Homologous Trans-Editing Factors with Broad Trna Specificity Prevent Mistranslation Caused by Serine/Threonine Misactivation

Homologous trans-editing factors with broad tRNA specificity prevent mistranslation caused by serine/threonine misactivation

Ziwei Liua,b,1, Oscar Vargas-Rodrigueza,b,1, Yuki Gotoc, Eva Maria Novoad,2, Lluís Ribas de Pouplanad,e, Hiroaki Sugac, and Karin Musier-Forsytha,b,3

aDepartment of Chemistry and Biochemistry and bCenter for RNA Biology, The Ohio State University, Columbus, OH 43210; cDepartment of Chemistry, Graduate School of Science, The University of Tokyo, Tokyo 113-0033, Japan; dInstitute for Research in Biomedicine, 08028 Barcelona, Catalonia, Spain; and eCatalan Institution for Research and Advanced Studies, 08010 Barcelona, Catalonia, Spain

Edited by Paul Schimmel, The Skaggs Institute for Chemical Biology, La Jolla, CA, and approved April 6, 2015 (received for review December 10, 2014) Aminoacyl-tRNA synthetases (ARSs) establish the rules of the genetic diverse and include the N-terminal domain of threonyl-tRNA syn- code, whereby each amino acid is attached to a cognate tRNA. thetase (ThrRS) (10), the related editing domain of alanyl-tRNA Errors in this process lead to mistranslation, which can be toxic synthetase (AlaRS) (11), the β3/ β4 domain of phenylalanine-tRNA to cells. The selective forces exerted by species-specific require- synthetase (12), and the insertion domain (INS) of prolyl- ments and environmental conditions potentially shape quality- tRNA synthetase (ProRS) (13, 14). Mutations in editing domains control mechanisms that serve to prevent mistranslation. A family can have detrimental effects on cells. For example, a mutation in of editing factors that are homologous to the editing domain of the editing site of AlaRS, which results in only a twofold increase bacterial prolyl-tRNA synthetase includes the previously characterized in misacylation in vitro, results in a severe neurodegeneration trans-editing factors ProXp-ala and YbaK, which clear Ala-tRNAPro and phenotype in mice (15). Cellular degradation and apoptosis Cys-tRNAPro, respectively, and three additional homologs of unknown caused by a mutation in the editing domain of ValRS have been function, ProXp-x, ProXp-y, and ProXp-z. We performed an in vivo reported in murine cells (16). In addition, editing defects in bac- screen of 230 conditions in which an Escherichia coli proXp-y de- teria often result in slower growth rates, delayed growth, or even letion strain was grown in the presence of elevated levels of amino death (17–22). acids and specific ARSs. This screen, together with the results of in In addition to the cis-editing domain appended to ARSs, free- vitro deacylation assays, revealed Ser- and Thr-tRNA deacylase func- standing homologs of editing domains are distributed throughout tion for this homolog. A similar activity was demonstrated for organisms in all three kingdoms of life as additional checkpoints to Bordetella parapertussis ProXp-z in vitro. These proteins, now maintain translational fidelity in trans. Autonomous trans-editing renamed “ProXp-ST1” and “ProXp-ST2,” respectively, recognize factors that are evolutionarily related to three class II ARSs, multiple tRNAs as substrates. Taken together, our data suggest namely, AlaRS, ThrRS and ProRS, have been identified. A ho- that these free-standing editing domains have the ability to pre- molog of the AlaRS editing domain, AlaXp, is widely distributed vent mistranslation errors caused by a number of ARSs, including and is shown to hydrolyze seryl-tRNA synthetase (Ser-tRNAAla) lysyl-tRNA synthetase, threonyl-tRNA synthetase, seryl-tRNA syn- (23, 24). ThrRS-ed is an autonomous editing domain of Ser- thetase, and alanyl-tRNA synthetase. The expression of these tRNAThr in crenarchaeal genomes where thrS genes are truncated multifunctional enzymes is likely to provide a selective growth (25). Bacterial ProRS INS domain homologs include five proteins advantage to organisms subjected to environmental stresses and other conditions that alter the amino acid pool. Significance aminoacyl tRNA synthetase | protein synthesis | ProXp | quality control | trans-editing A key step in protein synthesis is catalyzed by aminoacyl-tRNA synthetases (ARSs), which attachspecificaminoacidstotRNAs. Errors caused by ARS amino acid misactivation require proofread- high level of accuracy in protein synthesis is essential for ing activities encoded in editing domains. Approximately half of normal cell function and proliferation. A critical step in this A the ARSs possess intrinsic editing functions, and single-domain process is pairing the correct amino acid with the cognate tRNA editingproteinsalsoareencodedinsomeorganisms.Thepresent species by aminoacyl-tRNA synthetases (ARSs). ARSs catalyze study used an in vivo screen coupled with in vitro assays to eluci- the aminoacyl-tRNA (aa-tRNA) formation in two steps involving date the substrate specificity of two free-standing editing domains, amino acid activation (step 1) and transfer of the activated revealing that they are multifunctional Ser/Thr-tRNA deacylases amino acid to tRNAs (step 2). Although ARSs have evolved to with the ability to prevent errors caused by a number of ARSs. exhibit specific tRNA-recognition capabilities with an estimated − Trans-editing proteins with broad substrate specificities provide a error frequency of 10 6, a greater number of mistakes arise from growth advantage to microbes under conditions in which amino the lack of discrimination of near-cognate amino acids, with an − − acid pools are altered and may represent novel targets. estimated error rate of 10 4 to 10 5 at this step (1). Misincorpora-

tion of amino acids into proteins can be harmful to both eukaryotic Author contributions: Z.L., O.V.-R., L.R.d.P., and K.M.-F. designed research; Z.L., O.V.-R., and prokaryotic cells (2, 3). and E.M.N. performed research; Y.G. and H.S. contributed new reagents/analytic tools; Quality control of aa-tRNA formation is achieved by hydrolysis Z.L. and O.V.-R. analyzed data; and Z.L., O.V.-R., E.M.N., and K.M.-F. wrote the paper. of the aminoacyl-adenylate (“pre-transfer editing”)ordeacylation The authors declare no conflict of interest. of the mischarged aa-tRNA (“post-transfer editing”)(4,5).Edit- This article is a PNAS Direct Submission. ing mechanisms are used by both classes of ARSs, and 7 of 22 1Z.L. and O.V.-R. contributed equally to this work. ARSs possess posttransfer editing sites that are distinct from the 2Present address: Computer Science and Artificial Intelligence Laboratory, Massachusetts aminoacylation active site. The connective polypeptide 1 editing Institute of Technology, Cambridge, MA 02139. domain is found in class I isoleucyl-tRNA synthetase (IleRS) (6), 3To whom correspondence should be addressed. Email: [email protected].

leucyl-tRNA synthetase (LeuRS) (7, 8), and valyl-tRNA synthe- This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10. BIOCHEMISTRY tase (ValRS) (9). Editing domains of class II ARSs are more 1073/pnas.1423664112/-/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.1423664112 PNAS | May 12, 2015 | vol. 112 | no. 19 | 6027–6032 Downloaded by guest on September 26, 2021 previously named “YbaK,”“ProXp-ala,”“ProXp-x,”“ProXp-y” (annotated YeaK), and “ProXp-z” (annotated PA2301) (Fig. 1) (26). These domains together with the INS domain of ProRS are collectively known as the “INS superfamily.” Although the INS domain edits mischarged Ala-tRNAPro (13, 14), YbaK deacylates Cys-tRNAPro (27), which is formed in the active site of ProRS because of the similar size of Cys and Pro (28). ProXp-ala is capable of clearing Ala-tRNAPro and compensates for the lack of an INS posttransfer editing domain in many bacteria (23, 26). Functions for ProXp-x, -y, and -z have not yet been reported. Here, for the first time to our knowledge, we explore the in vivo and in vitro substrate specificities of ProXp-y and ProXp-z. These two domains are phylogenetically distinct (26) and differ in length and in the identity of several highly conserved residues. Most Fig. 2. Comparison of growth curves of E. coli WT (closed circles) and notably, proXp-y encodes a functionally critical Lys residue that is proXp-y–null strains (open circles). E. coli WT BW25113 and proXp-y–null present in all other INS superfamily members with the exception strains (Keio collection) were grown in M9 liquid minimal medium contain- of proXp-z, which instead contains a strictly conserved Asn (26). ing 10 mM Ser, AlaRS overexpressed (A); 10 mM Ser, LysS overexpressed (B); 10 mM Ser, LysU overexpressed (C); and 10 mM Thr, LysU overexpressed (D). Results Effect of Escherichia coli proXp-y Deletion on Cell Growth in the To confirm that intracellular amino acid concentrations were Presence of Elevated Amino Acid Levels. We first tested the in elevated when excess Ser and Thr were added to the extracel- vitro editing activities of E. coli ProXp-y using Pro-, Ala-, and SI Experimental Pro lular medium, we used an LC-MS/MS approach ( Cys-tRNA . However, no deacylation activity was detected for Procedures). The intracellular Ser concentration was elevated 3- these substrates (Fig. S1).We next took advantage of the avail- to 10-fold, whereas the Thr concentration was increased ∼1.5- E coli proXp-y– ability of an . null strain (Keio collection) to fold when 10 mM of each amino acid was added to the medium screen for conditions under which growth defects in the null (Fig. S3). To confirm that the growth defects observed in Fig. 2 strain could be observed. We hypothesized that increased ex- were caused by the lack of ProXp-y function, we performed a pression of certain error-prone synthetases together with ele- complementation assay. As shown in Fig. S2 E and F, the growth vated levels of a near-cognate amino acid may amplify any difference was abolished when both WT and null strains were growth defect caused by mistranslation in the null strain. The grown in the presence of elevated amino acids and overexpressed proXp-y– WT and null strains were transformed with over- ARS and ProXp-y. Taken together, these results indicate that E coli expression plasmids encoding each of the . class II ARS growth defects were observed for the proXp-y–null strain in the E coli genes. The . genome encodes two class II lysyl-tRNA presence of an altered intracellular amino acid pool and elevated lysS synthetase (LysRS) genes, constitutive and stress-inducible ARS levels and suggest that ProXp-y is likely to function in lysU . The two corresponding enzymes share 88% identity but aa-tRNA quality control. Interestingly, addition of homoserine possess different affinities for Lys and different thermostabilities (Hse) to the growth medium, which was tested because of its (29). The mischarging activities of these two LysRS isoforms structural similarity to both Ser and Thr, also resulted in a have not been investigated extensively, and both were included in phenotype when LysS, LysU, and seryl-tRNA synthetase (SerRS) our screen. Among the 231 growth conditions screened, growth were overexpressed (Table S1). In all cases, the main differ- defects were observed in the null strain in the presence of Ser ence in growth observed in the WT and null strain was a dif- when AlaRS and LysU were overexpressed and with Thr when ference in the lag phase before exponential growth. This effect LysS and LysU were overexpressed (Fig. 2 and Table S1). Upon is similar to growth defects observed with editing-deficient syn- the addition of 10 mM Ser alone, a much smaller growth defect proXp-y– A thetases (22, 30). was observed in the null strain (Fig. S2 ), whereas AlaRS is known to misactivate Ser (31), and LysRS has been without the amino acid addition, overexpression of AlaRS, LysS, reported to misactivate Ser, Thr, and Hse (32). Although E. coli and LysU delayed growth in the WT and proXp-y–null strains to B–D SerRS has been reported not to misactivate Hse (33), we ob- the same extent (Fig. S2 ). served weak mischarging of tRNASer in the presence of 5 mM Hse (Fig. S4). The differences observed between the WT and proXp-y–null strains in Fig. 2 and Table S1 suggest roles for ProXp-y in editing tRNAs mischarged by LysRS, AlaRS, SerRS, and possibly ThrRS, which misactivates Ser (10). LysU overexpression appeared to result in a much larger growth defect than LysS overexpression. To explore these differences further, a dose-dependent analysis was carried out by varying the concentrations of added amino acids (Fig. S5 A–G). As summarized in Fig. 3A for the 5-mM concentrations of Ser, Thr, and Hse, the proXp-y–null strain expressing the stress-inducible lysU gene Fig. 1. Domain structure of E. coli and B. parapertussis ProRS and the INS showed a dramatic increase in growth time to half-maximal superfamily. Conserved motifs 1, 2, and 3 (M1–M3, purple), the anticodon- OD600 compared with the WT strain. The largest effect was binding domain (ABD, green), and the editing domain (INS domain, blue) observed for Hse, where no growth was observed with LysU in are shown. Single-domain INS-like proteins YbaK, ProXp-ala, ProXp-x, the null strain. SerRS overexpression in the proXp-y–null strain ProXp-y, and ProXp-z encoded in the indicated species are shown together showed a similar growth defect in the presence of Hse. Taken with the INS and a truncated mini-INS present in the corresponding ProRS. together, these results indicate that LysU may be more error C. crescentus is Caulobacter crescentus,andR. palustris is Rhodopseudomonas prone than LysS, and ProXp-y may be required under physio- palustris. The known activities of INS, YbaK, and ProXp-ala are color coded lysU as follows: blue, Ala deacylation; red, Cys deacylation. The domains in- logical conditions where the gene is induced. To test this vestigated in this work are in orange, and domains of unknown function hypothesis, we grew cells in 1 mM Leu in the absence of ARS are in gray. overexpression, because increased Leu is known to induce LysU

6028 | www.pnas.org/cgi/doi/10.1073/pnas.1423664112 Liu et al. Downloaded by guest on September 26, 2021 tRNA is not inhibitory and does not impact the deacylation activity of ProXp-y or -z.

Deacylation Activity of ProXp-y and ProXp-z Toward Cognate Thr- tRNAThr. Based on the robust deacylation of various Thr-tRNAs by ProXp-y and ProXp-z, we wondered whether cognate Thr- Fig. 3. Effect of amino acids and antibiotics on the growth of E. coli WT and tRNAThr also was hydrolyzed by these two enzymes. Indeed, Thr- proXp-y–null strains. (A) Effect of Ser, Thr, and Hse on the growth of WT tRNAThr is effectively deacylated by ProXp-y and ProXp-z (Fig. – (black bars) and proXp-y null (gray bars) strains with LysS, LysU, or SerRS S6 C and D). However, the presence of even a substoichiometric overexpression. E. coli WT and proXp-y–null strains were transformed with amount of ThrRS protects Thr-tRNAThr from hydrolysis, whereas overexpression plasmids encoding lysS, lysU,orserS genes and were cultured Thr at 37 °C overnight. Cells then were washed and inoculated into M9 minimal excess ProRS does not. This result suggests that Thr-tRNA is medium containing no amino acids (No) or 5 mM Ser, Thr, or Hse. Growth not deacylated by these two free-standing editing domains in a curves were monitored using the Omnilog system. The bars indicate the physiological context.

mean time to half-maximal OD600 of three independent growth curves. N.G., no growth of cells during the time (48 h) monitored. (B) Growth assays of WT Deacylation Activity of ProXp-y and ProXp-z Toward aa-tRNASer. In (black bars) and proXp-y–null (gray bars) strains in the presence of kana- contrast to the deacylation of Thr-tRNAThr by ProXp-y and mycin (Kan), neomycin (Neo), or streptomycin (Strep). The y axis indicates the ProXp-z, Ser-tRNASer is not cleared efficiently by ProXp-z, al- area of the clearing zone, which was calculated by measuring the diameter though it is a moderate substrate for ProXp-y (Fig. 5). Compared of the circle in which no growth was observed. Error bars indicate the SD Lys Ser based on three independent experiments. with the deacylation of Ser-tRNA , Ser-tRNA is deacylated with a threefold and 50-fold reduced rate by ProXp-y and ProXp-z, respectively. In both cases, despite the cell-based re- expression (34). We observe that the proXp-y–null strain displays sults showing a dramatic growth defect with added Hse (Fig. 3A), Ser a reduced growth phenotype even though ProXp-y does not edit very similar deacylation rates were observed for Hse-tRNA Ser Ser Leu-tRNAs (Figs. S1 and S5H). This result is consistent with the relative to Ser-tRNA . In addition, Thr-tRNA is a good requirement for ProXp-y under LysU overexpression conditions. substrate for ProXp-y, but not for ProXp-z, the only major dif- If ProXp-y functions as a quality-control factor in vivo, we hy- ference in specificity observed for these two homologs. pothesize that the editing-defective null strain would be hyper- mutagenic and therefore would develop antibiotic resistance more Distribution of AlaXp, ProXp-y, ProXp-z, and LysU. The editing domain of AlaRS and the trans-editing domain AlaXp display rapidly than the WT strain. To test this idea, halo assays were similar capabilities for deacylating Ser-tRNAAla (23). Because performed in the presence of three aminoglycoside antibiotics ProXp-y and -z also deacylate Ser-tRNAAla, we investigated the (neomycin, streptomycin, and kanamycin). Indeed, compared with distribution of the three trans-editing proteins. Of 1,597 genomes the WT strain, the proXp-y–null strain showed a smaller clearing B examined representing all kingdoms of life, 44% (37% when only zone (Fig. 3 ). The reduced antibiotic sensitivity of the null strain bacteria are considered) encode at least one of these trans- is consistent with an editing function for ProXp-y and with the editing domains. Because ProXp-y and -z are not encoded in notion that when cells are stressed by ribosome ambiguity (i.e., archaea and are present in only a very few eukaryotes, only trans antibiotics), the regulated expression of editing domains bacteria were considered further. No bacteria encode all three provides adaptive variation to sustain growth (17). trans-editing domains, and there is very little overlap between E coli Bordetella parapertussis ProXp-y and ProXp-z (10 species encode both) or between In Vitro Deacylation by . ProXp-y and ProXp-z and AlaXp (11 species encode both). However, 51 ProXp-z. Based on the in vivo results using the proXp-y–null E coli species encode both AlaXp and ProXp-y (Fig. S7). An analysis of strain, we next examined the editing capability of both . all bacterial genomes reveals 10 species (two Clostridium species ProXp-y and the closely related ProXp-z in vitro. Because the and eight species of Gammaproteobacteria) that encode both later domain is not encoded in E. coli, we cloned and purified the homolog from B. parapertussis for these studies. For both proteins, moderate levels of deacylation were observed for mischarged Ser-tRNALys, Ser-tRNAAla,andHse-tRNALys, and more robust activities were measured for Thr-tRNALys (Fig. 4 A and B). No deacylation of cognate Lys-tRNALys and Ala-tRNAAla was observed. In addition to the formation of mischarged Ser-tRNAAla, Ser-tRNALys, and Thr-tRNALys by AlaRS and LysRS, Thr and Ser are known to be misactivated by ValRS, IleRS, and ThrRS to form Thr-tRNAVal, Thr-tRNAIle, and Ser-tRNAThr, respectively. Thus, these substrates were prepared to test whether the activities of ProXp-y and ProXp-z extend beyond tRNAAla and tRNALys. Deacylation assays showed that ProXp-y and ProXp-z also edit Thr-tRNAVal,Thr-tRNAIle, and Ser-tRNAThr, whereas cognate Val-tRNAVal and Ile-tRNAIle species are not deacylated (Fig. 4 C and D). These results suggest the role of ProXp-y and ProXp-z as general deacylases with dual specificities for mischarged Ser/Thr- tRNAs. To ensure that the various levels of deacylation observed for the different substrates tested in Fig. 4 are caused by different Fig. 4. Deacylation of 0.75 μM aa-tRNA by 0.5 μM E. coli ProXp-y or amounts of uncharged tRNA in the aa-tRNA preparations, we Lys, Ala Lys Lys Lys B. parapertussis ProXp-z. Ser-tRNA , Thr-tRNA , and Hse-tRNA performed deacylation assays for both Ser-tRNA and Thr- Thr Val, Ile Lys (A and B) and Ser-tRNA and Thr-tRNA (C and D) were deacylated by tRNA using ProXp-x and ProXp-y with various ratios of ProXp-y (A and C) or ProXp-z (B and D). Cognate aa-tRNALys, Ala, Val, Ile were

uncharged tRNA added to the reaction (Fig. S6 A and B). The tested also. All assays were carried out in triplicate as described in Experi- BIOCHEMISTRY results clearly show that the presence of high amounts of uncharged mental Methods; error bars indicate SD.

Liu et al. PNAS | May 12, 2015 | vol. 112 | no. 19 | 6029 Downloaded by guest on September 26, 2021 may have been retained for a function other than Ser-tRNAAla deacylation. The ability to differentiate Ser and Thr from cognate substrates appears to be particularly challenging for ARSs, because at least 6 of the 20 synthetases misactivate these amino acids. ThrRS readily misactivates Ser, and in most species ThrRS encodes an N-terminal posttransfer editing domain to clear Ser- tRNAThr (10). In several crenarchaeal species, the editing domain ThrRS-ed is encoded as a separate trans-editing factor (25). Interestingly, in E. coli the ThrRS editing function is severely impaired under oxidative stress conditions, and this phenotype is Fig. 5. Deacylation of 0.75 μM Ser-, Thr-, and Hse-tRNASer by 0.5 μM E. coli attributed to an oxidized Cys residue in the editing domain active ProXp-y or B. parapertussis ProXp-z. All assays were carried out in triplicate site (22). We hypothesize that trans-editing factors may be re- as described in Experimental Methods; error bars indicate SD. quired to clear Ser-tRNAThr when ThrRS editing function is lost. All trans-editing factors identified to date are related to class lysS and lysU. Although the Clostridium species do not encode II synthetase editing domains (AlaRS, ThrRS, and ProRS) and proXp-y or proXp-z, 100% of the lysU-containing genomes in are believed to assist in editing of tRNAs mischarged by these Gammaproteobacteria also encode proXp-y. same synthetases. However, we show here that ProXp-type editing domains are able to deacylate a wide variety of tRNAs, including Discussion those potentially mischarged by class I IleRS and ValRS. Although Ile Val As a key step in maintaining high fidelity in protein synthesis, Thr-tRNA and -tRNA were robustly deacylated by both catalysis of aa-tRNA formation by ARSs is balanced between ProXp-ST1 and -ST2, both these synthetases possess editing do- efficiency and specificity. Although the active sites of ARSs re- mains that clear these mischarged tRNAs under normal conditions. ject most noncognate amino acids in the “primary sieve,” their Although it is possible that the ProXp-type domains perform a ability to differentiate proteinogenic and nongenetically encoded redundant editing function under certain conditions, differences amino acids from cognate substrates is challenged by smaller and in aminoacylation kinetics between class I and class II ARSs trans structurally related amino acids. Thus, “double-sieve” and “tri- suggest that editing may not be required for tRNAs charged ple-sieve” editing mechanisms have evolved in all three kingdoms of by class I synthetases (40). Product release is rapid in class II life and become essential under certain growth conditions, including ARSs, unlike class I enzymes in which this step is rate limiting environmental stress (22, 35). (41). Thus, class I ARSs remain bound to their substrates long Previously, two trans-editing domain homologs of the bacterial enough to edit any misacylated tRNA before release and normally ProRS INS domain, YbaK and ProXp-ala, were shown to would not require proofreading by a trans-editing factor (42). In deacylate known ProRS editing errors, with specificity for Cys- the assay demonstrating ThrRS protection, we also expect that class and Ala-tRNAPro, respectively. YbaK gains tRNAPro specificity II ThrRS releases the substrate rapidly; however, ThrRS competes through interaction with ProRS (36), whereas ProXp-ala recognizes effectively with ProXp-ST1 and -ST2 in rebinding the cognate tRNAPro through interaction with the same acceptor stem identity Thr-tRNAThr substrate and therefore protects it from hydrolysis. elements as bacterial ProRS (26, 37). Here, we show that two other LysRS is a class II synthetase with no known posttransfer INS domain homologs, ProXp-y and ProXp-z, have specificity for editing capability, despite reported mischarging of a variety of Ser- and Thr-tRNAs and therefore are unlikely to clear ProRS protein and nonprotein amino acids in vitro (32). Based on the mischarging errors. Instead, based on in vitro studies, these en- results reported here, we propose that mischarging errors caused zymes are capable of clearing mischarging errors caused by LysRS by LysRS are the primary substrates for ProXp-ST1 and -ST2 (Ser/Hse- and Thr-tRNALys), AlaRS (Ser-tRNAAla), and ThrRS (Fig. 6). We hypothesize that these domains are required under (Ser-tRNAThr) as well as class I IleRS (Thr-tRNAIle) and ValRS conditions of cellular stress, which is when LysU levels are (Thr-tRNAVal). Given their in vitro specificity, we renamed ProXp-y elevated. The overlapping distribution of lysU and proXp-y in “ProXp-ST1” and ProXp-z “ProXp-ST2” (Fig. 6). ProXp-ST1 and -ST2 have very similar specificities in vitro and are more similar in sequence to each other than to other INS homologs (26), although they differ in a residue that is critical for activity in all INS homologs investigated to date (38). With the exception of ProXp-ST2, a Lys residue (K52 in E. coli ProXp-ST1) is strictly conserved in the active site of all INS-like domains and is proposed to play a role in A76 phosphate backbone recognition and substrate positioning (38, 39). In the case of ProXp-ST2, this residue is a strictly conserved Asn (see sequence alignment in Fig. S8A). B. parapertussis N66A ProXp-ST2 failed to deacylate Ser-tRNAThr unless very high concentrations (∼8 μM) of enzyme were added, as is consistent with a role for N66 in substrate binding (Fig. S8B). With the exception of LysRS and SerRS, all the other synthetases listed in Fig. 6 have known cis posttransfer editing capability, and in some cases trans-editing domains exist also. For example, redundant mechanisms exist for correcting errors made Fig. 6. Summary of aa-tRNA deacylation capabilities of ProXp-ST1 (ProXp-y) by AlaRS caused by the similar structure of Ser relative to cognate and ProXp-ST2 (ProXp-z) and the ARSs that generate the mischarged substrates. Ala (2). AlaXp, ProXp-ST1, and ProXp-ST2 can all deacylate The major substrates appear to be Ser- and Thr-tRNAs with weaker activity Ala toward the nonprotein amino acid Hse. Based on the results of our analysis, the Ser-tRNA , and their distribution is not always exclusive (Fig. most likely candidates for substrates are (i)Ser-tRNAThr, which is formed by S7). In the organisms where they do not overlap, differing en- mischarging errors of ThrRS under oxidative stress (bold circle), (ii)Ser/Hse-and vironmental conditions may determine which enzyme is needed. Thr-tRNALys, which are formed by LysRS (bold circle), and (iii) Thr-tRNASer, In cases where there is overlap with AlaXp, the ProXp domains which is formed by SerRS (dotted circle indicating ProXp-ST1 only).

6030 | www.pnas.org/cgi/doi/10.1073/pnas.1423664112 Liu et al. Downloaded by guest on September 26, 2021 Gammaproteobacteria supports a functional relationship be- Dye Mix MA and 0.005 mM β-D-1-thiogalactopyranoside (IPTG) and in the ab- tween these two genes in this phylum. However, many species that sence or presence of the indicated concentration of specific amino acids. encode ProXp-ST1 and/or ProXp-ST2 do not encode LysU (e.g., Cell growth was monitored at 37 °C for 48 h using the Omnilog system B. parapertussis). In these cases, LysS may be more error prone, (Biolog Inc.). especially under conditions of cellular stress, requiring the exis- – trans Halo Assays. Overnight cultures of WT and proXp-y null strains grown in LB tence of a -editing factor. medium were diluted 1:100, and 1-mL aliquots were pipetted onto LB plates. Environmental stress and other factors may result in an im- Plates were dried, and a 200-μL plug was made in the middle of the plates. balance of amino acid levels in the cell, and these imbalances can Solutions of kanamycin (33 mg/mL), neomycin (10 mg/mL), or streptomycin pose a difficult challenge for the fidelity of ARSs. In addition to (50 mg/mL) (100 μL each) were allowed to absorb into the plates, which were Ser/Thr ambiguity, Hse is a structurally related nonprotein incubated at 37 °C overnight followed by measurement of the diameter of amino acid that is misactivated by LysRS (5) and is readily the clearing zone. All conditions were tested in triplicate. imported into the cell, resulting in toxic effects. Although the exact mechanism of Hse toxicity is unknown, Thr can rescue Enzyme Preparation. The E. coli proXp-y gene was cloned into pET15b with the effect, and it is proposed that the toxic effects may be caused an N-terminal His-tag and was overexpressed in E. coli BL21 (DE3) by the addition of 0.1 mM IPTG and incubation for 12 h at room temperature. The by the misincorporation or misfolding of proteins (43). In addi- B. parapertussis pa2301 gene sequence encoding ProXp-z was codon- tion to LysU, our results showed that Hse is detrimental to the optimized for expression in E. coli and cloned into pET15b (Novagen) by proXp-y–null strain in the presence of overexpressed SerRS. Genewiz. ProXp-z overexpression was carried out in E. coli BL21 (DE3) as However, our in vitro deacylation studies showed that Hse- described for ProXp-y. To prepare ProXp-y and ProXp-z proteins, cells were tRNASer is not a very good substrate for ProXp-ST1 or -ST2 (Fig. pelleted by centrifugation at 4,300 × g for 20 min at 4 °C. Buffer containing β 5). We hypothesize that overexpression of SerRS together with 50 mM NaH2PO4 (pH 7), 300 mM NaCl, and 10 mM -mercaptoethanol was Hse may result in mischarging of noncognate tRNAs in vivo, as used for cell lysis, followed by sonication. The His-tagged proteins were previously reported (44). Alternatively, because Hse is an in- purified using HIS-select nickel resin and were eluted with an imidazole termediate in the biosynthetic pathway to Thr (43), elevated gradient. Fractions containing protein were concentrated using Amicon ul- tra centrifugal filters and were stored in 50 mM NaH2PO4 (pH 7), 150 mM levels of Hse may result in increased Thr in the cell. Metabolomic NaCl, and 40% glycerol at −20 °C. WT E. coli ThrRS, E. coli C666A/Q584H studies to measure the levels of amino acids inside cells directly AlaRS (11), E. coli K279A ProRS (14), E. coli T222P ValRS (50), E. coli LysRS, under a variety of conditions will be needed to address these E. coli SerRS, and E. coli tRNA nucleotidyltransferase (51) were prepared as open questions. described above. Protein concentrations were determined by the Bradford In summary, although ProXp-ST1 and -ST2 are homologs of assay (Bio-Rad). the ProRS editing domain, they do not edit tRNAPro substrates. These multifunctional domains edit multiple amino acids (Ser, Hse, aa-tRNA Substrate Preparation. E. coli tRNASer, tRNAAla, tRNAPro, tRNAIle, Val Thr Lys and Thr) and lack significant tRNA specificity, although ProXp-ST2 tRNA , tRNA , and tRNA were prepared by in vitro transcription using ′ 32 – does not readily deacylate aa-tRNASer. This tRNA encodes a T7 RNA polymerase as previously described (13). All tRNAs were 3 -[ P] labeled by tRNA nucleotidyltransferase (51). Preparation of E. coli Ser-tRNASer, universally conserved G73 in bacteria (45), which may prevent Thr Pro Pro Ile E coli Lys Ser-tRNA , Ser-tRNA , Leu-tRNA , and Ile-tRNA was carried out using deacylation by this factor. Indeed, mutation of . tRNA A73 biotinylated dinitro-flexizyme and Ser-3,5-dinitrobenzyl ester (Ser-DBE), Leu- to G73 abolished the deacylation activity of ProXp-ST2 but had DBE, or Ile-DBE as described (52). Thr-tRNASer and Thr-tRNAIle were prepared little effect on the activity of ProXp-ST1 (Fig. S6 E and F). These using enhanced-flexizyme and Thr-4-chlorobenzyl thioester (Thr-CBT) (52). results indicate either that A73 serves as an important positive E. coli C666A/Q584H AlaRS was used for preparation of E. coli Ala-, Gly-, and recognition element for ProXp-ST2 or, alternatively, that the Ser-tRNAAla; E. coli K279A ProRS was used for preparation of Ala- and Pro- Pro Pro presence of other nucleotides such as G at this position acts as an tRNA ; E. coli WT ProRS was used for preparation of Cys-tRNA ; E. coli T222P ValRS was used for preparation of Val- and Thr-tRNAVal; E. coli anti-determinant to prevent deacylation. Both domains fail to Lys deacylate Thr-tRNAThr in the presence of cognate ThrRS, and LysRS was used for preparation of Lys-, Ser-, and Thr-tRNA ; E. coli ThrRS was used for preparation of Thr-tRNAThr;andE. coli SerRS was used for promiscuous editing of Hse, a nonprotein amino acid, would not preparation ofHse-tRNASer. Following aminoacylation reactions using be expected to be detrimental to the cell. Studies to explore the standard protocols (26), aa-tRNAs were phenol chloroform-extracted fol- tRNA and protein-binding partners of E. coli ProXp-ST1 lowed by ethanol precipitation. Substrates for deacylation assays were under different conditions are under way to help define the stored as pellets at −80 °C. Mischarging assays of Hse-tRNASer were per- biological function of these proteins. It has been estimated that formed at 37 °C with WT E. coli SerRS (3 μM) in the presence of trace nearly 37% of E. coli enzymes are generalists with multiple amounts of E. coli [32P]tRNASer, unlabeled E. coli tRNASer (8 μM), and Hse substrates, shaped by the metabolic network under certain envi- (5 mM). Reactions were initiated by addition of E. coli SerRS following ronmental circumstances (46). Trans-editing proteins such as standard protocols (24). ProXp-ST1 and ProXp-ST2 clearly fit into this category, with Deacylation Assays. In vitro deacylation assays were performed at 20–25 °C in the ability to act on multiple aa-tRNA substrates depending on reactions containing ∼0.75 μM[32P]-labeled aa-tRNA, 50 mM Hepes (pH 7.5), the pressures experienced by the cellular translational machinery. 2 mM DTT, 20 mM KCl, 5 mM MgCl2, 0.1 mg/mL BSA, and 15 μg/mL inorganic pyrophosphatase. Reactions were initiated by addition of 0.5 μM E. coli Experimental Procedures ProXp-y or B. parapertussis ProXp-z and were quenched at the indicated Materials and Strains. All amino acids and chemicals were purchased from time points by putting 2 μL of the reaction into 6 μL of a solution con- Sigma-Aldrich unless otherwise noted. [α-32P]ATP was from PerkinElmer Life taining 200 mM NaOAc (pH 5.5) and 0.5 U/μL of P1 nuclease. The reaction q 32 32 Sciences. E. coli strain BW25113 (lacI rrnBT14 ΔlacZWJ16 hsdR514 ΔaraBADAH33 products (aa-[ P]AMP and [ P]AMP) were analyzed on polyethyleneimine- ΔrhaBADLD78) and the proXp-y–null strain (Δyeak, JW1776) from the Keio cellulose TLC plates as previously described (51). Graphical analysis was collection were obtained from the Coli Genetic Stock Center (47). The plasmid performed using SigmaPlot (Systat Software) with error bars representing pCP20, which carries the FLP recombinase, was used to remove the kanamycin the SD of triplicate measurements. Reactions to measure spontaneous (no cassette from JW1776 mutants (48). Plasmids encoding each ARS gene in enzyme) hydrolysis were performed for each aa-tRNA substrate and were vector pCA24N (-gfp) were from the ASKA collection (49). subtracted from deacylation reactions with enzyme. In addition, all aa-tRNA preparations were shown to be fully deacylated under basic conditions Growth Curve Assays. M9 minimal medium supplemented with 0.4% glucose (0.1 mM NaOH). and indicated concentrations of amino acids was used in all growth curve assays. E. coli cells were grown in LB medium overnight (for strains carrying ARS ex- Bioinformatic Analysis of the Distribution of ProXp-y, ProXp-z, AlaXp, and LysU pression plasmids, 20 μg/mL chloramphenicol was added), collected, and in All Kingdoms of Life. Complete annotated and curated reference pro- BIOCHEMISTRY washed twice with minimal medium. Cells were diluted to an OD600 of 0.01 and teomes were downloaded from UniProt (www.uniprot.org/). Selecting only were grown at 37 °C in M9 minimal medium in the presence of Biolog Redox one strain per species, we obtained a final dataset of 1,597 complete

Liu et al. PNAS | May 12, 2015 | vol. 112 | no. 19 | 6031 Downloaded by guest on September 26, 2021 proteomes, which included 119 Archaea, 1143 Bacteria, and 335 Eukarya. cluster in our analysis and thus could not be distinguished from LysS. Structure-guided multiple sequence alignments of experimentally vali- Therefore, to identify LysU members, we first filtered the LysRS hits by size dated AlaXp, ProXp-y, ProXp-z, and LysU sequences were built using (removing LysX and PoxA members), leaving only LysU and LysS candidates. T-coffee (53) and were converted into hidden Markov model (HMM) profiles Among this set of filtered LysRS hits, we defined LysU-containing species as using HMMER 3.0 (54). Each HMM profile then was used to query our those for which we found two LysRS genes in the same species within our dataset of reference proteomes using HMMER. To distinguish hits that be- filtered dataset of LysRS hits. long to the functional group of interest (i.e., AlaXp, ProXp-y, ProXp-z, and LysU) from those that are homologous editing domains that belong to a ACKNOWLEDGMENTS. We thank Drs. Michael Ibba, Margaret Saks, and Paul functionally different group (e.g., YbaK), the hits obtained by the HMM Schimmel for providing the plasmids encoding E. coli tRNAIle, E. coli tRNASer, profile searches were annotated manually and curated using phylogenetic and E. coli C666A/Q584H AlaRS, respectively; Drs. Jo Marie Bacusmo and analysis, using the HMM alignments of the hits as input for the PHYLIP Mom Das for preparation of E. coli tRNAIle and T222P ValRS, respectively; package (55). Therefore, ProXp HMM hits were classified into ProXp-x, and Ms. Rachel Simari for technical support. This work was funded by Na- ProXp-y, ProXp-z, ProXp-ala, INS, and YbaK, and AlaXp HMM hits were tional Institutes of Health Grants R01 GM049928 (to K.M.-F.) and SAF2012- classified into trans-editing AlaXp domains AlaRS, ThrRS, and TrpRS. In 32200 (to L.R.d.P.). E.M.N. was supported by a La Caixa/Institute for Research contrast, LysU sequences did not form an independent phylogenetic in Biomedicine International Ph.D. Program Fellowship.

1. Pan T (2013) Adaptive translation as a mechanism of stress response and adaptation. 28. Kamtekar S, et al. (2003) The structural basis of cysteine aminoacylation of tRNAPro by Annu Rev Genet 47(1):121–137. prolyl-tRNA synthetases. Proc Natl Acad Sci USA 100(4):1673–1678. 2. Schimmel P (2011) Mistranslation and its control by tRNA synthetases. Philos Trans R 29. Brevet A, Chen J, Lévêque F, Blanquet S, Plateau P (1995) Comparison of the enzy- Soc Lond B Biol Sci 366(1580):2965–2971. matic properties of the two Escherichia coli lysyl-tRNA synthetase species. J Biol Chem 3. Perona JJ, Gruic-Sovulj I (2014) Synthetic and editing mechanisms of aminoacyl-tRNA 270(24):14439–14444. synthetases. Top Curr Chem 344:1–41. 30. Pezo V, et al. (2004) Artificially ambiguous genetic code confers growth yield ad- 4. Martinis SA, Boniecki MT (2010) The balance between pre- and post-transfer editing vantage. Proc Natl Acad Sci USA 101(23):8593–8597. in tRNA synthetases. FEBS Lett 584(2):455–459. 31. Tsui WC, Fersht AR (1981) Probing the principles of amino acid selection using the 5. Jakubowski H (2012) Quality control in tRNA charging. Wiley Interdiscip Rev RNA 3(3): alanyl-tRNA synthetase from Escherichia coli. Nucleic Acids Res 9(18):4627–4637. 295–310. 32. Jakubowski H (1999) Misacylation of tRNALys with noncognate amino acids by lysyl- 6. Schmidt E, Schimmel P (1994) Mutational isolation of a sieve for editing in a transfer tRNA synthetase. Biochemistry 38(25):8088–8093. RNA synthetase. Science 264(5156):265–267. 33. Gruic-Sovulj I, Rokov-Plavec J, Weygand-Durasevic I (2007) Hydrolysis of non-cognate 7. Chen JF, Guo NN, Li T, Wang ED, Wang YL (2000) CP1 domain in Escherichia coli leucyl- aminoacyl-adenylates by a class II aminoacyl-tRNA synthetase lacking an editing do- tRNA synthetase is crucial for its editing function. Biochemistry 39(22):6726–6731. main. FEBS Lett 581(26):5110–5114. 8. Mursinna RS, Martinis SA (2002) Rational design to block amino acid editing of a tRNA 34. Lévêque F, Gazeau M, Fromant M, Blanquet S, Plateau P (1991) Control of Escherichia synthetase. J Am Chem Soc 124(25):7286–7287. coli lysyl-tRNA synthetase expression by anaerobiosis. J Bacteriol 173(24):7903–7910. 9. Lin L, Hale SP, Schimmel P (1996) Aminoacylation error correction. Nature 384(6604): 35. Bullwinkle TJ, et al. (2014) Oxidation of cellular amino acid pools leads to cytotoxic 33–34. mistranslation of the genetic code. eLife 3:e02501. 10. Dock-Bregeon A, et al. (2000) Transfer RNA-mediated editing in threonyl-tRNA syn- 36. An S, Musier-Forsyth K (2005) Cys-tRNA(Pro) editing by Haemophilus influenzae YbaK thetase. The class II solution to the double discrimination problem. Cell 103(6): via a novel synthetase.YbaK.tRNA ternary complex. J Biol Chem 280(41):34465–34472. 877–884. 37. Das M, Vargas-Rodriguez O, Goto Y, Suga H, Musier-Forsyth K (2014) Distinct tRNA 11. Beebe K, Ribas De Pouplana L, Schimmel P (2003) Elucidation of tRNA-dependent recognition strategies used by a homologous family of editing domains prevent editing by a class II tRNA synthetase and significance for cell viability. EMBO J 22(3): mistranslation. Nucleic Acids Res 42(6):3943–3953. 668–675. 38. Bartholow TG, et al. (2014) Strictly conserved lysine of prolyl-tRNA Synthetase editing 12. Roy H, Ling J, Irnov M, Ibba M (2004) Post-transfer editing in vitro and in vivo by the domain facilitates binding and positioning of misacylated tRNA(Pro.). Biochemistry beta subunit of phenylalanyl-tRNA synthetase. EMBO J 23(23):4639–4648. 53(6):1059–1068. 13. Beuning PJ, Musier-Forsyth K (2000) Hydrolytic editing by a class II aminoacyl-tRNA 39. Kumar S, Das M, Hadad CM, Musier-Forsyth K (2013) Aminoacyl-tRNA substrate and synthetase. Proc Natl Acad Sci USA 97(16):8916–8920. enzyme backbone atoms contribute to translational quality control by YbaK. J Phys 14. Wong FC, Beuning PJ, Nagan M, Shiba K, Musier-Forsyth K (2002) Functional role of Chem B 117(16):4521–4527. the prokaryotic proline-tRNA synthetase insertion domain in amino acid editing. 40. Yadavalli SS, Ibba M (2012) Quality control in aminoacyl-tRNA synthesis: Its role in Biochemistry 41(22):7108–7115. translational fidelity. Adv Protein Chem Struct Biol 86:1–43. 15. Lee JW, et al. (2006) Editing-defective tRNA synthetase causes protein misfolding and 41. Zhang CM, Perona JJ, Ryu K, Francklyn C, Hou YM (2006) Distinct kinetic mechanisms neurodegeneration. Nature 443(7107):50–55. of the two classes of Aminoacyl-tRNA synthetases. J Mol Biol 361(2):300–311. 16. Nangle LA, Motta CM, Schimmel P (2006) Global effects of mistranslation from an 42. Ling J, et al. (2009) Resampling and editing of mischarged tRNA prior to translation editing defect in mammalian cells. Chem Biol 13(10):1091–1100. elongation. Mol Cell 33(5):654–660. 17. Bacher JM, de Crécy-Lagard V, Schimmel PR (2005) Inhibited cell growth and protein 43. Kingsbury JM, McCusker JH (2010) Homoserine toxicity in Saccharomyces cerevisiae functional changes from an editing-defective tRNA synthetase. Proc Natl Acad Sci and Candida albicans homoserine kinase (thr1Δ) mutants. Eukaryot Cell 9(5):717–728. USA 102(5):1697–1701. 44. Swanson R, et al. (1988) Accuracy of in vivo aminoacylation requires proper balance of 18. Bacher JM, Schimmel P (2007) An editing-defective aminoacyl-tRNA synthetase is tRNA and aminoacyl-tRNA synthetase. Science 242(4885):1548–1551. mutagenic in aging bacteria via the SOS response. Proc Natl Acad Sci USA 104(6): 45. Jühling F, et al. (2009) tRNAdb 2009: Compilation of tRNA sequences and tRNA genes. 1907–1912. Nucleic Acids Res 37(Database issue):D159–D162. 19. Chong YE, Yang XL, Schimmel P (2008) Natural homolog of tRNA synthetase editing 46. Nam H, et al. (2012) Network context and selection in the evolution to enzyme domain rescues conditional lethality caused by mistranslation. J Biol Chem 283(44): specificity. Science 337(6098):1101–1104. 30073–30078. 47. Baba T, et al. (2006) Construction of Escherichia coli K-12 in-frame, single-gene 20. Karkhanis VA, Mascarenhas AP, Martinis SA (2007) Amino acid toxicities of Escherichia knockout mutants: The Keio collection. Mol Syst Biol 2(1):2006.0008. coli that are prevented by leucyl-tRNA synthetase amino acid editing. J Bacteriol 48. Datsenko KA, Wanner BL (2000) One-step inactivation of chromosomal genes in Es- 189(23):8765–8768. cherichia coli K-12 using PCR products. Proc Natl Acad Sci USA 97(12):6640–6645. 21. Cvetesic N, Palencia A, Halasz I, Cusack S, Gruic-Sovulj I (2014) The physiological target 49. Kitagawa M, et al. (2005) Complete set of ORF clones of Escherichia coli ASKA library for LeuRS translational quality control is norvaline. EMBO J 33(15):1639–1653. (a complete set of E. coli K-12 ORF archive): Unique resources for biological research. 22. Ling J, Söll D (2010) Severe oxidative stress induces protein mistranslation through DNA Res 12(5):291–299. impairment of an aminoacyl-tRNA synthetase editing site. Proc Natl Acad Sci USA 50. Döring V, et al. (2001) Enlarging the amino acid set of Escherichia coli by infiltration 107(9):4028–4033. of the valine coding pathway. Science 292(5516):501–504. 23. Ahel I, Korencic D, Ibba M, Söll D (2003) Trans-editing of mischarged tRNAs. Proc Natl 51. Ledoux S, Uhlenbeck OC (2008) [3′-32P]-labeling tRNA with nucleotidyltransferase for Acad Sci USA 100(26):15422–15427. assaying aminoacylation and peptide bond formation. Methods 44(2):74–80. 24. Schimmel P, Ribas De Pouplana L (2000) Footprints of aminoacyl-tRNA synthetases are 52. Murakami H, Ohta A, Ashigai H, Suga H (2006) A highly flexible tRNA acylation everywhere. Trends Biochem Sci 25(5):207–209. method for non-natural polypeptide synthesis. Nat Methods 3(5):357–359. 25. Korencic D, et al. (2004) A freestanding proofreading domain is required for protein 53. Notredame C, Higgins DG, Heringa J (2000) T-Coffee: A novel method for fast and synthesis quality control in Archaea. Proc Natl Acad Sci USA 101(28):10260–10265. accurate multiple sequence alignment. J Mol Biol 302(1):205–217. 26. Vargas-Rodriguez O, Musier-Forsyth K (2013) Exclusive use of trans-editing domains 54. Eddy SR (2009) A new generation of homology search tools based on probabilistic prevents proline mistranslation. J Biol Chem 288(20):14391–14399. inference. Genome Inform 23(1):205–211. 27. An S, Musier-Forsyth K (2004) Trans-editing of Cys-tRNAPro by Haemophilus influen- 55. Felsenstein J (1988) Phylogenies from molecular sequences: Inference and reliability. zae YbaK protein. J Biol Chem 279(41):42359–42362. Annu Rev Genet 22:521–565.

6032 | www.pnas.org/cgi/doi/10.1073/pnas.1423664112 Liu et al. Downloaded by guest on September 26, 2021