Восстановление ТРНС последнего универсального общего пастора (Общества) на пути к восстановлению системы перевода Древних.

Introduction

The inference of ancestral protein or nucleic acid sequences using molecular phylogenetic analysis, commonly referred to as ancestral sequence reconstruction (ASR), has become a widely used approach in evolutionary studies and is often followed by synthesis and functional characterization of the inferred molecules (Jermann et al. 1995 ; Gaucher et al. 2003 , 2008 ; Finnigan et al. 2012 ; Kacar et al. 2017a , b ). ASR is a powerful approach for reconstructing ancestral biomolecules and experimentally investigating the physiology of extinct organisms. For example, Gaucher et al. ( 2008 ) reconstructed the translational elongation factor EF-Tu from various evolutionary ages, analyzed its thermal stability, and used this information to infer that ancestral organisms were thermophilic, supporting hypotheses about elevated global temperatures in ancient environments. In the same manner, our laboratory analyzed the thermostability of an inferred ancestral nucleoside diphosphate kinase (NDK) and inferred that the common ancestor of Archaea and Bacteria—as well as the last universal common ancestor (LUCA, referred to as “Commonote” in our previous studies such as Furukawa et al. 2017 )—were likely (hyper) thermophiles (Akanuma et al. 2013 , 2015 ). We also inferred ancient global surface temperature trends based on the properties of ancestral NDKs (Garcia et al. 2017 ; Harada et al. 2019 ).

The translation systems of Archaea, Bacteria, and Eukarya generally utilize the standard genetic code, with a few exceptions (Watanabe and Yokobori 2014 ). Thus, the last archaeal common ancestor (LACA), last bacterial common ancestor (LBCA), and LUCA are all presumed to have used the standard code. During translation, specific amino acids are attached to their corresponding tRNAs by aminoacyl-tRNA synthetases (ARSs), and the anticodon of tRNAs pairs with the codon of mRNA on ribosomes, preserving the codon-amino acid relationship.

Prior to LUCA, the early translation system likely utilized a limited set of prebiotically synthesized amino acids, with the repertoire expanding over time (cf. Wong 1975 ). Functional proteins formed from such prebiotic amino acids—identified in meteorites, asteroids such as Ryugu (Lawless 1973 ; Cronin and Pizzarello 1983 ; Naraoka et al. 2023 ; Parker et al. 2023 ) and chemical evolution experiments such as those of Stanley Miller (Miller 1953 )—have been studied experimentally (e.g. Patel et al. 2015 ; Shibue et al. 2018 ).

Regarding aminoacylation, it has been hypothesized that, in the RNA world, a limited set of amino acids was recognized and bound by proto-tRNAs (Koonin 2017 ). As proto-tRNAs diversified, different RNAs began to recognize similar amino acids, driving the emergence of polypeptide cofactors and protein domains that catalyzed aminoacylation, ultimately giving rise to the modern ARS system (Furukawa et al. 2022 ).

Despite their different amino acid specificities, all tRNAs share a conserved structure, suggesting that they are derived from a common ancestor. This structural conservation allows the reconstruction of composite phylogenetic trees of tRNAs. However, due to their short sequence length and limited phylogenetic signal, tRNA-based phylogenetic analysis remains challenging. In one such effort, de Farias ( 2013 ) inferred ancestral tRNA sequences and a composite phylogenetic tree but did not address the co-evolution of tRNAs and ARSs. However, this study emphasized the evolutionary significance of the second anticodon position. Other studies have investigated early tRNA evolution prior to the establishment of the canonical L-shaped (cloverleaf-like) structure (Tamura 2015 ; Root-Bernstein et al. 2016 ).

Phylogenetic studies of ARSs suggest that GlnRS originated from an archaeal/eukaryal-type GluRS (Woese et al. 2000 ; Nair et al. 2016 ). GlnRS forms a distinct clade that is closely related to archaeal and eukaryal GluRS, rather than bacterial GluRS. Similarly, AsnRS is thought to have evolved from archaeal AspRS (Woese et al. 2000 ; Nair et al. 2016 ). GluRS and AspRS exist in two forms: non-discriminating (ND-GluRS and ND-AspRS) and discriminating (D-GluRS and D-AspRS), based on their ability to distinguish between tRNA Glu and tRNA Gln or tRNA Asp and tRNA Asn . These observations imply that LUCA may have lacked dedicated GlnRS and AsnRS enzymes, instead relying on ND-GluRS and ND-AspRS in combination with tRNA-dependent transamidosomes to synthesize Gln and Asn from Glu and Asp directly on tRNA Gln and tRNA Asn .

In Bacteria, methionine-attached initiator tRNA is formylated (fMet-tRNA i fMet ) by methionyl-tRNA formyltransferase, forming a complex with IF2-GTP to initiate translation. In contrast, archaeal and eukaryal initiator tRNAs (Met-tRNA i Met ) are not formylated. Although formylation enhances bacterial translation initiation, it is not essential; disruption of the formyltransferase gene in Escherichia coli ( E. coli ) is non-lethal but reduces growth (Guillon et al. 1992a ). Therefore, it is likely that Met-tRNA i Met in LUCA was not formylated and lacked features for interaction with the methionyl-tRNA formyltransferase.

Another important aspect of tRNA evolution is codon decoding. While non-Watson–Crick base pairs (e.g., G: U) are tolerated at the wobble position (the first position of the tRNA anticodon and the third position of the mRNA codon) (Crick 1966 ), wobble rules vary among systems (Osawa et al. 1992 ; Watanabe and Yokobori 2011 , 2014 ).

Unmodified U at the first anticodon position enables decoding of all four codons in a 4-codon box in translation systems with a limited tRNA repertoire, such as mycoplasmas and mitochondria (Osawa et al. 1992 ; Watanabe and Yokobori 2011 , 2014 ). In contrast, in Eukarya, unmodified U typically pairs only with A. Modified U nucleotide, such as 5-carboxymethylaminomethyluridine (cmnm⁵U), 5-methylaminomethyluridine (mnm⁵U), or 2-thiouridine derivatives—are common in bacterial and eukaryal tRNAs, while archaeal systems tend to use simpler modifications (Masuda and Hou 2024 ). Modified U at the first anticodon position decodes only A and G at the third codon position.

Other unmodified nucleotides (A, G, and C) also appear at the first anticodon position in some modern tRNAs, enabling broader decoding. For instance, unmodified A is found in mycoplasma tRNAs and in the mitochondrial tRNA Arg ACG from the nematode, Ascaris suum , which decodes all CGN codons (Watanabe et al. 1997 ). Inosine (I34), a deaminated A, expands decoding capacity in Bacteria (decoding U, C, and A) and Eukarya (decoding U and C) but is absent in Archaea (Masuda and Hou 2024 ). Unmodified G can decode C and U in general, but also A as seen in Drosophila mitochondrial tRNA Ser GCU (Tomita et al. 1999 ). Queuosine (Q), a modified base found in bacterial and eukaryal tRNAs for Tyr, His, Asn, and Asp, is also absent in Archaea (Masuda and Hou 2024 ). Unmodified C can also function in decoding, although it decodes only G in Crick’s wobble rule, e.g., the mitochondrial tRNA Met CAU decodes both AUG and AUA codons in the squid, Loligo bleekeri (Ohira et al. 2013 ). Modifications like lysidine (Bacteria) and agmatidine (Archaea) further diversify decoding (decoding A instead of G) (Muramatsu et al. 1988 ; Mandal et al. 2010 ; Ikeuchi et al. 2010 ).

Given the differences in modification between bacterial and archaeal tRNAs (in particular, A and C), it is plausible that the number of anticodon modifications in LUCA was likely minimal or absent. Estimating wobble rules in this pre-modification era is essential for understanding tRNA evolution. Watanabe and Yokobori ( 2014 ) proposed that 4-codon boxes were decoded by single tRNAs with unmodified U (or A) at the first anticodon position. For NNY 2-codon families, tRNAs with G 34 are likely responsible, and for NNR families, U 34 was hypothesized. U 34 could potentially decode both NNY and NNR, possibly competing with G 34 tRNAs. The roles of unmodified A and C remain to be clarified. A central unanswered question concerns the identity of the nucleotides that were used at the first anticodon position in the unmodified era and whether this minimum set was sufficient to decode all the sense codons.

As described above, ASR is widely used to reconstruct ancient genes and proteins and to infer ancient life and environments (Gaucher et al. 2003 , 2008 ; Thornton 2004 ; Akanuma et al. 2015 ). Ancestral proteins from LACA, LBCA, and LUCA have been reconstructed and their properties inferred to evaluate the thermal condition of their host’s growth environment by several groups, including ours (Akanuma et al. 2013 , 2015 ; Garcia et al. 2017 ; Harada et al. 2019 ).

ASR approaches have also been applied to study translational system evolution. For instance, ancestral EF-Tu proteins have been analyzed not only for their thermostability but also for their ability to function within extant protein synthesis systems (Gaucher et al. 2008 ; Kacar et al. 2017a , b ). In the case of aminoacyl-tRNA synthetases (ARSs), Fournier et al. ( 2011 ) inferred the ancestral sequence of a common ancestor of IleRS and ValRS based on a composite phylogenetic tree of IleRS, ValRS, and LeuRS. Their analysis suggested that the ancestral enzyme could have discriminated between isoleucine and valine, implying that both amino acids were already utilized in protein synthesis at that evolutionary stage. Similarly, Fournier and Alm ( 2015 ) reconstructed the ancestral protein sequences of TyrRS and TrpRS and proposed that tryptophan was incorporated into the genetic code after the divergence of these two enzymes.

In another line of research, we aim to elucidate the evolutionary pathway of the genetic code prior to LUCA by reconstructing ancestral sequences of Class IIa ARSs based on composite phylogenetic trees (Furukawa et al. 2022 ). Through comparative analyses of these ancestral sequences, we explored the evolution of amino acid specificity among Class IIa ARSs (Furukawa et al. 2022 ) and experimentally reconstructed and characterized several ancestral ARSs from Classes I and II (S. Yokobori et al., in preparation).

Like ARSs and other translation-related proteins, tRNAs have also undergone evolutionary changes. To investigate the substrate specificities of ancestral translation factors and ARSs, it is necessary to examine tRNAs that are contemporaneous with these ancestral proteins. Only through such complementary analyses can ancient translation systems be experimentally validated. Therefore, we argue that, like elongation factors and ARSs, tRNAs should also be investigated using ASR-based approaches. However, as noted above, most prior studies of ancestral tRNAs remain theoretical (e.g. Tamura 2015 ; Root-Bernstein et al. 2016 ), and empirical ASR-based research on tRNAs is still quite limited (de Farias 2013 ).

As a first step toward reconstructing an early translation system, we inferred the ancestral tRNA sequences corresponding to LUCA, LACA, and LBCA for all 20 standard amino acids and characterized their features. Our ultimate goal is to reconstruct a “primitive translation system” composed entirely of tRNAs from LUCA or earlier lineages, providing a foundation to explore the nature of early translation. Two key challenges are (1) determining the minimal set of anticodons and (2) acquiring functional tRNAs for all amino acids.

To evaluate the functionality and anticodon requirements, we focused on LUCA tRNA Ala and tRNA Ser . Since their anticodons are not recognized by their respective ARSs (Hou and Schimmel 1988 ; Francklyn and Schimmel 1989 ), their anticodons can be modified without affecting aminoacylation specificity. We engineered LUCA and E. coli tRNA Ala and tRNA Ser genes with anticodons (UGU, CGU, AGU, and GGU) to decode Thr codons. We also synthesized Thr codon-free reporter genes beginning with ACA, ACC, ACG, or ACT codons for T7 in vitro transcription (“ACN codon test genes”).

Using these tRNAs and ACN reporter constructs in a Thr-free version of the E. coli cell-free protein synthesis system (Shimizu et al. 2001 ), we assessed the functionality of the modified tRNAs and examined their codon–anticodon pairing properties.

Materials and Methods

Data Collection

Genome sequences from selected species representing Bacteria, Archaea, and Eukarya were retrieved from GenBank (Supplementary Table 1). Species were selected to maximize phylogenetic diversity across the three domains of life while minimizing redundancy from closely related taxa. tRNA genes were extracted from these sequences and classified according to their amino acid specificity, as inferred from their anticodon sequences. Twenty distinct tRNA groups were defined for the initial analysis.

Phylogenetic Analyses and Ancestral Sequence Inference

After excluding tRNA sequences with potentially incorrect annotations and those that may have been recruited from other tRNA groups, each set was aligned using MAFFT version 7.017 (Katoh and Standley 2013 ). The initial alignments were refined manually based on the standardized tRNA numbering system (Giegé and Eriani 2023 ). Positions containing gaps were retained in the alignment unless they were ambiguously aligned, in which case they were excluded during the manual curation. Poorly aligned sequences were removed from the datasets. The resulting alignments of tRNAs, grouped by amino acid specificity, were then used to reconstruct phylogenetic trees.

Phylogenetic analysis of each tRNA species was performed using IQ-TREE version 1.6.12 (Nguyen et al. 2014 ). For each tRNA dataset, the best-fit substitution model was selected based on the Bayesian Information Criterion (BIC) implemented in IQ-TREE (see the legend of Supplementary Fig. S1 ). Branch support was assessed using nonparametric bootstrap analysis with 1,000 replicates.

tRNAs with the CAU anticodon include initiator tRNA (tRNA i fMet or tRNA i Met ), elongator tRNA e Met , and tRNA Ile CAU . However, many genome annotations do not distinguish among these types and often label them all as tRNA Met . Therefore, for tRNA Met , we first constructed a phylogenetic tree by aligning all genes with a CAU anticodon annotated as tRNA Met . Based on the resulting tree, we inferred the respective clades corresponding to the initiator and elongator tRNAs, as these two types have distinct conserved sequence features. Experimentally validated sequences of E. coli initiator and elongator tRNAs were used as references to identify these clades. For tRNA Ile , genes annotated as tRNA Ile CAU , tRNA Ile GAU , tRNA Ile UAU , and tRNA Ile AAU were aligned, followed by phylogenetic analysis.

Ancestral sequence estimation was performed using MEGA version 10.1.7 (Kumar et al. 2018 ), with the tree topologies inferred by IQ-TREE specified as user-defined trees. Ancestral sequences of LACA and LBCA were inferred under the GTR+G4 model (Nei and Kumar 2000 ; Kumar et al. 2018 ; Stecher et al. 2020 ). Given the limited length and information content of tRNA sequences, the GTR+G4 model was adopted as a practical approximation for ancestral sequence reconstruction, as more complex models are constrained by the weak phylogenetic signal in short tRNA alignments. In MEGA, gaps were treated as character states rather than ignored during ancestral sequence estimation. After excluding the terminal CCA sequence and the − 1 nucleotide of tRNA His , the aligned tRNA sequences were used for phylogenetic analysis and ancestral sequence reconstruction of LACA and LBCA for each tRNA species.

Ancestral sequences were inferred using a maximum likelihood approach. Because the phylogenetic trees are unrooted, we placed LACA and LBCA at opposite ends of the branch linking the archaeal cluster (containing mainly archaeal but some eukaryal sequences) and the bacterial cluster (containing mainly bacterial but with some eukaryal sequences). During ancestral sequence reconstruction in MEGA, the maximum probability of the inferred nucleotide at each site was recorded for each reconstructed ancestral tRNA sequence, in order to assess site-wise uncertainty in the ASR results.

Next, the ancestral sequences of LACA and LBCA for each tRNA were aligned, and phylogenetic inference and ancestral sequence reconstruction were performed as described above. The position of LUCA was inferred as the node joining the LACA and LBCA sequences for each tRNA species.

In Vitro Transcription of LUCA and E. coli tRNA Genes

tRNA Ala and tRNA Ser genes from both LUCA and E. coli were selected for anticodon sequence modifications. tRNA Ala and tRNA Ser were selected because their anticodons are not primary identity elements for aminoacylation, allowing anticodon modification without disrupting amino acid specificity. Genes in which the anticodons were modified to AGT, CGT, GGT, or TGT were synthesized and inserted into the pTwist Amp High Copy vector (Twist Bioscience, South San Francisco, CA, USA). As shown in Fig. 1 A, pairs of tRNA genes from LUCA and E. coli , corresponding to the same amino acid and sharing the same modified anticodon, were each placed under a T7 promoter and inserted into the vector. These constructs were used to amplify the respective tRNA genes. For tRNA Ser with anticodons AGT, CGT, and TGT, new gene constructs were synthesized and inserted into the pUC-GW-Amp vector (GENEWIZ, Azenta Life Sciences, South Plainfield, NJ, USA). The resulting plasmids were designated as: pTwist-Ala-AGU, pTwist-Ala-CGU, pTwist-Ala-GGU, pTwist-Ala-UGU, pUC-GW-Ser-AGU, pUC-GW-Ser-CGU, pUC-GW-Ser-GGU, and pUC-GW-Ser-UGU.

Monitor Genes of Codon Recognition

The expression vector for Arc1ΔThr gene (Shibue et al. 2018 ) was kindly provided by Dr. Satoshi Akanuma at Waseda University. The Arc1ΔThr gene, which lacks codons for threonine (Thr) and cysteine (Cys), was inserted between the Nde I and Bam HI sites of the pET23a vector, resulting in the plasmid pET23a- Arc1ΔThr . The Arc1ΔThr gene does not contain any tag sequences.

Five threonine “monitor” codons (ACA, ACC, ACG, or ACT) followed by ten histidine codons were added using the following procedure to the 5′ end of the Arc1ΔThr gene. First, inverse PCR was done on pET23a- Arc1ΔThr using primers Arc1ΔThr_inverse_first_to_5end and Arc1ΔThr_inverse_first_to_3end (see Supplementary Table S2). In this step, five Thr codons (ACA–ACC–ACG–ACT–ACG) were added to the 5′ end of the gene, along with three codons (GGT–TCT–TCT, encoding Gly–Ser–Ser) immediately downstream of the initiation codon. The resulting PCR product was self-ligated (circularized) and used for transformation of E. coli JM109 competent cells (ECOSTM Competent E. coli JM109; NIPPON GENE, Tokyo, Japan).

Plasmids were purified using the QIAquick Miniprep Kit (QIAGEN, Hilden, Germany), following the manufacturer’s instructions. The purified plasmid was then used as a template for a second inverse PCR using primers Arc1ΔThr_inverse_second_to_5end and Arc1ΔThr_inverse_second_to_3end, in order to insert ten His codons between the five Thr codons and the Arc1ΔThr gene. The resulting PCR product was again self-ligated and cloned.

Subsequently, the 5′ Thr codons were replaced with five copies of a single type of codon (either ACA, ACC, ACG, or ACT) via a third round of inverse PCR. This was done using primer Arc1ΔT_inverse_third_to_5end and one of the following reverse primers: Arc1ΔT_ACA5_inverse_third_to_3end, Arc1ΔT_ACC5_inverse_third_to_3end, Arc1ΔT_ACG5_inverse_third_to_3end, or Arc1ΔT_ACU5_inverse_third_to_3end. Each PCR product was again self-ligated and cloned.

All inverse PCR and self-ligation steps described above were completed using the KOD Plus Mutagenesis Kit (Toyobo Co., Ltd., Osaka, Japan). Oligonucleotides were synthesized by means of Hokkaido System Science Co., Ltd. (Sapporo, Japan), Eurofins Scientific SE (Luxembourg City, Luxembourg), or Integrated DNA Technologies, Inc. (Coralville, IA, USA). Self-ligated PCR products were used for the transformation of E. coli JM109 competent cells. Overnight cultures from selected colonies were used for plasmid purification using the QIAquick Miniprep Kit (QIAGEN).

The resulting plasmids were named: pET23a-ACA5-Arc1ΔThr, pET23a-ACC5-Arc1ΔThr, pET23a-ACG5-Arc1ΔThr, and pET23a-ACU5-Arc1ΔThr.

Furthermore, the HiBiT tag sequence was added to the 3′ end of each modified gene (as previously done for the test codons and His codons at the 5′ end) using a similar inverse PCR strategy. The resulting plasmids were designated: pET23a-ACA5-Arc1ΔThr-HiBiT, pET23a-ACC5-Arc1ΔThr-HiBiT, pET23a-ACG5-Arc1ΔThr-HiBiT, and pET23a-ACU5-Arc1ΔThr-HiBiT (see Fig. 1 B).

In Vitro Transcription of tRNA

Each vector carrying a tRNA gene was used to transform E. coli JM109 chemically competent cells (ECOSTM Competent E. coli JM109). Plasmids were then purified using the QIAquick Miniprep Kit. Each tRNA gene, along with its upstream T7 promoter, was amplified by PCR using primers listed in Supplementary Table S2.

PCR products were purified by agarose gel electrophoresis followed by extraction using the QIAquick Gel Purification Kit (QIAGEN, Hilden, Germany). In vitro transcription of each purified tRNA gene was done using the ScriptMAX ® Thermo T7 Transcription Kit (Toyobo Co., Ltd., Osaka, Japan). The resulting tRNA transcripts were purified with the NucleoSpin ® miRNA Kit (MACHEREY-NAGEL GmbH & Co. KG, Düren, Germany), according to the manufacturer’s protocol.

Purified transcribed tRNAs were analyzed by polyacrylamide gel electrophoresis (PAGE) with and without 7 M urea to assess the accuracy of transcription termination and to examine possible tertiary structural variations in the transcripts. The concentration of each tRNA was measured using a Qubit 3.0 Fluorometer (Thermo Fisher Scientific Inc., Waltham, MA, USA) with the Qubit RNA BR Assay Kit (Thermo Fisher Scientific).

In Vitro Translation and Analysis

In vitro translation was done using a customized PURE frex 1.0 system (GeneFrontier Corporation, Kashiwa, Chiba, Japan). In this customized version, threonine (Thr) was omitted from Solution I of the kit. Each reaction mixture (total volume: 20 µL) was prepared in a 200 µL PCR tube and contained a tRNA transcript with a test anticodon, a plasmid vector encoding a monitor gene containing the test codon, and the customized PURE frex solutions lacking Thr. The mixture was incubated at 37 °C for 4 h. As controls, a reaction containing the ACU-test codon monitor plasmid and supplemented with Thr was included, using the same customized PURE frex solution lacking endogenous Thr, and a reaction containing only customized PURE frex solution lacking endogenous Thr.

The results of the in vitro translation assays were analyzed by SDS–polyacrylamide gel electrophoresis (SDS–PAGE), followed by Coomassie Brilliant Blue (CBB) staining. The gel consisted of a 5% stacking layer and a 15% separating layer, using a Tris–Glycine buffer system.

Protein Quantification with HiBiT System

Luminescence (LUM) was measured using the Nano-Glo ® HiBiT Detection System (Promega Corporation, Madison, WI, USA). In each well of a white, opaque 96-well ProxiPlate (shallow well; Revvity, Inc., Waltham, MA, USA), 25 µL of stepwise-diluted cell-free protein synthesis reaction solution or HiBiT control protein was dispensed. Next, 25 µL of Nano-Glo ® HiBiT Lytic Reagent was added to each well. After mixing, the plate was incubated for 10 min to allow equilibration between LgBiT and HiBiT. Luminescence intensity was then measured using a VICTOR Nivo™ multimode plate reader (Revvity).

To address the possibility that luminescence values might exceed the detection range, each sample was subjected to serial dilutions. A buffer containing 0.05 M Tris-HCl (pH 6.5) and 0.05 M NaCl was used for dilution. For each sample, three plates were prepared, with three technical replicates per plate. The average luminescence from each plate was calculated, and these averages constituted one data set. A total of four such independent sets were obtained. Among the dilution series, values from the 500-fold dilution—falling within the linear detection range—were used for the final analysis.

Results and Discussion

In this study, we addressed three key questions regarding the early evolution of tRNA and the translation system: whether the evolutionary history of tRNA reflects that of ARSs; whether the inferred ancestral tRNA retains the identity elements necessary for aminoacylation; and whether such ancestral tRNA is functionally compatible with the modern translation system such as E. coli . Our analyses focus on these functional and evolutionary features rather than on detailed phylogenetic relationships among tRNAs.

Individual tRNA Phylogenies and Ancestral tRNA Sequences for LACA and LBCA

The phylogenetic analyses presented here are primarily intended to provide a framework for ancestral sequence reconstruction, rather than to resolve detailed evolutionary relationships among different tRNA species. Given the short length of tRNA sequences and their limited phylogenetic signal, the inferred topologies are interpreted cautiously and are used primarily for inferring ancestral sequences of LACA and LBCA.

Figure 2 provides an overview of the phylogenetic trees for each tRNA (for details, see Supplementary Fig. S1 ), under the assumption that LUCA is positioned on the branch connecting the archaeal and bacterial clusters (see above for details). Bootstrap support values are provided in Supplementary Fig. S1 . As expected given the short length of the tRNA alignment, the support for some deep internal nodes is limited. However, broader clustering patterns—particularly the separation between archaeal and bacterial lineages into distinct clusters—were generally recovered, with bootstrap support values often exceeding 80 in most tRNA datasets (except for tRNA Glu and tRNA Gln ). Accordingly, interpretations based on weakly supported nodes are treated with caution.

In this framework, several eukaryal tRNAs—namely tRNA Ala , tRNA Arg , tRNA Asn , tRNA Cys , and tRNA Lys —are placed within the bacterial clades in the inferred trees. In contrast, tRNA Gln , tRNA Glu , tRNA Gly , tRNA Ile , tRNA Phe , tRNA Pro , tRNA Ser , tRNA Thr , tRNA Trp , and tRNA Val tend to cluster within the archaeal clades. However, in the cases of tRNA Asp , tRNA His , and tRNA Tyr , archaeal sequences are positioned within eukaryal clades. In addition, eukaryal tRNA Leu forms a distinct third group.

Previous studies (e.g., Furukawa et al. 2017 ) have proposed that the eukaryal cytoplasmic ARSs are mosaic. Several ARSs (SerRS, GlyRS (α₂ type), GluRS, LeuRS, TrpRS, PheRS, CysRS, TyrRS, ProRS, and HisRS) are thought to have archaeal origins, whereas others (ThrRS, ValRS, IleRS, AspRS, ArgRS, and MetRS) are inferred to derive from bacterial lineages, in some cases involving horizontal gene transfer from archaea (IleRS, AspRS, and ArgRS—as well as one subgroup of eukaryal MetRS). In addition, eukaryal AsnRS and AlaRS include multiple evolutionary lineages of distinct origins (i.e., bacterial and archaeal). For LysRS, Class II LysRS is found in eukaryotes, most bacteria, and some archaea, whereas Class I LysRS is found in most archaea and a few bacteria. The bacterial affinity observed for eukaryal tRNA Lys in our analyses may therefore reflect co-evolution with Class II LysRS. Similarly, the archaeal affinities observed for several eukaryal tRNAs (tRNA Glu , tRNA Gly , tRNA Phe , tRNA Pro , tRNA Ser , and tRNA Trp )—and the bacterial affinity of eukaryal tRNA Arg —are broadly consistent with the hypothesis that tRNAs co-evolved with their corresponding ARSs.

However, discrepancies between tRNA and ARS phylogenetic patterns are also evident. For several tRNA species, the inferred placements do not match the expected origins based on ARS evolution, suggesting that the evolutionary histories of tRNAs and their cognate synthetases are not fully coupled.

While some topological relationships remain uncertain due to limited phylogenetic signal, these trees provide a consistent and practical framework for ancestral sequence inference and for exploring the early evolution of the translation system. Using this framework, ancestral tRNA sequences corresponding to LACA and LBCA were inferred for each tRNA species. To evaluate site-wise uncertainty in these reconstructions, we recorded the maximum probability of the inferred nucleotide at each site. These site-wise maximum probabilities for LACA and LBCA tRNAs are summarized in Supplementary Fig. S2.

Composite tRNA Tree and Ancestral tRNA Sequences of LUCA

To integrate the ancestral sequences inferred for individual tRNA species and to define the position of LUCA, we constructed a composite phylogenetic tree based on the ancestral tRNAs of LACA and LBCA. We inferred 21 pairs of ancestral tRNA sequences corresponding to the 20 standard amino acids, including both initiator and elongator tRNAs for methionine, from LACA and LBCA. Using these 42 ancestral tRNA sequences, we reconstructed a composite phylogenetic tree of ancestral tRNAs from LACA and LBCA (Fig. 3 ), following the approach of de Farias ( 2013 ). Phylogenetic inference was done using IQ-TREE with the K3Pu + F+I+G4 substitution model (Nguyen et al. 2014 ). The nodes connecting LACA and LBCA for each tRNA were interpreted as representing the positions of the LUCA.

However, tRNA Lys , tRNA Phe , and tRNA Thr appeared as paraphyletic groups in the ML tree (Fig. 3 A). To address this, a constraint tree was generated (Fig. 3 B), enforcing the monophyly of each tRNA species while preserving the inter-species relationships as closely as possible to those in the ML tree. Both the ML and constraint trees were inferred under the same model. The log-likelihoods of ML and constraint trees were − 1797.5 and − 1802.0, respectively—a difference of only 4.5. The constraint tree is not rejected by Kishino-Hasegawa test (Kishino and Hasegawa 1989 ) (unweighted and weighted), Shimodaira-Hasegawa test (Shimodaira and Hasegawa 1999 ) (unweighted and weighted), or the approximately unbiased test (Shimodaira and Goldman 2002 ). Therefore, we used the constraint tree (Fig. 3 B) for subsequent analyses, including estimation of the LUCA sequence for each tRNA species. These inferred sequences are presented in Fig. 4 . To evaluate site-wise uncertainty in the LUCA reconstructions, we recorded the maximum probability of the inferred nucleotide at each site. These site-wise maximum probabilities for LUCA tRNAs are summarized in Supplementary Fig. S3.

This composite tree should be regarded as an operational framework for ancestral sequence inference rather than a definitive representation of evolutionary relationships among tRNAs. Therefore, the following observations should be interpreted with caution.

The topology of the tRNA composite tree (Fig. 3 A and B), which served as the basis for ancestral sequence inference, does not match that of the ARS tree. In the tRNA composite tree, tRNAs corresponding to Class I and Class II ARSs do not form distinct clades. tRNA Ala and tRNA Asp appear in the upper half of the trees (Fig. 3 A and B), which roughly corresponds to tRNAs for Class I ARSs, while tRNA Cys , tRNA Leu , and tRNA Met s are found in the lower half, corresponding to Class II ARSs.

For subclass Ia ARSs, the relationship (MetRS, (LeuRS, (IleRS, ValRS)) is widely accepted (Nagel and Doolittle 1995 ). However, in our tRNA tree, tRNA Met , tRNA Leu , tRNA Ile , and tRNA Val are located in separate branches (Fig. 3 A and B). Similarly, tRNA Arg and tRNA Cys are apart from other tRNAs for Class Ia ARSs. Notably, tRNA Leu clusters with tRNA Ser , whose synthetase (SerRS) belongs to subclass IIa. Both tRNAs have long variable arms, which may reflect a shared evolutionary origin. In Candida albicans , tRNA Ser CAG is known to be leucylated (Suzuki et al. 1997 ; Zhou et al. 2013 ). However, this observed sister relationship could be an artifact, potentially due to the inclusion of gap-containing sites in the alignment or convergent features associated with their long variable arms.

In contrast, tRNA Trp and tRNA Tyr —both substrates of subclass Ic ARSs, form a group, despite bacterial tRNA Tyr having a long variable arm. If our phylogenetic inference is accurate, the long variable arm of bacterial tRNA Tyr may have emerged after divergence from the archaeal/eukaryal lineage.

tRNA Glu and tRNA Gln , which correspond to subclass Ib ARSs, also form a group. Their sister group is tRNA Asp , a substrate of a subclass IIb ARS. However, tRNA Asn appears in a separate position in the tree. Interestingly, Helicobacter pylori AspRS has been reported to possess glutamylation activity toward tRNA Glu in addition to its canonical role in aspartylation of tRNA Asp and tRNA Asn (Rathnayake and Hendrickson 2019 ), which may be related to the proximity of tRNA Asp and tRNA Glu in our phylogenetic tree.

All tRNAs corresponding to Class IIa ARSs—namely GlyRS (α₂), HisRS, ThrRS, ProRS, and SerRS, are in “Class II clade” in the tRNA composite tree, along with tRNA Asn and some tRNAs associated with Class I ARSs (e.g., tRNA Me ts, tRNA Cys , and tRNA Ser ) (Fig. 3 A and B). Notably, tRNA His and tRNA Gly form a group. HisRS and GlyRS (α 2 ) are members of subclass IIa ARS. Although GlyRS is believed to have diverged early within subclass IIa, GlyRS (α₂) has been shown to group with HisRS under certain conditions in molecular trees (Furukawa et al. 2022 ), consistent with the topology of our tree, although this pattern may also be influenced by methodological factors.

It should be noted that the ASR approach used here relies on a homogeneous substitution model (GTR+G4), which may not fully capture site-specific evolutionary constraints. More sophisticated models, such as site-heterogeneous models, may improve reconstruction accuracy and should be explored in future studies. Additionally, the treatment of gaps in the alignment may influence inferred ancestral sequences, representing another source of uncertainty. These methodological limitations, together with the relatively short length of tRNA sequences, suggest that some inferred relationships and ancestral states should be interpreted with caution. The observed incongruence between tRNA and ARS phylogenies may reflect genuine biological differences, but could also be influenced by methodological limitations inherent to short tRNA sequences and model assumptions.

tRNA Identity Elements in Ancestral tRNAs and Their Evolution

We next examined whether ancestral tRNAs retain identity elements required for aminoacylation, as this provides insight into the functional continuity of the translation system from the LUCA to extant organisms. In evaluating these features, we took into account the site-wise maximum probabilities obtained during ancestral sequence reconstruction. Most reconstructed sites showed maximum probabilities greater than 0.9, and most of the conserved residues and identity elements discussed below fell within these high-probability sites. Thus, the main conclusions concerning conserved sequence features are based largely on sites with relatively high reconstruction confidence.

Identity elements of tRNA are sequences and structural features recognized by their cognate ARSs. In Fig. 4 , we show that LUCA, LACA, and LBCA tRNAs seem to retain identity elements mostly studied for E. coli tRNAs and less so for Thermus thermophilus , Saccharomyces cerevisiae , Aeropyrum pernix , and other species’ tRNAs (see Giegé and Eriani 2023 ). In most cases, these identity elements are well conserved in ancestral tRNAs. In addition, in Fig. 4 , the LUCA, LACA, and LBCA tRNAs are compared with some bacterial, archaeal, and eukaryal tRNAs including E. coli , T. thermophilus , and S. cerevisiae .

tRNA Ala : The major identity determinant of tRNA Ala is the G 3 -U 70 wobble pair, rather than the anticodon, which governs the alanylation of E. coli tRNA Ala (Hou and Schimmel 1988 ; McClain et al. 1988 ; Park et al. 1989 ). This wobble pair, along with G 2 -C 71 and A 73 , is well conserved across LACA, LBCA, and LUCA tRNA Ala (Fig. 4 ).

tRNA Arg : The major identity determinants in E. coli tRNA Arg include A 20 , C 35 and U 36 /G 36 and the discriminator base A 73 /G 73 (McClain and Foss 1988 ; Schulman and Pelka 1989 ; Tamura et al. 1992 ) A 20 is the key determinant for arginylation by E. coli ArgRS. These elements are conserved in LUCA, LBCA and LACA tRNA Arg (discriminator G 73 ) (Fig. 4 ).

tRNA Asn . The major identity determinants of E. coli tRNA Asn are the anticodon (G 34 , U 35 , and U 36 ) and the discriminator base G 73 (Li et al. 1993 ); LACA, LBCA and LUCA tRNA Asn all retain G 34 , U 35 , U 36 , A 37 and G 73 (Fig. 4 ).

tRNA Asp : Conserved identity elements of tRNA Asp across S. cerevisiae , E. coli , and T. thermophilus include 73 , G 34 , U 35 , C 36 , and C 38 (Nameki et al. 1992 ; Giegé et al. 1996 , 1998 ). All these are present in LACA, LBCA and LUCA tRNA Asp (Fig. 4 ).

tRNA Cys : The discriminator U 73 and the anticodon G 34 -C 35 -A 36 are major identity determinants in E. coli (Pallanck et al. 1992 ; Komatsoulis and Abelson 1993 ), and all of these are conserved in LUCA, LBCA and LACA tRNA Cys (Fig. 4 ). However, G 15 :G 48 of an important tRNA element in E. coli tRNA Cys (Hou et al. 1993 ; Hauenstein et al. 2004 ) is absent in LUCA, LBCA and LACA tRNA Cys .

tRNA Gln : Identity elements for tRNA Gln in E. coli include G 73 , U 1 -A 72 , G 2 -C 71 , G 3 -C 70 , U 34 , U 35 , and G 36 (Jahn et al. 1991 ). The base pair of 10th and 25th nucleotides and the U 35 are also important binding sites for E. coli GlnRS, with G 36 contributing to both binding and recognition (Ibba et al. 1996 ). LACA, LBCA and LUCA tRNA Gln retain the G 2 -C 71 , G 10 , and anticodon sequences (U 34 /C 34 U 35 G 36 ), respectively (Fig. 4 ). However, LACA and LUCA tRNA Gln have A 73 instead of G 73 . Ancestral trait of tRNA Gln identity might be G 2 -C 71 , G 10 , anticodon sequences (U 34 /C 34 U 35 G 36 ) and discriminator A 73 .

tRNA Glu : Strong identity determinants of tRNA Glu are U 34 , U 35 , C 36 and A 37 , along with U 11 -A 24 , U 13 -G 22 :A 46 and the absence of 47th nucleotide serve as major identity determinants for tRNA Glu (Sekine et al. 1996 ). LACA tRNA Glu retains U 34 , while LBCA and LUCA tRNA Glu substitute it with C 34 (Fig. 4 ). The presence of U 11 -A 24 and U 13 -G 22 :A 46 and the absence of 47th nucleotide is also common to the three ancestral tRNA Glu s. The ancestral trait identity of tRNA Glu appears to be C 34 /U 34 , U 35 , C 36 and A 37 , U 11 -A 24 , U 13 -G 22 :A 46 , and the absence of 47th nucleotide.

tRNA Gly : In bacterial tRNA Gly , the discriminator base U 73 is a critical identity element; however, A 73 is typical of eukaryal and archaeal tRNA Gly . Additionally, the sequences G 1 -C 72 pair, C 35 , C 36 , and the (G 10 -Y 25 ):G 45 triplet are conserved (Mazauric et al. 1998 ). The tRNA Gly of LBCA retains the bacterial U 73 , while those of LACA and LUCA possess A 73 . The mutation of U 73 of E. coli tRNA Gly reduces glycylation activity, indicating that U 73 functions as an identity element for GlyRS (Shimizu et al. 1992 ). By contrast, in the archaeon A. pernix , the A 73 discriminator is not essential for tRNA Gly identity (Okamoto et al. 2005 ). Further mutational analyses in E. coli , T. thermophilus , and S. cerevisiae have identified a broader identity set that includes the first two base pairs of the acceptor stem (G 1 –C 72 and C 2 –G 71 ) as well as the anticodon bases C 35 and C 36 (Mazauric et al. 1999 ). In T. thermophilus , additional identity elements have been identified, including the C 50 –G 64 base pair, G 10 , and U 16 (Mazauric et al. 1999 ). tRNA Gly in LACA, LBCA, and LUCA retain several conserved features: G 1 –C 72 , C 2 –G 71 , G 3 –C 70 , and the anticodon nucleotides C 35 and C 36 . The identity elements C 50 –G 64 and G 10 are also conserved among these ancestral lineages.

tRNA His : In E. coli (Himeno et al. 1989 )d cerevisiae tRNA His (Nameki et al. 1995 ), G − 1 and anticodon (G 34 , U 35 , G 36 ) have been shown to play an essential role in identity. In Bacteria, the discriminator C 73 forms a base pair with G − 1 , while the discriminator is A 73 in Eukarya (cytoplasm). Both C 73 and A 73 are found in archaeal tRNA His (Fig. 4 ). The G − 1 -C 73 pair is important for the recognition by the HisRS in E. coli . LACA and LUCA tRNA His have A 73 , while LBCA tRNA His has C 73 .

tRNA Ile : In tRNA Ile , key identity determinants include the discriminator base (A 73 ) and anticodon nucleotides (G 34 /C 34 , A 35 , U 36 ) followed by A 37 (Muramatsu et al. 1988 ; Normanly et al. 1990 ; Pallanck and Schulman 1991 ). In E. coli , additional elements critical for isoleucylation and editing are found in the stem region (C 4 -G 69 , U 12 -G 23 , C 29 -G 41) and the D-loop (G 16 , D 20 , D 21 ) (Nureki et al. 1994 ; Farrow et al. 1999 ; Hendrickson 2001 ). These are conserved in LUCA, LBCA and LACA tRNA Ile except for G 16 .

tRNA Leu : In E. coli tRNA Leu , A 73 is important for leucine identity, while the anticodon and long variable arm are less significant (Asahara et al. 1993 ). Mutagenesis studies have emphasized the importance of tertiary interactions, such as G 18 :U 55 , G 19 :C 56 and U 54 :A 58 in maintaining leucine identity (Du and Wang 2003 ). These are conserved in LUCA, LACA, and LBCA tRNA Leu .

tRNA Lys : The A 73 of E. coli tRNA Lys plays a critical role in lysine identity (McClain et al. 1990 ). Additionally, substitution experiments indicate that all three anticodon bases (U 34 , U 35 , U 36 ) contribute to lysine identity (Tamura et al. 1992 ). LACA, LBCA and LUCA tRNA Lys share U 34 , U 35 and U 36 . However, unlike LBCA and LUCA tRNA Lys , LACA tRNA Lys has G 73 instead of A 73 . This suggests that A 73 is an ancestral trait for tRNA Lys .

tRNA Met : The anticodon (C 34 , A 35 , U 36 ) is a major identity element in tRNA Met (Schulman and Pelka 1983 , 1988 ). Additionally, the tRNA fMet -formyltransferase recognition elements (1–72 mismatch pairs and R 11 :Y 24 base pairs) are known in tRNA i fMet (Guillon et al. 1992b ). Furthermore, G 29 -C 41 , G 30 -C 40 , and G 31 -G 39 are important for E. coli tRNA i fMet to play a role as initiator tRNA but not as elongator (Guillon et al. 1993 ). Among these, the 1–72 mismatch pair is not retained in LACA and LUCA tRNA i Met . This matches with the observation that in Archaea and Eukarya, the amino acid corresponding to the start codon is methionine (Met) rather than formylmethionine (fMet) (Ramesh and RajBhandary 2001 ; Schmitt et al. 2020 ).

tRNA Phe : The tRNA Phe identity elements are the anticodon triplet G 34 , A 35 , and A 36 and the discriminator A 73 (Peterson and Uhlenbeck 1992 ). Additional elements critical for phenylalanylation include U 20 and U 59 (Peterson and Uhlenbeck 1992 ; Peterson et al. 1993 ). LACA, LBCA and LUCA tRNA Phe retain G 34 , A 35 , A 36 , U 59 , and A 73 . However, U 20 is conserved only in LBCA and LUCA tRNA Phe , and is absent in LACA tRNA Phe .

tRNA Pro : Identity elements of E. coli tRNA Pro include G 35 , G 36 , A 37 , G 72 , A 73 , and the G 15 :C 48 pair (Liu et al. 1995 ; Hasegawa and Yokogawa 2000 ). In archaeon A. pernix tRNA Pro , identity is also influenced by the G 1 –C 72 base pair and anticodon (Yokozawa et al. 2003 ). All ancestral tRNAs retain G 15 :C 48 , G 35 , G 36 , and A 73 . However, A 37 is absent from all ancestral sequences, suggesting a later acquisition. LBCA tRNA Pro additionally retains C 1 -G 72 , while LACA and LUCA tRNA Pro have G 1 -C 72 , suggesting that G 1 -C 72 is the ancestral trait of tRNA Pro .

tRNA Ser : Serine identity is predominantly determined by the presence of a large variable arm rather than the anticodon (Normanly et al. 1992 ). In E. coli , conserved features include G 1 -C 72 and G 2 -C 71 across tRNA Ser isoacceptors, (Jühling et al. 2009 ). Both E. coli and human SerRS recognize tRNA Ser through its large variable arm and the discriminator G 73 (Asahara et al. 1994 ; Breitschopf and Gross 1994 , 1996 ; Breitschopf et al. 1995 ). All ancestral tRNAs exhibit large variable arms and retain G 1 -C 72 , G 4 -C 69 , C 11 -G 24 , G 30 -C 40 , and G 73 . LBCA retains G 2 -C 71 , while LACA and LUCA possess the reverse pair C 2 -G 71 .

tRNA Thr : Unlike many other tRNAs, the discriminator base is not essential for tRNA Thr identity, though exceptions exist in species such as Haloferax volcanii and S. cerevisiae (Nameki 1995 ; Ishikura et al. 2000 ). Across five species ( E. coli , T. thermophilus , H. volcanii , A. pernix and S. cerevisiae ), the identity elements comprise acceptor stem base pairs (G 1 -C 72 , C 2 -G 71 , G 4 -C 69 ) and anticodon nucleotides G 35 and G 36 (Hasegawa et al. 1992 ; Nameki 1995 ; Nameki et al. 1996 ; Ishikura et al. 2000 ; Iwaki et al. 2002 ). The primary identity elements consist of base pairs in the acceptor stem (G 1 -C 72 , C 2 -G 71 , G 4 -C 69 ) and anticodon nucleotides (G 35 and G 36 ). All ancestral tRNA Thr sequences examined retain these features.

tRNA Trp : In E. coli and Bacillus subtilis , identity elements for tRNA Trp include the discriminator base G 73 , the anticodon (C 34 , C 35 , A 36 ), and the A 1 -U 72 base pair (Himeno et al. 1991 ; Xue et al. 1993 ). Additional acceptor stem base pairs (G 2 -C 71 , G 3 -C 70 , G 4 -C 69 ) are also essential for B. subtilis tRNA Trp (Xu et al. 2002 ). These base pairs are also present in E. coli tRNA Trp . In ancestral tRNA Trp s, G 2 -C 71 , G 3 -C 70 , and the three anticodon bases are conserved, but they have G 1 -C 72 instead of A 1 -U 72 . In addition, LACA and LUCA tRNA Trp have A 73 instead of G 73 . The combination of A 73 and G 1 -C 72 in LACA and LUCA tRNA Trp mirrors the identity pattern observed in A. pernix tRNA Trp (Tsuchiya and Hasegawa 2009 ), suggesting that archaeal-type traits were retained in tRNA Trp .

tRNA Tyr : Bacterial tRNA Tyr is a type II tRNA with a long extra arm, whereas archaeal and eukaryal tRNA Tyr is a type I tRNA with a short extra arm. Our phylogeny suggests that LUCA tRNA Tyr was a type I tRNA and that bacterial type II tRNA Tyr emerged after the divergence of Bacteria from the archaeal/eukaryal lineage. Identity depends on the discriminator A 73 and anticodon U 35 (Bedouelle 1990 ), together with the first base pair (C 1 -G 72 in archaea/eukaryotes; G 1 -C 72 in bacteria) (Fechter et al. 2000 ; Iwaki et al. 2002 ; Bonnefond et al. 2005 ). LACA, LBCA and LUCA tRNA Tyr share G 34 , U 35 and A 73 . LACA tRNA Tyr retains the archaeal-type C 1 -G 72 , while LBCA and LUCA tRNA Tyr share the bacterial-type G 1 -C 72 .

tRNA Val identity: The identity elements of tRNA Val overlap with those of tRNA Ile and tRNA Met , sharing A 73 and A 35 . Additionally, C 36 is a critical identity element, while G 20 and G 45 are minor recognition elements (Horowitz et al. 1999 ). In E. coli , mutations of base pairs G 3 -C 70 or U 4 -A 69 in the acceptor stem significantly affect activity (Tamura et al. 1991 ). Both LUCA and LBCA tRNA Val lack U 4 -G 69 . LACA, LBCA, and LUCA tRNA Val lacks G 20 , but G 20 is also not conserved among E. coli tRNA Val isoacceptors. Ancestral traits of tRNA Val identity seem to be G 3 -C 70 , A 35 , C 36 , and A 73 .

Identity of tRNAs for transamidosome: The genes for AsnRS and GlnRS genes are absent in various bacterial and archaeal genomes (Curnow et al. 1996 ; Furukawa et al. 2017 ). In those organisms, Asn-tRNA Asn and Gln-tRNA Gln are synthesized via a two-step indirect process. First, non-discriminatory AspRS (ND-AspRS) and GluRS (ND-GluRS) aminoacylate the tRNA to form Asp-tRNA Asn and Glu-tRNA Gln , respectively. Second, Asp-tRNA Asn and Glu-tRNA Gln are amidated in transamidosomes to Asn-tRNA Asn and Gln-tRNA Gln , respectively (aaRS: tRNA: AdT, AdT is amidotransferase)(Bailly et al. 2007 ).

In the transamidosome, amidation of Asp-tRNA Asn is promoted by the U 1 -A 72 , whereas the presence of G 1 -C 72 pair and U 20a in the D-loop of tRNA Asp inhibits this process (Bailly et al. 2006 ). The bacterial AdT (GatCAB) has a region that distinguishes between U 1 -A 72 and G 1 -C 72 , while archaeal GatCAB lacks this specificity and can accept aminoacyl-tRNA with a G 1 -C 72 base pair (Nakamura et al. 2010 ). In archaeal AdTs, the Asp-to-Asn conversion is promoted primarily by G 46 and U 47 in the variable region (Bailly et al. 2006 ); a similar recognition mechanism is thought to operate for Gln-tRNA Gln formation.

AsnRS and GlnRS are believed to have evolved secondarily from AspRS and GluRS, respectively (Woese et al. 2000 ). AsnRS shares a common ancestor with archaeal AspRS, while GlnRS shares a common ancestor with eukaryal GluRS. Among AspRS, ND-AspRS is considered ancestral (Nair et al. 2016 ), and similarly, ND-GluRS is thought to be the ancestral form of GluRS (Siatecka et al. 1998 ). Therefore, the AspRS and GluRS enzymes of LACA, LBCA, and LUCA are likely to have been the ND type.

Regarding sequence features, LACA, LBCA and LUCA tRNA Asn all retain G 46 and U 47 in the variable region. However, while LACA and LUCA tRNA Asn have G 1 -C 72 , LBCA tRNA Asn contains U 1 -A 72 , promoting amidation. Conversely, all three ancestral tRNA Asp retain U 20a , a feature that inhibits recognition by bacterial transamidosomes. These findings suggest that LACA and LUCA tRNA Asn were not compatible with bacterial-type AdTs but instead were likely recognized by archaeal-type AdTs. In contrast, LBCA tRNA Asn is likely to have been compatible with the bacterial-type transamidosome.

An important limitation of ancestral sequence reconstruction is that the maximum likelihood sequence represents only one of many possible ancestral states. Therefore, the inferred ancestral tRNAs analyzed here should be interpreted as representative sequences rather than exact historical reconstructions. Although most conserved residues and identity elements discussed above were inferred with high site-wise maximum probabilities, alternative ancestral states at some positions may still influence the interpretation. Future work could incorporate probabilistic sampling of ancestral sequences to be better account for this uncertainty and to evaluate the robustness of these inferences.

Can Inferred tRNAs Function in E. coli Translation System?

To assess the functional relevance of the inferred sequences, we evaluated whether LUCA tRNAs are compatible with a modern translation system ( E. coli ). When the inferred LUCA tRNA sequences were compared with their cognate E. coli counterparts (Supplementary Fig. S4), sequence identities ranged from 50 to 85.1% (Supplementary Table S3). The lowest sequence identity was observed for tRNA Glu with only 50.0–51.4% identity between LUCA and E. coli sequences.

As discussed above, most of the inferred LUCA tRNAs retain bacterial identity elements, particularly those characterized in E. coli (Fig. 4 ), suggesting that they may be partially compatible with the E. coli system. Specifically, the following LUCA tRNAs were found to include all known bacterial identity elements: tRNA Ala , tRNA Asn , tRNA Asp , tRNA Lys , tRNA i Met , tRNA e Met , and tRNA Phe . Other LUCA tRNAs did not retain the full set of bacterial identity elements, which is not unexpected given that identity elements are not necessarily conserved among bacteria, archaea, and eukaryotes.

However, several LUCA tRNAs may lack key residues required for efficient aminoacylation in E. coli . One such case is LUCA tRNA Gly , which has A 73 instead of the canonical U 73 . Previous studies have shown that this substitution reduces glycylation activity approximately 11-fold in the E. coli system (Shimizu et al. 1992 ; Nameki et al. 1997 ). Another example is LUCA tRNA His , which carries A 73 instead of C 73 , which is highly conserved in bacterial tRNA His . A study by Rosen et al. ( 2006 ) on the minihelix of tRNA His suggested that the substitution of C 73 by A 73 in the minihelix does not affect the histidylation activity of E. coli HisRS, indicating that LUCA tRNA His may still function in the E. coli system.

A more critical exception is the LUCA tRNA i Met . In E. coli , the initiator tRNA i fMet contains a conserved mismatch at the first base pair of the acceptor stem (C 1 -A 72 ) – a signature feature essential for selective recognition by methionyl-tRNA formyltransferase (Lee et al. 1991 ). The absence of N-formylation of Met on tRNA fMet was reported to reduce translation activity in the E. coli system ((Guillon et al. 1992a ). It is also important for the recognition by translational initiation factor 2 (IF2) (Guenneugues et al. 2000 ). In contrast, LUCA tRNA i Met has A 1 -U 72 instead of the C 1 -A 72 mismatch. This suggests that LUCA tRNA i Met may not be a suitable substrate for E. coli ’s formylation machinery and could be inefficient in supporting translation initiation.

Furthermore, previous studies have shown that tRNAs lacking post-transcriptional modifications often exhibit reduced translational activity, particularly for tRNAs corresponding to cysteine, glutamate, isoleucine, lysine, and phenylalanine in E. coli , and isoleucine and phenylalanine in S. cerevisiae (Giegé and Eriani 2023 ). Modifications at the wobble position (position 34) and adjacent to the anticodon (position 37) are known to play crucial roles in codon recognition, identity, and aminoacylation (e.g. Nureki et al. 1994 ; Senger et al. 1997 ; Madore et al. 1999 ; Liu et al. 2007 ; Giegé and Lapointe 2009 ; Wang et al. 2022 ). In particular, modifications at residue 34 of the wobble position or at position 37 adjacent to the anticodon can be important as identity elements.

In summary, while many LUCA tRNAs appear to retain sufficient identity elements to function in modern translation systems such as that of E. coli , it is crucial to experimentally assess the functionality of each individual LUCA tRNA species in such systems. Therefore, compatibility of LUCA tRNAs with the E. coli translation system should be regarded as a testable inference based on representative reconstructed sequences.

Testing the Function of LUCA tRNA in the Translation System

We experimentally tested whether selected LUCA tRNAs can decode codons and support protein synthesis in vitro. Arc1 is the inferred nucleotide diphosphate kinase (NDK) of LACA (Akanuma et al. 2013 ). Shibue et al. ( 2018 ) used Arc1 variants to investigate the minimal number of amino acid species required to form a functional protein. One such mutant, Arc1ΔThr, lacks codons for threonine (Thr) and cysteine (Cys).

When a particular amino acid is excluded from the translation system, its corresponding codon cannot be decoded—even if the relevant ARS and tRNA are present—because aminoacylation does not occur. The uncharged tRNA cannot form a ternary complex with EF-Tu·GTP and is therefore not delivered to the ribosome. Even if the uncharged tRNA enters the A-site, translation is halted due to the absence of an attached amino acid. However, if an exogenous tRNA is introduced that carries a different amino acid and can decode the excluded codon, translation may resume.

To evaluate decoding ACN codons (where N = A, C, G, or T), we inserted a sequence encoding five consecutive test codons (ACT X 5, ACC X 5, ACA X 5, or ACG X 5) and ten His codons at the 5′ end of the Arc1ΔThr gene in the pET23a vector (see Materials and Methods and Fig. 1 B). This design allows testing whether specific tRNA species can decode ACN codons and thereby enable synthesis of Arc1ΔThr.

As discussed above, the functionality of individual LUCA tRNA species in modern translation systems such as that of E. coli needs to be evaluated. As a first step, we selected two LUCA tRNA species—tRNA Ala and tRNA Ser , for functional testing. It is well established that the anticodon sequences of tRNA Ala and tRNA Ser are not identity elements for their respective aminoacylation activities (Hou and Schimmel 1988 ; McClain et al. 1988 ; Park et al. 1989 ; Normanly et al. 1992 ). Therefore, altering the anticodon of tRNA Ala and tRNA Ser does not affect their alanylation or serylation efficiency, nor their amino acid specificity.

To test whether LUCA tRNA Ala and tRNA Ser can function in the E. coli translation system and to explore a minimal anticodon set functional without posttranscriptional modifications, we synthesized E. coli and LUCA tRNA Ala and tRNA Ser variants with four different anticodon sequences (either of AGU, CGU, GGU, and UGU), resulting in 16 tRNA species in total (Fig. 5 ).

The 7 M Urea acrylamide gel electrophoresis confirmed that the in vitro transcripts of E. coli and LUCA tRNA Ala and tRNA Ser variants were successfully amplified as single bands (Supplementary Fig. S5). Native acrylamide gel electrophoresis also showed predominantly single bands, although faint additional bands were observed in some cases (data not shown). These results suggest that the tRNA transcripts generally adopt a single, stable structural form.

In Vitro Translation of Arc1ΔThr Containing Test Codons Using E. coli and LUCA tRNA Ala Anticodon Sequence Variants

We first tested decoding ACN codons using in vitro transcripts of E. coli and LUCA tRNA Ala variants with anticodons UGU, CGU, AGU, and GGU. As shown in Fig. 6 , the ACC and ACU codons were clearly decoded by E. coli tRNA Ala AGU and tRNA Ala GGU . Similarly, LUCA tRNA Ala AGU and tRNA Ala GGU appeared to decode ACC and ACU codons as well. These results indicate that LUCA tRNA Ala can function in the E. coli translation system.

Although faint bands corresponding to Arc1ΔThr were observed in other tRNA–codon combinations, their translation efficiency remained unclear. To evaluate translation activity with tRNA transcripts quantitatively with higher sensitivity we used HiBiT luminescence detection system (Dixon et al. 2016 ). The HiBiT tag sequence was added to the test genes as described in Materials and Methods section (Fig. 1 B). Luminescence intensities were used to estimate the relative amounts of translated Arc1ΔThr (Fig. 7 and Supplementary Fig. S6).

In E. coli tRNA Ala mutants (Fig. 7 A and B), all four anticodon variants—UGU, CGU, AGU, and GGU—were able to decode ACU and ACC codons. E. coli tRNA Ala UGU and tRNA Ala CGU weakly decoded the ACA and ACG codon, respectively. Decoding of ACA codon by E. coli tRNA Ala GGU was marginal. Similar decoding patterns were observed with the LUCA tRNA Ala mutants (Fig. 7 C and D), where all four variants could decode ACU and ACC codons. LUCA tRNA Ala CGU also weakly decoded ACG codon, while the decoding of ACA by LUCA tRNA Ala GGU was marginal.

In contrast to the tRNA Ser mutants, the translation efficiencies of tRNA Ala were substantially higher (Supplementary Fig. S6). For E. coli tRNA Ser mutants (Supplementary Fig. S6A and S6B), tRNA Ser AGU and tRNA Ser GGU appeared to decode ACU and ACC codons, while tRNA Ser CGU weakly decoded ACU and ACG codons. Among LUCA tRNA Ser mutants (Supplementary Fig. S6C and S6D), tRNA Ser AGU appeared to decode ACU codon, and tRNA Ser UGU , tRNA Ser CGU , and tRNA Ser GGU were able to decode both ACU and ACC codons. Because experiments with tRNA Ser mutants were conducted only once, these results should be interpreted with caution, and standard error bars are not provided.

The lower translation efficiency of both E. coli and LUCA tRNA Ser variants compared to tRNA Ala variants may be due to several factors. (1) Differences in the catalytic efficiency of aminoacylation between AlaRS and SerRS in the PURE frex working solution. (2) Structural stability of tRNA Ala and tRNA Ser in vitro transcripts. tRNA Ala is a class I tRNA with a small variable loop (76 nucleotides including the CCA terminus), while tRNA Ser is a class II tRNA with a large variable arm (88 nucleotides including the CCA terminus), which may reduce folding stability or translational efficiency. (3) Differences in the binding affinity of tRNA Ala and tRNA Ser to EF-Tu·GTP may also impact their delivery to the ribosome and overall translation efficiency.

The estimated abundance of native E. coli tRNA Thr capable of decoding each ACN codon is as follows (Dong et al. 1996 ). ACU: tRNA Thr GGU + tRNA Thr UGU = 3.28%, ACC: tRNA Thr GGU = 1.86%, ACA: tRNA Thr UGU = 1.42%, and ACG: tRNA Thr CGU + tRNA Thr UGU = 2.26%. The concentration of total tRNA in the PURE frex 1.0 reaction solution in this study was calculated as approximately 46 µM. Therefore, the tRNA Thr concentration decoding each ACN codon ranges from 0.6 to 1.5 µM. In the positive control with ACU-codon test gene, the decoding tRNA Thr species were present at an estimated concentration of ~ 1.5 µM. By comparison, the concentration of the synthetic tRNA transcripts in the reaction solution was 0.8 µM—roughly half that of the native ACU-decoding E. coli tRNA Thr .

Direct comparison of translation efficiency between native E. coli tRNA Thr and engineered E. coli or LUCA tRNA Ala and tRNA Ser (lacking post-transcriptional modifications) is inherently challenging. Nonetheless, several E. coli and LUCA tRNA Ala variants, depending on their anticodon–codon pairing, exhibited translation efficiencies comparable to native E. coli tRNA Thr . In contrast, tRNA Ser variants from both sources generally exhibited lower translation efficiency than tRNA Ala variants. This may be attributed to the structural complexity of Class II tRNA Ser , including its large variable arm, which could lead to misfolded or translationally incompetent forms. Differences in the activity of AlaRS and SerRS in the PURE frex system may also have contributed.

Wobble Basepairs Between LUCA tRNA and mRNA

Finally, we examined codon–anticodon pairing properties of LUCA tRNAs to infer characteristics of primitive decoding systems. As described above, the correspondence between the first anticodon position and the third codon position to the decoding of the ACN codon using tRNA Ala variants in this study can be summarized as shown in Fig. 8 . Our results demonstrate that the anticodons with U, C, A, or even G at the wobble position can decode codons with U and C in the third position. These observations do not strictly conform to either Crick’s original wobble hypothesis or the extended wobble rules (Table 1 ; Fig. 8 ).

The ability of G at the first anticodon position to decode U and C at the third codon position is consistent with Crick’s wobble rule. Likewise, its potential to decode A at the third codon position agrees with the extended wobble framework (Watanabe and Yokobori 2014 ). In contrast, the observation that all tested anticodon variants—including those with U, C, and A at the first position, can decode both U and C codon endings is contradictory to conventional expectation.

According to the extended wobble rule, U at the first anticodon position is capable of decoding any of the four nucleotides at the third codon position (U, C, A, or G), a phenomenon reported in metazoan mitochondria as well as certain bacteria and archaea, such as Mycoplasma capricolum (Inagaki et al. 1995 ; Yokobori et al. 2013 ). However, in M. capricolum tRNA Thr UGU , the ACC codon is read less efficiently than ACU, ACA, and ACG (Inagaki et al. 1995 ). Similarly, studies have shown that U at the first anticodon position can misread U and C at the third codon position in vivo with relatively high efficiency (Manickam et al. 2014 ), especially in systems lacking competing tRNAs—as was the case in our in vitro condition. This is consistent with the fact that U 34 can decode the third letter of the codons U and C.

Unmodified E. coli tRNA Gly UCC has also been shown to decode all four GGN codons, whereas unmodified E. coli tRNA Gly GCC and tRNA Gly CCC fail to decode at least one of them (Kompatscher et al. 2023 ). Also, in Crick’s wobble rule, unmodified U at anticodon first position also decodes A and G in the codon third position. In our study, neither E. coli nor LUCA tRNA Ala variants efficiently decoded A or G at the third codon position, even when U was present at the wobble position. This may reflect structural limitations in forming U 34 :A 3 and U 34 :G 3 pairs in our system (A 3 and G 3 are at the third codon position), possibly due to conformational constraints in the anticodon loop.

The case of C at the first anticodon position is particularly noteworthy. According to both Crick’s and the extended wobble rules, C at the first anticodon position should pair only with G at the third codon position. While rare exceptions suggest that A at the third codon position may be decoded via context-dependent mechanisms such as base 37 modification or loop sequence effects (Watanabe and Yokobori 2011 , 2014 ), there is no known example of C at the first anticodon position decoding U or C at the third codon position. Our results, which show the apparent decoding of U and C at the third codon position by C at the first anticodon position, are not readily explained by established wobble rules and suggest the possibility of broader flexibility in decoding under simplified or primitive conditions.

Similarly, A at the first anticodon first position has been proposed to decode all four nucleotides at the third codon position in certain mitochondrial and bacterial systems with limited number of tRNA species (Inagaki et al. 1995 ; Watanabe and Yokobori 2014 ). In the case of M. capricolum tRNA Thr AGU , the ACA codon is read poorly so when compared with ACU, ACC, and ACG (Inagaki et al. 1995 ). Therefore, it is not surprising that the A at the first anticodon position decodes the U and C at the third codon position. Yet in our results, both E. coli and LUCA tRNA Ala with A at the first anticodon position showed no decoding efficiency for A and G at the third codon position (Fig. 7 ). This again suggests that the intrinsic properties of the anticodon loop—rather than base pairing rules alone, may impose constraints on wobble flexibility.

It is also worth noting that the test constructs used in our experiments contained five consecutive instances of the same test codon. This repetition amplifies the effects of small differences in decoding efficiency: for example, a per-codon efficiency of 0.1, 0.5, and 0.9 results in cumulative translation success of only ~ 10 − 5 , 0.03, and 0.69 for five identical codons. Conversely, even low per-codon misreading rates (e.g., 1%) yield virtually no detectable full-length product over five codons (probability ≈ 10⁻ 10 ). Thus, our experimental system is highly sensitive to decoding accuracy and selectivity.

In summary, our results suggest that unmodified synthetic tRNAs, especially those reconstructed from (inferred) LUCA sequences, can exhibit broader wobble decoding flexibility than previously expected, particularly in simplified in vitro translation systems. These findings may reflect ancestral decoding mechanisms before the evolutionary emergence of anticodon modifications and refinements in codon discrimination.

Concluding Remarks

In this study, we inferred a set of LUCA, LACA, and LBCA tRNAs for all 20 standard protein amino acids. In addition, we have also shown that LUCA tRNA Ala and tRNA Ser , each bearing modified anticodons for ACN codons, can function within the E. coli translation system. Whether similar codon–anticodon pairing patterns are maintained in other codon boxes, and whether other LUCA tRNAs can operate effectively in the E. coli translational context, remains to be determined. A next goal of our future work is to reconstruct a translation system composed entirely of LUCA-derived tRNAs.

We believe that ASR based on molecular phylogenetics may enable the inference of pre-LUCA tRNA sequences, allowing deeper exploration of the evolutionary origins of the translational apparatus. The phylogenetic tree shown in Fig. 3 illustrates how such reconstructions may proceed. In parallel, progress has been made in reconstructing ARSs (Fournier et al. 2011 ; Fournier and Alm 2015 ), including our efforts to infer ARSs predating the LUCA (Furukawa et al. 2022 ; S. Yokobori, in preparation). Functional analysis of ancestral EF-Tu proteins has already been reported (Gaucher et al. 2008 ; Kacar et al. 2017a , b ), and further studies of additional translational factors are likely to follow.

Importantly, the feasibility of replacing E. coli rRNA with that from other organisms (Asai et al. 1999 ; Kitahara et al. 2012 ), together with the in vitro reconstitution of the E. coli small ribosomal subunit (Tamaru et al. 2018 ), suggests that partial ancestral reconstruction of the ribosome is now within reach.

By combining ancestral components—tRNAs, ARSs, elongation factors, and ribosomal RNAs—derived from LUCA or earlier lineages, it may be possible to experimentally reconstitute primordial translation systems. Such reconstructions will provide valuable insights into how modern translation machinery evolved and became highly optimized.

Supplementary Information

Below is the link to the electronic supplementary material.

Supplementary Material 1

Acknowledgements

This work was supported by the Astrobiology Center Program of the National Institutes of Natural Sciences (NINS) (Grant Number AB311010) to SY. We thank Dr Satoshi Akanuma for his kind gift of the expression vector of the Arc1ΔThr gene. We also thank Dr Daisuke Kiga for his valuable comments on this study. The authors used ChatGPT and Paperpal during the revision process for assistance with Japanese-to-English translation, and editing and proofreading of English text, and proofreading. All AI-assisted revisions were reviewed and verified by the authors, who take full responsibility for the content of the manuscript.

Author Contributions

All authors contributed to the conception and design of this study. Phylogenetic analyses and ancestral sequence reconstruction were performed by Shin-ichi Yokobori. Kei Yamanouchi and Hinata Moriya were responsible for material preparation and data collection, and both Kei Yamanouchi and Shin-ichi Yokobori carried out the data analysis. The first draft of the manuscript was written by Shin-ichi Yokobori, and all authors provided critical feedback and helped to revise earlier versions of the manuscript. All the authors have read and approved the final version of the manuscript.

Footnotes

Publisher’s Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

References

Akanuma S, Nakajima Y, Yokobori S, Kimura M, Nemoto N, Mase T, Miyazono K, Tanokura M, Yamagishi A (2013) Experimental evidence for the thermophilicity of ancestral life. Proc Natl Acad Sci USA 110:11067–11072. 10.1073/pnas.1308215110

Akanuma S, Yokobori S, Nakajima Y, Bessho M, Yamaghishi A (2015) Robustness of predictions of extremely thermally stable proteins in ancient organisms. Evolution 69:2954–2962. 10.1111/evo.12779

Asahara H, Himeno H, Tamura K, Hasegawa T, Watanabe K, Shimizu M (1993) Recognition nucleotides of Escherichia coli tRNA Leu and its elements facilitating discrimination from tRNA Ser and tRNA Tyr . J Mol Biol 231:219–229. 10.1006/jmbi.1993.1277

Asahara H, Himeno H, Tamura K, Nameki N, Hasegawa T, Shimizu M (1994) Escherichia coli seryl-tRNA synthetase recognizes tRNA Ser by its characteristics tertiary structure. J Mol Biol 236:738–748. 10.1006/jmbi.1994.1186

Asai T, Zaporojets D, Squires C, Squires CL (1999) An Escherichia coli strain with all chromosomal rRNA operons inactivated: Complete exchange of rRNA genes between bacteria. Proc Natl Acad Sci USA 96:1971–1976. 10.1073/pnas.96.5.1971

Bailly M, Giannouli S, Blaise M, Stathopoulos C, Kern D, Becker HD (2006) A single tRNA base pair mediates bacterial tRNA-dependent biosynthesis of asparagine. Nucleic Acids Res 34:6083–6094. 10.1093/nar/gkl622

Bailly M, Blaise M, Lorber B, Becker HD, Kern D (2007) The transamidosome: A dynamic ribonucleoprotein particle dedicated to prokaryotic tRNA-dependent asparagine biosynthesis. Mol Cell 28:228–239. 10.1016/j.molcel.2007.08.017

Bedouelle H (1990) Recognition of tRNA Tyr by tyrosyl-tRNA synthetase. Biochimie 72:589–598. 10.1016/0300-9084(90)90122-w

Bonnefond L, Giegé R, Rudinger-Thirion J (2005) Evolution of the tRNA Tyr /TyrRS aminoacylation systems. Biochimie 87:873–883. 10.1016/j.biochi.2005.03.008

Breitschopf K, Gross HJ (1994) The exchange of the discriminator base A73 for G is alone sufficient to convert human tRNA Leu into a serine-acceptor in vitro . EMBO J 13:3166–3169. 10.1002/j.1460-2075.1994.tb06615.x

Breitschopf K, Gross HJ (1996) The discriminator bases G73 in human tRNA Ser and A73 in tRNA Leu have significantly different roles in the recognition of aminoacyl-tRNA synthetases. Nucleic Acids Res 24:405–410. 10.1093/nar/24.3.405

Breitschopf K, Achsel T, Busch K, Gross HJ (1995) Identity elements of human tRNA Leu : structural requirements for converting human tRNA Ser into a leucine acceptor in vitro . Nucleic acids Res 23:3633–3637. 10.1093/nar/23.18.3633

Crick FHC (1966) Codon—anticodon pairing: The wobble hypothesis. J Mol Biol 19:548–555. 10.1016/s0022-2836(66)80022-0

Cronin JR, Pizzarello S (1983) Amino acids in meteorites. Adv Space Res 3:5–18. 10.1016/0273-1177(83)90036-4

Curnow AW, Ibba M, Söll D (1996) tRNA-dependent asparagine formation. Nature 382:589–590. 10.1038/382589b0

de Farias ST (2013) Suggested phylogeny of tRNAs based on the construction of ancestral sequences. J Theor Biol 335:245–248. 10.1016/j.jtbi.2013.06.033

Dixon AS, Schwinn MK, Hall MP, Zimmerman K, Otto P, Lubben TH, Butler BL, Binkowski BF, Machleidt T, Kirkland TA, Wood MG, Eggers CT, Encell LP, Wood KV (2016) NanoLuc complementation reporter optimized for accurate measurement of protein interactions in cells. ACS Chem Biol 11:400–408. 10.1021/acschembio.5b00753

Dong H, Nilsson L, Kurland CG (1996) Co-variation of tRNA abundance and codon usage in Escherichia coli at different growth rates. J Mol Biol 260:649–663. 10.1006/jmbi.1996.0428

Du X, Wang E (2003) Tertiary structure base pairs between D- and TψC‐loops of Escherichia coli tRNA Leu play important roles in both aminoacylation and editing. Nucleic Acids Res 31:2865–2872. 10.1093/nar/gkg382

Farrow MA, Nordin BE, Schimmel P (1999) Nucleotide determinants for tRNA-dependent amino acid discrimination by a Class I tRNA synthetase. Biochemistry 38:16898–16903. 10.1021/bi9920782

Fechter P, Rudinger-Thirion J, Théobald-Dietrich A, Giegé R (2000) Identity of tRNA for yeast tyrosyl-tRNA synthetase: Tyrosylation is more sensitive to identity nucleotides than to structural features. Biochemistry 39:1725–1733. 10.1021/bi992276t

Finnigan GC, Hanson-Smith V, Stevens TH, Thornton JW (2012) Evolution of increased complexity in a molecular machine. Nature 481:360–364. 10.1038/nature10724

Fournier GP, Alm EJ (2015) Ancestral reconstruction of a pre-LUCA aminoacyl-tRNA synthetase ancestor supports the late addition of Trp to the genetic code. J Mol Evol 80:171–185. 10.1007/s00239-015-9672-1

Fournier GP, Andam CP, Alm EJ, Gogarten JP (2011) Molecular evolution of aminoacyl tRNA synthetase proteins in the early history of life. Origins Life Evol Biosph 41:621–632. 10.1007/s11084-011-9261-2

Francklyn C, Schimmel P (1989) Aminoacylation of RNA minihelices with alanine. Nature 337:478–481. 10.1038/337478a0

Furukawa R, Nakagawa M, Kuroyanagi T, Yokobori S, Yamagishi A (2017) Quest for ancestors of eukaryal cells based on phylogenetic analyses of aminoacyl-tRNA synthetases. J Mol Evol 84:51–66. 10.1007/s00239-016-9768-2

Furukawa R, Yokobori S, Sato R, Kumagawa T, Nakagawa M, Katoh K, Yamagishi A (2022) Amino acid specificity of ancestral aminoacyl-tRNA synthetase prior to the Last Universal Common Ancestor Commonote commonote . J Mol Evol 90:73–94. 10.1007/s00239-021-10043-z

Garcia AK, Schopf JW, Yokobori S, Akanuma S, Yamagishi A (2017) Reconstructed ancestral enzymes suggest long-term cooling of Earth’s photic zone since the Archean. Proc Natl Acad Sci USA 114:4619–4624. 10.1073/pnas.1702729114

Gaucher EA, Thomson JM, Burgan MF, Benner SA (2003) Inferring the palaeoenvironment of ancient bacteria on the basis of resurrected proteins. Nature 425:285–288. 10.1038/nature01977

Gaucher EA, Govindarajan S, Ganesh OK (2008) Palaeotemperature trend for Precambrian life inferred from resurrected proteins. Nature 451:704–707. 10.1038/nature06510

Giegé R, Eriani G (2023) The tRNA identity landscape for aminoacylation and beyond. Nucleic Acids Res 51:1528–1570. 10.1093/nar/gkad007

Giegé R, Lapointe J (2009) Transfer RNA aminoacylation and modified nucleosides. In: Grosjean H (ed) DNA and RNA modification enzymes: Structure, mechanism, function, and evolution. Landes Bioscience, Austin, Texas, pp 475–492

Giegé R, Florentz C, Kern D, Gangloff J, Eriani G, Moras D (1996) Aspartate identity of transfer RNAs. Biochimie 78:605–623. 10.1016/s0300-9084(96)80007-1

Giegé R, Sissler M, Florentz C (1998) Universal rules and idiosyncratic features in tRNA identity. Nucleic Acids Res 26:5017–5035. 10.1093/nar/26.22.5017

Guenneugues M, Caserta E, Brandi L et al (2000) Mapping the fMet-tRNA fMet binding site of initiation factor IF2. EMBO J 19:5233–5240. 10.1093/emboj/19.19.5233

Guillon JM, Mechulam Y, Schmitter JM, Blanquet S, Fayat G (1992a) Disruption of the gene for Met-tRNA fMet formyltransferase severely impairs growth of Escherichia coli . J Bacteriol 174:4294–4301. 10.1128/jb.174.13.4294-4301.1992

Guillon J-M, Meinnel T, Mechulam Y, Lazennec C, Blanquet S, Fayat G (1992b) Nucleotides of tRNA governing the specificity of Escherichia coli methionyl-tRNA fMet formyltransferase. J Mol Biol 224:359–367. 10.1016/0022-2836(92)91000-f

Guillon JM, Mechulam Y, Blanquet S, Fayat G (1993) Importance of formylability and anticodon stem sequence to give a tRNA Met an initiator identity in Escherichia coli . J bacteriol 175:4507–4514. 10.1128/jb.175.14.4507-4514.1993

Harada M, Nagano A, Yagi S, Furukawa R, Yokobori S, Yamagishi A (2019) Planktonic adaptive evolution to the sea surface temperature in the Neoproterozoic inferred from ancestral NDK of marine cyanobacteria. Earth Planet Sc Lett 522:98–106. 10.1016/j.epsl.2019.06.014

Hasegawa T, Yokogawa T (2000) Escherichia coli proline tRNA: structure and recognition sites for prolyl-tRNA synthetase. Nucleic acids Symp Ser 44:7–8. 10.1093/nass/44.1.7

Hasegawa T, Miyano M, Himeno H, Sano Y, Kimura K, Shimizu M (1992) Identity determinants of E. coli threonine tRNA. Biochem Biophys Res Commun 184:478–484. 10.1016/0006-291x(92)91219-g

Hauenstein S, Zhang C-M, Hou Y-M, Perona JJ (2004) Shape-selective RNA recognition by cysteinyl-tRNA synthetase. Nat Struct Mol Biol 11:1134–1141. 10.1038/nsmb849

Hendrickson TL (2001) Recognizing the D-loop of transfer RNAs. Proc Natl Acad Sci USA 98:13473–13475. 10.1073/pnas.251549298

Himeno H, Hasegawa T, Ueda T, Watanabe K, Miura K, Shimizu M (1989) Role of the extra G-C pair at the end of the acceptor stem of tRNA His in aminoacylation. Nucleic Acids Res 17:7855–7863. 10.1093/nar/17.19.7855

Himeno H, Hasegawa T, Asahara H, Tamura K, Shimizu M (1991) Identity determinants of E. coli tryptophan tRNA. Nucleic acids Res 19:6379–6382. 10.1093/nar/19.23.6379

Horowitz J, Chu WC, Derrick WB, Liu JC-H, Liu M, Yue D (1999) Synthetase recognition determinants of E. coli valine transfer RNA. Biochemistry 38:7737-7746. 10.1021/bi990490b

Hou YM, Schimmel P (1988) A simple structural feature is a major determinant of the identity of a transfer RNA. Nature 333:140–145. 10.1038/333140a0

Hou YM, Westhof E, Giegé R (1993) An unusual RNA tertiary interaction has a role for the specific aminoacylation of a transfer RNA. Proc Natl Acad Sci USA 90:6776–6780. 10.1073/pnas.90.14.6776

Ibba M, Hong KW, Sherman JM, Sever S, Söll D (1996) Interactions between tRNA identity nucleotides and their recognition sites in glutaminyl-tRNA synthetase determine the cognate amino acid affinity of the enzyme. Proc Natl Acad Sci USA 93:6953–6958. 10.1073/pnas.93.14.6953

Ikeuchi Y, Kimura S, Numata T, Nakamura D, Yokogawa T, Ogata T, Wada T, Suzuki T, Suzuki T (2010) Agmatine-conjugated cytidine in a tRNA anticodon is essential for AUA decoding in archaea. Nat Chem Biol 6:277–282. 10.1038/nchembio.323

Inagaki Y, Kojima A, Bessho Y, Hori H, Ohama T, Osawa S (1995) Translation of synonymous codons in family boxes by Mycoplasma capricolum tRNAs with unmodified uridine or adenosine at the first anticodon position. J Mol Biol 251:486–492. 10.1006/jmbi.1995.0450

Ishikura H, Nagaoka Y, Yokozawa J, Umehara T, Kuno A, Hasegawa T (2000) Threonyl-tRNA synthetase of archaea: Importance of the discriminator base in the aminoacylation of threonine tRNA. Nucleic Acids Symp Ser 44:83–84. 10.1093/nass/44.1.83

Iwaki J, Asahara H, Nagaoka Y, Yokozawa J, Umehara T, Kawarabayasi Y, Koyama Y, Sako Y, Kuno A, Hasegawa T (2002) Differences in tyrosine tRNA identity between Escherichia coli and archaeon, Aeropyrum pernix K1. Nucleic Acids Symp Ser 2:225–226. 10.1093/nass/2.1.225

Jahn M, Rogers MJ, Söll D (1991) Anticodon and acceptor stem nucleotides in tRNA Gln are major recognition elements for E. coli glutaminyl-tRNA synthetase. Nature 352:258–260. 10.1038/352258a0

Jermann TM, Opitz JG, Stackhouse J, Benner SA (1995) Reconstructing the evolutionary history of the artiodactyl ribonuclease superfamily. Nature 374:57–59. 10.1038/374057a0

Jühling F, Mörl M, Hartmann RK, Sprinzl M, Stadler PF, Pütz J (2009) tRNAdb 2009: compilation of tRNA sequences and tRNA genes. Nucleic Acids Res 37:D159–D162. 10.1093/nar/gkn772

Kacar B, Garmendia E, Tuncbag N, Andersson DI, Hughes D (2017a) Functional constraints on replacing an essential gene with Its ancient and modern homologs. mBio 8:e01276–e01217. 10.1128/mbio.01276-17

Kacar B, Ge X, Sanyal S, Gaucher EA (2017b) Experimental evolution of Escherichia coli harboring an ancient translation protein. J Mol Evol 84:69–84. 10.1007/s00239-017-9781-0

Katoh K, Standley DM (2013) MAFFT Multiple Sequence Alignment Software Version 7: Improvements in performance and usability. Mol Biol Evol 30:772–780. 10.1093/molbev/mst010

Kishino H, Hasegawa M (1989) Evaluation of the maximum likelihood estimate of the evolutionary tree topologies from DNA sequence data, and the branching order in hominoidea. J Mol Evol 29:170–179. 10.1007/bf02100115

Kitahara K, Yasutake Y, Miyazaki K (2012) Mutational robustness of 16S ribosomal RNA, shown by experimental horizontal gene transfer in Escherichia coli . Proc Natl Acad Sci USA 109:19220–19225. 10.1073/pnas.1213609109

Komatsoulis GA, Abelson J (1993) Recognition of tRNA Cys by Escherichia coli cysteinyl-tRNA synthetase. Biochemistry 32:7435–7444. 10.1021/bi00080a014

Kompatscher M, Bartosik K, Erharter K, Plangger R, Juen FS, Kreutz C, Micura R, Westhof E, Erlacher MD (2023) Contribution of tRNA sequence and modifications to the decoding preferences of E. coli and M. mycoides tRNA Gly UCC for synonymous glycine codons. Nucleic Acids Res 52:1374–1386. 10.1093/nar/gkad1136

Koonin EV (2017) Frozen Accident pushing 50: Stereochemistry, expansion, and chance in the evolution of the genetic code. Life 7:22–13. 10.3390/life7020022

Kumar S, Stecher G, Li M, Knyaz C, Tamura K (2018) MEGA X: Molecular Evolutionary Genetics Analysis across computing platforms. Mol Biol Evol 35:1547–1549. 10.1093/molbev/msy096

Lawless JG (1973) Amino acids in the Murchison meteorite. Geochim Cosmochim Acta 37:2207–2212. 10.1016/0016-7037(73)90017-3

Lee CP, Seong BL, RajBhandary UL (1991) Structural and sequence elements important for recognition of formylmethionine tRNA by methionyl-tRNA transformylase are clustered in the acceptor stem. J Biol Chem 266:18012–18017. 10.1016/s0021-9258(18)55230-3

Li S, Pelka H, Schulman LH (1993) The anticodon and discriminator base are important for aminoacylation of Escherichia coli tRNA Asn . J Biol Chem 268:18335–18339

Liu H, Peterson R, Kessler J, Musier-Forsyth K (1995) Molecular recognition of tRNA Pro by Escherichia coli proline tRNA synthetase in vitro . Nucleic Acids Res 23:165–169. 10.1093/nar/23.1.165

Liu C, Gamper H, Shtivelband S, Hauenstein S, Perona JJ, Hou Y-M (2007) Kinetic quality control of anticodon recognition by a eukaryotic aminoacyl-tRNA synthetase. J Mol Biol 367:1063–1078. 10.1016/j.jmb.2007.01.050

Madore E, Florentz C, Giegé R, Sekine S, Yokoyama S, Laoiubte J (1999) Effect of modified nucleotides on Escherichia coli tRNA Glu structure and on its aminoacylation by glutamyl-tRNA synthetase. Predominant and distinct roles of the mnm 5 and s 2 modifications of U34. Eur J Biochem 266:1128–1135. 10.1046/j.1432-1327.1999.00965.x

Mandal D, Köhrer C, Su D, Russell SP, Krivos K, Castleberry CM, Blum P, Limbach PA, Söll D, RajBhandary UL (2010) Agmatidine, a modified cytidine in the anticodon of archaeal tRNA Ile , base pairs with adenosine but not with guanosine. Proc Natl Acad Sci USA 107:2872–2877. 10.1073/pnas.0914869107

Manickam N, Nag N, Abbasi A, Patel K, Farabaugh PJ (2014) Studies of translational misreading in vivo show that the ribosome very efficiently discriminates against most potential errors. RNA 20:9–15. 10.1261/rna.039792.113

Masuda I, Hou Y-M (2024) A tRNA modification pattern that facilitates interpretation of the genetic code. Front Microbiol 15:1415100. 10.3389/fmicb.2024.1415100

Mazauric M, Keith G, Logan D, Kreutzer R, Giegé, Kern D (1998) Glycyl-tRNA synthetase from Thermus thermophilus . Eur J Biochem 251:744–757. 10.1046/j.1432-1327.1998.2510744.x

Mazauric M-H, Roy H, Kern D (1999) tRNA glycylation system from Thermus thermophilus . tRNA Gly identity and functional interrelation with the glycylation systems from other Phylae. Biochemistry 38:13094–13105. 10.1021/bi991392t

McClain WH, Foss K (1988) Changing the acceptor identity of a transfer RNA by altering nucleotides in a variable pocket. Science 241:1804–1807. 10.1126/science.2459773

McClain WH, Chen YM, Foss K, Schneider J (1988) Association of transfer RNA acceptor identity with a helical irregularity. Science 242:1681–1684. 10.1126/science.2462282

McClain WH, Foss K, Jenkins RA, Schneider J (1990) Nucleotides that determine Escherichia coli tRNA Arg and tRNA Lys acceptor identities revealed by analyses of mutant opal and amber suppressor tRNAs. Proc Natl Acad Sci USA 87:9260–9264. 10.1073/pnas.87.23.926

Miller SL (1953) A production of amino acids under possible primitive Earth conditions. Science 117:528–529. 10.1126/science.117.3046.528

Muramatsu T, Nishikawa K, Nemoto F, Kuchino Y, Nishimura S, Miyazono T, Yokoyama S (1988) Codon and amino-acid specificities of a transfer RNA are both converted by a single post-transcriptional modification. Nature 336:179181. 10.1038/336179a0

Nagel GM, Doolittle RF (1995) Phylogenetic analysis of the aminoacyl-tRNA synthetases. J Mol Evol 40:487–498. 10.1007/bf00166617

Nair N, Raff H, Islam MT, Feen M, Garofalo DM, Sheppard K (2016) The Bacillus subtilis and Bacillus halodurans aspartyl-tRNA synthetases retain recognition of tRNA Asn . J Mol Biol 428:618–630. 10.1016/j.jmb.2016.01.014

Nakamura A, Sheppard K, Yamane J, Yao M, Söll D, Tanaka I (2010) Two distinct regions in Staphylococcus aureus GatCAB guarantee accurate tRNA recognition. Nucleic Acids Res 38:672–682. 10.1093/nar/gkp955

Nameki N (1995) Identity elements of tRNA Thr towards saccharomyces cerevisiae threonyl-tRNA synthetase. Nucleic Acids Res 23:2831–2836. 10.1093/nar/23.15.2831

Nameki N, Tamura K, Himeno H, Asahara H, Hasegawa T, Shimizu M (1992) Escherichia coli tRNA Asp recognition mechanism differing from that of the yeast system. Biochem Biophys Res Comm 189:856–862. 10.1016/0006-291X(92)92282-3

Nameki N, Asahara H, Shimizu M, Okada N, Himeno H (1995) Identity elements of Saccharomyces cerevisiae tRNA His . Nucleic Acids Res 23:389–394. 10.1093/nar/23.3.389

Nameki N, Asahara H, Hasegawa T (1996) Identity elements of Thermus thermophilus tRNA Thr . FEBS Lett 396:201–207. 10.1016/0014-5793(96)01094-0

Nameki N, Tamura K, Asahara H, Hasegawa T (1997) Recognition of tRNA Gly by three widely diverged glycyl-tRNA synthetases. J Mol Biol 268:640–647. 10.1006/jmbi.1997.0993

Naraoka H, Takano Y, Dworkin JP et al (2023) Soluble organic molecules in samples of the carbonaceous asteroid (162173) Ryugu. Science 379:eabn9033. 10.1126/science.abn9033

Nei M, Kumar S (2000) Molecular Evolution and Phylogenetics. Oxford University Press, New York

Nguyen L-T, Schmidt HA, von Haeseler A, Minh BQ (2014) IQ-TREE: A fast and effective stochastic algorithm for estimating maximum-likelihood phylogenies. Mol Biol Evol 32:268–274. 10.1093/molbev/msu300

Normanly J, Kleina LG, Masson JM, Abelson J, Miller JH (1990) Construction of Escherichia coli amber suppressor tRNA genes. III. Determination of tRNA specificity. J Mol Biol 213:719–726. 10.1016/s0022-2836(05)80258-x

Normanly J, Ollick T, Abelson J (1992) Eight base changes are sufficient to convert a leucine-inserting tRNA into a serine-inserting tRNA. Proc Natl Acad Sci USA 89:5680–5684. 10.1073/pnas.89.12.5680

Nureki O, Niimi T, Muramatsu T, Kanno H, Kohno T, Florentz C, Giegé R, Yokoyama S (1994) Molecular recognition of the identity-determinant set of isoleucine transfer RNA from Escherichia coli . J Mol Biol 236:710–724. 10.1006/jmbi.1994.1184

Ohira T, Suzuki T, Miyauchi K, Suzuki T, Yokobori S, Yamagishi A, Watanabe K (2013) Decoding mechanism of non-universal genetic codes in Loligo bleekeri mitochondria. J Biol Chem 288:7645–7652. 10.1074/jbc.m112.439554

Okamoto K, Kuno A, Hasegawa T (2005) Recognition sites of glycine tRNA for glycyl-tRNA synthetase from hyperthermophilic archaeon, Aeropyrum pernix K1. Nucleic Acids Symp Ser 49:299–300. 10.1093/nass/49.1.299

Osawa S, Jukes TH, Watanabe K, Muto A (1992) Recent evidence for evolution of the genetic code. Microbiol Rev 56:229–264. 10.1128/mr.56.1.229-264.1992

Pallanck L, Schulman LH (1991) Anticodon-dependent aminoacylation of a noncognate tRNA with isoleucine, valine, and phenylalanine in vivo . Proc Natl Acad Sci USA 88:3872–3876. 10.1073/pnas.88.9.3872

Pallanck L, Li S, Schulman LH (1992) The anticodon and discriminator base are major determinants of cysteine tRNA identity in vivo . J Biol Chem 267:7221–7223. 10.1016/S0021-9258(18)42508-2

Park SJ, Hou YM, Schimmel P (1989) A single base pair affects binding and catalytic parameters in the molecular recognition of a transfer RNA. Biochemistry 28:2740–2746. 10.1021/bi00432a056

Parker ET, McLain HL, Glavin DP et al (2023) Extraterrestrial amino acids and amines identified in asteroid Ryugu samples returned by the Hayabusa2 mission. Geochim Cosmochim Acta 347:42–57. 10.1016/j.gca.2023.02.017

Patel BH, Percivalle C, Ritson DJ, Duffy CD, Sutherland JD (2015) Common origins of RNA, protein and lipid precursors in a cyanosulfidic protometabolism. Nat Chem 7:301–307. 10.1038/nchem.2202

Peterson ET, Uhlenbeck OC (1992) Determination of recognition nucleotides for Escherichia coli phenylalanyl-tRNA synthetase. Biochemistry 31:10380–10389. 10.1021/bi00157a028

Peterson ET, Blank J, Sprinzl M, Uhlenbeck OC (1993) Selection for active E. coli tRNA Phe variants from a randomized library using two proteins. EMBO J 12:2959–2967. 10.1002/j.1460-2075.1993.tb05958.x

Ramesh V, RajBhandary UL (2001) Importance of the anticodon sequence in the aminoacylation of tRNAs by methionyl-tRNA synthetase and by valyl-tRNA synthetase in an Archaebacterium. J Biol Chem 276:3660–3665. 10.1074/jbc.m008206200

Rathnayake UM, Hendrickson TL (2019) Bacterial aspartyl-tRNA synthetase has glutamyl-tRNA synthetase Activity. Genes 10:262. 10.3390/genes10040262

Root-Bernstein R, Kim Y, Sanjay A, Burton ZF (2016) tRNA evolution from the proto-tRNA minihelix world. Transcription 7:153–163. 10.1080/21541264.2016.1235527

Rosen AE, Brooks BS, Guth E, Francklyn CS, Musier-Forsyth K (2006) Evolutionary conservation of a functionally important backbone phosphate group critical for aminoacylation of histidine tRNAs. RNA 12:1315–1322. 10.1261/rna.78606

Schmitt E, Coureux P-D, Kazan R, Bourgeois G, Lazennec-Schurdevin C, Mechulam Y (2020) Recent advances in archaeal translation initiation. Front Microbiol 11:584152. 10.3389/fmicb.2020.584152

Schulman LH, Pelka H (1983) Anticodon loop size and sequence requirements for recognition of formylmethionine tRNA by methionyl-tRNA synthetase. Proc Natl Acad Sci USA 80:6755–6759. 10.1073/pnas.80.22.6755

Schulman LH, Pelka H (1988) Anticodon switching changes the identity of methionine and valine transfer RNAs. Science 242:765–768. 10.1126/science.3055296

Schulman LH, Pelka H (1989) The anticodon contains a major element of the identity of arginine transfer RNAs. Science 246:1595–1597. 10.1126/science.2688091

Sekine S, Nureki O, Sakamoto K, Niimi T, Tateno M, Gö M, Kohno T, Brisson A, Lapointe J, Yokoyama S (1996) Major identity determinants in the augmented D helix of tRNA Glu from Escherichia coli . J Mol Biol 256:685–700. 10.1006/jmbi.1996.0118

Senger B, Auxilien S, Englisch U, Cramer F, Fasiolo F (1997) The modified wobble base inosine in yeast tRNA Ile is a positive determinant for aminoacylation by isoleucyl-tRNA synthetase. Biochemistry 36:8269–8275. 10.1021/bi970206l

Shibue R, Sasamoto T, Shimada M, Zhang B, Yamagishi A, Akanuma S (2018) Comprehensive reduction of amino acid set in a protein suggests the importance of prebiotic amino acids for stable proteins. Sci Rep 8:1227. 10.1038/s41598-018-19561-1

Shimizu M, Asahara H, Tamura K, Hasegawa T, Himeno H (1992) The Role of anticodon bases and the discriminator nucleotide in the recognition of some E. coli tRNAs by their aminoacyl-tRNA synthetases. J Mol Evol 35:436–443. 10.1007/bf00171822

Shimizu Y, Inoue A, Tomari Y, Suzuki T, Yokogawa T, Nishikawa K, Ueda T (2001) Cell-free translation reconstituted with purified components. Nat Biotechnol 19:751–755. 10.1038/90802

Shimodaira H, Goldman N (2002) An approximately unbiased test of phylogenetic tree selection. Syst Biol 51:492–508. 10.1080/10635150290069913

Shimodaira H, Hasegawa M (1999) Multiple comparisons of log-likelihoods with applications to phylogenetic inference. Mol Biol Evol 16:1114–1114. 10.1093/oxfordjournals.molbev.a026201

Siatecka M, Rozek M, Barciszewski J, Mirande M (1998) Modular evolution of the Glx-tRNA synthetase family–rooting of the evolutionary tree between the bacteria and archaea/eukarya branches. Eur J Biochem 256:80–87. 10.1046/j.1432-1327.1998.2560080.x

Stecher G, Tamura K, Kumar S (2020) Molecular Evolutionary Genetics Analysis (MEGA) for macOS. Mol Biol Evol 37:1237–1239. 10.1093/molbev/msz312

Suzuki T, Ueda T, Watanabe K (1997) The polysemous codon–a codon with multiple amino acid assignment caused by dual specificity of tRNA identity. EMBO J 16:1122–1134. 10.1093/emboj/16.5.1122

Suzuki T, Nagao A, Suzuki T (2011) Human mitochondrial tRNAs: Biogenesis, function, structural aspects, and diseases. Annu Rev Genet 45:299–329. 10.1146/annurev-genet-110410-132531

Tamaru D, Amikura K, Shimizu Y, Nierhaus KH, Ueda T (2018) Reconstitution of 30S ribosomal subunits in vitro using ribosome biogenesis factors. RNA 24:1512–1519. 10.1261/rna.065615.118

Tamura K (2015) Origins and early evolution of the tRNA molecule. Life 5:1687–1699. 10.3390/life5041687

Tamura K, Himeno H, Asahara H, Hasegawa T, Shimizu M (1991) Identity determinants of E. coli tRNA Val . Biochem Biophys Res Comm 177:619–623. 10.1016/0006-291X(91)91833-X

Tamura K, Himeno H, Asahara H, Hasegawa T, Shimizu M (1992) In vitro study of E. coli tRNA Arg and tRNA Lys identity elements. Nucleic Acids Res 20:2335–2339. 10.1093/nar/20.9.2335

Thornton JW (2004) Resurrecting ancient genes: experimental analysis of extinct molecules. Nat Rev Genet 5:366–375. 10.1038/nrg1324

Tomita K, Ueda T, Ishiwa S, Crain PF, McCloskey JA, Watanabe K (1999) Codon reading patterns in Drosophila melanogaster mitochondria based on their tRNA sequences: a unique wobble rule in animal mitochondria. Nucleic Acids Res 27:4291–4297. 10.1093/nar/27.21.4291

Tsuchiya W, Hasegawa T (2009) Molecular recognition of tryptophan tRNA by tryptophanyl-tRNA synthetase from Aeropyrum pernix K1. J Biochem (Tokyo) 145:635–641. 10.1093/jb/mvp020

Wang J-T, Zhou J-B, Mao X-L, Zhou L, Chen M, Zhang W, Wang E-D, Zou X-L (2022) Commonality and diversity in tRNA substrate recognition in t 6 A biogenesis by eukaryotic KEOPSs. Nucleic Acids Res 50:2223–2239. 10.1093/nar/gkac056

Watanabe K, Yokobori S (2011) tRNA modification and genetic code variations in animal mitochondria. J Nucleic Acids 2011:Article ID 623095. 10.4061/2011/623095

Watanabe K, Yokobori S (2014) How the early genetic code was established? Inference from the analysis of extant animal mitochondrial decoding systems. In: Erdmann V, Markiewicz W, Barciszewski J (eds) Chemical Biology of Nucleic Acids. RNA Technologies. Springer, Berlin Heidelberg, pp 25–40

Watanabe Y, Tsurui H, Ueda T, Furusihima-Shimogawara R, Takoamiya S, Kita K, Nishikawa K, Watanabe K (1997) Primary sequence of mitochondrial tRNA Arg of a nematode Ascaris suum : occurrence of unmodified adenosine at the first position of the anticodon. Biochim Biophys Acta Gene Struct Expr 1350:119–122. 10.1016/s0167-4781(96)00211-4

Woese CR, Olsen GJ, Ibba M, Söll D (2000) Aminoacyl-tRNA synthetases, the genetic Code, and the evolutionary process. Microbiol Mol Biol Rev 64:202–236. 10.1128/mmbr.64.1.202-236.2000

Wong JT-F (1975) A co-evolution theory of the genetic code. Proc Natl Acad Sci USA 72:1909–1912. 10.1073/pnas.72.5.1909

Xu F, Jiang G, Li W, He X, Jin Y, Wang D (2002) Three G·C base pairs required for the efficient aminoacylation of tRNA Trp by tryptophanyl-tRNA synthetase from Bacillus subtilis . Biochemistry 41:8087–8092. 10.1021/bi015881g

Xue H, Shen W, Giegé R, Wong JT (1993) Identity elements of tRNA Trp . Identification and evolutionary conservation. J Biol Chem 268:9316–9322. 10.1016/s0021-9258(18)98352-3

Yokobori S, Kitamura A, Grosjean H, Bessho Y (2013) Life without tRNA Arg –adenosine deaminase TadA: evolutionary consequences of decoding the four CGN codons as arginine in Mycoplasmas and other Mollicutes. Nucleic Acids Res 41:6531–6543. 10.1093/nar/gkt356

Yokozawa J, Okamoto K, Kawarabayasi Y, Kuno A, Hasegawa T (2003) Molecular recognition of proline tRNA by prolyl-tRNA synthetase from hyperthermophilic archaeon, Aeropyrum pernix K1. Nucleic acids Symp Ser 3:247–248. 10.1093/nass/3.1.247

Zhou X-L, Fang Z-P, Ruan Z-R, Wang M, Liu R-J, Tan M, Anella FM, Wang ED (2013) Aminoacylation and translational quality control strategy employed by leucyl-tRNA synthetase from a human pathogen with genetic code ambiguity. Nucleic Acids Res 41:9825–9838. 10.1093/nar/gkt741

Associated Data

Supplementary Materials

Supplementary Material 1