Human Thg1 displays tRNA-inducible GTPase activity.

tRNAHis guanylyltransferase (Thg1) catalyzes the 3'-5' incorporation of guanosine into position -1 (G-1) of tRNAHis. G-1 is unique to tRNAHis and is crucial for recognition by histidyl-tRNA synthetase (HisRS). Yeast Thg1 requires ATP for G-1 addition to tRNAHis opposite A73, whereas archaeal Thg1 requires either ATP or GTP for G-1 addition to tRNAHis opposite C73. Paradoxically, human Thg1 (HsThg1) can add G-1 to tRNAsHis with A73 (cytoplasmic) and C73 (mitochondrial). As N73 is immediately followed by a CCA end (positions 74-76), how HsThg1 prevents successive 3'-5' incorporation of G-1/G-2/G-3 into mitochondrial tRNAHis (tRNAmHis) through a template-dependent mechanism remains a puzzle. We showed herein that mature native human tRNAmHis indeed contains only G-1. ATP was absolutely required for G-1 addition to tRNAmHis by HsThg1. Although HsThg1 could incorporate more than one GTP into tRNAmHisin vitro, a single-GTP incorporation prevailed when the relative GTP level was low. Surprisingly, HsThg1 possessed a tRNA-inducible GTPase activity, which could be inhibited by ATP. Similar activity was found in other high-eukaryotic dual-functional Thg1 enzymes, but not in yeast Thg1. This study suggests that HsThg1 may downregulate the level of GTP through its GTPase activity to prevent multiple-GTP incorporation into tRNAmHis.

INTRODUCTION

Two distinct pathways for acquisition of G-1 to tRNAHis

Nuclear tRNA genes are transcribed by RNA polymerase III as nascent precursor tRNAs (1,2), which then go through a series of modifications to yield mature tRNAs. A unique modification, 3′-5′ addition of a guanosine at position −1 (G-1), has been shown to occur at the 5′ end of tRNAHis. Note that position −1 is normally unoccupied in mature tRNAs. In bacteria, G-1 is encoded by the tRNAHis genes opposite C73 and retained after removal of the 5′ leader sequence by ribonuclease P (3). An exception to this rule is a small group of α-proteobacteria, whose tRNAsHis lack a G-1 residue (4). In eukaryotes, however, G-1 is added posttranscriptionally by tRNAHis guanylyltransferase (Thg1) after removal of the 5′ leader sequence (5). The presence of two distinct pathways to ensure the acquisition of G-1 highlights the key role of G-1 for tRNAHis. G-1 is a unique distinguishing feature of tRNAHis throughout biology (6–9) and is essential for aminoacylation of tRNAHis by histidyl-tRNA synthetase (HisRS) (9–11). Most archaea acquire G-1 through a pathway similar to that for bacteria (12), whereas some archaea acquire G-1 through a posttranscriptional pathway (13). Paradoxically, both pathways appear to take place in some organellar tRNAsHis (14–16).

tRNAHis recognition by Thg1

Unlike HisRS, which identifies its cognate tRNA primarily through recognition of G-1, Thg1 identifies tRNAHis through recognition of its anticodon (GUG). Despite the fact that both eukaryal- and archaeal-type Thg1 enzymes recognize the anticodon, the former prefers tRNAHis with A73, whereas the latter prefers tRNAHis with C73. Eukaryal-type Thg1, such as yeast Thg1, adds G-1 to tRNAHis across from A73 via a template-independent mechanism, producing a G:A mismatch (5). In fact, this enzyme can add any nucleotide to tRNAHis opposite A73 in vitro, albeit with a strong preference for GTP (17). In contrast, archaeal-type Thg1, such as M. thermoautotrophicus and M. kandleri Thg1 enzymes, adds G-1 to tRNAHis across from C73 through a template-dependent mechanism, producing an extra Watson-Crick base pair in the acceptor stem (12,18). Many bacterial, archaeal, and certain eukaryotic microbes, such as Dictyostelium discoideum, possess a Thg1-like protein, which is not specific to tRNAHis and is involved in tRNA 5′ end repair instead (13,19).

Mechanisms of G-1 addition in eukaryotes and archaea

In yeast, Thg1-catalyzed G-1 addition to ribonuclease P-processed tRNAHis, 5′-monophosphorylated tRNAHis (p-tRNAHis), involves three consecutive steps: (i) adenylylation, in which p-tRNAHis is activated by ATP, forming an activated intermediate, App-tRNAHis; (ii) guanylylation, in which GTP replaces the 5′ AMP, producing 5′-triphosphorylated G-1-tRNAHis intermediate (ppp-G-1-tRNAHis) and (iii) pyrophosphate removal, in which the PPi group is removed from the 5′ end of ppp-G-1-tRNAHis, yielding a mature tRNAHis, p-G-1-tRNAHis (17) (Figure 1A). In vitro, ppp-tRNAHis can be used as the substrate for G-1 addition, bypassing the requirement for the activation step (13). In contrast, ATP is not absolutely required for archaeal Thg1, as GTP can also be used to activate p-tRNAHis (Figure 1A) (13,20).

Figure 1.

Incorporation of G-1 into tRNAHis. (A) Mechanisms of Thg1-catalyzed G-1 incorporation. Yeast Thg1 adds G-1 to tRNAHis opposite A73 through an ATP-activation mechanism, whereas archaeal Thg1 adds G-1 to tRNAHis opposite C73 through an ATP(GTP)-activation mechanism. RTP represents ATP or GTP. (B) Cloverleaf structures of human cytoplasmic and mitochondrial tRNAHis isoacceptors. (C) Primer extension analysis of in vivo-isolated human mitochondrial tRNAHis. Aminoacylation of tRNAmHis by human (D) mitochondrial and (E) cytoplasmic HisRSs. tRNA substrates used included native tRNAmHis and in vitro-transcribed tRNAmHis with or without G-1.

Each Thg1 subunit possesses two nucleotide-binding pockets

To date, Thg1 is the only enzyme known to catalyze templated reverse polymerization. Its catalytic domain shares unexpected structural similarity with canonical 5′-3′ DNA polymerases (13,21). Each Thg1 subunit consists of a palm domain and a finger domain. The former is responsible for catalysis, whereas the latter is responsible for binding the anticodon (20,22). Human Thg1 exists as a homotetramer (subunits ABCD), but appears as a dimer of dimers (dimers AB and CD) (21,23). Such an architecture allows the enzyme to contact the opposite ends of the L-shape tRNA simultaneously, with one dimer (AB) capturing the acceptor stem and the other dimer (CD) binding the anticodon (23). Each subunit possesses two nucleotide-binding pockets, one for adenylylation and the other for guanylylation (13,21). The adenylylation site can fit both ATP and GTP, whereas the guanylylation site can fit all four nucleotides. In the adenylylation site, two conserved amino acid residues (D29 and D76) interact with two Mg2+ ions, which in turn coordinate the triphosphate of the incoming nucleotide. In the guanylylation site, four conserved amino acid residues (R27, R92, K95 and R130) interact with the triphosphate of the second nucleotide, which in turn holds another Mg2+ ion (21). Mutation of any of these residues distinctly reduces G-1 addition to tRNAHis (21,24).

A dual-functional human Thg1

A recent study showed that a single HsThg1 gene is responsible for G-1 addition to both cytoplasmic and mitochondrial tRNAHis isoacceptors (tRNAnHis and tRNAmHis, respectively) (25). Notably, addition of G-1 to tRNAnHis forms a G-1:A73 mismatch, whereas addition of G-1 to tRNAmHis forms a G-1:C73 base pair (Figure 1B). As a CCA end immediately follows the discriminator base (N73), G-2 and G-3 can be incorporated easily via a template-dependent mechanism into tRNAHis with C73, but not A73. Unfortunately, tRNAHis with an additional G-2:C74 base pair perturbs the universally conserved CCA end of tRNA. This additional tRNA base pair might interfere with a conformational change important for translation (26). This raised the question of how HsThg1 can prevent successive incorporation of GTP into tRNAmHisin vivo. We showed herein that a single-GTP incorporation prevails when the GTP level is low (or the ATP/GTP ratio is high), and that HsThg1 possesses a tRNA-inducible GTPase activity. Perhaps HsThg1 can down-regulate the relative GTP level to preclude continuous incorporation of GTP into tRNAmHis.

MATERIALS AND METHODS

Gene cloning and protein purification

Cloning of the gene encoding HsThg1 into pET21b (or pADH) followed a standard protocol. In brief, a pair of gene-specific primers was used to amplify the gene via a polymerase chain reaction using human cDNA as the template. The forward primer (containing an SpeI site) was located just upstream of the open reading frame, while the reverse primer (containing an XhoI site) was located just upstream of the stop codon. The amplified DNA fragment was digested with SpeI and XhoI and then cloned into the multiple cloning sites of pET21b (or pADH). Cloning of Saccharomyces cerevisiae, Drosophila melanogaster and Bombyx mori Thg1 genes followed a similar strategy. To construct Mut A (D76A) and Mut G (R27A) of HsThg1, two sets of mutagenesis primer were designed accordingly. Site-directed mutagenesis followed protocols provided by the manufacturer (Agilent, Santa Clara, CA, USA).

For protein purification, the plasmid harboring the Thg1 gene was transformed into an Escherichia coli expression strain, BL21-CodonPlus(DE3) and the His6-tagged protein was purified to homogeneity through Ni-NTA affinity chromatography as previously described (27).

Primer extension analysis

To sequence the 5′ end of mature human tRNAmHis, this tRNA was isolated from a 293T cell culture and sequenced by primer extension analysis. Primer used for this assay contains sequences complementary to the anticodon- and D-arm region of HstRNAmHis (5′ TCTGATGTTTTGGTT). Primer was 5′ end labeled with [γ-32P]-ATP (Perkin Elmer, 3000 Ci/mmol) by T4 polynucleotide kinase and purified as described (28). Annealed primer was extended for 5 minute at room temperature, followed by 1-h reverse-transcription at 37°C with 20 μM dNTP and 0.4 U/μl AMV-reverse transcriptase in AMV-RT reaction buffer (Promega). Sequencing reaction additionally contained 10 μM of each corresponding ddNTP. Reactions were terminated by addition of an equal volume of RNA loading buffer (80% formamide and 1 mM EDTA). Products were resolved by 12% PAGE with 4 M urea and visualized by PhosphoImager (Typhoon FLA 9000) (5).

In vitro transcription of tRNAHis

Preparation of tRNAHis transcripts followed a previously described protocol (9). Briefly, a synthetic DNA duplex containing a T7 promoter (class III) and a gene encoding tRNAHis (GUG) was cloned into the SmaI site of pUC18 using a pair of complementary primers. The transcription template was enriched by polymerase chain reaction amplification of the insert (containing the T7 promoter and target gene). The in vitro transcription reaction of tRNAHis with a 5′-GMP was performed at 37°C for 3 h with 0.3 μM T7 RNA polymerase in 20 mM Tris–HCl (pH 8.0), 150 mM NaCl, 20 mM MgCl2, 5 mM DTT, 1 mM spermidine, 2 mM of each NTP, and 20 mM GMP. To generate tRNAHis with a 5′-GTP, 20 mM GMP was omitted from the above reaction mixture. The tRNAHis transcript was purified by a 10% denaturing urea-polyacrylamide gel. After ethanol precipitation and vacuum-drying, the tRNA pellet was dissolved in TE buffer and then refolded by heating to 80°C and gradually cooled to room temperature after addition of 2 mM MgCl2. Approximately 80% of in vitro-transcribed tRNAHis containing G-1 was active in aminoacylation.

Aminoacylation assay

Aminoacylation was carried out at ambient temperature in a buffer containing 50 mM HEPES (pH 7.5), 50 mM KCl, 15 mM MgCl2, 5 mM dithiothreitol, 10 mM ATP, 0.1 mg/ml bovine serum albumin, and 26.25 μM histidine (6.25 μM 14C-histidine; Perkin Elmer, Waltham, MA, USA), 5 μM tRNAHis and 50 nM HisRS (29). At various time points, reactions were quenched by spotting 10-μl aliquots of the reaction mixture onto Whatman filter pads (Maidstone, Kent, UK) that had been presoaked in a solution containing 5% trichloroacetic acid and 2 mM histidine. The filter pads were washed three times for 15 min each in ice-cold 5% trichloroacetic acid before liquid scintillation counting. Aminoacylation data were obtained from three independent experiments and averaged.

GTP incorporation assay

A GTP incorporation assay was carried out in a buffer containing 25 mM HEPES (pH 7.5), 10 mM MgCl2, 125 mM KCl, 3 mM DTT, 0.5 μM p-tRNAHis, 1–1000 μM GTP and 0 or 1000 μM ATP. The reaction was initiated by addition of HsThg1 to a final concentration of 0.5 μM. At various times, reactions were quenched by mixing 4-μl aliquots of the reaction mixture with an equal volume of 2× loading buffer (16 M urea, 0.5 M EDTA, 2% bromophenol blue, 2% xylene cyanol and 10% glycerol), and analyzed by 10% urea-polyacrylamide gel electrophoresis. The gel was then stained with SYBR Gold Nucleic Acid Gel Stain (Thermo Fisher Scientific) and visualized using a Typhoon Imager (GE Healthcare).

GTPase assay

The GTPase assay was carried out in a buffer containing 1 μM GTP (0.125 μM α-32P-GTP; PerkinElmer, Boston, USA), 25 mM HEPES (pH 7.4), 125 mM KCl, 10 mM MgCl2, 3 mM DTT, 0 or 5 μM tRNA, 0 or 1000 μM ATP and 0.5 μM Thg1 (unless otherwise indicated). The reaction mixture was incubated at 37°C for 40 min and quenched by addition of 2 μl 0.5 M EDTA. Two-μl aliquots of the quenched reaction mixture were spotted onto a polyethyleneimine cellulose plate (Merck, Germany) and the nucleotides were separated by thin-layer chromatography using 0.4 M KH2PO4 (pH 3.65) as the mobile phase for 40 min. The wet TLC plate was air dried in a laminar air flow for 30 min and then covered with plastic wrap before being inserted into a cassette with an X-ray film. The cassette was incubated at either −80°C (for 3–5 h) or room temperature (for overnight) (30,31). To mark the relative positions of GTP, GDP GMP and Pi in the TLC plate, GTP was partially hydrolyzed in a buffer containing 1 μM GTP (0.125 μM α-32P-GTP) and calf intestinal phosphatase (CIP) (10–5 unit) at 37°C for 1 min. The relative signal intensities of GTP and GDP were quantified using an ImageJ.

Complementation assay

Construction of a yeast haploid THG1 knockout (KO) strain (MATα, thg1::kanMX4, his3Δ1 leu2Δ0 lys2Δ0 ura3Δ0) was reported earlier (32). To examine whether a heterologous THG1 gene can substitute functionally for yeast THG1, a test plasmid harboring the target gene was constructed and transformed into the KO strain, and the resultant transformant was tested on a 5-FOA (1 mg/ml) plate for its ability to grow. Because yeast can convert 5-FOA to a toxic compound in the presence of URA3, the transformant must get rid of the maintenance plasmid (carrying a yeast THG1 gene and a URA3 marker) on 5-FOA to survive. Without the maintenance plasmid, the transformant will die on 5-FOA unless the test plasmid encodes a functional Thg1 enzyme.

RESULTS

Mature human mitochondrial tRNAHis contains G-1

Despite G-1 playing a key role in aminoacylation of in vitro-transcribed HstRNAmHis by HisRSm (29), the mitochondrial gene encoding HstRNAmHis lacks a G residue at position -1, and no experimental evidence has been provided so far to verify that the mature HstRNAmHis actually contains G-1. Pursuant to this objective, total RNA was isolated from a human 293T cell culture, followed by purification of tRNAmHis through a streptavidin column using a biotin-conjugated complementary oligonucleotide as the probe. Purified HstRNAmHis was then subjected to sequencing via primer extension. As shown in Figure 1C, the mature HstRNAmHis indeed contained G-1 but no G-2 or G-3. In addition, a prominent band was found across the gel at position G10. G10 is normally modified by human Trm11/Trm12 complex, yielding N-methylguanosine (m2G) (33). This modification is involved in L-shaped structure stabilization by preventing incorrect folding (34), which might hinder primer extension, leading to early termination of polymerization.

An aminoacylation assay was carried out to explore whether in vivo-isolated native HstRNAmHis is a suitable substrate for human mitochondrial HisRS (HsHisRSm). As shown in Figure 1D, HsHisRSm charged the native HstRNAmHis with an efficiency slightly higher than that for in vitro-transcribed HstRNAmHis with G-1, but it could hardly charge in vitro-transcribed HstRNAmHis without G-1. Therefore, G-1 is an important identity element for tRNAHis. On the other hand, because human cytoplasmic HisRS (HsHisRSc) strongly prefers tRNAHis with A73 (29), it was much less efficient at charging the aforementioned HstRNAmHis (Figure 1E).

ATP/GTP ratios affects GTP incorporation into tRNAmHis by human Thg1

As yeast and archaeal Thg1 enzymes require different nucleotides for tRNAHis activation (Figure 1A), we wondered whether ATP is absolutely required for HsThg1-catalyzed G-1 addition to tRNAmHis and whether ATP/GTP ratios affect the number of GTPs incorporated into HstRNAmHis. To this end, a G-1 addition assay was carried out using p-tRNAmHis as the substrate under conditions with various concentrations of ATP and GTP, and aliquots of the reaction products were resolved by 8 M urea−10% polyacrylamide gel electrophoresis. Figure 2 shows that G-1 incorporation occurred only when ATP was present (compared lanes 10–12 and 1–9), suggesting that ATP is absolutely required for HsThg1-catalyzed GTP incorporation. In addition, GTP concentrations or ATP/GTP ratios affected the number of GTPs incorporated into tRNAmHis. When the reaction was carried out at a relatively low ATP/GTP ratio (e.g. 1000/1000), more than one GTP was incorporated into p-tRNAmHis (see lanes 1–3). In contrast, when the reaction was carried out at a relatively high ATP/GTP ratio (e.g. 1000/10 or 1000/1), no multiple-GTP incorporation was observed (see lanes 4–9). Thus, the ratios of ATP to GTP affect the number of GTPs incorporated into tRNAHis with C73. Inspired by this finding, we sought to unveil mechanisms that can regulate the ATP/GTP ratio.

Figure 2.

Incorporation of GTP into p-HstRNAmHis by human Thg1. A G-1 addition assay was carried out under conditions with various concentrations of ATP and GTP. The reactions were quenched by addition of 0.5 M EDTA at various times, and the products were resolved by 8 M urea–10% polyacrylamide gel electrophoresis.

Human Thg1 displays tRNA-inducible GTPase activity

We next explored whether HsThg1 displays activities other than adenylylation and guanylylation. As HsThg1 closely interacts with mitofusin 2, a dynamin-like GTPase on the mitochondrial outer membrane (35), we set out to check whether HsThg1 also demonstrates a GTPase activity. Interestingly, HsThg1 could hydrolyze GTP to GDP to a certain extent in the absence of ATP and tRNAHis. Further addition of HstRNAnHis or HstRNAmHis to the reaction enhanced its activity (∼5-fold) (Figure 3A). In contrast, a similar activity was not detected in S. cerevisiae Thg1 (ScThg1), regardless of the presence or absence of SctRNAnHis. To check whether HsThg1 can further hydrolyze GDP to GMP or even guanosine, GTP was treated with various concentrations of HsThg1, ranging from 16 to 0.5 μM. As shown in Figure 3B, no phosphate (Pi) signal was observed at any enzyme concentration, and only a poor GMP signal was observed at the highest enzyme concentration. Thus, HsThg1 predominantly cleaves γ-phosphate from GTP, yielding GDP (Figure 3B).

Figure 3.

Human Thg1 possesses a tRNA-inducible GTPase activity. A GTPase assay was carried out under conditions (A) with or without tRNAHis, (B) with various concentrations of HsThg1, (C) with various tRNA species and (D) with mRNA. Relative positions of GTP, GDP, GMP and Pi are marked by partial hydrolysis of α-32P-labeled GTP by CIP. Quantitative data for (A) and (C) were obtained from three independent experiments and averaged.

We next investigated whether this GTPase activity is tRNAHis-specific. Surprisingly, tRNAThr and tRNAAla were almost as effective as tRNAHis in inducing the GTPase activity (∼5-fold increase), but tRNAPhe was less effective than HstRNAmHis (∼2-fold increase) (Figure 3C). Thus, the GTPase activity of HsThg1 is not tRNAHis-specific. Additionally, we checked whether an activated tRNAHis intermediate (ppp-HstRNAmHis) and a mature tRNAmHis (p-G-1-HstRNAmHis) can induce GTPase activity at a similar level. The results showed that these two tRNAmHis species could also induce the GTPase activity (3.6- and 2.7-fold, respectively) (Figure 3C). On the contrary, mRNA had little effect on the GTPase activity (Figure 3D). Together, these results suggest that HsThg1 can regulate the cellular GTP level (or ATP/GTP ratio) through its GTPase activity.

ATP inhibits the GTPase activity of human Thg1

As both ATP and GTP are substrates of HsThg1, we wondered whether ATP affects its GTPase activity. Pursuant to this objective, a GTPase assay was carried out with various concentrations of ATP, ranging from 1000 to 10 μM. As shown in Figure 4A, ATP inhibited the GTPase activity of HsThg1 in a dose-dependent manner. The higher the ATP concentration, the stronger the inhibition was. In addition, dATP, an ATP analog, was as effective as ATP in inhibiting the GTPase activity. To map the active site for the GTPase, we made two mutants, D76A (Mut A) in the adenylylation site and R27A (Mut G) in the guanylylation site, and then tested their G-1 addition and GTPase activities. Consistent with an earlier report (21), mutation at D76 impaired both adenylylation and guanylylation, whereas mutation at R27 impaired only guanylylation (Supplementary Figure S1). Our results also showed that mutation at either site abolished its ability to rescue a yeast THG1 KO strain on 5-FOA (Figure 4B). Western blotting showed that these two mutants could be stably expressed in the yeast, albeit to a level slightly lower than that of its WT counterpart (Figure 4C).

Figure 4.

GTPase activity resides in the adenylylation site of human Thg1. (A) Inhibition of GTPase activity by ATP. GTPase activity of HsThg1 was tested with various concentrations of ATP. (B) Complementation assay. Complementation was determined by transforming the test plasmid harboring a WT or mutant Thg1 gene into a yeast THG1 KO strain and plating the resultant transformants on 5-FOA. (C) Western blotting. The protein expression levels of the constructs used in (B) were determined by Western blotting using an anti-His6-tagged antibody as the probe. Constructs used in (B, C) are numbered for clarity. (D) GTPase activities of WT and mutant human Thg1 enzymes. GTPase activities of WT and mutant HsThg1 enzymes were determined in the absence or presence of ATP, CTP, and UTP. Mut A and Mut G refer to D76A and R27A, respectively. Quantitative data for (A) and (D) were obtained from three independent experiments and averaged.

As described earlier, the adenylylation site can accommodate only ATP and GTP, while the guanylylation site can accommodate all four nucleotides (21). We next analyzed the WT, Mut A, and Mut G for their relative GTPase activities under conditions with ATP, CTP or UTP. As shown in Figure 4D, all three nucleotides inhibited the GTPase activity of WT HsThg1 to a certain level, with ATP, UTP, and CTP reducing the activity 6-, 4- and 2-fold, respectively. Mutation at the guanylylation site (yielding Mut G) had little effect on its GTPase activity, and the mutant behaved like its WT counterpart in responding to nucleotide inhibition. In contrast, mutation at the adenylylation site (yielding Mut A) distinctly reduced its GTPase activity (∼2-fold), and the mutant was relatively insensitive to inhibition by ATP, UTP, or CTP. These results suggest that the GTPase activity resides in the adenylylation site of HsThg1.

GTPase activity exists in other high-eukaryotic Thg1 enzymes

As shown earlier, GTPase activity exists in HsThg1 but not in ScThg1 (Figure 3A), this prompted us to ask whether a similar activity exists in other high-eukaryotic Thg1 enzymes, especially those responsible for G-1 addition to both tRNAsHis with A73 and C73. To this end, we cloned the genes encoding D. melanogaster and B. mori Thg1 homologues (DmThg1 and BmThg1, respectively) and tested their activities in vivo and in vitro. As shown in Figure 5A, DmThg1 efficiently rescued the growth defect of the yeast KO strain on 5-FOA. In contrast, BmThg1 only weakly restored the growth phenotype of the KO strain, probably due to its poor expression in the yeast (Figure 5B). We next tested the purified Thg1 homologues for their GTPases activities in vitro. Remarkably, both DmThg1 and BmThg1 exhibited a strong GTPase activity, but the major product for DmThg1 was guanosine (removing α-, β- and γ-phosphates) instead of GDP (removing γ-phosphate) (Figure 5C). Thus, the GTPase activity is not unique to HsThg1, but instead is present in many other high-eukaryotic Thg1 homologues.

Figure 5.

GTPase activity exists in other high-eukaryotic Thg1 homologues. (A) Complementation assay. Complementation was determined by transforming the test plasmid harboring a Thg1 homologue into a yeast THG1 KO strain and plating the resultant transformants on 5-FOA. (B) Western blotting. The protein expression levels of the constructs used in (A) were determined by Western blotting using an anti-His6-tagged antibody as the probe. Constructs used in (A, B) are numbered for clarity. (C) GTPase assay for various Thg1 enzymes.

DISCUSSION

ATP/GTP ratios regulate the number of GTPs incorporated into HstRNAmHis

A previous study showed that yeast Thg1 adds only one GTP (as G-1) to tRNAnHis (with A73), but it adds up to three GTPs (as G-1/G-2/G-3) to tRNA once its A73 mutates to C73 (17). Similarly, human Thg1 adds more than one GTP to tRNAmHis (with C73) in vitro. This poses a challenge to HisRS, as only p-G-1-tRNAHis is a suitable substrate for aminoacylation. GMP can be used as a substrate and incorporated into an activated tRNAmHis intermediate (ppp-tRNAmHis), producing mature p-G-1-tRNAmHisin vitro. However, GMP is 22-fold less efficient than GTP in incorporation, which makes GMP an inferior substrate (36,37). In addition, the concentration of GMP is much lower than that of GTP in human cells (38). We demonstrated herein that multiple-GTP incorporation prevailed when the ATP/GTP ratio was low (Figure 2). In contrast, a single-GTP incorporation prevailed when the ATP/GTP ratio was high. As the concentration of ATP is ∼10-fold higher than that of GTP inside the cell (∼1000 μM and ∼100 μM, respectively) (38,39), such a condition might favor a single-GTP incorporation (Figure 2). However, under circumstances when the relative mitochondrial GTP level is increased, reducing the GTP level becomes necessary and pertinent to ensure proper maturation of tRNAmHis.

Human Thg1 displays a tRNA-inducible GTPase activity

The GTPase-family enzymes contain a classical G domain that folds structurally into five α helices, six β strands, and five loops interconnecting the helices and strands (40). Interestingly, the palm domain of a human Thg1 monomer holds a similar architechure, with four α helices, six β strands, and five loops interconnecting the helices and strands (21). A structural study revealed that HsThg1 appears as a dimer of dimers (AB and CD), holding two tRNAHis molecules facing one another in a parallel fashion. Each tRNAHis molecule occupies three subunits, where its acceptor stem is positioned in between subunits A and B, while its anticodon is positioned in subunit D or C (23). Although HsThg1 specifically recognizes the anticodon of tRNAHis, its GTPase activity can be induced not only by tRNAHis of various statuses, including p-tRNAHis (an immature tRNAHis), ppp-tRNAHis (an activated tRNAHis intermediate), and p-G-1-tRNAHis (a mature tRNAHis), but also by non-tRNAHis species, such as tRNAAla, tRNAThr and tRNAPhe, suggesting that this induction activity is not anticodon specific. (Figure 3). As tRNAPhe is not as efficient as other tRNAs tested, other parts of tRNA might be involved in this activity. Sequence and structural comparison showed that tRNAPhe is different from others by having a larger D-loop, a G at position 37, a U at position 52, and an A at position 62. Whether or not these differences play a role in this activity requires further verification. As most GTPase-family enzymes are induced by DNA (41), K+ (42), or ribosome (43), HsThg1 represents an interesting example of a tRNA-inducible GTPase.

Human Thg1′s GTPase activity resides in the adenylylation site

Despite the fact that yeast and human Thg1 enzymes follow a similar pathway for G-1 addition (17,21), only human Thg1 can efficiently hydrolyze GTP to GDP. Our results showed that the GTPase activity of HsThg1 resides in the adenylylation site. Therefore, mutation at the adenylylation site (D76A), but not the guanylylation site (R27A), distinctly reduces its GTPase activity (Figure 4). Moreover, ATP is much more potent than UTP and CTP in inhibiting the GTPase activity (Figure 4). As depicted in Figure 6, the amino acid residues in the adenylylation site make direct contacts with the nitrogen base (A43, D47 and S75), ribose (H34), and triphosphate (N32, H34 and D76) of the purine nucleotide. Therefore, the pyrimidine nucleotides (CTP or UTP) cannot properly fit the pocket. Moreover, mutation at D76 disrupts its interaction with the triphosphate of the purine nucleotide and impairs the GTPase activity. On the contrary, the base and sugar moieties are not seen in the electron density map of the guanylylation site presumably due to lack of specific interactions (21). As a result, the adenylylation site can fit only purine nucleotides (ATP or GTP), while the guanylylation site can accommodate all four nucleotides (ATP, UTP, CTP or GTP) with a preference toward GTP/UTP (37). When the ATP/GTP ratio is high, the enzyme preferentially catalyzes G-1 addition to tRNAHis, producing a mature tRNAHis for protein synthesis. Conversely, when the ATP/GTP ratio is low, the enzyme preferentially hydrolyzes GTP to GDP, precluding multiple-GTP incorporation into tRNAmHis. Human Thg1 interacts strongly with both Mfn1 and Mfn2, two dynamine-like GTPase homologues that play a central role in regulating mitochondrial fusion and cell metabolism. Thg1 increases Mfn2′s GTP-binding affinity significantly (44). However, It is unclear whether Mfn2 also increases Thg1′s G-1 addition or GTPase activity.

Figure 6.

A possible interplay between G-1 addition and GTP hydrolysis by human Thg1. The protein structure was obtained from dimeric human Thg1-dGTP complex (PDB ID: 3OTB). Subunits A and B of human Thg1 are colored in wheat and turquoise, respectively. The adenylylation and guanylylation sites are marked as ‘A’ and ‘G’, respectively. Under conditions with a high ATP/GTP ratio, HsThg1 preferentially catalyzes G-1 incorporation into tRNAmHis (reactions 1 to 4), whereas under conditions with a low ATP/GTP ratio, HsThg1 preferentially catalyzes the hydrolysis of GTP to GDP (reactions a to b).

tRNA-inducible GTPase activity exists in high-eukaryotic Thg1 enzymes

Yeast Thg1 adds G-1 to tRNAnHis across from A73, resulting in a G-1:A73 mismatch, which would terminate continuous 3′-5′ polymerization through a template-dependent mechanism (Figure 1A) (17). However, human Thg1 needs to add G-1 not only to tRNAnHis (with A73) but also to tRNAmHis (with C73) (25). As N73 is immediately followed by a CCA end (Figure 1B), multiple-GTP incorporation (as G-2/G-3) through a template-dependent mechanism can occur with tRNAmHisin vitro (Figure 2) (17,36). Because tRNAHis with an additional G-2:C74 base pair perturbs the universally conserved CCA end of tRNA and might interfere with a conformational change important for translation (27), an approach to generate mature tRNAHis with only G-1 becomes crucial. Our data showed that human Thg1 possesses a tRNA-inducible GTPase activity, which could be involved in down-regulating the GTP concentration inside mitochondria (Figure 6). This activity is particularly pertinent for Thg1 enzymes that must cope with tRNAHis carrying C73. Indeed, many high-eukaryotic Thg1 enzymes that act on both tRNAsHis with A73 and C73 (or U73) (29) possess a similar activity (Figure 5). However, regardless of the detailed interpretation, our study suggests that HsThg1 exhibits an ATP-inhibitable, tRNA-inducible GTPase activity. Additionally, an earlier report showed that HsThg1 interacts with a mitochondrial dynamin-like GTPase (Mfn1 or Mfn2) to regulate the GTP level. Such an interaction might allow Thg1 to participate in mitochondrial fusion and bioenergetic function (35).

DATA AVAILABILITY

The authors confirm that the data supporting the findings of this study are available within the article and its supplementary materials.

Click here for additional data file.