The regulatory network comprising ArcAB‐RpoS‐RssB influences motility in Vibrio cholerae

The diarrheal disease cholera is caused by the versatile and responsive bacterium Vibrio cholerae, which is capable of adapting to environmental changes. Among others, the alternative sigma factor RpoS activates response pathways, including regulation of motility‐ and chemotaxis‐related genes under nutrient‐poor conditions in V. cholerae. Although RpoS has been well characterised, links between RpoS and other regulatory networks remain unclear. In this study, we identified the ArcAB two‐component system to control rpoS transcription and RpoS protein stability in V. cholerae. In a manner similar to that seen in Escherichia coli, the ArcB kinase not only activates the response regulator ArcA but also RssB, the anti‐sigma factor of RpoS. Our results demonstrated that, in V. cholerae, RssB is phosphorylated by ArcB, which subsequently activates RpoS proteolysis. Furthermore, ArcA acts as a repressor of rpoS transcription. Additionally, we determined that the cysteine residue at position 180 of ArcB is crucial for signal recognition and activity. Thus, our findings provide evidence linking RpoS response to the anoxic redox control system ArcAB in V. cholerae.

RpoS was first described as being associated with the survival of V. cholerae following exposure to hydrogen peroxide or artificial seawater (Yildiz & Schoolnik, 1998).Later, RpoS was linked to a more defined role, namely the mucosal escape response (Nielsen et al., 2006).Furthermore, rpoS is known to be expressed under nutrient-poor conditions, such as late-stage infections, and induces transcription of motility-and chemotaxis-related genes.Nielsen et al. (2006) showed that an rpoS deletion mutant is impaired to efficiently detach from its site of colonisation and return to the environment.Additionally, RpoS was found to facilitate detachment from biofilms (Muller et al., 2007).Similar to that seen in E. coli, stability of the RpoS protein is regulated by the anti-sigma factor RssB (Wurm et al., 2017), while rssB transcription is regulated via RpoS.
Subsequent elevation of RssB levels in cells initiates the proteolysis of RpoS.In contrast to its status in E. coli, the RssB protein in V. cholerae is degraded and so far no Ira proteins have been identified in V. cholerae (Wölflingseder et al., 2022).Furthermore, RssB appears to act as a substrate for phosphorylation.Recently, we reported that the mutant rssB D75A (amino acid exchange mutation of the possible phosphorylation site) exhibits phenotypes similar to those corresponding to an rssB-knockout mutant.This suggests that RssB is fully activated only when phosphorylated by a kinase (Wölflingseder et al., 2022).Until now, no such kinase has been recognised in V. cholerae, and thus the identification and characterisation of an RssB kinase was the objective of the current study.
ArcB, the sensor histidine kinase of the anoxic redox control (Arc) two-component system, phosphorylates ArcA and RssB in E. coli (Georgellis et al., 1997;Mika & Hengge, 2005).As the name suggests, the Arc system responds to the redox state of a cell (Brown et al., 2022).It not only senses the availability of oxygen (Malpica et al., 2004;Malpica et al., 2006) but also the energy state of the cell, thereby contributing to starvation strategies (Mika & Hengge, 2005).During the transition between exponential to stationary growth phase, bacterial cultures switch to starvation, where TCA cycle activity is reduced via ArcAB-dependent regulation to prevent generation of potentially harmful metabolic by-products (Nyström et al., 1996).The signal to inactivate ArcB is transmitted via its two cysteine residues (C180 and C241).Electrons are transferred from C180 of ArcB to the oxidised quinone pool, entailing an intermolecular disulphide bond between the respective cysteine residues of the two ArcB monomers.This intermolecular disulphide bond drastically reduces ArcB activity.If oxygen availability is high, the electrons are further withdrawn from the quinone pool, preserving its oxidised state.This introduces a second disulphide bond between C241 of these two ArcB proteins, resulting in an inactive ArcB kinase (Georgellis et al., 2001;Malpica et al., 2004;Malpica et al., 2006).Similar ArcAB homologues exist in V. cholerae (Heidelberg et al., 2000).Thus far, it is known that ArcA is involved in the regulation of virulence, biofilm formation and motility in V. cholerae (Li et al., 2022;Sengupta et al., 2003;Xi et al., 2020;Zhou et al., 2021).Aside from phosphorylation by ArcB, ArcA activity is modulated by the formation of intramolecular disulphide bonds.This reflects the flexibility of the Arc system as a regulatory pathway potentially important for survival in aquatic environments as well as in human hosts (Zhou et al., 2021).
In this study, we describe ArcB as the sensor histidine kinase responsible for phosphorylating RssB in V. cholerae.Regulation of RpoS protein stability via the modulation of RssB activity does not appear to be the only connection between the ArcAB system and the RpoS-RssB network because rpoS transcription is also altered in arc-knockout mutants.These two regulatory networks, RpoS-RssB and the ArcAB system, are interwoven and modulate various signalling pathways in V. cholerae.In this study, we observed certain similarities in the transcriptional and post-transcriptional regulation of rpoS between E. coli and V. cholerae.However, the differences between the output activities of RpoS in these two bacterial species indicate that regulatory consequences may be species specific.

| RpoS proteolysis depends on the ArcB kinase
To better understand the mechanisms regulating RpoS proteolysis in V. cholerae, we characterised the activation pathway of RssB.As mentioned earlier, RssB must be phosphorylated to become fully active, which is substantiated by the fact that a point mutation of a putative phosphorylation site in RssB (D75A) leads to decreased RpoS proteolysis (Wölflingseder et al., 2022).Previous studies have demonstrated that the ArcB kinase is responsible for the phosphorylation of RssB in E. coli (Mika & Hengge, 2005).However, it was suggested that an ArcB (VC2369) homologue did not influence RpoS stability in V. cholerae (Wurm et al., 2017).A re-evaluation of the ΔarcB mutant revealed a clear phenotype of a less active RssB, as indicated not only by a more stable RpoS protein under nutrient-shift conditions but also by increased motility (see below).Nutrient-shift conditions, which are known to be important for RpoS expression and turnover, were selected to monitor protein stability (Wölflingseder et al., 2022).Thus, the bacteria were shifted from nutrient-rich (LB medium) to nutrient-poor (M9 minimal-medium +0.2% maltose) conditions.A polyclonal RpoS antibody was used to detect RpoS protein levels.Samples for immunoblot analysis were obtained at defined time points and analysed using densitometry (shown as the ratio of RpoS level at time point 180 to 20 min).Polyacrylamide gels stained with CBB as described (Kang et al., 2002) served as loading control (Figure S1).Although the RpoS protein was more stable in the ΔarcB mutant than in the WT, it was not as stable as in the ΔrssB deletion strain (Figure 1a,b), suggesting that ArcB may be involved in RpoS degradation, but not solely so.

| ArcB kinase is phosphorylating RssB
In vitro phosphorylation experiments were performed to provide further evidence that ArcB functions as a kinase of RssB.For this purpose, increasing amounts of a recombinant and truncated form of purified ArcB were incubated with RssB in the presence or absence of ATP, according to methods described previously (Georgellis et al., 1997;Mika & Hengge, 2005).Following incubation, protein samples were separated using phosphate-affinity SDS-PAGE (Phos-tag).RssB is detectable in two protein forms, an upper minor and a lower major band (Wölflingseder et al., 2022).
However, as ArcB concentration increases in the presence of ATP, a third RssB band identified as RssB-P, became visible (Figure 2).Incubation of RssB with ATP alone did not yield the RssB-P band.
Using the same procedure, the aspartate at position 75 (D75) of RssB was identified as the phosphorylation site, since no assignable protein band corresponding to RssB-P was detectable for the RssB D75A protein (Figure 2).The same samples were also applied to SDS-PAGE as a control and no differences in the protein pattern could be detected (Figure S2).Furthermore, it could also be excluded that the additional protein band is the truncated form of ArcB, since ArcB has a molecular weight of 88 kDa (truncated form approximately 80 kDa, see polyacrylamide gel stained with CBB of XL1 pQE30arcB 78-785 in Figure S3), while the RssB protein has approximately 40 kDa.

| ArcB-dependent motility phenotype is linked to RssB function
Next, the motility of the ΔarcB mutant strain was investigated under nutrient-poor conditions on M9-based swarming agar plates.
Under these conditions, the alternative sigma factor RpoS is crucial for regulating the expression of motility-and chemotaxis-related genes (Nielsen et al., 2006).Thus, changes in the protein stability of RpoS can be monitored via altered motility behaviour.The ΔarcB, rssB D75A and ΔrssB mutant strains showed higher motility compared to the WT, with a gradual increase in motility from ΔarcB to rssB D75A and ΔrssB according to RpoS stability (Figure 3a,b) (Wölflingseder et al., 2022).Furthermore, the hypermotile phenotype of the ΔarcB mutant was complemented by the expression of arcB from the pBAD plasmid (Figure 3c,d).
F I G U R E 1 RpoS stability in WT, ΔarcB and ΔrssB strains.(a) Nutrient-shift experiments (shift from LB medium to M9 minimal medium supplemented with 0.2% maltose) were performed with the WT, ΔarcB and ΔrssB strains.At indicated time points (as stated in the figure), samples for immunoblot analyses were harvested and RpoS levels were detected by using a polyclonal RpoS antibody.(b) Immunoblots were used to perform densitometry.Therefore, the ratio of the RpoS levels between time points 180 min and 20 min was calculated, and RpoS levels were normalised to an RpoS-independent background band (•).Mean values ± SD of six replicates are shown.Significant differences are marked by an asterisk (one-way anova followed by a Dunnett's multiple-comparisons test, *p < 0.05).

F I G U R E 2
In vitro phosphorylation assays of RssB.Purified RssB or RssB D75A proteins (1 μg) were incubated with ArcB (0.025-0.1 μg, as indicated in the figure) in the absence and presence of ATP (2 mM).As control for specific phosphorylation via ArcB, RssB and RssB D75A alone were incubated with ATP.After 30 min of incubation, samples were separated using phosphate affinity SDS-PAGE (Phos-tag) and stained with CBB.Further experimental details can be found in the Materials and Methods section.

| ArcA negatively regulates rpoS transcription, while arcA-knockout mutation leads to hypomotility
In E. coli, ArcA, ArcB and RssB reportedly form part of a "threecomponent system" that keeps the RpoS levels low in the logarithmic growth phase.On the one hand, ArcB phosphorylates RssB and activates RpoS proteolysis, while ArcA, on the other hand, is phosphorylated by ArcB, and acts as a transcriptional repressor of rpoS (Mika & Hengge, 2005).To investigate whether the ArcAB system functions similarly in V. cholerae, rpoS transcription was measured in WT, ΔarcA (VC2368), ΔarcB and ΔarcAB strains via PhoA assays using rpoS-phoA transcriptional fusions.The rpoS transcription rates in all tested deletion mutants were significantly enhanced compared to those in the WT (Figure 4a).Additionally, the effects exerted by ArcA on arcB transcription were tested (Figure S4).A transcriptional fusion of arcB to phoA similar to that for rpoS was constructed in the WT and the ΔarcA mutant, and the transcription rate was measured.The results did not show a significant difference between the two strains tested.
To further elucidate the role of the Arc system in V. cholerae, the motility behaviour of arc deletion mutants was tested under nutrient-poor conditions (Figure 4b,c).As stated above, the ΔarcB mutant displayed a hypermotile phenotype.Hypermotility of the ΔarcAB double-knockout mutant was similar to that of the ΔarcB mutant.By contrast, the ΔarcA deletion strain was less motile than the WT.Notably, the ΔarcA, ΔarcB and ΔarcAB mutants exhibited a slight delay in growth in M9 minimal medium compared to that in WT (Figure S5a).Expression of arcA derived from the pBAD plasmid expression system resulted in complementation of the ΔarcA motility phenotype (Figure 5a,b).Furthermore, the hypermotile phenotype of ΔarcAB was complemented by expressing arcAB from the pBAD plasmid (Figure 5c,d).Although only 0.001% arabinose was used to induce plasmid-derived expression, the phenotype of ΔarcAB pBADarcAB appeared to be slightly over-complementing compared to the WT pBAD strain.

| arcA-knockout mutation leads to enhanced (RssB depending) RpoS proteolysis
To further characterise the apparent hypomotility phenotype of the ΔarcA mutant, RpoS protein stability was examined via chloramphenicol (Cm)-dependent stability assays.Cm, an antibiotic that inhibits the synthesis of new proteins (Studemann et al., 2003), was added to the bacterial cultures; samples were harvested and subjected to immunoblot analysis at defined time points.Polyacrylamide gels stained with CBB (Kang et al., 2002) served as a loading control (Figure S6).
Compared to WT, less RpoS and accelerated proteolysis were observed in the ΔarcA mutant (Figure 6a).Almost no RpoS protein could be detected in the ΔarcA mutant 15 min after the addition of Cm.
This suggested that the RpoS protein was more labile without ArcA, which was not observed when rssB was knocked out in addition to arcA (ΔarcAΔrssB in Figure 6a).These Cm-dependent stability assays were performed in LB medium, while the swarming agar plates were based on M9 minimal-medium conditions.For added consistency, the RpoS protein levels were also determined under nutrient-shift conditions.Interestingly, 60 min following the shift to M9 minimal medium (+0.2% maltose), a lower amount of RpoS was detectable in the ΔarcA mutant compared to WT, whereas the RpoS level in the ΔarcAΔrssB double mutant was elevated (Figure 6b, Figure S7).Cm-dependent protein stability assays performed under the same nutrient-shift conditions (Figure S8), Cm being added 60 min after shift to M9 minimal medium (+0.2% maltose), revealed similar tendencies as those that were observed under nutrient-rich conditions.
In addition, the motility behaviour of these mutants was determined under nutrient-poor conditions.Although the ΔarcAΔrssB mutant was more motile than the WT and ΔarcA strains, it was not as motile as the ΔrssB mutant (Figure 6c,d).Since the growth patterns of the ΔarcAΔrssB mutant were similar to those of WT (Figure S5a), the effect of growth could be excluded.Furthermore, the enhanced RpoS level of the ΔarcAΔrssB strain was indirectly measured by determining the transcription rate of VC1248 in the WT, ΔarcA and ΔarcAΔrssB strains (Figure S9).VC1248 is an RpoS-regulated gene (Nielsen et al., 2006).The transcription rate of VC1248 in the ΔarcAΔrssB double mutant was increased compared to that in the WT and the ΔarcA mutant, which expressed similarly high levels of VC1248-phoA.
Considered together, these data revealed that although rpoS transcription in the arcA-knockout mutant was enhanced, its RpoS protein level was decreased compared to that in WT.

| Regulation of ArcB activity is dependent on the redox state of its two cysteine residues
Similar to those seen in E. coli, ArcB of V. cholerae has a total of two cysteine residues, C180 and C427, the sequence alignments of which are shown (Figure S10).Kinase activity of E. coli depends on dimer formation mediated by a leucine zipper domain (Nuñez Oreza et al., 2012) as well as on the redox state of the cysteine residues (Malpica et al., 2004).ArcB becomes inactive in the oxidised form by the formation of intermolecular disulphide bonds between two ArcB proteins, whereas it is active in the reduced form due to the presence of SH/thiol groups (Malpica et al., 2004).Thus, in order to generate putatively active ArcB forms, we constructed the following point mutations in arcB: arcB C180S , arcB C427S and arcB C180SC427S (further termed as arcB CC ), and checked for apparent phenotypes.
First, the motility behaviour under nutrient-poor conditions signified the importance of C180 for ArcB activity, as evidenced by the arcB C180S mutant being hypomotile while the arcB C427S mutant showed a motility similar to that of the WT (Figure 7a,b).When both cysteine residues were mutated (arcB CC ), the motility behaviour was comparable to that of arcB C180S .Furthermore, the growth of all three arcB point mutants was similar to that of the WT (Figure S5b).
Next, we aimed to determine whether the hypomotility of arcB CC was caused by altered rpoS transcription via ArcA, changed RpoS stability via RssB or both.For this purpose, the rpoS transcription rates in WT, arcB C180S , arcB C427S and arcB CC strains were measured (Figure 7c).Transcription of rpoS in the putatively active arcB C180S and arcB CC mutant strains was significantly lower compared to that in WT, whereas no difference was found between the rpoS expression levels in the arcB C427S mutant and WT.Additionally, the stability of the RpoS protein in the WT and arcB CC strains was determined using Cm-dependent stability assays, followed by the detection of RpoS via immunoblot analysis (Figure S11a).Polyacrylamide gels stained with CBB (Kang et al., 2002) served as a loading control (Figure S11b).Under the conditions tested, only low levels of RpoS F I G U R E 6 RpoS proteolysis in the ΔarcA mutant.(a) RpoS protein stability was analysed in WT, ΔarcA and ΔarcAΔrssB strains, by Cmdependent stability assays.Respective strains were grown in LB medium to an OD 600 value of 1.5-2.Cm (100 μg ml −1 ) was added to the cultures, which were further incubated at 37°C and 180 rpm.At indicated time points (as stated in the figure), samples were harvested for immunoblot analysis.Cultures grown without Cm (-Cm) served as expression controls for the 60 min time point.A polyclonal RpoS antibody was used for protein detection.An RpoS-independent background band (•) is marked in the figure.(b) RpoS protein levels 60 min after the shift from LB to M9 minimal medium (+0.2% maltose) in WT, ΔarcA and ΔarcAΔrssB strains are shown.Cells were harvested for immunoblot analysis (using a polyclonal RpoS antibody, immunoblots are depicted in Figure S7).RpoS protein levels were normalised to an RpoSindependent background band and then the ratios compared to the mean value of the WT RpoS level were calculated.Shown are mean values ± SD of six independent samples.(c, d) M9 minimal medium-based swarming agar plates (+0.2% maltose) inoculated with the WT, ΔrssB, ΔarcA and ΔarcAΔrssB strains and incubated for 48 h at 37°C.(c) A representative picture of the swarming agar plates is depicted, and (d) mean values ± SD of eight independent measurements of the swarming diameter of the respective strains is shown.(b, d) Significant differences are marked by an asterisk (one-way anova followed by a Tukey's multiple-comparisons test, *p < 0.05).
protein were detectable in the WT and arcB CC mutant, indicating that maximum detectable RpoS proteolysis may have occurred.This causes difficulties in interpreting protein stability.However, it appeared that the amount of RpoS protein in the arcB CC mutant was actually lower than that in the WT, which may have been due to the higher repression of rpoS transcription.

| DISCUSS ION
Bacteria utilise different mechanisms to sense changing conditions as also to respond to them.These regulatory mechanisms are crucial for their survival since environmental changes may be of a drastic nature, a fine case in point being bacteria that survive the transition between environment and host.V. cholerae is affected by such transitions during its life cycle, as it persists in aquatic ecosystems before successfully infecting a human host.In this study, we focused on the interactions between the alternative sigma factor RpoS and the ArcAB two-component system, which enable V. cholerae to effectively adapt to such altering environments.
Regulation of the half-life of the RpoS protein by its anti-sigma factor RssB and the ClpXP protease has been studied (Wurm et al., 2017).Recently, we provided evidence indicating that RpoS is more stable in the rssB D75A mutant strain, which carries a point mutation in a putative phosphorylation site (Wölflingseder et al., 2022).
This result indicated possible kinase-dependent regulation of RpoS stability.To identify this kinase, we assumed that the RpoS protein would be more stable in a knockout mutant of the respective kinase and that this mutant would therefore show similar phenotypes as an rssB deletion strain.Despite earlier findings (Wurm et al., 2017), we, in this study, revisited knockout mutants of putative sensor histidine kinases and found that the RpoS protein was more stable in the ΔarcB strain compared to WT.Interestingly, the RpoS protein was not as stable as it was in the ΔrssB mutant.Remaining RpoS proteolysis observed in ΔarcB may have been either a result of RssB phosphorylation by other kinases or the residual activity of non-phosphorylated RssB proteins.In addition, another protease may recognise RpoS as a substrate, similar to what has been previously reported in E. coli (Mika & Hengge, 2005;Ranquet & Gottesman, 2007).Considering that the regulatory circuits of RpoS and RssB in V. cholerae and E. coli are closely related, it seems likely that RssB is phosphorylated by an ArcB kinase homologue in both species.Additionally, Sengupta et al. (2003) have previously shown that the ArcAB two-component system is active in V. cholerae, as it was found to play an important role in regulating the expression of the virulence transcription factor ToxT under anaerobic conditions.
An increase in the half-life of the RpoS protein in an arcB-knockout mutant cannot be considered direct proof of RssB phosphorylation by the ArcB kinase.It may also be an effect of the ArcA response pathway or other proteins phosphorylated by ArcB.Therefore, we conducted in vitro phosphorylation experiments with purified ArcB, RssB and RssB D75A proteins.A truncated version of ArcB lacking the membrane-spanning domain (amino acids 1-77), which was similar to truncated ArcB proteins used in studies performed with other bacteria, was used (Georgellis et al., 1997;Mika & Hengge, 2005).
It seems that in E. coli, the membrane-spanning domain may be acting as an anchor instead of being involved in sensing signals (Kwon et al., 2000).The ArcB-dependent appearance of a phosphorylated RssB protein band (RssB-P) in the presence of ATP demonstrated that ArcB is phosphorylating RssB.Performing the same experiment with the point-mutated protein RssB D75A did not result in an additional phosphorylated RssB band.Although the phosphorylation assays were performed under in vitro conditions, we consider this result, together with the altered RpoS protein stability, as proof of RssB phosphorylation by ArcB.Consequently, the aspartate residue at position 75 was identified as the phosphorylation site in RssB.
Considering that ArcB functions as a kinase of the RssB protein, we attempted to further characterise the effects exerted by ArcB on the regulatory RpoS network and dependent phenotypes in V. cholerae.First, motility behaviour, which is known to play an important role during infection, was investigated.Motility is not only essential for colonising the small intestine but also for the mucosal escape response at the end of infection (Echazarreta & Klose, 2019;Guentzel & Berry, 1975;Nielsen et al., 2006).In this study, we always determined motility phenotypes under nutrient-poor conditions, where RpoS is essential for inducing the transcription of motility-and chemotaxisrelated genes (Nielsen et al., 2006).Thus, changes in RpoS stability can be monitored via motility behaviour.The intermediate ΔarcB motility phenotype between WT and rssB D75A or ΔrssB strains was positively correlated with the RpoS protein stability phenotypes of these strains.As stated above, RpoS was more stable in the arcBknockout mutant compared to its stability in the WT strain, although residual RpoS degradation continued to be observed.The stability of RpoS in the rssB D75A mutant is elevated and RpoS proteolysis is most inefficient in the absence of RssB (ΔrssB) (Wölflingseder et al., 2022).
However, the possibility that ArcB may exert RpoS-independent effects on motility behaviour cannot be excluded.
As mentioned previously, the sensor histidine kinase ArcB in E. coli phosphorylates the response regulator ArcA (Bauer et al., 1999;Teran-Melo et al., 2018) and thereby exerts an immense impact on various gene expression patterns and other regulatory pathways.For example, over 1100 E. coli genes are regulated directly or indirectly by ArcA (Salmon et al., 2005).ArcA also acts as a global regulator in V. cholerae, affecting virulence, biofilm formation and motility (Li et al., 2022;Sengupta et al., 2003;Xi et al., 2020;Zhou et al., 2021).
It is known that in Vibrio fischeri, phosphorylated ArcA inhibits bioluminescence by binding in close proximity to the lux box (Septer & Stabb, 2012).Similar to that seen in E. coli, we found that ArcA was a repressor of rpoS transcription.Furthermore, ΔarcA, ΔarcB as well as ΔarcAB mutants showed higher rpoS transcription rates compared to that observed in WT.The elevated rpoS transcription rate in the ΔarcB strain indicated that ArcA is able to effectively repress rpoS transcription only in the presence of its kinase ArcB.Thus, ArcB seems to activate two proteins involved in the regulatory cascade of RpoS, thereby affecting its gene transcription and protein halflife.In that sense, ArcB must be active under conditions when rpoS transcription should be prevented or when the RpoS protein should be proteolytically removed.These two layers of regulation appear to be necessary to effectively inhibit the RpoS-dependent regulatory network.Similar to that seen in E. coli, signals that activate the Arc system in V. cholerae appear not only during the transition from aerobic to microaerophilic conditions but also during high nutrient availability (Mika & Hengge, 2005).During the infection process, V. cholerae faces changes in both oxygen availability and nutrient levels (Hsiao & Zhu, 2020;Nielsen et al., 2006).Furthermore, the motility behaviour of ΔarcA and ΔarcAB mutants was investigated.The arcA-knockout mutant exhibited a hypomotile phenotype under nutrient-poor conditions.Due to the derepression of rpoS transcription in the arcA-knockout mutant (see above) and previous experiments by Li et al. (2022), showing an increase in the motility phenotype of the ΔarcA mutant, caused by the repression of flrA transcription by ArcA, a hypermotile phenotype was expected.FlrA is the master regulator of flagella biosynthesis and part of the Class I flagella transcription hierarchy.In the stationary growth phase, RpoS reduces flrA promoter occupancy by H-NS, resulting in elevated FlrA levels, which are together with σ 54 responsible for transcription of the Class II genes (Klose & Mekalanos, 1998;Wang et al., 2012).The discrepancy to the study of Li and colleagues may be attributed to several factors as follows: firstly, derepressed rpoS transcription may not automatically lead to more RpoS protein in the cell, which subsequently induces the transcription of motilityand chemotaxis-related genes.Secondly, Li et al. (2022) used different conditions, namely LB medium-based swarming agar plates, from the conditions used by us to determine motility behaviour.
Under these nutrient-rich conditions, motility is reportedly independent of RpoS (Nielsen et al., 2006).Finally, different El Tor strains were used by the two studies and genetic variabilities affecting motility may exist.For these reasons, similar results compared to ours are not necessarily expected.To explore the regulatory mechanism in more detail, we further tested the motility of the double-knockout mutant ΔarcAB.Similar to the ΔarcB mutant, ΔarcAB was also hypermotile.Thus, generation of the hypermotile phenotype of ΔarcB was not dependent on the presence or absence of ArcA in the cell.
By contrast, for the hypomotile phenotype of ΔarcA, it is essential that ArcB is present in the cell.This suggested that the hypermotile phenotype may not be due to a regulatory effect resulting from the absence of the ArcA protein, but rather to one resulting from ArcB activity.Since all three deletion mutants, ΔarcA, ΔarcB and ΔarcAB, had a growth disadvantage compared to WT, we did not consider this as the sole reason for the hypomotile phenotype of ΔarcA.Interestingly, the arcB transcription in V. cholerae seems to be independent of ArcA, whereas in E. coli it appeared that ArcA may be involved in arcB transcription (Shalel-Levanon et al., 2005).
To further elucidate the reasons for the formation of the hypomotile phenotype of the ΔarcA mutant, the stability of the RpoS protein was examined.The RpoS protein level in the ΔarcA mutant appeared to be reduced compared to that in WT and its degradation appeared to be more rapid.The low RpoS level in the ΔarcA mutant was correlated with its hypomotile phenotype, and this may be attributed to RpoS being essential for inducing the motility of V. cholerae under nutrientpoor conditions (Nielsen et al., 2006;Wurm et al., 2017).However, further studies may be needed to explain the decreased RpoS protein levels in the ΔarcA mutant.It may be speculated that the absence of ArcA increases the phosphorylation of RssB by ArcB and that this increase in RssB phosphorylation accelerates RpoS degradation.This hypothesis is supported by two observations: first, the ΔarcAB deletion strain was found to be hypermotile, indicating that ArcB must be present in the cell for the ΔarcA mutant to express its hypomotility.
Second, the ΔarcAΔrssB-knockout mutant not only exhibits increased motility behaviour but also enhanced RpoS protein levels and activity, as indicated by the transcription rate of VC1248, which solely depends on RpoS (Nielsen et al., 2006).However, due to the lack of an appropriate method that could be used to directly analyse protein phosphorylation in vivo, the ArcB-induced increase in RssB phosphorylation in the arcA mutant remains unexplored.
Finally, we examined the regulation of ArcB activity.Two cysteine residues of ArcB are involved in regulating kinase activity in E. coli.Depending on the redox state of the cell, these cysteine residues form intermolecular disulphide bonds between two ArcB monomers, thereby decreasing kinase activity.The ArcB protein in E. coli becomes constitutively active when the cysteine residues are reduced (Malpica et al., 2004;Nuñez Oreza et al., 2012).To determine whether the regulation of ArcB activity in V. cholerae was similar, we constructed the point mutations arcB C180S , arcB C427S and arcB CC .Both arcB C180S and arcB CC mutants exhibited hypomotile phenotypes, whereas the motility of arcB C427S was similar to that of the WT.Consequently, the cysteine residue at position 180 (C180) appeared to be important for regulating ArcB activity under the conditions tested.Interestingly, in the ArcB homologue of E. coli, the role played by C180 in regulating ArcB activity in response to the redox state was also more substantial than that played by C241, as shown by the measurement of ArcB autophosphorylation and the repression of lldP transcription by phosphorylated ArcA (Malpica et al., 2004).Thus, the point-mutated ArcB C180S protein is held in a sort of active conformation, unable to form the respective intermolecular disulphide bond.Subsequently, ArcA and RssB may be phosphorylated during different growth phases, regardless of the redox or energy state of the cell.This appears to be true for both E. coli and V. cholerae.Phosphorylated ArcA is expected to repress the transcription of rpoS.We found that, under the conditions tested, rpoS transcription rates in the arcB C180S and arcB CC mutants were lower compared to those in the WT and arcB C427S strains.This finding suggested that ArcA in arcB C180S and arcB CC cells is phosphorylated, and phosphorylated ArcA subsequently represses rpoS expression.In addition, rpoS transcription in V. cholerae is regulated via other pathways, such as those involving the stringent response factors (p)ppGpp (Wurm et al., 2017), cAMP-CRP (Silva & Benitez, 2004) and quorum sensing via HapR (Joelsson et al., 2007).Additionally, we tested the impact exerted by the continuously active ArcB CC protein on RpoS stability.However, difficulties were encountered when comparing protein stability in the WT and arcB CC strains due to the low protein amounts found in both strains under the conditions tested.A comparison of the total protein amounts indicated that less RpoS protein detected in the arcB CC mutant was a consequence of reduced rpoS transcription, faster protein degradation or both.Overall, the hypomotile phenotype was positively correlated with the rate of repressed rpoS transcription as well as with the low RpoS protein amounts in the arcB CC mutant strain.
In summary, we identified ArcB as the kinase, which phosphorylates RssB, and thereby induces accelerated RpoS proteolysis in V. cholerae.Similar to that seen in E. coli, the response regulator ArcA, which is also a substrate of ArcB, represses rpoS transcription.The energy status of the cell is transmitted via the redox state of the C180 of ArcB via ArcA as well as RssB to RpoS and dependent genes.Although we found a significant overlap between ArcAB and RpoS-RssB input network regulation as seen in E. coli, the findings of this study provide deeper insights into the regulatory circuit of RpoS and the mechanisms underlying the motility behaviour of V.
cholerae.It appears that V. cholerae uses nutritional stress as an alarm signal to escape from unfavourable niches.This involves "RpoS activated motility-driven foraging" behaviour aimed at discovering and conquering new nutritional spots via the activation of chemotaxis and motility.This contrasts with the behaviour seen in E. coli, which follows the RpoS pathway centred on an adhesion-based non-motile lifestyle (Pesavento et al., 2008).Ultimately, such differences in the output scenario of RpoS regulation, among others, probably prevent V. cholerae from becoming a truly commensal bacterium in humans.

| Bacterial strains and growth conditions
All bacterial strains and plasmids used in this study are listed in Table 1.V. cholerae CO968 (O1 El Tor Ogawa) was used as the wild-type (WT) strain, and all mutant strains are derivates of strain CO968.The E. coli strains DH5αλpir and SM10λpir were used for genetic manipulations and the XL1-Blue strain for protein expression.Bacteria were cultivated in lysogeny broth (LB) or M9 minimal medium (+0.2% maltose), as stated in the text, under shaking conditions (180 rpm) at 37°C or in LB agar plates with aeration at 37°C.For nutrient-shift experiments, cells were cultivated in LB medium until an OD 600 of 1.5, then washed and inoculated in M9 minimal medium supplemented with 0.2% maltose with a starting OD 600 of 1 (Wurm et al., 2017).To determine the growth behaviour, cells grown overnight in LB medium were washed and diluted in M9 minimal medium (+0.2% maltose) to an OD 600 of ~0.01 and further cultivated at 37°C and 200 rpm for 16 h.OD 600 values were determined every 5 min using a plate reader.If required, antibiotics or other supplements were used in the following final concentrations: streptomycin (Sm, 100 μg ml −1 ), kanamycin (Km, 50 μg ml −1 ), ampicillin (Ap, 100 or 50 μg ml −1 in combination with other antibiotics), L-arabinose (0.001%), maltose (0.2%), sucrose (10%) and isopropyl ß-thiogalactopyranoside (IPTG, 1 mM).

| Construction of deletion mutants, arcB point mutants, expression plasmids and transcriptional fusions
Isolation of chromosomal DNA was carried out as previously described (Grimberg et al., 1989).The following kits were used for purification of PCR products, digested plasmids or plasmid isolation: QIAquick® PCR purification, QIAquick® gel extraction or the QIAprep Spin Mini Kit (Qiagen).Q5® High-Fidelity DNA Polymerase (New England Biolabs) was used for subcloning and sequencing, while Taq DNA polymerase (New England Biolabs) was used for all other PCRs.All oligonucleotides used in this study are listed in Table 2.

TA B L E 2
Oligonucleotides used in this study.
knockout mutants.After connecting and amplification of up-and downstream fragments via splicing by overlap extension PCR (SOE-PCR) (Horton et al., 1989), the PCR products were digested with the restriction enzymes SacI and XbaI (New England Biolabs) and ligated into the identically digested pCVD442 plasmid.Ligation products were transformed in electro-competent DH5αλpir cells and then checked for correct constructs by PCR.Subsequently, SM10λpir cells containing the gene-specific suicide plasmids were used to deliver these plasmids into V. cholerae via conjugation.After selection for Sm R and Ap R clones, implying that the V. cholerae chromosome had recombined the deletion construct, clones were grown on 10% sucrose LB agar plates to obtain Ap S clones (excision of the suicide plasmid from the chromosome) (Miller & Mekalanos, 1988).Chromosomal deletion mutants were verified by PCR and agarose (1%) gel electrophoresis.
The suicide plasmid pCVD442 was also used to generate arcB point mutations.The generation of these point mutants was performed similarly to knockout mutants.The codons for the amino acid exchange were encoded on the overlapping sequence of the oligonucleotides.
The oligonucleotide pairs arcB_C180S_F1_XbaI_fw/arcB_C180S_F1_ SOE_rv and arcB_C180S_F2_SOE_fw/arcB_C180S_F2_SacI_rv were used for the construction of arcB C180S and arcB_C427S_F1_XbaI_fw/ arcB_C427S_F1_SOE_rv and arcB_C427S_F2_SOE_fw/arcB_C427S_ F2_Xma_rv for arcB C427S .For the double cysteine mutation in arcB (arcB CC ), chromosomal DNA of the arcB C180S mutant was used to generate F1 and F2 of arcB C427S for the SOE-PCR.For conjugation into V.
cholerae, the arcB C180S strain was used as recipient strain.Sequencing was carried out to verify the point mutations.
To obtain plasmid-derived expression of arcA, arcB and arcAB, the arabinose-inducible pBAD18-Kan plasmid was used (Guzman et al., 1995).PCR fragments of the genes including the ribosome binding site were generated with the following oligonucleotides: Chromosomal transcriptional phoA fusions with rpoS, arcB or VC1248 were constructed as described recently (Seper et al., 2011).In short, gene fragments including the stop codons were amplified using respective oligonucleotide pairs.The oligonucleotides arcB_phoA_ Genes used in this work can be found with following N16961 accession numbers: rpoS/VC0534, rssB/VC1050, arcA/VC2368, arcB/ VC2369 and VC1248.

| Whole cell lysates, SDS-PAGE and immunoblot analysis
To obtain whole cell lysates, bacterial cells were cultivated under specific conditions (as indicated in the text) and cell equivalents of 2 OD 600 units were harvested (5 min, 5000 g).After resuspending the pellets in 80 μL Laemmli buffer (Laemmli, 1970), the samples were boiled for 30 min at 100°C and further used for sodium dodecyl sulphate-polyacrylamide gel electrophoresis (SDS-PAGE).The standard procedure for SDS-PAGE was performed using 12% polyacrylamide gels (Mini-PROTEAN Tetra cell system, BIO-RAD).The PageRuler™ Prestained Protein Ladder (10-180 kDa, Thermo Fisher Scientific) was used to determine the protein molecular weight.The gels were then either stained with Coomassie brilliant blue (CBB, Kang et al., 2002) or further used for immunoblot analysis as described recently (Wölflingseder et al., 2022).α-RpoS rabbit serum and horseradish peroxidase-conjugated goat α-rabbit IgG were used as primary and secondary antibodies respectively.The chemiluminescence signals were detected using the Molecular Imager ChemiDoc™ XRS system (BIO-RAD).If required, immunoblots were analysed with the Image Lab Software (BIO-RAD).

| Alkaline phosphatase assays (PhoA assays)
To determine the transcription levels of rpoS, arcB or VC1248, transcriptional fusions of the respective genes to phoA were constructed (Seper et al., 2011).The enzymatic activity of PhoA was measured in overnight cultures (M9 minimal medium +0.2% maltose).After cell lysis, the enzymatic reaction was started by the addition of the substrate p-nitrophenyl phosphate.PhoA activities were expressed in Miller units and calculated as follows: Miller units = (OD 405 × 1000)/ (OD 600 × t min × v ml × 0.486) (Miller, 1992).

| Motility assays
M9 minimal medium-based plates with 0.3% agar were used to determine the motility behaviour of V. cholerae mutants.Strains grown overnight (LB plates, 37°C) were inoculated into these plates and further incubated for 48 h at 37°C.Images were taken with the Molecular Imager ChemiDoc™ XRS system (BIO-RAD) at indicated time points.For statistical analysis, eight independent plates were inoculated and the diameters of the swarming extensions were measured.

F
Motility phenotype of the ΔarcB mutant.(a, b) M9 minimal medium-based swarming agar plates (+0.2% maltose) were inoculated with the WT, ΔarcB, rssB D75A and ΔrssB strains and incubated for 48 h at 37°C.(c, d) M9 minimal medium-based swarming agar plates (+0.2% maltose) supplemented with 0.001% arabinose for plasmid-derived arcB expression were inoculated with the WT pBAD, ΔarcB pBAD and ΔarcB pBADarcB strains and then incubated for 48 h at 37°C.Representative images are shown in (a, c).Mean values ± SD of eight independent measurements of the swarming diameter of the respective strains are shown in (b, d), and significant differences are marked by an asterisk (one-way anova followed by a Tukey's multiple comparisons test, *p < 0.05).

F
Influence of ArcA on rpoS transcription and motility behaviour.(a) The rpoS transcription rate was determined in the WT, ΔarcA, ΔarcB and ΔarcAB strains by alkaline phosphatase activities (in Miller units).The respective strains harbour an rpoS-phoA transcriptional fusion.Cells were grown in M9 minimal medium (+0.2% maltose) overnight.Shown are the median values ± IR of six independent biological replicates.Significant different values are marked by an asterisk (one-way anova followed by a Dunnett's multiple comparisons test, *p < 0.05).(b, c) M9 minimal medium-based swarming agar plates (+0.2% maltose) were inoculated with the WT, ΔarcA, ΔarcB and ΔarcAB strains and further incubated for 48 h at 37°C.(b) A representative picture of the swarming agar plates is depicted and (c) mean values ± SD of eight independent measurements of the swarming diameter of the respective strains are shown.Significant differences are marked by an asterisk (one-way anova followed by a Tukey's multiple comparisons test, *p < 0.05).F I G U R E 5 Complementation studies of the ΔarcA and ΔarcAB motility phenotypes.The motility phenotypes (a, b) of the ΔarcA and (c, d) ΔarcAB mutants were complemented by expression of arcA and arcAB, respectively, from the pBAD plasmid.For these complementation studies, M9 minimal medium-based swarming agar plates (+0.2% maltose) supplemented with 0.001% arabinose were used.The plates were inoculated with (a) WT pBAD, ΔarcA pBAD and ΔarcA pBADarcA or (c) WT pBAD, ΔarcAB pBAD and ΔarcAB pBADarcAB strains and cultivated for 48 h at 37°C.Representative images are depicted.(b, d) Shown are median values ± IR of eight independent measurements of the swarming diameter of the respective strains.Significant different values are marked by an asterisk (Kruskal-Wallis followed by Dunn's multiple-comparison tests, *p < 0.05).

F
ArcB regulation via its cysteine residues.(a, b) M9 minimal medium-based swarming agar plates (+0.2% maltose) were inoculated with the WT, arcB C180S , arcB C427S and arcB CC strains and incubated for 48 h at 37°C.(b) The median values ± IR of eight independent measurements are shown.(c) The rpoS transcription rate was measured in the WT, arcB C180S , arcB C427S and arcB CC strains by alkaline phosphatase activities (in Miller units).The respective strains harbour an rpoS-phoA transcriptional fusion.Cells were harvested from overnight cultures (M9 minimal medium +0.2% maltose).Median values ± IR of six independent biological replicates are shown.(b, c) Significant different values are marked by an asterisk (Kruskal-Wallis followed by Dunn's multiple-comparison tests, *p < 0.05).
arcA_KpnI_fw/arcA_XbaI_rv for arcA, arcB_KpnI_fw/arcB_XbaI_rv for arcB and arcAB_KpnI_fw/arcAB_XbaI_rv for arcAB.After digestion with the restriction enzymes KpnI and XbaI (New England Biolabs), the fragments were ligated in the identically digested pBAD18-Kan plasmid.The ligations were transformed into electrocompetent DH5αλpir cells and positive clones were identified by PCR and agarose (1%) gel electrophoresis.The respective isolated plasmids were then transformed into V. cholerae strains.
SacI_fw/arcB_phoA_KpnI_rv were used for the generation of arcB-phoA.The digested PCR fragments were ligated into the pGP704phoA plasmid and the ligation products were transformed into DH5αλpir cells.Positive clones were identified by PCR and agarose (1%) gel electrophoresis and then used for plasmid isolations.After plasmid transformation into the SM10λpir cells and transfer into V. cholerae mutants by conjugation, selection for Sm R and Ap R clones was carried out.The overexpression plasmid pQE-30 (Qiagen) was used for protein purification.To obtain pQE-30arcB78-785 , arcB (without the predicted membrane-spanning domain, amino acids 1-77) was amplified using the oligonucleotide pair arcB_pQE30_BamHI_fw/ arcB_pQE30_SalI_rv.The digested PCR fragment was ligated into the identically digested pQE-30 plasmid.After transformation into electro-competent XL-1 blue cells, positive clones were identified by PCR and agarose (1%) gel electrophoresis.