Corynebacterium glutamicum was routinely cultivated at 30°C and under shaking conditions (130 rpm). For growth experiments, C. glutamicum strains were cultivated overnight (O/N) in 2xTY medium. Bacteria of these O/N cultures were washed with sterile 0.9% (w/v) NaCl and used to inoculate a modified CGXII minimal medium without urea containing glucose at 2% (w/v) as carbon source (CGXII-U) (Desiderato et al., 2022) to an optical density (OD₆₀₀) of ∼2. For optimization of recombinant GarQ production, 0.5 g L^–1 CaCl₂ and/or 0.5% (v/v) Tween 80 were supplemented as indicated. These media will be referred to as CGXII-U+C (CGXII-U with CaCl₂), CGXII-U+T (CGXII-U with Tween 80), and CGXII-U+CT (CGXII-U with CaCl₂ and Tween 80). Lactococcus lactis IL1403/pNZ-pHin2^Lm was cultivated in M17 medium with 0.5% (w/v) glucose at 30°C with aeration on an orbital shaker. Listeria monocytogenes EGDe/pNZ-pHin2^Lm was cultivated in BHI medium at 37°C with aeration on an orbital shaker. For GarQ production by the native producer, L. petauri B1726 was cultivated statically in M17 medium supplemented with 2% (w/v) glucose at 30°C overnight in Schott glass bottles. Media were supplemented with 25 μg kanamycin ml^–1 (C. glutamicum), 10 or 7.5 μg ml^–1 chloramphenicol (L. lactis or C. glutamicum) and/or 0.2 mM IPTG (C. glutamicum) as appropriate. All strains are listed in Table 1.

Cloning procedures were performed using standard reagents following instructions of the manufacturers. Primers were purchased from a commercial service provider (Eurofins Genomics, Germany) and sequences are listed in Supplementary Table 1. For construction of the Sec-SP library (plasmids pCMEx8-SP_XYZ-garQ), an automated protocol for plasmid construction using an OT-2 robot (Opentrons, USA) was applied as previously described (Müller et al., 2022). Briefly, the garQ gene lacking the nucleotide sequence for its native SP was amplified from pPBEx2-garQICD^Cgl and inserted into pCMEx8-[SP_XYZ] by Golden Gate Assembly using the type II restriction enzyme BsaI (New England Biolabs). Restriction digestion and capillary electrophoresis were used to check for positive clones. For construction of pXMJ19-SP_ywaD-garQ-v1, the gene coding for a fusion peptide consisting of the Sec-dependent SP of YwaD of B. subtilis and GarQ was amplified together with its ribosomal binding site from pCMEx8-SP_ywa-garQ and new restriction sites were introduced by primers with appropriate adapters containing PstI and BamHI restriction sites. The PCR fragment was digested with these enzymes and ligated into linearized pXMJ19. For construction of pXMJ19-SP_ywaD-garQ-v2, SP_ywaD and garQ were separately amplified from pXMJ19-SP_ywaD-garQ-v1 lacking the codons for the two amino acid linker between SP and GarQ. The two fragments were fused by overlapping extension PCR and sub-cloned into pJET1.2 (ThermoFisher, Germany) by blunt-end ligation. Then SP_ywaD-garQ-v2 was excised from the sub-cloning vector using restriction by XbaI and PstI and ligated into linearized pXMJ19. For construction of pXMJ19-SP_ywaD-garQ^M5F, a SP_ywaD-garQ^M5F gene with the mutations for the M5F amino acid exchange was synthesized by a commercial service provider (Eurofins Genomics) and inserted into XbaI/PstI-linearized pXMJ19. All constructs were verified by Sanger sequencing by a commercial service provider (Microsynth Seqlab, Germany).

The pHluorin2 assay was performed according to a previously published protocol using reporter strains L. lactis IL1403/pNZ-pHin2^Lm or L. monocytogenes EGDe/pNZ-pHin2^Lm (Desiderato et al., 2022; Reich et al., 2022) and Listeria minimal buffer (LMB) with slight modifications. Instead of 100 mM 3-(N-morpholino)propanesulfonic acid (MOPS) as a buffering agent, 200 mM 2-(N-morpholino)ethanesulfonic acid were used. The assay is based on the two distinct excitation peaks of pHluorin2 at 400 and 480 nm. Fluorescence intensity (FI) at these two excitation peaks shifts in response to changes in pH in a ratiometric manner: with increasing pH, FI after excitation at 400 nm increases and FI after excitation at 480 nm decreases. Under steady-state conditions, biosensor bacteria expressing pHluorin2 are able to maintain an intracellular pH (pH_i) of 7.6–8.0 even in LMB with a pH of 6.2. However, when exposed to compounds that disrupt membrane integrity, pH_i will rapidly change to the pH of LMB, and this shift can be detected by the changes in FIs after excitation at 400 and 480 nm.

For quantification of GarQ activities, standards with defined concentration (range 0.03–2.0 μg ml^–1) of chemically synthesized peptides were used to establish a calibration curve by plotting ratios of pHluorin2 fluorescence intensities at 520 nm after excitation at 400 and 480 nm (ratio RFU 400/480) against peptide concentration. The equation of the calibration curve was calculated as a nonlinear regression using a semilogarithmic fit with least square fitting. The coefficient of determination R² was > 0.96 in all cases. These calibration curves were used to calculate mass concentration of GarQ in unknown samples based on measurements of ratio RFU 400/480. Statistical analysis was performed by ANOVA with the Bonferroni post-test to calculate p-values adjusted for multiple comparisons. GraphPad Prism 6 software was sued to compute nonlinear regression and ANOVA. Chemically synthesized GarQ was obtained from the Functional Peptidomics core facility of Ulm University. Dried, synthetic GarQ had a purity of > 95% and was dissolved in HPLC water for further use.

GarQ was purified from culture supernatants of C. glutamicum by sequential hydrophobic interaction chromatography (HIC) and reverse-phase chromatography (RPC) as previously described (Desiderato et al., 2022). After HIC, GarQ was eluted with HPLC-grade water. Elution buffer in RPC was 80% acetonitrile and 0.05% TFA in water and GarQ eluted at 30% elution buffer.

Mass spectrometry analysis of purified GarQ peptide fractions was performed based on a previously described method (Desiderato et al., 2022). Lyophilized peptide fractions were concentrated to 1 mg ml^–1. Aliquots of peptide solutions were subjected to tryptic digestion, with 1 μg trypsin in a total volume of 100 μl for 5 h at 42°C as recommended by the supplier. Solutions of intact and digested peptides were diluted 1:2 with LC-MS-grade H₂O prior to Liquid chromatography mass spectrometry (LC-MS).

Liquid chromatography mass spectrometry was conducted with an Agilent 1260 Infinity system (Agilent Technologies, Waldbronn, Germany) coupled to a quadrupole time-of-flight mass spectrometer (TripleTOF6600, AB Sciex, Darmstadt, Germany). LC was performed with an Ascentis^® Express Peptide ES-C18, 2.7 μm HPLC column (53307-U, Merck, Darmstadt, Germany) with a flow rate of 200 μL min^–1 and the mobile phases (A) 0.1% formic acid in water and (B) acetonitrile. The elution gradient was as follows: 0 min, 3% B; 70 min, 40% B; 78 min 40% B, 79 min 60% B, 89 min 60% B, 90 min 3% B followed by a 12 min equilibration time between injections. Column temperature was set to 21 °C and injection volume to 10 μL. MS was conducted with a TurboV ion source operated in positive ionization mode. Ion spray voltage was set to 5.5 kV, source temperature to 450 °C, curtain gas to 35 psi, and the support gases GS1/GS2 to 50 psi/50 psi. All gases were nitrogen.

For analysis of undigested peptide, the quadrupole time-of-flight (QToF) mass spectrometer was operated in ToF mode with a dwell time of 250 ms and for digested peptide, the QToF mass spectrometer was operated in data-dependent acquisition (DDA) mode. Based on a ToF survey scan with a dwell time of 250 ms, product ion scans with a dwell-time of 100 ms were automatically performed for a maximum of 40 ions with a mass-to-charge ratio > 300 m/z and a charge of 2 to 4 passing an intensity threshold of 150 cps. After acquisition, mass-to-charge ratios were excluded from the potential candidate-ion list for 12 s. Declustering potential was set to 120 V, collision energy spread to 5 V and mass tolerance to 25 ppm.

Acquired mass spectra were analyzed with PeakView 2.1 (AB Sciex, Darmstadt, Germany). Protein identification was performed with the ProteinPilot 5.1 software (AB Sciex, Darmstadt, Germany).

The adsorption assays were performed as previously described (Desiderato et al., 2022) with slight modifications. Overnight cultures of C. glutamicum were washed three times with phosphate-buffered saline. The universal buffer UB3 (20 mM HEPES, 20 mM Bis-Tris and 20 mM sodium acetate) was used to allow adsorption studies at different pH values without changing the buffer composition (Brooke et al., 2015). Bacteria were resuspended at an OD₆₀₀ of 3 in UB3 buffer set to different pH containing HIC purified GarQ. After incubation of the peptide with bacteria for 1 h at 30°C, bacteria were pelleted by centrifugation (11,000 × g, 5 min, room temperature) and the adsorption supernatant (ASN) was collected. In parallel, GarQ was incubated at the same conditions and buffer composition without bacteria as a control. Antimicrobial activity of two-fold serial dilution of ASN and control samples was then determined in growth inhibition assay using L. lactis IL1403 as indicator strain as described previously (Desiderato et al., 2022). To calculate bacteriocin activity, a nonlinear fit based on the Gompertz function was calculated using GraphPad Prism 6 software. The coefficient of determination R² of the Gompertz function was above 0.97 in all cases, indicating that the function is suitable. Arbitrary units (AU) of bacteriocin activity of sample was defined as the X-value of the inflection point (1/K) of the Gompertz function. The percentage of adsorbed GarQ was calculated according to following formula:

Corynebacterium glutamicum strains were cultivated in CGXII-U or normal CGXII minimal medium (with urea). Samples were taken at the end of exponential phase and in stationary phase. Cells were diluted 1:100 in PBS and 50 μl of cell suspension was analyzed for fluorescence intensity of the eYFP reporter (FI eYFP) of C. glutamicum K9 strains in an Amnis^® CellStream^® device (Luminex) equipped with a 488 nm laser (emission wavelength 528 nm). The flow speed was set to “slow.” The laser powers were: 15% for forward scatter (FSC) and side scatter (SSC) and 40% for the 488 nm laser (excitation of eYFP). Bacteria were gated based on FSC and SSC and FI eYFP was recorded for at least 5,000 bacteria per sample. Flow cytometry data was analyzed using the FlowJo analysis software (v10).

For proteomic analysis, C. glutamicum ATCC13032 was cultivated in a four-fold parallel bioreactor setup (DASGIP^® Parallel Bioreactor System, Eppendorf SE, Germany) with a filling volume of 1 L each. Bacteria were grown in CGXII-U medium without MOPS and pH was maintained at defined levels (pH = 5 or 8) using 4 M KOH or 6 M H₃PO₄. Ample oxygen supply was guaranteed by using an oxygen controller set to 30% dissolved oxygen and using a cascade of stirrer speed, gas flow and oxygen concentration in inlet air. The bioreactors were inoculated to an OD₆₀₀ of 1.5. Samples for proteome analysis were taken in exponential and stationary growth phase in technical triplicates. For each replicate, 1.6 ml culture broth were samples to a proteome vial and centrifuged for 5 min at 4°C and 21,500 × g. The supernatant was discarded, and pellets were frozen in liquid nitrogen and stored at −80°C. Proteomic analysis was carried out as described elsewhere (Unrean et al., 2018). The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium via the PRIDE (Perez-Riverol et al., 2022) partner repository with the dataset identifier PXD046499.

To establish Sec-dependent secretion of GarQ, an existing library of known Sec-dependent SPs of Bacillus subtilis was used to replace the native double-glycine-type signal sequence by automized plasmid construction (Müller et al., 2022). This yielded constructs of 23 different N-terminal SPs fused to the GarQ-coding sequence. Following transformation into C. glutamicum, all strains were cultivated in CGXII-U medium and GarQ activity was determined in culture supernatants after 24 h of cultivation by pHluorin2 assay (Figure 1A). A significant response of the sensor L. lactis IL1403/pNZ-pHin2^Lm was detected with supernatants of a total of 10 strains of the library. Highest activity and thus most efficient secretion was observed with the SP of the ywaD gene (SP_ywaD) and the corresponding SP_ywaD-garQ fusion was selected for further experiments.

As a remnant of the cloning procedure, GarQ derivatives expressed using the constructs of the SP library contain a linker of two amino acid residues (glutamate and phenylalanine) at the N-terminus of GarQ following cleavage by the signal peptidase. To obtain GarQ by Sec-dependent secretion without these two amino acids, we generated a second version of our construct. Both strains, C. glutamicum CR099/pXMJ19-SP_ywaD-garQ-v1 (with linker) and C. glutamicum CR099/pXMJ19-SP_ywaD-garQ-v2 (without linker), showed comparable growth characteristics and had comparable levels of GarQ activity in their supernatants (Supplementary Figure 1). This indicates that presence/absence of the linker had no effect on activity and/or secretion. To ensure that GarQ is secreted and processed properly by the Sec translocon, the active compound was purified from culture supernatants of C. glutamicum CR099/pXMJ19-SP_ywaD-garQ-v2 by sequential HIC and RPC (Figures 1B, C). Three distinct peaks with mass/charge ratios (m/z) of 1069.1316, 891.1182, or 763.9558, respectively, were identified by LC-MS of pooled RPC fractions containing antimicrobial activity (Figures 1D, E). These peaks correspond to m/z of the five-, six and seven-fold protonated ions of GarQ and are in good agreement with the calculated monoisotopic mass of natural GarQ (5340.63 Da). Trypsin digestion and subsequent mass spectrometry analysis further confirmed the presence and correct processing of GarQ in culture supernatants of the recombinant C. glutamicum strain (data not shown).

For better estimation of mass concentrations of GarQ, we established calibration curves for quantification of GarQ by the pHluorin2 assay using standards of chemically synthesized GarQ of known concentration serially diluted in CGXII-U medium (Supplementary Figure 2). Using this approach, we determined that C. glutamicum CR099/pXMJ19-SP_ywaD-garQ-v2 produces 7.2 ± 1.5 mg L^–1 GarQ after cultivation in CGXII-U medium for 24 h. This is about 3.5-fold higher than levels observed with the previously published GarQ production strain C. glutamicum CR099/pPBEx2-garQICD^Cgl (Desiderato et al., 2022) under the same conditions (2.0 ± 0.3 mg L^–1). However, it is still by far lower than GarQ levels in supernatants of the native producer L. petauri B1726 grown in GM17 for 24 h (198.6 ± 5.6 mg L^–1).

Production of bacteriocins by their native producers is affected by pH and generally LAB produce higher titers in acidic conditions (Abbasiliasi et al., 2017). Likewise, we previously reported that acidic pH enhances recombinant production of pediocin PA-1 and GarQ by C. glutamicum (Desiderato et al., 2022; Christmann et al., 2023). Also, GarQ is strongly adsorbed by C. glutamicum and other bacterial cells (Desiderato et al., 2022) at neutral pH and we have made similar observations for other bacteriocins including nisin Z and pediocin PA-1 (data not shown). To better understand the adsorption of GarQ by C. glutamicum cells, we tested the impact of different pH values (4, 5, 6, 7, and 8) on adsorption (Table 2; Supplementary Figure 3).

In general, we observed a decrease in adsorption at lower pH. At pH = 6–8, about 90% of the peptide adsorbed to the cells. At pH = 5 adsorption was reduced to 66% and at pH = 4 only 18% of the peptide were adsorbed by bacteria. Concomitantly, we also observed that activity of GarQ was reduced at pH = 7–8.

To reduce adsorption of GarQ to biomass of C. glutamicum and improve release into the supernatant, we sought to modify cultivation conditions. In previous studies, CaCl₂ was shown to improve the heterologous production of secreted proteins with C. glutamicum (Teramoto et al., 2011; Freier et al., 2016), to improve resistance of C. glutamicum to nisin, and to increase recombinant pediocin PA-1 production (Weixler et al., 2022; Christmann et al., 2023). Tween 80 is a non-ionic detergent that is used as a solubilizer for proteins and as emulsifying agent in food (Linke, 2009; Duquesne and Sturgis, 2010). Thus, we tested the effects of CaCl₂ and Tween 80 on adsorption of GarQ to C. glutamicum (Table 3; Supplementary Figure 4).

The effects of Tween 80 and CaCl₂ individually were rather moderate. Supplementation with 0.5–2 g L^–1 CaCl₂ reduced adsorption to ∼80% and the addition of 0.5% (w/v) Tween 80 reduced adsorption to 64%. Interestingly, a combination of 0.5 g L^–1 CaCl₂ and 0.5% (w/v) Tween 80 strongly reduced adsorption of GarQ to 11%, suggesting a synergistic effect.

Based on these results, we cultivated C. glutamicum CR099/pXMJ19-ywaDSPgarQ-v2 in CGXII-U, CGXII-U+C (containing 0.5 g L^–1 CaCl₂), CGXII-U+T [with 0.5% (w/v) Tween 80], or CGXII-U+CT (containing both supplements) and compared GarQ levels in supernatants (Figure 2). Supplementation with CaCl₂ alone increased GarQ titers at all timepoints analyzed and a maximum of 15.1 ± 0.4 mg L^–1 GarQ was observed after 24 h. When only Tween 80 was supplemented, the GarQ production in exponential phase was comparable to that CGXII-U without supplements (2.6 ± 0.4 mg L^–1) at t = 4 h but decreased in stationary phase. In CGXII-U+CT, we observed increased GarQ levels after 4 h of cultivation (13.3 ± 0.7 mg L^–1) but the activity decreased drastically to 2.0 ± 0.5 mg L^–1 after 24 h.

Corynebacterium glutamicum is generally believed to have little to no extracellular protease activity (Vertès, 2013). However, the dramatic decrease in GarQ activity in supernatants of C. glutamicum CR099/pXMJ19-ywaDSPgarQ-v2 grown in CGXII-U+CT at later timepoints of the cultivation prompted us to reconsider proteolytic degradation as a potential problem under these conditions. For other bacteria, it is described that high level heterologous expression of secreted proteins and presence of bacteriocins induces cell envelop/secretion stress and expression of proteases for cleavage of unfolded or misfolded proteins (Jordan et al., 2008; Neef et al., 2020). A protease previously associated with Sec-dependent secretion stress in C. glutamicum is HtrA (Jurischka et al., 2020). This was reported for C. glutamicum K9, a biosensor that reports secretion stress by expression of the yellow fluorescent protein eYFP from a gene cassette replacing the chromosomal htrA gene. To investigate whether HtrA may play a role in reducing GarQ levels, we performed similar experiments under the conditions of GarQ production. C. glutamicum K9 was transformed with pXMJ19-SP_ywaD–garQ-v2 and the obtained strain as well as its two parental strains K9 and ATCC13032 were grown in standard CGXII or CGXII-U, and eYFP fluorescence was analyzed by flow cytometry (Figure 3).

In normal CGXII medium, pH was slightly increased to 8.5 in stationary growth phase and the two K9 strains showed only marginally increased eYFP fluorescence (approx. 2-fold higher mean fluorescence intensity; MFI) compared to the negative control, i.e., wildtype strain ATCC13032. As observed previously, pH in cultures grown in CGXII-U dropped markedly to 5.6 in exponential and to 4.5 in stationary growth phase. At the same time, eYFP expression of the two K9 strains was markedly increased. Compared to the wildtype strain ATCC13032, MFI of C. glutamicum K9 was increased approx. 10-fold and for C. glutamicum K9/pXMJ19-SP_ywaD-garQ-v2 a 7-fold increase in eYFP fluorescence was observed. Although for a subpopulation of these strains eYFP fluorescence returned to baseline in stationary growth phase, still a considerable fraction remains eYFP positive. No clear difference was observed between C. glutamicum K9 and K9/pXMJ19-SP_ywaD–garQ-v2. These results indicate that HtrA expression is increased in acidic environments but is not affected by expression/secretion of GarQ.

Additionally, LC-MS proteome analysis of C. glutamicum ATCC13032 cultivated in bioreactors in CGXII-U without MOPS at defined pH values confirmed that HtrA levels are increased at acidic extracellular pH. At pH = 5, signals for HtrA protein were increased 7.6-fold (p = 2.09E-08) and 2.4-fold (p = 0.01) in exponential and stationary growth phase, respectively, compared to pH = 8.

Based on these results, we hypothesized that increased expression of htrA under the conditions of GarQ production may cause proteolytic degradation of the product and result in decreased activity in supernatants at later timepoints of the fermentation. To test this hypothesis, we analyzed GarQ production by C. glutamicum K9/pXMJ19-SP_ywaD–garQ-v2 lacking a functional htrA gene (Figure 4). When grown in CGXII-U+CT, GarQ titers of 38.6 ± 3.3 mg L^–1 were observed after 4 h of cultivation. This is a ∼3-fold increase compared to the highest GarQ levels observed with C. glutamicum CR099/pXMJ19-SP_ywaD–garQ-v2, i.e., the strain harboring an intact htrA gene (Figure 2). Also, GarQ concentrations remained high at t = 8 h of fermentation. Of note, lack of a functional htrA gene had no effect on GarQ levels in CGXII-U, CGXII-U+C or CGXII-U+T. This suggests that HtrA causes proteolytic degradation of GarQ when expressed and deletion of htrA improves production only under the optimized conditions, i.e., growth in CGXII-U+CT, i.e., when both CaCl₂ and Tween 80 are supplemented. Nevertheless, we still observed a decrease in GarQ activities at later timepoints of the fermentation (14.6 ± 2.8 mg L^–1 GarQ at t = 24 h) indicating that there are also other factors that reduce levels of the active product.

Recently, we were able to show that fermentation under low oxygen conditions in bioreactor or non-baffled flasks improves production of the class IIa bacteriocin pediocin PA-1 (Goldbeck et al., 2021; Christmann et al., 2023). A possible explanation for this effect is that pediocin PA-1 is (partially) inactivated by oxidation of the methionine residue in position 31 (Bédard et al., 2018). Similar to pediocin PA-1, GarQ contains a methionine residue at position 5 of the active peptide. Hence, inactivation of the peptide by oxidation of this methionine may occur. Indeed, re-evaluation of the LC-MS data of the purified GarQ fraction (Figure 1) revealed that mass spectra also contained signals with of 1072.326 m/z (spacing: 0.200) and 893.772 m/z (spacing: 0.167) corresponding to the five- and six-fold protonated ions of oxidized GarQ (Supplementary Figure 5A). We also observed a loss of GarQ activity by about 50% after overnight incubation at 30°C, 130 rpm compared to samples stored at −20°C (Supplementary Figure 5B).

To test if reduced aeration might positively impact on GarQ production, we cultivated C. glutamicum K9/pXMJ19-SP_ywaD–garQ-v2 in non-baffled shake flasks (Figure 5) and observed a slight increase in GarQ titers to 47.6 ± 1.3 mg L^–1 in cultures grown in CGXII-U+CT at 4 h of fermentation. However, in contrast to all previous experiments GarQ concentrations further increased toward the end of the fermentation and reached a maximum of 102.8 ± 1.1 mg L^–1 after 24 h of cultivation. Similarly, increased levels of GarQ were observed in CGXII-U+T at 4 h (18.6 ± 3.2 mg L^–1) and 8 h (56.0 ± 12.1 mg L^–1) of cultivation. Of note, reduced aeration had no impact on GarQ levels in CGXII-U or CGXII-U+C.

Activity of the class II bacteriocin pediocin PA-1 was shown to be sensitive to oxidation of a methionine residue. As GarQ contains a methionine in position 5 of the active peptide, we hypothesized that GarQ may be inactivated by oxidation of this methionine and cultivation in low oxygen conditions may reduce oxidative inactivation Garvicin AG2 is a natural variant of GarQ, in which the methionine in position 5 is changed to a phenylalanine (Maldonado-Barragán et al., 2022). To test if this peptide may be protected against oxidative inactivation, we purchased a synthetic GarQ variant with the M5F amino acid exchange (GarQ^M5F) from a commercial supplier and compared its activity to native GarQ in pHluorin2 assays using L. lactis LMG2785/pNZ-pHin^Lm or L. monocytogenes EGDe/pNZ-pHin^Lm (Figure 6A).

Surprisingly, GarQ^M5F has markedly higher activity against both biosensors than the native peptide as concentration of GarQ^M5F to achieve a complete shift in fluorescence ratios was 64-fold lower than for GarQ. To test if GarQ^M5F is also protected against oxidative inactivation, we performed activity assays of GarQ^M5F after incubation O/N at 30°C with aeration or stored at −20°C. In contrast to our predictions, activity of GarQ^M5F was reduced by aeration (Figure 6B) and this effect was even more pronounced than oxidative inactivation of the wildtype GarQ (Supplementary Figure 5). This indicates that GarQ^M5F has markedly improved activity, but oxidation of the methionine seems not to be the reason for reduced activity of GarQ after prolonged cultivation or incubation under oxygenated conditions.

Based on the interesting properties of GarQ^M5F, we generated a C. glutamicum production strain again employing Sec-dependent secretion using the YwaD SP. The strain C. glutamicum K9/pXMJ19-SP_ywaD-garQ^M5F was cultivated in CGXII-U+CT in baffled flasks and activities were determined in supernatants at different timepoints during cultivation (Figure 7). Surprisingly, activities were approx. 40-fold lower than with the isogenic strain producing wildtype GarQ cultivated under the same conditions (Figure 4). As observed in all other experiments conducted in baffled flasks, GarQ^M5F activity was decreased by 10-fold at 24 h of incubation again supporting that methionine oxidation is not the reason for loss in activity of GarQ and GarQ^M5F during recombinant production under high oxygen conditions.