When Contemporary Aminoacyl-Trna Synthetases Invent Their Cognate Amino Acid Metabolism

When contemporary aminoacyl-tRNA synthetases SEE COMMENTARY invent their cognate amino acid metabolism

Herve´ Roy, Hubert Dominique Becker, Joseph Reinbolt, and Daniel Kern*

De´partement Mećanismes et Macromolećules de la Synthe`se Proteíque et Cristallogene`se, Unite´Propre de Recherche 9002, Institut de Biologie Molećulaire et Cellulaire du Centre National de la Recherche Scientifique, 15 Rue Rene´Descartes, F-67084 Strasbourg Ce´dex, France

Edited by Paul R. Schimmel, The Scripps Research Institute, La Jolla, CA, and approved June 9, 2003 (received for review April 14, 2003) Faithful protein synthesis relies on a family of essential enzymes genomes drew our attention to the fact that all archaea that called aminoacyl-tRNA synthetases, assembled in a piecewise fash- possess AsnRS (7 of 13) display a second ORF highly homolo- ion. Analysis of the completed archaeal genomes reveals that all gous to the catalytic core of AsnRS (AsnRS2) but lacking the archaea that possess asparaginyl-tRNA synthetase (AsnRS) also anticodon-binding domain. We analyzed the functional proper- display a second ORF encoding an AsnRS truncated from its ties of AsnRS2 from Pyrococcus abyssi, a hyperthermophilic anticodon binding-domain (AsnRS2). We show herein that Pyro- archaeon. The results show that this truncated AsnRS does not coccus abyssi AsnRS2, in contrast to AsnRS, does not sustain form Asn-tRNAAsn but produces Asn by amidation of Asp. asparaginyl-tRNAAsn synthesis but is instead capable of converting Phylogeny suggests this asparagine synthetase (AS) appeared aspartic acid into asparagine. Functional analysis and complemen- recently in archaea and allowed emergence of a direct and tation of an Escherichia coli asparagine auxotrophic strain show autonomous pathway of tRNA asparaginylation. This study that AsnRS2 constitutes the archaeal homologue of the bacterial describes an example of emergence of a new tRNA aminoacy- ammonia-dependent asparagine synthetase A (AS-A), therefore lation specificity and of the enzyme forming the homologous aa named archaeal asparagine synthetase A (AS-AR). Primary se- from a contemporary aaRS. quence- and 3D-based phylogeny shows that an archaeal AspRS ancestor originated AS-AR, which was subsequently transferred Materials and Methods Ϫ Ϫ into bacteria by lateral gene transfer in which it underwent General. Escherichia coli ER strain (asnA , asnB ) was from structural changes producing AS-A. This study provides evidence Genetic Stock Center (Yale University, New Haven, CT), and P. that a contemporary aminoacyl-tRNA synthetase can be recruited abyssi DNA was a gift from J.-C. Thierry (Institut de Ge´ne´tique to sustain amino acid metabolism. et de Biologie Mole´culaireet Cellulaire, Illkirch, France). The pKKET vector is a modified version of pKK223-3 (Amersham ccurate protein translation requires a complete set of 20 Pharmacia) in which the EcoRI site was replaced by the NdeI Asp Asn Aspecies of perfectly paired aminoacyl-tRNAs (aa-tRNAs). site. Thermus thermophilus tRNA , tRNA , and AspRS2 It was therefore expected that each organism should possess 20 were purified from E. coli overproducing strains. aa-tRNA synthetases (aaRSs), each capable of matching a particular amino acid (aa) to the cognate tRNA (reviewed in ref. Cloning of the P. abyssi asnS2 Gene. The ORF was amplified by 1). However, the vast number of sequences that emerged from PCR of genomic DNA with 22- and 20-nt-long sense and genome sequencing and biochemical investigations revealed antisense primers extended by NdeI and PstIorKpnI restriction anomalies that forced a revision of this assumption (2). Archaea sites for cloning of the genes in pKKET and pET vectors, respectively. AsnRS2 was expressed in E. coli ER strain (ER͞ and various eubacteria are deprived of glutaminyl-tRNA syn- ͞ thetase and more exceptionally of asparaginyl-tRNA synthetase pKKET-asnS2) and in BL21-CodonPlus-RIL strain (BL21 (AsnRS). The homologous aa-tRNAs are formed indirectly by pET-asnS2). Gln CELL BIOLOGY amidation of Glu and Asp, respectively, mischarged on tRNA ͞ and tRNAAsn by a glutamyl-tRNA synthetase (GluRS) or an Overexpression and Purification of AsnRS2. The BL21 pET-asnS2 aspartyl-tRNA synthetase (AspRS) of relaxed specificity (3–6). strain was grown overnight at 37°C in LB medium containing ͞ ampicillin, and AsnRS2 was expressed by induction with isopro- In bacteria of the Thermus Deinococcus group, deprived of the ␤ enzymes forming free asparagine (Asn), the indirect route to pyl -D-thiogalactoside. The cells were disrupted by sonication; the thermolabile proteins were flocculated by heat treatment of tRNA asparaginylation also constitutes the sole pathway of Asn ϫ biosynthesis (6, 7). the 105,000 g supernatant during 30 min at 70°C and removed The functional interrelation between the direct and the indi- by centrifugation. AsnRS2 was further purified by chromatog- rect pathways of amide aa-tRNA formation is not well under- raphies on DEAE-cellulose and hydroxyapatite followed by stood, and the phylogenetic relationship of their partners has not FPLC chromatography on an UNO-Q6 (Bio-Rad) column. been explored. Emergence of the direct pathways of tRNA From 27 g of cells, 80 mg of pure protein was isolated. glutaminylation and asparaginylation had to be accompanied by AsnRS2 Activities. ATP–PP exchange. concomitant apparition of the enzymes forming the free aa and i The reaction mixture contained 100 mM NaHepes (pH 7.2), 10 mM MgCl2,2mM the homologous aa-tRNA. The paths leading to the emergence 32 [ P]PPi,2mMATP,5mML-Asp (0.05–1.35 mM for Km of glutaminyl-tRNA synthetase and AsnRS have been docu- ␮ mented (8, 9), but the origin of the Gln- and Asn-forming measurements) or L-Asn, and 1 M P. abyssi AsnRS2. For Ki enzymes remains unclear. In particular, it is not known whether measurements, the concentration of Asn varied from 12 to 95 the aaRS and the enzyme forming the cognate free aa originated

from the same ancestor. This paper was submitted directly (Track II) to the PNAS ofﬁce. The microbial genomes are sprinkled with polypeptide chains Abbreviations: aa, amino acid; aa-tRNA, aminoacyl-tRNA; aaRS, aa-tRNA synthetase; consisting of one of the functional domains of aaRSs (10). In AsnRS, asparaginyl-tRNA synthetase; AspRS, aspartyl-tRNA synthetase; AS, asparagine some cases, the participation of these domains in a variety of synthetase; AS-A, bacterial ammonia-dependent AS; AS-AR, archaeal AS-A; AdT, tRNA- cellular processes, such as translation regulation (11), stimula- dependent amidotransferase. tion of DNA polymerase processivity (12), and aa synthesis (13), See commentary on page 9650. could be assessed. Analysis of the aaRS ORFs in archaeal *To whom correspondence should be addressed. E-mail: [email protected].

www.pnas.org͞cgi͞doi͞10.1073͞pnas.1632156100 PNAS ͉ August 19, 2003 ͉ vol. 100 ͉ no. 17 ͉ 9837–9842 Downloaded by guest on September 24, 2021 Table 1. Occurrence of AsnRS2 and enzymes involved in Asn and Asn-tRNAAsn synthesis in archaea Organism asnA* asnB* gatCAB*† asnS† asnS2

Crenarchaeota Aeropyrum pernix Ϫϩ ϩ ϪϪ Sulfolobus solfataricus ϪϪ ϩ ϪϪ Sulfolobus tokodaii Ϫϩ ϩ ϪϪ Pyrobaculum aerophilum Ϫϩ Ϫ ϩϩ Euryarchaeota Archaeoglobus fulgidus Ϫϩ ϩ ϪϪ Halobacterium sp. NRC1 Ϫϩ ϩ ϪϪ Methanobacterium thermoautotrophicum Ϫϩ ϩ ϪϪ Methanococcus jannaschii Ϫϩ ϩ ϪϪ Methanosarcina barkeri Ϫϩ ϩ ϪϪ P. abyssi Ϫϩ Ϫ ϩϩ Pyrococcus furiosus Ϫϩ Ϫ ϩϩ Pyrococcus horikoshii Ϫϩ Ϫ ϩϩ Ferroplasma acidarmanus Ϫϩ Ϫ ϩϩ Thermoplasma acidophilum Ϫϩ Ϫ ϩϩ Thermoplasma volcanium ϪϪ Ϫ ϩϩ

Only archaeal genome sequences for which the presence (ϩ) or absence (Ϫ) of the genes could be ascertained are listed. Genes encoding AsnRS2 (asn2) and the enzymes involved in either Asn (*) or Asn-tRNAAsn (†) formation, asnA, asnB, gatCAB, and asnS encoding, respectively, asparagine synthetases A and B, amidotransferase (AdT), and AsnRS are presented.

␮M, and Asp was fixed at 88, 176, or 293 ␮M. The [32P]ATP that Version 4.0.2 (15). Trees were edited with the TREEVIEW pro- formed after 2–10 min at 70°C was determined in 20-␮l gram, Version 1.5. aliquots (6). Aminoacylation. The reaction mixture contained 100 mM NaHepes Results ␮ 14 (pH 7.2), 30 mM KCl, 10 mM MgCl2,10 M L-[ C]Asp or Existence of Two Genes Encoding AsnRS in P. abyssi. Analysis of the L-[3H]Asn, 320 ␮M unfractionated E. coli tRNA or 30 ␮M annotated genome from P. abyssi reveals two ORFs related to enriched Thermus thermophilus tRNAAsp or tRNAAsn, and 0.1 AsnRS, one encoding the full-length enzyme and the other one, ␮M P. abyssi AsnRS2 or Thermus thermophilus AspRS2. The a shorter isoform, identified as AsnRS2 sharing 39% identity aa-tRNA formed after 1–30 min at 50°C was determined in 40-␮l with the canonical AsnRS. The gene encoding this truncated aliquots (6). AsnRS (asnS2) was found in six other archaeal genomes (16) but Amidation. The reaction mixture contained 100 mM NaHepes (pH not in bacterial or eukaryal genomes. Surprisingly, all archaea 7.2), 10 mM MgCl2,10mMNH4Cl, L-Asn or L-Gln, 10 mM ATP, that encode AsnRS2 also possess the full-length AsnRS (Table ␮ 0.5 mM L-Asp (0.5–10 mM for Km measurements), and 2 M 1). Alignment of AsnRS2 with the archaeal AsnRS and AspRS AsnRS2. The reaction was conducted at 70°C. For character- shows this isoform lacks the 100 first aa, which in other AsnRS ization of Asn, [14C]Asp was used and 4.5-␮l aliquots were are organized in the anticodon-binding domain (Fig. 1); never- transferred in 1 ␮l of acetic acid after a 5-min to 3-h incubation. theless, the catalytic core displays the three consensus motifs For characterization of the ATP products, the reaction mixture characterizing class II aaRS in which the invariant and semi- contained 40 ␮M[␥-or␣-32P]ATP, 40 ␮M L-Asp, and 0.6 ␮M invariant residues are mostly conserved (17). To assign a func- AsnRS2; after a 10-min to 5-h incubation, 20-␮l aliquots were tion to these truncated AsnRS, we expressed P. abyssi AsnRS2 diluted in 80 ␮l of water; after phenol͞chloroform and chloro- in E. coli and analyzed its properties. form extractions, the aqueous phase was dried and the products were suspended in 20 ␮l of water. The reaction products were Cloning, Expression, and Analysis of Substrate Specificity of P. abyssi analyzed by TLC on cellulose plates of 0.5-␮l aliquots. Asp and AsnRS2. Analysis of the extract of the BL21͞asnS2 strain showed Asn were separated with the solvent 2-propanol͞formic acid͞ presence of a thermostable protein corresponding to the 294- acetic acid͞water, 80͞10͞10͞4 (vol͞vol͞vol͞vol); ATP was sep- aa-long polypeptide chain (molecular mass of 36,061 Da) en- arated from its products with the solvent isobutyric acid͞25% coded by asnS2. Gel filtration of the purified protein revealed a ammonia͞water, 50͞1.1͞28.9 (vol͞vol͞vol). The labeled prod- molecular mass of 60 kDa, whereas SDS͞PAGE showed ucts were revealed by image plate with a Fuji Bioimager. The polypeptide chains of 34 kDa, suggesting a homodimeric struc- phenylthiohydantoin derivatized products were separated on a ture for the native protein. Brownlee PTC, C-18, 5-␮m column (220 ϫ 2.1 mm2) and Aminoacylation assays on either unfractionated E. coli tRNA analyzed on an Applied Biosystems 420A Derivatizer. or enriched Thermus thermophilus tRNAAsn were unsuccessful with AsnRS2 under conditions where Thermus thermophilus Phylogenetic Analysis. Alignments were performed by using the AsnRS forms Asn-tRNA (not shown). Further, AsnRS2 was also CLUSTALX program, Version 1.81 (14), and refined with the 3D unable to form asparaginyl adenylate (AsnϳAMP), as revealed structures of the proteins (Protein Data Bank). The trees were by the absence of Asn-dependent ATP–PPi exchange (not constructed from a bootstrap of 500 replicates obtained from shown). Surprisingly, the enzyme stimulated the exchange in the analysis of an ungapped alignment of 215 residues inferred by the presence of Asp with a rate constant comparable to that of neighbor-joining method by using the PHYLIP package, Version AspRS2 from Thermus thermophilus, an archaeal-type AspRS 3.57c (J. Felsenstein, University of Washington, Seattle), with (respectively 1.0 and 1.75 sϪ1 at 70°C), demonstrating its capacity the Dayhoff PAM matrix; they were reconstructed with the to form AspϳAMP. However, AsnRS2 was unable to aspartylate maximum-likelihood method by using the PUZZLE program, tRNA. Despite its inability to activate Asn, AsnRS2 binds Asn,

9838 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.1632156100 Roy et al. Downloaded by guest on September 24, 2021 SEE COMMENTARY

Fig. 1. Alignments of AsnRS2 with archaeal AsnRS and AspRS. The alignment of 7 archaeal AsnRS (N),19 archaeal AspRS (D), and 7 AsnRS2 (2) is summarized. The sequences of the enzymes of P. abyssi (P࿝abyssi) and the consensus sequences encompassing the consensus motifs 1, 2, and 3 of class II aaRSs are shown. The motifs are schematized by black boxes located at the top or bottom of the sequences, and the invariant residues are indicated in white letters. The percentage homology is symbolized by shading: black and dark-gray shadings denote 100% and 80% of homology, and light-gray shading denotes conservation of the chemical nature in at least 80% of residues; dots symbolize lack of a residue in the detailed sequences and homology lower than 80% in the consensus sequences. Nomenclature of the consensus sequences is the following: a, aromatic; h, hydrophobic; Ϫ and ϩ, negatively and positively charged; p, polar or small residues; ˆ refers to a gap. Residues interacting with the aa (}) or the adenylate (F) moiety of AspϳAMP or with both (■) in the 3D structure of the complex of Pyrococcus AspRS (22) are indicated.

Ϫ1 which acts as a competitive inhibitor of Asp (not shown). Affinity for Asp of 4–5mMandakcat of 2–3s were determined. The ϩ of AsnRS2 for Asn exceeds 10 times that for Asp (Km for Asp is affinity for NH4 was high because the contaminating ions ␮ ␮ 180 M and Ki for Asn is 16 M), contrasting with the strict partially stimulated the amidation. Finally, among the various specificity of AspRS and AsnRS for their cognate aa. nucleoside triphosphates only ATP promotes amidation.

Characterization of the Reaction Catalyzed by AsnRS2. Inability of P. abyssi AsnRS2 Confers Asn Prototrophy. To unambiguously prove AsnRS2 to aspartylate tRNA narrowed the possible role for this that AsnRS2 ensures Asn synthesis in vivo, we attempted protein down to AS or argininosuccinate synthase, which, like complementation of the E. coli ER strain, auxotrophic for Asn ϳ AsnRS2, use Asp and ATP as substrates and form Asp AMP. (19), by transformation with pKKET-asnS2. Fig. 4 shows that P. Inspection of the biochemical background of archaea shows that abyssi asnS2 was able to confer Asn prototrophy to the trans- all are missing asnA, the gene that encodes, in eubacteria, the formed E. coli strain. This result constitutes strong evidence that ammonia-dependent AS-A (Table 1). However, with two excep- AsnRS2 serves as AS-A in P. abyssi. tions, all archaea possess the glutamine-dependent AS-B; half of CELL BIOLOGY them possess also the AdT able to catalyze the tRNA-dependant Asn formation, whereas the other half encode AsnRS2 (Table 1). We therefore hypothesized that AsnRS2 might be an enzyme catalyzing the ammonia-dependent formation of free Asn. Fig. ϩ 2 shows that AsnRS2 was able, in the presence of NH4 ,to convert Asp into a component of the same chromatographic mobility as Asn (lane 3). Formation of Asn was confirmed by (i) derivatization of the compounds of the reaction mixture with phenylisothiocyanate, showing formation of phenylthiohydantoin-Asn (not shown) and (ii) transfer of the radiolabeled aa formed in the amidation reaction and isolated by TLC, onto tRNA by Thermus thermophilus AsnRS, which exclusively charges tRNA with Asn (data not shown). Selectivity of AsnRS2 for free ammonia as an amide group donor (Fig. 2, lanes 2 and 3) and absence of glutaminase or asparaginase activities able to produce free NH3 (lanes 4 and 5) reveal a mechanistic resemblance to AS-A (18). Formation of Asn is correlated with conversion of ATP into AMP (Fig. 3). Quantification of the end products along the ͞ kinetic curves shows formation of Asn and AMP in a 1 1 Fig. 2. Identiﬁcation of the end products formed by P. abyssi AsnRS2 by TLC. stoichiometry. However, consumption of ATP exceeds forma- Phosphor images of TLC plates of the amidation mixture containing [14C]Asp tion of AMP (Fig. 3B) because of an accumulation of a small without amide group donor (lane 2) or with NH4Cl (lane 3) or containing amount of ADP. This side-product appears only under amida- unlabeled Asp and either [3H]Asn (lane 4) or [3H]Gln (lane 5); lanes 1 and 6, Ϸ tion conditions and represents 8% of the ATP consumed. A Km controls with labeled aa.

Roy et al. PNAS ͉ August 19, 2003 ͉ vol. 100 ͉ no. 17 ͉ 9839 Downloaded by guest on September 24, 2021 Fig. 3. Kinetics of substrate consumption and of end-product formation by AsnRS2. (A) Consumption of [14C]Asp (Œ) and formation of [14C]Asn (F). (B) Consumption of [␣-32P]ATP (᭜) and formation of [32P]AMP (■) and [32P]ADP (ϫ). The labeled reactants were fractionated by TLC (Inset) and quantiﬁed by using the IMAGE GAUGE software (Fuji), and their amounts are expressed as the percentage of the total radioactivity in the assay.

Retracing Appearance and Evolution of the Autonomous tRNA Aspar- found in 6 of 12 species of bacteria possessing AS-A (20) and in aginylation Pathway. The tree reconstructed by alignment of two archaea. AsnRS2 with AspRS and AsnRS sequences shows two main groupings, one constituted by the bacterial AspRS and the other Mapping AsnRS2 Active Site. The alignments of AsnRS2 with AS-A one in which archaeal and eukaryal AspRS are intermixed with (Fig. 6A) refined by the 3D structure of E. coli AS-A complexed AsnRS and AsnRS2 (Fig. 5). Branching of the AsnRS2 clade with AMP and Asn (21) allowed identification of the essential suggests that this enzyme derives from an AspRS ancestor that residues of P. abyssi AsnRS2 and mapping them in the active site also evolved AsnRS and the archaeal͞eukaryal AspRS. The of E. coli AS-A. All residues of AS-A contacting the Asn most probable event that could explain apparition of AsnRS2 substrate are strictly conserved at almost the same position in would be two gene duplications (Fig. 5). The first was that of the AsnRS2 with only one exception (A74 in AS-A replaced by I77 AspRS ancestor, from which one copy evolved into archaeal͞ in AsnRS2, Fig. 6 A and B). The same strategy was used to map eukaryal AspRS and the other one into AsnRS. The second was AsnRS2 residues (Figs. 1 and 6A) in the 3D structure of the that of AsnRS, from which one copy originated AsnRS2 and the active site of Pyrococcus AspRS complexed with AspϳAMP other one conserved the function of AsnRS. The fact that AsnRS (22). Most of the residues conserved in the active sites of AS-A, is present in the three phyla whereas AsnRS2 is archaeal-specific AsnRS2, and contacting Asn are conserved in AspRS and suggests that AsnRS appeared just before the split of archaea contact Asp. Indeed, of the 11 residues involved in recognition and eukaryotes and AsnRS2 after the split. The high bootstrap values obtained along the evolutionary path leading to AsnRS2 (in thick bars) supports this scenario. Because the topology of the trees was independent of the method used, bias inherent to the use of a particular method was minimized. Because of the absence of significant sequence similarity of AS-A with AspRS and AsnRS, introduction of AS-A sequences in the alignment resulted in (i) a significant sensitivity of the topology of the trees to the method used for their reconstruction and (ii) a drop of Ϸ50% of the bootstrap values of all branches (data not shown). Nevertheless, the AS-A clade was always branched to the AsnRS2 lineage, suggesting that bacteria did not evolve AS-A but acquired this enzyme by lateral gene transfer from archaea. This event is supported by the conservation of a synteny involving asnA in bacteria or asnS2 in archaea and the gene encoding a transcription regulator (asnC). This synteny is

Fig. 5. Attempt to pinpoint the origin of the archaeal ammonia-dependent Asn metabolism. The tree is based on an ungapped alignment done by CLUSTALX Fig. 4. Complementation of the Asn auxotrophy of E. coli ER strain by the P. of 7 AsnRS2, 60 AspRS, 49 AsnRS, and 61 lysyl-tRNA synthetase (LysRS) se- abyssi asnS2 gene. The E. coli ER strain was transformed with either the pKKET quences. LysRS sequences were used to root the tree. Nodes statistically vector or the recombined pKKET-asnS2 vector and grown on minimal M9 relevant [bootstrap value of 100% (᭛)orϾ80% (E)] are indicated. The medium agar plates supplemented with ampicillin, and 0.5 mM isopropyl evolutionary path of AsnRS2 is indicated by thick bars. The arrows indicate the ␤-D-thiogalactoside in the presence (ϩ) or absence (Ϫ)of40␮g͞liter L-Asn. gene duplications. The scale bar represents 10 substitutions per 100 aa.

9840 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.1632156100 Roy et al. Downloaded by guest on September 24, 2021 SEE COMMENTARY

Fig. 6. Alignment-based identification of P. abyssi AsnRS2 active site residues. (A) Comparative alignment of AS-A (A), AsnRS2 (2), and AspRS (D). Only blocks of sequences encompassing the active-site residues involved in binding of Asp (arrowheads) are shown. The equivalent of AspRS flipping- and motif 2- loops are indicated by black boxes located at the top of the alignment. Nomenclature for the alignment is indicated in the Fig. 1 legend. (B) Comparison of the 3D structures of Pyrococcus AspRS and E. coli AS-A active sites. Coordinates of AspRS complexed to AspϳAMP (22) and of AS-A complexed to Asn and ATP (21) were used. Hydrogen bonds are symbolized by dashed lines and water molecules by small spheres, and ‡ refers to residues of the same chemical nature spatially conserved in AspRS and AS-A active sites. Residues in white letters over black boxes are those of P. abyssi AsnRS2 identified in the alignment with AS-A shown in A as potential candidates for recognition of Asp. (C) Schematized 2D representation comparing recognition of Asp by Pyrococcus AspRS and AsnRS2. The residues interacting with Asp in AspRS and suggested to interact with Asp in AsnRS2 are displayed. Residues in ellipses are spatially conserved in AspRS and AsnRS2; residues in white letters in black ellipses determine different orientations of Asp in the two active sites; hydrogen bonds are symbolized by dashed lines.

of Asn in AS-A (and in AsnRS2) and of Asp in AspRS, 8 are (AS-A and AsnRS2) and AspRS. The flipping loop, which strictly conserved but differ in the interactions they make with entraps AspϳAMP in the active site of AspRS (22, 23) and flips the substrate. Additionally, of the 8 residues involved in recog- back in its open position upon binding of the cognate tRNAAsp nition of AMP in the AspRS complex (23), 6 are conserved in to allow the nucleophilic attack of the activated ␣-COOH by AsnRS2 and AS-A and contact AMP (Fig. 1). However, the 3 tRNA, might entrap AspϳAMP and the ammonia substrate in ␣ ␣ ␤ residues contacting the -COOH, -NH2, and -CO groups of AS. Motif 2 loop, which in AspRS is involved in ATP binding and the aa substrate are not conserved (Fig. 6B). In AsnRS2, D195, establishes base-specific hydrogen bonds with G73 and C74 of equivalent to D219 in AS-A and involved in recognition of the the tRNA acceptor stem (23), might prevent recognition of ␣ -NH2 group of the substrate, is replaced in Pyrococcus AspRS tRNA by AS. by I340. In the second strand of motif 2 of Pyrococcus AspRS, S229, which binds the ␣-COOH group of Asp with a water- Discussion mediated bond, is replaced by Q116 in both AsnRS2 and AS-A, This investigation proves that AsnRS2 is the archaeal orthologue

which now interact with the ␤-COOH group of the substrate. of the bacterial ammonia-dependent AS-A. Therefore we re- CELL BIOLOGY Finally, Q192, the last residue of the QSPQ motif of AspRS, named this protein archaeal AS-A (AS-AR). Previous studies ␣ ͞ which contacts the -NH2 group of Asp, is replaced in AS-A and suggested that AS-A might have evolved from an archaeal AsnRS2 by A74 and I77 and does not interact with the substrate. eukaryal ancestor (24, 25). This hypothesis was supported by the Nonconservation in AsnRS2 (or AS-A) and AspRS of the structural resemblance at the 3D level between E. coli AS-A and residues contacting identical substrate groups provides the struc- the catalytic core of yeast AspRS (21). However, the lack of tural basis for the reversed orientation of the Asp substrate in the significant sequence similarities did not allow us to establish the active centers (Fig. 6C), and suggests that conversion of the phylogeny between these enzymes. We show that AS-AR con- ␣-COOH-activating site of AspRS into the ␤-COOH-activating stitutes the link in the evolutionary path that led to emergence site of AS-A or AsnRS2 was achieved by only three substitutions: of ammonia-dependent Asn synthesis. An AspRS ancestor orig- I340 to D195, Q192 to I77, and S229 to Q116 (Fig. 6C). inated AS-AR through two gene duplication events, one in the The active sites of AS-A and AsnRS2 differ additionally from archaeal͞eukaryal progenitor and another one later in archaea. that of AspRS by the flipping and motif 2 loops that might be The phylogenetic analysis suggests that bacterial AS-A results related to differences in the reactions they catalyze: AS-A and from a lateral gene transfer of archaeal AS-AR. Thus the AsnRS2 transfer the activated Asp onto ammonia and AspRS absence of sequence homology of archaeal AspRS with AS-A, onto tRNAAsp. Comparison of the 3D structures of E. coli AS-A contrasting with their high degree of identity with AS-AR, and Pyrococcus AspRS and alignments with AsnRS2 show suggests that whereas the gene in bacteria underwent structural conservation of an acidic residue in the flipping loops, Glu in changes, the archaeal gene, for reasons that would have to be AspRS (E170 in Pyrococcus AspRS) and Asp in AS (D46 in clarified, remained untouched. A similar phenomenon occurred AS-A and D47 in AsnRS2), both contacting by hydrogen bond in the evolutionary history of bacterial AsnRSs, which segregate ␣ the -NH2 group of the substrate (Fig. 6B). In addition, the in two subgroups, one typically bacterial (Fig. 5, group 1) and flipping and motif 2 loops differ in size. In AS-A the flipping another one archaeal-like (Fig. 5, group 2). loop is five residues longer than in AspRS and AsnRS2, whereas Three prerequisites were needed for evolving AS from AspRS: in AsnRS2 the motif 2 loop is longer than in AspRS and in AS-A. (i) switch in activation of the ␣-COOH group of the Asp These differences probably reflect distinct functionalities of AS substrate (for AspRS) to the ␤-COOH group (for AS), (ii) loss

Roy et al. PNAS ͉ August 19, 2003 ͉ vol. 100 ͉ no. 17 ͉ 9841 Downloaded by guest on September 24, 2021 of transfer of the ␤-COOH-activated Asp onto tRNA, and (iii) aminoacylation capacity of an AspRS activating group other binding of ammonia and transfer onto the activated ␤-COOH than the ␣-COOH of Asp. Because in subclass IIb aaRS residues group of Asp. Acquisition by AspRS of the capacity to activate from anticodon constitute the major elements responsible for this COOH group had to be accompanied by the loss of tRNA efficient aminoacylation (reviewed in refs. 30 and 31), loss of the aminoacylation to prevent isopeptide bond formation in proteins anticodon-binding domain will impair the capacity of these and premature termination of peptide elongation on ribosomes. aaRSs to bind and aminoacylate their tRNA substrate. Further, because the activated Asp can react with nucleophilic Synthesis of Asn is distinctive because it can be produced by acceptors (26), loss of the tRNA charging capacity of the three different and unrelated enzymes, AS-A (or AS-AR), AS-B ␤-COOH-activating AspRS had to be correlated with acquisition and AdT. AS-B and AdT are found in the three domains of life, of the amidation reaction to prevent release of AspϳAMP. suggesting that they were already present in the latest common So far, four aaRSs truncated from their anticodon-binding ancestor. AS-A, however, is found only in a subset of pathogenic domain have been described: GenX (27), BirA (28), HisZ (13), bacteria and in seven archaea. Furthermore, almost all organ- and AlaX (10), respectively homologues of Lys-, Ser-, His-, and isms that possess AS-A also contain AS-B or AdT. Only three AlaRS. Their cellular functions are diverse, and they are dis- organisms, two bacteria (Haemophilus influenzae and Pasteurella tributed in all kingdoms of life. HisZ functionally most resembles multocida) and one archaea, (Thermoplasma volcanium) escape AS-AR. Although this enzyme does not form aa directly, it this rule, and strictly depend on AS-A or AS-AR for Asn constitutes the functional subunit of the ATP-phosphoribosyl- synthesis. The reason why AS-A predominantly coexists with transferase, involved in the His biosynthetic pathway (13). Thus, another AS is unclear. Presence of AS-AR in all archaea that to our knowledge, AS-AR constitutes the first example of a display AsnRS suggests that this Asn synthetase has especially truncated aaRS capable of synthesizing its cognate aa. been acquired by archaea to allow emergence of the direct and Absence of anticodon-binding domain raises the question autonomous pathway of tRNA asparaginylation. whether these isoforms evolved from complete aaRSs that lost their anticodon-binding domain or from simpler aaRS ancestors We thank E. Schmitt (Ecole Polytechnique, Palaiseau, France) and D. that had not yet acquired this domain. Because AS-AR evolved Moras (Institut de Geńe´tique et de Biologie Molećulaire et Cellulaire, from an archaeal AspRS ancestor that also originated AsnRS Illkirch, France) for the coordinates of the 3D structures of Pyrococcus Ϸ (Fig. 5), and AspRS and AsnRS both display an anticodon- AspRS complexed to Asp AMP and R. Giege´ (Institut de Biologie binding domain organized in an OB-fold (29), the AspRS Molećulaire et Cellulaire, Strasbourg, France) for suggestions and critical reading of the manuscript. This work was supported by the ancestor had to already possess this domain. Otherwise, one has Universite´ Louis Pasteur (Strasbourg), the Centre National de la Re- to consider the highly improbable event that bacterial, archaeal, cherche Scientifique et Technique, and a grant from the Association and eukaryal AsnRS and AspRS independently acquired the de la Recherche contre le Cancer. H.R. is a recipient of a fellowship same anticodon-binding domain. The loss of this module was from Ministe`redel’Education Nationale de la Recherche et de la probably imposed by a strong selection pressure to destroy Technologie.

1. Ibba, M., Francklyn, C. & Cusack, S., eds. (2003) The Aminoacyl-tRNA 16. Altschul, S. F., Gish, W., Miller, W., Myers, E. W. & Lipman, D. J. (1990) J. Synthetases (Landes Bioscience, Georgetown, TX), in press. Mol. Biol. 215, 403–410. 2. Ibba, M., Becker, H. D., Stathopoulos, C., Tumbula, D. L. & So¨ll, D. (2000) 17. Eriani, G., Delarue, M., Poch, O., Gangloff, J. & Moras, D. (1990) Nature 347, Trends Biochem. Sci. 25, 311–316. 203–206. 3. Wilcox, M. & Nirenberg, M. (1968) Proc. Natl. Acad. Sci. USA 61, 229–236. 18. Cedar, H. & Schwartz, J. H. (1969) J. Biol. Chem. 244, 4122–4127. 4. Curnow, A. W., Hong, K., Yuan, R., Kim, S., Martins, O., Winkler, W., Henkin, 19. Felton, J., Michaelis, S. & Wright, A. (1980) J. Bacteriol. 142, 221–228. T. M. & So¨ll, D. (1997) Proc. Natl. Acad. Sci. USA 94, 11819–11826. 20. Kolling, R., Gielow, A., Seufert, W., Kucherer, C. & Messer, W. (1988) Mol. 5. Curnow, A. W., Ibba, M. & So¨ll, D. (1996) Nature 382, 589–590. Gen. Genet. 212, 99–104. 6. Becker, H. D. & Kern, D. (1998) Proc. Natl. Acad. Sci. USA 95, 12832–12837. 21. Nakatsu, T., Kato, H. & Oda, J. (1998) Nat. Struct. Biol. 5, 15–19. 7. Curnow, A. W., Tumbula, D. L., Pelaschier, J. T., Min, B. & So¨ll, D. (1998) 22. Schmitt, E., Moulinier, L., Fujiwara, S., Imanaka, T., Thierry, J.-C. & Moras, Proc. Natl. Acad. Sci. USA 95, 12838–12843. D. (1998) EMBO J. 17, 5227–5237. 8. Lamour, V., Quevillon, S., Dirion, S., NЈGuyen, V. C., Lipinski, M. & Mirande, 23. Moulinier, L., Eiler, S., Eriani, G., Gangloff, J., Thierry, J.-C., Gabriel, K., M. (1994) Proc. Natl. Acad. Sci. USA 91, 8670–8674. McClain, W. H. & Moras, D. (2001) EMBO J. 20, 5290–5301. 9. Woese, C. R., Olsen, G. J., Ibba, M. & So¨ll, D. (2000) Microbiol. Mol. Biol. Rev. 24. Gatti, D. L. & Tzagoloff, A. (1991) J. Mol. Biol. 218, 557–568. 64, 202–236. 25. Hinchman, S. K., Henikoff, S. & Schuster, S. M. (1992) J. Biol. Chem. 267, 144–149. 10. Schimmel, P. & Ribas De Pouplana, L. (2000) Trends Biochem. Sci. 25, 207–209. 26. Kern, D., Lorber, B., Boulanger, Y. & Giege´, R. (1985) Biochemistry 24, 11. Kaniga, K., Compton, M. S., Curtiss, R. & Sundaram, P. (1998) Infect. Immun. 1321–1332. 66, 5599–5606. 27. Kong, L., Fromant, M., Blanquet, S. & Plateau, P. (1991) Gene 108, 163–164. 12. Carrodeguas, J. A., Kobayashi, R., Lim, S. E., Copeland, W. C. & Bogenhagen, 28. Artymiuk, P. J., Rice, D. W., Poirrette, A. R. & Willet, P. (1995) Nat. Struct. D. F. (1999) Mol. Cell. Biol. 19, 4039–4046. Biol. 1, 758–760. 13. Sissler, M., Delorme, C., Bond, J., Ehrlich, S. D., Renault, P. & Francklyn, C. 29. Murzin, A. G. (1993) EMBO J. 12, 861–867. (1999) Proc. Natl. Acad. Sci. USA 96, 8985–8990. 30. Giege´, R., Florentz, C., Kern, D., Gangloff, J., Eriani, G. & Moras, D. (1996) 14. Thompson, J. D., Gibson, T. J., Plewniak, F., Jeanmougin, F. & Higgins, D. G. Biochimie 78, 605–623. (1997) Nucleic Acids Res. 25, 4876–4882. 31. Kern, D., Roy, H. & Becker, H. D. (2003) in The Aminoacyl-tRNA Synthetases, 15. Schmidt, H. A., Strimmer, K., Vingron, M. & von Haeseler, A. (2002) eds. Ibba, M., Francklyn, C. & Cusack, S. (Landes Bioscience, Georgetown, Bioinformatics 18, 502–504. TX), in press.

9842 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.1632156100 Roy et al. Downloaded by guest on September 24, 2021