg1 with yeast tRNA^His, steady-state initial rates of 3^′–5^′ nucleotide addition were measured using each of the four possible NTPs and ppp-tRNA^His with yeast and MaThg1. For yeast Thg1, it was previously demonstrated that any nucleotide can be added to the 5^′ end of wild-type tRNA^His after prolonged incubation at high concentrations of Thg1 (15), although, consistent with the physiological role of Thg1 in yeast, the rate of G_-1 addition under steady-state conditions is significantly faster than the rate of addition of any other N_-1 residue (Fig. 6B). In contrast, for MaThg1 the NTP preference was essentially reversed, with UTP added most efficiently to the A₇₃-containing tRNA^His substrate, consistent with a requirement for formation of a Watson–Crick base pair for catalysis of 3^′–5^′ nucleotide addition (Fig. 6B). Thus, template-dependent 3^′–5^′ addition activity of archaeal Thg1 is not limited to formation of G-C base pairs. The rate of addition of U_-1(v_o/[E] = 0.04 hr^-1) is nearly 10-fold slower than the k_cat for G_-1 addition to C₇₃-tRNA^His by MaThg1 (Table 1), suggesting an additional effect of the identity of the base pair being formed on templated activity with tRNA^His substrates. The slower rate of MaThg1-catalyzed U-A vs. G-C addition likely contributes to the lack of complementation of the yeast thg1Δ phenotype observed by archaeal Thg1 homologs in the presence of A₇₃-tRNA^His (Fig. 5).

Discussion

In this study, we demonstrated that archaeal homologs of yeast Thg1 catalyze solely template-dependent 3^′–5^′ nucleotide addition to tRNA^His substrates, and do not catalyze the nontemplated G opposite A addition reaction that is the hallmark reaction of eukaryal Thg1 enzymes. Purified Thg1 homologs from four different archaea do not efficiently add G_-1 to wild-type yeast tRNA^His (Fig. 3), but instead add G_-1 to a tRNA^His variant that contains C₇₃ and is thus a substrate for template-dependent G_-1 addition (Fig. 4). Moreover, archaeal Thg1 homologs substitute for the essential G_-1 addition function of yeast Thg1 in vivo, as long as they are provided with the C₇₃-containing tRNA^His substrate that reflects the preference of these enzymes to catalyze templated 3^′–5^′ nucleotide addition (Fig. 5). 3^′–5^′ addition catalyzed by archaeal Thg1 is not restricted to forming G-C base pairs, but also forms U-A base pairs, as evidenced by the preference for MaThg1 to efficiently add U_-1 over other non-Watson–Crick pairing N_-1 residues to wild-type (A₇₃-containing) tRNA^His, thus distinguishing these enzymes from yeast Thg1 that prefers to add the nontemplated G_-1 to A₇₃-tRNA^His (Fig. 6). However, because yeast Thg1 is also capable of a previously described template-dependent 3^′–5^′ addition activity (15), these results demonstrate that the ability to catalyze Watson–Crick base pair-dependent nucleotide addition is a universal property of Thg1 family members from multiple domains of life. In contrast, the ability to add G_-1 opposite A₇₃ with eukaryal tRNA^His is so far a unique feature of eukaryal Thg1 enzymes.

Biological Function for Template-Dependent Thg1 Addition Activity.

Although template-dependent 3^′–5^′ nucleotide addition was originally discovered as a property of yeast Thg1, the physiological relevance of this activity in yeast remains unknown (15). Nonetheless, unexplained interactions of yeast Thg1 with the Orc2 component of the origin recognition complex (24) and cell-cycle progression defects associated with decreased Thg1 expression in yeast and human cells (24, 25) suggest that the role of the enzyme in eukarya may be more complicated than is currently understood. However, this work provides an opportunity to explore possible physiological roles for template-directed nucleotide addition activity in other organisms. Certain archaea (represented here by M. thermoautotrophicus and M. kandleri) lack a genomically encoded G-residue at the -1 position of the tRNA^His gene (Fig. 2), and thus can not incorporate G_-1 via an RNase P-dependent pathway like the one used in bacteria. Because the discriminator nucleotide in archaeal tRNA^His is universally a C₇₃, our results are consistent with a role for archaeal Thg1 in this G_-1 addition reaction, thus this work provides unique evidence for a likely physiological role for Thg1-catalyzed template-dependent 3^′–5^′ nucleotide addition in any organism.

We note that, whereas the measured k_cat values for archaeal Thg1 activity are quite low (0.3–1.1 hr^-1) (Table 1), the ability of archaeal enzymes to support growth of the yeast thg1Δ strain in the presence of C₇₃-tRNA^His (Fig. 5) demonstrates that these enzymes catalyze G_-1 addition on a biologically relevant timescale. Because the tRNA substrates used here are derived from yeast tRNA^His sequences, it is possible that the in vitro activities of the archaeal enzymes could be improved using tRNA^His substrates derived from archaea. However, because the k_cat/K_M values measured for yeast Thg1 activity are similarly low (Table 1), it is possible that the measured in vitro activities of both yeast and archaeal enzymes do not accurately reflect the in vivo rates. Such a discrepancy could be caused by the use of nonoptimal in vitro reaction conditions, lack of important tRNA modifications on the in vitro transcribed substrates, or lack of additional components that stimulate Thg1 activity in vivo. While this remains a possibility, investigations of alternative assay conditions have not yet yielded increased rates with the yeast enzyme, and Thg1 purified from E. coli in the absence of any other yeast proteins is catalytically active and not stimulated by inclusion of yeast crude extracts in the assay.

A role for archaeal Thg1 homologs in G_-1 addition to C₇₃-containing tRNA^His in archaea accounts for the presence of the enzyme in the majority of archaea in which it has been identified, including M. thermoautotrophicus and M. kandleri. However, a second group of archaea, including M. barkeri and M. acetivorans, is particularly intriguing in that, whereas we have shown here that enzymes from these species catalyze G_-1 addition in vitro and in vivo, these organisms also contain a genomically encoded G_-1 residue in tRNA^His (Fig. 2). This raises the possibility that, in some archaea, G_-1 addition could occur by either or both of the two known G_-1 incorporation pathways. It is also possible that the presence of Thg1 in these organisms is required for another activity that is independent of a role for the enzyme in maturation of tRNA^His.

Alternative Biological Roles for Template-Dependent 3^′–5^′ Nucleotide Addition.

Possible biological roles for templated nucleotide addition activity in the 3^′–5^′ direction, opposite to that of all known DNA and RNA polymerases, are not limited to tRNA^His maturation. For example, templated 3^′–5^′ addition is required for a tRNA editing activity that functions in the mitochondria of certain lower eukarya, where it is used to repair up to three mismatches found in many of the mitochondrial tRNA genes (18, 19). The editing reaction requires stepwise 3^′–5^′ addition, using the nucleotides of the 3^′ half of the aminoacyl-acceptor stem as a template (19, 26–28). Similarities between the enzymatic activities required for G_-1 addition, now known to be catalyzed by Thg1, and 5^′-tRNA editing, for which the identity of the enzyme(s) involved remains a mystery, were noted at the time of the discovery of the editing activity (18, 19). It is tempting to speculate that Thg1 enzymes from archaea that do not require enzymatic addition of G_-1 to tRNA^His may participate in a similar tRNA editing or repair activity in their respective organisms, which would necessarily take advantage of their ability to catalyze template-dependent 3^′–5^′ addition. Although 5^′-editing or repair of tRNA has not been identified in archaea, the existence of such activities may mirror tRNA 3^′-end surveillance mechanisms that are found in multiple domains of life (29–31). The example of the ubiquitous 3^′-CCA addition enzyme is particularly intriguing in this respect, because it catalyzes the essential de novo addition of the 3^′-CCA end to the majority of eukaryal tRNAs, but in bacteria and archaea, predominantly functions in 3^′-end repair pathways, because the 3^′-CCA sequence is normally encoded in the genome (32, 33).

Implications for Thg1 3^′–5^′ Nucleotide Addition Mechanism.

As the only known enzyme family that catalyzes nucleotide addition in the opposite direction to all known DNA and RNA polymerases, the Thg1 molecular mechanism is of great interest, but is currently unknown. The results presented here add an additional dimension to this issue, raising the question of how enzymes from archaea and yeast exhibit distinct biochemical properties despite sharing a significant degree of overall sequence similarity (even the most distantly related archaeal enzymes share approximately 30% identity with yeast Thg1), including many residues that play critical roles in yeast Thg1 activity (23). A second issue to be explained is the apparently inefficient addition of multiple G nucleotides to C₇₃-tRNA^His observed with most archaeal Thg1 enzymes. Although consistent with the presumed biological need for only a single G_-1 residue at the 5^′ end of tRNA^His, this suggests the existence of additional mechanism(s) used by some archaeal Thg1 enzymes to limit the addition of multiple nucleotides to the 5^′ end of tRNA^His. It remains to be seen whether multiple nucleotide additions analogous to yeast reverse polymerase activity are catalyzed by these archaeal Thg1 homologs with other substrates.

Implications for Evolution of 3^′–5^′ Addition Activities.

Despite the fact that eukaryal G_-1 addition required for tRNA^His maturation was the first Thg1 activity to be described, these results indicate that this is not likely the ancestral activity of Thg1 family members. Instead, template-dependent nucleotide addition is catalyzed by diverse Thg1 family members from multiple domains of life, and therefore these results suggest a scenario in which template-dependent 3^′–5^′ addition was a property of the earliest Thg1 family members. Further investigation of Thg1 homologs from other organisms, including > 20 bacteria in which Thg1 homologs can now be found, will be important for understanding the evolution of 3^′–5^′ nucleotide addition activities. Interestingly, Thg1-containing bacteria all contain a genomically encoded G_-1 residue, suggesting that they are similar to the 7 archaea that do not require enzymatic G_-1 addition to tRNA^His (Fig. 2), and therefore the function of bacterial Thg1 may be the same as in these archaea. Particularly, it will be interesting to determine the evolutionary basis for the apparent acquisition of the ability to catalyze nontemplated G_-1 addition opposite A₇₃ by eukaryal Thg1 family members.

Methods

Expression and Purification of Thg1 Homologs.

Cloned archaeal Thg1 enzymes were expressed in and affinity-purified from E. coli, essentially as previously described for yeast Thg1. For detailed descriptions of cloning, expression and purification protocols, see SI Text.

3^′–5^′ Nucleotide Addition Assays with p-tRNA Substrates.

Assays contained approximately 1 nM 5^′-³²P-tRNA (specific activity 6,000 Ci/mmol, prepared as described in SI Text) in 25 mM Hepes, 10 mM MgCl₂, 3 mM DTT, 125 mM NaCl, 0.2 mg/mL BSA, pH 7.5 buffer, 0.1 mM ATP and 1 mM GTP, unless otherwise indicated, according to the previously published protocol (21). Reactions were initiated by addition of either 1 µl enzyme (final concentration 0.1–1 μL depending on concentration of the purified stock), or 5-fold serial dilutions of enzyme starting with the same purified stock, as indicated, and were performed for 2 h at 25 °C.

U_-1 addition activity was tested using a variation of the previously described nuclease protection assay with 5^′-³²P-labeled tRNA^His. Assays were performed identically to those described above, except that reactions were treated with RNase T1 (Ambion) in place of RNase A, to release a labeled U_-1p ∗ G₊₁ dimer product if U_-1 is added, but still yield labeled phosphate if no nucleotide has been added. The identity of the Up*G reaction product was confirmed by further subjecting the reaction products to RNase T2 digestion that released Up* after reaction of the tRNA with Thg1.

Steady-State Kinetic Parameters for Thg1 Activity.

Triphosphorylated ³²PppC₇₃-tRNA^His (SI Text) was assayed as described previously (21) in Thg1 assay buffer, in the presence of 1 mM GTP at room temperature, unless otherwise indicated. Steady-state reactions contained 0.05–8 µM C₇₃-tRNA^His and 1–500 nM of each purified Thg1 such that at least 5-fold excess tRNA was maintained in each reaction. Time courses measured linear initial rates (< 10%) of the release of labeled ³²PP_i due to G_-1 addition to the ³²Ppp-tRNA for each [tRNA]. Steady-state kinetic parameters k_cat, K_M, and k_cat/K_M for each enzyme were determined by a fit of the initial velocities obtained as a function of [C₇₃-tRNA^His] to the Michaelis–Menten equation using KaleidaGraph (Synergy). Initial rates of N_-1 addition to A₇₃-tRNA^His (Fig. 6) were measured in the same way, with ppp-tRNA^His substrate.

Analysis of In Vivo Thg1 Function of Archaeal Thg1 Homologs.

Plasmids for galactose-inducible expression of archaeal Thg1 [CEN LEU2 P_GAL - thg1^∗] (SI Text), along with analogous yeast THG1-containing and empty vector control plasmids, were transformed into yeast strain JJY20 [relevant genotype MATα thg1Δ::kanMX his3–1 leu2Δ met15Δ ura3Δ (CEN URA3 P_THG1 - THGI)] (34). Positive (Leu⁺) transformants were selected at 30 °C on synthetic dextrose (SD) media (35), and tested by replica plating at 30 °C to synthetic galactose (SGal) media containing FOA, to induce expression of the various Thg1 homologs and select against the wild-type yeast THG1 URA3 plasmid. To test the tRNA requirement for Thg1 function, CEN HIS3 plasmids containing either wild-type (A₇₃)-tRNA^His, C₇₃-tRNA^His, or no tRNA were transformed into each of the Leu⁺ strains using the same method. Positive transformants were selected on SD media lacking leucine and histidine, and further subjected to replica plating as described above.