Stable integration of the Mrx1-roGFP2 biosensor to monitor dynamic changes of the mycothiol redox potential in Corynebacterium glutamicum

Mycothiol (MSH) functions as major low molecular weight (LMW) thiol in the industrially important Corynebacterium glutamicum. In this study, we genomically integrated an Mrx1-roGFP2 biosensor in C. glutamicum to measure dynamic changes of the MSH redox potential (EMSH) during the growth and under oxidative stress. C. glutamicum maintains a highly reducing intrabacterial EMSH throughout the growth curve with basal EMSH levels of ~− 296 mV. Consistent with its H2O2 resistant phenotype, C. glutamicum responds only weakly to 40 mM H2O2, but is rapidly oxidized by low doses of NaOCl. We further monitored basal EMSH changes and the H2O2 response in various mutants which are compromised in redox-signaling of ROS (OxyR, SigH) and in the antioxidant defense (MSH, Mtr, KatA, Mpx, Tpx). While the probe was constitutively oxidized in the mshC and mtr mutants, a smaller oxidative shift in basal EMSH was observed in the sigH mutant. The catalase KatA was confirmed as major H2O2 detoxification enzyme required for fast biosensor re-equilibration upon return to non-stress conditions. In contrast, the peroxiredoxins Mpx and Tpx had only little impact on EMSH and H2O2 detoxification. Further live imaging experiments using confocal laser scanning microscopy revealed the stable biosensor expression and fluorescence at the single cell level. In conclusion, the stably expressed Mrx1-roGFP2 biosensor was successfully applied to monitor dynamic EMSH changes in C. glutamicum during the growth, under oxidative stress and in different mutants revealing the impact of Mtr and SigH for the basal level EMSH and the role of OxyR and KatA for efficient H2O2 detoxification under oxidative stress.


Introduction
The Gram-positive soil bacterium Corynebacterium glutamicum is the most important industrial platform bacterium that produces millions of tons of L-glutamate and L-lysine every year as well as other value-added products [1][2][3][4]. In addition, C. glutamicum serves as model bacterium for the related pathogens Corynebacterium diphtheriae and Corynebacterium jeikeium [5]. In its natural soil habitat and during industrial production, C. glutamicum is exposed to reactive oxygen species (ROS), such as hydrogen peroxide (H 2 O 2 ) which is generated as consequence of the aerobic lifestyle [6][7][8]. The low molecular weight (LMW) thiol mycothiol (MSH) functions as glutathione surrogate in detoxification of ROS and other thiol-reactive compounds in all actinomycetes, including C. glutamicum and mycobacteria to maintain the reduced state of the cytoplasm [9][10][11]. Thus, MSH-deficient mutants are sensitive to various thiol-reactive compounds, although the secreted histidine-derivative ergothioneine (EGT) also functions as alternative LMW thiol [12][13][14][15][16].
The enzymes for MSH biosynthesis and the Trx/TrxR systems are under control of the alternative extracytoplasmic function (ECF) sigma factor SigH which is sequestered by its cognate redox-sensitive anti sigma factor RshA in non-stressed cells [28][29][30]. RshA is oxidized under disulfide stress leading to structural changes and relief of SigH to initiate transcription of the large SigH disulfide stress regulon [16,[31][32][33]. In addition, the LysR-type transcriptional repressor OxyR plays a major role in the peroxide response in C. glutamicum which controls genes encoding antioxidant enzymes for H 2 O 2 detoxification and iron homeostasis, such as the catalase (katA), two miniferritins (dps, ftnA), the Suf machinery and ferrochelatase (hemH) [30,34]. Thus, SigH and OxyR can be regarded as main regulatory systems for the defense under disulfide and oxidative stress to maintain the redox balance in actinomycetes.
The standard thiol-redox potential of MSH was previously determined with biophysical methods as E 0′ (MSSM/MSH) of − 230 mV which is close to that of glutathione (GSH) [35]. However, Mrx1 was also recently fused to redox-sensitive green fluorescent protein (roGFP2) to construct a genetically encoded Mrx1-roGFP2 redox biosensor for dynamic measurement of E MSH changes inside mycobacterial cells. E MSH values of~-300 mV were calculated using the Mrx1-roGFP2 biosensor in mycobacteria that were much lower compared to values obtained with biophysical methods [35,36]. This Mrx1-roGFP2 biosensor was successfully applied for dynamic E MSH measurements in the pathogen Mycobacterium tuberculosis (Mtb). Using Mrx1-roGFP2, E MSH changes were studied in drug-resistant Mtb isolates, during intracellular replication and persistence in the acidic phagosomes of macrophages [36][37][38]. Mrx1-roGFP2 was also applied as tool in drug research to screen for ROS-generating anti-tuberculosis drugs or to reveal the mode of action of combination therapies based on E MSH changes [36,[39][40][41]. The Mtb population exhibited redox heterogeneity of E MSH during infection inside macrophages which was dependent on sub-vacuolar compartments and the cytoplasmic acidification controlled by WhiB3 [36,38]. Thus, application of the Mrx1-roGFP2 biosensor provided novel insights into redox changes of Mtb. However, Mrx1-roGFP2 has not been applied in the industrial platform bacterium C. glutamicum.
In this work, we designed a genetically encoded Mrx1-roGFP2 biosensor that was genomically integrated and expressed in C. glutamicum. The biosensor was successfully applied to measure dynamic E MSH changes during the growth, under oxidative stress and in various mutant backgrounds to study the impact of antioxidant systems (MSH, KatA, Mpx, Tpx) and their major regulators (OxyR, SigH) under basal and oxidative stress conditions. Our results revealed a highly reducing basal E MSH of~-296 mV that is maintained throughout the growth of C. glutamicum. H 2 O 2 stress had only little effect on E MSH changes in the wild type due to its H 2 O 2 resistance, which was dependent on the catalase KatA supporting its major role for H 2 O 2 detoxification. Confocal imaging further confirmed equal Mrx1-roGFP2 fluorescence in all cells indicating that the biosensor strain is well suited for industrial application to quantify E MSH changes in C. glutamicum at the single cell level.

Bacterial strains and growth conditions
Bacterial strains, plasmids and primers are listed in Tables S1 and S2. For cloning and genetic manipulation, Escherichia coli was cultivated in Luria Bertani (LB) medium at 37°C. The C. glutamicum ATCC13032 wild type as well as the ΔmshC, Δmtr, ΔoxyR, ΔsigH, ΔkatA, Δmpx, Δtpx and Δmpx tpx mutant strains were used in this study for expression of the Mrx1-roGFP2 biosensor which are described in Table  S1. All C. glutamicum strains were cultivated in heart infusion medium (HI; Difco) at 30°C overnight under vigorous agitation. The overnight culture was inoculated in CGC minimal medium supplemented with 1% glucose to an optical density at 500 nm (OD 500 ) of 3.0 and grown until OD 500 of 8.0 for stress exposure as described [16]. C. glutamicum mutants were cultivated in the presence of the antibiotics nalidixic acid (50 μg/ml) and kanamycin (25 μg/ml).

Construction, expression and purification of His-tagged Mrx1-roGFP2 protein in E. coli
The mrx1 gene (cg0964) was amplified from chromosomal DNA of C. glutamicum ATCC13032 by PCR using the primer pair Cgmrx1-roGFP2-NdeI-FOR and pQE60-Cgmrx1-roGFP2-SpeI-REV. The PCR product was digested with NdeI and SpeI and cloned into plasmid pET11b-brx-roGFP2 [42] to exchange the brx sequence by mrx1 with generation of plasmid pET11b-mrx1-roGFP2 (Table S1). The correct sequence was confirmed by PCR and DNA sequencing.
The E. coli BL21 (DE3) plysS expression strain containing the plasmid pET11b-mrx1-roGFP2 was grown in 1 l LB medium until OD 600 of 0.6 at 37°C, followed by induction with 1 mM IPTG (isopropyl-β-Dthiogalactopyranoside) for 16 h at 25°C. Recombinant His 6 -tagged Mrx1-roGFP2 protein was purified using His Trap™ HP Ni-NTA columns (5 ml; GE Healthcare, Chalfont St Giles, UK) and the ÄKTA purifier liquid chromatography system (GE Healthcare) according to the instructions of the manufacturer (USB). The purified protein was dialyzed against 10 mM Tris-HCl (pH 8.0), 100 mM NaCl and 30% glycerol and stored at − 80°C. Purity of the protein was analyzed after sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and Coomassie brilliant blue (CBB) staining.

Construction of katA, mtr, mpx and tpx deletion mutants in C. glutamicum
The vector pK18mobsacB was used to create marker-free deletions in C. glutamicum (1). The gene-SOEing method of Horton (2) was used to construct pK18mobsacB derivatives to perform allelic exchange of the katA and mtr genes in the chromosome of C. glutamicum ATCC13032 using the primers listed in Table S2. The constructs include the katA and mtr genes with flanking regions and internal deletions (ΔkatA [1555 bp] and Δmtr [1382 bp]). The pK18mobsacB derivatives were sub-cloned in E. coli JM109 (Table S1) and transformed into C. glutamicum ATCC13032. The pK18mobsacB::Δtpx plasmid containing the tpx flanking regions was constructed previously (3) and transformed into the C. glutamicum Δmpx mutant (3). The gene replacement in the chromosome of C. glutamicum ATCC13032 resulted in ΔkatA and Δmtr single deletion mutants and the gene replacement of tpx in the chromosome of C. glutamicum Δmpx resulted in the C. glutamicum Δmpx tpx double deletion mutant. The deletions were confirmed by PCR using the primers in Table S2.

Construction of C. glutamicum Mrx1-roGFP2 biosensor strains
For construction of the genomically integrated Mrx1-roGFP2 biosensor, a 237 bp fragment of mrx1 (cg0964) was fused to roGFP2 containing a 30-amino acid linker (GGSGG) 6 under control of the strong P tuf promoter of the C. glutamicum tuf gene encoding the translation elongation factor EF-Tu. The P tuf -Mrx1-roGFP2 fusion was codon-optimized, synthesized with flanking MunI and XhoI restriction sites and sub-cloned into PUC-SP by Bio Basic resulting in PUC-SP::P tuf -mrx1-roGFP2. For genomic integration of the biosensor into the cg1121-cg1122 intergenic region of C. glutamicum (Table S1), the vector pK18mobsacB-cg1121-cg1122 was used [43], kindly provided by Julia Frunzke, Forschungszentrum Jülich. The vector was PCR amplified with primers pk18_MunI and pk18_XhoI to swap the restrictions sites. After digestion of the pk18mobsacB-cg1121-cg1122 PCR product and the PUC-SP::P tuf -mrx1-roGFP2 plasmid with MunI and XhoI, both digestion products were ligated to obtain pK18mobsacB-cg1121-cg1121-P tuf -mrx1-roGFP2. The resulting plasmid was sequenced with biosensor_seq_-primer_1 and biosensor_seq_primer_2. Transfer of the plasmid into C. glutamicum strains (Table S1) was performed by electroporation and screening for double homologous recombination events using the conditional lethal effect of the sacB gene as described [16,43]. Correct integration of P tuf -mrx1-roGFP2 into the cg1121-cg1122 intergenic region was verified by colony PCR using 2 primer pairs (pk18_INT_Cg_-Test_rev, pk18_INT_Cg_Test_fwd and FUB_7_seq_wo_linker_fwd; FUB_8_-seq_wo_linker_rev) ( Table S2).

Characterization of recombinant Mrx1-roGFP2 biosensor in vitro
The purified Mrx1-roGFP2 protein was reduced with 10 mM dithiothreitol (DTT) for 20 min, desalted with Micro-Bio spin columns (Bio-Rad), and diluted to a final concentration of 1 µM in 100 mM potassium phosphate buffer, pH 7.0. The oxidation degree (OxD) of the biosensor was determined by calibration to fully reduced and oxidized probes which were generated by treatment of the probes with 10 mM DTT and 5 mM diamide for 5 min, respectively [42]. The thiol disulfides and oxidants were injected into the microplate wells (BD Falcon 353219) 60 s after the start of measurements. Emission was measured at 510 nm after excitation at 400 and 488 nm using the CLARIOstar microplate reader (BMG Labtech) with the Control software version 5.20 R5. Gain setting was adjusted for each excitation maximum. The data were analyzed using the MARS software version 3.10 and exported to Excel. Each in vitro measurement was performed in triplicate.

Measurements of Mrx1-roGFP2 biosensor oxidation in C. glutamicum in vivo
C. glutamicum wild type and mutant strains expressing stably integrated Mrx1-roGFP2 were grown overnight in HI medium and inoculated into CGC medium with 1% glucose to a starting OD 500 of 3.0. For stress experiments, the strains were cultivated for 8 h until they have reached an OD 500 of 14-16. Cells were harvested by centrifugation, washed twice with CGC minimal medium, adjusted to an OD 500 of 40 in CGC medium and transferred to the microplate reader. Aliquots were treated for 15 min with 10 mM DTT and 20 mM cumene hydroperoxide (CHP) for fully reduced and oxidized controls, respectively. Injection of the oxidants was performed 5 min after the start of microplate reader measurements.
For the OxD measurements along the growth curves, cells were harvested by centrifugation at different time points and washed in 100 mM potassium phosphate buffer, pH 7.0. Aliquots were treated with 20 mM CHP and 10 mM DTT for fully reduced and oxidized controls, respectively. Samples and controls were incubated with 10 mM Nethylmaleimide (NEM) to block free thiols and transferred to microplate wells. The Mrx1-roGFP2 biosensor fluorescence emission was measured at 510 nm after excitation at 400 and 488 nm using the CLARIOstar microplate reader (BMG Labtech). The OxD of biosensor was calculated for each sample and normalized to fully reduced and oxidized controls as described previously [42,44] based to the following Eq. (1).
The values of I400 sample and I488 sample are the observed fluorescence excitation intensities at 400 and 488 nm, respectively. The values of I400 red , I488 red , I400 ox and I488 ox represent the fluorescence intensities of fully reduced and oxidized controls, respectively.

Confocal laser scanning microscopy of Mrx1-roGFP2 biosensor strains
C. glutamicum wild type expressing Mrx1-roGFP2 was grown in HI medium for 48 h, exposed to 80 mM H 2 O 2 for different times and washed in potassium phosphate buffer, pH 7.0. Cells were blocked with 10 mM NEM, and imaged using a LSM 780 confocal laser-scanning microscope with a 63 × /1.4 NA Plan-Apochromat oil objective controlled by the Zen 2012 software (Carl-Zeiss, Jena, Germany). Fluorescence excitation was performed at 405 and 488 nm with laser power adjustment to 15% and 25%, respectively. For both excitation wavelengths, emission was collected between 491 and 580 nm. Fully reduced and oxidized controls were prepared with 10 mM DTT and 10 mM diamide, respectively. Images were analyzed by the Zen 2 software and Fiji/ImageJ [42,46]. Fluorescent intensities were measured after excitation at 405 and 488 nm and the images false-colored in red and green, respectively. Auto-fluorescence was recorded and subtracted. Quantification of the OxD and E MSH values was performed based on the 405/488 nm excitation ratio of mean fluorescence intensities as described [42,46]

The Mrx1-roGFP2 biosensor of C. glutamicum responds most specifically to MSSM in vitro
Previous studies have revealed a specific response of the Mrx1-roGFP2 biosensor to MSSM in vitro, which was based on a fusion of mycobacterial Mrx1 to roGFP2 [36]. Here we aimed to engineer a related Mrx1-roGFP2 biosensor for the MSH-producing industrially important bacterium C. glutamicum. Mrx1 (Cg0964) of C. glutamicum exhibits a similar redox-active CxxC motif and shares 46.8% and 42.1% sequence identity with Mrx1 homologs of M. tuberculosis H37Rv (Rv3198A) and M. smegmatis mc 2 155 (MSMEG_1947), respectively (Fig. 1AB) [27]. The principle of the Mrx1-roGFP2 biosensor to measure intrabacterial E MSH changes was shown previously [14,36]. MSSM reacts with Mrx1 to form S-mycothiolated Mrx1, followed by the transfer of the MSH moiety to roGFP2 which rearranges to the roGFP2 disulfide resulting in ratiometric changes of the 400/488 excitation ratio [14,36] (Fig. 1C).
Mrx1 of C. glutamicum was fused to roGFP2 and first purified as Histagged Mrx1-roGFP2 protein to verify the specific Mrx1-roGFP2 biosensor response to MSSM in vitro. In addition, Mrx1-roGFP2 was integrated into the genome of C. glutamicum wild type in the intergenic region between cg1121-cg1122 and placed under control of the strong P tuf promoter using the pK18mobsacB-int plasmid as constructed previously [43]. First, the Mrx1-roGFP2 biosensor response of the purified biosensor and of the stably integrated Mrx1-roGFP2 fusion were compared under fully reduced (DTT) and fully oxidized (diamide) conditions. The Mrx1-roGFP2 biosensor fluorescence excitation spectra were similar under in vitro and in vivo conditions exhibiting the same excitation maxima at 400 and 488 nm for fully reduced and oxidized probes (Fig. 1DE). Thus, the Mrx1-roGFP2 probe is well suited to monitor dynamic E MSH changes during the growth and under oxidative stress in C. glutamicum. In addition, it was verified that purified Mrx1-roGFP2 reacts very fast and most strongly to low levels of 100 µM MSSM, although weaker responses were also observed with bacillithiol disulfide (BSSB) and glutathione disulfide (GSSG) which are, however, not physiologically relevant for C. glutamium (Fig. 1F).
Furthermore, we assessed the direct response of Mrx1-roGFP2 and unfused roGFP2 to the oxidants H 2 O 2 and NaOCl to compare the sensitivities of the probes for direct oxidation (Fig. 2). This was important since a previous study showed a high sensitivity of fused Grx-roGFP2 and roGFP2-Orp1 to 10-fold molar excess of 2 µM NaOCl [47]. In our in vitro experiments, the Mrx1-roGFP2 and roGFP2 probes did not respond For fully reduced and oxidized Mrx1-roGFP2, 10 mM DTT and 5 mM diamide were used in vitro as well as 10 mM DTT and 20 mM CHP in vivo (n = 5). The fluorescence excitation spectra were monitored using the microplate reader. (F) The purified Mrx1-roGFP2 biosensor (1 µM) responds most strongly to 100 µM of MSSM, but only weakly to BSSB and GSSG in vitro (n = 3). The thiol disulfides were injected into the microplate wells 60 s after the start of the measurements of the Mrx1-roGFP2 biosensor response. The control (Co) indicates the measurement of the Mrx1-roGFP2 biosensor response without thiol-disulfides. The OxD was calculated based on the 400/488 nm excitation ratio with emission measured at 510 nm. Mean values and standard error of the mean (SEM) are shown in all graphs.
to 100 µM H 2 O 2 as in previous studies. Only 1-5 mM H 2 O 2 lead to a direct oxidation of both probes with a faster response of the Mrx1-roGFP2 fusion. Both probes were rapidly oxidized by 10-40 µM NaOCl in vitro, and again Mrx1-roGFP2 was more sensitive to thiol-oxidation by NaOCl compared to unfused roGFP2 (Fig. 2). The rapid oxidation of roGFP2 and fused roGFP2 biosensors to low levels of HOCl is in agreement with previous studies [47] and was also observed using the Brx-roGFP2 biosensor in S. aureus [42]. The higher sensitivity of fused roGFP2 biosensors (Brx-roGFP2, Mrx1-roGFP2) to NaOCl indicates that the redox active Cys residues of Brx or Mrx1 are more susceptible for thiol-oxidation compared to the thiols of roGFP2. In conclusion, our Mrx1-roGFP2 probe is highly specific to low levels of MSSM. The response of Mrx1-roGFP2 to higher levels of 1 mM H 2 O 2 in vitro are not expected to occur inside C. glutamicum cells due to its known H 2 O 2 resistance mediated by the highly efficient catalase.

The intracellular redox balance was affected in mutants with defects of MSH, Mtr and SigH
Next, we applied the genomically expressed Mrx1-roGFP2 biosensor to monitor the perturbations of basal level E MSH along the growth curve in various C. glutamicum mutant backgrounds, which had deletions of major antioxidant systems (MSH, Mtr, KatA, Tpx, Mpx) and redox-sensing regulators (OxyR, SigH) (Figs. 3 and 4). The oxidation degree was calculated in C. glutamicum wild type and mutants during the 5-12 h time points representing the log phase and transition to stationary phase in defined CGC medium. The biosensor oxidation of each C. glutamicum sample was normalized between 0 and 1 based on the fully reduced (DTT) and oxidized (CHP) controls. It is interesting to note, that C. glutamicum wild type cells maintained a highly reducing and stable E MSH of~-296 mV with little fluctuations during the log and stationary phase (Table S3). Thus, this basal level E MSH of C. glutamicum is very similar to that measured in M. smegmatis previously (E MSH of −300) [36].
In agreement with previous studies of bacillithiol (BSH)-and GSHdeficient mutants, the absence of MSH resulted in constitutive oxidation of the Mrx1-roGFP2 biosensor in the mshC mutant (Fig. 3A). This indicates an impaired redox state in the mshC mutant and the importance of MSH as major LMW thiol to maintain the redox balance in C. glutamicum (Fig. 3A). We hypothesize that increased levels of ROS may lead to constitutive biosensor oxidation in the MSH-deficient mutant since the mshC mutant had a H 2 O 2 -sensitive phenotype in previous studies [48]. The high MSH/MSSM redox balance is maintained by the NADPH-dependent mycothiol disulfide reductase Mtr which reduces MSSM back to MSH [9]. The importance of Mtr to maintain a reduced E MSH was also supported by our biosensor measurements which revealed an oxidative shift in E MSH to −280.2 mV in the mtr mutant during all growth phases (Fig. 3B, Table S3).
The alternative ECF sigma factor SigH controls a large disulfide stress regulon mainly involved in the redox homeostasis, including genes for thioredoxins and thioredoxin reductases (TrxAB), mycoredoxin-1 (Mrx1) and genes for MSH biosynthesis and recycling (MshA, Mca, Mtr) [9,28,29,32]. The C. glutamicum sigH mutant showed an increased sensitivity to ROS and NaOCl stress [16,28,29]. Mrx1-roGFP2 biosensor measurements confirmed a slightly more oxidized E MSH of − 286 mV in the sigH mutant supporting the regulatory role of SigH for the redox balance (Fig. 3C, Table S3). However, the oxidative E MSH shift was lower in the sigH mutant compared to the mtr mutant. In conclusion, our Mrx1-roGFP2 biosensor results document the important role of MSH, Mtr and SigH to maintain the redox homeostasis in C. glutamicum during the growth.
In addition to MSH, C. glutamicum encodes many antioxidant enzymes that are involved in H 2 O 2 detoxification and confer strong resistance of C. glutamicum to millimolar levels of H 2 O 2 . The H 2 O 2 scavenging systems in C. glutamicum are the major vegetative catalase (KatA) and the peroxiredoxins (Tpx, Mpx). The catalase is highly efficient for detoxification at high H 2 O 2 levels while Tpx and Mpx are more involved in reduction of physiological low levels of H 2 O 2 generated during the aerobic growth [49]. In C. glutamicum, expression of katA is induced by H 2 O 2 and controlled by the redox-sensing OxyR repressor which is inactivated under H 2 O 2 stress [34]. Thus, the oxyR mutant exhibits increased H 2 O 2 resistance due to constitutive derepression of katA [34]. Here, we were interested in the contribution of OxyR, and the antioxidant enzymes KatA, Tpx and Mpx to maintain the reduced basal level E MSH in C. glutamicum. In all mutants with deletions of oxyR, katA, tpx and mpx, the basal level of E MSH was still highly reducing and comparable to the wild type during different growth phases (Fig. 3D,  Fig. 4A-D, Table S3). Thus, we can conclude that the major antioxidant enzymes for H 2 O 2 detoxification (KatA, Mpx and Tpx) do not contribute to the reduced basal E MSH level in C. glutamicum during aerobic growth. These results further point to the main roles of these H 2 O 2 scavenging systems under conditions of oxidative stress to recover the reduced state of E MSH which was investigated in the next section.

Mrx1-roGFP2 biosensor responses in C. glutamicum under oxidative stress in vivo
Next, we were interested to determine the kinetics of Mrx1-roGFP2 biosensor oxidation in C. glutamicum under H 2 O 2 and NaOCl stress and the recovery of reduced E MSH . C. glutamicum can survive even 100 mM H 2 O 2 without killing effect which depends on the very efficient catalase KatA [34]. In accordance with the H 2 O 2 resistant phenotype, the Mrx1-roGFP2 biosensor did not respond to 10 mM H 2 O 2 in C. glutamicum wild type cells and was only weakly oxidized by 40 mM H 2 O 2 (Fig. 5A). C. glutamicum cells were able to recover the reduced E MSH within 40-60 min after H 2 O 2 treatment. Importantly, even 100 mM H 2 O 2 did not further enhance the biosensor oxidation degree, indicating highly efficient antioxidant systems (data not shown).
In contrast, C. glutamicum was more sensitive to sub-lethal doses of NaOCl stress and showed a moderate biosensor oxidation by 0.5-1 mM NaOCl, while 1.5 mM NaOCl resulted in the fully oxidation of the probe. Moreover, cells were unable to regenerate the reduced basal level of E MSH within 80 min after NaOCl exposure, which could be only restored with 10 mM DTT (Fig. 5B). Since H 2 O 2 is the more physiological oxidant in C. glutamicum, we studied the biosensor response under 40 mM H 2 O 2 stress in the various mutants deficient for MSH and Mtr, antioxidant enzymes (KatA, Mpx, Tpx) and redox regulators (SigH, OxyR). The sigH mutant showed an increased basal level of E MSH of~-286 mV as noted earlier (Fig. 3C), but a similar oxidation increase with 40 mM H 2 O 2 and recovery of the reduced state after 40 min compared to the wild type (Fig. 6A). The similar kinetics of biosensor oxidation and regeneration in wild type and  sigH mutant cells may indicate that MSH is not directly involved in H 2 O 2 detoxification. In contrast, the oxyR mutant showed a lower H 2 O 2 response than the wild type, but required the same time of 40 min for recovery of the reduced state of E MSH (Fig. 6B). The derepression of katA in the oxyR mutant is most likely responsible for the lower biosensor oxidation under H 2 O 2 stress [34,50]. This hypothesis was supported by the very fast response of katA mutant cells to 40 mM H 2 O 2 stress, resulting in fully oxidation of the biosensor due to the lack of H 2 O 2 detoxification in the absence of KatA (Fig. 6C). Exposure of katA mutant cells to 40 mM H 2 O 2 might cause enhanced oxidation of MSH to MSSM leading to full biosensor oxidation with no recovery of the reduced state. In contrast, kinetic biosensor measurements under H 2 O 2 stress revealed only slightly increased oxidation in the tpx mutant while the mpx mutant showed the same oxidation increase like the wild type (Fig. 6DE). However, the H 2 O 2 response of the mpx tpx mutant was similar compared to the wild type, indicating that Tpx and Mpx do not contribute significantly to H 2 O 2 detoxification during exposure to high levels of 40 mM H 2 O 2 stress, while KatA plays the major role (Fig. 6F). The small oxidation increase in the tpx mutant might indicate additional roles of Tpx for detoxification of low levels of H 2 O 2 as found in previous studies [51]. Altogether, our studies on the kinetics of the Mrx1-roGFP2 biosensor response under H 2 O 2 stress support that KatA plays the most important role in H 2 O 2 detoxification in C. glutamicum.
To correlate increased biosensor responses under H 2 O 2 stress to peroxide sensitive phenotypes, we compared the growth of the wild type and mutants after exposure to 80 mM H 2 O 2 (Fig. 7). Exposure of the wild type to 80 mM H 2 O 2 did not significantly affect the growth rate indicating the high level of H 2 O 2 resistance in C. glutamicum. Of all mutants, only the katA mutant was significantly impaired in growth under non-stress conditions and lysed after exposure to 80 mM H 2 O 2 (Fig. 7C). In contrast, deletions of sigH, oxyR, tpx and mpx did not significantly affect the growth under control and H 2 O 2 stress conditions (Fig. 7AB, DE). However, we observed a slightly decreased growth rate of the mpx tpx mutant in response to 80 mM H 2 O 2 stress supporting the residual contribution of thiol-dependent peroxiredoxins in the peroxide stress response (Fig. 7F). Overall, the growth curves are in agreement with the biosensor measurements indicating the major role of KatA for detoxification of high levels of H 2 O 2 and the recovery of cells from oxidative stress.

Single cell measurements of E MSH changes under H 2 O 2 stress using confocal imaging
To verify the biosensor response under H 2 O 2 stress in C. glutamicum at the single cell level, we quantified the 405/488 nm fluorescence excitation ratio in C. glutamicum cells expressing stably integrated Mrx1-roGFP2 using confocal laser scanning microscopy (CLSM) (Fig. 8A). For control, we used fully reduced and oxidized C. glutamicum cells treated with DTT and diamide, respectively. In the confocal microscope, most cells exhibited similar fluorescence intensities at the 405 and 488 nm excitation maxima, respectively, indicating that the Mrx1-roGFP2 biosensor was equally expressed in 99% of cells. Fully reduced and untreated C. glutamicum control cells exhibited a bright fluorescence intensity at the 488 nm excitation maximum which was false- colored in green, while the 405 nm excitation maximum was low and false-colored in red (Fig. 8A). In agreement with the microplate reader results, the basal E MSH was highly reducing and calculated as −307 mV for the single cell population (Fig. 8B, Table S4). Treatment of cells with 80 mM H 2 O 2 for 20 min resulted in a decreased fluorescence intensity at the 488 nm excitation maximum and a slightly increased signal at the 405 nm excitation maximum, causing an oxidative shift of E MSH . Specifically, the E MSH of control cells was increased to −263 mV after 20 min H 2 O 2 treatment. The recovery phase could be also monitored at the single cell level after 40 and 60 min of H 2 O 2 stress, as revealed by the regeneration of reduced E MSH of −271 mV and −293 mV, respectively (Fig. 8B, Table S4). The oxidative E MSH shift after H 2 O 2 treatment and the recovery of reduced E MSH were comparable between the microplate reader measurements and confocal imaging (Fig. 8B). This confirms the reliability of biosensor measurements at both single cell level and for a greater cell population using the microplate reader.

Discussion
Here, we have successfully designed the first genome-integrated Mrx1-roGFP2 biosensor that was applied in the industrial platform bacterium C. glutamicum which is of high biotechnological importance. During aerobic respiration and under industrial production processes, C. glutamicum is frequently exposed to ROS, such as H 2 O 2 . Thus, C. glutamicum is equipped with several antioxidant systems, including MSH and the enzymatic ROS-scavengers KatA, Mpx and Tpx. Moreover, Mpx and Tpx are dependent on the MSH cofactor required for recycling during recovery from oxidative stress [16,21,22]. The kinetics of H 2 O 2 detoxification has been studied for catalases and peroxiredoxins in many different bacteria. However, the roles of many H 2 O 2 detoxification enzymes are unknown and many seem to be redundant and not essential [49]. There is also a knowledge gap to which extent the H 2 O 2 detoxification enzymes contribute to the reduced redox balance under aerobic growth conditions and under oxidative stress.
Thus, we applied this stably integrated Mrx1-roGFP2 biosensor to measure dynamic E MSH changes to study the impact of antioxidant systems (MSH, KatA, Mpx, Tpx) and their major regulators (OxyR, SigH) under basal conditions and ROS exposure. The basal E MSH was highly reducing with~-296 mV during the exponential growth and stationary phase in C. glutamicum wild type, but maintained reduced also in the katA, mpx and tpx mutants. In contrast, the probe was strongly oxidized in mshC and mtr mutants indicating the major role of MSH for the overall redox homeostasis under aerobic growth conditions. While the enzymatic ROS scavengers KatA, Mpx and Tpx did not contribute to the reduced basal level of E MSH during the growth, the catalase KatA was essential for efficient H 2 O 2 detoxification and the recovery of the reduced E MSH under H 2 O 2 stress. In contrast, both MSHdependent peroxiredoxins Tpx and Mpx did not play a significant role in the H 2 O 2 defense and recovery from stress, which was evident in the tpx mpx double mutant. These results were supported by growth phenotype analyses, revealing the strongest H 2 O 2 -sensitive growth phenotype for the katA mutant, while the growth of the mpx tpx double mutant was only slightly affected under H 2 O 2 stress. These biosensor and phenotype results clearly support the major role of the catalase KatA for H 2 O 2 detoxification.
Since expression of katA is controlled by the OxyR repressor, we observed even a lower H 2 O 2 response of the oxyR mutant, due to the constitutive derepression of katA as determined previously [34]. In contrast, the sigH mutant showed an enhanced basal E MSH during aerobic growth, since SigH controls enzymes for MSH biosynthesis and recycling (MshA, Mca, Mtr) which contribute to reduced E MSH [29,32]. However, the sigH mutant was not impaired in its H 2 O 2 response of Mrx1-roGFP2, since H 2 O 2 detoxification is the role of KatA. Thus, we have identified unique roles of SigH and Mtr to control the basal E MSH level, while OxyR and KatA play the major role in the recovery of reduced E MSH under oxidative stress.
In previous work, the kinetics for H 2 O 2 detoxification by catalases and peroxiredoxins was been measured using the unfused roGFP2 biosensor in the Gram-negative bacterium Salmonella Typhimurium [52]. The deletion of catalases affected the detoxification efficiency of H 2 O 2 strongly, while mutations in peroxidases (ahpCF, tsaA) had only a minor effect on the H 2 O 2 detoxifying power. These results are consistent with our data and previous results in E. coli, which showed that catalases are the main H 2 O 2 scavenging enzymes at higher H 2 O 2 concentrations, while peroxidases are more efficient at lower H 2 O 2 doses [53]. The reason for the lower efficiency of H 2 O 2 detoxification by peroxidases might be due to low NAD(P)H levels under oxidative stress that are not sufficient for recycling of oxidized peroxidases under high H 2 O 2 levels [53]. Overall, these data are in agreement with our Mrx1-roGFP2 measurements in the katA, tpx and mpx mutants in C. glutamicum. However, C. glutamicum differs from E. coli by its strong level of H 2 O 2 resistance since C. glutamicum is able to grow with 100 mM H 2 O 2 and the biosensor did not respond to 10 mM H 2 O 2. In contrast, 1-5 mM H 2 O 2 resulted in a maximal roGFP2 biosensor response with different detoxification kinetics in E. coli [52]. Since the high H 2 O 2 resistance and detoxification power was attributed to the catalases, it will be interesting to analyze the differences between activities and structures of the catalases of C. glutamicum and E. coli. Of note, due to its remarkable high catalase activity, KatA of C. glutamicum is even commercially applied at Merck (CAS Number 9001-05-2). However, the structural features of KatA that are responsible for its high catalase activity are unknown.
While our biosensor results confirmed the strong H 2 O 2 detoxification power of the catalase KatA [51], the roles of the peroxiredoxins Mpx and Tpx for H 2 O 2 detoxification are less clear in C. glutamicum. Both Tpx and Mpx were previously identified as S-mycothiolated proteins in the proteome of NaOCl-exposed C. glutamicum cells [16]. Smycothiolation inhibited Tpx and Mpx activities during H 2 O 2 detoxification in vitro, which could be restored by the Trx and Mrx1 pathways [16,21,22]. Moreover, Tpx displayed a gradual response to increasing H 2 O 2 levels and was active as Trx-dependent peroxiredoxin to detoxify low doses H 2 O 2 while high levels H 2 O 2 resulted in overoxidation of Tpx [51]. Overoxidation of Tpx caused oligomerization to activate the chaperone function of Tpx. Since mpx and katA are both induced under H 2 O 2 stress, they were suggested to compensate for the inactivation of Tpx for detoxification of high doses of H 2 O 2. Previous analyses showed that the katA and mpx mutants are more sensitive to 100-150 mM H 2 O 2 [21,22]. In our analyses, the mpx mutant was not more sensitive to 80 mM H 2 O 2 and displayed the same H 2 O 2 response like the wild type, while the katA mutant showed a strong H 2 O 2 sensitivity and responded strongly to H 2 O 2 in the biosensor measurements. Thus, our biosensor and phenotype results clearly support the major role of KatA in detoxification of high doses H 2 O 2 in vivo.
Finally, we confirmed using confocal imaging further that the genomically expressed Mrx1-roGFP2 biosensor shows equal fluorescence in the majority of cells indicating that the biosensor strain is suited for industrial application to quantify E MSH changes in C. glutamicum at the single cell level or under production processes. Previous Mrx1-roGFP2 biosensor applications involved plasmid-based systems which can result in different fluorescence intensities within the cellular population due to different copy numbers. Moreover, plasmids can be lost under long term experiments when the selection pressure is decreased due to degradation or inactivation of the antibiotics.
We also compared the fluorescence intensities of the plasmid-based expression of Mrx1-roGFP2 using the IPTG-inducible pEKEx2 plasmid with the stably integrated Mrx1-roGFP2 strain in this work (Fig. S1). Using confocal imaging, the plasmid-based Mrx1-roGFP2 biosensor strain showed only roGFP2 fluorescence in < 20% of cells, while the genomically expressed biosensor was equally expressed and fluorescent in 99% of cells. The integration of the Mrx1-roGFP2 biosensor was performed into the cg1121-1122 intergenic region and the biosensor was expressed from the strong P tuf promoter using the pK18mobsacB construct designed previously for an Lrp-biosensor to measure L-valine production [54]. Previous live cell imaging using microfluidic chips revealed that only 1% of cells with the Lrp-biosensor were non-fluorescent due to cell lysis or dormancy [54]. Thus, expression of roGFP2 fusions from strong constitutive promoters should circumvent the problem of low roGFP2 fluorescence intensity after genomic integration. The advantage and utility of a stably integrated Grx1-roGFP2 biosensor has been also recently demonstrated in the malaria parasite Plasmodium falciparum which can circumvent low transfection frequency of plasmidbased roGFP2 fusions [55]. Moreover, quantifications using the microplate reader are more reliable, less time-consuming and reproducible with strains expressing genomic biosensors compared to measurements using confocal microscopy [55]. Thus, stably integrated redox biosensors should be the method of the choice for future applications of roGFP2 fusions to monitor redox changes in a greater cellular population.
In conclusion, in this study we designed a novel Mrx1-roGFP2 biosensor to monitor dynamic E MSH changes in C. glutamicum during the growth, under oxidative stress and in mutants with defects in redoxsignaling and H 2 O 2 detoxification. This probe revealed the impact of Mtr and SigH to maintain highly reducing E MSH throughout the growth and the main role of KatA and OxyR for efficient H 2 O 2 detoxification and the regeneration of the redox balance. This probe is now available for application in engineered production strains to monitor the impact of industrial production of amino acids on the cellular redox state. In addition, the effect of genome-wide mutations on E MSH changes can be followed in C. glutamicum in real-time during the growth, under oxidative stress and at the single cell level.