Regulation of Translation by the Redox State of Elongation Factor G in the Cyanobacterium Synechocystis sp. PCC 6803^*

Preparation of Recombinant Proteins

EF-G, encoded by slr1463 in Synechocystis, was expressed in Escherichia coli BL21 (DE3) as a histidine-tagged recombinant protein and purified, in the reduced form, as described previously (17). The m-type of thioredoxin, encoded by slr0623 in Synechocystis (18, 19), was also purified as described previously (20). These proteins were stored in a buffer that contained 20 mm HEPES-KOH (pH 7.5), 50 mm NaCl, and 20% (w/v) glycerol.

Site-directed Mutagenesis of Proteins

Site-directed mutagenesis was performed with a QuikChange site-directed mutagenesis kit (Stratagene, La Jolla, CA) by PCR with a pair of complementary oligonucleotides of 32–44 bases that contained the desired mutations and pET21b/slr1463 (17) as the template (sequences of primers are available on request). Parental DNA was digested with DpnI to remove methylated parent strands, and the obtained plasmid DNA was used to transform E. coli JM109.

Modification of Thiol Groups of Cysteine Residues

The redox state of the cysteine residues in EF-G was monitored by modifying the thiol groups in EF-G with a maleimidyl reagent, methoxypoly(ethylene glycol) maleimide, which has an average molecular mass of 5 kDa (Nihon Yushi, Tokyo, Japan), with subsequent separation of proteins by non-reducing SDS-PAGE on a 7.5% polyacrylamide gel, as described previously (17).

Quantitative Analysis of Thiol Groups

The number of thiol groups per EF-G molecule was determined as described previously (20). From 10 to 50 μm EF-G was incubated with 0.4 mm 5,5′-dithio-bis(2-nitrobenzoic acid) (Sigma) in buffer that contained 100 mm Tris-HCl (pH 8.0), 10 mm EDTA, and 6 m guanidine HCl. Changes in absorbance at 412 nm were then measured, and the number of reactive thiol groups was determined as described elsewhere (21).

Preparation of Extracts for Translation in Vitro

Extracts for translation in vitro were prepared as described previously (17). Cells of Synechocystis at the late exponential phase of growth were broken open with glass beads. After centrifugation of the resultant homogenate, the blue-green supernatant that contained thylakoid membranes was used as the cell extract for translation in vitro.

Translation in Vitro

Translation in vitro was performed by incubation, at 30 °C, of the above mentioned cell extract with psbA2 mRNA as the template, [¹⁴C]leucine, the other 19 amino acids, and various compounds required for translation, as described previously (17). The incorporation of [¹⁴C]leucine into proteins was quantified by liquid scintillation counting, and the extent of incorporation of the radiolabeled amino acid was taken as a measure of translational activity. For inhibition of translation by H₂O₂, the cell extract was incubated for 10 min at 30 °C in the presence of 10 mm H₂O₂ together with 40 mm NaN₃, which was necessary for inactivation of catalase and peroxidases that had accumulated at high levels together with other proteins in the extract (17).

Reduction of EF-G by Thioredoxin

EF-G (2 μm) was oxidized by incubation with 500 μm H₂O₂ for 15 min at 25 °C. Then 0.75 μm catalase (Nacalai Tesque, Kyoto, Japan) was added to remove residual H₂O₂, and the oxidized EF-G was incubated for 15 min at 25 °C with dithiothreitol (DTT) at various concentrations in the presence and in the absence of 2 μm thioredoxin. Proteins were then precipitated with 10% trichloroacetic acid and subjected to the thiol modification assay for detection of thiol groups on cysteine residues. To assay the activity of EF-G in translation, EF-G that had been reduced by thioredoxin was separated from thioredoxin, DTT, and catalase on a HisTrap column (GE Healthcare) by the same method as used for the purification of EF-G (17).

Determination of Redox Potential

The redox potential of EF-G was determined as described by Motohashi and Hisabori (22) with a minor modification. EF-G (1 μm) was incubated at 25 °C for 16 h in 50 mm potassium phosphate buffer (pH 7.0) that contained 100 mm oxidized DTT and various concentrations of reduced DTT (0.5 μm to 2 mm) under an atmosphere of nitrogen gas. Proteins were then precipitated with 10% trichloroacetic acid and subjected to the thiol modification assay. Intensities of bands, after staining of proteins with Coomassie Brilliant Blue R-250, were determined with the Scion Image system (available on the World Wide Web). The equilibrium constant and the standard redox potential were calculated, taking a value of −330 mV for the standard redox potential of DTT as a reference, as described in the above cited reference (22).

Redox State of EF-G in Vivo

Wild-type cells and cells of a mutant of Synechocystis that lacked NADPH-thioredoxin reductase (19) were grown photoautotrophically at 32 °C in light at 70 μmol of photons m⁻² s⁻¹, as described previously (6). Cells were disrupted by homogenization with glass beads, and cell extracts were obtained as described previously (17). Proteins in cell extracts (2 μg/assay) were precipitated with 10% trichloroacetic acid and subjected to the thiol modification assay. After separation of proteins by non-reducing SDS-PAGE on a 7.5% polyacrylamide gel and subsequent transfer of proteins to a nitrocellulose membrane, EF-G was detected immunologically with an antiserum that had been raised against EF-G (Slr1463) of Synechocystis, as described previously (6).

Oxidation of Two Cysteine Residues by H₂O₂

We monitored the redox state of five cysteine residues of EF-G (Slr1463) by modifying free thiol groups with a maleimidyl reagent (average molecular mass, 5 kDa). Modification of the fully reduced form of EF-G with this reagent resulted in a decrease in the electrophoretic mobility of EF-G on a non-reducing gel (Fig. 1A, lane 2). Incubation of EF-G with 500 μm H₂O₂ prior to the modification decreased this change in mobility (Fig. 1A, lane 3), indicating that some of the five cysteine residues were oxidized by H₂O₂.

In order to estimate the number of oxidized cysteine residues in EF-G, we compared the mobility of wild-type EF-G with that of mutated proteins in which specific cysteine residues were replaced by serine residues. The mobility of the mutated protein C105S fell between the mobilities of the reduced and oxidized forms of wild-type EF-G (Fig. 1A, lane 4). The mobility of a double-mutated protein, C68S/C105S, was the same as that of the oxidized EF-G (Fig. 1A, lane 5), indicating that two cysteine residues in EF-G were oxidized by H₂O₂. Since a dimeric form of EF-G was not detected after oxidation of EF-G with H₂O₂ (data not shown), it seems likely that two oxidized cysteine residues formed an intramolecular disulfide bond. Note that Cys⁶⁸ was not a target of oxidation, as discussed below.

We also determined the number of free thiol groups on cysteine residues in EF-G and its mutated derivatives by monitoring changes in absorption after reaction of thiol groups with 5,5′-dithio-bis(2-nitrobenzoic acid). The numbers of free thiol groups per molecule of reduced EF-G, oxidized EF-G, C105S, and C68S/C105S were 4.6, 3.1, 3.9, and 2.9, respectively (Fig. 1B), confirming that two cysteine residues were oxidized by H₂O₂.

To identify the cysteine residues that were oxidized by H₂O₂, we replaced individual cysteine residues in EF-G by serine residues and investigated the redox state of cysteine residues by the thiol modification assay. In the presence of 500 μm H₂O₂, the cysteine residues in C68S, C242S, C388S, and C676S were oxidized, whereas the cysteine residues in C105S were not oxidized (Fig. 1C). We therefore concluded that Cys¹⁰⁵ is a target of oxidation by H₂O₂. However, we failed to identify the counterpart of Cys¹⁰⁵ for formation of a disulfide bond by this method, perhaps because Cys¹⁰⁵ reacted with another cysteine residue in the mutated protein in which the true counterpart had been replaced by serine.

Effects of Mutation of EF-G on Translation in Vitro

To elucidate the relationship between the redox state of each cysteine residue and translation, we investigated the effects of mutated derivatives of EF-G on translation in vitro in a translation system prepared from Synechocystis. When we added psbA2 mRNA, which encodes the D1 protein, to the translation system together with components required for translation, which included [¹⁴C]leucine, we observed the incorporation of radioactivity into newly synthesized proteins (Fig. 2, WT, closed diamonds). In this assay, the major protein synthesized is the D1 protein; traces of other proteins reflect the presence of intrinsic mRNAs in the translation system (17). The addition of 10 mm H₂O₂ to the translation system inhibited translation (Fig. 2, WT, open circles). This apparent robustness of the translation system against H₂O₂ is due to the presence of catalase and peroxidases in the cell extract (17). The addition of the reduced form of wild-type EF-G to the translation system that had been incubated with H₂O₂ reversed the inhibition of translation to some extent (Fig. 2, WT, closed circles). By contrast, the oxidized form of wild-type EF-G, generated by incubation of EF-G with 2 mm H₂O₂, was unable to restore translational activity (Fig. 2, WT, open triangles). These results indicate that EF-G is active in the reduced form and inactive in the oxidized form.

Among mutated derivatives of EF-G, C68S, C388S, and C676S exhibited essentially the same redox response as the wild-type protein (Fig. 2). However, C105S and C242S that had been incubated with H₂O₂ were able to support translation to the same extent as their reduced forms (Fig. 2), indicating that Cys¹⁰⁵ and Cys²⁴² in EF-G are the possible targets of oxidation by H₂O₂ as well as the sites of regulation of the activity of EF-G in a redox-dependent manner.

Identification of Oxidized Cysteine Residues by Peptide Mapping Analysis

To confirm that Cys¹⁰⁵ and Cys²⁴² are the critical targets of oxidation by H₂O₂, we performed peptide mapping analysis of EF-G. We digested the oxidized form of EF-G, which had been obtained by incubation of EF-G with 2 mm H₂O₂, with lysyl endopeptidase and separated the resulting peptide fragments by reversed-phase HPLC. To obtain peptides in their reduced form, we incubated a portion of the peptide fragments with DTT prior to reversed-phase HPLC. We compared the patterns of elution of peptides derived from oxidized EF-G with patterns of elution of the reduced peptides. Three specific peaks, namely LR1, LR2, and LR3, which were present after HPLC of the reduced peptides, were absent from the chromatographic profile of the peptides derived from oxidized EF-G (Fig. S1). Mass spectrometry revealed that LR1, LR2, and LR3 corresponded to peptides that contained Cys²⁴², Cys³⁸⁸, and Cys¹⁰⁵, respectively (Table S1). Thus, peptide mapping analysis indicated that these three cysteine residues were potential targets of oxidation. We failed to detect peaks that corresponded to disulfide-linked peptides under oxidizing conditions, probably due to incomplete digestion of EF-G or insufficient separation of peptide fragments. The absence of LR2, which corresponded to the peptide that contained Cys³⁸⁸, in the oxidized EF-G was probably due to further oxidation under strong oxidizing conditions, such as 2 mm H₂O₂.

Regulation of the Redox State of EF-G by Thioredoxin

The redox behavior of EF-G that is mediated by cysteine residues suggested that EF-G might interact with thioredoxin, an oxidoreductase that acts by reducing disulfide bonds in target proteins. We examined whether thioredoxin was able to reduce the disulfide bond in the oxidized form of EF-G. The results of a thiol modification assay showed that the oxidized cysteine residues in 2 μm EF-G that had been incubated with 500 μm H₂O₂ were reduced by subsequent exposure to a high concentration of DTT, such as 5 mm (Fig. 3A). By contrast, complete reduction of EF-G was achieved by the addition of 2 μm thioredoxin, even in the presence of only 50 μm DTT (Fig. 3A). Thus, thioredoxin can interact with EF-G and reduce an intramolecular disulfide bond.

A triple-mutated derivative of EF-G, in which Cys⁶⁸, Cys³⁸⁸, and Cys⁶⁷⁶ had been replaced by serine, had a redox response similar to that of wild-type EF-G; the mutated protein was oxidized by H₂O₂ and reduced by thioredoxin (Fig. 3A). This result clearly indicated that the remaining (non-mutated) Cys¹⁰⁵ and Cys²⁴² residues formed a disulfide bond under oxidative conditions and that the disulfide bond was reduced by thioredoxin (Fig. 3B). By contrast, a different mutated derivative of EF-G, in which Cys⁶⁸, Cys²⁴², and Cys⁶⁷⁶ had been replaced by serine, was no longer susceptible to oxidation (Fig. 3A), an observation that excluded the possibility that a disulfide bond was formed between Cys¹⁰⁵ and Cys³⁸⁸ under oxidative conditions.

Determination of the Redox Potential of EF-G

To evaluate the possibility of a putative cascade of electrons from thioredoxin to EF-G, we determined the redox potential of EF-G with redox buffers that contained various ratios of the reduced form of DTT to the oxidized form. The results of a thiol modification assay showed that EF-G was oxidized in two steps when we decreased the relative level of the reduced form of DTT (Fig. 4). From its migration on the gel, the first oxidized form of EF-G appeared to be identical to the form generated upon oxidation by H₂O₂; the second form might have been produced by further oxidation of other cysteine residues by the oxidized form of DTT. Since the two-step oxidation complicated the accurate determination of redox potential, we determined the redox potential of the redox-active cysteine residues, namely Cys¹⁰⁵ and Cys²⁴², using the triple-mutated derivative of EF-G, C68S/C388S/C676S. From the ratios of the reduced form of EF-G to the oxidized form, we determined the redox potential of the target cysteine residues of EF-G to be −221 mV. We also determined the redox potential of thioredoxin-m (Slr0623) of Synechocystis to be −282 mV, which is the same as that of thioredoxin-m (−282 mV) of spinach (22). Thus, EF-G can naturally accept electrons from thioredoxin-m.

Functional Regulation of EF-G by Thioredoxin

We investigated whether the reduction of EF-G by thioredoxin affected the activity of EF-G during translation. We incubated oxidized EF-G with 2 μm thioredoxin and 50 μm DTT, namely under conditions used for the thioredoxin-assisted reduction of EF-G (Fig. 3A), and then we purified the reduced EF-G. When the reduced protein was added to the translation system in vitro that had been incubated with 10 mm H₂O₂ for 10 min, translational activity was restored (Fig. 5A). The direct addition of 2 μm thioredoxin plus 10 μm DTT to the translation system that had been incubated with 10 mm H₂O₂ for 10 min also restored translational activity. By contrast, the addition of 10 μm DTT alone did not restore translation (Fig. 5B). Thus, the reduction of oxidized EF-G by thioredoxin allowed the EF-G to be active in the translation system.

Redox State of EF-G in Vivo

We investigated the redox state of EF-G in cells of Synechocystis with antibodies specific to EF-G (Slr1463). In cells grown in light at 70 μmol of photons m⁻² s⁻¹, the optimum light condition for growth, the majority of EF-G was present in the reduced form, whereas the oxidized form was also present to a lesser extent (Fig. 6). From its migration on the gel, the oxidized form was considered to be identical to the form that had been generated upon oxidation by H₂O₂ in vitro, namely the form that possessed an intramolecular disulfide bond between the two specific cysteine residues (Fig. 6). When cells were incubated for 1 h in light at 200 μmol of photons m⁻² s⁻¹, which is considered to be moderately strong light for this organism, levels of the reduced form of EF-G increased to about 90%. This reduction must be due to the acceleration of the photosynthetic transport of electrons. When we further increased the intensity of light to 1,500 μmol of photons m⁻² s⁻¹, strong light that causes the photoinhibition of PSII (6), some of the EF-G was converted to the oxidized form, most probably as a result of the generation of ROS and the consequent oxidation of EF-G.

In Synechocystis, thioredoxins are reduced by two enzymes, namely NADPH-thioredoxin reductase and ferredoxin-thioredoxin reductase (19). Of the two reducing pathways, both of which are derived from the photosynthetic transport of electrons, the pathway via NADPH-thioredoxin reductase is more important for the antioxidant system (19). In a mutant that lacked NADPH-thioredoxin reductase, Δntr, levels of the reduced form of thioredoxins fell significantly (19). In this same mutant, we found that about 40% of the EF-G was present in the oxidized form, even after incubation in light at 200 μmol of photons m⁻² s⁻¹ (Fig. 6). In strong light at 1,500 μmol of photons m⁻² s⁻¹, levels of the oxidized form of EF-G in the mutant rose to about 50%, indicating that reduction of EF-G by thioredoxin had been interrupted. These observations suggest that EF-G is reduced by thioredoxin-mediated reducing signals that are derived from the photosynthetic transport of electrons.

Oxidation of Two Specific Cysteine Residues Inactivates EF-G

Studies of the effects of ROS on the photoinhibition of PSII have revealed that ROS act, at the elongation step of translation, by inhibiting the synthesis of proteins that are required for the repair of PSII (6, 7). Biochemical studies in vitro with a translation system derived from Synechocystis have demonstrated that EF-G is a primary target, within the translational machinery, of inactivation by ROS (17). EF-G was identified initially as one of the proteins that were most abundantly oxidized in E. coli, as assessed by the formation of carbonyl groups, in cells that had been treated with H₂O₂ (23) as well as in cells that lacked a superoxide dismutase (24). However, the chemical nature of the oxidation of EF-G by ROS remained to be determined. In the present study, we demonstrated that inactivation by H₂O₂ of EF-G (Slr1463) in Synechocystis was due to the oxidation of two specific cysteine residues and the formation of a disulfide bond.

Replacement of two cysteine residues, namely Cys¹⁰⁵ and Cys²⁴², by serine rendered EF-G insensitive to inactivation by H₂O₂ and allowed EF-G to function in translation in vitro even after exposure to H₂O₂ (Fig. 2). These results support the hypothesis that these two cysteine residues are the targets of oxidation by ROS, and furthermore, they suggest that genetic engineering of these cysteine residues of EF-G might alter the sensitivity of the translation system to oxidative stress.

A model of the three-dimensional structure of EF-G suggests that Cys¹⁰⁵ and Cys²⁴² are located close to each other and to the GTP-binding site (Fig. S2). Oxidation of these cysteine residues with subsequent formation of a disulfide bond might induce a conformational change that decreases the binding affinity of EF-G for GTP or its capacity for hydrolysis of GTP. Alternatively, the putative conformational change might prevent EF-G from entering the P site on the ribosome.

The Target Cysteine Residues Are Conserved in Many Organisms

Alignment of the deduced amino acid sequences of EF-G from various organisms revealed that a pair of cysteine residues that correspond to Cys¹⁰⁵ and Cys²⁴² in EF-G (Slr1463) is strongly conserved in the EF-G proteins of several species of cyanobacteria and other prokaryotes as well as in the chloroplast EF-G proteins of some plants (Fig. S3). Moreover, these cysteine residues are conserved even in anaerobic bacteria, such as Rhodobacter sphaeroides ATCC 17025 and Salmonella typhi strain CT18. By contrast and unexpectedly, the EF-G proteins of Thermus thermophilus and Bacillus subtilis lack these cysteine residues. The conserved cysteine residues might play an important role in the regulation of the activity of EF-G, and the susceptibility of these cysteine residues to oxidation might serve as a negative regulator for suppression of translation under oxidative stress in vivo.

The Function of EF-G Is Modified by Thioredoxin

Various metabolic processes in chloroplasts and cyanobacteria are controlled by regulatory systems that involve redox reactions (25–27). A fundamental example of such a redox reaction is the reversible reduction of disulfide bonds of cysteine residues by thioredoxin. The redox behavior of two of the cysteine residues in EF-G of Synechocystis prompted us to investigate the interaction of EF-G with thioredoxin. Assays involving thiol modification and translation in vitro revealed that the disulfide bond in peroxide-oxidized EF-G was reduced by thioredoxin and that the newly reduced form of EF-G was able to function in translation (Figs. 3 and 5).

Thioredoxin affinity chromatography captured a number of putative targets of thioredoxin (25–32). EF-G in spinach chloroplasts (28) and E. coli (29) and EF-2 in Arabidopsis thaliana (30) and Chlamydomonas reinhardtii (31) were identified as candidates for targets of thioredoxin, as was EF-G (Slr1463) in Synechocystis (32). In the present study, we demonstrated for the first time, to our knowledge, that EF-G is an actual and functional target of thioredoxin.

Physiological Implications of the Redox Regulation of EF-G by Thioredoxin

In photosynthesis, reducing equivalents that are generated by the photosynthetic transport of electrons are transmitted to various proteins via thioredoxin and regulate their activity. Several enzymes involved in the Calvin cycle, such as glyceraldehyde-3-phosphate dehydrogenase, fructose-1,6-bisphosphatase, sedoheptulose-1,7-bisphosphatase, and phosphoriburokinase, are activated by thioredoxin (33). ATP synthase in chloroplasts is activated upon reduction of its γ subunit by thioredoxin (34, 35).

The activation of EF-G by the reduced form of thioredoxin suggests that the activity of EF-G might also be regulated by the reducing power that is generated by the photosynthetic transport of electrons and mediated by thioredoxin. We found that the translational activity of EF-G was regulated by its redox state (Fig. 2) (17). Furthermore, Western blotting analysis revealed that the redox state of EF-G is regulated by thioredoxin in vivo (Fig. 6). Thus, it seems possible that the translational machinery might be regulated by the photosynthetic transport of electrons in a redox pathway that includes thioredoxin and EF-G. This phenomenon might also explain aspects of the light-dependent regulation of translation. Synthesis of the D1 protein is induced by light at the translational level in plants (36, 37) and algae (38) and at both the transcriptional and the translational level in cyanobacteria (39, 40). In addition, the activation of the synthesis of the D1 protein requires reducing power derived from the photosynthetic transport of electrons (41–43).

Two other homologs of EF-G, namely Sll1098 and Sll0830, have been found in Synechocystis (Fig. S3), and they have been shown to function as EF-G (17). Sll1098 has the pair of conserved cysteine residues found in EF-G, but Sll0830 does not. It is possible that Sll1098 might be redox-regulated in the same way as EF-G (Slr1463), whereas Sll0830 might function constitutively. The roles of these two homologs in translation remain to be elucidated.

ROS oxidize a specific pair of cysteine residues in EF-G, and these residues are also targets of thioredoxin. When EF-G is oxidized by ROS, it is no longer able to support translation. The specific cysteine residues in EF-G are strongly conserved in cyanobacteria and chloroplasts (Fig. S3). These observations together suggest a strategy whereby photosynthetic organisms are able to survive under oxidative conditions. Environmental stress, such as strong light, often give rise to oxidative stress, producing ROS via photosynthesis (1). Under severe oxidative conditions, translation might be suppressed to prevent the synthesis of proteins that would otherwise enhance oxidative stress by accelerating the repair of PSII and reactivating or stimulating the photosynthetic transport of electrons. In fact, suppression of translation has been observed both under oxidative stress (6, 7) and when excess reducing power accumulates in the photosynthetic machinery (13–15). Thus, EF-G might act as a safety valve to suppress translation under oxidative stress.