An Isolated Class II Aminoacyl-tRNA Synthetase Insertion Domain Is Functional in Amino Acid Editing*

Aminoacyl-tRNA synthetases are responsible for activating specific amino acids and transferring them onto cognate tRNA molecules. Due to the similarity in many amino acid side chains, certain synthetases misactivate non-cognate amino acids to an extent that would be detrimental to protein synthesis if left uncorrected. To ensure accurate translation of the genetic code, some synthetases therefore utilize editing mechanisms to hydrolyze non-cognate products. Previously class II Escherichia coli proline-tRNA synthetase (ProRS) was shown to exhibit pre- and post-transfer editing activity, hydrolyzing a misactivated alanine-adenylate (Ala-AMP) and a mischarged Ala-tRNAPro variant, respectively. Residues critical for the editing activity (Asp-350 and Lys-279) are found in a novel insertion domain (INS) positioned between motifs 2 and 3 of the class defining aminoacylation active site. In this work, we present further evidence that INS is responsible for editing in ProRS. We deleted the INS from wild-type E. coli ProRS to yield ΔINS-ProRS. While ΔINS-ProRS was still capable of misactivating alanine, the truncated construct was defective in hydrolyzing non-cognate Ala-AMP. When the INS domain was cloned and expressed as an independent protein, it was capable of deacylating a mischarged Ala-microhelixPro variant. Similar to full-length ProRS, post-transfer editing was abolished in a K279A mutant INS. We also show that YbaK, a protein of unknown function from Haemophilus influenzae with high sequence homology to the prokaryotic INS domain, was capable of deacylating Ala-tRNAPro and Ala-microhelixPro variants but not cognate Pro-tRNAPro. Thus, we demonstrate for the first time that an independently folded class II synthetase editing domain and a previously identified homolog can catalyze a hydrolytic editing reaction.

Aminoacyl-tRNA synthetases are the family of enzymes responsible for activating cognate amino acids with ATP to form aminoacyl-adenylates and subsequently transferring the activated amino acids onto their corresponding tRNAs (1). The aminoacylated or charged tRNA is then delivered to the ribosome for use in protein synthesis. Each aminoacyl-tRNA synthetase must select and activate its cognate amino acid from the cellular pool of 20 different proteinaceous amino acids. High fidelity in this selection process is necessary to ensure faithful protein translation as an accumulation of mistakes in the aminoacylation process will eventually lead to cell death (2)(3)(4).
Some amino acids have chemical structures that closely resemble each other, making accurate discrimination by synthetases difficult. In particular, smaller non-cognate amino acids may enter into the active sites of synthetases and be misactivated and subsequently transferred onto the wrong tRNA. Some synthetases are capable of correcting mistakes in amino acid selection through hydrolysis of the misactivated aminoacyl-adenylate prior to amino acid transfer to the tRNA (pretransfer editing) and hydrolysis of the ester linkage of the misacylated tRNA (post-transfer editing) (5).
Editing has been well documented in the case of class I synthetases. For example, class I isoleucine-tRNA synthetase (IleRS) 1 and valine-tRNA synthetase (ValRS) activate and edit the non-cognate amino acids leucine and threonine, respectively (6,7). More recently, class I leucine-tRNA synthetase (LeuRS) has also been shown to misactivate and edit isoleucine as well as a series of non-standard amino acids (8 -11). In all of these class I enzymes, a structural domain distinct from the aminoacylation active site known as connective polypeptide 1 (CP1) (12) is responsible for the editing activity (9,11,13,14). Biochemical and structural studies have also identified editing activities in class II synthetases, including editing of serine by threonine-tRNA synthetase (ThrRS) (15,16), editing of alanine by proline-tRNA synthetase (ProRS) (17,18), and editing of glycine and serine by alanine-tRNA synthetase (AlaRS) (19,20).
Synthetases are modular proteins composed of domains that have distinct functional roles (21,22). The core catalytic domain is responsible for amino acid activation and tRNA acceptor stem docking. It has been proposed that ancestral aminoacyl-tRNA synthetases consisted only of the core catalytic domain, while the anticodon-binding domain was recruited later to improve the binding and discrimination of tRNA substrates (21)(22)(23). The catalytic domain alone of some synthetases is capable of aminoacylating tRNA or microhelix substrates that mimic the acceptor stem of full-length tRNA (24,25), supporting such an evolutionary scenario. In some cases, an editing domain may have been recruited to improve amino acid specificity and eliminate non-cognate product formation (22,23). "Extra" domains such as the editing domains of certain class I synthetases have been cloned and expressed as independent functional domains. In particular, the independently expressed CP1 domains from IleRS and ValRS were shown to deacylate Val-tRNA Ile and Thr-tRNA Val , respectively (14). Interestingly the CP1 domain of LeuRS has been shown to play a critical role in group I intron splicing (26).
Editing domains found in class II aminoacyl-tRNA synthetases are significantly different from the CP1 domain responsible for editing in class I synthetases. Moreover, in the case of class II enzymes, editing domains identified to date are believed to be distinct from each other, although there is weak homology between the editing domains of ThrRS and AlaRS (15,20). Based on sequence alignments, class II ProRSs can be divided into two distinct groups (27)(28)(29). The "prokaryotic-like" group contains synthetases from bacteria and eukaryotic mitochondrial enzymes, whereas the "eukaryotic-like" group contains synthetases from Eukaryotae, Archaea, and Bacteria. In the case of Escherichia coli ProRS, a representative member of the prokaryotic-like grouping, the editing active site is a novel insertion domain (INS) located between motifs 2 and 3, which together constitute the aminoacylation active site (18). Surprisingly this ϳ200-amino acid INS is absent from eukaryotic-like ProRS. This observation suggests either that the INS domain was recruited late in evolution to enhance the performance of the prokaryotic-like enzymes (23) or that it was present early in both groupings of ProRS and then lost from the eukaryoticlike group, which generally exhibits higher amino acid specificity (30). Whereas the three-dimensional structures of two members of the eukaryotic-like ProRS group have been solved by x-ray crystallography (27,31,32), the structure of a bacterial ProRS containing the INS domain is not yet known.
Previously we have shown that E. coli ProRS is capable of hydrolyzing non-cognate Ala-AMP in a tRNA-independent pretransfer editing reaction and of deacylating Ala-tRNA Pro via post-transfer editing (17). Using alanine-scanning mutagenesis, we have also identified residues within the INS domain that are critical for both pre-and post-transfer editing (18). More recently, ProRS from the bacterium Aquifex aeolicus was shown to possess post-transfer editing activity, whereas a mutant lacking 117 residues of the insertion domain also lacked detectable deacylation activity (33). Taken together with the mutagenesis data obtained with the E. coli enzyme, these data provide strong support for the role of the prokaryotic INS domain in amino acid editing by ProRS.
By analogy to the class I CP1 domain, we hypothesized that the INS domain alone might be capable of hydrolytic editing activity. To test this hypothesis, we cloned and expressed the E. coli INS domain and tested its post-transfer editing capability. In addition, we deleted the INS from full-length E. coli ProRS to yield ⌬INS-ProRS. We also tested the Haemophilus influenzae YbaK protein, previously reported to have high homology to the bacterial ProRS INS domain (34,35), for editing activity. The data presented here provide further support for the functional role of the INS domain in translational editing and demonstrate for the first time that an independently folded domain derived from a class II synthetase can catalyze hydrolytic editing.

EXPERIMENTAL PROCEDURES
Plasmid Construction-A plasmid encoding ⌬INS-ProRS was constructed from plasmid pCS-M1S, which encodes wild-type E. coli ProRS with an N-terminal histidine tag (28). This plasmid contains a single KpnI restriction site at the N terminus of the INS domain. A second KpnI restriction site was introduced at the C-terminal end of the INS domain using the QuikChange site-directed mutagenesis kit (Stratagene) and the following primers: 5Ј-GCGCGCTGCTGGTACCCAG-GAAATG-3Ј and 5Ј-CATTTCCTGGGTACCAGCAGCGCGC-3Ј. After digesting with KpnI to excise the insertion domain, the digested plasmid was then ligated to yield the ⌬INS-ProRS with amino acid residues 249 -418 of wild-type E. coli ProRS deleted.
The plasmid containing the E. coli INS domain was prepared by PCR amplifying the INS fragment from full-length plasmid pCS-M1S using the following primers: 5Ј-CGCGGATCCATGGGGCTGGATTTCCGCG-C-3Ј and 5Ј-CCCAAGCTTGTTACGGCCATCTTCACCC-3Ј. After digesting with BamHI and HindIII, the PCR fragments were ligated into the corresponding sites of the pMAL-c2E vector (New England Biolabs) immediately following the maltose-binding protein (MBP) coding region. Following ligation, plasmids were transformed into XL-1 Blue supercompetent cells (Stratagene). Results of all cloning and deletion steps were confirmed by automated DNA sequencing (Microchemical Facility, University of Minnesota). RNA Preparation-Wild-type E. coli tRNA Pro and a G1:C72/U70-tRNA Pro triple mutant were prepared by in vitro transcription from BstNI-linearized plasmids using T7 RNA polymerase as described previously (28,36). The G1:C72/U70-microhelix Pro variant was synthesized using automated chemical RNA synthesis as described previously (37,38). Mischarged G1:C72/U70-tRNA Pro and G1:C72/U70-microhelix Pro for use in post-transfer deacylation assays were prepared as described previously (17,18). All reactions contained 2 units/ml inorganic pyrophosphatase, and reactions were quenched by the addition of acetic acid to 1% final concentration. Purified charged tRNA and microhelix were quantified by scintillation counting.
Protein Preparation-Expression of wild-type and truncated ⌬INS-ProRS was induced with 1 mM isopropyl-␤-D-thiogalactopyranoside in SG13009(pREP4) or BL21(DE3) pLysE competent cells. His-tagged proteins were then purified and prepared using a Talon cobalt affinity resin (Clontech) as described previously (39). The MBP-INS fusion protein was purified using amylose affinity chromatography according to the manufacturer's protocol (New England Biolabs). In brief, sonicated protein solution was loaded onto the amylose affinity column followed by extensive washing with 15 column volumes of column buffer (20 mM Tris-HCl, pH 7.4, 200 mM NaCl, 1 mM EDTA). The MBP-INS fusion protein was then eluted with column buffer containing 10 mM maltose. Fractions containing the fusion protein were identified via SDS-PAGE, pooled, and subjected to cleavage by 0.1% enterokinase (New England Biolabs). The latter was performed at 37°C for 1 h followed by incubation at 25°C overnight. Following the removal of maltose from the cleaved fusion protein by hydroxyapatite chromatography (Bio-Rad), the solution was loaded onto a second amylose affinity column. The cleaved INS domain was then eluted using column buffer, and fractions containing the purified protein were pooled and concentrated using Centricon 10 microconcentrators (Amicon). Purified enzyme concentrations were determined either by active site titration (full-length and ⌬INS-ProRS) or by Bradford protein assay (INS) (Bio-Rad). Concentrated protein was stored in column buffer with the addition of an equal volume of 80% glycerol.
Plasmid pCYB2_HI1434 encoding H. influenzae YbaK protein (HI1434) was a gift from Prof. Osnat Herzberg (University of Maryland). The HI1434 is expressed as a fusion protein consisting of an N-terminal HI1434 polypeptide followed by a self-splicing intein and a chitin-binding domain. Protein purification was performed essentially as described previously (35) using the IMPACT TM I system (New England Biolabs). In brief, protein expression was induced in the presence of 1 mM isopropyl-␤-D-thiogalactopyranoside, and cell pellets were harvested after 4 h of growth at 37°C. Cell pellets were then lysed by sonication and clarified by centrifugation. Cell-free lysate was then passed through a chitin column. After washing with column buffer (20 mM HEPES, pH 8.0, 500 mM NaCl, 0.1 mM EDTA, and 0.1% Triton), intein-mediated cleavage was initiated by the addition of column buffer containing 50 mM dithiothreitol. Protein was then eluted with column buffer, and fractions containing the YbaK protein were pooled and concentrated using Centriprep 10 and Centricon 10 concentrators (Amicon). Concentrated protein was stored in column buffer containing an equal volume of 80% glycerol.
Circular Dichroism (CD) Spectroscopy-CD spectra of proteins were obtained at room temperature using a J-710 spectropolarimeter (Jasco). Protein samples at a concentration of 0.5 mg/ml in 50 mM HEPES, pH 7.0 were analyzed using a 0.1-cm-path length cuvette, and spectra were accumulated over six scans. Subtraction of the INS domain spectrum from the wild-type ProRS spectrum was performed using Excel and Origin programs.
Enzyme Assays-Active site titration was performed using the adenylate burst assay as described previously (40). Cognate tRNA aminoacylation assays were performed at room temperature using the published conditions (41,42). Mischarging assays were performed using 22.5 M [ 3 H]alanine, 2 M enzyme, and 10 M tRNA as described previously (18). ATP-PP i exchange assays were performed at 37°C using the published conditions (43). The amino acid concentrations used were 0.05-2 mM (proline) and 25-500 mM (alanine), and the final enzyme concentrations were 1 nM (with proline) and 20 nM (with alanine). Kinetic parameters were determined from Lineweaver-Burk plots and represent the average of three determinations. Editing assays were performed at room temperature using the published conditions (17). In brief, ATPase assays for assessing pretransfer editing were initiated with a 1-5 M final concentration of enzymes in the presence of 3 mM [␥-32 P]ATP, 4 units/ml inorganic pyrophosphatase, 2 mM proline or 500 mM alanine. Reactions were quenched with 25 volumes of 7% HClO 4 , 10 mM NaPP i , and 3% charcoal, and the charcoal-bound ATP/ AMP was separated from the [ 32 P]P i in solution by centrifugation. A 50-l aliquot of the supernatant was quantified by liquid scintillation counting. Deacylation assays for assessing post-transfer editing were performed using 2.5-21 M protein, as indicated in the figure legends, and 20,000 -100,000 cpm mischarged RNA substrate prepared as described above. Aliquots of the reaction mixture were quenched on 5% trichloroacetic acid-soaked Whatman No. 3MM filter pads and washed extensively to remove free 3 H-labeled amino acid prior to quantifying by scintillation counting. Fig. 1 shows a schematic illustration of the domain architecture of full-length E. coli ProRS (top). Two additional proteins derived from full-length E. coli ProRS were constructed for this work. ⌬INS-ProRS, a variant with a deletion from residues 249 to 418 resulting in removal of 86% of the INS domain, is shown in Fig. 1 (bottom). The end points for this internal deletion were chosen to encompass the majority of the INS domain and also reflected convenient sites for cloning without disrupting sequences within the class II consensus motifs 2 and 3. A protein that contains the entire 183-amino acid INS domain and 65 flanking residues chosen to help stabilize the isolated domain (residues 188 -436) was also constructed (Fig. 1, middle). This construct was appended to a cleavable MBP fusion protein.

Expression and Stability of Truncated ProRS Constructs-
To evaluate the structure and folding of the truncated ⌬INS-ProRS and the cloned INS domain in comparison to full-length E. coli ProRS, we used CD spectroscopy. Fig. 2 shows the CD spectra of wild-type ProRS ( Fig. 2A, solid curve), ⌬INS-ProRS ( Fig. 2A, dashed curve), and INS (Fig. 2B, solid curve). The wild-type and ⌬INS-ProRS spectra are consistent with a combination of ␣-helical and ␤-sheet secondary structural elements with minima near 208 and 220 nm. The INS domain has a single minimum at 205 nm, suggesting a higher content of ␤-sheet and random coil elements than the full-length protein.
Subtraction of the INS domain spectrum from the wild-type ProRS spectrum yields the spectrum shown in Fig. 2A, inset (WT-INS). The close resemblance between the calculated WT-INS spectrum and the experimental spectrum for ⌬INS-ProRS ( Fig. 2A, dashed curve) provides strong support for the structural integrity of the engineered proteins.
Misactivation of Non-cognate Amino Acids-Misactivation of non-cognate alanine by wild-type E. coli ProRS and ⌬INS-ProRS was tested by assaying ATP-PP i exchange. Wild-type ProRS misactivates alanine with a k cat of 3 Ϯ 2 s Ϫ1 and K m of 300 Ϯ 30 mM, while ⌬INS-ProRS was found to misactivate alanine with a k cat of 0.4 Ϯ 0.2 s Ϫ1 and K m of 80 Ϯ 30 mM. Thus, relative to wild-type ProRS, the k cat /K m of the deletion mutant is only reduced by ϳ2-fold. These data indicate that the deletion of the INS domain results in only a moderate decrease in the ability of the ⌬INS-ProRS to activate alanine. As expected, no activation of alanine was detected with the cloned INS domain, which lacks the aminoacylation active site. We also tested the purified ⌬INS-ProRS for misactivation of glycine in the ATP-PP i exchange assay as wild-type E. coli ProRS is known to weakly misactivate non-cognate glycine (30). However, no glycine activation was detected.
To ensure that the observed alanine misactivation by ⌬INS-ProRS was not due to trace contamination of E. coli AlaRS during protein purification, we tested the purified ⌬INS-ProRS for its ability to charge an E. coli tRNA Ala transcript with alanine. No aminoacylation activity was detected using 0.5 M ⌬INS-ProRS and 0.5 M tRNA Ala . Under the same conditions, aminoacylation activity was readily detected using 0.1 M pu-  Activation of Cognate Proline, Cognate tRNA Aminoacylation, and Mischarging-Under standard reaction conditions, no activation of proline to form the aminoacyl-adenylate is observed with ⌬INS-ProRS. Therefore, although the ⌬INS-ProRS protein appeared to be globally folded with secondary structural elements that resemble those of full-length ProRS ( Fig. 2A), we speculate that deletion of the INS domain without insertion of a flexible linker sequence may cause misfolding of the adjacent (motifs 2 and 3) active site structure and thus prevent the cognate amino acid from being properly bound and activated. As expected, cognate proline activation was not detected with the independently cloned INS domain. Addition of the INS domain to the ⌬INS-ProRS construct was also attempted, but no stimulation of proline activation was detected under any conditions used.

FIG. 1. Schematic representation of full-length E. coli ProRS (top), the cloned MBP-INS fusion protein (middle), and ⌬INS-ProRS
Both the INS domain and ⌬INS-ProRS were tested for their ability to aminoacylate E. coli tRNA Pro with proline. No tRNA charging activity was detected with either protein. This is not surprising given the proline activation data described above. In addition, no non-cognate alanine mischarging activity was detected using as high as 2 M ⌬INS-ProRS and 10 M tRNA Pro .
Pretransfer Editing-ATP hydrolysis in the presence of noncognate amino acid is indicative of pretransfer editing. In the presence of 500 mM alanine, ATP hydrolysis catalyzed by wildtype ProRS was readily detected at room temperature (Fig. 3). Under the same conditions, no pretransfer editing activity was detected with ⌬INS-ProRS (Fig. 3) despite the observation that the deletion construct was capable of misactivating non-cognate alanine. Addition of purified INS (up to 3 M) did not stimulate editing by ⌬INS-ProRS (Fig. 3).
Post-transfer Editing by INS and ⌬INS-ProRS-Wild-type, ⌬INS-ProRS, and the cloned INS were assayed for their ability to deacylate mischarged Ala-tRNA Pro or Ala-microhelix Pro variants (Fig. 4). For these experiments, a triple mutation was introduced into the acceptor stem of tRNA Pro and microhelix Pro to stimulate alanine charging by E. coli AlaRS as described previously (17). No deacylation of the mischarged G1:C72/U70-Ala-tRNA Pro was detected in the presence of up to 5 M ⌬INS-ProRS or INS. Under the same conditions, the mischarged Ala-tRNA Pro variant was readily deacylated by wild-type E. coli ProRS (data not shown) (17). We also tested the ProRS deletion constructs for their ability to deacylate the mischarged G1:C72/U70-Ala-microhelix Pro . With this substrate, the INS domain alone was found to catalyze deacylation at rates similar to wild-type E. coli ProRS (Fig. 5). In contrast, no deacylation of Ala-microhelix Pro was detected in the presence of ⌬INS-ProRS.
A single K279A mutation results in a severe reduction of post-transfer editing by E. coli ProRS (18). To establish whether the observed INS deacylation activity depends on the same active site residue as in the full-length enzyme, the analogous mutant was prepared in the isolated INS domain. No deacylation of the mischarged microhelix was detected with the K279A-INS mutant (Fig. 5). CD spectroscopy confirmed that INS and K279A-INS have similarly folded structures (Fig. 2B). ogy to any other known synthetase domain with the exception of an N-terminal extension present in yeast ProRS (35). However, Wolf et al. (34) first noted significant homology between the prokaryotic INS and the YbaK family of proteins. Although the function of the YbaK proteins is unknown, the structure of the H. influenzae YbaK protein, HI1434, has recently been solved (35). We tested this INS domain homolog for deacylation of both G1:C72/U70-Ala-tRNA Pro and the Ala-microhelix Pro variant. Interestingly the YbaK protein was capable of deacylating mischarged Ala-tRNA Pro (Fig. 6A) while preserving the aminoacyl linkage of cognate Pro-tRNA Pro (Fig. 6B). The mischarged Ala-microhelix Pro was also deacylated by H. influenzae YbaK (Fig. 6B). DISCUSSION CP1 editing domains from two class I synthetases, E. coli IleRS and Bacillus stearothermophilus ValRS, have been cloned and expressed as isolated protein domains. The stably folded 211-amino acid CP1 Ile and 275-amino acid CP1 Val segments are capable of performing specific post-transfer editing of Val-tRNA Ile and Thr-tRNA Val , respectively (14). In contrast, the 260-amino acid editing domain of E. coli LeuRS (CP1 Leu ), which also appears to fold into a stable protein, was reported to lack editing function (9). Thus, in this case, it was proposed that the CP1 domain is active only in the context of the fulllength LeuRS enzyme. The present work is the first study to examine the activity of an isolated editing domain from a class II synthetase.
To determine whether the editing domain of class II E. coli ProRS is functional when separated from the rest of the enzyme, we cloned and expressed the 248-amino acid INS domain as an independent protein. In addition, ⌬INS-ProRS, a mutant construct with 86% of the INS domain deleted, was also examined for its aminoacylation and editing activities (Fig. 1). The ⌬INS-ProRS deletion construct failed to activate cognate proline or non-cognate glycine (data not shown). The crystal structure of Thermus thermophilus ProRS complexed to the prolyladenylate intermediate suggests that residues that align with Glu-410, His-413, and Phe-415, which are missing in our ⌬INS-ProRS construct, are critical for prolyl-adenylate formation (44). Since the deleted INS domain lies between motifs 2 and 3, which constitute the amino acid binding pocket, it is also possible that the large 169-amino acid deletion results in local misfolding of the active site. This structural change must be quite subtle since the active site is sufficiently formed to be able to misactivate alanine, and the CD spectrum shows the overall fold is maintained (Fig. 2A). Despite its inability to activate proline, the ⌬INS-ProRS construct was able to misactivate alanine but was defective in pretransfer editing of Ala-AMP (Fig. 3). This observation is consistent with a previous study showing that residue Asp-350 in the INS domain is critical for pretransfer editing activity as D350A-ProRS is able to activate alanine but is completely defective in hydrolyzing Ala-AMP (18). The ⌬INS deletion construct was not active in aminoacylation or in post-transfer editing (Fig. 5) either alone or in the presence of the isolated INS domain. The lack of post-transfer editing is also consistent with a previous report showing that deletion of 65% of the bacterial A. aeolicus ProRS insertion sequence abolished post-transfer editing (33).
In contrast to the ⌬INS-ProRS construct, the cloned INS domain did carry out post-transfer editing of Ala-microhelix Pro (Fig. 5). The anticodon domain of tRNA Pro is an important site of interaction with ProRS (27,28,42,45), and the inability of the INS domain to deacylate full-length mischarged tRNA may be due to the absence of binding determinants that facilitate correct positioning of the tRNA acceptor stem in the editing active site. To ensure that the observed microhelix deacylation by the isolated INS domain was dependent upon the same active site residues as in the full-length protein (18), we showed that a K279A mutation abolished post-transfer editing (Fig. 5) without disrupting the protein fold (Fig. 2B). The demonstration of microhelix editing by the ProRS INS domain emphasizes the modular architecture of synthetases and tRNAs as well as the use of appended or extra domains for amino acid editing (21)(22)(23). Although class I E. coli IleRS is also capable of deacylating a mischarged minihelix Ile (46), class II E. coli AlaRS does not edit Gly-minihelix Ala (47). In both cases, it has been shown that the entire tRNA structure is required to activate the editing reaction (47,48).
Based on phylogenetic and structural arguments, it has been hypothesized that the class II ThrRS and ProRS editing domains may have been added later in evolution than the class I CP1 domains or the class II AlaRS editing domain (20,23). The latter may be one of the most ancient editing domains and appears to have co-evolved with the active site domain (20). The unique editing domain of class II ProRS is not conserved through evolution but is present in the majority of prokaryoticlike ProRSs (27)(28)(29). A similar distribution is found with the ThrRS editing domain, which is not present in Archaea but is found in bacteria and eukaryotes.
Despite the lack of a domain homologous to the INS domain, some eukaryotic-like ProRSs, such as the archaeabacterial enzymes from Methanococcus jannaschii (30) and Methanobacterium thermoautotrophicum, 2 have been shown to possess pretransfer editing activity similar to the E. coli enzyme as well as weak post-transfer activity (30). In contrast, human ProRS lacks an INS domain, and no editing activity was detected (30). The eukaryotic-like ProRSs examined to date have an ϳ10-fold higher initial specificity for their cognate amino acid than the prokaryotic enzymes (30). In the latter case, when the relative in vivo concentration of proline and alanine are taken into account, the "effective discrimination factor" is only about 1200 (17,49), which is below the threshold value (of ϳ3300) where editing is expected to be required in vivo (50). The lack of accurate data on cellular concentrations of amino acids makes it difficult to predict the need for editing in Eukaryotae and Archaea.
The weaker editing activity detected in some of the ProRSs from the eukaryotic-like group may be a functional remnant of an activity that is no longer essential. Some enzymes within this group may therefore have either lost or never needed alanine editing capability. Alternatively a separate protein may act as a cofactor to hydrolyze mischarged products in these species. Our observation that the H. influenzae YbaK protein (HI1434), which shares significant homology with the prokaryotic ProRS INS domain, can carry out post-transfer editing of full-length Ala-tRNA Pro and Ala-microhelix Pro but not cognate Pro-tRNA Pro (Fig. 6) provides support for this proposal. sequences of the YbaK family. As shown in the structure of the H. influenzae protein (Fig. 7B), this residue (Lys-46) is located in a crevice formed by a highly curved mixed seven-stranded ␤-sheet motif (35).
The present day function of the microbial YbaK protein family (also known as EbsC) is still unknown, and the structure of HI1434 provides few insights into this open question, although a remote structural relationship between HI1434 and C-lectins has been noted (35). Using the Conserved Domain Architecture Retrieval Tool (CDART) (ncbi.nlm.nih.gov), in addition to the bacterial ProRS INS domain, 155 proteins from bacteria, 14 from Eukaryotae, and two from Archaea are identified as being homologous to HI1434. The thermophilic Archaea Aeropyrum pernix is an example of an organism that possesses a YbaK homolog along with a ProRS that lacks the INS domain. In this case, it is reasonable to hypothesize that the YbaK homolog, which shares 32% sequence identity over 105 residues with the ProRS INS from the thermophilic bacteria Thermotoga maritima (35), may be responsible for editing of Ala-tRNA Pro . Whether the YbaK proteins perform an essential editing function in vivo is an intriguing question and will require further investigation.