The Campylobacter concisus BisA protein plays a dual role: oxide-dependent anaerobic respiration and periplasmic methionine sulfoxide repair

ABSTRACT Campylobacter concisus, an emerging pathogen found throughout the human oral-gastrointestinal tract, is able to grow under microaerobic or anaerobic conditions; in the latter case, N- or S-oxides could be used as terminal electron acceptors (TEAs). Analysis of 23 genome sequences revealed the presence of multiple (at least two and up to five) genes encoding for putative periplasmic N- or S-oxide reductases (N/SORs), all of which are predicted to harbor a molybdopterin (or tungstopterin)-bis guanine dinucleotide (Mo/W-bisPGD) cofactor. Various N- or S-oxides, including nicotinamide N-oxide, trimethylamine N-oxide , biotin sulfoxide, dimethyl sulfoxide, and methionine sulfoxide (MetO), significantly increased anaerobic growth in two C. concisus intestinal strains (13826 and 51562) but not in the C. concisus oral (type) strain 33237. A collection of mutants was generated to determine each N/SOR substrate specificity. Surprisingly, we found that disruption of a single gene, annotated as “bisA” (present in strains Cc13826 and Cc51562 but not in Cc33237), abolished all N-/S-oxide-supported respiration. Furthermore, ΔbisA mutants showed increased sensitivity to oxidative stress and displayed cell envelope abnormalities, suggesting BisA plays a role in protein MetO repair. Indeed, purified recombinant CcBisA was able to successfully repair MetO residues on a commercial protein (β-casein), as shown by mass spectrometry. Our results suggest that BisA plays a dual role in C. concisus, by allowing the pathogen to use N-/S-oxides as TEAs and by repairing periplasmic protein-bound MetO residues, therefore essentially being a periplasmic methionine sulfoxide reductase (Msr). This is the first report of a Mo/W-bisPGD-containing Msr enzyme in a pathogen. IMPORTANCE Campylobacter concisus is an excellent model organism to study respiration diversity, including anaerobic respiration of physiologically relevant N-/S-oxides compounds, such as biotin sulfoxide, dimethyl sulfoxide, methionine sulfoxide (MetO), nicotinamide N-oxide, and trimethylamine N-oxide. All C. concisus strains harbor at least two, often three, and up to five genes encoding for putative periplasmic Mo/W-bisPGD-containing N-/S-oxide reductases. The respective role (substrate specificity) of each enzyme was studied using a mutagenesis approach. One of the N/SOR enzymes, annotated as "BisA", was found to be essential for anaerobic respiration of both N- and S-oxides. Additional phenotypes associated with disruption of the bisA gene included increased sensitivity toward oxidative stress and elongated cell morphology. Furthermore, a biochemical approach confirmed that BisA can repair protein-bound MetO residues. Hence, we propose that BisA plays a role as a periplasmic methionine sulfoxide reductase. This is the first report of a Mo/W-bisPGD-enzyme supporting both N- or S-oxide respiration and protein-bound MetO repair in a pathogen.

Analysis of all C. concisus genomes available to date highlights the presence of several (i.e., at least two, often three or four, and sometimes up to five) genes encoding for putative N-or S-oxide reductases (hereafter referred to as "N/SORs"); all of which are predicted to harbor a molybdopterin (or tungstopterin)-bis guanine dinucleotide (Mo/ W-bisPGD) cofactor.In comparison, the reference strain C. jejuni NCTC 11168, previously shown to reduce both DMSO and TMAO, contains only one related N/SOR protein complex (21)(22)(23).Recently, Yeow et al. found that the anaerobic growth of three C. concisus strains was significantly better when DMSO was added on plates, suggesting that this particular S-oxide can be used as an alternative electron acceptor by C. concisus (16).However, DMSO reductase activity was not assigned to any specific gene or enzyme in any of these three strains (16).In the present study, we selected three unrelated C. concisus sequenced strains, i.e., 13826, 33237, and 51562, to study N-/S-oxide respira tion for the following reasons: (i) they are representative of the two main C. concisus genomospecies (GS), as both Cc51562 and Cc33237 belong to GS1, while Cc13826 belongs to GS2 (24,25); (ii) they come from distinct human geographic niches, as both Cc13826 and Cc51562 were isolated from feces (6), whereas Cc33237 was isolated from the oral cavity (25); (iii) finally, they harbor very distinct sets of N/SOR genes, representa tive of C. concisus as a species.
The three C. concisus strains studied herein (13826, 33237, and 51562) have only one N/SOR complex in common, the TorA-TorC/Y complex conserved in all C. concisus strains (Table S1).The torA gene encodes for a putative 813 amino-acid (aa)-long catalytic subunit, while the adjacent torC/Y gene encodes for a putative 189 aa-long, monohemecontaining cytochrome c, predicted to be periplasmic, based on the presence of a Secdependent signal peptide (28,29).Another putative N/SOR gene, annotated as "torZ," is usually found by itself (i.e., without any adjacent cytochrome-encoding gene).This gene, which encodes for a 817 aa-long periplasmic reductase, is present in almost all C. concisus strains (e.g., 21 out of 23 sequenced genomes; Table S1), including Cc13826 and Cc33237; however, torZ was not found in the genome of Cc51562.
Other putative N/SOR genes encountered in many (87%) C. concisus genomes include genes annotated as "bisA" and "bisB"; both are present in Cc13826 and Cc51562; however, neither can be found in Cc33237 (Table S1).The bisA gene encodes for a putative 855 aa-long catalytic subunit, while the bisB gene encodes for a 189 aa-long putative monoheme type c-cytochrome.Both BisA and BisB are expected to be transported to the periplasm, via the Tat and Sec translocation system, respectively (28, 29) (Fig. S3).Among all putative N/SORs found in C. concisus, BisA is the catalytic subunit that shares the highest homology (65% identity, 78% similarity) with the C. jejuni Cj0264c protein found in strain NCTC 11168 (21) (Table S2).
In addition, one-fourth of C. concisus genomes harbor the dmsAB-torC-dmsC operon, predicted to encode for a DmsABC-type heterotrimeric membrane-bound complex (Fig. S3) similar to that found in E. coli (30) and in select strains of C. jejuni (31,32) (Table S2).This operon is present in strains Cc13826 and Cc33237; however, it was not detected in the Cc51562 genome.Finally, Cc13826 contains yet another putative N/SOR gene: the 13826_1396 gene, adjacent to bisA (1395) and bisB (1398) (Fig. 1), encodes for a 835 aa-long protein with high-sequence homology to BisA (62% identity).The presence of this seemingly redundant gene, herein annotated as bisA′, is restricted to a few (~22%) C. concisus strains (Table S1).

Addition of various S-or N-oxides enhanced anaerobic respiration in the intestinal strains Cc13826 and Cc51562 but not in the oral strain Cc33237
C. concisus wild-type (WT) strains 13826, 33237, and 51562 were incubated in rich liquid broth supplemented with biotin sulfoxide (BSO), dimethyl sulfoxide (DMSO), methionine sulfoxide (MetO), nicotinamide N-oxide (NANO), or trimethylamine N-oxide (TMAO) (10 mM each), and their growth yield after 24 h was compared to that achieved in non-supplemented "plain" medium (Fig. 2).Supplementation of the medium with any of the five N-/S-oxides significantly increased the growth yield in both intestinal strains, Cc13826 and Cc51562, as determined by OD 600 read (Fig. 2A through C ) and colonyforming unit (CFU) counts (Fig. 3).By contrast, no increase in growth was observed in the presence of 10 mM methionine sulfone (MetO 2 ), a compound structurally related to MetO but which cannot be reduced by methionine sulfoxide reductases (data not shown).Surprisingly, none of the five N-or S-oxides had any effect on the growth of the (oral) type strain Cc33237 (Fig. 2B), even when oxide substrates were added in excess (up to 50 mM, data not shown).These unexpected results suggest Cc33237 cannot use any of the N-or S-oxide compounds for anaerobic respiration despite the presence of three putative N/SOR complexes (i.e., dmsABC, torAC/Y, and torZ) in its genome.Finally, the addition of nitrate and oxygen to the growth medium and the overhead space, respectively, resulted in a significant growth increase (data not shown), suggesting that the failure to use N-or S-oxide compounds is not linked to a broader respiration problem in this particular strain.

C. concisus ΔbisA mutants are deficient in N-/S-oxide-supported anaerobic growth
To determine which enzyme complex is responsible for N-/S-oxide respiration in C. concisus, as well as to assign substrate specificity (e.g., N-vs S-oxide), nine different genes were inactivated among the three strains, as previously described (15) (Fig. 1).Each mutant was then compared to its respective parental strains for its ability to use any of the N-/S-oxide compounds as terminal electron acceptor (TEA) during anaerobic growth (Fig. 2).In Cc13826, the C. concisus strain with the most diverse N/SOR gene content (i.e., five genes), mutagenesis of bisA′, dmsA, torA, or torZ did not lead to any growth deficiency.On the contrary, supplementation with either N-/S-oxide resulted in a significant anaerobic growth increase (compared to plain medium control) for 13826 ΔbisA′, ΔdmsA, ΔtorA, or ΔtorZ mutant (Fig. 2A), suggesting that none of these four genes is involved in N-/S-oxide reduction in Cc13826.Likewise, disruption of torA in strain Cc51562 did not lead to any deficiency in N-or S-oxide-supported anaerobic growth; the 51562 ΔtorA mutant grew as well as, and sometimes better than, the WT in the presence of N-/S-oxides (Fig. 2C).
By contrast, disruption of the bisA gene in strains Cc13826 (i.e., 13826_1395) and Cc51562 (i.e., 51562_0033) decreased or abolished N-/S-oxide-mediated growth stimuli, as shown by OD 600 (Fig. 2) and live cells counts (Fig. 3).These results demonstrate that the BisA protein plays an essential role in N-/S-oxide respiration in both strains Cc13826 and Cc51562.Finally, despite the fact that no N-/S-oxide had any effect on the anaerobic growth of WT strain 33237 (see previous paragraph), two putative N/SOR genes (i.e., torA and torZ) were inactivated in this strain (Fig. 1), and their N-or S-oxide-depend ent anaerobic growth was studied (Fig. 2B).As expected, none of the aforementioned compounds had any effect on the anaerobic growth of either mutant (Fig. 2B).

C. concisus ΔbisA mutants have cell division defects
Upon routine microscopic observation of mutants, we noticed ΔbisA mutant cells had strikingly different morphology compared to their parental WT strain (Fig. 4).Indeed, bright field microscopy analysis revealed that both cells from strain 13826 ΔbisA (Fig. 4B) and 51562 ΔbisA (Fig. 4D) were significantly longer compared to their parental strains, WT strain 13826 (Fig. 4A) and 51562 (Fig. 4C), respectively.This unexpected phenotype is likely linked to the deficiency of the Msr activity of BisA (rather than its N-/S-oxide respiratory role).We hypothesize that one or several periplasmic proteins involved in cell division in C. concisus are prone to methionine oxidation, and subsequent repair by BisA, as further discussed below.

C. concisus ΔbisA mutants are more sensitive to oxidative stress
The inability to repair oxidized methionine residues usually correlates with increased sensitivity to reactive oxygen species, including superoxide anion (O 2 − .)and hypochlo rous acid (HOCl) (33).Agar plate disk diffusion assays were used to compare the ability of C. concisus WT and ΔbisA mutant strains to combat methyl viologen (MV; a source of O 2 − .)or sodium hypochlorite (NaClO; a source of HOCl) (Table 1).Overall, WT strain Cc13826 appeared to be more sensitive (i.e., larger inhibition zones) than WT strain

In vitro repair of oxidized β-casein MetO residues by CcBisA
Electron spray ionization-mass spectrometry (ESI-MS) was used to determine whether CcBisA can repair oxidized β-casein in vitro.This approach has been successfully used to characterize several Msr enzymes, including Rhodobacter sphaeroides DorA and Msr and Saccharomyces cerevisiae MsrB (34)(35)(36).Two variants of C. concisus BisA (native and hexahistidine-tagged) were expressed in Escherichia coli and purified to near homoge neity (Fig. S4).The N-/S-oxide reductase activity of both purified proteins was spectro photometrically assayed, using reduced benzyl viologen (BV) as electron donor, and DMSO, MetO, or TMAO as electron acceptor (data not shown).Although both purified proteins showed measurable BV-dependent oxide reductase activity, including MetO reductase activity, the His-tagged version (CcBisA-His 6 ) was more active and, therefore, selected for the protein-bound MetO repair experiment.As mentioned above, the repair experiment involved the use of bovine β-casein, an intrinsically disordered protein that for less than 24 h on BA plates under H 2 -enriched microaerobic conditions, resuspended in BHI broth, and examined with a bright field microscope (1,000 × magnification).Pictures have been digitally processed (brightness and contrast) for ease of interpretation.contains 6 Met residues readily oxidizable as either Met-R-O or Met-S-O epimer upon incubation with H 2 O 2 (34).ESI-MS was used to measure the total mass of β-casein, before and after H 2 O 2 treatment and, finally, after incubation with CcBisA-His 6 (Fig. 5).Commercial β-casein consists of a mixture of genetic variants (34); in our case, up to 11 peaks were detected between 24,050 and 24,450 Da in the mixture (Fig. 5A).Prolonged treatment with H 2 O 2 led to an overall mass increase of 96 Da for each variant (peak) corresponding to oxidation (i.e., +16 Da) of each of the 6 Met residues, as previously reported (34) (Fig. 5B).Subsequent incubation of oxidized β-casein with CcBisA-His 6 in the presence of dithionite-reduced BV (electron donor) resulted in a decrease in mass of approximately 16, 32, or 48 Da for most of the variants, corresponding to one, two, or even three MetO residues being repaired (i.e., recycled into Met), respectively (Fig. 5C).Similar changes (mass decrease) were not observed in the no-BisA control (oxidized β-casein with dithionite-reduced BV, data not shown), indicating BisA is responsible for the change in mass.Most (but not all) β-casein variants were repaired; however, the fact that only one, two, or sometimes three MetO residues were repaired indicates that CcBisA is stereospecific, i.e., it can only repair Met-R-O or Met-S-O epimers, but not both.Nevertheless, our results are in good agreement with previous reports.For instance, only one or two β-casein MetO residues were found to be repaired by R. sphaeroides DorA (34), and an average of 2.5 MetO per oxidized β-casein was reduced by S. cerevisiae MsrA (36).In conclusion, our proteomic results indicate that C. concisus BisA is able to repair protein-bound MetO residues in vitro.

Role of BisA in N-/S-oxide-supported anaerobic respiration
The ability to use alternative electron acceptors for respiration confers an advantage to the organisms that are capable to do so.For instance, both DMSO and TMAO have been known for a long time to be excellent electron acceptors for a variety of bacteria, including purple photosynthetic bacteria such as R. sphaeroides and R. capsulatus (37), γ-proteobacteria such as Escherichia coli (38) and related ε-proteobacteria such as C. jejuni (22,23) and Wolinella succinogenes (39).Based on results presented herein and elsewhere (16), it is now clear that the list of DMSO-reducing bacteria should include C. concisus.We must, however, restrict our conclusion to the two intestinal strains used in our study (13826 and 51562) (24) since the oral strain Cc33237 (25) did not respond to the DMSO supplementation treatment.Besides DMSO, four other N-/S-oxides, BSO, MetO, NANO, and TMAO, were shown to significantly bolster the growth of both Cc13826 and Cc51562 but not that of Cc33237.Although puzzling at first, the apparent inability of strain Cc33237 to respire any N-/S-oxide compounds appears to correlate with the absence of a bisA homolog in the genome of this strain, hence reinforcing the role of BisA as the main N/SOR enzyme in C. concisus.To confirm this hypothesis, it would be of interest to determine whether other bisA-lacking C. concisus strains, including the previously studied P10CDO-S2 and H1O1 strains (40), fail to respond to the N-/S-oxide supplementation treatment.Alternatively, introducing and expressing bisA in strain Cc33237 would provide us with useful information.However, such heterologous expression has never been done in C. concisus, and it is anticipated to be challenging due to the paucity of genetic tools.Unexpectedly, disruption of each of the four other putative N/SOR genes encoun tered in C. concisus (i.e., bisA′, dmsA, torA, torZ) did not lead to any detectable phenotype, leaving the respective role of each of these enzymes unanswered for now.In this context, previous studies conducted in the related ε-proteobacterium C. jejuni can be informative although there are significant differences between N/SOR homologs found in the two Campylobacter species.Indeed, most C. jejuni strains, (e.g., NCTC 11168) contain only one N/SOR protein complex (Cj0264/0265 in strain NCTC11168) (21)(22)(23).In addition, a few C. jejuni strains (e.g., Cj81-176) possess a second N/SOR complex (31,32) with significant homology to the DmsABC-type of DMSO reductase found in E. coli (30) and in some C. concisus strains (e.g., Cc13826 and Cc33237).Though Cj 81-176 ΔdmsA mutants have been found to display significant mouse colonization defects, suggesting dmsA plays a critical role in vivo (31), the DmsA enzyme has not been biochemically characterized and its preferred substrate is not known.Hence, the physiological role of DmsA(BC) paralogs remains to be determined in Campylobacter spp.Regarding the Cj0264/0265 complex, the WT strain NCTC 11168 was shown to metabolize DMSO and TMAO (into DMS and TMA, respectively), a property lost in the cj0264 mutant (23).Furthermore, both DMSO and TMAO were shown to serve as alternative electron acceptors by the WT but not by the cj0264 mutant (22,23).Cj0264 is expected to get electrons from a monoheme c-type soluble cytochrome encoded by the adjacent cj0265 gene, which is highly conserved in all C. jejuni strains (22,41); Cj0265 itself is expected to receive electrons from the QcrABC membrane-bound menaquinol-cytochrome c reductase complex (42).In C. concisus, the concomitant presence of qcrABC, bisA, and bisB genes (the latter encoding for a monoheme c-type soluble cytochrome, highly similar to Cj0265) suggests a topological model and an electron transfer route similar to that described for C. jejuni Cj0264/264 (42), i.e., QcrC→BisB→BisA (Fig. S3).

Role of BisA in periplasmic methionine sulfoxide repair
Based on various phenotypes associated with the disruption of bisA in both Cc13826 and Cc51562, we conclude that BisA plays a role as periplasmic Msr, able to repair both free MetO residues and protein-bound MetO.The first conclusion (repair of periplasmic free MetO) is based on the observation that free MetO can be used as a respiratory substrate by BisA in the periplasm, i.e., MetO recycling into Met would be a direct consequence of the respiration process.Future studies using pure (BisA) enzyme will be needed to determine substrate stereospecificity (for instance, Met-R-O vs Met-S-O), as well as affinity constants and additional kinetic parameters.The second conclusion (repair of periplasmic protein-bound MetO) is based on the following results.First, we found that the purified recombinant BisA-His 6 enzyme was able to repair between one and three MetO residues out of six MetO residues found on oxidized β-casein, as shown by ESI-MS; these results are in agreement with that of previous studies using the same target protein (oxidized β-casein) and Msr-type enzymes such as R. sphaeroides DorA and S. cerevisiae MsrA (34,36).Second, ΔbisA mutant cells failed to properly divide, suggesting BisA-mediated MetO repair is crucial for cell division in C. concisus; fully periplasmic or membrane-bound periplasmic-oriented Met-rich proteins that could account for such phenotype include two Met-rich proteins, annotated as "septum formation initiator" (6.3% Met content) and "cell division protein" (5.3% Met content); for reference, the average protein Met content in C. concisus is 2.4%.However, both the oxidation of these two periplasmic Met-rich proteins and their putative repair by BisA will need to be experimentally proven, using pure components and ESI-MS.Lastly, we observed an increased sensitivity of ΔbisA mutants toward oxidative stress, a hallmark of msr mutants in many eukaryotic and prokaryotic organisms (33), including S. cerevisiae (43), E. coli (44), H. pylori (45), and C. jejuni (46,47).
C. concisus contains an additional periplasmic Msr enzyme; however, it is not of the MsrPQ-type originally described in E. coli (48) and widely encountered in many Campylobacter sp., including in C. jejuni (42,49).Instead, C. concisus contains a homolog of the multifunctional MsrAB-type fusion protein, similar to that found in H. pylori (45), but with one notable difference: HpMsrAB is cytoplasmic, whereas CcMsrAB is predicted to be periplasmic (29).Surprisingly, genome sequence analysis suggests there is no cytoplasmic Msr in C. concisus.This finding raises the question of the fate of cytoplasmic oxidized proteins, especially in the catalase-negative organism that is C. concisus.In the absence of a dedicated cytoplasmic MetO repair pathway, a non-specific quenching mechanism could be involved, similar to that previously uncovered in H. pylori (50) or in other organisms, including in eukaryotes (51,52).

Some BisA homologs play similar dual roles in other bacteria
Two periplasmic molybdoenzymes with significant homology to CcBisA have been shown to play a similar dual role (N-/S-oxide respiration and MetO repair) in two unrelated organisms.In Haemophilus influenzae, a protein annotated as "MtsZ" can use BSO and MetO, and repair free MetO (53).However, purified HiMtsZ could not repair calmodulin-bound MetO, suggesting that the primary role of HiMtsZ is respiratory, rather than MetO-protein repair (53).This is a key difference with CcBisA since a repair role of both free MetO and protein-bound MetO is proposed here.In the photosynthetic bacterium R. sphaeroides, a protein named "DorA" also plays a role in both MetO respiration and repair (34).Indeed, RsDorA was shown to repair two physiologically relevant periplasmic oxidized Met-rich proteins, including the copper chaperone PCu A C (34).Interestingly, a periplasmic Met-rich (i.e., 6.4% Met) PCu A C homolog is present in C. concisus: it could be a preferred target for CcBisA.There are differences between CcBisA and RsDorA: CcBisA is a soluble protein, predicted to be associated with only one protein, the soluble periplasmic cytochrome c annotated here as CcBisB (Fig. S3), whereas RsDorA is associated with two membrane subunits, DorB and DorC (34,54), making the R. sphaeroides DorABC more similar to a DmsABC-type complex described above.
In summary, it appears that, among the many putative N/SORs found in C. concisus, only one, annotated as BisA, is actually involved in N-/S-oxide-supported respiration.Additionally, CcBisA is expected to repair protein-bound MetO in the periplasm.This dual role might be surprising, especially given the fact that C. concisus has another periplas mic MsrAB enzyme.But one has to remember that the average genome size of Campy lobacter sp., including C. concisus (i.e., 1.8-2.1 Mb), is significantly smaller compared to that of other groups of commensal or pathogenic bacteria living in similar environmental niches, including Staphylococci (~3 Mb), Bacteroides (~5-7 Mb), or Enterobacteriaceae such as E. coli and Salmonella enterica (~5 Mb).Thus, to adapt to and thrive in the extremely diverse environment that is the human OGIT, C. concisus has to rely on both versatile respiratory pathways and multifunctional enzymes, as exemplified herein by BisA, a bifunctional enzyme capable of both N-/S-oxide respiration and protein-bound methionine sulfoxide repair.

Bacterial strains and plasmids
C. concisus strains used in this study are listed in Table S4.The three C. concisus paren tal strains (13826, 33237, and 51562) were purchased from ATCC (Manassas, VA, USA).Genomic DNA from each C. concisus parental strain was used as template for PCR to construct each strain-specific mutant.All PCR products were sequenced at ETON (Research Triangle Park, NC, USA).

Growth conditions
Campylobacter concisus was routinely grown on Brucella agar (Becton Dickinson, Sparks, MD, USA) plates supplemented with 10% defibrinated sheep blood (Hemostat, Dixon, CA, USA) (BA plates).Chloramphenicol (Cm, 8-25 μg/mL) was added as needed.BA plates were incubated at 37°C under H 2 -enriched microaerobic conditions, consisting of sealed pouches filled with anaerobic mix, a commercial gas mixture containing 10% H 2 , 5% CO 2 , and 85% N 2 (Airgas, Athens, GA, USA) and some residual air.For liquid cultures, 165 mL sealed bottles were filled with 2-10 mL of 1.25× concentrated Brain-Heart Infusion (BHI, Becton Dickinson), autoclaved, and then supplemented with 0.2-1 mL (10% final) of fetal calf serum (FCS, Gibco Thermo Fisher) and 0.2-1 mL (10% final) of either sterile deionized water (plain, control) or 100 mM BSO, DMSO, MetO, MetO 2 , NANO, or TMAO (each at a final concentration of 10 mM).Cells were grown under H 2 -enriched anaerobic conditions, as follows.First, bottles (overhead space) were sparged with N 2 for 10 min and then with anaerobic mix (see above) for 10 min.Additional H 2 (10%) was added, to achieve 20% H 2 partial pressure.The inoculum was prepared as follows: C. concisus cells grown on BA plates for less than 24 h were harvested, resuspended in BHI-FCS, and standardized to the same OD 600 before being inoculated.The starting OD 600 was between 0.04 and 0.05, corresponding to approxi mately 5 × 10 7 to 8 × 10 7 CFU/mL, respectively (15).Cells were grown (triplicate for each strain and condition) for 24 h at 37°C under vigorous shaking (200 rpm).Growth yield was estimated by both measuring OD 600 and counting CFUs.For this purpose, samples from each bottle were serially diluted in PBS (up to 10 −8 ), and 5 μL of each dilution was spotted in triplicate on BA plates.CFU was counted after 48-72 h of incubation under H 2 -enriched microaerobic conditions.Results shown are the mean and standard deviation (SD) of at least triplicate biological replicates, each with technical triplicate.

Construction of C. concisus mutants
Each mutant was constructed by following a three-step strategy, as previously described (15).Briefly, DNA sequences (400-850 bp long) flanking each target gene (i.e., bisA, bisA′, dmsA, torA, torZ) were PCR amplified, using genomic DNA from each C. concisus parental strain (e.g., 13826, 33237, 51562; see Table S4) and specific primers for each targeted gene (Table S5).Each set of two PCR products was combined with a 740-bp-long DNA sequence containing a cat (chloramphenicol resistance) cassette that has its own promoter, and the final PCR step yielded a product containing both flanking sequences with the cat cassette in between (more detailed information on the construction of each mutant is provided in Supplementary data).In the second step, each tripartite PCR product was purified and methylated, by incubating approximately 25 μg of DNA with 150-250 μg of (cell-free extract) total protein from C. concisus for 2 h at 37°C in the presence of 0.4 mM S-adenosylmethionine (SAM, New England Biolabs, Ipswich, MA, USA).After methylation, each PCR product was purified again (Qiaquick purification kit, Qiagen, Valencia, CA, USA), and the methylated DNA (1-5 μg) was introduced into C. concisus, by using natural transformation or electroporation (BTX Transporator Plus, 2,500 V/pulse).Transformed cells were first plated on BA plates and incubated (H 2 -enriched microaerobic conditions, see above) for 8-12 h before being transferred onto BA supplemented with 8-10 μg/mL Cm.Colonies appeared after 3-5 days.The concomitant deletion of the gene of interest and the insertion of cat was confirmed by PCR, using genomic DNA from mutants as template and appropriate primers.

Microscopy analysis
C. concisus cells (13826 WT, 51562 WT, and ΔbisA mutant strains) were grown for less than 24 h on BA plates under H 2 -enriched microaerobic conditions (see above), resuspended in BHI broth, and examined with a bright field microscope (Nikon Eclipse Ni).Digital images were obtained at 1,000× magnification using a Nikon cool snap camera.

N-/S-oxide reductase enzyme assays
BisA reductase activity was measured as previously described (34,35) with modification.Briefly, Britton-Robinson buffer (55), pH 6.0, was sparged for 20 min with N 2 to create semi-anaerobic conditions.The N 2 -infused buffer was then mixed with 0.2 mM of benzyl viologen (BV, electron donor) and 5-10 mM of DMSO, MetO, or TMAO (N-/S-oxide electron acceptor) in a cuvette fitted with a silicone seal to prevent gas exchange (1 mL final volume).A few microliters of freshly prepared sodium dithionite (Sigma) solution were injected into the cuvette to reduce BV until the absorbance at 600 nm was stable and between 1 and 1.5 units.The reaction was initiated by the addition of 0.2-1 mM of purified CcBisA or CcBisA-His 6 .Oxidation of BV (i.e., A 600 decrease) was followed for 1 min.Rates of N-/S-oxide reductase activity were calculated using a molar extinction coefficient (ε 600 ) of 10,400 M −1 •cm −1 , with 2 moles of BV oxidized for 1 mole of reduced oxide substrate.

ESI-MS analysis of β-casein oxidation and subsequent repair by CcBisA
The protein-bound MetO repair ability of CcBisA was assessed using commercial bovine β-casein as target, and dithionite-reduced BV as electron donor, as previously described by Tarrago and collaborators, with modifications (34).Briefly, 100 μM of bovine β-casein (Sigma # C-6905) was oxidized overnight at room temperature in 10 mL PBS (pH 7.5) with 200 mM H 2 O 2 .The protein was concentrated and the oxidant was removed by using an Amicon concentrator with a MWCO of 10 kDa, followed by a PBS wash and a second concentration step.Finally, β-casein was resuspended in 1 mL of PBS, and the protein concentration was determined with the BCA protein assay kit.Some oxidized β-casein was kept aside for further analysis.For repair of oxidized β-casein by BisA, N 2 -sparged Britton-Robinson buffer, pH 6.0, was mixed with 100 μM β-casein, 1 μM of purified CcBisA-His 6 and 200 μM dithionite-reduced BV.Repair was conducted at 37°C for approximately 2 h, with 1.6 mM dithionite injected into the solution every 20 min, as previously described (34).The repair mix was frozen and stored at −80°C before being sent for analysis, along with appropriate controls (unoxidized β-casein; oxidized β-casein; and BisA-free, oxidized β-casein with BV and dithionite).Finally, β-casein was purified by liquid chromatography (LC) and then analyzed by electron spray ionization (53)-mass spectrometry (MS) on a Bruker Esquire 3,000 plus ion trap mass spectrometer, at the Proteomics and Mass Spectrometry facility (PAMS), University of Georgia, Athens, GA, USA.

FIG 1
FIG 1 Location and organization of genes encoding for putative N-and S-oxide reductases in various C. concisus strains and mutagenesis sites.The name of each C. concisus strain (13826, 33237, and 51562) is indicated above each column.Each gene is represented by an arrow.Colored arrows represent genes encoding for N/SOR components, and white arrows represent genes not relevant to the study.Putative gene names and locus tag numbers are indicated underneath and above each arrow, respectively.Gene annotations are according to JGI-IMG/M website (img.jgi.doe.gov).Genes targeted in this study (bisA, bisA′, dmsA, torA, torZ) are indicated by hatched arrows.An approximate scale is shown bottom left.

FIG 2
FIG 2 Growth yield of WT and mutant strains in the presence of various N-or S-oxides.C. concisus WT strains 13826 (A), 33237 (B), 51562 (C), and their isogenic N/SOR mutants were inoculated (OD 600 of 0.05, dashed line) in plain growth medium or in medium supplemented with 10 mM of BSO, DMSO, MetO, NANO, or TMAO, as indicated by each colored square on the right.Cells were incubated under H 2 -enriched anaerobic atmosphere at 37°C, with permanent shaking (220 rpm), and OD 600 was recorded after 24 h.Results shown represent means and standard deviations from at least three biological replicates.A single asterisk above a bar indicates the bacterial growth yield (OD 600 ) for the mutant is significantly lower compared to that of the WT strain, for the same (N-/S-oxide) condition (P < 0.01, Student's t-test).Double asterisk, P < 0.02.The effect of BSO on the 13826 ΔbisA′ mutant strain was not tested.

FIG 3
FIG 3 Quantitative growth yield of WT and ΔbisA mutant strains in the presence of various N-or S-oxides.C. concisus WT 13826 and 51562 and ΔbisA mutant strains were inoculated (~5 × 10 7 CFUs/mL, dashed line) in plain growth medium or in medium supplemented with 10 mM of BSO, DMSO, MetO, NANO, or TMAO, as indicated by each colored square on the right.After 24 h, cells were serially diluted in PBS, spotted on BA plates in triplicate, and incubated under H 2 -enriched microaerobic conditions for 2-3 days, after which CFUs were counted.Results shown represent means and standard deviations of CFU/mL from three biological replicates, each with three technical replicates.Each color corresponds to a different growth medium (plain or 10 mM of each N-/S-oxide), as indicated on the right.The asterisk above each bar indicates the bacterial cell growth (CFU/mL) of the ΔbisA mutant under the indicated (N-/S-oxide) condition is significantly lower compared to that of its isogenic WT strain grown under the same (N-/S-oxide) conditions (P < 0.01, Student's t-test).

TABLE 1
Sensitivity of C. concisus WT and ΔbisA to methyl viologen and sodium hypochlorite a C. concisus strains 13826 and 51562 WT and isogenic ΔbisA mutant cells were grown on BA plates for 24 h before being resuspended in sterile PBS buffer to a final OD 600 of 2.Then, 0.2 mL of cells was homogenously spread on top of 25-mL standardized BA plates.A sterile paper disk (7.5 mm diameter) was placed in the center of each plate, and 10 μL of 100 mM methyl viologen (MV) or 10 μL of 1.4 M sodium hypochlorite (NaClO) was added onto the disk.Cells were allowed to grow for 48 h, and the diameter (in mm) of the inhibition zone was measured.Results shown are means and standard deviations from three to five replicates.Inhibition zones (diameter) measured for each ΔbisA mutant were significantly larger than zones measured for each respective WT, for both MV and NaClO (P < 0.001, Student's t-test). a