Heme A Synthase Deficiency Affects the Ability of Bacillus cereus to Adapt to a Nutrient-Limited Environment

The branched aerobic respiratory chain in Bacillus cereus comprises three terminal oxidases: cytochromes aa3, caa3, and bd. Cytochrome caa3 requires heme A for activity, which is produced from heme O by heme A synthase (CtaA). In this study, we deleted the ctaA gene in B. cereus AH187 strain, this deletion resulted in loss of cytochrome caa3 activity. Proteomics data indicated that B. cereus grown in glucose-containing medium compensates for the loss of cytochrome caa3 activity by remodeling its respiratory metabolism. This remodeling involves up-regulation of cytochrome aa3 and several proteins involved in redox stress response—to circumvent sub-optimal respiratory metabolism. CtaA deletion changed the surface-composition of B. cereus, affecting its motility, autoaggregation phenotype, and the kinetics of biofilm formation. Strikingly, proteome remodeling made the ctaA mutant more resistant to cold and exogenous oxidative stresses compared to its parent strain. Consequently, we hypothesized that ctaA inactivation could improve B. cereus fitness in a nutrient-limited environment.


Introduction
Bacillus cereus is a ubiquitous endospore-forming bacterium, which mainly affects humans as a food-borne pathogen [1]. Recently, we examined the ability of the emetic strain B. cereus AH187 (F4810/72) to survive oligotrophic conditions encountered in groundwater. Our results showed that vegetative B. cereus cells rapidly evolved to produce a mixed population composed of endospores and asporogenic variants bearing mutations in the spo0A gene, which encodes a master regulator for entry into sporulation [2]. The whole genome of one of the variants isolated was sequenced, and the mutations identified included an alteration to codon 178 of ctaA gene (GCT→ACT, Ala→Thr). This variant survives in sterilized groundwater over a long period in a vegetative form and has a competitive advantage compared to its parental strain [2].
The ctaA gene encodes heme A synthase (CtaA), a membrane-bound enzyme that converts heme O to heme A. CtaA is required for cytochrome caa3 oxidase biosynthesis and sporulation in Bacillus subtilis [3]. B. cereus cytochrome caa3 is made up of four proteins (CtaCDEF), with Cu A and a cytochrome c domain in subunit II (CtaC). This protein may form a supercomplex with the cytochrome bc complex (QcrABC) and cytochrome c 550 (CccA) or cytochrome c 551 [4] in the membrane, as reported in B. subtilis [5]. Cytochrome caa3 is one of the two heme-copper terminal oxidases in the branched B. cereus aerobic respiratory chain [6,7] (Figure 1). The other is cytochrome aa3, which uses menaquinol as electron donor instead of cytochrome c. In contrast to cytochrome caa3, cytochrome aa3 is not strictly dependent on heme A for its activity, as it can also use heme B and heme O to produce a novel bo 3 cytochrome [8]. The third terminal oxidase in the B. cereus respiratory chain is a cytochrome bd menaquinol oxidase that requires neither copper nor heme A for dase was shown to contribute to microaerobic respiration in B. subtilis [12]. Respiratory flexibility is an important factor that allows bacteria to cope with changing oxygen and nutrient conditions.
Here, we investigated the impact of functional loss of ctaA on B. cereus, first discovered in an environment with limited nutrients. Proteomics analysis revealed that the mutant strain adapts its respiratory network. Thus, lack of CtaA disrupted electron flow through the bc-caa3 pathway, and upregulated the menaquinol-cytochrome aa3 oxidase pathway. Re-routing of respiratory chain electron transport causes endogenous redox stress and is accompanied by changes at the bacterial surface. , electrons (e − ) are transferred to cytochrome caa3, aa3, and cyd terminal oxidases to reduce oxygen to H2O. The resulting electrochemical gradient is used by ATP synthase to produce ATP. QH2 provides electrons to cytochrome caa3 (green) via cytochrome bc (purple) and cytochrome c (orange). Alternatively, electrons from QH2 can be delivered directly to cytochrome aa3 (blue) or cytochrome bd (pink) menaquinol oxidases, the latter is not a proton pump. Cytochromes caa3 and aa3 have four subunits encoded by the ctaCDEF and qoxABCD operons, respectively. Subunit I of both enzymes carries the electron-accepting heme a that delivers electrons to the active site-composed of heme a3 and a copper center (CuB). (b) Heme a synthesis pathway. Heme a synthesis is catalyzed by CtaB and CtaA-heme o and a synthases, respectively.

B. cereus AH187 CtaA Is Required for Cytochrome Aa3 Oxidase Activity, and Optimal Growth
To investigate the role of CtaA in wild-type B. cereus AH187 (WT), the ctaA gene (BCAH187_A4064) was deleted. On Lysogeny Broth (LB) agar plates, ΔctaA strain colonies were smaller than those of its parent strain, indicating a growth defect on solid medium ( Figure 2a). Colonies were tested for N, N, N′, N′-tetramethyl-p-phenylenediamine (TMPD) oxidation capacity. This artificial electron donor interacts specifically with , electrons (e − ) are transferred to cytochrome caa3, aa3, and cyd terminal oxidases to reduce oxygen to H 2 O. The resulting electrochemical gradient is used by ATP synthase to produce ATP. QH 2 provides electrons to cytochrome caa3 (green) via cytochrome bc (purple) and cytochrome c (orange). Alternatively, electrons from QH 2 can be delivered directly to cytochrome aa3 (blue) or cytochrome bd (pink) menaquinol oxidases, the latter is not a proton pump. Cytochromes caa3 and aa3 have four subunits encoded by the ctaCDEF and qoxABCD operons, respectively. Subunit I of both enzymes carries the electron-accepting heme a that delivers electrons to the active site-composed of heme a3 and a copper center (Cu B ). (b) Heme a synthesis pathway. Heme a synthesis is catalyzed by CtaB and CtaA-heme o and a synthases, respectively.
The composition of the respiratory chain is regulated by growth conditions [10]. Thus, the cytochrome aa3-terminating branch of the B. subtilis respiratory chain is the major contributor to respiration in most aerobic growth conditions, whereas the cytochrome caa3-terminating branch is a minor contributor [11]. The branch terminating at the bdoxidase was shown to contribute to microaerobic respiration in B. subtilis [12]. Respiratory flexibility is an important factor that allows bacteria to cope with changing oxygen and nutrient conditions.
Here, we investigated the impact of functional loss of ctaA on B. cereus, first discovered in an environment with limited nutrients. Proteomics analysis revealed that the mutant strain adapts its respiratory network. Thus, lack of CtaA disrupted electron flow through the bc-caa3 pathway, and upregulated the menaquinol-cytochrome aa3 oxidase pathway. Re-routing of respiratory chain electron transport causes endogenous redox stress and is accompanied by changes at the bacterial surface.

B. cereus AH187 CtaA Is Required for Cytochrome Aa3 Oxidase Activity, and Optimal Growth
To investigate the role of CtaA in wild-type B. cereus AH187 (WT), the ctaA gene (BCAH187_A4064) was deleted. On Lysogeny Broth (LB) agar plates, ∆ctaA strain colonies were smaller than those of its parent strain, indicating a growth defect on solid medium (Figure 2a). Colonies were tested for N, N, N , N -tetramethyl-p-phenylenediamine (TMPD) oxidation capacity. This artificial electron donor interacts specifically with cytochrome caa3 oxidase [13], producing blue staining of enzymatically-active colonies. Oxidation activity was detectable in WT and complemented strains, but not in the ∆ctaA strain (Figure 2b), confirming a lack of cytochrome caa3 activity in these colonies. cytochrome caa3 oxidase [13], producing blue staining of enzymatically-active colonies. Oxidation activity was detectable in WT and complemented strains, but not in the ∆ctaA strain (Figure 2b), confirming a lack of cytochrome caa3 activity in these colonies. In liquid MOD medium supplemented with 30 mM glucose (MODG) [14], the ∆ctaA mutant grew at the same rate and reached the same final biomass as its parent strain (Figure S1). However, it excreted more acetate (yield 1.11 ± 0.15 mol/mol glucose) than its parent strain (yield 0.48 ± 0.06 mol/mol glucose), indicating increased overflow metabolism. When glucose was replaced by glycerol, the ∆ctaA mutant showed altered growth ( Figure 3), suggesting that lack of CtaA and caa3 activity reduced the capacity of B. cereus to use non-phosphotransferase-system-dependent glycerol as a carbon source [15]. To confirm the role of caa3 activity in carbon metabolism, antimycin A-which selectively binds to the bc complex and interrupts cytochrome caa3′s function [16]-was added to growth medium. As expected in these conditions [6], growth of WT strain was altered on glycerol but not on glucose. Unexpectedly, antimycin A impaired ∆ctaA mutant growth on both glucose and glycerol (Figure 3), suggesting that in the absence of CtaA, this compound inhibits membrane-centered transport processes [17]. In liquid MOD medium supplemented with 30 mM glucose (MODG) [14], the ∆ctaA mutant grew at the same rate and reached the same final biomass as its parent strain ( Figure S1). However, it excreted more acetate (yield 1.11 ± 0.15 mol/mol glucose) than its parent strain (yield 0.48 ± 0.06 mol/mol glucose), indicating increased overflow metabolism. When glucose was replaced by glycerol, the ∆ctaA mutant showed altered growth (Figure 3), suggesting that lack of CtaA and caa3 activity reduced the capacity of B. cereus to use nonphosphotransferase-system-dependent glycerol as a carbon source [15]. To confirm the role of caa3 activity in carbon metabolism, antimycin A-which selectively binds to the bc complex and interrupts cytochrome caa3's function [16]-was added to growth medium. As expected in these conditions [6], growth of WT strain was altered on glycerol but not on glucose. Unexpectedly, antimycin A impaired ∆ctaA mutant growth on both glucose and glycerol (Figure 3), suggesting that in the absence of CtaA, this compound inhibits membrane-centered transport processes [17].

Proteome and Exoproteome Response to CtaA Deficiency
To determine how CtaA deficiency affects cellular metabolism at distinct growth phases on MODG medium, we performed shotgun proteomics assays on three biological replicates for ∆ctaA and WT strains at three time-points. The time-points were: EE (early exponential growth phase, OD = 0.1), LE (late exponential growth phase, OD = 1), and S (stationary growth phase, OD = 1.5) ( Figure S1). The proteomics dataset acquired on the 36 samples (2 strains × 3 time-points × 3 replicates for cellular proteome and exoproteome) comprised 843,332 MS/MS spectra and a total of 27,804 validated peptide sequences. From these results, based on the confident detection of at least two distinct peptides per protein, 1922 proteins were identified in the cellular proteome (Table S1), and 998 proteins in the exoproteome (Table S2).

Proteome and Exoproteome Response to CtaA Deficiency
To determine how CtaA deficiency affects cellular metabolism at distinct growth phases on MODG medium, we performed shotgun proteomics assays on three biological replicates for ΔctaA and WT strains at three time-points. The time-points were: EE (early exponential growth phase, OD = 0.1), LE (late exponential growth phase, OD = 1), and S (stationary growth phase, OD = 1.5) ( Figure S1). The proteomics dataset acquired on the 36 samples (2 strains × 3 time-points × 3 replicates for cellular proteome and exoproteome) comprised 843,332 MS/MS spectra and a total of 27,804 validated peptide sequences. From these results, based on the confident detection of at least two distinct peptides per protein, 1922 proteins were identified in the cellular proteome (Table S1), and 998 proteins in the exoproteome (Table S2).

Cellular Proteome
Principal component analyses (PCA) revealed good homogeneity of the replicates for each growth phase (Figure 4a). PCA also indicated that EE samples clearly segregated from LE and S samples, and that ΔctaA and WT samples showed poor convergence at LE compared to EE and S growth phases. These results indicate that the cellular proteome undergoes relatively few changes to support B. cereus growth in the absence of CtaA. Pairwise comparisons identified differentially accumulated proteins (DAPs) between ΔctaA and WT strains at each growth phase. Only proteins with an adjusted p-value ≤ 0.05 and at least a 1.5-fold-change (|log2 fold-change| ≥ 0.56) were considered to be differentially accumulated between the two strains. Overall, 58 DAPs were identified with high confidence, with 24 proteins less abundant (down-DAPs), and 34 more abundant (up-DAPs) in ΔctaA compared to WT (Table S3). The distribution of these proteins at the three growth phases is illustrated in Figure 4b. Among the 24 down-DAPs, whatever the growth phase, FliC flagellin (B7HLW0)-a major component of the surface-associated flagellum [18]was significantly less abundant in ΔctaA than in the WT strain (adjusted p-value < 10 −4 ). Flagellin-based motility may thus be compromised in the absence of CtaA. Another surface-associated protein, B7HXP4, which is potentially one of the two components of the B.

Cellular Proteome
Principal component analyses (PCA) revealed good homogeneity of the replicates for each growth phase ( Figure 4a). PCA also indicated that EE samples clearly segregated from LE and S samples, and that ∆ctaA and WT samples showed poor convergence at LE compared to EE and S growth phases. These results indicate that the cellular proteome undergoes relatively few changes to support B. cereus growth in the absence of CtaA. Pairwise comparisons identified differentially accumulated proteins (DAPs) between ∆ctaA and WT strains at each growth phase. Only proteins with an adjusted p-value ≤ 0.05 and at least a 1.5-fold-change (|log 2 fold-change| ≥ 0.56) were considered to be differentially accumulated between the two strains. Overall, 58 DAPs were identified with high confidence, with 24 proteins less abundant (down-DAPs), and 34 more abundant (up-DAPs) in ∆ctaA compared to WT (Table S3). The distribution of these proteins at the three growth phases is illustrated in Figure 4b. Among the 24 down-DAPs, whatever the growth phase, FliC flagellin (B7HLW0)-a major component of the surface-associated flagellum [18]-was significantly less abundant in ∆ctaA than in the WT strain (adjusted p-value < 10 −4 ). Flagellin-based motility may thus be compromised in the absence of CtaA. Another surface-associated protein, B7HXP4, which is potentially one of the two components of the B. cereus S-layer [2], was also less abundant in ∆ctaA than in WT strains at both EE and S growth phases, suggesting altered surface integrity. The down-DAP B7HLA5, identified at EE growth phase, shares homologies with the B. subtilis RicA protein. RicA is a component of the RicAFT complex that senses the cellular redox status, and accelerates phosphorylation of the Spo0A transcriptional regulator [19]. Decreased abundance of RicA in the absence of CtaA could affect the ability of B. cereus to form biofilm and/or to sporulate [20,21]. Eight other down-DAPs identified at EE growth phase are linked to the iron acquisition system (Table S3). The proteins B7HR44 (DhbA/BacA), B7HR45 (DhbB/BacC), B7HR46 (DhbC/BacE), B7HR47 (DhbE/BacB) and B7HR48 (DhbF/BacF) are involved in synthesizing the iron-binding bacillibactin siderophore [22,23]. Related proteins were also down-regulated, including FeuA (B7HKU2), a siderophore-binding protein [24], and IlsA (B7HK52), a surface protein which plays an important role in iron acquisition in B. cereus [25]. In addition, Dps2 (B7HVX5), an iron-binding protein playing a role in iron storage and resistance to oxidative stress [26] was also among the down-DAPs. The significant decrease in abundance of these eight proteins in the ∆ctaA strain suggests down-regulation of siderophore-mediated iron uptake, potentially preventing accumulation of an excess of free intracellular iron which could lead to ROS generation via the Fenton reaction [27]. Interestingly, we noted the presence of a nitroreductase-like protein (B7HMT1) among the down-DAPs. Low amounts of nitroreductase could prevent the accumulation of its reduced substrate: quinones [28], which have also been reported to enhance the Fenton reaction [29].

Exoproteome Analysis
According to our quantitative and statistical criteria, only 21 exoproteins differentially accumulated between the strains (Table S4). These DAPs were mainly related to intracellular processes, and consequently are not classical secreted proteins [39]. However, six of them were identified both in the exoproteome and in the cellular proteome, includ- Among the 34 up-DAPS, three proteins were significantly more abundant in the ∆ctaA compared to the WT strain whatever the growth phase (Table S3). These were the UCPA oxidoreductase B7HLZ6 of unknown function, the disulfide bond-formation protein D precursor (BdbD, B7HU77), and the flavohemoglobin Hmp (B7HKH2). BdbD is involved in disulfide bond-formation in extra-cytoplasmic proteins, and contributes to various cellular processes, including maturation of cytochrome c and endospore maturation [30,31]. Hmp is known to protect bacteria from NO/redox stress [32], and is activated by the twocomponent ResDE system in response to reduced menaquinone accumulation [33]. ResDE also activates the cytochrome and heme biogenesis pathways. Accordingly, among up-DAPs, we found the first two subunits of quinol oxidase caa3-QoxB (B7HWC4) and QoxA (B7HWC3) (Figure 4)-alongside HemC (B7HQM4), which is involved in the heme subpathway that synthesizes coproporphyrinogen-III from 5-aminolevulinate. The other up-DAPs included: (i) two sulfatases (B7HKE1 and B7HKV9) and one sulfate adenyltransferase Sat (B7HKE6) that regulate and contribute to the sulfate assimilation pathway [34], which is known to be upregulated in response to oxidative stress [35]. (ii) An enzyme involved in molybdopterin co-factor (MoCo) biosynthesis (MoeB, B7HW37) that contributes to maintaining intracellular sulfur and thiol homeostasis [36] and prevents ROS damage [37]. (iii) a KefF quinone oxidoreductase-like protein (B7HUU9), that could decrease the redox toxicity of quinones and activate the potassium efflux system [38]. (iv) Eight sporulationassociated proteins comprising four spore components (B7I0D2, B7HXF1, B7HU59, B7I057) and four regulators of stages in the sporulation process-Spo0M (B7I0D2, stage 0), SpoII0Q (B7HY40, stage II), and RfsA and SpoIIIAN (B7HYE5 and B7HNU9, stage III). Up-regulation of these sporulation-associated proteins during exponential growth of the ∆ctaA mutant suggests that a signal that normally makes sporulation a post-exponential growth-phase response in WT strains could be detected earlier in ∆ctaA. Our proteomics data suggest that this signal could be redox stress, which ∆ctaA cells are exposed to from the beginning of growth.

Exoproteome Analysis
According to our quantitative and statistical criteria, only 21 exoproteins differentially accumulated between the strains (Table S4). These DAPs were mainly related to intracellular processes, and consequently are not classical secreted proteins [39]. However, six of them were identified both in the exoproteome and in the cellular proteome, including the membrane-associated iron ABC transporter (FeuA, B7HKU2) and the transmembrane quinol oxidase subunit 2 (QoxA, B7HWC4) ( Figure 5). These corroborating results confirm their abundance-changes in ∆ctaA compared to WT strain.

Resistance to Stress
Proteomics data suggested that in the absence of CtaA, having already activated their stress-response, cells should better resist additional exogenous stress. We assessed the ability of WT, ∆ctaA and complemented strains to survive at temperatures below the minimal growth temperature (i.e., below 8 • C [40]), and to resist exogenous oxidant. Viable colony forming units (CFU) counts decreased regularly over time during incubation at 4 • C for all strains, but the viability loss was significantly greater for WT and complemented strains than for the ∆ctaA strain (Figure 6a). Similarly, aerobically-grown ∆ctaA was more resistant to the deleterious effects of H 2 O 2 than either WT or complemented strains (Figure 6b). Taken together, these results suggest that CtaA deficiency effectively makes B. cereus more resistant to cold and oxidative stressors.

Resistance to Stress
Proteomics data suggested that in the absence of CtaA, having already activated their stress-response, cells should better resist additional exogenous stress. We assessed the ability of WT, ∆ctaA and complemented strains to survive at temperatures below the minimal growth temperature (i.e., below 8 °C [40]), and to resist exogenous oxidant. Viable colony forming units (CFU) counts decreased regularly over time during incubation at 4 °C for all strains, but the viability loss was significantly greater for WT and complemented strains than for the ∆ctaA strain (Figure 6a). Similarly, aerobically-grown ΔctaA was more resistant to the deleterious effects of H2O2 than either WT or complemented strains ( Figure  6b). Taken together, these results suggest that CtaA deficiency effectively makes B. cereus more resistant to cold and oxidative stressors.

Motility
Proteomics data relating to altered expression of flagellum-related proteins suggested that motility could be modified in the ∆ctaA strain. When inoculated in semisolid medium, the motile parental strain produced diffuse turbidity as it grew. In contrast, the ΔctaA mutant grew only along the line of inoculation (Figure 7), indicating that CtaA is required for B. cereus motility.

Motility
Proteomics data relating to altered expression of flagellum-related proteins suggested that motility could be modified in the ∆ctaA strain. When inoculated in semisolid medium, the motile parental strain produced diffuse turbidity as it grew. In contrast, the ∆ctaA mutant grew only along the line of inoculation (Figure 7), indicating that CtaA is required for B. cereus motility. coxon test, p <0.0001 for WT vs. ∆ctaA and complemented ∆ctaA vs. ∆ctaA. (b) B. cereus AH187 WT (blue), ∆ctaA (red) and complemented ∆ctaA(pctaA) (green) cells were grown in MODG at 30 °C to the mid-exponential growth phase, before exposure to 10 mM H2O2 for 5, 10, or 20 min. Surviving CFU were counted and expressed as (N/N0) × 100. Values correspond to mean ± SD measured for three biological replicates. Statistical analysis was performed by one-way ANOVA followed by Tukey's post hoc analysis. *: p < 0.05.

Motility
Proteomics data relating to altered expression of flagellum-related proteins suggested that motility could be modified in the ∆ctaA strain. When inoculated in semisolid medium, the motile parental strain produced diffuse turbidity as it grew. In contrast, the ΔctaA mutant grew only along the line of inoculation (Figure 7), indicating that CtaA is required for B. cereus motility.

Surface Properties
Proteomics analysis revealed several modifications to the abundance of proteins involved in cell wall, membrane and envelope biogenesis in the ∆ctaA mutant compared to its parental strain, suggesting surface alterations in this mutant.
In particular, a putative component of the S-layer (B7HXP4) was expressed at lower levels in the ∆ctaA mutant compared to the WT strain. To confirm these findings, we extracted non-covalently attached proteins (S-layer fraction) from the surface of bacteria for

Surface Properties
Proteomics analysis revealed several modifications to the abundance of proteins involved in cell wall, membrane and envelope biogenesis in the ∆ctaA mutant compared to its parental strain, suggesting surface alterations in this mutant.
In particular, a putative component of the S-layer (B7HXP4) was expressed at lower levels in the ∆ctaA mutant compared to the WT strain. To confirm these findings, we extracted non-covalently attached proteins (S-layer fraction) from the surface of bacteria for western blot analysis. B7HXP4 was confirmed to be less abundant in the S-layer fraction from the ∆ctaA strain compared to WT and complemented strains ( Figure S2).
As autoaggregation is related to cell-surface characteristics [41], we assessed the capacity of ∆ctaA and WT cells to autoaggregate by performing sedimentation assays. The OD 600 of bacterial culture suspensions was monitored during static incubation [41]. Most WT cells settled to the bottom of the tube over the course of incubation for 8 h, whereas ∆ctaA mutant cultures remained turbid (Figure 8a). Sedimentation kinetics revealed that WT cells aggregated rapidly, reaching 81.1 ± 4.4% autoaggregation after 8 h, whereas ∆ctaA have lost the autoaggregation phenotype (Figure 8b).

Adhesion/Biofilm
As both motility and autoaggregation can contribute to biofilm formation [42], we used the BioFilm Ring Test ® [43] to assess whether the ∆ctaA mutant could attach to a solid surface and develop a sessile biomass. Both ∆ctaA and WT strains formed biofilm on microplates after 24 h of incubation. However, the kinetics of biofilm formation differed between the two strains ( Figure 9). Thus, although after 4 h at 25 • C and 30 • C ∆ctaA cells had formed a biofilm (∆BFI > 17, Figure 9c,d), this biofilm did not persist over time, as revealed by the decrease in ∆BFI at 8 h. In contrast, this value increased progressively for WT cells from 8 h incubation, reflecting continual, persistent biofilm formation. Taken together, these results indicate that ∆ctaA cells form a biofilm faster than WT cells, but that the biofilm formed is weaker than that formed by WT cells. This weakness could be linked to the inability of ∆ctaA cells to autoaggregate [44].
As autoaggregation is related to cell-surface characteristics [41], we assessed the capacity of ∆ctaA and WT cells to autoaggregate by performing sedimentation assays. The OD600 of bacterial culture suspensions was monitored during static incubation [41]. Most WT cells settled to the bottom of the tube over the course of incubation for 8 h, whereas ΔctaA mutant cultures remained turbid (Figure 8a). Sedimentation kinetics revealed that WT cells aggregated rapidly, reaching 81.1 ± 4.4% autoaggregation after 8 h, whereas ΔctaA have lost the autoaggregation phenotype (Figure 8b). . Sedimentation assays. Overnight cultures in BHI medium were adjusted to an OD600 of 1, and then incubated under static conditions in a spectrometry cuvette at room temperature. Absorbance (OD600) was monitored for 8 h. Data are expressed as percentages of initial OD600. Experiments were performed in triplicate. Statistical significance of the observed differences was determined by two-way ANOVA with Bonferroni post hoc test. ****: p < 0.0001.

Adhesion/Biofilm
As both motility and autoaggregation can contribute to biofilm formation [42], we used the BioFilm Ring Test ® [43] to assess whether the ΔctaA mutant could attach to a solid surface and develop a sessile biomass. Both ΔctaA and WT strains formed biofilm on microplates after 24 h of incubation. However, the kinetics of biofilm formation differed between the two strains ( Figure 9). Thus, although after 4 h at 25 °C and 30 °C ∆ctaA cells had formed a biofilm (ΔBFI > 17, Figure 9c,d), this biofilm did not persist over time, as revealed by the decrease in ΔBFI at 8 h. In contrast, this value increased progressively for WT cells from 8 h incubation, reflecting continual, persistent biofilm formation. Taken together, these results indicate that ∆ctaA cells form a biofilm faster than WT cells, but that the biofilm formed is weaker than that formed by WT cells. This weakness could be linked to the inability of ∆ctaA cells to autoaggregate [44].

Sporulation
Like biofilm formation, sporulation is a strategy used by B. cereus to adapt to and survive a variety of stresses [45,46]. We found no difference in the ability of WT and ∆ctaA strains to sporulate (data not shown), in contrast to reports for B. subtilis [3].

Sporulation
Like biofilm formation, sporulation is a strategy used by B. cereus to adapt to and survive a variety of stresses [45,46]. We found no difference in the ability of WT and ∆ctaA strains to sporulate (data not shown), in contrast to reports for B. subtilis [3].

Discussion
The aim of this study was to investigate how mutations in the ctaA gene-identified during a proteomics screen of bacteria surviving in groundwater-affected the capacity of B. cereus AH187 to grow and resist exogenous stress.
Pathogenic variants often emerge due to increased fitness, which can result from point mutations [47,48]. CtaA is a target gene for spontaneous mutation in both B. cereus [2] and B. subtilis [3]. This gene encodes an integral transmembrane protein displaying heme O monooxygenase activity. The results presented here show that ctaA gene deletion inactivates cytochrome caa3 oxidase in B. cereus AH187 (Figure 2b). Consequently, we can conclude that CtaA plays a role in cytochrome caa3 biogenesis in B. cereus AH187, and that cytochrome caa3 activity requires heme A. Like caa3, cytochrome aa3 quinol oxidase binds heme A. However, in the absence of heme A, the active bo3 variant, which binds heme B and heme O instead of heme A can replace cytochrome aa3 oxidase [8]. Our proteomics data indicate that cytochrome aa3 apoprotein, encoded by QoxABCD, is over-produced in ctaA mutants. We hypothesize that disruption of the bc-caa3 branch of the respiratory network could redirect electron flow toward menaquinol-cytochrome bo3 oxidase. This redirection probably generates redox stress, as reflected by increased levels of key proteins involved in the redox stress-response. These proteins include the flavohemoglobin Hmp, which may promote cytochrome bo3 activity [49]. Redox stress is associated with reorganization of the respiratory network, and could result from excess ROS production due to blocked electron flow within the bc complex leading to accumulation of reduced quinone [50,51]. However, it is also possible that loss of CtaA disrupts plasma membrane integrity, leading to secondary redox stress [52].
B. cereus AH187 ∆ctaA exhibited a small-colony phenotype on LB solid media (Figure 2a). A similar phenotype was observed in B. subtilis and S. aureus ctaA mutants [3,53], and has been demonstrated to allow bacteria to survive in hostile environments, by reducing metabolic needs [54]. When grown in liquid medium supplemented with glucose, B. cereus AH187 ∆ctaA displayed no significant growth defect, in line with what was reported for the spontaneous ctaA mutant of B. cereus strain 9373 [55]. This capacity to grow in suspension is probably linked to the ability of bacteria to increase acetate overflow to overcome respiration dysfunction. Indeed, increased acetate overflow could prevent NADH/NAD + imbalance in the respiratory chain and allow faster ATP production, through acetate kinase activity, to meet the requirements for biomass [56,57]. Overflow metabolism has been suggested to be advantageous for bacteria using this strategy to compete with fully respiration-competent cells in both nutrient-rich and -poor medium [58]. Increased overflow metabolism, due to loss of CtaA, could thus increase the overall fitness of bacteria.
Indeed, our results show that CtaA deficiency makes B. cereus more resistant to cold and oxidative stresses, suggesting that ctaA deletion generates an effective redox response to various types of stress [59].
Interestingly, loss of CtaA also changed the surface composition of B. cereus. For example, levels of one of the putative S-layer proteins (B7HXP4) was decreased 2-fold (Table S3). Production of the S-layer demands a high metabolic investment from micro-organisms [60], mainly due to the high synthesis rate of its proteins (means of normalized spectral abundance factor (NSAF) for B7HXP4 = 0.59 ± 0.18%, 0.50 ± 0.10% and 0.73 ± 0.13% at EE, LE and S-growth phase in WT strain, Table S1). The flagellin protein FliC is also produced in large quantities (means of NSAF = 0.37 ± 0.03%, 1.70 ± 0.45% and 1.50 ± 0.03% at EE, LE and S-growth phase in WT strain, Table S1), and was similarly found to be 2-to 4-fold less abundant in ∆ctaA compared to the WT strain (Table S3). Thus, ctaA deletion affected B. cereus motility, as reported in S. aureus [61]. By decreasing the synthesis of S-layer protein and flagellin, bacteria may be attempting to conserve energy that can then be diverted to mount a stress-response [62] and promote survival through other processes, such as sporulation.
Flagella and surface proteins are involved in autoaggregation of planktonic bacteria, and biofilm formation [41]. Both these processes play important roles in bacterial survival in their natural environment and in diseases [63]. Our results indicate that the ∆ctaA strain lost the autoaggregation phenotype but retained its ability to form biofilm on a solid surface. However, the biofilm formed by ∆ctaA cells appeared weaker than that formed by WT cells, probably due to the loss of the autoaggregation phenotype. Lack of autoaggregation within a biofilm could be a survival advantage for bacterial cells in limited nutrient conditions [41].
In conclusion, CtaA synthesis is dispensable for B. cereus growth under oxic conditions as bacteria can implement an effective compensation strategy. This strategy is closely linked to the flexibility of the respiratory network and the bacteria's ability to generate a respiration-dysfunction-mediated stress-response. Considering the phenotypes of ∆ctaA cells, we propose that spontaneous inactivation of ctaA could improve B. cereus' fitness in limited nutrient conditions, such as those encountered in groundwater, as in the study where we initially identified mutation of this gene.
Better understanding how bacterial aerobic respiration, and terminal oxidases, participate in pathogen fitness in nutrient-limited environment is an exciting research challenge. Proteomics approaches offer great potential for characterizing respiratory chain in model strains. However, further studies utilizing multi-omics approaches are necessary to address the mechanisms by which bacteria regulate their metabolism in response to respiration dysfunction.

Bacterial Strain and Mutant Construction
The wild-type Bacillus cereus AH187 (F4810/72) strain used in this study originates from emetic food poisoning outbreak [64]. Deletion of the ctaA (BCAH187_A4064) gene was achieved by allelic replacement, using the temperature-sensitive pMAD plasmid [65]. Briefly, 1-kbp flanking DNA sequences upstream and downstream of the BCAH187_A4064 gene were amplified using the appropriate oligonucleotide primers (Table S5). The recombinant PCR products containing DNA sequences upstream and downstream of the BCAH187_A4064 gene were cloned into the pCR-TOPO 2.1 plasmid (TOPO T/A cloning kit, Invitrogen). The resulting plasmid, pctaA-KO, was digested with SmaI (Promega, Charbonnières-les-Bains, France) and ligated with a SmaI-digested DNA fragment encoding spectinomycin resistance. The new plasmid, pctaA-KO-spec, was digested with EcoRI (Promega, Charbonnières-les-Bains, France), and the resulting fragment was cloned into pMAD digested by the same enzyme. The recombinant plasmid, pMAD ctaA-KO-spec, was used to transform B. cereus AH187. Chromosomal allele exchange was confirmed by PCR with oligonucleotide primers located upstream and downstream of the DNA regions used for allelic exchange (ExF4064 and ExR4064, Table S5, Figure S3). For complementation assays, the plasmid pHT304-ctaA [2] was electroporated in ∆ctaA strain and then transformants were selected on LB plate containing erythromycin (Sigma Aldrich, Saint Louis, CA, USA) and confirmed by PCR.

Growth Parameters and Analytical Procedures
Growth of B. cereus WT and ∆ctaA strains was performed as described previously [2], and monitored spectrophotometrically at 600 nm (BioSpec-mini, Shimadzu Biotech). Growth parameters in MOD medium [14] supplemented with 30 mM glucose or 60 mM glycerol (all from Sigma Aldrich, Saint Louis, CA, USA), in the presence or absence of 5 µM antimycin A (from Streptomyces sp., Sigma Aldrich, Saint Louis, CA, USA) were studied on microtiter plates in a temperature-controlled, automated optical density reader (Flx-Xenius XMA, Safas, Monaco). The maximal specific growth rate (µ max ) was calculated using the modified Gompertz equation [66]. Glucose and acetate concentrations were determined in filtered supernatants using enzymatic kits purchased from Biolabo (Maizy, France) and BioSenTec (Auzeville Tolosane, France), respectively. Kits were used according to the manufacturer's protocols.

TMPD Oxidase Staining
The presence of active c-type cytochrome oxidase was verified by a colorimetric assay using N, N, N , N -tetramethyl-p-phenylenediamine (TMPD) as an artificial electron donor that can be oxidized by cytochrome caa3 to a blue colored product that stains colonies. TMPD oxidase staining was performed by adding drops of oxidase reagent (bioMerieux, Craponne, France) to B. cereus colonies grown overnight at 30 • C on LB plates.

Sample Preparation for Shotgun Proteomics
WT and ∆ctaA strains were grown in MODG medium, as previously described [14]. For proteomics analysis, cultures (three biological replicates) were performed in 2-L flasks containing 500 mL culture medium. Flasks were incubated with shaking (200 rpm) at 30 • C. The inoculum was a sample of an overnight culture harvested by centrifugation, washed and diluted in fresh medium to obtain an initial optical density at 600 nm (OD 600 ) of 0.02. Samples (100 mL) were collected at EE (OD 600 = 0.1), LE (OD 600 = 1) and S growth phases (OD 600 = 1.5) ( Figure S1). Protein and peptide samples from cells and culture supernatants were prepared, as previously described [2].

Protein Identification by LC-MS/MS and Label-Free Quantification
Peptides were separated on an Ultimate 3000 nano LC system coupled to a Q-Exactive HF mass spectrometer (Thermo Fisher Scientific, Illkirch-Graffenstaden, France) for analysis. Briefly, peptide mixtures (10 µL) were loaded, desalted online on a reverse-phase Acclaim PepMap 100 C18 precolumn (5 mm, 100 Å, 300 µm i.d. × 5 mm), and then resolved according to their hydrophobicity on a nanoscale Acclaim Pepmap 100 C18 column (3-µm bead size, 100-Å pore size, 75 µm i.d. × 500 mm) at a flow rate of 200 nL·min −1 using a bi-modal gradient combining buffer B (0.1% HCOOH, 80% CH 3 CN, 20% H 2 O) and buffer A (0.1% HCOOH, 100% H 2 O). Peptide digests of cellular proteins were eluted by applying a 90-min gradient (4-25% B in 75 min, followed by 25-40% B in 15 min), whereas extracellular proteins were eluted by applying a 60-min gradient (4-25% B in 50 min, followed by 25-40% B in 10 min). A Top-20 method was used, with full MS scans acquired in the Orbitrap mass analyzer over an m/z range from 350 to 1500, at 60,000 resolution. After each scan, the 20 most abundant precursor ions were sequentially selected for fragmentation and MS/MS acquisition at 15,000 resolution. A 10-s dynamic exclusion window was applied to increase the detection of low-abundance peptides. Only double-and triple-charged ions were selected for MS/MS analysis.
Sequences were assigned using the Mascot Daemon search engine (version 2.5.1, Matrix Science) against the B. cereus AH187 NCBI_20200622 database (7100 sequences). Peptide mass tolerance and MS/MS fragment mass tolerance were set to 5 ppm and 0.02 Da, respectively. The search included carbamidomethylation of cysteine residues (C) as fixed modification; oxidation of methionine (M) and deamidation of asparagine and glutamine (NQ) were included as variable modifications. All peptide matches with a peptide score associated with a Mascot p-value lower than 0.05 were retained. Proteins were considered valid when at least two distinct peptides were detected in the same sample, resulting in a false discovery rate lower than 1%. NSAF values were calculated by dividing the number of spectra assigned to a protein in a given sample by its molecular weight, as recommended [67]. Results were then statistically analyzed. The R tool Bioconductor DEP package (version 1.12.0) was used to perform PCA and determine changes in protein abundance between WT and ∆ctaA mutant at the different growth phases [68]. Significant changes were selected where the adjusted p-value was lower than 0.05 and the |fold-change| ≥ 1.5 (|log 2 fold-change| ≥ 0.56).
The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium via the PRIDE partner repository with the dataset identifier PXD030118 and 10.6019/PXD030118 for cellular proteome of B. cereus AH187, PXD030114 and 10.6019/ PXD030114 for cellular proteome of ∆ctaA mutant, PXD030165 and 10.6019/PXD030165 for exoproteome of B. cereus AH187 and PXD030163 and 10.6019/PXD030163 for exoproteome of ∆ctaA mutant.

S-Layer Extraction and Western Blot Analysis
S-layer extraction was performed as described in [69]. Culture samples of WT, ∆ctaA and complemented ∆ctaA(pctaA) strains harvested at EE, LE and S growth phases were centrifuged for 5 min at 8000× g. Pellets were washed with PBS and boiled (100 • C) for 10 min in 110 µL PBS-3 M urea to extract S-layer and S-layer-associated proteins. Extracts were then centrifuged 10 min at 16,000× g, and the S-layer extracts were separated from bacteria pellets. S-layer extracts were added to 1/5 volume of sample buffer (10% SDS, 1% β-mercaptoethanol, 50% glycerol, 300 mM Tris-HCl (pH 6.8), 0.1% bromophenol blue). The protein content of cell lysates was estimated by BCA assay (Pierce). S-layer extract aliquots (2 µg) were separated on 10% SDS-PAGE gels, and transferred to nitrocellulose membranes (Thermo Fisher Scientific, Illkirch-Graffenstaden, France) for immunoblotting. Proteins were detected using rabbit antiserum raised against purified B7HXP4 (Boutonnet et al., in preparation). Immunoreactive products were revealed by chemiluminescent detection after incubation with horseradish peroxidase (HRP)-conjugated anti-rabbit antibody (Sigma-Aldrich, Saint-Louis, CA, USA).

Cell Survival at 4 • C
Overnight cultures of WT, ∆ctaA and complemented ∆ctaA(pctaA) strains were inoculated in 15 mL MODG media in 50 mL tubes at an initial OD 600 of 0.02. Cultures were incubated with shaking (200 rpm) at 30 • C. Once cultures had reached exponential phase, the tubes were incubated at 4 • C with shaking for up to 576 h (24 days). The number of surviving CFU was determined by plating 100-µL volumes of 10-fold serial dilutions of cultures on LB agar plates. The colonies formed after incubation for 18 h at 30 • C were counted. All experiments were performed in triplicate.

Hydrogen Peroxide Killing Assays
WT, ∆ctaA and complemented ∆ctaA(pctaA) strains were grown to mid-log phase (OD 600~0 .3) in MODG medium. Hydrogen peroxide challenge assays were performed by exposing samples to 10 mM H 2 O 2 for 5, 10 or 20 min. Cells were then centrifuged and resuspended in an equal volume of H 2 O. Sample aliquots (100 µL) were appropriately diluted in H 2 O and plated on LB agar. CFUs were counted after overnight incubation at 30 • C. All experiments were performed at least in triplicate.

Autoaggregation
WT and ∆ctaA strains were grown overnight in Brain Heart Infusion (BHI) at 30 • C. OD 600 values were adjusted to 1, and 1 mL of culture was placed in spectrophotometer cuvette. Optical density (OD 600 ) was monitored over 8 h static incubation. Results were expressed as percentage of initial OD 600 . Experiments were performed in triplicate.

Bacterial Adhesion (BioFilm Ring Test ® )
The BioFilm Ring test [43] measures the immobilization of magnetic beads by attached cells. The more beads entrapped by cells, the fewer remain mobile. The size of the spot of free beads formed upon magnetization of the plate can be used to estimate the number of cells engaged in forming biofilm on the plate's surface.
WT and ∆ctaA strains were grown in BHI at 25 and 30 • C for 24 h. Cultures were adjusted to 2 × 10 6 CFU/mL, and 200 µL was distributed in 96-well polystyrene plates. No bacteria were added to control wells. Adhesion was measured after 4, 8 and 24 h incubation. The ability of each strain to adhere was assessed based on the BioFilm Index (BFI), calculated using the Biofilm Control software (Biofilm Control, Saint Bauzire, France) from the size of the black spot in the bottom of the wells detected by the Scan Plate Reader. BFI values are inversely proportional to attached cell number. ∆BFI (BFIcontrol−BFIsample) was calculated by subtracting the BFI for each sample from the mean BFI obtained for the controls-containing no bacteria-to assess the ability of strains to adhere. Four replicates (four wells) were analyzed for each strain and condition tested.

Statistical Analyses
Data from three biological replicates were pooled for statistical analyses. Comparisons of multiple data were analyzed by analysis of variance (ANOVA) followed by post hoc analysis (one-way ANOVA followed by Tukey's post hoc analysis for stress response studies, two-way ANOVA followed by Bonferroni post hoc analysis for biofilm studies). Changes in autoaggregation ability and metabolite production were evaluated using Student's t-test. Statistical analyses were performed using GraphPad Prism software version 6.0 (GraphPad Software, San Diego, CA, USA). p-values ≤ 0.05 were considered significant.