Severe oxidative stress induces protein mistranslation through impairment of an aminoacyl-tRNA synthetase editing site

Abstract

Oxidative stress arises from excessive reactive oxygen species (ROS) and affects organisms of all three domains of life. Here we present a previously unknown pathway through which ROS may impact faithful protein synthesis. Aminoacyl-tRNA synthetases are key enzymes in the translation of the genetic code; they attach the correct amino acid to each tRNA species and hydrolyze an incorrectly attached amino acid in a process called editing. We show both in vitro and in vivo in Escherichia coli that ROS reduced the overall translational fidelity by impairing the editing activity of threonyl-tRNA synthetase. Hydrogen peroxide oxidized cysteine182 residue critical for editing, leading to Ser-tRNA^Thr formation and protein mistranslation that impaired growth of Escherichia coli. The presence of major heat shock proteases was required to allow cell growth in medium containing serine and hydrogen peroxide; this suggests that the mistranslated proteins were misfolded.

Cellular oxidative stress arises from an excess of reactive oxygen species (ROS), such as superoxide radicals, hydrogen peroxide (H₂O₂), hydroxyl radicals, and hypochlorous acid (1–3). High intrinsic levels of oxidative stress occur in some mammalian cells; for instance in HeLa cells over 190 proteins with sulfenic acid modifications have been identified (4). Oxidative stress has been associated with the pathogenesis and progression of many human diseases (5). Although it is well established that ROS are able to damage biomolecules like proteins, DNA, RNA, and lipids, the molecular mechanisms of how oxidative stress contributes to human diseases remain largely elusive. In addition to the cell damaging role, recent studies have shown that ROS also participate in cell signaling pathways via reversible oxidation of critical cysteine (Cys) residues of regulatory proteins, such as OxyR and Hsp33 (6–8).

Extensive studies have revealed that oxidative stress impacts cells at least partially through targeting the protein quality control machinery (9, 10). Protein quality control refers to the cellular regulation of protein folding and degradation (11). Aberrant proteins resulting from translation errors or posttranslational damage are normally misfolded; they then undergo unfolding and refolding by chaperones, degradation by proteases or the proteasome, or aggregation to cause deleterious effects to cells (11, 12). Oxidatively damaged proteins are carefully monitored by chaperones and are rapidly degraded (10, 13). Nevertheless, during severe oxidative stress conditions or in aged cells, the protein quality control capacity becomes saturated by the large amount of misfolded proteins, and protein aggregation prevails (5, 11). In addition to posttranslational damage, misfolded proteins can also be caused by errors in translation (12). Surprisingly, little is known as to whether oxidative stress affects translational fidelity, leading to an increased level of misfolded proteins.

Fidelity in protein synthesis (with an error rate of one in 10³–10⁴) is maintained via correct codon-anticodon recognition on the ribosome and accurate pairing of each amino acid with its cognate tRNA by the aminoacyl-tRNA synthetase (aaRS) (14, 15). An aaRS catalyzes aminoacylation in two steps: (a) activation of the amino acid with ATP to form an aminoacyl adenylate (aa-AMP), and (b) subsequent transfer of the amino acid moiety from aa-AMP to the tRNA (16). The error rate of tRNA selection by an aaRS is as low as one in 10⁶ due to the complexity of tRNA molecules and a kinetic proofreading mechanism (16, 17). In contrast, selection of the correct amino acid is more challenging for the aaRS. While the aminoacylation active site excludes most noncognate amino acids, the enzyme activates certain amino acids or analogues that are structurally similar to the cognate amino acid (15) (Fig. 1). For example, threonyl-tRNA synthetase (ThrRS) misactivates serine (Ser) (18). Such errors are further corrected by a proofreading activity named editing (18–26). A double-sieve model that suggested that the active and editing sites are structurally distinct (27) was later confirmed by biochemical and structural analyses (28, 29). The editing activity of aaRSs is critical for maintaining the overall translational accuracy, as misacylated tRNAs that escape the editing step are effectively used by the ribosome (30–33). AaRS editing deficiency causes growth defects in bacteria (23, 34, 35), apoptosis in mammalian cells (36), and neurodegeneration in mice (12), mainly due to an increased level of misfolded proteins. In HeLa cells a number of aaRS enzymes (ThrRS, glutamyl-tRNA synthetase, and methionyl-tRNA synthetase) have been found to contain sulfenic acid modifications (4). Here we show that oxidative stress directly impaired the editing site of ThrRS, resulting in protein mistranslation. The lack of major heat shock proteases severely exacerbated the growth defect caused by mistranslation, suggesting that oxidative stress-induced mistranslation may harm cells through accumulation of misfolded proteins. Thus, our work revealed a previously unknown pathway through which ROS affects faithful protein synthesis.

Fig. 1.

Double-sieve model of editing by aaRSs. The aaRS aminoacylation active site as the first sieve accepts the cognate (shown in green) and structurally similar near-cognate (i.e., Ser for ThrRS, shown in red) amino acids but rejects the majority of noncognate amino acids. The editing site serves as the second sieve to hydrolyze only the misactivated products (aminoacyl adenylate or aa-tRNA). The correct aa-tRNA is excluded from the editing site and participates in protein synthesis.

Results

H₂O₂ Induces ThrRS Editing Defect and Ser-tRNA^Thr Formation.

Active site cysteines have been found susceptible to oxidation by ROS (6, 37), prompting us to search in the editing sites of aaRSs for critical Cys residues. Previous studies have revealed that ThrRS depends on a Cys residue (Fig. S1) for efficient editing and that replacing the editing site Cys with other amino acids causes Ser to be misacylated to tRNA^Thr species (18, 38). To study the impact of ROS on aminoacylation and editing activities, we tested Escherichia coli ThrRS in an ATP consumption assay, which would determine the overall editing activity of an aaRS against a noncognate amino acid. AaRSs hydrolyze ATP to form aa-AMP and aa-tRNA. In the presence of editing activity, aa-AMP or aa-tRNA is subjected to hydrolysis, which initiates multiple rounds of ATP consumption. We showed that ThrRS could effectively misactivate (Table 1) and edit Ser (Fig. 2A), while the ATP consumption rate in the presence of Thr is negligible. Treating ThrRS with 4 mM H₂O₂ for 5 min significantly reduced the ATP consumption rate, and addition of dithiothreitol (DTT) after H₂O₂ treatment restored the editing activity, demonstrating that ThrRS oxidation by H₂O₂ is reversible. A reduced ATP consumption rate could result from a decrease in either aminoacylation or editing efficiency. To distinguish the two possibilities, we employed pyrophosphate exchange and aminoacylation assays. The pyrophosphate exchange assay measures the radioactivity transferred from pyrophosphate to ATP during reversible amino acid activation and has been traditionally used to determine amino acid activation rates (39). Treatment of ThrRS with 1–4 mM H₂O₂ did not affect the number of active sites or activation of Thr and Ser (Table 1, Fig. S2A), and the steady-state threonylation activities were similar in the presence or absence of H₂O₂ (Table 2, Fig. 2B). In contrast, 0.2 mM H₂O₂ started to cause Ser to be misacylated to tRNA^Thr (Fig. 2C). Treating H₂O₂-oxidized ThrRS with DTT suppressed the misacylation activity (Fig. 2D). Together, these data reveal that the ThrRS editing activity is impaired by H₂O₂ treatment, likely through reversible oxidation of the editing site Cys.

Table 1.

Pyrophosphate exchange activities of E. coli ThrRS with or without H₂O₂ treatment

	Thr			Ser
H2O2 treated	kcat (sec-1)	KM (mM)	kcat/KM (mM-1 sec-1)	kcat (sec-1)	KM (mM)	kcat/KM (mM-1 sec-1)	Selectivity
Yes	11 ± 1	0.082 ± 0.003	130 ± 10	10 ± 1	58 ± 7	0.17 ± 0.03	760
No	13 ± 1	0.090 ± 0.015	140 ± 30	12 ± 1	53 ± 11	0.23 ± 0.05	610

These values are the average of three repeats with standard deviations indicated.

Fig. 2.

ROS cause editing defect and misacylation by WT ThrRS. (A) Total editing activity of ThrRS (1.5 μM) in the presence of tRNA^Thr transcript (2 μM) and cold Ser (10 mM) or cold Thr (10 mM). Treatment of ThrRS with H₂O₂ (4 mM) at 37 °C for 5 min significantly reduces ThrRS editing efficiency. Subsequent treatment with DTT (11 mM) recovers the editing activity. (B) Threonylation (25 μM [¹⁴C]Thr, 5 mg/ml total E. coli tRNA containing 15 μM tRNA^Thr) by ThrRS (37.5 nM). Treating ThrRS with H₂O₂ (4 mM) and/or DTT (11 mM) does not affect the aminoacylation activity. (C) Serylation (30 μM [¹⁴C]Ser, 5 mg/ml total E. coli tRNA containing 15 μM tRNA^Thr) by ThrRS (2.3 μM) in the presence and absence of H₂O₂. Treatment of ThrRS with H₂O₂ (0.2–4 mM) at 37 °C for 5 min induces Ser-tRNA^Thr formation; 0.2 mM is the lowest H₂O₂ concentration at which misacylation is detected. (D) Serylation (30 μM [¹⁴C]Ser, 5 mg/ml total E. coli tRNA containing 15 μM tRNA^Thr) by ThrRS (2.3 μM) is suppressed by DTT and sodium arsenite. ThrRS is treated with H₂O₂ (1 mM) at 37 °C for 5 min and followed by addition of DTT (11 mM) or sodium arsenite (11 mM). tRNA is added last to initiate the aminoacylation reaction.

Table 2.

Threonylation of tRNA^Thr by E. coli ThrRS with or without H₂O₂ treatment

H2O2 treated	kcat (min-1)	KM (μM)	kcat/KM (μM-1 min-1)
Yes	3.2 ± 0.3	1.2 ± 0.4	2.9 ± 1.2
No	3.0 ± 0.2	1.5 ± 0.6	2.3 ± 0.8

These values are the average of three repeats with standard deviations indicated.

Oxidation of the Editing Site Cys182 Residue.

Oxidation of Cys residues can generate several products, e.g., disulfides, sulfenic (SOH), sulfinic (SO₂H), and sulfonic (SO₃H) acids (Fig. S3) (6). To analyze the oxidation product, we used sodium arsenite (NaAsO₂), which specifically reduces sulfenic acids (40), to treat H₂O₂-oxidized ThrRS. NaAsO₂ partially restored the editing efficiency of H₂O₂-treated ThrRS (Fig. 2D), suggesting that a Cys residue was at least partially modified to a sulfenic acid. In ThrRS, Cys182 has been shown to be essential for editing (38, 41). The C182A ThrRS variant displayed misacylation activity in the absence of ROS, confirming the critical role of Cys182 in editing (Fig. S2). Treatment of ThrRS C182A with H₂O₂ did not further enhance misacylation. Using Ellman’s reagent 5, 5′-dithiobis-(2-nitrobenzoic acid) (42), we determined that the wild-type (WT) ThrRS contained 1.9 ± 0.1 free thiols per monomer, whereas the C182A variant contained only 1.0 ± 0.1 free thiol per monomer. Treating the WT ThrRS with H₂O₂ decreased the number of free thiols to 0.7 ± 0.1, consistent with oxidation of roughly one Cys residue. This reveals that Cys182 is one of the two Cys residues exposed to solvent and is thus likely the target for oxidation. To confirm Cys182 oxidation, the WT ThrRS was treated with 1 mM H₂O₂, dialyzed to remove residue H₂O₂, digested with chymotrypsin, and subjected to mass spectrometry [liquid chromatography (LC) MS/MS] analyses. The MS data revealed that Cys182 was oxidized to a Cys sulfinic acid (Fig. S4), a higher order oxidation product derived from Cys sulfenic acids. In an attempt to detect Cys sulfenic acids, we labeled the oxidized ThrRS with dimedone and 7-chloro-4-nitrobenz-2-oxa-1,3-diazole (NBD-Cl), respectively, which forms adducts with protein sulfenic acids (43, 44). LC MS/MS analyses did not identify dimedone or NBD-labeled Cys derivatives, suggesting that the sulfenic acid form of Cys182 is transient and can be further oxidized to Cys sulfinic acids or even disulfides.

H₂O₂ Causes Missense Suppression of the β-Lactamase Gene in E. coli.

In vitro experiments clearly demonstrated that H₂O₂ could damage the ThrRS editing site, resulting in misacylation of Ser to tRNA^Thr. To test whether such mistakes are conveyed to protein synthesis in vivo, a β-lactamase reporter system was employed to detect misincorporation of Ser at Thr codons. Previous studies have shown that Ser68 is critical for the activity of TEM β-lactamase and that an editing-defective ThrRS rescues the lactamase activity of the S68T variant via misincorporation of Ser at the codon Thr68 (33). E. coli strain XL1-blue expressing the WT β-lactamase could grow on ampicillin (Amp) plates with or without H₂O₂, while the strain expressing the S68T variant displayed Amp resistance only in the presence of H₂O₂ (Fig. 3). Replacing the Ser68 codon (AGC) with a near-cognate asparagine codon (AAC) abolished the lactamase activity, which could not be rescued by H₂O₂. Colonies expressing S68T β-lactamase that appeared on plates with Amp and H₂O₂ were streaked on Amp plates without H₂O₂ and showed no growth. This confirmed that H₂O₂-induced β-lactamase activity from the S68T variant was not due to changes at the DNA level but was rather caused by ThrRS editing defects and missense suppression of the ACA codon68 with Ser.

Fig. 3.

H₂O₂ induces Ser misincorporation at Thr codons. E. coli strain XL1-blue expressing β-lactamase variants (1, S68T; 2, WT; 3, S68N) are tested on the following plates: (Top) minimal media with Cm (25 μg/ml) and Amp (30 μg/ml); (Middle) minimal media with Cm (25 μg/ml), Amp (30 μg/ml), and 0.3 mM H₂O₂ (compared to Top, cells grow less well due to the toxic effect of H₂O₂); (Bottom) LB media with Cm (25 μg/ml).

H₂O₂-Induced Ser-tRNA^Thr Formation Causes Protein Mistranslation.

ThrRS editing deficiency results in large-scale protein mistranslation, with Ser randomly inserted at Thr codons. To investigate the impact of H₂O₂-induced protein mistranslation, we tested the growth of E. coli strains in minimal media. The XL1-blue strain displayed a longer lag phase in the presence of 0.8 mM H₂O₂ (Fig. 4A). Addition of Ser to the growth medium further prolonged the lag phase in the presence of H₂O₂ but did not affect the growth in the absence of H₂O₂; this suggests that oxidative stress-induced mistranslation has a negative effect on bacterial growth. The Ser toxicity in the presence of H₂O₂ was suppressed by addition of Thr. In vitro experiments have shown that the ThrRS aminoacylation active site prefers Thr over Ser (Table 1). Therefore, in vivo, Thr would decrease the level of Ser-tRNA^Thr and improve translational accuracy under oxidative stress conditions. The inhibitory effect of Ser in the presence of H₂O₂ was observed only in the lag phase but not in the exponential growth phase. E. coli cells exposed to H₂O₂ trigger an OxyR stress response pathway that decreases the intracellular H₂O₂ level and enhances chaperone activities (1). The prolonged lag phase may reflect the time cells take to reduce the initial H₂O₂ levels, repair damage, and better tolerate an increased level of aberrant proteins.

Fig. 4.

Growth curves of E. coli strains in the presence (closed symbols) and absence (open symbols) of H₂O₂ (0.8–1 mM). (A) E. coli strain XL1-blue. (B) WT E. coli strain MG1655. (C) E. coli strain KY2350, a protease-deficient derivative of MG1655. Squares, minimal media without Ser or Thr; circles, minimal media with Ser (5 mM) but without Thr; triangles, minimal media with Ser (5 mM) and Thr (5 mM). Plots represent the average of three repeats with standard deviations indicated.

Proteases Protect E. coli from the Deleterious Effects of Mistranslation.

Mistranslated proteins are more susceptible to misfolding and degradation by proteases. We reasoned that oxidative stress-induced protein mistranslation might lead to an increased level of misfolded proteins. While the WT E. coli strain MG1655 appeared to tolerate a higher level of oxidative stress and H₂O₂-induced mistranslation (Fig. 4B), the growth of KY2350, a protease-deficient derivative of MG1655, was inhibited by H₂O₂ and an excess of Ser in minimal media (Fig. 4C). Addition of Thr was only able to partially rescue the growth defect of KY2350 in the presence of H₂O₂. It is reasonable to speculate that due to the lack of protease activities in KY2350, the misfolded proteins resulting from mistranslation would overwhelm the protein quality control machinery and form harmful aggregates. A dose-response growth curve analysis revealed that Ser exhibited an inhibitory effect on KY2350 in the presence of 200–1,000 μM H₂O₂, suggesting that ThrRS editing activity might be partially impaired by a minimal level of 200 μM H₂O₂ in the media (Fig. S5A).

Discussion

Physiological Impact of ROS-Induced ThrRS Editing Defect.

A common type of stress in organisms from all three domains of life is oxidative stress accompanied by an increasing level of ROS. Here we show that H₂O₂ can oxidize a critical editing site Cys of ThrRS, resulting in Ser-tRNA^Thr formation and protein mistranslation. The minimal H₂O₂ concentration required to impair ThrRS editing is about 200 μM both in vitro and in vivo (Figs. 2 and 3 and Fig. S5A). The initial bacterial defense against peroxide stress senses low in vivo concentrations (< 10 μM) of H₂O₂ and activates a transcription factor OxyR to induce the expression of several proteins (e.g., HPI catalase, DnaK and GroES) (1, 45). Catalases convert H₂O₂ to H₂O and O₂, whereas chaperones are critical for protein quality control. Thus, OxyR stress response both effectively reduces intracellular H₂O₂ concentrations and enhances the tolerance of bacteria to aberrant proteins, derived from either direct oxidative damage of proteins or an increased level of mistranslation. However, under severe oxidative stress conditions, such as exposure of bacteria to phagocyte-derived oxidants during host defense, the level of exogenous H₂O₂ may exceed the capacity of the OxyR system and exert toxic effects (46). Killing of E. coli by H₂O₂ is bimodal. Low H₂O₂ concentrations (1–3 mM) are fatal for E. coli cells by inducing DNA damage, whereas the mechanism of bacteria killing by high concentrations (> 20 mM) is unclear (46, 47). At the concentrations required to induce protein mistranslation (0.2–0.5 mM) H₂O₂ does not significantly kill E. coli as indicated by a survival assay (Fig. 4B), suggesting that ROS-induced mistranslation might be bacteriostatic but not bactericidal.

Oxidation of Cys Residues in ThrRS by H₂O₂.

Chemical and mass spectrometry experiments suggested that Cys182 of ThrRS was oxidized to a Cys sulfenic acid by H₂O₂, which could be reduced by sodium arsenite or DTT. While DTT completely restored the editing activity of the oxidized ThrRS, sodium arsenite only partially rescued the editing efficiency (Fig. 2). It is therefore possible that the Cys sulfenic acid further forms a disulfide, which is reducible by DTT but not by sodium arsenite. The ThrRS editing site residues (including Cys182 and several histidines) are similar to the active sites of Cys proteases (38), which are known to form sulfenic acids (6), further supporting the notion that a Cys sulfenic acid forms in the oxidized ThrRS editing site. In addition to the editing site Cys, the aminoacylation active site contains a Cys (Cys334) involved in zinc binding (48), which might also be oxidized by ROS. However, under tested conditions neither the aminoacylation activity nor the selectivity for Thr over Ser was changed upon ThrRS oxidation (Tables 1 and 2), suggesting that the ROS damage of ThrRS mainly causes an editing defect.

ROS, Cell Damage, and Signaling.

ROS causes protein oxidation which generates carbonyls, indicators of damaged proteins (10). A moderate level of carbonylation enhances degradation of damaged proteins by proteases and the proteasome, but heavily oxidized proteins can form aggregates and harm cells (10, 13) (Fig. 5). The protein synthesis machinery might also be affected by ROS. For example, glutamyl-tRNA synthetase oxidized by H₂O₂ is shown to be a substrate for the thiol-disulfide oxidoreductase DsbA (49); three HeLa cell aaRSs (ThrRS, glutamyl-tRNA synthetase, and methionyl-tRNA synthetase) have been found to carry sulfenic acid modifications (4); recently it has also been shown that ROS increase translational errors by causing methionine to be misacylated to noncognate tRNAs (50).

Fig. 5.

Model for oxidative stress-induced mistranslation. ROS cause editing defects to ThrRS, leading to global protein mistranslation. While certain mistranslated proteins may participate in signal transduction, the majority of incorrectly made proteins are misfolded, overwhelming the protein quality control system. Accumulation of protein aggregates will eventually damage the cell.

Increasing evidence has supported that ROS play regulatory roles as cell signals (7, 37, 51). Among the well-characterized ROS sensor proteins are OxyR, Hsp33, and protein tyrosine phosphatases (7, 8). A common switch for ROS-signaled pathways is the reversible oxidation of Cys residues, which allows the sensor protein to detect the flux of cellular ROS levels. The active site Cys residues in signaling proteins are often first oxidized to sulfenic acids, which can be stabilized by forming disulfide bonds with intramolecular or intermolecular thiols (6, 51). Both sulfenic acids and disulfide bonds are reducible by cellular reductants such as thioredoxin and glutaredoxin (6). Oxidation of ThrRS results in misincorporation of Ser, a common site for phosphorylation, at Thr codons. It is thus possible that regulation of editing activity by ROS might involve signaling pathways.

Editing and Protein Synthesis Fidelity.

Four decades have passed since editing has been initially discovered, yet the physiological roles of editing are only beginning to be understood (15). The editing activities of many aaRSs are strictly conserved in all three domains of life, implying a critical role of editing throughout evolution. Disruption of the editing activity in certain aaRSs causes bacterial growth defects and apoptosis in mammalian cells (34, 36). Furthermore, an editing site mutation in alanyl-tRNA synthetase only slightly enhances misacylation but leads to an increased level of protein misfolding and severe neurodegeneration in sticky mice (12). Terminally differentiated neurons are particularly sensitive to misfolded proteins, presumably because such cells cannot dilute misfolded proteins through cell division. Further, the nervous system produces a high level of ROS, as neurons are low in antioxidant capacity and consume more oxygen for energy production than other cells (5). Recent studies have shown that in HeLa cells with high oxidative stress levels, ThrRS contains sulfenic acid modifications (4), suggesting that mammalian ThrRSs are susceptible to oxidation. It remains unclear whether oxidative stress affects ThrRS aminoacylation fidelity in mammalian cells.

Although editing is a crucial activity, loss of editing can be beneficial under certain extreme conditions (33, 52). For example, when the cognate amino acid is depleted, the growth rate of Acinetobacter baylyi increased in the presence of an editing-defective isoleucyl-tRNA synthetase (52). It has also been shown that E. coli can survive certain selective pressures only if the protein mistranslation level is increased (30, 33). The interplay between stresses, editing, and adaptation of organisms during evolution raises a variety of questions in the field of protein synthesis quality control. The proposed pathway of ROS-induced mistranslation provides a fascinating example of such interactions between organisms and stresses.

Experimental Procedures

Strains, Plasmids, and Protein Expression.

E. coli strain MG1655 and its protease-deficient derivative KY2350 [ΔhslVU, Δ(clpPX-lon)] (53) were gifts from M. Kanemori (HSP Research institute, Kyoto, Japan). ThrRS was cloned into pET28a expression vector (Novagen) with an N-terminal six-His tag. Expression of ThrRS was induced at 37 °C for 4 h with 0.5 mM isopropyl β-D-1-thiogalactopyranoside (IPTG) in E. coli strain BL21-codon plus in Luria-Bertani (LB) media. His-tagged ThrRS were purified according to standard procedures. β-lactamase gene variants were cloned into a pTech vector with a constitutive lipoprotein (lpp) promoter and a chloramphenicol (Cm) resistance gene.

In Vitro Assays.

AaRS experiments were performed as described (23, 54). The ATP consumption reaction mix contained 2 U/ml of yeast pyrophosphatase (Roche), 10 mM Ser, 2–14 μM cognate tRNA, 2 mM [γ-³²P]ATP (10 mCi/ml in the stock, and 5 cpm/pmole in the final reaction), 100 mM Na-HEPES (pH 7.2), 30 mM KCl, 10 mM MgCl₂, and 1.5 μM ThrRS. The pyrophosphate exchange experiment contained 100 mM Na-HEPES pH 7.2, 30 mM KCl, 10 mM MgCl₂, 2 mM KF, 2 mM ATP, 2 mM [³²P]-pyrophosphate (10 mCi/ml in the stock and 1 cpm/pmole in the final reaction), 10–1,000 μM Thr or 5–500 mM Ser, and 150 nM ThrRS. Aminoacylation experiments were done in the presence of 100 mM Na-HEPES pH 7.2, 30 mM KCl, 10 mM MgCl₂, 2 mM ATP, 25 μM [¹⁴C]-Thr (44 μCi/ml or 190 cpm/pmole) or 30 μM [¹⁴C]-Ser (49 μCi/ml or 180 cpm/pmole), 2–14 μM tRNA transcript or 5 mg/ml total tRNA (from E. coli MRE 600, Roche), and 37–2,300 nM ThrRS.

For mass spectrometry analyses, ThrRS was oxidized with 1 mM H₂O₂ for 5 min at 37 °C. The remaining H₂O₂ was removed by extensive dialysis, and the resulting sample was labeled with 5 mM dimedone (Sigma-Aldrich) or 1 mM NBD-Cl (Sigma-Aldrich) for 1 h at room temperature in the anaerobic chamber. LC MS/MS was performed on labeled ThrRS at W. M. Keck Foundation Biotechnology Research Lab at Yale University.

In Vivo Assays.

For growth curve measurement, the minimal media contained 48 mM Na₂HPO₄, 22 mM KH₂PO₄, 8.5 mM NaCl, 23 mM NH₄Cl, 2 mM MgSO₄, 0.1 mM CaCl₂, 1% glucose, 0.1 mM thiamine, 20 μg/ml of each amino acid except Cys, Thr, and Ser, and indicated concentrations of Ser, Thr, and H₂O₂. E. coli cells were grown in LB media overnight, collected with centrifugation, washed twice with minimal media, diluted to A₆₀₀ ∼ 0.02, and grown at 37 °C in minimal media. In the β-lactamase reporter assay, E. coli XL1-blue cells expressing β-lactamase variants were grown in LB media with 25 μg/ml Cm overnight, collected with centrifugation, washed twice with minimal media, and streaked on minimal medium plates with 25 μg/ml Cm, 30 μg/ml Amp, with or without 0.3 mM H₂O₂. Plates were grown at 37 °C for 24–48 h.