Diversity of the biosynthesis pathway for threonylcarbamoyladenosine (t⁶A), a universal modification of tRNA

Abstract

The tRNA modification field has a rich literature covering biochemical analysis going back more than 40 years, but many of the corresponding genes were only identified in the last decade. In recent years, comparative genomic-driven analysis has allowed for the identification of the genes and subsequent characterization of the enzymes responsible for N6-threonylcarbamoyladenosine (t⁶A). This universal modification, located in the anticodon stem-loop at position 37 adjacent to the anticodon of tRNAs, is found in nearly all tRNAs that decode ANN codons. The t⁶A biosynthesis enzymes and synthesis pathways have now been identified, revealing both a core set of enzymes and kingdom-specific variations. This review focuses on the elucidation of the pathway, diversity of the synthesis genes, and proposes a new nomenclature for t⁶A synthesis enzymes.

Introduction

tRNAs (tRNAs) are the central adaptors in the translation process responsible for decoding mRNAs. tRNAs harbor numerous post-transcriptional modifications that fine-tune their function. To date, more than 90 modifications¹ have been identified in tRNA, and most organisms devote more genetic information to modifying tRNAs than to the tRNAs themselves.² While modifications have been shown to affect different aspects of tRNA metabolism and shape interactions of tRNA molecules with the rest of the translation apparatus,^3-5 most modifications to the anticodon-stem-loop (ASL) are required for accurate decoding.^6,7 The diversity of tRNA modifications and how modifications affect function has been the topic of recent reviews,^8-10 and for a summary of modifications to the ASL in Escherichia coli see the recent review by Helm and Alfonzo¹¹ and Table 4 in de Crécy-Lagard, et al.¹⁰ In this review, we address the enzymes responsible for the formation of N6-threonyl-carbamoyl-adenosine (t⁶A) and its derivatives. This complex modification of adenosine is located at position 37, next to the anticodon (t⁶A₃₇), and is one of the few universal modifications of the ASL.¹²

The hypermodified base t⁶A is present in nearly all ANN decoding tRNAs and has been studied in vitro and in vivo for more than 40 y.^13-19 Since the first discovery of the modification by Schweizer, et al. in 1969,¹⁸ sporadic studies established the basic requirements for the synthesis of this universal modification, identifying the requirement for ATP, threonine and carbonate,^15,17,20-22 but fell short of elucidating the multi-step path to its formation. Subsequent studies in which native E. coli tRNA^fMet (harboring an unmodified A₃₇) and yeast tRNA^iMet transcripts were converted to t⁶A₃₇ after microinjection into Xenopus laevis oocytes demonstrated that the formation of t⁶A occurred in the oocyte cytoplasm and used a conserved machinery.^13,14 These studies also demonstrated that A₃₇ and U₃₆ were strict determinants for t⁶A formation, and that A₃₈ enhances the efficiency of modification of A₃₇ to t⁶A₃₇.¹³ Finally, structural studies showed that t⁶A enhances anticodon-codon base-pairing by cross-strand base-stacking of the t⁶A base with the first position of the codon,²³ and influences the structure of the ASL by preventing across the loop base-pairing between U33-A37, as well as stacking of bases A₃₇ and A₃₈.^23-27

Only in the last 5 y have the t⁶A biosynthesis enzymes and the pathways been elucidated, revealing both a core set of enzymes and kingdom-specific variations (Fig. 1).^28-33 Elucidation of this multi-step pathway, which requires the formation of an activated carbon dioxide intermediate, and the diversity of the enzymes required for synthesis are the focus of this review.

Figure 1.

Diversity in the synthesis of the universal tRNA modification t⁶A. Two types of enzymes families, TsaC (YrdC) and TsaC2 (Sua5) in Bacteria and Tcs1 (YrdC) and Tcs2 (Sua5) in Eukarya and Archaea, catalyze the formation of TC-AMP. TsaC2 and Tcs2 contain a TsaC-domain plus an additional C-terminal Sua5-domain. To transfer threonylcarbamoyl (TC) to tRNA, Bacteria require TsaBDE, while Archaea and Eukarya use the KEOPS complex composed of Tcs3 (Kae1), Tcs5 (Bud32), Tcs6 (Pcc1) and Tcs7 (Cgi121) proteins. Tcs8 (Gon7) is found exclusively in Fungi. Mitochondria use the nuclear encoded Tcs4 (Qri7) for transfer of the TC to tRNA. Colors represent homology and correspond with Figure 2.

Discovery of the First t⁶A Synthesis Genes

The first enzyme of the t⁶A pathway was discovered in 2009 when it was found that a universal protein family, YrdC/Sua5 (COG0009), was involved in t⁶A modification.³¹ Based on the assumption that because t⁶A was universally conserved the t⁶A biosynthetic enzymes would also be universally conserved, this work used comparative genomic analysis to focus on universally conserved protein families of unknown function. At the time of this study, 9 universally conserved protein families were of unknown function. Of these, the YrdC/Sua5 family was judged the most likely candidate for involvement in t⁶A biosynthesis due to: 1) its similarity to HypF, which catalyzes a carbamoylation reaction³⁴ similar to a putative step in t⁶A biosynthesis proposed in 1974 by both Elkins and Keller¹⁵ and Körner and Söll;¹⁷ 2) mutations in the yeast yrdC ortholog SUA5 led to translation defects (initiation at non-AUG codons);³⁵ and 3) E. coli YrdC was found to bind RNA and tRNA.³⁶ The involvement of the YrdC/Sua5 family in t⁶A synthesis was experimentally validated using E. coli and Saccharomyces cerevisiae.³¹

In E. coli, yrdC is essential, but SUA5 can be deleted from S. cerevisiae, although the growth of the mutant is severely compromised. tRNAs analyzed from this mutant were devoid of t⁶A. The levels of t⁶A could be restored through complementation with SUA5_Sc, yrdC_Ec, ywlC_Bs (Bacillus subtilis SUA5 homolog), and yrdC_Mm (Methanococcus maripaludis, an archaeal yrdC homolog). The essentiality phenotype of a E. coli yrdC deletion could be complemented by expressing orthologs from yeast, B. subtilis, and M. maripaludis in trans.³¹ Analysis of tRNAs in the complemented strains confirmed the presence of t⁶A.³¹ This work identified the first gene family involved in t⁶A synthesis, established that its function is universally conserved, and that members of the family could bind ATP, but because purified YrdC alone was not sufficient to produce t⁶A in tRNA transcripts in vitro, it also suggested that additional enzymes were needed for t⁶A synthesis, or that the role of YrdC/Sua5 family was indirect.

A second protein family involved with t⁶A synthesis, YgjD/Kae1/Qri7 (COG0533), was discovered in 2011.³⁰ Like YrdC/Sua5, the YgjD/Kae1/Qri7 family of proteins is universally conserved, and also exhibited similarity to HypF, which harbors a fusion of YrdC-like and YgjD-like domains. Kae1 had first been described as a member of the KEOPS complex (Kinase, putative Endopeptidase and Other Proteins of Small size)³⁷ also known as EKC (Endopeptidase-like Kinase Chromatin-associated complex),³⁸ and had been proposed to be involved in a variety of phenomena unrelated to RNA modification.^37,38 A phylogeny of this family revealed that yeast harbored 2 members of the family.³⁹ The first, Kae1 had homologs in other eukaryotes and archaea, and the second, Qri7 was targeted to the mitochondria and was part of the bacterial YgjD clade.³⁹

The hypothesis that the YgjD/Kae1/Qri7 family was involved in t⁶A synthesis was confirmed by extracting tRNAs from S. cerevisiae kae1Δ and showing they were devoid of t⁶A and that t⁶A levels could be restored by complementation with either ygjD_Ec or a version of QRI7_Sc designed to remain in the cytoplasm. These results also indicated that members of the YgjD/Kae1/Qri7 family were isofunctional for t⁶A synthesis, at least in yeast. To test if YgjD/Kae1/Qri7 were isofunctional in E. coli, a P_TET::ygjD strain was constructed (ygjD is only expressed when anhydrotetracycline, aTc, is added). Only the expression of the ygjD gene from E. coli allowed complementation of the essentiality phenotype in the absence of aTc. In contrast to the YrdC/Sua5 complementation results, expression of the KAE1_Sc and QRI7_Sc genes from yeast, the PRPK_Mm from Methanococcus maripaludis (PRPK is a fusion of Kae1-Bud32 in Archaea) or the B. subtilis ygjD_Bs did not complement the essentiality phenotype of the absence of ygjD.

While the protein families TsaC/Sua5 and Kae1/Qri7/TsaD were found to be strictly required for the biosynthesis of t⁶A, and a homolog of at least one member of each family is found in all domains of life (Fig. 2),^30,31 YrdC and YgjD failed to produce t⁶A in vitro with transcript or t⁶A-deficient tRNA purified from yeast sua5Δ,³⁰ suggesting that the biosynthetic machinery for t⁶A biosynthesis required more than these 2 proteins. Over the last 2 years, a flurry of papers have reported the identification of these missing proteins, and elucidated the complete bacterial,^28,33 eukaryotic/archaeal,^29,40-42 and mitochondrial^32,43,44 biosynthetic pathways to t⁶A.

Figure 2.

Distribution of genes for biosynthesis of t⁶A and derivatives. Representative organisms from each domain of life were used to build a taxonomic tree in iToL (http://itol.embl.de).^73,74 Filled circles indicate presence of genes. Genes for formation of ct⁶A are collapsed into a single column.

Synthesis of t⁶A Varies With Domains of Life

The observation that the YrdC/Sua5 family members were functionally interchangeable between domains³¹ while YgjD/Kae1/Qri7 were not³⁰ lead to a model in which t⁶A biosynthesis occurred in 2 steps with kingdom, species, or organelles specific partners for the second step.

Bacteria

The identity of the remaining enzymes in bacterial t⁶A synthesis was predicted from 3 pieces of evidence. First, YgjD was shown to form an association network with YeaZ (a paralog of YgjD) and YjeE, based on physical interaction between the proteins and physical clustering of the genes;^30,45 second, like YrdC and YgjD, YeaZ and YjeE were essential in E. coli;⁴⁶ and third, complementation of the E. coli yjgD^- essentiality phenotype required expression of both B. subtilis ygjD and yeaZ genes, suggesting that a YgjD/YeaZ interaction was necessary for t⁶A synthesis.³⁰ Notably, it's been shown that only YeaZ-YgjD pairs from closely related organisms form complexes in vitro.⁴⁷ The final evidence that YeaZ and YjeE were the missing proteins in t⁶A bacterial synthesis was provided by in vitro reconstitution experiments,²⁸ which demonstrated that recombinant YrdC, YgjD, YeaZ, and YjeE proteins from E. coli²⁸ were collectively both necessary and sufficient to generate t⁶A in reactions with threonine, bicarbonate, ATP, and either E. coli tRNA^Thr or tRNA^Lys transcripts, or unfractionated tRNA from yeast sua5Δ. Notably, t⁶A formation was not observed when a tRNA transcript corresponding to tRNA^Gln from Methanothermaobacter thermautotrophicus, which does not naturally contain t⁶A, or a 17-mer corresponding to an unmodified ASL of E. coli tRNA^Lys were used as a substrates.²⁸ While the former was consistent with the natural lack of t⁶A in this tRNA, the latter was surprising since this ASL had previously been show to bind specifically to E. coli YrdC.²⁸ The t⁶A synthesis pathway was subsequently reconstituted using the B. subtilis enzymes YwlC (an ortholog of yeast Sua5), and YdiBCE (orthologs of E. coli YjeE, YeaZ, and YgjD, respectively),³³ demonstrating the universality of these enzymes in bacteria. With the newly established enzymatic role for YeaZ, YrdC, YgjD, and YjeE (and their orthologs) in the biosynthesis of threonylcarbamoyl-6-adenosine (t⁶A), these enzymes were renamed TsaB, TsaC, TsaD, and TsaE, respectively (Fig. 1).²⁸

Archaea and Eukarya

The TsaE and TsaB proteins in Bacteria have no homologs in Eukarya or Archaea. The identification of the additional components of t⁶A biosynthesis in these last 2 kingdoms came from the fact that Kae1 was part of the KEOPS/EKC complex. The other subunits of the KEOPS/EKC complex (Bud32, Cgi121, Pcc1, plus the fungal specific Gon7) were tested for a potential role in t⁶A synthesis, first genetically then in vitro. Mutation of PCC1 and BUD32, but not in CGI121, in yeast eliminated t⁶A on tRNA^Ile reportedly by a primer extension assay.⁴¹ However, the primer extension method reported by the Sternglanz laboratory has not been repeated by others. (de Crécy-Lagard laboratory, Goldberg laboratory, and Glavic laboratory, personal communication). Indeed, as recently shown by the Alfonzo laboratory, reverse-transcriptase bypassed t⁶A to stop downstream at m³C₃₂.⁴⁸ Analysis of bulk tRNA from a pcc1–4 allele by LC-MS/MS found t⁶A was reduced 30%.⁴² HPLC and LC-MS/MS analysis of bulk tRNAs purified from whole gene deletions in S. cerevisiae found that the bud32Δ and gon7Δ strains were devoid of t⁶A, while t⁶A levels in the pcc1Δ and cgi121Δ strains were reduced 30% and 60%, respectively, versus wild-type (Thiaville and de Crécy-Lagard, unpublished data). In the halophilic Archaea Haloferax volcanii, the kae1‑bud32 (gene fusion) and cgi121 are essential, precluding a direct genetic test of their role in t⁶A biosynthesis, and deletion of pcc1 had only a small decrease (∼16%) in total t⁶A content.^40,49

Confirmation that the KEOPS/EKC complex was responsible for t⁶A formation came with in vitro reconstitution experiments. It was shown that both the KEOPS/EKC complex from Pyrococcus abyssi (Kae1, Bud32, Pcc1, and Cgi121), reconstituted from the individual genes expressed in E. coli, as well as the S. cerevisiae KEOPS complex (Kae1, Bud32, Pcc1, Cgi121, and Gon7), genes expressed in E. coli as a synthetic operon,⁵⁰ can form t⁶A in vitro, when combined with Sua5 from yeast or Archaea.²⁹

Mitochondria

Yeast mitochondrial tRNAs contain t⁶A,⁵¹ and while none the subunits of the KEOPS/EKC complex or Sua5 have paralogs targeted to the mitochondria,^52,53 the Kae1 homolog Qri7 was found to be targeted to the mitochondria in yeast,^39,53 Caenorhabditis elegans,³⁹ human,⁵⁴ rat,⁵⁴ and Arabidopsis thaliana.⁵⁴ It was subsequently demonstrated that the nuclear encoded Sua5 can localize to both the cytoplasm and to the mitochondria in yeast through the use of alternative translation initiation at 2, in-frame AUG sites.⁴⁴ Translation from the first AUG encoded a mitochondrial signal peptide, and Sua5 was localized to the mitochondria. Sua5 translated from the second AUG remained in the cytoplasm.⁴⁴ Co-expression of both Sua5 and Qri7 in E. coli complemented the TsaD essentiality when the expression of Qri7 alone did not,⁴⁴ suggesting that Qri7 could substitute for the KEOPS complex or the TsaBDE proteins. In addition, expression of QRI7 in the cytoplasm of a bud32Δ yeast strain restored growth defects.³² This was confirmed when it was demonstrated that a minimal system comprised of only Sua5 and Qri7 is sufficient to synthesize t⁶A in vitro.^32,44

Thus, the enzyme families TsaC/Sua5 and TsaD/Kae1/Qri7 are shared in all organisms, and Bacteria additionally require TsaBE, while Archaea and Eukarya use the other components of the KEOPS complex. Interestingly, although TsaBDE, KEOPS, and Qri7 are functional analogs, only the TsaD/Kae1/Qri7 protein is shared among the 3 systems (Fig. 2), suggesting that this protein family along with the TsaC/Sua5 family were part of the ancestral t⁶A synthesis core present in the last universal common ancestor (LUCA).³⁹

Mechanistic Analysis of the t⁶A Synthesis Machineries

Experiments probing the role of ATP in bacterial t⁶A formation demonstrated that both AMP and ADP were products, and that ATP consumption could be uncoupled from RNA modification, with TsaC being the source of AMP (in a threonine dependent process) and TsaD/TsaB/TsaE together producing ADP.²⁸ These observations were consistent with earlier mechanistic hypotheses³⁰ in which the ATP requirement in t⁶A biosynthesis was rationalized on the presumed need for 2 activated acyl intermediates during the course of t⁶A formation (i.e. acyl-phosphate and/or acyl-adenylate), the first a phosphocarboxy species (e.g. carboxyphosphate or carboxyadenylate; Fig. 3, intermediate I) activated for transfer to the nitrogen of either threonine or adenosine-37, and the second an N-carboxyphospho species activated for transfer to the remaining component (threonine or adenosine-37;Fig. 3, intermediate III). However, closer scrutiny of the TsaC reaction in the B. subtilis system revealed that the product was threonylcarbamoyl-adenylate (TC-AMP, Fig. 1),³³ an intermediate already activated for condensation with adenosine-37 of tRNA, thus obviating ADP formation as part of the activation steps proposed to be necessary in the biosynthesis of t⁶A. Furthermore, AMP formation by TsaC was shown to arise exclusively from hydrolysis of TC-AMP,³³ and that PP_i was the other product of the TsaC reaction, implying that formation of TC-AMP itself proceeds through an unusual direct carboxylation of threonine by CO₂ or HCO₃^- (Fig. 4).

Figure 3.

Early mechanistic proposal for the formation of t⁶A. P* refers to an activated acylphosphate species, either a simple acyl monophosphate or an acyl AMP.

Figure 4.

Stepwise formation of t⁶A illustrating the intermediates in the pathway.

The role of ADP production in t⁶A biosynthesis remains cryptic; in the bacterial system, t⁶A can be generated without formation of ADP by supplying purified TC-AMP to a reaction containing TsaD/TsaB/TsaE in the absence of ATP,³³ while in Archaea and Eukarya formation of t⁶A appears to require the reaction of ATP to ADP,⁵⁵ although it does not appear to serve a direct role in the reaction. In Archaea, Pcc1, Kae1, and Bud32 are minimally required to produce t⁶A in vitro,⁵⁵ with Kae1 comprising the catalytic subunit responsible for condensing TC-AMP with tRNA. Bud32 was shown to be an ATPase in the presence of Kae1, while it autophosphorylates when it is in complex with Cgi121. Thus, Kae1 apparently modifies the phosphotransferase activity of Bud32 and switches it from a kinase to an ATPase.⁵⁵ It's unclear what the specific role of this ATPase activity is, as neither Bud32 or Cgi121 participate directly in the t⁶A reaction. Cgi1321 appears to regulate activity by acting as an effector, where it's binding significantly changes the conformation of Bud32.⁵⁵

Overall, the chemistry of t⁶A formation bears similarities to the TobZ system,⁵⁶ an O-carbamoyltransferase comprised of a TsaC-like domain fused to a Kae1-like domain that carries out the carbamoylation of tobramycin to form nebramycin-5′. In the TobZ reaction, the TsaC domain catalyzes the conversion of carbamoylphosphate to carbamoyladenylate, while the Kae1 domain condenses the latter with tobramycin. Likewise, in t⁶A formation the TsaC/Sua5 homologs generate an adenylated intermediate, which then is condensed with tRNA by the TsaD/Kae1/Qri7 proteins. A notable difference in the systems is that in TobZ the initial substrate is carbamoyl phosphate, which undergoes a phosphotransfer reaction to generate carbamoyl adenylate. This interchange of phosphoryl moieties is chemically not necessary for the subsequent condensation of the carbamoyl group with tobramycin, as both species are activated for the coupling reaction. In t⁶A biosynthesis, the initial substrates for TsaC/Sua5 are threonine and CO₂/HCO₃^-, which react first to form N-carboxythreonine followed by reaction with ATP to form TC-AMP.

Structural Organization of t⁶A Biosynthetic Proteins

The highly reactive nature of TC-AMP is not compatible with a freely diffusible intermediate in the biosynthesis of t⁶A, and argues for the evolution of systems in which this intermediate is instead channeled, as in TobZ,⁵⁶ from its site of production in TsaC/Sua5 to a second active-site (presumably in TsaD/Kae1/Qri7) where it undergoes reaction with tRNA. A number of observations are consistent with this proposal. First is the fact that the KEOPS complex from both Eukarya and Archaea is known to be a stable quaternary complex.²⁹ Second, while the bacterial proteins do not form an isolable complex analogous to KEOPS, they do interact with one another as demonstrated by the analysis of the E. coli proteins in pull-down experiments, which demonstrated binding of TsaC to both TsaB and TsaD,²⁸ and binding of TsaB to both TsaE and TsaD.^28,45 Furthermore, the ability of B. subtilis YdiD (TsaD homolog) to complement the ΔtsaD essentiality phenotype in E. coli was dependent on co-expression of YdiB (TsaB homolog),³⁰ while the ability of Qri7 to complement this phenotype was dependent on co-expression of Sua5.⁴⁴ These observations are consistent with the requirement for specific physical interactions between these proteins necessary for function.

Additionally, crystallographic analysis has shown that protein-protein associations are conserved across the systems. For example, the crystal structure of Qri7 shows that dimerization is required for Qri7 function, and the dimerization surfaces for Qri7 are used by the archaeal/eukaryotic Kae1 binding to Pcc1 and the bacterial TsaD binding to TsaB.³² Interestingly, although the Pcc1 subunit of KEOPS/EKC shares no sequence similarity to Qri7 or TsaD, Pcc1 engages Kae1 in a manner surprisingly similar to dimerization of Qri7 and TsaD-TsaB.³²

Thus, in all 3 systems the ability of the constituent proteins to physically interact with one another appears to be a requirement for t⁶A biosynthesis.

Naming Convention

The literature is polluted with a variety of names for each t⁶A synthesis protein and even for the complexes. With the defined enzymatic and biological function now established it is appropriate to unify the t⁶A nomenclature. For all Bacteria, we recommend the following suggestions, in agreement with Ken Rudd (Curator of EcoGene, U. of Miami) and published in Deutsch, et al., of TsaB, TsaC, TsaD and TsaE, to replace YeaZ, YrdC, YgjD, and YjeE, respectively. Additionally, Sua5 in bacteria should be renamed TsaC2. TsaC2 is defined as a protein containing both a TsaC and the additional C-terminal Sua5 domain. For Eukarya and Archaea, the use of Tcs (threonyl-carbamoyl synthesis) is recommended (in yeast, TSA1 and TSA2 are in use in yeast for thioredoxin). We recommend the following nomenclature: Tcs1 (YrdC), Tcs2 (Sua5), Tcs3 (Kae1), Tcs4 (Qri7), Tcs5 (Bud32), Tcs6 (Pcc1), Tcs7 (Cgi121), and Tcs8 (Gon7). A summary of the new and old names, as well as recommended functional descriptions can be found inTable 1. Additionally, we recommend naming the bacterial TsaBDE complex as well as the archeal/eukaryotic KEOPS/EKC complex to Threonyl-carbamoly Transferase Complex (TCTC), which will be in keeping with nomenclature of other members of the carbamoyl transferase family. The TCTC family can be further subdivided into bacterial (bTCTC), archaeal (aTCTC), and eukaryal (eTCTC).

Table 1.

Proposed names and functional roles for t⁶A synthesis genes.

New Name	Old Names	Function
Bacteria
TsaB	YeaZ / YdiC	tRNA adenosine(37) threonylcarbamoyltransferase complex, dimerization subunit type 1
TsaC	YrdC	L-threonylcarbamoyladenylate synthase (EC 2.7.7.87) type 1
TsaC2	Sua5 / YwlC	L-threonylcarbamoyladenylate synthase (EC 2.7.7.87) type 2
TsaD	YgjD / YdiE	tRNA adenosine(37) threonylcarbamoyltransferase complex, transferase subunit
TsaE	YgjE / YdiB	tRNA adenosine(37) threonylcarbamoyltransferase complex, ATPase subunit type 1
Archaea / Eukaryotes
Tcs1	YrdC	L-threonylcarbamoyladenylate synthase (EC 2.7.7.87) type 1
Tcs2	Sua5	L-threonylcarbamoyladenylate synthase (EC 2.7.7.87) type 2
Tcs3	Kae1 / gcp / OSGEP	tRNA adenosine(37) threonylcarbamoyltransferase complex, transferase subunit
Tcs4	Qri7 / OSGEPL1	tRNA adenosine(37) threonylcarbamoyltransferase, mitochondrial
Tcs5	Bud32	tRNA adenosine(37) threonylcarbamoyltransferase complex, ATPase subunit type 2
Tcs6	Pcc1	tRNA adenosine(37) threonylcarbamoyltransferase complex, dimerization subunit type 2
Tcs7	Cgi121	tRNA adenosine(37) threonylcarbamoyltransferase complex, regulator subunit
Tcs8	Gon7	tRNA adenosine(37) threonylcarbamoyltransferase complex, fungal specific subunit

Distribution of the t⁶A Synthesis Genes Vary in Different Organisms

Annotation for the first enzyme of t⁶A synthesis, in bacteria, is complicated by the fact that 2 forms are found (the TsaC or TsaC2) and that 50% of the genomes analyzed harbor a TsaC paralog, YciO, that does not have the same function and does not contain the conserved KRSN tetrad.^31,57 We reannotated all members of the COG0009 family (in 9176 bacterial genomes and all contained a TsaC or a TsaC2: 6745 contain TsaC (73%), 2846 contain TsaC2 (31%), and 859 (9%) contained both. In addition, 54% contained YciO (Fig. 2, and http://tinyurl.com/t6A-bacteria). To date, no clear pattern (phylogenetic or lifestyle) has emerged in terms of presence of TsaC or TsaC2, in any given genome and the functional differences between the 2 are not understood. Most bacteria contain both TsaB and TsaE; however, TsaE can be lost in symbiotic or intracellular bacteria, such as Wolbachia or Mycoplasmas (e.g., Mycoplasma genitalium and Mycoplasma pneumoniae). To date only 2 bacteria, Mycoplasma haemofelis and Mycoplasma suis strain Illinois, are missing both TsaB and TsaE⁵⁸ (Fig. 2 and Table 2). It seems these organisms harbor a mitochondrial like minimal t⁶A synthesis system (unless another unidentified protein has been recruited).

Table 2.

Homologs of t⁶A biosynthetic genes in Bacteria.

Organism	TsaC	TsaC2 (Sua5)	TsaB	TsaD	TsaE
E. coli K12	b3282		b1807	b3064	b4168
Vibrio cholerae O1 El Tor	VC0054	VC1079	VC1989	VC0521	VC0343
Caulobacter crescentus NA1000	CCNA_03501		CCNA_00057	CCNA_00069	CCNA_03648
Mycoplasm gentialium G37	MG259*		MG208	MG046	N.P.
Mycoplasm pulmonis	MYPU_6130#		MYPU_1190	MYPU_1180	MYPU_1200
Bacillus subtilis subsp. subtilis str. 168		BSU36950	BSU05920	BSU05940	BSU05910
Haemophilus influenzae Rd	HI0656		HI0388	HI0530	HI0065
Acinetobacter baylyi APD1	ACIAD0208		ACIAD0677	ACIAD1332	ACIAD2376
Salmonella Typhii TY2	STY4395		STY1950	STY3387	STY4714
Francisella novisida U112	FTN_0158		FTN_1148	FTN_1565	FTN_0274
Pseudomonas aeruginosa PA01	PA0022		PA3685	PA0580	PA4948
Burkholderia thailandensis E264	BTH_I0669		BTH_I2001	BTH_II0616	BTH_I0723
Stapholcoccus aureus subsp aureus MW2	MW0860	MW2040	MW1975	MW1973	MW1976

^*M. gentialium MG259 is a TsaC/HemK fusion.

^#M. pulmonis TsaC (MYPU_6130) and HemK (MYPU_1060).

N.P.: Not Present.

Like Bacteria, all Eukarya and Archaea contain either a homolog of Tcs1 (TsaC/YrdC) or Tcs2 (TsaC2/Sua5). We have found one organism that has both, the fungi Pseudocercospora fijiensis CIRAD86, also known as Mycosphaerella fijiensis CIRAD86 (NCBI Taxonomic ID: 383855). As with bacteria, there is not a clear phylogenetic inheritance between organisms with Tcs1 or Tcs2 in Archaea or Eukarya, but a taxonomic relationship does exists in eukaryotes. Fungi exclusively contain Tcs2 (with P. fijiensis as an exception), while all Plants (including Chlamydomonas reinhardtii) and all Metazoans exclusively contain Tcs1. Of the 53 Archaea analyzed, 25 contain Tcs1 and 28 contain Tcs2. The only taxonomic relationship found is in the order Halobacteriales that exclusively contain Tcs1, Figure 2 and Table 3 (http://tinyurl.com/t6A-Arc-Euk).

Table 3.

Homologs of t⁶A biosynthetic genes in Archaea and Eukarya.

Organism	Tcs1 YrdC)	Tcs2 (Sua5)	Tcs3 (Kae1)	Tcs4 (Qri7)	Tcs5 (Bud32)	Tcs6 (Pcc1)	Tcs7 (Cgi121)	Tcs8 (Gon7)
Haloferax volcanii DS2	HVO_0253		HVO_1895+		HVO_1895 +	HVO_0652	HVO_0013
Homo sapiens	1p34.3		14q11.2	2q32.2	20q13.2	Xq28	2p24.3-p24.1
Drosophila melanogaster	CG10438		CG4933	CG14231	CG10673	CG42498	N.P.
Plasmodium falciparum		PFL0175c	PF3D7_1030600 PF3D7_0408900.1‡	N.P.	MAL7P1.26	N.P.	PFE0580w
Saccharomyces cerevisiae S228C		YGL169w	YKR038c	YDL104c	YGR262c	YKR095w-A	YML036w	YJL184w
Schizosaccharomyces pombe		SPCC895.03c	SPBC16D10.03	SPCC1259.10	SPAP27G11.07c	SPAC4H3.13	SPCC24B10.12	SPAC6B12.18
Arabidopsis thaliana	AT5G60590		AT4G22720	AT2G45270	AT5G26110	AT5G53045	AT4G34412

⁺H. volcanii Tcs3 and Tcs5 occur as a gene fusion (HVO_1895).

^‡P. falciparum PF3D7_0408900.1 (Tcs3b) targets to the apicoplast and is similar to Tcs3.

N.P.: Not Present.

All Archaea contain a single Tcs3 (Kae1) homolog, while Eukarya also contain a single Tcs3 homolog and also have a Tcs4 (Qri7) homolog (evolutionarily related to the bacterial TsaD), which will function in the organelles. In all genomes analyzed, both Tcs3 and Tcs4 were found in the nuclear genome and not in the organelle. Specifically, the human nuclear genome contains Tcs3 (OSGEP) for cytoplasmic t⁶A synthesis and Tcs4 (OSGEPL1) was shown to target to the mitochondria.⁵⁴ Note to the reader, the Oberto, et al. paper incorrectly referred to OSGEPL as the Tcs4 homolog (instead of OSGEPL1), the human mitochondrial targeting protein. As an example for plants, Arabidopsis thaliana contains nuclear encoded Tcs3 (AT4G22720) and Tcs4 (AT2G45270). Tcs4_At contains a strong chloroplast targeting signal, but has only been detected in the mitochondria.⁵⁴ The human pathogen Plasmodium falciparum (causative agent of malaria) presents an interesting case for t⁶A synthesis, as the mitochondria utilize fully modified cytoplasmic tRNAs for mitochondrial translation (requirement for t⁶A machinery is unknown), and P. falciparum contains an apicoplast originating from secondary endosymbiosis of an alage.⁵⁹ P. falciparum contains 2 nuclear encoded homologs of Tcs3 (Table 3): a Tcs3 that is similar to the yeast Tcs3, and an apicoplast-targeting Tcs3b, that is similar to Tcs3,⁶⁰ but is phylogenetically distant from all known Tcs3 and from the bacterial TsaD (Thiaville and de Crécy-Lagard, unpublished). Tcs3 interacts with both Tcs5 (Bud32) and Tcs7 (Cgi121), and Tcs3b interacts with multiple proteins associated with the apicoplast ribosome (Mallari and Goldberg, personal communication). Tcs2 has not been detected in the apicoplast, and it is currently unknown how the first step in t⁶A synthesis occurs. (Mallari and Goldberg, personal communication).

Tcs5 (Bud32) is found in all Eukarya and Archaea sequenced to date. In the 53 Archaea analyzed, Tcs5 and Tcs3 are adjacent ORFs in 13 genomes and are fused in 25 genomes, demonstrating a strong functional linkage between the proteins of these genes. Tcs6 (Pcc1) and Tcs7 (Cgi121) are found in nearly all Archaea and Eukarya. Notable exceptions are the absence of Tcs6 in P. falciparum and the absence of Tcs7 in Drosophila melanogaster. Tcs8 (Gon7) is a fungal specific protein. Tcs8 is required for t⁶A formation in yeast (Thiaville and de Crécy-Lagard, unpublished), but the function of Tcs8 is currently unknown.

Derivatives of t⁶A

Currently, there are 3 known derivatives of t⁶A: ct⁶A (cyclic t⁶A), m⁶t⁶A (N⁶-methyl-N⁶-threonycarbamoyladenosine), and ms²t⁶A (2-methylthio-N⁶-threonycarbamoyladenosine).¹

A new twist in the t⁶A field was recently discovered with the identification of cyclic form of t⁶A (ct⁶A), a cyclized active ester of t⁶A with an oxazolone ring.⁶¹ Renumber starting here throughout the end of the manuscript. TcdA (previously CsdL in E. coli) catalyzes an ATP-dependent dehydration of t⁶A to ct⁶A; this reaction is performed by Tcd1 (YHR003c) and by Tcd2 (YKL027w) in yeast.⁶¹ The harsh treatment for preparing tRNAs for LC-MS/MS analysis had masked the presence of the true arrangement of t⁶A, and ct⁶A appears to help tRNA^Lys decode the noncognate codons AGA and UAG.⁶¹ At least for E. coli, ct⁶A appears to occur on all t⁶A-modified tRNAs.⁶¹ Unlike TsaB, C, D, and E, TcdA is not essential for E. coli (minor growth defect), and is not universally conserved in bacteria (Fig. 2).⁶¹ Whether this represents the final functional form of t⁶A, or if this a species-specific solution for a particular problem has not been addressed. Additionally, Tcd1 and Tcd2 localize to the outer membrane, not to the surface of yeast mitochondria,⁶² and mutations in either render cells mitochondrial deficient.⁶¹ How yeast cytoplasmic tRNAs would be converted to ct⁶A is currently unknown. Also, neither ct⁶A nor homologs of TcdA have been found in Archaea, and ct⁶A does not occur in humans⁶¹ (Fig. 2).

The second known derivative of t⁶A, m⁶t⁶A, was initially thought to only occur in E. coli on the 2 tRNA^Thr_(GGU) species that decode ACC and ACU.⁶³ The limited distribution of m⁶t⁶A is confounded by the small number of organisms in which tRNAs have been sequenced.⁷ m⁶t⁶A is formed by TsaA (E. coli YaeB was recently identified as the gene responsible for TsaA activity, and renamed TrmO)⁶⁴ by transferring a methyl group from S-adenosylmethionine (AdoMet) to tRNA^Thr_(GGU) containing t⁶A.⁶³ TrmO has a unique single-sheeted β-barrel structure and does not belong to any known classes of methyltransferases, representing a novel category of AdoMet-dependent methyltransferase (Class VIII). Interestingly, t⁶A is required for the formation of the m⁶ moiety at position 37 of tRNA^Thr_(GGU), and in ΔtrmO, tRNA^Thr_(GGU) A₃₇ will be modified to ct⁶A, suggesting that t⁶A is a common precursor to both m⁶t⁶A and ct⁶A.⁶⁴ m⁶t⁶A slightly improves translational efficiency at the codon ACY.^63,64 TrmO is widely distributed throughout life and cross-kingdom functional analysis was performed to show the activity was conserved.⁶⁴ (Fig. 2)

The third known derivative of t⁶A, ms²t⁶A, is found only on tRNA^Lys_(UUU) in a subset of organisms.¹ Particularly, ms²t⁶A is found in B. subtilis, some Archaea, and in human, but not in E. coli. YqeV (MtaB) in B. subtilis and Cdkal1 in humans are responsible for the insertion of the sulfur moiety and methylation at position 2 of the adenosine containing t⁶A.^65-66 MtaB has been shown to increase the accuracy of decoding lysine codons.^66,67 Loss of the Cdkal1 homolog in mice is correlated to increase Type 2 diabetes.^67,68 It is not clear if ct⁶A is the base to form ms²t⁶A, or like m⁶t⁶A, ms²t⁶A is formed from t⁶A. (Fig. 2).

Concluding Remarks

The biosynthesis of t⁶A is just one example of a “rediscovery” of tRNA modifications in the genomic era, which has allowed for the discovery of globally unknown genes for enzyme reactions that were discovered more than 40 y earlier. In Bacteria, the 4 genes involved in t⁶A biosynthesis, due to their prokaryotic-specific essentiality and because tsaB and tsaE are found only in bacteria, had been identified as potential antibacterial and inhibitor targets prior to the discovery of their role in t⁶A synthesis was even established.^45,69-72 The unique Tcs3b found in P. falciparum also presents itself as an attractive anti-malarial target. For these proteins to be viable targets, it is critical to understand their distribution profile and potential range of action as well as the mechanisms underlying the essentiality phenotypes to predict resistance mechanisms. Clearly, one should use caution in designing drugs targeting TsaB and TsaE in Mycoplasmas spp. since the genes are absent. Caution would also be needed for drugs targeting TsaC, due to the possibility of cross reactivity in humans, although TsaC2 and TsaD may be viable options.

The discovery of the t⁶A pathways now allows us to address more systematically the causes of the pleiotropic phenotypes caused by the absence of t⁶A synthesis enzymes. Are these due to mistranslation of target proteins, to a role of t⁶A as a determinant for other components of the translation apparatus, or to a role of t⁶A or of t⁶A synthesis proteins in other processes than translation? Indeed, the recent discovery of a molecule similar to the t⁶A nucleoside in dauer signaling in nematodes⁷³ opens an unforeseen role for t⁶A derivatives in biology.

Disclosure of Potential Conflicts of Interest

No potential conflicts of interest were disclosed.

Acknowledgments

We thank Basma El Yacoubi and Jennifer Thiaville for insightful comments and editing the manuscript. We thank Jeremy Mallari, Dan Goldberg, Diego Rojas-Benítez, and Álvaro Glavic for sharing results ahead of publication.

Funding

This work was supported by the National Institutes of Health (grant number R01 GM70641 to V. de C.-L.). P.C.T. was funded in part by the Chateaubriand Fellowship from the French Embassy in the United States.