Evolutionary Divergence of the Archaeal Aspartyl-tRNA Synthetases into Discriminating and Nondiscriminating Forms*

Asparaginyl-tRNA (Asn-tRNA) is generated in nature via two alternate routes, either direct acylation of tRNA with asparagine by asparaginyl-tRNA synthetase (AsnRS) or in a two-step pathway that requires misacylated Asp-tRNAAsn as an intermediate. This misacylated aminoacyl-tRNA is formed by a nondiscriminating aspartyl-tRNA synthetase (AspRS), an enzyme that in addition to forming Asp-tRNAAsp also misacylates tRNAAsn. In contrast, a discriminating AspRS cannot acylate tRNAAsn. It has been suggested that the archaeal AspRS enzymes are nondiscriminating, whereas the bacterial ones discriminate. The archaeal and bacterial AspRS proteins are indeed distinct in sequence and structure. However, we show that both discriminating and nondiscriminating forms of AspRS exist among the archaea. Using unfractionated methanobacterial and pyrococcal tRNA, theMethanothermobacter thermautotrophicus AspRS acylated approximately twice as much tRNA as did AspRS from Pyrococcus kodakaraensis or Ferroplasma acidarmanus. Proof that Asp-tRNAAsn was generated by the methanogen synthetase was the conversion of Asp-tRNA formed by M. thermautotrophicusAspRS to Asn-tRNA by M. thermautotrophicusAsp-tRNAAsn amidotransferase. In contrast, Asp-tRNA formed by the Pyrococcus or Ferroplasma enzymes was not a substrate for the amidotransferase. Also, although all three AspRS enzymes charged tRNAAsp transcripts, only M. thermautotrophicus AspRS aspartylated the tRNAAsntranscript. Genomic analysis provides a rationale for the nature of these enzymes. The mischarging AspRS correlates with the absence in the genome of AsnRS and the presence of Asp-tRNAAsnamidotransferase, employed by the transamidation pathway. In contrast, the discriminating AspRS correlates with the absence of the amidotransferase and the presence of AsnRS, forming Asn-tRNA by direct aminoacylation. The high sequence identity, up to 60% between discriminating and nondiscriminating archaeal AspRSs, suggests that few mutational steps may be necessary to convert the tRNA-discriminating ability of a tRNA synthetase.

by each aminoacyl-tRNA synthetase is essential (1). The remarkable ability of these enzymes to discriminate between numerous tRNA molecules has long been the focus of studies in protein-nucleic acid recognition (reviewed in Refs. 2 and 3). Despite the need for precision, at least two misacylating tRNA synthetases happily exist in nature. Both aspartyl-tRNA synthetase (AspRS) 1 and glutamyl-tRNA synthetase (GluRS) occur in mischarging forms, often in organisms that lack asparaginyl-tRNA synthetase (AsnRS) or glutaminyl-tRNA synthetase (GlnRS) (reviewed in Ref. 4). These organisms generate Asn-tRNA or Gln-tRNA by pre-translational amino acid modification. For instance, in place of direct Asn-tRNA formation by AsnRS (Fig. 1A), a nondiscriminating AspRS misacylates tRNA Asn with aspartate (Fig. 1B). A second enzyme, the tRNA amidotransferase GatCAB, converts Asp-tRNA Asn by amidation to the correct translational substrate, Asn-tRNA Asn (Fig. 1B). The heterotrimeric bacterial GatCAB enzyme can also form Gln-tRNA from Glu-tRNA Gln (4). The absence of the genes encoding AsnRS (asnS) or GlnRS (glnS) in most completed microbial genome sequences (4 -6) emphasizes that the transamidation pathways are widespread in nature.
Sequence analyses indicate that the AspRSs form two distinct phylogenetic branches (5-7), essentially a bacterial and an archaeal/eukaryotic branch. An AsnRS branch also separates both AspRS groups in some phylogenetic analyses (6,7). The phylogenetic segregation of the AspRSs into two groups is also supported by comparisons of AspRS structures from the three domains (8 -11). Several residues critical for binding of the Asp-AMP adenylate are unique to the eukaryote-and archaea-type AspRSs and are distinct from those used in the bacteria-type enzyme (10). Comparison of AspRS amino acid sequences reveals a relatively long (ϳ570 residue) enzyme common in the bacteria, compared with a shorter (ϳ430 residue) enzyme in the archaea. The size difference is mainly due to an insertion domain in the bacteria-type AspRS that is absent in the archaea-type enzyme (6,12,13).
Biochemical analyses led to the proposal that the bacterial AspRS was a strictly discriminating enzyme, whereas the archaeal AspRS could aspartylate tRNA Asp and tRNA Asn (10,13,14). Both Thermus thermophilus and Deinococcus radiodurans were shown to contain a discriminating bacteria-type AspRS, which charged only tRNA Asp , and a nondiscriminating ar-chaea-type AspRS, which also misacylated tRNA Asn (14 -16). However, a recent study indicated that the 583-residue, bacteria-type AspRS found in Chlamydia trachomatis could mischarge the chlamydial tRNA Asn when it was expressed along with Escherichia coli tRNA (17). This result suggests that the evolution and discriminating functions of the AspRSs may be more complex than originally envisioned.
Here we show that the archaea-type AspRSs are also divided into discriminating and nondiscriminating types based on activity and function. The nature of tRNA discrimination by the archaeal AspRS enzymes appears to be correlated with the genomic presence of asnS (18 Preparation of tRNA Transcripts-The genes encoding the F. acidarmanus tRNA Asn and tRNA Asp were contained in contig 145 (tRNA 2) and contig 144 (tRNA 1), respectively, of the organism's sequence data (genome.ornl.gov/microbial/faci/12oct00/trna_summary.html). For the F. acidarmanus tRNA Asn , position 68 in the tRNA was changed from A to G to yield a C-G base pair in the acceptor stem. The M. thermautotrophicus tRNA Asn is from MTH1276 (20). The tRNA Asn sequences of Pyrococcus horikoshii (PHtRNA32; Ref. 21), P. furiosus (www. genome.utah.edu/Pfu102000.gb), and Pyrococcus abyssi (www.genoscope.cns.fr/Pab/Pabyssi_genetic_elements.fasta) were identical, as were all of the tRNA Asp sequences (PHtRNA42; Ref. 21). The tRNA genes were synthesized from two overlapping oligonucleotides that included a T7 promoter and a BstNI site as described previously (16). Following the in vitro transcription reaction (18), the transcripts were extracted with phenol and chloroform, precipitated with ethanol, and purified on a Q-Sepharose column (20 cm ϫ 0.2 cm 2 ) in Buffer A (10 mM HEPES, pH 7.5, 1 mM Na 2 EDTA, 7 M urea). Nucleotides were removed with 3 column volumes of Buffer A plus 0.4 M NaCl, and transcripts were eluted with a linear gradient of 0.4 -2 M NaCl in Buffer A. Fractions containing the transcripts as judged by a denaturing polyacrylamide gel were pooled, extracted with phenol and chloroform, and precipitated with ethanol. Excess urea was removed by a Nap-10 gel filtration column (Amersham Biosciences). Transcripts were resuspended in water and stored at Ϫ80°C.
Preparation of Unfractionated tRNA-Unfractionated tRNA was purified from 5 g of M. marburgensis and P. furiosus cells as described (15) using extractions by phenol and chloroform:isoamyl alcohol (24:1), precipitation in ethanol, and purification on DEAE-cellulose (DE52, Whatman). Phenol extractions were continued until the aqueous phase was clear. For purification from P. furiosus cells, 0.1 M NaCl was included in the cell resuspension. The yields based on A 260 measurements were 0.38 mg of tRNA/g of M. marburgensis cells and 0.61 mg of tRNA/g of P. furiosus cells.
Cloning . Oligonucleotide primers introduced an NdeI site at the 5Ј-end of each gene and a BamHI site at the 3Ј-end of the M. thermautotrophicus and P. kodakaraensis genes, or an XhoI site at the 3Ј end of the F. acidarmanus gene. After cloning into the pCR2.1-TOPO vector (Invitrogen, Carlsbad, CA) and sequence confirmation, the genes were cloned into pET20b or pET28a (Novagen, Madison, WI). The NdeI-BamHI fragment containing the M. thermautotrophicus AspRS gene and the NdeI-XhoI fragment containing the F. acidarmanus AspRS gene were cloned into pET28a to yield Nterminally His 6 -tagged proteins. The NdeI-BamHI fragment containing the P. kodakaraensis gene was cloned into the pET20b vector to yield native enzyme. The operon encoding the M. thermautotrophicus Asp-tRNA Asn amidotransferase (18) was also cloned into the NdeI-SacI sites of pET20b without the stop codon of the gatB gene. All proteins were expressed from the E. coli BL21-CodonPlus(DE3)-RIL strain (Stratagene), which expresses the E. coli minor tRNA genes argU, ileY, and leuW from a chloramphenicol-resistant vector. The cells were grown at 37°C in LB (5 g of yeast extract, 10 g of tryptone, 10 g of NaCl, pH 7.0, per liter) with 34 g/ml chloramphenicol. For pET28a clones, 25 g/ml kanamycin was included, the cells were induced with 0.4 mM isopropyl-1-thio-␤-D-galactopyranoside at an A 600 of 0.5, and growth was continued at 15°C for 16 h before harvesting. For pET20b clones, 100 g/ml ampicillin was included, the cells were induced with 1 mM isopropyl-1thio-␤-D-galactopyranoside at an A 600 of 1.3, and growth was continued at 37°C for 4 h before harvesting.
Enzyme Purification-All steps were performed at 4°C except where noted. All enzymes were Ͼ95% pure as judged by Coomassie-stained SDS-polyacrylamide gel electrophoresis (not shown). The His 6 -tagged AspRSs were purified from 5 g of cells on Ni-NTA-agarose (Qiagen, Valencia, CA). The cells were resuspended in Buffer B (25 mM HEPES-KOH, pH 7.2, 20 mM imidazole, 0.3 M NaCl, 1 mM benzamidine-HCl, 5 mM ␤-mercaptoethanol (␤-ME)) plus 0.5 mg/ml lysozyme and 50 M phenylmethylsulfonyl fluoride, and sonicated on ice with a Branson 250 sonifier (Branson Ultrasonics Corp., Danbury, CT), at an output of 50, three times for 30 s with a 2-min resting. The supernatant from a 100,000 ϫ g centrifugation of the cell suspension was mixed for 1 h with 1.5 ml of the Ni-NTA-agarose that was pre-equilibrated in Buffer B. A column (2.5 cm ϫ 0.5 cm 2 ) containing the resin and protein was washed with 20 volumes of Buffer B and then with 10 volumes of Buffer B containing 0.5 M NaCl and 10% glycerol. The AspRSs were eluted with Buffer B containing 250 mM imidazole. The M. thermautotrophicus AspRS was further purified on a Superdex 200 column (11 cm ϫ 3 cm 2 , Amersham Biosciences) in Buffer C (25 mM HEPES-KOH, pH 7.2, 0.4 M KCl, 0.1 mM Na 2 -EDTA, 5 mM ␤-ME, 5% glycerol). The AspRSs were concentrated on polyethylene glycol (PEG; 15,000 -20,000 M r in dialysis tubing (12,000 -14,000 molecular weight cut-off) and dialyzed into Buffer C containing 20 mM KCl and 50% glycerol (v/v) for storage at Ϫ20°C. The M. thermautotrophicus Asp-tRNA Asn amidotransferase containing a C-terminal His 6 tag on the GatB subunit was also purified by this method, but it included an additional wash of the Ni-NTA column with 10 volumes of Buffer B containing 50 mM imidazole.
P. kodakaraensis AspRS purification (10) was performed with modifications. Cells (26 g) were resuspended in 250 ml of lysis buffer (10 mM KH 2 PO 4 , pH 7.5, 20 mM KCl, 5 mM ␤-ME, 4 g/ml lysozyme, and one Complete protease inhibitor mixture tablet (Roche Molecular Biochemi- cals)) and incubated overnight. Cells were lysed by 10 min of sonication, as described above, for 2 s plus 2 s of resting. An extract was prepared by centrifugation at 11,000 ϫ g for 30 min. The extract was heat-treated at 80°C for 15 min to precipitate the majority of the E. coli proteins and centrifuged as above. Additional proteins were removed with a 0 -50% (saturated) ammonium sulfate precipitation plus centrifugation, and the AspRS was precipitated with 50 -80% (saturated) ammonium sulfate. The centrifuged precipitate was resuspended in 10 ml of water, dialyzed overnight against Buffer D (10 mM Tris-HCl, pH 7.5, 100 mM KCl, 5 mM ␤-ME) and applied to a 20-ml UnoQR column (Bio-Rad) pre-equilibrated with Buffer D. Protein was eluted at 2 ml/min with a 300-ml linear gradient of 100 -600 mM KCl in Buffer D. Active fractions were pooled, dialyzed overnight against Buffer D, concentrated on PEG as described above, and dialyzed again. The protein (1 ml aliquots) was loaded onto the Superdex 200 column (see above) pre-equilibrated in Buffer D. The active fractions were pooled, treated with 0.3% (w/v) streptomycin sulfate to precipitate residual nucleic acids, filtered (0.2 m), and dialyzed overnight against Buffer D. The enzyme was concentrated on PEG and dialyzed into Buffer D containing 50% glycerol (w/v) for storage at Ϫ20°C.
Aminoacyl-tRNA Synthetase Assays-Because the organisms used here grow in very diverse environments, the aminoacylation reaction conditions were optimized for each enzyme as well as for the tRNA source. The pH of each buffer stock, at 10-fold concentration, was adjusted at the temperature used in the assay. M. thermautotrophicus grows optimally at 65-70°C at a neutral pH (22). Therefore, for the M. thermautotrophicus AspRS, pH 7.0 (HEPES-KOH) and 37-60°C were used for the assays. Using unfractionated tRNA, aminoacylation activity with the P. kodakaraensis AspRS reportedly drops above 65°C (23). Under our conditions, the optimal activity for this enzyme was at 60°C and pH 6.0 (not shown), and therefore pH 6.0 (MES-KOH) and 45-60°C were used. F. acidarmanus was isolated from a habitat at 40°C and at a pH between 0 and 1. The intracellular pH values of Ferroplasma species have not been reported. However, the intracellular pH for their phylogenetic relative, Thermoplasma acidophilum, was pH 6.0 when the extracellular pH ranged from 0.5 to 3.0 (24). At 37°C, the optimal activity of the F. acidarmanus AspRS was at pH 6.0 (not shown), and so pH 6.0 (MES-KOH) and 37°C were used.
For asparaginylation and aspartylation assays in cell extracts, cells Prior to the assay, the transcripts were reannealed by heating to 90°C, followed by gradual cooling to 25°C, and including the addition After incubation at the relevant temperature for 40 min, the substrate was isolated by phenol extraction and ethanol precipitation as described (15). The tRNA amidotrans-ferase assay has been described also (15). Although the optimal growth temperature of M. thermautotrophicus is near 65°C, the reaction with the His-tagged M. thermautotrophicus Asp-tRNA Asn amidotransferase was performed at 37°C as a compromise between optimal enzyme activity and the stability of the reaction product.

Formation of Archaeal Asn-tRNA by Two Different Routes-
The transamidation route to Asn-tRNA was suggested in the archaea when direct asparaginylation of tRNA could not be obtained in Haloferax volcanii or in Methanothermobacter (18,25,26). However, direct asparaginylation of unfractionated tRNA has been observed in the pyrococcal species P. furiosus (18). Using a tRNA Asn transcript, we tested the ability of archaeal cell extracts to directly charge the tRNA with asparagine or, alternatively, with aspartate to provide a substrate for the GatCAB amidotransferase. The P. furiosus extract was able to asparaginylate the transcript directly, whereas the Methanothermobacter extracts had no such activity ( Fig. 2A). However, the Methanothermobacter extracts could form Asp-tRNA Asn , whereas the P. furiosus extract lacked this ability (Fig. 2B). This result confirms the nondiscriminating nature of M. thermautotrophicus AspRS that was predicted by the presence of a GatCAB tRNA amidotransferase (18). Therefore, some archaea have a discriminating AspRS and an AsnRS (Fig.  1A), whereas others use a nondiscriminating AspRS and the GatCAB amidotransferase for Asn-tRNA formation (Fig. 1B).
The AspRSs from P. kodakaraensis and F. acidarmanus Are Discriminating, whereas the AspRS from M. thermautotrophicus Is Nondiscriminating-To identify whether the archaeal AspRSs possess different forms of tRNA Asn discrimination, the enzymes from three euryarchaea, M. thermautotrophicus, P. kodakaraensis, and F. acidarmanus, were expressed in E. coli and purified. For the substrate, unfractionated tRNA was purified from M. marburgensis (the same genus as M. thermautotrophicus) and P. furiosus (the same genus as P. kodakaraen- sis). The aspartylation levels of both tRNA preparations were tested with the AspRSs because discrimination can arguably be affected by differences in the sequence or modification of the tRNA. All three enzymes charged M. marburgensis (Fig. 3A) and P. furiosus (Fig. 3B) tRNA with aspartate. However, the M. thermautotrophicus AspRS aspartylated both tRNA sources to a higher level (approximately twice as much) than the enzymes from either P. kodakaraensis or F. acidarmanus (Fig. 3,  A and B). Although the amounts of tRNA Asp and tRNA Asn have not been described in these archaea, the charging levels shown in Fig. 3, A and B, are comparable with the amounts of tRNA Asn (0.6 -1.2%) and tRNA Asp (0.8 -1.3%) that were determined in Bacillus subtilis, E. coli, and Mycoplasma capricolum using two-dimensional gel electrophoresis (Ref. 27 and references therein). These results suggest that the M. thermautotrophicus AspRS aminoacylated tRNA Asn and tRNA Asp , in contrast to the AspRSs from P. kodakaraensis or F. acidarmanus.
To determine whether tRNA Asn was the additional tRNA species aminoacylated by M. thermautotrophicus AspRS, the Asp-tRNAs described in Fig. 3, A and B, were subjected to the action of M. thermautotrophicus tRNA amidotransferase. The transamidation product, [ 14 C]Asn-tRNA, was generated only in the presence of the GatCAB enzyme and only with the [ 14 C]Asp-tRNA formed with the M. thermautotrophicus AspRS, not with the AspRSs from P. kodakaraensis and F. acidarmanus (Fig. 3, C and D). The [ 14 C]Asn-tRNA was also formed regardless of whether the unfractionated tRNA was from Methanothermobacter (Fig. 3C) or Pyrococcus (Fig. 3D), indicating that the ability to misacylate the tRNA was a property of the AspRS and not of the tRNA. Each lane on the TLC also contains a spot corresponding to [ 14 C]Asp, presumably representing the Asp-tRNA Asp that was formed upon aspartylation of unfractionated tRNA and that is not a substrate for the amidotransferase (15,17). Thus, the tRNA amidotransferase assay indicates that of the three archaeal AspRSs, only the M. thermautotrophicus enzyme can mischarge tRNA Asn .
The data given above were confirmed by testing the aspartylation of the tRNA Asn and tRNA Asp transcripts. The P. kodakaraensis and M. thermautotrophicus tRNA transcripts were poorly charged (Ͻ10%) under a range of conditions by all As-pRSs tested except at high enzyme concentrations (not shown). The poor aminoacylation levels of these transcripts may result from incorrect folding because of a lack of modifications that have been shown to be necessary for stability of the tRNA at high temperatures (28). However, the F. acidarmanus tRNA transcripts could be highly charged. The F. acidarmanus (Fig.  4A) and P. kodakaraensis (Fig. 4B) AspRS could aspartylate the tRNA Asp transcript but not the tRNA Asn transcript. However, the tRNA Asn transcript could be aspartylated by the M. thermautotrophicus AspRS (Fig. 4, A and B). Taken together, the data show that the P. kodakaraensis and F. acidarmanus AspRSs are discriminating, unlike previously characterized archaea-type AspRSs (14 -16), whereas the M. thermautotrophicus enzyme can make both Asp-tRNA Asp and Asp-tRNA Asn .

Archaea Differ in Their Uses of the Redundant Pathways to
Asn-tRNA Formation-The synthesis of amide aminoacyl-tRNA in bacteria and archaea occurs by one of the two redundant routes depicted in Fig. 1 (16, 18). Crucial for this task is either AsnRS or a nondiscriminating AspRS combined with the GatCAB tRNA amidotransferase. Most organisms have only one of the redundant pathways. However, the bacteria D. radiodurans and T. thermophilus are the best studied examples of species possessing both pathways. However, in these organisms GatCAB-dependent Asn-tRNA formation is required for asparagine synthesis (14 -16). Surveying archaeal genomes for the genes encoding these enzymes revealed a striking correlation between the presence of asnS and the lack of gatCAB (Table I).
The archaea are apparently divided in their use of the two Asn-tRNA Asn pathways (Fig. 1). Some archaea generate Asn-tRNA directly using AsnRS, whereas others use the Asp-tRNA Asn amidotransferase, which is encoded by gatCAB. The latter pathway requires a nondiscriminating AspRS to form Asp-tRNA Asn , the amidotransferase substrate. We have demonstrated this mischarging ability in M. thermautotrophicus  (Table I), a misacylating AspRS is unnecessary and possibly harmful. We have shown in vitro that the AspRSs from two such representative archaea, P. kodakaraensis and F. acidarmanus, are discriminating (Figs. 3 and 4). No genomic sequence is available for P. kodakaraensis. However, the three other pyrococcal genomes contain asnS homologs and lack homologs of GatCAB. In addition, we have demonstrated AsnRS activity in P. furiosus (Fig. 2 and Ref. 18). We suggest that a discriminating AspRS is common to the pyrococcal group, despite the attractive observation that misacylation may be achieved by a shortened loop identified in the P. kodakaraensis AspRS structure (10), which corresponds to a loop in the yeast AspRS structure that interacts with the C36 in the tRNA Asp anticodon (41). Position 36 is possibly a major discriminating element because it is the only anticodon position to differ between tRNA Asp and tRNA Asn . For example, the effects of changes at this position on the E. coli AspRS were significant even though such changes were secondary to changes at positions 34 and 35 (42). The discriminating mechanism of the archaea-type enzyme will best be addressed by biochemical experiments that are based on structural complexes of the archaea-type AspRS and tRNA.
Table I further illustrates that there is no redundancy of archaeal Asn-tRNA formation as is found in T. thermophilus and D. radiodurans (14,15). These bacteria can alternately use exogenously supplied asparagine, which is directly aminoacylated by AsnRS, or can biosynthesize the amino acid in a strictly tRNA-dependent manner (14 -16). Instead, all archaea likely generate asparagine using asparagine synthetase, encoded by the archaeal asnB homolog, although this has not been explicitly demonstrated. The asnB homolog appears to be ubiquitous in the archaea (20, 21, 30 -40), despite the apparent ability of many archaea to also biosynthesize asparagine via transamidation.
Finally, Table I suggests a division of the roles of the two archaeal tRNA amidotransferases. GlnRS has not been identified in any archaeon (4,18,25), and so Gln-tRNA Gln is predicted to be formed by a Glu-tRNA Gln amidotransferase. Two such enzymes with this activity have been described in the archaea, the GatCAB and GatDE amidotransferases (18). The archaeal GatCAB can generate in vitro both Gln-tRNA Gln and Asn-tRNA Asn , whereas GatDE can synthesize only Gln-tRNA Gln . However, although GatDE is found in all completed archaeal genomes, corresponding to a complete lack of GlnRS in the archaea, GatCAB is found only in the absence of AsnRS (Table I). The mechanism underlying this possible division in asparaginyl-tRNA formation is unclear. It is possible that this extraordinary dichotomy between GatCAB and AsnRS is a chance occurrence. This possibility seems less likely for two reasons. In archaea that have GatCAB, GatDE appears redundant, yet GatDE remains in these archaea. Also, the two routes to Asn-tRNA Asn (Fig. 1) are not grouped together phylogenetically but exist in both the euryarchaeal and crenarchaeal kingdoms (Table I).
In contrast, the bacterial GatCAB possesses the ability to form in vivo both Gln-tRNA Gln and Asn-tRNA Asn (15), and no GatDE exists in the bacterial domain (18). These facts explain why the nature of the bacterial AspRS will not be so easily predictable from genomic analyses as in the archaea. The presence of GatCAB in a bacterial genome suggests the possibility of either the Asn-tRNA or the Gln-tRNA transamidation pathway and therefore the possibility of either a misacylating As-pRS or a misacylating GluRS. A misacylating aminoacyl-tRNA synthetase is supported by the absence of AsnRS or GlnRS (4), however, the presence of AsnRS or GlnRS does not preclude a misacylating enzyme, as was shown in T. thermophilus and D. radiodurans (14,15). These bacteria contain both GlnRS and AsnRS and yet also have a nondiscriminating AspRS. Therefore, the discriminating nature of many bacterial AspRSs will only be revealed upon biochemical analysis (14,15,17).
Biochemical and Evolutionary Bases of AspRS Discrimination-A comparison of the tRNA Asn sequences from the archaea listed in Table I indicates little in the tRNA sequence that would convincingly affect discrimination by the AspRSs. There is no single base difference in the tRNA Asn sequences between archaea that are predicted in Table I to have discriminating AspRSs compared with those that have misacylating AspRSs (not shown). This observation is consistent with our results indicating that the source of the tRNA did not affect formation of the substrate for the amidotransferase (Fig. 3, C and D). These results also minimize the possibility that a tRNA modification is involved in archaeal AspRS discrimination. At present, the effect of modifications is difficult to address in the absence of known modifications in more than one archaeal tRNA Asn (25).
Because discriminating and nondiscriminating forms apparently exist in both the bacteria-type AspRSs (17) and the archaea-type AspRSs, it appears that AspRSs have diverged at least twice with respect to their tRNA discriminating abilities. The fact that two discriminating forms are now found in both the bacteria-type (17) and archaea-type AspRSs may not be surprising upon comparison of their N-terminal domains, which are responsible for the specific recognition of the tRNA anticodon loop. These ϳ130-residue domains are closer in sequence between the bacterial and the archaeal AspRSs than either is with the eukaryotic AspRS (10,12), which is still a presumably discriminating AspRS. No eukaryon is known to lack AsnRS in the cytoplasm, and no mischarging eukaryal AspRS has yet been described. In fact, there may be much similarity in the mechanisms distinguishing the discriminating forms in both the archaeal and bacterial AspRSs. The investigation of such mechanisms in tRNA discrimination will likely illustrate the developments in codon recognition that can occur naturally (5,6) or through genetic engineering (43).