A thioredoxin-dependent peroxiredoxin Q from Corynebacterium glutamicum plays an important role in defense against oxidative stress

Peroxiredoxin Q (PrxQ) that belonged to the cysteine-based peroxidases has long been identified in numerous bacteria, but the information on the physiological and biochemical functions of PrxQ remain largely lacking in Corynebacterium glutamicum. To better systematically understand PrxQ, we reported that PrxQ from model and important industrial organism C. glutamicum, encoded by the gene ncgl2403 annotated as a putative PrxQ, played important roles in adverse stress resistance. The lack of C. glutamicum prxQ gene resulted in enhanced cell sensitivity, increased ROS accumulation, and elevated protein carbonylation levels under adverse stress conditions. Accordingly, PrxQ-mediated resistance to adverse stresses mainly relied on the degradation of ROS. The physiological roles of PrxQ in resistance to adverse stresses were corroborated by its induced expression under adverse stresses, regulated directly by the stress-responsive ECF-sigma factor SigH. Through catalytical kinetic activity, heterodimer formation, and bacterial two-hybrid analysis, we proved that C. glutamicum PrxQ catalytically eliminated peroxides by exclusively receiving electrons from thioredoxin (Trx)/thioredoxin reductase (TrxR) system and had a broad range of oxidizing substrates, but a better efficiency for peroxynitrite and cumene hydroperoxide (CHP). Site-directed mutagenesis confirmed that the conserved Cys49 and Cys54 are the peroxide oxidation site and the resolving Cys residue, respectively. It was also discovered that C. glutamicum PrxQ mainly existed in monomer whether under its native state or functional state. Based on these results, a catalytic model of PrxQ is being proposed. Moreover, our result that C. glutamicum PrxQ can prevent the damaging effects of adverse stresses by acting as thioredoxin-dependent monomeric peroxidase could be further applied to improve the survival ability and robustness of the important bacterium during fermentation process.


Bacterial strains and culture conditions
The bacterial strains and plasmids used in this study are listed in S1 Table. C. glutamicum and Escherichia coli strains were cultured in Luria-Bertani (LB) broth aerobically on a rotary shaker (220 rpm) or on LB plates at 30˚C or 37˚C, respectively, as previously reported [25,31]. To construct the ΔprxQ in-frame deletion mutants, the pK18mobsacB-ΔprxQ plasmids were transferred into C. glutamicum by electroporation and chromosomal integration was selected by single crossover on LB agar plates containing 25 μg/ml kanamycin and 40 μg/ml nalidixic acid. After the above kanamycin-resistant (Km R ) colonies were cultured overnight in LB for a second cross-over, the ΔprxQ deletion mutants were subsequently screened on LB agar plates containing 20% sucrose and 40 μg/ml nalidixic acid and then confirmed by PCR and DNA sequencing as previously described [25,31]. For complementation in relevant C. glutamicum strains, the pXMJ19 derivatives were transferred into relevant C. glutamicum strains by electroporation and the expression in C. glutamicum was induced by addition of 0.5 mM isopropyl β-D-1-thiogalactopyranoside (IPTG). Sensitivity assays for peroxides were investigated as described [24][25][26][27]. To express and purify His 6 -or GST-tagged proteins, these pET28a and GST derivatives were transferred into E. coli BL21(DE3) strains. Recombinant proteins were obtained as described [24]. Cleavage of the His 6 tag was performed by adding 10 units of Enterokinase-Max (Invitrogen, Karlruhe, Germany) and incubation at 4˚C overnight to conduct subsequent polymerization analysis via PAGE and gel filtration chromatography. Ni-NTA agarose was used to remove the cleaved tag and uncleaved protein from the tag-free protein. All enzymes were purchased from Sigma-Aldrich (St. Louis, MO). All antibiotics were purchased from Gold Biotechnology (Shanghai, China). All chemicals, oxidants, and heavy metals were purchased from aladdin (Shanghai, China). When needed, antibiotics were used at the following concentrations: kanamycin, 50 μg ml -1 for E. coli and 25 μg ml -1 for C. glutamicum; nalidixic acid, 40 μg ml -1 for C. glutamicum; chloramphenicol, 20 μg ml -1 for E. coli and 10 μg ml -1 for C. glutamicum.
The plasmid pK18mobsacB-ΔprxQ was made by overlap PCR with primers listed in S2 Table [32]. Briefly, the 904-bp upstream fragment and the 890-bp downstream fragment of prxQ were amplified with primer pairs DPrxQ-F1/DPrxQ-R1 and DPrxQ-F2/DPrxQ-R2, respectively. The upstream and downstream PCR fragments were fused together with the primer pair DPrxQ-F1/ DPrxQ-R2 by overlap PCR [32]. The resulting products were digested with BamH I and Xho I, and inserted into similar digested plasmid pK18mobsacB to produce pK18mobsacB-ΔprxQ.
To produce pXMJ19-prxQ, PrxQ-F/PrxQ-R as primers and C. glutamicum genomic DNA as templates were used. The obtained DNA fragments were digested and cloned into similar digested pXMJ19 vector.
To make the cysteine residue at position 49 of PrxQ into a serine residue (PrxQ:C49S), sitedirected mutagenesis was performed by overlap PCR [24]. In brief, the mutant prxQ:C49S DNA segment was amplified by two rounds of PCR. In the first round of PCR, primer pairs DPrxQ-F1/PrxQ-C49S-R and PrxQ-C49S-F/DPrxQ-R2 were used to amplify segments 1 and 2, respectively. The second round of PCR was performed by using PrxQ-F/PrxQ-R as primers and fragment 1 and fragment 2 as templates to get the prxQ:C49S fragment. The prxQ:C49S DNA fragment was digested and cloned into similar digested pET28a or pXMJ19 plasmid, obtaining plasmid pET28a-prxQ:C49S or pXMJ19-prxQ:C49S. The prxQ:C54S, trx1:C32S, trx1: C35S, trx2:C30S, trx2:C33S, prxQ:C49SC54S fragments were gained using similar procedure as described above and also cloned into plasmid pET28a, pXMJ19, or pUT18CM, obtaining corresponding plasmids. All primers used in this study are listed in S2 Table. For getting the lacZ fusion reporter vector pK18mobsacB-P prxQ ::lacZ, the fusion of prxQ promoter to the lacZY reporter gene by overlap PCR was made [32][33]. Firstly, the primers PPrxQ-F1/PPrxQ-R1 and lacZY-F/lacZY-R were used in the first round of PCR to amplify the 700-bp prxQ promoter DNA fragments and the lacZY DNA fragments, respectively. Secondly, PPrxQ-F1/lacZY-R as primers and the first round PCR products as template were used to perform the second round of PCR, and the resulting fragments were digested with Sal I/Pst I and inserted into similar digested pK18mobsacB to obtain the pK18mobsacB-P prxQ ::lacZ fusion construct [26].
The fidelity of all constructs was confirmed by DNA sequencing (Sangon Biotech, Shanghai, China).

Catalytic kinetic activity measurement
Peroxidase activity of PrxQ toward hydrogen peroxide (H 2 O 2 ), Cumene-OOH (CHP), and t-Butyl-OOH (t-BOOH) was measured using an NADPH-coupled spectrophotometric method as described [34]. The assays were performed in the reaction mixture (500 μl) containing 50 mM Tris-HCl buffer (pH 8.0), 2 mM EDTA, 250 μM NADPH, 1 μM PrxQ (WT or its variants), and the reduced Trx-generating system (15 μM thioredoxin reductase (TrxR) and 40 μM Trx), or the Mrx1-generating system [500 μM MSH, 15 μM mycothione reductase (Mtr), and 40 μM mycoredoxin 1 (Mrx1)]. After a 5 min preincubation, the addition of 1 mM H 2 O 2 , CHP, or t-BOOH started the reaction. The catalytic kinetic parameters for one substrate were acquired by changing its concentration in the case of saturating concentrations of the other substrate (between 0 and 120 μM for Trx or between 0 and 2 mM for peroxide). The decrease of NADPH absorbance was monitored at 340 nm. The activity was determined after subtracting the spontaneous reduction rate detected without PrxQ, and the micromoles amount of NADPH oxidized per second per micromole of enzyme (i.e. turnover number, s −1 ) was calculated by the molar absorption coefficient of NADPH at 340 nm (ε 340 ) of 6220 M −1 cm −1 . Three independent experiments were performed at each substrate concentration. The k cat and K m values of PrxQ were acquired from a non-linear fit with the Michaelis-Menten equation by the program GraphPad Prism 5.

Determination of the reaction activity of PrxQ with peroxynitrite
The reaction rate of PrxQ with peroxynitrite was investigated using a competition approach by measuring the inhibitor effect of increasing PrxQ concentrations on peroxynitrite-mediated HRP oxidation to compound I as described previously [35][36][37][38][39].

Determination of the oligomerization and redox states
The oligomerization state of PrxQ was analyzed with 15% native-PAGE according to the method described previously [26]. The mixtures of the tag-free native PrxQ and the loading buffer containing 250 mM Tris-HCl (pH6.8), 0.5% (W V -1 ) bromophenol blue (BPB) and 50% (V V -1 ) glycerol were separated on 15% PAGE and subsequently stained with Coomassie Brilliant Blue.
The oligomerization state of PrxQ was also analyzed with gel filtration method on Superdex 75 10/300 GL column connected to an FPLC system (GE Healthcare, Piscataway, NJ) as previously described [26]. The tag-free proteins were loaded onto the column pre-equilibrated with 0.15 M NaCl-containing potassium phosphate buffer (50 mM, pH 7.2). The molecular mass of standard proteins used was: aprotinin (6.5 KDa), ribonuclease A (13.7 KDa), carbonic anhydrase (29 KDa), ovalbumin (44 KDa) and conalbumin (75 KDa) (GE Healthcare, Piscataway, NJ). The K av values for each of the standard protein were calculated by the equation: K av = (V e -V 0 )/(V c -V 0 ), where V 0 = column void volume, V e = elution volume and V c = geometric column volume. The flow rate was stetted as 0.5 ml min -1 and the absorbance was monitored at 280 nm. The calibration curve was obtained by the plots of K av values from the above standard proteins against the log of the molecular weight of the standard protein. These data were suited with a linear equation. The molecular weight of the unknown protein could be calculated by its measured elution volume, its K av value and the calibration curve.
The redox states of PrxQs wild type (WT) or its variants were analyzed as described [26]. Briefly, 10 μM PrxQs incubated with 50 mM DTT (dithiothreitol) or 1 mM H 2 O 2 for 30 min were mixed with the loading buffer containing DTT and β-mercaptoethanol (for reducing conditions) or 50 mM iodoacetamide (IAM) (for non-reducing conditions), and then separated on reducing-or non-reducing 15% SDS-PAGE.

Examination of heterodimers
The heterodimers assays were determined as described with minor modifications [45]. 20 μM PrxQ variants and 15 μM Trx (WT and its variants) were mixed in TE buffer containing 30 mM Tris-HCl (pH 8.0) and 1 mM EDTA. After the reaction mixture was incubated at room temperature for 15 min, 50 μM H 2 O 2 was added and then continued to culture at room temperature for 30 min. The resulting samples were resolved on 15% non-reducing SDS-PAGE.

Construction of chromosomal fusion reporter strains and β-galactosidase activity analysis
Chromosomal fusion reporter strains were obtained as described [33]. Briefly, the lacZ fusion reporter plasmid pK18mobsacB-P prxQ ::lacZ was transferred into the relavant C. glutamicum by electroporation, and the pK18mobsacB-P prxQ ::lacZ fusion reporter strain was selected on LB agar plates containing kanamycin. β-galactosidase activity was determined with o-nitrophenyl-β-galactoside (ONPG) as substrate [46].

Electrophoretic mobility shift assay (EMSA)
EMSA was performed according to the method described previously [47]. In brief, P prxQ fragments were obtained by prxQ promoter region of pK18mobsacB-P prxQ ::lacZ reporter vector as templates and PPrxQ-F2/PPrxQ-R as primers. Each 10-μl EMSA reaction solutions contained 20 mM Tris-HCl (pH 7.4), 4 mM MgCl 2 , 100 mM NaCl, 1 mM DTT, 10% (W V -1 ) glycerol, 20 ng PPrxQ DNA fragments, and different concentrations of His 6 -SigH (0-4 μg). After reaction for 20 min, the protein-probes mixture was separated on a 6% native polyacrylamide gel and then stained with SYBR Green (Promega, Fitchburg, WI). Fragments from the prxQ coding region obtained with primer pair Control-F and Control-R instead of the prxQ promoter were added to the binding assays to act as negative controls.

Western blot analysis
Western blot analysis was performed as described previously [32]. Samples subjected on SDS-PAGE were transferred onto polyvinylidene fluoride (PVDF) membranes. After blocking with 4% (w v -1 ) milk for 2 h at room temperature, membranes were incubated with the primary antibody at 4˚C overnight: anti-His; anti-DNP; anti-GST (Zhongshan Gold Bridge, Beijing). The blots were washed with 0.2% (v v -1 ) Tween 20-containing PBST buffer and then incubated with horseradish peroxidase conjugated secondary antibody (Shanghai Genomics Inc., Shanghai, China). The protein bands were visualized with ECL plus kit (GE Healthcare, Piscataway, NJ).

Quantitative RT-PCR analysis
Quantitative RT-PCR analysis (7500 Fast Real-Time PCR; Applied Biosystems, Foster City, CA) was performed as described previously [48]. The primers used are listed in S2 Table. To obtain standardization of results, the relative abundance of 16S rRNA was used as the internal standard.

GST pull-down assay
The GST pull-down assay was performed as described previously [49]. Retained proteins were examined after SDS-PAGE using the anti-His antibody (Millipore, MA, USA).

Statistical analysis
Statistical analyses of survival rate, ROS level, catalytic kinetic activity and transcription level were determined with paired two-tailed Student's t-test. GraphPad Prism Software was used to carry out statistical analyses (GraphPad Software, San Diego California USA).

Indentification of prxQ gene from C. glutamicum genome and oligomeric state analysis
A blast search of C. glutamicum genome with a consensus PrxQ sequence identified a gene annotated as prxQ (ncgl2403) encoding a protein of 158 amino acids with a theoretical molecular mass of 17 kDa. The C. glutamicum PrxQ showed 49%, 39% and 34% amino acid identities with Mycobacterium tuberculosis PrxQ B, Synechococcus PrxQ, and Escherichia coli PrxQ, respectively (S1 Fig). Previous studies showed that PrxQ usually behave as atypical 2-Cys Prx, which means that it mainly exists in monomer whether under its native state or functional state [15]. To determine the oligomeric form exhibited by C. glutamicum PrxQ, it was expressed in E. coli BL21 (DE3), purified by homogeneity, and cleaved with Enterokinase-Max to remove His 6 tag. As shown in Fig 1A, the tag-free recombinant native PrxQ protein showed a single band with a molecular size of approximately 17.0 KDa on 15% polyacrylamide gel electrophoresis (PAGE) (pH 8.8). To further confirm the oligomeric properties of the tag-free native PrxQ, we performed analytical gel filtration chromatography, which enables the molecular mass to be determined independently of the molecular shape. In the gel filtration chromatogram, a sharp peak occurred at an elution volume of 14.56 ml (Fig 1C). Using the standard curve (Fig 1B), the molecular mass of PrxQ was estimated to be 17 KDa, corresponding well to the molecular mass of PrxQ deduced from its amino acid sequence, indicating that the PrxQ eluted from the column was entirely monomeric. Next, we determined the oxidized form of purified wild-type (WT) PrxQ under H 2 O 2 treatment. As shown in Fig 1D, the oxidized PrxQ migrated as a monomer as judged by its behavior on 15% non-reducing SDS-PAGE. Notably, the monomeric forms of H 2 O 2 -treated PrxQ showed two bands, one of which migrated more rapidly. This would be agreement with the formation of the intramolecular disulfide bonds imparting the proteins more compact configuration and more rapid migration. Together, these findings showed that C. glutamicum PrxQ mainly exists in monomer both under reduced and oxidized states, similar to polpar PrxQ and M. Tuberculosis PrxQ B [29][30].

Roles of PrxQ in adverse stresses responses in C. glutamicum
For evaluating the physiological role of PrxQ in resistance to adverse stresses, we compared the viability of the wild type (WT) C. glutamicum, ΔprxQ mutant, and complementary strains in the presence of H 2 O 2 (100 mM), cumene hydroperoxide (CHP, 11 mM), CdCl 2 (0.3 mM), and NiSO 4 (6 mM). As shown in Fig 2A, deletion of prxQ gene obviously decreased the resistance of C. glutamicum to these conditions compared with the wild type. However, the hypersensitive phenotype of the ΔprxQ mutant to adverse stresses was almost completely reversed by complementation with the wild type prxQ gene, which was very similar to the case for the WT(Vector) strain. Hence, PrxQ is critical for protection against adverse stresses.
It has been proposed that various stresses, such as heavy metals, antibiotics, alkylating agents, and acids, can cause cellular death by a prevalent oxidative damage mechanism that depends on the ROS levels [1]. Previous studies showed that peroxiredoxin cleared ROS [5,50], so we examined whether PrxQ played a role in removing ROS under ROS-inducing stress conditions ( Fig  2B). For all these stresses tested, WT(Vector) strain showed obviously lower ROS levels as compared to that in the ΔprxQ(Vector) strain. However, the complementation strain showed a greater decrease in ROS contents than the ΔprxQ(Vector) strain, and showed the same level of ROS generation as the WT(Vector) strain. ROS escaping from the antioxidant defense system are more easily to react with Cys residue of proteins, leading to the reversible formation of molecular disulfide bonds and mixed disulfide bond, or the irreversible formation of sulfoxidation products and protein carbonylation [51][52]. To test whether PrxQ protects cells through reducing the oxidative damage suffered by proteins under oxidative stress, carbonyl groups on total proteins isolated from WT(Vector), ΔprxQ(Vector), and ΔprxQ(prxQ) strains under exposure to adverse stresses were derivatized with 2, 4-dinitrophenyl hydrazine (DNPH) and detected by Western blot using the anti-DNP antibody. As shown in Fig 2C, the ΔprxQ(Vector) strain showed more increase in the carbonylation level of protein extracts as compared to that in the WT(Vector) strain. However, it is noteworthy that the carbonylation level in the complementary strain was almost completely reduced up to the level of WT(Vector) strain.

Peroxidase activities of PrxQ using Trx as an electron donor
Although the putative C. glutamicum PrxQ was suspected to act as an atypical 2-Cys Prx that receives electrons from Trx/TrxR system to reduce peroxides, the experimental evidence was lacking. To explore this activity, we performed an in vitro assay system by adding different oxidants as substrate and the coupled NADPH oxidation by Trx/TrxR system was monitored at 340 nm. Since C. glutamicum contains three alternative Trxs, i.e. Trx1, Trx2 and Trx3, we examined the ability of PrxQ to use each reducing equivalents to sustain peroxidase activity at saturating concentration of substrates and different concentration reducing power (0-120μM). As shown in Fig 3, the apparent affinity of PrxQ towards Trx1 was obtained in steady-state conditions, mildly higher than the value determined with Trx2. For example, the k cat values of PrxQ for CHP with Trx1 and Trx2 systems were 1.91 ± 0.05 s -1 and 1.58 ± 0.04 s -1 , respectively, while the respective K m values were 4.41 ± 0.75 μM and 5.43 ± 0.99 μM. These correspond to a catalytic efficiency of 4.3×10 5 M -1 Á s -1 and 2.9×10 5 M -1 Á s -1 , respectively. Unlike Trx1 and Trx2, PrxQ cannot use Trx3 as an electron donor. So Trx1 and Trx2 were active with an apparent efficiency order Trx1 > Trx2. Moreover, experiments carried out with different Trx concentrations indicated that the reaction saturate at Trx concentrations between 4 and 14 μM.
We further asked whether PrxQ was able to reduce peroxides via the mycoredoxin 1/ MSH/ mycothione reductase (Mxr1/MSH/Mtr) pathway, a novel electron transfer pathway identified in MSH producing high-(G+C)-content Gram-positive Actinobacteria that functional equivalent to the widely distributed glutaredoxin/glutathione/ glutathione reductase (Grx/GSH/GR) pathway [23]. However, no NADPH consumption was observed by using the Mrx1/Mtr/MSH reducing system (S2 Fig), indicating that this system cannot provide electrons to PrxQ. Together, these results showed that although PrxQ can employ the two Trxs regeneration systems in reducing oxidant, it has higher affinity for Trx1. That is, Trx1 was the most efficient electron donor for PrxQ.
Next, enzymatic activities of PrxQ toward various oxidizing substrates were analyzed using the Trx/TrxR system as the reducing power (Fig 4 and S3 Table). Results showed that the catalytic efficiency (k cat /K peroxide ) of PrxQ towards H 2 O 2 , CHP, and t-BOOH, is quite variable with values of 1. (Fig 4A and 4B). The distinct difference observed with CHP was relevant with a difference in the apparent substrate affinity, with K m values of about 93.74~115.8 μM for CHP, and 220.9~333.4 and 316.8~473.9 μM for H 2 O 2 and t-BOOH. . Experimental data were fitted to single exponentials from which observed rate constants of HRP compound I formation were determined. The latter were plotted against PrxQ concentrations (lower right). The data were presented as means of the values obtained from three independent assays.
The above values obtained from C. glutamicum PrxQ showed comparability with those from Poplar PrxQ and E. coli BCP, indicating that these enzymes have very similar properties [29,17].
Apart from H 2 O 2 , CHP, and t-BOOH, peroxynitrite has also been reported for the oxidizing substrate of M. tuberculosis PrxQ B [30]. Peroxynitrite was formed in vivo from the diffusion-controlled reaction between NO and O 2 •−, and has been implicated in promoting the development of various pathologies [53]. Moreover, the amount of peroxynitrite-induced damage was not inferior to that by the other peroxides [53]. Thus, the ability of C. glutamicum PrxQ to eliminate peroxynitrite was detected. After PrxQ (15 μM) was mixed with peroxynitrite (15 μM), peroxynitrite was quickly reduced, confirming its peroxynitrite reductase ability (S3 Fig). However, peroxynitrite is extremely unstable and decays mostly to nitrate with a half life of 0.3 s -1 at physiological pH or of 0.078 s -1 at pH 8. Therefore, the capacity of PrxQ to reduce peroxynitrite was analyzed by competition approaches as described but not an NADPHcoupled spectrophotometric method [35][36][37][38][39]. As shown in Fig 4C, the experiment that HRP (5 μM) was rapidly mixed with peroxynitrite (1 μM) in the absence of PrxQ resulted in the stoichiometric formation of HRP-Compound I, while addition of PrxQ obviously inhibited the HRP-compound I formation and increased the observed rate constant of HRP compound I formation (Fig 4C). The slope of the plot of k obs vs PrxQ concentration showed that the rate constant of the reduction of peroxynitrite by PrxQ was of (1.  [30]. Together, the preferred substrate of C. glutamicum PrxQ is CHP and peroxynitrite, followed by H 2 O 2 , but t-BOOH is the poorest substrate in all cases.

Enzymatic properties of the PrxQ variants
Amino acid sequence showed that PrxQ contains two conserved Cys residues at positions 49 and 54 (S1 Fig). In order to determine whether C. glutamicum PrxQ acts through a 1-Cys or 2-Cys mechanism, the peroxidase activities of single mutated forms of C. glutamicum PrxQ in each of its two Cys residues (two single-substitution mutations of Cys to Ser, i.e. PrxQ:C49S and PrxQ:C54S) and the double variants PrxQ:C49SC54S were assayed. As shown in Fig 5A, mutant of Cys49 led to a complete loss of the ability of PrxQ to detoxify peroxides, and peroxidase activity was substantially all lost when PrxQ Cys 54 was replaced with Ser, indicating Cys49 and Cys54 had a very crucial role in catalysis process. Thus, it is concluded that Trx can efficiently reduce PrxQ containing Cys49 and Cys54 under oxidative stress, while Trx is not able to effectively reduce 1-Cys PrxQ variant (PrxQ:C49S and PrxQ:C54S) (Fig 5A).
These findings promoted us to compare the ability of PrxQ variants in vivo to complement the oxidation-sensitive phenotype of the ΔprxQ mutant (Fig 5B). While the wild-type prxQ gene completely efficiently recovered the sensitivity of the prxQ mutant to H 2 O 2 , CHP, CdCl 2 , and NiSO 4 , complementation of the prxQ:C49S mutant genes had marginal effects on recovery of the survival rates of the ΔprxQ mutant under adverse stress conditions and ΔprxQ(prxQ: C54S) strains only mildly enhanced such tolerance (Fig 5B). These findings indicate that, in vivo, the Cys49 and Cys54 residues of PrxQ played an antioxidant role upon exposure to different kinds of oxidative stress.

Cys49 and Cys54 are the peroxidative cysteine that is oxidized to sulfenic acid and the resolving Cys residue, respectively
Amino acid sequence comparison showed that Cys49 is the peroxidative cysteine residue (C P ) that is homologous to catalytic Cys present in all selected PrxQ containing a P-(X) 3 [30]. However, Cys54 might be the resolving Cys residues (C R ), which was because it located five residues downstream from the C P . Therefore, to investigate whether Cys49 and Cys54 have the above speculative function, variants PrxQ:C49S and PrxQ: C54S with and without previous exposure to H 2 O 2 were used, which were then treated with NBD-Cl. NBD-Cl can exclusively react with thiol groups and sulfenic acids, and the covalent attachment of NBD-Cl generated an absorption peak at about 420 nm upon reaction with thiol groups, whereas it peaked at about 347 nm upon reaction with sulfenic acids [42]. Following the reaction with NBD-Cl, the absorption spectrum of the PrxQ:C49S variant was unchanged before and after exposure to CHP (Fig 6A), exhibiting only the peak at 420 nm. The PrxQ: C54S variant with H 2 O 2 treatment showed soret bands at 347 nm and 420 nm, indicating the simultaneous reaction of NBD-Cl with sulfenic acids and free thiol groups. Similarly, Cys oxidation was also quantified by measuring the free thiol content per PrxQ monomer by 5, 5-dithiobis (2-nitrobenoic acid) (DTNB) assay. As expected, the PrxQ:C49S variant showed one thiol per monomer before and after H 2 O 2 treatment. However, the PrxQ:C54S under H 2 O 2 treatment lost one thiol groups, compared to the thiol content of DTT-treated states ( Fig  6B). TThe redox state of thiol was further examined using AMS covalent modification. Since the molecular mass of AMS is 0.5 KD and covalent modification of AMS is irreversible, the apparent electrophoretic mobility delay would emerge in proportion to the amount of free thiol groups [43]. The H 2 O 2 -treated, AMS-modified PrxQ:C54S migrated as same as the H 2 O 2 -treated, AMS-unmodified states and faster than its H 2 O 2 -untreated, AMS-modified form (Fig 6C), indicating that Cys49 was not existed in thiols in the H 2 O 2 -treated PrxQ:C54S variant (Fig 6C, lanes 7 and 8). These results suggest that Cys49 is the peroxidatic Cys, which is more susceptible to oxidation upon exposure to H 2 O 2 .

-(T/S)-[P/ F]-(X)-C P -[T/S/P] motif that had been reported to be involved in catalysis via the transient formation of a sulfenic acid
However, H 2 O 2 -treated, AMS-modified PrxQ:C49S migrated slower compared to that of H 2 O 2 -treated, AMS-unmodified form and the same as that of H 2 O 2 -untreated, AMS-modified form (Fig 6C, lanes 2-4), indicating that the Cys54 was still in thiol form under exposure to H 2 O 2 . NBD-Cl assay also revealed that the absorption spectra of the PrxQ:C49S proteins were unvaried before or after exposure to H 2 O 2 (Fig 6A). In line with the above phenomena, the thiol content of DTT-treated states was 0.93 ± 0.09 thiol groups per monomer, while the H 2 O 2 -treated PrxQ:C49S proteins lost zero thiol group compared to its reduced states ( Fig  6B). These data indicate that the PrxQ:C49S variants are not oxidized under H 2 O 2 treatment and Cys54 still remain be the thiol state.

Cys49 and Cys54 form an intramolecular disulfide bond under oxidative stress
To test if these two Cys could undergo disulfide bonding after oxidation, AMS modification and the free thiol content assay were used. As shown in Fig 6C, PrxQ WT treated with H 2 O 2 was not modified by AMS, because AMS and H 2 O 2 -treated PrxQ WT migrated more rapidly than that of AMS-modified, H 2 O 2 -untreated form, and the same as that of H 2 O 2 -treated, AMS-unmodified form. The result was further confirmed by measuring the thiol content with the DTNB assay (Fig 6B). The DTT-treated PrxQ WT contained 1.74 ± 0.13 thiol groups per monomer, but the thiol content decreased to 0.16 ± 0.03 when PrxQ WT was treated with H 2 O 2 . The difference of 1.56 thiol groups between the two preparations is linked to the fully oxidation of PrxQ WT after H 2 O 2 treatment. These data indicate that PrxQ WT is fully oxidized by H 2 O 2 to form disulfide bond between Cys49 and Cys54.

PrxQ forms a mixed disulfide with Trx
To further confirm the results that PrxQ uses Trx as a reducing power, we investigated the direct interaction between Trx1/Trx2 and PrxQ by glutathione S-transferase (GST) pull-down and bacterial two-hybrid assays. Early reports suggested that sulfhydryl group regeneration at the active-site Cys of Prx was performed by the N-terminal Cys of Trx at the CXXC active sites (Trx1_C 32 XXC 35 and Trx2_C 30 XXC 33 ), and a transient intermolecular disulfide bond linking Prx with Trx can be stabilized by removing an internal cysteine of Trx at the CXXC active sites [54][55]. Therefore, to avoid attacking by the C-terminal Cys of Trx at the CXXC active sites, single mutants of Trx protein, namely, Trx1:C35S and Trx2:C33S, were constructed. As shown in Fig 7A, the Trx1:C35S and Trx2:C33S mutants showed a very strong interaction with PrxQ, similar to the positive control. To verify the Trx-PrxQ interaction seen in the bacterial twohybrid assay, we produced recombinant His 6 -Trx:CXXS/GST-PrxQ and tested their interactions with GST-PrxQ using the GST pull-down assay in vitro. As shown in Fig 7B, we observed the specific interaction between GST-PrxQ and His 6 -Trx1:C35S/His 6 -Trx2:C33S.
At the stage of PrxQ regeneration, the formation of a transient intermolecular disulfide bond between PrxQ and Trx1-Cys 32 /Trx2-Cys 30 was identified (Fig 7A and 7B)  Although the mechanisms remain unrevealed, the formation of the nonspecific dimers upon strong oxidative stress treatment has also been reported for the thiol-dependent peroxidase Ohr [56] and AphC [57]. The observation that WT Trx did not form stable heterodimers with PrxQ:C54S was in line with the formation of a transient intermolecular disulfide bridge that is subsequently reduced by the second active site Cys of Trx. In addition, the various Trxs were unable to form dimers with PrxQ:C49S (Fig 7C). These results suggest that Cys-49 of PrxQ is favored by Trx1-Cys 32 /Trx2-Cys 30 . The above in vitro results indicate that PrxQ can indeed be regenerated by Trx.

Oxidant-induced prxQ expression and its positive regulation by SigH
As PrxQ is involved in eliminating cellular ROS induced by adverse stresses, prxQ expressions in response to adverse stresses were investigated by chromosomal P prxQ ::lacZ fusion reporter and qRT-PCR analysis. The lacZ activity of P prxQ ::lacZ chromosomal promoter fusion reporter in the RES167 wild type strain was quantitatively measured under H 2 O 2 , CHP, CdCl 2 , and NiSO 4 at different concentrations ( Fig 8A). Concentrations of adverse stress applied were able to reduce the growth rate but under sub-lethal concentrations (S6 Fig). Compared to the untreated samples, the promoter activity of prxQ was approximately 2.57-, 2.92-, 2.0-, and 2.15-fold in the wild type RES167 reporter strain treated with H 2 O 2 (25 mM), CHP (5.5 mM), CdCl 2 (75 μM), and NiSO 4 (3 mM), respectively ( Fig 8A). Further, expression of the P prxQ ::lacZ Exponentially growing relevant C. glutamicum strains were exposed to different adverse stresses at indicated concentrations for 30 min. The expression levels were measured by qRT-PCR. The mRNA levels are presented relative to the value obtained from wild-type cells without treatment. Relative transcript levels of the gene without treatment were set at a value of 1.0.The values represent the mean results from three independent cultivations, with standard errors. ÃÃ : P 0.01; Ã , P 0.05. (c) Binding of SigH to the prxQ promoter examined by EMSA. The increasing amounts of SigH used were 0, 0.2, 0.5, 1.5, 2.0, 2.5, and 3.5 μg (lane1, 2, 3, 4, 5, 6, and 7, respectively). A 300 bp fragment from the prxQ coding region was added in the binding assay to determine the binding specificity of His 6 -SigH (lane 8).
https://doi.org/10.1371/journal.pone.0192674.g008 fusion displayed a dose-dependent increase in response to these adverse environmental conditions (Fig 8A). A similar dose-dependent pattern of prxQ expression in response to H 2 O 2 , CHP, CdCl 2 , and NiSO 4 was also observed in qRT-PCR analysis (Fig 8B). The strong induction of prxQ by adverse stress suggests that a stress sensor/transcriptional regulator might be involved in the regulation of its expression. As SigH, the stress-responsive extracytoplasmic functionsigma (ECF-σ) factor, was reported to respond to thiol-oxidative stress and regulate the expression of multiple resistance genes [48,58], we examined whether prxQ expression was subjected to SigH regulation by measuring the transcription of chromosomal P prxQ ::lacZ fusions. Under exposure to H 2 O 2 (25 mM), CHP (4.5 mM), CdCl 2 (75 μM), and NiSO 4 (3 mM) for 30 min, a marked decrease of prxQ promoter activity was detected in the ΔsigH mutant compared to the wild-type strain (Fig 8A). However, the prxQ expression in the ΔsigH mutant was almost fully recovered when the regulatory protein was complemented either under adverse stress-treated or -untreated conditions (Fig 8A). A similar pattern of prxQ expression in response to various adverse stresses was observed in the qRT-PCR analysis (Fig 8B). The SigH-depdendent prxQ expression was further confirmed with in vitro EMSA assay by determining the direct interaction between His 6 -SigH and prxQ promoter region. Incubation of a 300 bp PCR fragment probe containing the prxQ promoter sequence with His 6 -SigH led to the formation of DNAprotein complexes and retard mobility of the probe (Fig 8C), and the substitution of objective probe by a 300 bp control DNA amplified from the prxQ coding region abolished the formation of the protein-DNA complex ( Fig 8C, lane 8). Furthermore, the DNA-protein complexes increased in response to more His 6 -SigH used in the reactions. Collectively, these results indicate that SigH activates prxQ expression by directly binding to the prxQ promoter.

Discussion
Prxs are Cys-based peroxidases that react very rapidly with H 2 O 2 , organic peroxides, and peroxynitrite. According to the number of conserved cysteine residues directly involved in catalysis, Prxs are classified into three types and six subgroups: typical 2-Cys (Prx1), atypical 2-Cys (Tpx, PrxQ, and Prx5), and 1-Cys Prxs (Prx6 and AhpE) [9][10][11]. Although Prxs have been extensively studied in many bacteria, especially Prx1 and Tpx, less is known about the PrxQ in C. glutamicum. Here, we present data concerning the molecular mechanism, substrate and donor specificity, and in vivo function of PrxQ from C. glutamicum, which is annotated as the putative PrxQ-BCP subfamily. As expected, C. glutamicum PrxQ acted as a Trx-dependent atypical prokaryotic 2-Cys Prx, Cys49 and Cys54 of whose was C P and C R , respectively. The result was supported by following several lines of evidence: 1) C. glutamicum PrxQ exclusively received electrons from the Trx/TrxR system to reduce peroxides but not the Mrx1/MSH/Mtr system (Fig 3); 2) The Cys49 is completely essential for activity, whereas the catalytic activity of C54S variant substantially all lost ( Fig 5); 3) The thiol content, AMS assay, and non-reducing SDS-PAGE mobility ratio supported the presence of a disulfide bridge between Cys49 and Cys54, because of the introduction of the Ser residue in place of Cys49 or Cys54 leading to be not form disulfide bond between Cys49 and Cys54. We propose here that in the WT form of the enzyme, the regeneration of Cys49 involves the transient formation of a disulfide with Cys54, which is subsequently reduced by Trx (Fig 6); 4) The Cys49 was the peroxidatic cysteine (C p ), forming a covalent NBD adduct and being not the free thiol group as proven by an absorbance maximum at 347 nm and DTNB analysis when PrxQ:C54S was exposed to H 2 O 2 ( Fig  6); 5) C. glutamicum PrxQ was indeed monomeric, which was verified by gel filtration chromatography and non-reducing SDS-PAGE (Fig 1); 6) The Cys49 was a necessary residue to the Trx/TrxR pathway, being residues of interaction between PrxQ and Trx. The absence of heterodimer formation was observed in the mixture of the PrxQ:C49S and Trx1:C35S/Trx2:C33S whereas the heterodimer occurred in the reaction with the PrxQ:C53S and Trx1:C35S/Trx2: C33S (Fig 7).
From the biochemical analysis of C. glutamicum PrxQ, we found that C. glutamicum PrxQ was able to reduce a broad spectrum of hydroperoxides including H 2 O 2 , t-BOOH, CHP and peroxynitrite (Fig 4), but kinetic measurements indicated a preference for peroxynitrite and CHP, while reduction of t-BOOH was much slower. This is consistent to the finding that M. tuberculosis PrxQ B had various oxidizing substrates [30]. However, M. tuberculosis PrxQ A lacked the capacity of reducing peroxynitrite [30]. It is shown that C. glutamicum PrxQ had high identities (49%) with M. tuberculosis PrxQ B, while no significant correlation with M. tuberculosis PrxQ A (S1 Fig). Therefore, we suggested that the different behavior of these kinds of PrxQ could be explained based on their structure difference. More notably, the catalytic efficiency of C. glutamicum PrxQ for H 2 O 2 (k cat /K m of about 4.2~9.6 ×10 3 M -1 ÁS -1 ) is lower compared to that of CHP and peroxynitrite, while it is comparable with the value obtained for the E. coli BCP and Poplar PrxQ (2.45×10 3 M -1 ÁS -1 and 7.98×10 3 M -1 ÁS -1 ) and to those of the 2-Cys Prx from Arabidopsis (36×10 3 M -1 Á S -1 ) [17,29,59]. Given the important roles postulated for H 2 O 2 , especially in the gene transcription, cellular signal transduction, bactericidal action of neutrophils and macrophages, these findings are likely to be of physiological importance that was the significance of PrxQ as a means to mediated H 2 O 2 -regulated normal biological functions.
PrxQ was reported to be regenerated by the classic Trx/TrxR reducing system [17,[29][30]. In the present study, we confirmed that C. glutamicum PrxQ can also be efficiently reduced by the Trx/TrxR system (Fig 3), suggesting that the Trx/TrxR-based reducing system is a common reducing system for PrxQ in different organisms. It was important to mention that C. glutamicum possesses three thioredoxin encoding genes. Therefore, we have carried out an extensive study of the electron donor specificity of PrxQ, testing a large number of Trx. Results showed that their efficiencies differed. The most efficient reductant was Trx1, similar to the reported result of Poplar PrxQ [29]. Actually, Poplar type II Prx, having the ability of interacting with Grx, has two Glu residues equivalent to the Asp residues in position 148 and 156, relating to the interaction between Prx and Grx [51,60]. When comparing C. glutamicum PrxQ with poplar type II Prx, we found that C. glutamicum PrxQ had also two Glu residues (at positions 149 and 154) in positions equivalent to Asp148 and 154 of the Prx/Grx fusion. Thus, the possibility that Mrx1/MSH/Mtr system, considered as functional equivalent to the GSH/Grx/GR system, donates electrons to PrxQ via similar Prx/Grx interaction way cannot be excluded in C. glutamicum. However, we found that regeneration of C. glutamicum PrxQ was not supported by the Mrx1/Mtr/MSH reducing system (S2 Fig). Consistently, the Grx/GSH/GR system failed to support PrxQ activity in Poplar [29] and confirmed in C. glutamicum in the present study. We suggested that the finding was related to the 2-Cys mechanism of the Trx/TrxR system but independent of 1-Cys mechanism (i.e. the mycothiolation/demycothiolation reduction mechanism) of the Mrx1/MSH/Mtr system [28,61]. MSH reduces the sulfenic acid intermediate of peroxidase via the formation of an S-mycothiolated peroxidase intermediate (Protein-SSM) which is then recycled by Mrx1 and the second molecule of MSH. More notably, an interesting phenomenon that PrxQ:C54S had mild effects on recovery of the survival rates of the ΔprxQ mutant under adverse stress conditions but not active in vitro with the Trx/TrxR system as electron donor was observed (Fig 5). The reason why PrxQ:C54S has different behaviors in vivo and in vitro is possibly due to the mechanism employed in reducing PrxQ:C54S. It is shown that upon H 2 O 2 treatment, a stable sulfenic acid intermediate occurred in Cys49 of PrxQ:C54S variant (Fig 6). Moreover, the sulfenic acid formed in C P of C R -lacked PrxQ is more efficiently reduced by Grx/GSH/GR system [18]. This could be an indication that the Cys-SOH derivative of PrxQ:C54S was regenerated by the Mrx1/MSH/Mtr sysem but not by the Trx/TrxR system under oxidative stress. The physiological and biochemical mechanisms of the phenomenon need to be further investigated in the future.
Although it has long been known that Prx is involved in physiological and biochemical functions in numerous bacteria, PrxQ in C. glutamicum remained uncharacterized [4][5]. According to a cell viability assay, when the WT and ΔprxQ strains were exposed to different kinds of stressors, the former was more sensitive than the latter and the hypersensitive phenotype could be completely reversed by complementation with the prxQ gene (Fig 2). This means that, to some extent, PrxQ helps bacteria endure diverse stressors. Numerous studies have described that the exposure of microorganisms to various stressors, such as heavy metals, antibiotics, alkylating agents, and acids, can increase the production of ROS and then induce oxidative stress [1]. Reports have also stated that Prx can remove intracellular ROS, thereby decreasing the amount of damage suffered by bacteria [5]. Against this background, we examined the intracellular ROS level and found that, under various stressors, the levels markedly increased, especially in the ΔprxQ mutant. In addition, previous studies have demonstrated that ROS accumulation results in the irreversible formation of sulfoxidation products, eventually leading to protein carbonylation [51][52]. Therefore, in this study, carbonyl assays were performed on total proteins isolated from ΔprxQ and WT strains. After exposure to different stresses, the carbonylation level of protein extracts was significantly lower in the WT strain than in the ΔprxQ strain (Fig 2). The physiological roles of PrxQ in resistance to multiple oxidative stresses were also corroborated by the induced expression of prxQ in C. glutamicum under various oxidative stresses (Fig 8). These results not only strongly suggest that prxQ expression may be important for bacteria growth under oxidative stress conditions, but also provide a promising strategy to engineer robust industrial strains in the future.
Based on our experimental data, a catalytic model for the alternatively reduction of PrxQ by Trx would be hypothesized (Fig 9). The first step consists of H 2 O 2 reduction with the concomitant formation of a stable sulfenic acid intermediate on catalytic Cys49 of PrxQ. Nucleophilic attack by Cys54 on the sulfenic acid intermediate leads to the release of one molecule of H 2 O and the formation of a transient disulfide bond between Cys49 and Cys54. This disulfide bond is reduced by the Trx/TrxR recycling system to reform the reduced peroxidatic Cys. Then, the regenerated PrxQ is ready for another catalytic cycle.
Supporting information S1  C. glutamicum wild type was grown in LB medium to an OD 600 of 1.0 and exposed to different concentrations of various peroxides at 30˚C for 30 min. After treatment, the cultures were serially diluted, spreaded on LB plates and incubated at 30˚C for 36 h. Survival percentages were calculated as [(CFU ml -1 with stress)/(CFU ml -1 without stress)]×100. Mean values with standard deviations (error bars) from at least three repeats are shown. (TIF)