A naturally occurring mini-alanyl-tRNA synthetase

Abstract

Alanyl-tRNA synthetase (AlaRS) retains a conserved prototype structure throughout its biology, consisting of catalytic, tRNA-recognition, editing, and C-Ala domains. The catalytic and tRNA-recognition domains catalyze aminoacylation, the editing domain hydrolyzes mischarged tRNA^Ala, and C-Ala―the major tRNA-binding module―targets the elbow of the L-shaped tRNA^Ala. Interestingly, a mini-AlaRS lacking the editing and C-Ala domains is recovered from the Tupanvirus of the amoeba Acanthamoeba castellanii. Here we show that Tupanvirus AlaRS (TuAlaRS) is phylogenetically related to its host’s AlaRS. Despite lacking the conserved amino acid residues responsible for recognition of the identity element of tRNA^Ala (G3:U70), TuAlaRS still specifically recognized G3:U70-containing tRNA^Ala. In addition, despite lacking C-Ala, TuAlaRS robustly binds and charges micro^Ala (an RNA substrate corresponding to the acceptor stem of tRNA^Ala) as well as tRNA^Ala, indicating that TuAlaRS exclusively targets the acceptor stem. Moreover, this mini-AlaRS could functionally substitute for yeast AlaRS in vivo. This study suggests that TuAlaRS has developed a new tRNA-binding mode to compensate for the loss of C-Ala.

Introduction

An aminoacyl-tRNA synthetase (aaRS) attaches a specific amino acid to one of its cognate tRNAs to form an aminoacyl-tRNA. This charged tRNA is then delivered to the ribosome for translation of the genetic code¹. Therefore, aminoacyl-tRNA plays a key role in DNA translation―the expression of genes to make proteins. Many eukaryotic aaRSs recruit N- or C-terminal appended domains to expand their nontranslational functions during evolution. Hence, even aaRSs of the same amino acid specificity often possess distinct structural organizations between prokaryotes and eukaryotes. However, AlaRS has never recruited a new domain during evolution². Its prototype structure consists of the catalytic, tRNA-recognition, editing, and C-terminal (C-Ala) domains³. AlaRS exists as a homodimer (or tetramer) in prokaryotes^4,5 and a monomer in eukaryotes⁶. For instance, the hyperthermophilic archaeon Archaeoglobus fulgidus AlaRS (AfAlaRS) is a homodimer of 906 amino acid residues and genetically divided into two parts. Residues 1-739 include the catalytic, tRNA-recognition, and editing domains, while residues 736-906 comprise the C-Ala domain⁷. The N-terminal two domains are often referred to as the aminoacylation domain, which is responsible for alanylation of tRNA^Ala, and the editing domain is responsible for removal of the mischarged amino acids from tRNA^Ala. Defects in editing activity leads to misacylation, mistranslation, and cytotoxicity in yeast⁸ and neural degeneration in mice⁹. In contrast to the other three domains, C-Ala highly diverges among species. Prokaryotic C-Ala consists of a helical and a globular subdomain⁴ and constitutes the major tRNA-binding module¹⁰. The helical subdomain mediates dimer formation, whereas the globular subdomain targets the elbow of the L-shaped tRNA^Ala11. Despite eukaryotic C-Ala possessing a similar architecture, its helical subdomain consists of three α-helices instead of two. Human C-Ala binds DNA robustly but not tRNA^Ala, whereas C. elegans C-Ala binds both DNA and tRNA^Ala robustly^6,11,12.

Identity elements of tRNA may be single nucleotides, base pairs, or post-transcriptional modifications, and often reside in the acceptor stem and anticodon loop. AlaRS establishes tRNA^Ala identity via a unique G3:U70 wobble base pair in the acceptor stem. G3:U70 exists in all tRNAs^Ala, ranging from E. coli to human cytoplasm, but is missing from all non-tRNAs^Ala13,14,15. Recognition of this identity element is mediated by two conserved amino acid residues, Asn (N) and Asp (D) (as N359 and D450 in A. fulgidus and as N303 and D400 in E. coli), in the tRNA-recognition domain. These two amino acid residues form specific hydrogen bonds with the G:U base pair, thereby establishing its identity as the alanine acceptor¹¹. The amide nitrogen of the Asn side chain contacts O4 of U70 (from major-grove side), while the carboxyl side chain and the backbone carbonyl of Asp make hydrogen bonds with the 2-amino group of G3 (from minor-grove side)^15,16.

Tupanvirus belongs to the family of giant viruses Mimiviridae. Its genome contains 1400–1500 kb of double-strand linear DNA¹⁷. Based on the size and sequence of its genome, Tupanvirus can be divided into two strains: soda lake and deep ocean. Both strains were first identified to replicate in Acanthamoeba castellanii and Vermamoeba vermiformis¹⁸, and later in Acanthamoeba polyphaga, Acanthamoeba griffin, Dyctistelium discoideum, and Willartia magna¹⁹. Interestingly, Tupanvirus possesses a complete set of translation-related genes, including 20 aaRSs and tRNA isoacceptors associated with all amino acids, tRNA and mRNA maturation enzymes, translation initiation, elongation, and releasing factors. Paradoxically, however, Tupanvirus does not share the same codon preference with its natural host A. castellanii. Instead, the Tupanvirus genome’s codons match its own tRNA isoacceptors¹⁸, suggesting that Tupanvirus tRNA isoacceptors may be responsible for translation of its own genome.

Among the 20 Tupanvirus aaRSs, AlaRS was of particular interest, as it contains only the aminoacylation domain. Moreover, it lacks the conserved N and D residues in its tRNA-recognition domain. This piqued our curiosity as to whether and how this mini-AlaRS can function normally. Our results showed that Tupanvirus AlaRS (TuAlaRS) can bind and charge micro^Ala (an RNA substrate corresponding to the acceptor stem of tRNA^Ala) as well as tRNA^Ala specifically and robustly, an unexpected feature for a C-Ala-defective AlaRS. Conceivably, this mini-AlaRS has evolved a new tRNA-binding mode to counterbalance the loss of its C-Ala.

Results

TuAlaRS harbors only the aminoacylation domain

Unlike its A. fulgidus homologue, TuAlaRS possesses only the catalytic and tRNA-recognition domains (Fig. 1a). Motifs 1, 2, and 3 are still conserved in its catalytic domain. However, the conserved G3:U70-recognition amino acid residues N and D are missing from its tRNA-recognition domain. Instead, P321 and T416 are located at the corresponding positions in sequence alignment (Fig. 1b). To get a deeper insight, a structural model of TuAlaRS (with only 461 amino acid residues) was built using AlphaFold^2 20 and compared to the aminoacylation domain of A. fulgidus AlaRS (AfAlaRS-N484) obtained from AfAlaRS/tRNA^Ala complex (PDB ID 3WQY). Both structural models were visualized and further modified using EzMol²¹. TuAlaRS folds into a three-dimensional structure closely resembling that of AfAlaRS-N484 (Fig. 1c). P321 and T416 of TuAlaRS are located at almost the same positions as the conserved N and D residues of AfAlaRS-N484. Nevertheless, there are still some notable differences between these two structures, in particular at their C-termini. The C-terminal region of AfAlaRS-N484 (aa 456–484) folds into two short α-helices, while the corresponding region of TuAlaRS (aa 432–461) folds into a long α-helix.

Fig. 1: Domain organization of TuAlaRS.

a Schematic representation of TuAlaRS and AfAlaRS. The relative positions of the four domains are marked. b Alignment of the sequences flanking the G3:U70-recognition amino acid residues N and D (highlighted in red) in the tRNA-recognition domain. c Structural modeling of TuAlaRS in comparison with AfAlaRS-N484. The catalytic and tRNA-recognition domains were colored in light-blue and purple, respectively. The C-terminal helix of the tRNA-recognition domain was highlighted in pink.

Four Tupanvirus tRNA^Ala isoacceptors have been recovered, each of which carries a G3:U70 base pair (Supplementary Fig. 1). One of them has a D-loop larger than normal, while the rest of them carry a normal-sized D-loop. The sequences in the TψC-loop are conserved, while those in the D-loop are relatively diverged. Nevertheless, the bases that are important for the elbow formation (such as G¹⁸G¹⁹ in the D-loop and ψ⁵⁵C⁵⁶ in the TψC-loop) are still conserved, suggesting that they are likely to fold into the L-shaped structure²².

TuAlaRS charges tRNA^Ala and micro^Ala efficiently and specifically

As TuAlaRS lacks the major tRNA-binding module (C-Ala), we wondered whether this enzyme can charge tRNA^Ala efficiently. To this end, we purified the His₆-tagged TuAlaRS protein from an E. coli transformant and carried out an aminoacylation assay using in vitro-transcribed E. coli tRNA^Ala (EctRNA^Ala) as the substrate. To our surprise, this mini-AlaRS was almost as efficient as EcAlaRS in charging EctRNA^Ala (with ~2-fold difference in charging activity) (Fig. 2a, b). In addition, despite lacking the conserved N and D residues in its tRNA-recognition domain, TuAlaRS was still specific to G3:U70 and failed to charge EctRNA^Ala with G3:C70 or A3:U70 to a detectable level (Fig. 2a).

Fig. 2: Aminoacylation assay.

The aminoacylation activities of (a) TuAlaRS and (b) EcAlaRS toward WT and mutant EctRNA^Ala transcripts were determined by testing their ability to incorporate ³H-alanine into the transcripts. The aminoacylation activities of (c) TuAlaRS and (d) EcAlaRS toward WT and mutant Ec-micro^Ala transcripts were determined under similar conditions. The enzyme concentration used in the reactions was 100 nM. Error bars are standard deviation from triplicates.

We next tested whether TuAlaRS can efficiently charge a tRNA^Ala mimic without the elbow. Pursuant to this objective, an E. coli microhelix^Ala (Ec-micro^Ala, which corresponds to the acceptor stem of EctRNA^Ala) was used as the substrate (Supplementary Fig. 1, TuAlaRS charged this microhelix to a higher level than did EcAlaRS when the same concentration of enzymes was used (Fig. 2c, d). As a matter of fact, TuAlaRS charged EctRNA^Ala and Ec-micro^Ala to a similar level (Fig. 2a, c). Not surprisingly, mutation of G3:U70 to G3:C70 or A3:U70 in Ec-micro^Ala dramatically impaired its aminoacylation by TuAlaRS (Fig. 2c). To determine whether Tupanvirus tRNA^Ala (TutRNA^Ala) isoacceptors are good substrates for TuAlaRS and EcAlaRS, we arbitrarily chose one of the TutRNA^Ala isoacceptors (second from the left in Supplementary Fig. 1A) as the substrate for aminoacylation. As shown in Supplementary Fig. 2, both enzymes could charge TutRNA^Ala, albeit to a level slightly lower than that of EctRNA^Ala (2–4-fold). In addition, we also checked whether TuAlaRS can efficiently charge its host (A. castellanii) tRNA_n^Ala (Supplementary Fig. 1). The results showed that TuAlaRS charges A. castellanii tRNA_n^Ala as well as EctRNA^Ala (Supplementary Fig. 3).

TuAlaRS binds both tRNA^Ala and micro^Ala robustly

TuAlaRS’s ability to charge both EctRNA^Ala and Ec-micro^Ala efficiently implies that this mini-AlaRS can bind both ligands robustly. To take a closer look, we carried out an electrophoretic mobility shift assay (EMSA) using EctRNA^Ala and Ec-micro^Ala as the ligands. As shown in Fig. 3a, TuAlaRS bound EctRNA^Ala robustly (with a K_d value of 0.3 µM). Mutation of G3:U70 to G3:C70 or A3:U70 reduced its binding affinity 2–3-fold (Fig. 3a). EcAlaRS could also bind EctRNA^Ala robustly (with a K_d value of 1.0 µM; Fig. 3b). However, unlike the scenario of TuAlaRS, mutation at G3:U70 had little effect on its binding by EcAlaRS (~1.3-fold reduction in binding affinity) (Fig. 3b). Thus, mutation at G3:U70 had a stronger impact on binding by TuAlaRS than by EcAlaRS. A similar scenario was observed in mutation of the 3′-CCA end, in which deletion of the 3′-CCA end or substituting the 3′-CCA end with UUG reduced its binding by TuAlaRS 3–4-fold (Fig. 3a), but hardly affected its binding by EcAlaRS (Fig. 3b). The fact that deletion and substitution of the 3′-CCA end showed a comparable effect on TuAlaRS’s binding implies that the 3′-CCA end is important for recognition by TuAlaRS.

Fig. 3: Binding of TuAlaRS and EcAlaRS to EctRNA^Ala and Ec-micro^Ala transcripts.

The binding affinities of (a) TuAlaRS and (b) EcAlaRS for WT and mutant EctRNA^Ala transcripts were determined by an EMSA with protein concentrations ranging from 32 to 0.0625 μM. The binding affinities of (c) TuAlaRS and (d) EcAlaRS for WT and mutant Ec-micro^Ala transcripts were determined under similar conditions. Error bars are standard deviation from triplicates.

We next checked whether TuAlaRS can bind Ec-micro^Ala robustly. Consistent with an earlier report²³, EcAlaRS bound Ec-micro^Ala poorly (with a K_d value of >32 µM) (Fig. 3d). However, it was interesting to find that TuAlaRS binds Ec-micro^Ala robustly (with a K_d value of 0.3 µM) (Fig. 3c). As a result, TuAlaRS bound to EctRNA^Ala and Ec-micro^Ala equally well (Fig. 3a, c), indicating that this mini-AlaRS mainly targets the acceptor stem instead of the elbow. Mutation of G3:U70 to G3:C70 or A3:U70 reduced its binding affinity distinctly, but not drastically (~3-fold increase in K_d; Fig. 3c). In contrast, an EcAlaRS fragment (N461) possessing the N-terminal 461 amino acid residues bound EctRNA^Ala and Ec-micro^Ala poorly (with K_d values of >32 µM) (Supplementary Fig. 4), emphasizing the indispensable role of C-Ala in tRNA^Ala binding by EcAlaRS. Hence, the major binding affinity of EcAlaRS for tRNA^Ala arises from C-Ala targeting the elbow¹⁰, whereas the major binding affinity of TuAlaRS for tRNA^Ala arises from the mini-synthetase targeting the acceptor stem (Fig. 3). This might also explain why mutations in the acceptor stem have a stronger effect on binding by TuAaRS.

TuAlaRS fails to charge C. elegans mitochondrial micro^Ala

As reported earlier, an extra G:U base pair, G1:U72, in the acceptor stem of C. elegans mitochondrial tRNA^Ala blocks its aminoacylation by E. coli, S. cerevisiae, and C. elegans cytoplasmic AlaRSs efficiently^12,24. We wondered whether this extra G:U base pair has a similar effect on TuAlaRS. To this end, we carried out an aminoacylation assay using C. elegans mitochondrial micro^Ala (Ce-micro^Ala) (Supplementary Fig. 1) as the substrate. As shown in Supplementary Fig. 5A, TuAlaRS charged Ec-micro^Ala efficiently but not Ce-micro^Ala, suggesting that G1:U72 blocks TuAlaRS’s aminoacylation of Ce-micro^Ala. We next checked whether this anti-determinant blocks its binding by the mini-AlaRS. Contrary to our anticipation, TuAlaRS bound Ce-micro^Ala with affinity similar to that for Ec-micro^Ala (with K_d values of 0.3–0.4 μM) (Supplementary Fig. 5B), suggesting that this anti-determinant affects catalysis instead of binding. Perhaps G1:U72 interferes with the positioning of the 3′-CCA end of the acceptor stem in the active-site pocket.

Kinetic parameters for aminoacylation of tRNA^Ala and micro^Ala by TuAlaRS

To obtain the kinetic parameters (K_M and k_cat) for TuAlaRS’s aminoacylation of EctRNA^Ala and Ec-micro^Ala, we performed a kinetic assay using EcAlaRS as a control. As shown in Table 1, TuAlaRS charged EctRNA^Ala and Ec-micro^Ala with similar efficiency (<2-fold difference in k_cat/K_M), whereas EcAlaRS highly preferred EctRNA^Ala over Ec-micro^Ala (~68-fold difference in k_cat/K_M). In addition, EcAlaRS and TuAlaRS charged EctRNA^Ala with similar efficiency (~2-fold difference in k_cat/K_M), but TuAlaRS charged Ec-micro^Ala with a much higher efficiency than did EcAlaRS (~26-fold difference in k_cat/K_M). Moreover, mutation of G3:U70 to G3:C70 or A3:U70 in EctRNA^Ala or Ec-micro^Ala drastically reduced its aminoacylation efficiency (~100-fold reduction in k_cat/K_M) by TuAlaRS, with a major impact on k_cat, a scenario reminiscent of the EcAlaRS-catalyzed tRNA^Ala aminoacylation¹⁵. Thus, discrimination of G3:U70 by TuAlaRS mainly occurs on the k_cat level rather than on the K_M level.

Table 1 Kinetic parameters for aminoacylation of EctRNA^Ala and Ec-micro^Ala by EcAlaRS and TuAlaRS.

P321 and T416 are dispensable for aminoacylation

As shown in Fig. 1, the highly conserved N and D are absent from the tRNA-recognition domain of TuAlaRS. Instead, P321 and T416 are located at the corresponding positions in sequence alignment (Fig. 1b). In addition, structural modeling showed that the P and T residues are located at almost the same positions as the N and D residues of AfAlaRS-N484 (Fig. 1c). To check whether these two amino acid residues still play a similar role in recognition of G3:U70, we mutated P and T to A individually and assayed their aminoacylation activities. As expected, mutation at P or T had little effect on the enzyme’s tRNA charging activity and specificity. Similar to the WT enzyme, P321A and T416A could efficiently charge G3:U70-containing Ec-micro^Ala, but not G3:C70 or A3:U70-containing Ec-micro^Ala (Supplementary Fig. 6). Thus, P321 and T416 appear to be dispensable for G3:U70 recognition.

TuAlaRS can functionally rescue a yeast ALA1 knockout (KO) strain

To investigate whether TuAlaRS can function in vivo as an AlaRS enzyme, we cloned the gene encoding TuAlaRS into a yeast shuttle vector (with a TEF1 promoter) and transformed the resultant construct into a yeast ALA1 (which encodes AlaRS) KO strain to test whether this mini-AlaRS can substitute for yeast AlaRS and rescue the growth defect of the KO strain. As shown in Fig. 4a, the KO strain could not grow in the absence of a functional AlaRS on 5-FOA. Remarkably, TuAlaRS was able to restore the growth phenotype of the KO strain on 5-FOA, albeit not as efficiently as yeast AlaRS. This result suggests that TuAlaRS can charge yeast tRNA_n^Ala to a level sufficient for growth and that a lack of the editing domain in this mini-AlaRS does not cause cytotoxicity. Western blotting further showed that TuAlaRS is expressed to a level distinctly lower than that of yeast AlaRS under the conditions used (Fig. 4b). Thus, TuAlaRS can be a functional enzyme in vivo. In contrast, while EcAlaRS effectively rescued the growth defect of the KO strain, N461, an EcAlaRS fragment comparable to TuAlaRS, failed to do so. We wondered whether TuAlaRS shows toxicity under conditions with extra serine or glycine. To this end, we carried out complementation assays on 5-FOA with additional serine. The results showed that addition of extra serine (up to 5 mM) does not cause any toxic phenotype in the yeast knockout strain carrying TuAlaRS (Supplementary Fig. 7), suggesting that this enzyme has tolerable or low mischarging activity.

Fig. 4: Rescue of a yeast ALA1 KO strain by TuAlaRS.

a Complementation assay. TuAlaRS’s rescue activity was determined by transforming the test plasmid into a yeast ALA1 KO strain and plating the resultant transformants on 5-FOA. The symbols “+” and “−” denote positive and negative complementation, respectively. b Western blotting. The protein expression levels of these constructs were determined by Western blotting using an anti-His₆-tagged antibody as the probe. Constructs used in (a, b) are numbered for clarity.

Phylogenetic relationship of TuAlaRS with other AlaRSs

To trace the evolutionary origin of TuAlaRS and its relationships with other AlaRSs, we aligned the protein sequences of representative AlaRSs recovered from all three domains of life and built a phylogenetic tree^25,26. As shown in Fig. 5, except for the mitochondrial AlaRSs of A. castellanii and D. discoideum, which were clustered into the bacterial group, all cytoplasmic and mitochondrial isoforms of eukaryotic AlaRSs tested were clustered into the same group, suggesting that these isoforms descended from common predecessors through gene duplication (Fig. 5). In addition, this main eukaryotic group showed a higher affinity for the bacterial than for the archaeal group, suggesting a bacterial origin for the eukaryotic AlaRSs. As for TuAlaRS, it was closely associated with A. castellanii and D. discoideum cytoplasmic AlaRSs and belonged to the main eukaryotic group. As A. castellanii and D. discoideum are its natural hosts, this finding implies that TuAlaRS was horizontally transferred from its host genome and the gene encoding this enzyme was reduced afterwards.

Fig. 5: Phylogenetic analysis of AlaRSs.

The evolutionary history was inferred by using the Maximum Likelihood method and Le_Gascuel_2008 model. Initial tree(s) for the heuristic search were obtained automatically by applying Neighbor-Join and BioNJ algorithms to a matrix of pairwise distances estimated using the JTT model. A discrete Gamma distribution was used to model evolutionary rate differences among sites. This analysis involved 26 amino acid sequences. Evolutionary analyses were conducted in MEGA11.

Discussion

TuAlaRS displays a G3:U70 specificity

AlaRS is the only aaRS that retains a prototype structure throughout its biology. However, Nanoarchaeum equitans harbors a split AlaRS gene that encodes an α and a β subunit²⁷. The α-subunit, comprising an intact catalytic domain and part of the tRNA-recognition domain, charges tRNA^Ala with a low activity and in a non-G3:U70-specific manner. The β-subunit comprises the rest of the tRNA-recognition domain along with the editing and C-Ala domains. Addition of the β subunit into the α subunit-catalyzed aminoacylation reaction enhances its activity drastically and confers a G3:U70 specificity. Unlike N. equitans AlaRS, TuAlaRS contains an intact catalytic domain and an intact tRNA-recognition domain but lacks the editing and C-Ala domains completely (Fig. 1a). This mini-AlaRS charges G3:U70-containing tRNA^Ala and micro^Ala efficiently and specifically (Fig. 2 and Table 1).

Paradoxically, TuAlaRS lacks the conserved N and D residues in its tRNA-recognition domain; instead P and T residues are located at the corresponding positions (Fig. 1). However, mutation at either amino acid residue had little effect on the enzyme’s tRNA charging activity or specificity (Supplementary Fig. 6). Clearly, TuAlaRS uses a recognition mode distinct from E. coli (which selects G3:U70 through a direct positive recognition) or eukarya/archaea (which repels a non-G3:U70 base pair) AlaRS¹⁵. Perhaps, these two amino acids play a role in the enzyme’s structural stability or flexibility. To facilitate the acceptor stem binding, the enzyme might select the universal identity element through a new site/mode. It is worth mentioning that C. elegans mitochondrial AlaRS also lacks the conserved N/D, but the enzyme still exhibits a G3:U70 specificity^12,24. Human mitochondrial AlaRS, on the other hand, presents a completely different scenario, in which the conserved N/D residues are replaced by M/G and its cognate tRNA lacks G3:U70 (Fig. 1). As it turns out, this enzyme discriminates human mitochondrial tRNA^Ala mainly through recognition of its variable loop²⁸. In addition, Drosophila melanogaster mitochondrial AlaRS recognizes a shifted G:U base pair ― G2:U71²⁹.

In the case of tyrosyl-tRNA synthetase (TyrRS), an insertion peptide that splits the active site-containing domain determines the species-specific acylation. Substrate specificity can be switched by transplanting part of the insertion from human TyrRS into E. coli TyrRS and, reciprocally, by transplanting the counterpart from the E. coli enzyme into the human enzyme³⁰. Unfortunately, there is no insertion peptide found in the active site-containing domain of AlaRS. Instead, an anti-determinant, G1:U72, is found in the acceptor stem of the C. elegans mitochondrial tRNA^Ala. E. coli and Tupanvirus AlaRSs fail to charge this tRNA because of the presence of G1:U72. Charging is effectively blocked by the 4-ketooxygen of U72 in the major grove³¹. However, adaptations make C. elegans mitochondrial AlaRS insensitive to the presence of the 4-ketooxygen. Extensive alanine-scanning mutagenesis of motif 2 of EcAlaRS provided little evidence for important contacts between tRNA^Ala and this segment of the protein³², which is largely consistent with the observation made in the cocrystal structure of A. fulgidus AlaRS:tRNA^Ala16. Perhaps, it is the flexibility of the motif 2 loop, instead of the amino acid residues therein, that enables the C. elegans mitochondrial enzyme to accommodate the 4-ketooxygen.

Despite lacking C-Ala, TuAlaRS binds tRNA^Ala robustly

In E. coli, C-Ala constitutes the major tRNA-binding module, targeting the elbow of the L-shaped tRNA^Ala10. As a result, EcAlaRS charges EctRNA^Ala efficiently but not Ec-micro^Ala (by a ~60-fold difference)²³ (also repeated and shown in Table 1). Deletion of C-Ala from EcAlaRS reduces its binding affinity for tRNA^Ala drastically (up to 13-fold)¹⁰. A similar scenario was observed for C. elegans cytoplasmic AlaRS¹². Despite the fact that human cytoplasmic AlaRS can charge tRNA^Ala efficiently in the absence of its C-Ala⁶, it is unclear whether this enzyme can also efficiently charge micro^Ala. In addition, a truncated form of Aquifex aeolicus AlaRS that contains only the aminoacylation domain (N453) can efficiently charge a G3:U70-containing minihelix (an RNA substrate corresponding to the acceptor stem plus the TψC arm)³³, suggesting that its C-Ala is dispensable for tRNA binding. We demonstrated herein that TuAlaRS can bind and charge micro^Ala and tRNA^Ala with a similarly high efficiency (Figs. 2 and 3 and Table 1). Clearly, TuAlaRS exclusively targets the acceptor stem of tRNA^Ala. However, it should be emphasized that G3:U70 in the acceptor stem still represents the major identity element for TuAlaRS’s recognition of tRNA^Ala (Fig. 2). Mutation at this base pair primarily affects k_cat (catalysis) instead of K_M (apparent affinity) for aminoacylation (Table 1), a scenario reminiscent of EcAlaRS³⁴. To the best of our knowledge, TuAlaRS is the only naturally occurring aaRS that contains only the aminoacylation domain and can function normally.

According to the A. fulgidus AlaRS:tRNA^Ala complex, the amino acid residues ¹⁹³GGG¹⁹⁵ act as the route separator between reactive and non-reactive CCA-ends¹⁶. Mutation of any of these three Gly residues causes mis-aminoacylation of G3:C70-containing tRNA^{Ala 34}. Interestingly, the second and third Gly residues are conserved in TuAlaRS and other AlaRSs, implying that the remaining Gly residues play a similar role as the route separator. Moreover, unlike E. coli AlaRS, which binds the CCA-containing and CCA-deletion (or substitution) tRNAs^Ala with a similar affinity, TuAlaRS distinctly prefers the CCA-containing tRNA^Ala (Fig. 3a, b), lending further support to the hypothesis that TuAlaRS has gained additional specific contacts with the acceptor stem through the CCA-end.

TuAlaRS may have lost its editing and C-Ala domains later during evolution

Our phylogenetic analysis suggested that TuAlaRS is closely related to its hosts’ cytoplasmic AlaRSs (Fig. 5), consistent with the hypothesis that Tupanvirus acquired most of its genes through horizontal transfer from its host genomes^18,35. However, gene acquisition in giant viruses such as Tupanvirus is often accompanied by gene loss or reduction in response to environmental changes during evolution. Thus, TuAlaRS’s smaller size may have resulted from gene reduction during evolution. Examples of gene reduction were also found in TyrRS of Acanthamoeba polyphaga mimivirus and many proteins of Chlorella virus^36,37. Regardless of the detailed interpretation, our study suggests that TuAlaRS has developed a new binding mode to counterbalance the loss of C-Ala. Instead of targeting the elbow, this mini-AlaRS targets the acceptor stem (Fig. 3). In a sense, TuAlaRS exemplifies an ancient-like aaRS, which possesses only the aminoacylation domain and identifies its cognate tRNA through recognition of an operational RNA code (a specific sequence or structure) in the acceptor stem³⁸.

Methods

Construction of plasmids

For expression in E. coli, the gene encoding TuAlaRS (GenBank: MF405918.2; protein ID: QKU33644.1) was codon optimized for E. coli and synthesized by Omics Biotechnology (Taiwan). The synthesized gene was cloned into the NdeI/XhoI sites of pET21b (an E. coli expression vector) and the resultant plasmid was transformed into an E. coli strain, BL21-CodonPlus(DE3). The resultant transformants were induced with 1 mM isopropyl β-D-1-thiogalactopyranoside for 4 h at 30 °C. The His₆-tagged protein was purified to homogeneity through Ni-NTA affinity chromatography followed the protocols provided by the manufacture of the Ni-NTA resin (Qiagen, Hilden, Germany)³⁹. For expression in yeast, the gene encoding TuAlaRS was codon optimized for S. cerevisiae and cloned in pTEF1 (a yeast shuttle vector with a strong TEF1 promoter).

Electrophoretic mobility shift assay (EMSA)

Protein-RNA binding affinities were determined by an EMSA, as described earlier¹⁰. The RNA substrate was ³²P-radiolabeled at the 5′-end⁴⁰. For each complex formation, ~10,000 cpm of RNA (diluted with cold RNA to 0.1 μM final concentration) was incubated for 10 min on ice with variable concentrations of AlaRS in a total volume of 20 μl containing 20 mM HEPES, pH 7.5, 20 mM KCl, 5 mM MgCl₂, and 2 mM DTT. Glycerol was added to a final concentration of 5% in each sample prior to loading on a native 5% 29:1 polyacrylamide gel. The gel was run at 100 V for 30–45 min at 4 °C in 0.5× TBE and then dried. The dried gel was exposed to X-ray film at −80 °C with an intensifying screen. The signal intensity was compared through ImageJ to measure the remaining free ³²P-RNA.

Complementation assay on FOA

Construction of a yeast haploid ALA1 KO strain (MATα, ala1::kanMX4, his3Δ1 leu2Δ0 lyS3Δ0 ura3Δ0) was described earlier⁴¹. To examine whether a heterologous AlaRS gene can functionally substitute for the cytoplasmic activity of yeast ALA1 (which encodes both the cytoplasmic and mitochondrial forms of AlaRS), a test plasmid carrying the target gene and a LEU2 marker was transformed into the KO strain, and the resultant transformant was tested for its ability to grow on 5-FOA (1 mg/ml) plates. Because 5-FOA can be converted to a toxic compound in the presence of URA3, the transformant must evict the maintenance plasmid (carrying a WT ALA1 gene and a URA3 marker) on 5-FOA to survive. Thus, the transformant cannot grow on 5-FOA unless the test plasmid encodes a functional cytoplasmic AlaRS.

In vitro transcription of tRNA^Ala and micro^Ala

In vitro transcription of tRNA followed a standard protocol. We synthesized and cloned a DNA fragment containing a T7 promoter followed by a sequence encoding tRNA^Ala into the SmaI site of pUC18. We enriched the transcription DNA template by PCR amplification of the cloned fragment. We performed in vitro transcription at 37 °C for 3 h with 0.3 μM T7 RNA polymerase in a buffer containing 20 mM Tris-HCl (pH 8.0), 150 mM NaCl, 20 mM MgCl₂, 5 mM DTT, 1 mM spermidine, and 2 mM of each NTP. We purified the tRNA transcript using an 8 M urea-15% polyacrylamide gel. Following purification, ethanol precipitation, and vacuum drying, we dissolved the tRNA pellet in 1× TE buffer (20 mM Tris-HCl [pH 8.0] and 1 mM EDTA), refolded the tRNA by heating it to 80 °C for 3 min, and gradually cooled it to 55 °C before adding 2 mM of MgCl₂. We collected the refolded tRNA when the temperature reached ~25 °C and stored it at -80 °C. Preparation of micro^Ala followed a similar protocol.

Aminoacylation assay

We performed in vitro aminoacylation assays at 37 °C 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, 5 μM tRNA^Ala (or micro^Ala), 20 μM alanine (including 1.34 μM ³H-alanine; PerkinElmer, Waltham, MA, USA), and 100 nM AlaRS. At various times, we quenched the reactions by spotting 10 μl aliquots of the reaction mixture onto Whatman filter pads (Maidstone, Kent, UK). The filter pads had been presoaked in a solution containing 5% trichloroacetic acid and 2 mM alanine. We washed the filter pads three times for 15 min each in ice-cold 5% trichloroacetic acid and dried the pads with a heat lamp before liquid scintillation counting. We obtained aminoacylation data from three independent experiments and averaged them.

Kinetic parameters for aminoacylation of tRNA^Ala and micro^Ala by AlaRS were determined by fitting the data points directly to the Michaelis-Menten equation. Initial rates of aminoacylation were determined at 37 °C with tRNA^Ala concentration ranging from 1 to 10 µM, and micro^Ala concentrations ranging from 1 to 10 µM (for TuAlaRS) or 10–400 µM (for EcAlaRS). AlaRS concentrations ranged from 200 to 4000 nM. Active protein concentrations were determined by active site titration in the reaction cocktail containing [γ-³²P]ATP, 10 mM MgCl₂, 144 mM Tris-Cl (pH 7.8), 1 mM cognate amino acid, and yeast inorganic pyrophosphate (2 units/ml). The reaction was then incubated at 25 °C⁴².

Statistics and reproducibility

The data were obtained from three independent experiments and averaged.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.