Mechanism of tRNA-dependent editingin translational quality controlJiqiang Ling*, Herve Roy†, and Michael Ibba*†‡

*Biochemistry Program and †Department of Microbiology, Ohio State University, Columbus, OH 43210

Edited by Dieter Soll, Yale University, New Haven, CT, and approved November 7, 2006 (received for review July 25, 2006)

Protein synthesis requires the pairing of amino acids with tRNAscatalyzed by the aminoacyl-tRNA synthetases. The synthetases arehighly specific, but errors in amino acid selection are occasionallymade, opening the door to inaccurate translation of the geneticcode. The fidelity of protein synthesis is maintained by the editingactivities of synthetases, which remove noncognate amino acidsfrom tRNAs before they are delivered to the ribosome. Althoughediting has been described in numerous synthetases, the reactionmechanism is unknown. To define the mechanism of editing,phenylalanyl-tRNA synthetase was used to investigate differentmodels for hydrolysis of the noncognate product Tyr-tRNAPhe.Deprotonation of a water molecule by the highly conserved resi-due �His-265, as proposed for threonyl-tRNA synthetase, wasexcluded because replacement of this and neighboring residueshad little effect on editing activity. Model building suggested that,instead of directly catalyzing hydrolysis, the role of the editing siteis to discriminate and properly position noncognate substrate fornucleophilic attack by water. In agreement with this model, re-placement of certain editing site residues abolished substratespecificity but only reduced the catalytic efficiency of hydrolysis 2-to 10-fold. In contrast, substitution of the 3�-OH group of tRNAPhe

severely impaired editing and revealed an essential function forthis group in hydrolysis. The phenylalanyl-tRNA synthetase editingmechanism is also applicable to threonyl-tRNA synthetase andprovides a paradigm for synthetase editing.

proofreading � translation � phenylalanine

Aminoacyl-tRNA synthetases (aaRSs) maintain fidelity duringprotein synthesis by attaching amino acids to their cognate

tRNAs. Quality control by aaRSs is monitored at the aminoacyla-tion step, with the active site distinguishing cognate from noncog-nate amino acids with great accuracy. Erroneous activation maybecome a risk to the cell when the cognate amino acid exhibitsstructural similarities to other natural compounds. In these aaRSs,an editing mechanism hydrolyzes the misactivated aminoacyl-adenylate (pretransfer editing) (1–3) or the mischarged aminoacyl-tRNA (aa-tRNA) (posttransfer editing) (4–10).

The aaRSs are divided into two structurally unrelated groups,classes I and II, both of which contain examples of enzymes withediting activities (11–13). Editing sites in the two classes share littlein common. Class I aaRSs including isoleucyl-tRNA synthetase(IleRS) (14, 15), leucyl-tRNA synthetase (LeuRS) (9), and valyl-tRNA synthetase (ValRS) (16) harbor the editing activity in thewell conserved CP1 domain, which is inserted in the catalyticdomain. Class II aaRSs contain more diverse editing sites, none ofwhich bears any resemblance to the class I CP1 domain. Within classII, threonyl-tRNA synthetase (ThrRS) and alanyl-tRNA synthetaseediting domains share strong sequence similarities with each other(8, 17, 18) but are not homologous to that of prolyl-tRNA syn-thetase (19). Escherichia coli phenylalanyl-tRNA synthetase(PheRS) possesses a predominant posttransfer editing activityagainst the misaminoacylated species Tyr-tRNAPhe (10). The ed-iting domain of this enzyme, and of its archaeal/eukaryal counter-part (20), is located in the B3/B4 region of the �-subunit and lacksstructural features resembling any known editing domains (21).

Despite the emerging knowledge on the biological functions andstructures of aaRS editing sites, little is known about the molecularmechanisms of either pretransfer or posttransfer editing. Therecently resolved crystal structure of LeuRS complexed with aposttransfer editing analog suggested that the editing site binds thesubstrate and positions a catalytic water molecule (ref. 9, discussedin ref. 22). In contrast, structural studies of the ThrRS editing sitesuggested that two water molecules are specifically activated byediting site residues and subsequently hydrolyze the posttransferediting substrate (18). Here we used available PheRS structures todevelop different models for editing and probed the possiblemechanisms using site-directed mutagenesis and biochemical anal-yses. We found that the moderate catalytic efficiency and specificityof PheRS editing are mainly achieved through substrate binding byseveral conserved residues, whereas the chemistry is not rate-limiting during the editing step. Hydrolysis is catalyzed by two watermolecules, which are positioned by editing site residues, and mostimportantly an indispensable interaction with the 3� OH of theterminal adenosine of tRNAPhe. The PheRS editing mechanism canalso be applied to ThrRS and provides a paradigm for synthetaseediting.

ResultsFunctional Domains in the E. coli PheRS Editing Site. We previouslyfound that the B3/B4 domain of the �-subunit of E. coli PheRSharbors an editing activity against misactivated Tyr (10). Therecently resolved crystal structure of Thermus thermophilus PheRScomplexed with Tyr supported this assignment of the B3/B4 domain(21), as did the absence of editing activity in mitochondrial PheRS,which lacks the �-subunit (23). To probe the roles of the differentdomains of the PheRS editing site we first replaced residues whoseside chains form putative interactions with Tyr (Fig. 1) and anumber of neighboring positions. To accurately determine kineticparameters for editing, all replacements were made in the contextof the �A294G variant, which has previously been shown to betteraccommodate Tyr in the active site than wild-type PheRS withoutaffecting editing (10). Editing activities of the resulting PheRSvariants were analyzed by using ATP consumption assays, whichmeasure both pre- and posttransfer editing of misactivated Tyr (cisediting), and Tyr-tRNAPhe hydrolysis assays, which measure rates oftrans editing of exogenous substrate (Table 1). Most replacementstested had little or no effect on editing, including �P263 and �Y360,which were suggested to form edge-to-face interactions with the Tyrring, and numerous other residues in the vicinity of the carboxyl

Author contributions: J.L., H.R., and M.I. designed research; J.L. and H.R. performedresearch; J.L., H.R., and M.I. analyzed data; and J.L. and M.I. wrote the paper.

The authors declare no conflict of interest.

This article is a PNAS direct submission.

Abbreviations: aaRS, aminoacyl-tRNA synthetase; aa-tRNA, aminoacyl-tRNA; ValRS, valyl-tRNA synthetase; IleRS, isoleucyl-tRNA synthetase; LeuRS, leucyl-tRNA synthetase; ThrRS,threonyl-tRNA synthetase; PheRS, phenylalanyl-tRNA synthetase.

‡To whom correspondence should be addressed at: Department of Microbiology, Ohio StateUniversity, 484 West Twelfth Avenue, Columbus, OH 43210-1292. E-mail: ibba.1@osu.edu.

This article contains supporting information online at www.pnas.org/cgi/content/full/0606272104/DC1.

© 2006 by The National Academy of Sciences of the USA

72–77 � PNAS � January 2, 2007 � vol. 104 � no. 1 www.pnas.org�cgi�doi�10.1073�pnas.0606272104

group of Tyr (21). These results indicate significant differences inthe binding of free Tyr and Tyr-tRNAPhe at the editing site,suggesting that the PheRS:Tyr complex depicts a postediting statedifferent from the transition state. This finding is consistent withprevious structural studies of ThrRS, where the product (Ser) anda substrate analog (SerA76) of posttransfer editing were shown tobind differently at the editing site (18).

In an effort to find other domains of the editing site, we

extended our search to include well conserved residues on thesurface of the B3/B4 region [supporting information (SI) Fig. 8and Table 1]. As with the structure-based replacements, themajority of the changes had little effect on editing. Takentogether, the editing site replacements revealed that �R244,�H265, �G318, �E334, �T354, and �A356 are all involved inediting (Table 1). �G318 replacements, as with the previouslydescribed �A356 changes (10), hinder access to the editing site,leading to the most significant decreases in activity. �R244,�H265, �E334, and �T354 replacements all showed less dra-matic changes and were further tested for editing defects withrespect to their tyrosylation activities (Fig. 2). In agreement withthe kinetic data, �G318W exhibited the best misacylation activ-ity and �E334A PheRS, which had 5-fold reduced trans editing

0

0.2

0.4

0.6

0.8

0 2 4 6 8 10 12

AN

Rt-ryT

ehP

( µ

)M

Time (min)

Fig. 2. Tyr-tRNAPhe synthesis by PheRS variants (100 nM). �, wild-type�-subunit; Œ, �G318W; ‚, �E334A; E, �H265A; �, �R244A; F, �R244A/�H265A;ƒ, �T354V. Plots represent the average of three independent experiments.

Fig. 1. T. thermophilus PheRS editing site complexed with Tyr; equivalentresidues in E. coli PheRS are shown in parentheses (adapted from ref. 21).

Table 1. Steady-state kinetics of editing for E. coli PheRS editing site replacements

PheRS�-subunit*

ATP consumed†,kobs, min�1

Loss of activity‡,fold

Tyr-tRNAPhe hydrolysis†,kcat/KM, �M�1�min�1

Loss of activity§,fold

Wild type 66 � 1 1.0 55 � 10 1.0Structure-based replacements

T253A 50 � 2 1.3 51 � 10 1.1N254A 47 � 4 1.4 33 � 10 1.6Y255A 45 � 3 1.5 56 � 10 1.0P263A 50 � 5 1.3 73 � 10 0.7P263A�Y360A 65 � 3 1.0 46 � 10 1.2H265A 29 � 2 2.2 17 � 4 3.2F267A 58 � 6 1.1 47 � 1 1.2S322A 45 � 2 1.5 38 � 10 1.4E334A 15 � 4 4.3 12 � 0.8 4.7T354V 20 � 1 3.3 16 � 2 3.4A356W 4.8 � 0.8 14 1.0 � 0.4 53

Alignment-based replacementsR244A 39 � 6 1.7 20 � 3 2.8R244A�H265A 7 � 2 10 2.9 � 0.6 19D251A 64 � 5 1.0 50 � 10 1.1Q262A 57 � 5 1.2 36 � 10 1.5D268A 54 � 4 1.2 55 � 6 1.0G318A 6.5 � 2 10 7.3 � 1 7.5G318W 5.2 � 1 13 0.7 � 0.2 78R359A 47 � 4 1.4 45 � 10 1.2361A 53 � 6 1.2 75 � 9 0.7

*PheRS concentrations used in ATP consumption and Tyr-tRNAPhe hydrolysis assays were normalized according totheir respective phenylalanylation activities.

†Data are the means of at least three independent experiments with standard deviations indicated.‡Loss of cis editing activity relative to wild type.§Loss of trans editing activity relative to wild type.

Ling et al. PNAS � January 2, 2007 � vol. 104 � no. 1 � 73

BIO

CHEM

ISTR

Y

activity, could also tyrosylate tRNAPhe. Although the transediting activities of �R244A, �H265A, and �T354V all droppedby �3-fold, no tyrosylation activity was detected, suggesting thatthey contain sufficient residual hydrolytic activity to preventTyr-tRNAPhe accumulation (Fig. 2). This was supported by theobservation that �R244A/�H265A PheRS, which showed a19-fold reduction in editing efficiency, was able to stably attachTyr to tRNAPhe. Finally, to exclude the possibility that replace-ment of these residues either affects tRNA binding or inducesglobal conformational changes that impair activity, we tested theaffinity of the editing-defective PheRS variants for freetRNAPhe. None of the replacements led to a significant changein the KD for tRNAPhe (SI Table 2), which, in combination withtheir similar phenylalanylation activities (data not shown), indi-cated that the replacements at the editing site did not influencethe global conformation of the enzyme.

Active-Site Residues Do Not Directly Catalyze Editing. To investigatethe roles in the editing reaction of the residues identified above, weconstructed several docking models of the PheRS editing site incomplex with Tyr-A76. The model most consistent with our data isshown in Fig. 3, in which �E334 interacts with the hydroxyl groupof Tyr, �R244 lies near the tRNA backbone, whereas �T354 is closeto the aminoacyl-ester bond. The crystal structure of the PheRS-Tyr complex suggested that �H265 may also be in the vicinity of theester bond (21). In a recent model proposed for the ThrRS editingmechanism, a critical histidine deprotonates a catalytic watermolecule, which then performs a nucleophilic attack on the esterbond (18). To test whether �H265 of PheRS performs a similar rolein catalysis, we attempted to rescue the editing defects of thecorresponding PheRS variants using imidazole. Previous studieshave demonstrated that imidazole is able to partially rescue catalyticdefects resulting from the replacement of essential His residues incertain enzymes (24–27). Imidazole did not restore editing activityin any of the PheRS tested (SI Fig. 9A), suggesting that �H265 doesnot catalyze hydrolysis directly but rather plays a structural role.This finding is consistent with the modest change in editingefficiency seen when �H265 was replaced and in contrast to thedramatic changes previously reported when a catalytic histidine wasreplaced in other enzymatic systems (24–27). The lack of a directrole in catalysis for �H265 is also supported by the similar profilesfor the pH dependence of Tyr-tRNAPhe hydrolysis upon replace-ment with Ala (SI Fig. 9B), indicating that the protonation state of�H265 does not significantly affect editing. These findings, togetherwith the docking model, suggest that, rather than driving catalysis,the imidazole ring of �H265 contributes to substrate binding mostlikely through stacking interactions with Tyr, replacing the roles of�P263 and �Y360 in the postediting complex.

In addition to �H265, structural modeling also identified �T354as a potential catalytic residue. One potential mechanism of catal-ysis would involve nucleophilic attack of the ester bond by the sidechain hydroxyl group of �T354, similar to the catalytic mechanismof many proteases. However, as with �H265A, the �T354V re-placement only moderately diminished both cis and trans editing,which does not support the possible role of this residue as anucleophile. To clarify this point, we replaced �T354 with serineand cysteine and characterized the resulting variants. The thiolgroup of cysteine is a better nucleophile but a poorer hydrogenbonder than the hydroxyl group of serine, which is a naturallyoccurring variant at this position in bacterial PheRSs (SI Fig. 8).Replacement of �T354 with serine resulted in a 2-fold decrease incis editing, whereas the cysteine variant showed a 3.5-fold decrease,similar to valine (Fig. 4). These data clearly demonstrate that �T354does not participate in the hydrolysis reaction as a nucleophile andsuggest that it is involved in either substrate binding or hydrogenbonding with a catalytic water molecule.

Tyr-tRNAPhe Hydrolysis Is Not Rate-Limiting in Posttransfer Editing.The PheRS variants �R244A, �H265A, and �T354V all showed an�3-fold decrease in Tyr-tRNAPhe hydrolysis activity but could nottyrosylate tRNAPhe, indicating that aminoacylation may be muchslower than the hydrolysis step of editing. To address this question,we probed the kinetics of cis and trans editing, as well as tyrosyl-ation. Cis editing kinetic parameters were directly determined withwild-type � PheRS by using the ATP consumption assay, whichrevealed a kcat of 110 min�1 and a KM of 852 �M for Tyr (SI Table3). The KM for Tyr was assigned to the aminoacylation step. Transediting was tested with wild-type � PheRS by using the Tyr-tRNAPhe hydrolysis assay. Because of experimental limitations, wewere not able to saturate the enzyme with the substrate. The upperlimit of Tyr-tRNAPhe used was 5.2 �M, which gave an observedhydrolysis rate of 140 min�1 (SI Fig. 10). By fitting the data to theMichalis–Menten equation, we estimated that the kcat was �270min�1 and the KM for Tyr-tRNAPhe was 4.7 �M. This kcat is�104-fold higher than the uncatalyzed hydrolysis rate. Previouslywe determined that the kcat value of �A294G PheRS was �120min�1 in phenylalanylation (10), which serves as a useful approx-imation for the kcat of tyrosylation given that Tyr and Phe werepreviously shown to have the same kcat in amino acid activation byE. coli PheRS (28). A simplified kinetic scheme for posttransferediting by PheRS is shown in Fig. 5A, which suggests that Tyr-tRNAPhe hydrolysis is not the rate-limiting step; the slowest step iseither tyrosylation or translocation, as previously proposed forIleRS (29). Next we tested whether proton transfer is involved inediting. The catalytic efficiency of Tyr-tRNAPhe hydrolysis wasmeasured in 75% deuterium oxide (D2O) and compared withhydrolysis in H2O. No significant solvent isotope effect was ob-

Fig. 3. Structural modeling of the PheRS editing site complexed withTyr-A76.

Fig. 4. Impact of PheRS �T354 replacements on ATP consumption. �,wild-type �-subunit; E, �T354V; �, �T354C; ‚, �T354S.

74 � www.pnas.org�cgi�doi�10.1073�pnas.0606272104 Ling et al.

served (SI Fig. 11), indicating that proton transfer is not rate-limiting during Tyr-tRNAPhe hydrolysis.

Discrimination of Cognate Phe-tRNAPhe at the Editing Site. Structuraland modeling studies indicated that the side chain of �E334forms a hydrogen bond with the para hydroxyl group of the Tyrmoiety during substrate recognition (Fig. 3). The role of thisresidue in modulating substrate specificity was supported by thefinding that replacing �E334 with Ala induced significant editingof cognate Phe-tRNAPhe while reducing activity against Tyr-tRNAPhe (Fig. 6 and Table 1). Increasing the hydrophobicity ofthe editing site binding pocket by replacing �E334 with Ilefurther increased editing of Phe-tRNAPhe. These findings dem-onstrate that �E334 determines substrate specificity both bydirect recognition of the hydroxyl group of Tyr and by hydro-philic exclusion of Phe. In addition to the �E334 variants,�P263A/�Y360A PheRS also displayed a relaxed specificity

toward Phe-tRNAPhe, indicating that edge-to-face interactionswith the Tyr ring contribute to editing site specificity (Fig. 1).

3� Hydroxyl Group of A76 Is Critical for Editing. Previous studiesindicated that the 3�-OH group is critical for the editing activitiesof IleRS and ValRS (30). IleRS requires transacylation of themischarged amino acid from the 2�-OH to the 3�-OH beforedeacylation, whereas in ValRS it was proposed that the 3�-OHmight be more directly involved in catalyzing hydrolysis (30).Interestingly, studies of the closely related LeuRS revealed that theoriginal 2� linked substrate analog (Nva2AA), but not the 3� linkedanalog (Nva3AA), can inhibit LeuRS editing, suggesting thattransacylation is not required by LeuRS (9). Our studies of thePheRS editing site did not reveal a residue directly involved incatalysis, prompting us to further investigate the role of the neigh-boring hydroxyl group. We modified tRNAPhe at the 2�-OH and3�-OH of A76 by replacing them each with a proton, respectively.The resulting 2�-dA76 tRNAPhe could be charged with neitherphenylalanine nor tyrosine (data not shown), whereas 3�-dA76tRNAPhe was tyrosylated by both wild-type �-subunit and �G318WPheRSs (Fig. 7 A and B). The low tyrosylation efficiency indicatesthat replacing the 3�-OH with H slows down the aminoacylationstep, because the editing-defective �G318W PheRS exhibited thesame charging profile. In contrast, under the same conditions, A76tRNAPhe could be mischarged only by �G318W but not by wildtype, demonstrating that the 3�-OH is crucial for the editing activity.To clarify whether the 3�-OH plays a role in transacylation or incatalysis, we prepared Tyr-3�-dA76 tRNAPhe and tested its hydro-lysis in the presence and absence of wild-type �-subunit PheRS. Asshown in Fig. 7C, PheRS significantly enhances the rate of hydro-lysis, indicating that transacylation is not required for editing.However, replacing the 3�-OH with H resulted in an �300-folddecrease in kcat/KM (0.19 � 0.02 �M�1�min�1) compared with wildtype, indicating that the 3�-OH plays a far more critical role incatalysis than any PheRS residues tested in this study. To furtherprobe the function of the 3�-OH, we substituted it with a fluorineatom. Fluorine is highly electronegative but unable to donate ahydrogen bond. As with 3�-dA76 tRNAPhe, 3�-F-A76 tRNAPhe

could be tyrosylated by both wild-type �-subunit and �G318WPheRSs (Fig. 7A). The hydrolysis of Tyr-3�-F-A76 tRNAPhe was notsignificantly different in the presence or absence of 0.5 �M wild-type �-subunit PheRS (Fig. 7C). These results indicate that the roleof the 3�-OH cannot be substituted by fluorine, suggesting that thisposition acts as a hydrogen bond donor helping with water moleculepositioning or with substrate binding in the editing site. Our resultsalso demonstrated that the 3�-OH plays a role in uncatalyzedhydrolysis. The spontaneous hydrolysis rate of Tyr-3�-dA76tRNAPhe was �5-fold slower than those of Tyr-3�-A76 tRNAPhe

and Tyr-3�-F-A76 tRNAPhe (Fig. 7C) probably because of theinductive effects (�I) of the electronegative hydroxyl and fluoride

Fig. 5. Mechanism of posttransfer editing by PheRS. (A) Kinetic scheme forPheRS editing. (B) Model for Tyr-tRNAPhe hydrolysis at the PheRS editing site(T. thermophilus numbering; see Fig. 3).

Fig. 6. Editing of PheRS variants against cognate Phe. (A) ATP consumption in the presence of Phe (10 mM), tRNAPhe (10 �M), and PheRS (1 �M). (B) Phe-tRNAPhe

hydrolysis by PheRS variants (0.5 �M). (C) Phe-tRNAPhe synthesis by PheRS variants (2 nM). E, no enzyme; �, wild-type �-subunit; ‚, �E334A; Œ, �E334I; �,�P263A/�Y360A.

Ling et al. PNAS � January 2, 2007 � vol. 104 � no. 1 � 75

BIO

CHEM

ISTR

Y

groups, which destabilize the aa-tRNA by withdrawing electronsfrom the ester linkage (22).

DiscussionMechanism of Posttransfer Editing by PheRS. Previous studies of theediting sites of IleRS (31) and LeuRS (32) did not identify catalyticresidues. Instead, the structure of LeuRS complexed with a post-transfer editing analog suggested that the editing site binds thesubstrate and positions a catalytic water molecule for hydrolysis ofthe ester bond (9). A different water-mediated hydrolytic mecha-nism has also been proposed based on structures of ThrRS with andwithout editing analogs (18), although this mechanism has not yetbeen tested experimentally. Functional analyses of the PheRSediting site defined several important residues, none of whichperforms a nucleophilic attack at the ester bond of Tyr-tRNAPhe.Analyses of available PheRS crystal structures showed that divalentspecies such as magnesium or manganese were found only at theinterface between the �- and �-subunits but not at the editing site,excluding a direct role for metal ions in hydrolysis (21, 33–35). Thisfinding is consistent with both ThrRS (18) and alanyl-tRNAsynthetase (36), where divalent ions are unlikely to play a role inposttransfer editing. Our structural modeling revealed two watermolecules (S99 and S112 in Fig. 1) that are correctly positioned ascandidates to catalyze hydrolysis (Fig. 3). These two water mole-cules are well conserved among available PheRS crystal structures,including the apo form and the PheRS:Tyr complex. S112 ispositioned by �T253 and �N254, and S99 is hydrogen-bonded bythe 3�-OH of A76, �T354, and �A356, all of which contribute toediting. The 3�-OH of A76, previously implicated in editing byLeuRS, IleRS, and ValRS (9, 30), was shown to be the most criticalfunctional group for editing by PheRS, whereas changes of �T253,�N254, �T354, and �A356 produced more modest losses in cata-lytic efficiency.

Taken together, our data provide strong support for a substrate-assisted mechanism of posttransfer editing (Fig. 5B). After Tyr-tRNAPhe synthesis at the active site, the CCA-Tyr end translocatesto the B3/B4 domain editing site, as implicated by other studies (7,15, 16). The editing site binds the substrate in such a configurationthat the ester bond is positioned close to two catalytic watermolecules. �R244 interacts with the base of C75; the side chain of�H265 stacks with the Tyr ring; and the carboxyl group of �E334hydrogen bonds with the hydroxyl group of Tyr, which ensuresediting site specificity. The water positioned by �T253 and �N254(S112) performs a nucleophilic attack on the ester bond, whereasthe other water hydrogen-bonded by the 3�-OH, �T354, and �A356(S99) donates a proton to stabilize the leaving group and completethe reaction. The catalytic power is mainly contributed by the3�-OH via activation of the water molecule. Such a mechanism,where the pivotal role of the enzyme is to determine specificityrather than to drive catalysis, may be broadly applicable to post-transfer editing given the generally moderate effects on hydrolysis

seen upon replacing editing site residues (31, 32). Structural andfunctional studies of LeuRS suggest that the 3�-OH might also playa role in catalysis by activating a water molecule (9, 30), which isconsistent with the study of PheRS presented here. In the closelyrelated IleRS, the misacylated amino acid transacylates from the2�-OH to the 3�-OH before being hydrolyzed (30). The posttransferediting then initiates subsequent pretransfer editing (37), butexactly how the 2�-OH participates in catalysis is not clear. On theother hand, in the proposed ThrRS model, two water moleculeshydrogen-bonded by editing site residues were proposed to mediatesubstrate hydrolysis (18), which is also analogous to PheRS, al-though contributions of individual ThrRS editing site residues havenot yet been quantified. Reexamination of the E. coli ThrRSstructure revealed that the 2�-OH of A76 lies close to one catalyticwater, and the recent crystal structure of Pyrococcus abyssi ThrRSrevealed that the 2�-OH indeed interacts with a candidate catalyticwater molecule (38), raising the intriguing possibility that the 2�-OHof tRNA may contribute to ThrRS editing in much the same waythe 3�-OH assists PheRS (39). The high similarity in the editingmechanisms of PheRS and ThrRS, both class II aaRSs, would notseem to extend to class I aaRSs such as IleRS and LeuRS, whichmay partly reflect the general observation that the rate-determiningstep differs between the two enzyme classes (40).

Comparison of Aminoacyl- or Peptidyl-tRNA Esterases. Aside fromaaRSs, D-Tyr-tRNATyr deacylase (DTD) and peptidyl-tRNAhydrolase (PTH) represent other examples of enzymes thatcatalyze the hydrolysis of aminoacyl- or peptidyl-tRNA esterbonds. A recent study showed that the archaeal P. abyssi ThrRSediting domain shares a marked structural similarity to DTD(41), suggesting that the two enzymes might share a commonmechanism for aa-tRNA hydrolysis. The kcat (6 s�1) and rateenhancement (3 � 104) of D-Tyr-tRNATyr hydrolysis by DTDare similar to those for L-Tyr-tRNAPhe hydrolysis by PheRS (42).The hydrolytic mechanism of DTD has not yet been determined,nor has any critical residue been identified. It is possible thatDTD employs a similar mechanism as aaRSs, in which the vicinalhydroxyl group of A76 plays a critical role in hydrolyzing theester bond. This mechanism, however, is less likely to be used byPTH; replacements of critical residues at the PTH active siteresulted in �100-fold decreases in kcat without affecting KM (43),suggesting that these residues play more crucial roles in catalysisthan do editing site residues in aaRSs and revealing a keydifference between aa-tRNA and peptidyl-tRNA hydrolysis.

Role of Editing in Translational Quality Control. The editing site hasapparently evolved to maintain moderate catalytic efficiencysufficient for the hydrolysis of noncognate species while mini-mizing the hydrolysis of cognate aa-tRNA. Our proposed sub-strate-assisted mechanism of editing, wherein the enzyme simplyaccelerates the rate of spontaneous hydrolysis of aa-tRNA by

Fig. 7. The role of the 3�-OH of A76 in editing. (A and B) Tyrosylation of A76 tRNAPhe (squares), 3�-dA76 tRNAPhe (triangles), and 3�-F-A76 tRNAPhe (circles) by1 �M wild-type PheRS (A) or 1 �M �G318W PheRS (B). (C) Deacylation of 0.2–0.4 �M tyrosylated A76 tRNAPhe, 3�-dA76 tRNAPhe, and 3�-F-A76 tRNAPhe in thepresence (filled symbols) or absence (open symbols) of 0.5 �M wild-type �-subunit PheRS.

76 � www.pnas.org�cgi�doi�10.1073�pnas.0606272104 Ling et al.

�104, also suggests that selectively retaining misaminoacylatedspecies is in fact the critical step in aaRS proofreading. The KMof Tyr-tRNAPhe for wild-type �-subunit PheRS is within themicromolar range; however, elongation factor Tu (EF-Tu) bindsaa-tRNAs several orders of magnitude more tightly (44–46).Previously we showed that overexpressing �H265A or �E334APheRS resulted in Tyr misincorporation at Phe codons in vivo(10), suggesting that EF-Tu promptly sequesters the mischargedTyr-tRNAPhe upon its synthesis and that there is no efficientproofreading mechanism once Tyr-tRNAPhe leaves PheRS. Re-cently, it has been found that low levels of mischarged tRNAs canlead to protein misfolding and neurodegeneration in mice (47).Our present findings show that a loss of quality control followsa relatively minor loss in editing activity, emphasizing the finebalance that exists between a tolerable error rate and a level ofmisaminoacylation that critically impacts translational fidelity.

Materials and MethodsStrains, Plasmids, Site-Directed Mutagenesis, and General Methods.Proteins and tRNAs were prepared as described previously (23). AllPheRS variants described in this work contain an A294G replace-ment in the �-subunit. 2�- and 3�-dA76 tRNAPhe and 3�-F-A76tRNAPhe were prepared as described (10), and 3�-fluro ATP waspurchased from IBA (Gottingen, Germany). 32P-labeled tRNApA76 transcripts were synthesized as described previously (23) andused to determine the KD for PheRS by using filter-binding assays(48). Structural modeling was performed with Autodock 3.0 (49).

Aminoacylation Assays. Aminoacylation was performed as de-scribed (10) with the addition of 2 mM ATP, 20 �M [14C] Phe (273cpm/pmol), 6 mg/ml total tRNA (from E. coli MRE 600; Roche),and 2 nM PheRS. Tyrosylation was performed with 2 units/ml yeastpyrophosphatase (Roche), 2 mM ATP, 10 �M E. coli tRNAPhe

transcript, 50 �M [3H]Tyr (464 cpm/pmol), and 0.1 �M PheRS.Yeast pyrophosphatase, added to mimic conditions in the ATPconsumption assay, had no effect on tyrosylation by �G318WPheRS (data not shown).

ATP Consumption Assay. The ATP consumption assay was used tomeasure cis editing by PheRS. A 15-�l reaction mix contained 2units/ml yeast pyrophosphatase (Roche), 2 mM Tyr or 10 mMPhe, 10 �M tRNAPhe, 2 mM [�-32P]ATP (5 cpm/pmol), 0.1 M

Na-Hepes (pH 7.2), 30 mM KCl, 10 mM MgCl2, and 1 �MPheRS. Steady-state kinetics were determined by using 20–3,000�M Tyr or 1–32 �M tRNAPhe. In imidazole rescue experiments,0.1 M imidazole (pH 7.2) was added into the above reaction mix.

Preparation of Tyr-tRNAPhe and Phe-tRNAPhe. Tyr- and Phe-tRNAPhe

were prepared as described (10) except that 2 �M �G318W PheRSwas used. After a 10-min incubation at 37°C, the reaction wasstopped by addition of 56 mM potassium acetate (pH 4.5) and 250mM KCl, followed by phenol/chloroform extraction and ethanolprecipitation. The aa-tRNA pellet was dried and resuspended inDEPC water with 2 mM MgCl2. The charging level was �25% asdetermined by radioactivity retained on 3-mm filter discs.

Tyr-tRNAPhe and Phe-tRNAPhe Hydrolysis Assays. Tyr-tRNAPhe hydro-lysis was performed in 0.1 M Na-Hepes (pH 7.2), 30 mM KCl, and10 mM MgCl2. The hydrolysis rate–substrate concentration profilewas determined with 2 nM wild-type � PheRS and 0.7–5.2 �MTyr-tRNAPhe. kcat/KM values of PheRS variants for Tyr-tRNAPhe

hydrolysis were determined with 0.6–0.9 �M substrate and 2–100nM enzyme. Reaction mixtures were incubated at 37°C, and 2- to9-�l aliquots were periodically spotted on 3-mm filter discs pre-soaked with 5% TCA, followed by extensive washing in 5% TCA,drying, and scintillation counting. Tyr-tRNAPhe hydrolysis was alsomonitored in the absence of PheRS at each substrate concentrationas a control, and these rates subtracted from the enzyme catalyzedvalues to determine initial velocities. To determine the pH depen-dence of Tyr-tRNAPhe hydrolysis, 1–250 nM wild-type � PheRS and5–500 nM �H265A PheRS were used. pH values determined werethose of the final reaction mix before Tyr-tRNAPhe addition.Investigation of solvent isotope effects was performed in thepresence of 75% deuterium oxide (D2O) and 2–10 nM wild-type �PheRS. The enzyme was preequilibrated in a buffer containing75% D2O for 2 h on ice. Tyr-3�-dA76 tRNAPhe and Phe-tRNAPhe

hydrolysis was performed in the presence of 0.4 �M substrate and0.5 �M PheRS variants.

We thank D. Tirrell (California Institute of Technology, Pasadena, CA)and O. Uhlenbeck (Northwestern University, Evanston, IL) for strainsand plasmids and S. Ataide, C. Hausmann, J. Levengood, N. Reynolds,and T. Rogers for critical reading of the manuscript. This work wassupported by National Science Foundation Grant MCB-0344002.

1. Baldwin AN, Berg P (1966) J Biol Chem 241:839–845.2. Fersht AR (1979) in Transfer RNA: Structure, Properties and Recognition, eds Schimmel PR,

Soll D, Abelson JN (Cold Spring Harbor Lab Press, Cold Spring Harbor, NY), pp 247–254.3. Hale SP, Auld DS, Schmidt E, Schimmel P (1997) Science 276:1250–1252.4. Eldred EW, Schimmel PR (1972) J Biol Chem 247:2961–2964.5. Yarus M (1972) Proc Natl Acad Sci USA 69:1915–1919.6. Beuning PJ, Musier-Forsyth K (2000) Proc Natl Acad Sci USA 97:8916–8920.7. Dock-Bregeon A, Sankaranarayanan R, Romby P, Caillet J, Springer M, Rees B, Francklyn

CS, Ehresmann C, Moras D (2000) Cell 103:877–884.8. Beebe K, Ribas De Pouplana L, Schimmel P (2003) EMBO J 22:668–675.9. Lincecum TL, Tukalo M, Yaremchuk A, Mursinna RS, Williams AM, Sproat BS, Van Den

Eynde W, Link A, Van Calenbergh S, Grøtli M, et al. (2003) Mol Cell 11:951–963.10. Roy H, Ling J, Irnov M, Ibba M (2004) EMBO J 23:4639–4648.11. Webster T, Tsai H, Kula M, Mackie GA, Schimmel P (1984) Science 226:1315–1317.12. Eriani G, Delarue M, Poch O, Gangloff J, Moras D (1990) Nature 347:203–206.13. Cusack S, Berthet-Colominas C, Hartlein M, Nassar N, Leberman R (1990) Nature

347:249–255.14. Nureki O, Vassylyev DG, Tateno M, Shimada A, Nakama T, Fukai S, Konno M,

Hendrickson TL, Schimmel P, Yokoyama S (1998) Science 280:578–582.15. Silvian LF, Wang J, Steitz TA (1999) Science 285:1074–1077.16. Fukai S, Nureki O, Sekine S, Shimada A, Tao J, Vassylyev DG, Yokoyama S (2000) Cell

103:793–803.17. Sankaranarayanan R, Dock-Bregeon AC, Romby P, Caillet J, Springer M, Rees B,

Ehresmann C, Ehresmann B, Moras D (1999) Cell 97:371–381.18. Dock-Bregeon AC, Rees B, Torres-Larios A, Bey G, Caillet J, Moras D (2004) Mol Cell

16:375–386.19. Beuning PJ, Musier-Forsyth K (2001) J Biol Chem 276:30779–30785.20. Sasaki HM, Sekine S, Sengoku T, Fukunaga R, Hattori M, Utsunomiya Y, Kuroishi C,

Kuramitsu S, Shirouzu M, Yokoyama S (2006) Proc Natl Acad Sci USA 103:14744–14749.21. Kotik-Kogan O, Moor N, Tworowski D, Safro M (2005) Structure (London) 13:1799–1807.22. Weinger JS, Strobel SA (2006) Biochemistry 45:5939–5948.23. Roy H, Ling J, Alfonzo J, Ibba M (2005) J Biol Chem 280:38186–38192.24. Newmyer SL, de Montellano PR (1996) J Biol Chem 271:14891–14896.

25. Huang S, Tu SC (1997) Biochemistry 36:14609–14615.26. Venkataraman P, Lamb RA, Pinto LH (2005) J Biol Chem 280:21463–21472.27. McCartney SA, Brignole EJ, Kolegraff KN, Loveland AN, Ussin LM, Gibson W (2005) J Biol

Chem 280:33206–33212.28. Ibba M, Kast P, Hennecke H (1994) Biochemistry 33:7107–7112.29. Nomanbhoy TK, Hendrickson TL, Schimmel P (1999) Mol Cell 4:519–528.30. Nordin BE, Schimmel P (2002) J Biol Chem 277:20510–20517.31. Hendrickson TL, Nomanbhoy TK, De Crecy-Lagard V, Fukai S, Nureki O, Yokoyama S,

Schimmel P (2002) Mol Cell 9:353–362.32. Mursinna RS, Lincecum TL, Jr, Martinis SA (2001) Biochemistry 40:5376–5381.33. Mosyak L, Reshetnikova L, Goldgur Y, Delarue M, Safro MG (1995) Nat Struct Biol 2:537–547.34. Reshetnikova L, Moor N, Lavrik O, Vassylyev DG (1999) J Mol Biol 287:555–568.35. Fishman R, Ankilova V, Moor N, Safro M (2001) Acta Crystallogr D Biol Crystallogr

57:1534–1544.36. Sokabe M, Okada A, Yao M, Nakashima T, Tanaka I (2005) Proc Natl Acad Sci USA

102:11669–11674.37. Nordin BE, Schimmel P (2003) Biochemistry 42:12989–12997.38. Hussain T, Kruparani SP, Pal B, Dock-Bregeon AC, Dwivedi S, Shekar MR, Sureshbabu

K, Sankaranarayanan R (2006) EMBO J 25:4152–4162.39. Igloi GL, Cramer F (1978) FEBS Lett 90:97–102.40. Zhang CM, Perona JJ, Ryu K, Francklyn C, Hou YM (2006) J Mol Biol 361:300–311.41. Dwivedi S, Kruparani SP, Sankaranarayanan R (2005) Nat Struct Mol Biol 12:556–557.42. Soutourina J, Plateau P, Delort F, Peirotes A, Blanquet S (1999) J Biol Chem 274:19109–19114.43. Schmitt E, Mechulam Y, Fromant M, Plateau P, Blanquet S (1997) EMBO J 16:4760–4769.44. LaRiviere FJ, Wolfson AD, Uhlenbeck OC (2001) Science 294:165–168.45. Asahara H, Uhlenbeck OC (2002) Proc Natl Acad Sci USA 99:3499–3504.46. Asahara H, Uhlenbeck OC (2005) Biochemistry 44:11254–11261.47. Lee JW, Beebe K, Nangle LA, Jang J, Longo-Guess CM, Cook SA, Davisson MT, Sundberg

JP, Schimmel P, Ackerman SL (2006) Nature 443:50–55.48. Roy H, Ibba M (2006) Biochemistry 45:9156–9162.49. Morris GM, Goodsell DS, Halliday RS, Hart WE, Belew RK, Olson AJ (1998) J Comput

Chem 19:1639–1662.

Ling et al. PNAS � January 2, 2007 � vol. 104 � no. 1 � 77

BIO

CHEM

ISTR

Y