Defining the regulatory mechanisms of sigma factor RpoS degradation in Azotobacter vinelandii and Pseudomonas aeruginosa

In several Gram‐negative bacteria, the general stress response is mediated by the alternative sigma factor RpoS, a subunit of RNA polymerase that confers promoter specificity. In Escherichia coli, regulation of protein levels of RpoS involves the adaptor protein RssB, which binds RpoS for presenting it to the ClpXP protease for its degradation. However, in species from the Pseudomonadaceae family, RpoS is also degraded by ClpXP, but an adaptor has not been experimentally demonstrated. Here, we investigated the role of an E. coli RssB‐like protein in two representative Pseudomonadaceae species such as Azotobacter vinelandii and Pseudomonas aeruginosa. In these bacteria, inactivation of the rssB gene increased the levels and stability of RpoS during exponential growth. Downstream of rssB lies a gene that encodes a protein annotated as an anti‐sigma factor antagonist (rssC). However, inactivation of rssC in both A. vinelandii and P. aeruginosa also increased the RpoS protein levels, suggesting that RssB and RssC work together to control RpoS degradation. Furthermore, we identified an in vivo interaction between RssB and RpoS only in the presence of RssC using a bacterial three‐hybrid system. We propose that both RssB and RssC are necessary for the ClpXP‐dependent RpoS degradation during exponential growth in two species of the Pseudomonadaceae family.

In addition, the RpoS protein levels increase during the stationary phase of growth, which has been correlated with nutrient limitation during this stage (Lange & Hengge-Aronis, 1994). Owing to its relevance, the expression and activity of RpoS are highly regulated in E. coli; this occurs at the transcriptional level but mainly at translational and proteolysis levels. Regulation of translation is carried out by small regulatory RNAs in response to different signals (reviewed by Gottesman, 2019).
Proteolysis of RpoS in E. coli is carried out by ClpXP, an ATPdependent proteolytic complex, where ClpX has been reported as the specific chaperone that recognizes and delivers RpoS to ClpP protease for its degradation during the exponential phase of growth (Schweder et al., 1996). The RpoS recognition by ClpX is mediated by the RssB adaptor, a protein belonging to the response regulators (RR) family (Muffler et al., 1996). Usually, the C-terminus (C-TD) in proteins of this family holds a DNA-binding domain whose activity is modulated by phosphorylation of the N-terminal domain (N-TD). However, in E. coli RssB, the C-TD seems to be a degenerate member of the PP2C phosphatase family because it is shorter than the classic PP2C domain and lacks one of three conserved Mg 2+ -binding amino acids necessary for enzymatic activity. In contrast to most RRs that after phosphorylation bind DNA to activate transcription, RssB mediates protein-protein interaction (Galperin, 2006). Although no cognate histidine kinase has been identified for RssB, it can be phosphorylated by the ArcB sensor kinase (Mika & Hengge, 2005). Nevertheless, phosphorylation site of RssB is not essential for its activity (Peterson et al., 2004). The RssB-RpoS complex binds ClpXP, and RpoS is degraded while RssB is released from the complex and becomes available for a new cycle of RpoS degradation (Zhou et al., 2001).
The role of E. coli RssB homologs in RpoS proteolysis has been reported in other bacteria such as Salmonella typhymurium (Bearson et al., 1996), Erwinia carotovora (Andersson et al., 1999) and Vibrio cholerae (Wurm et al., 2017). In contrast, in species belonging to the Pseudomonadaceae family, such as Pseudomonas and Azotobacter, the recognition of RpoS by ClpXP has not been elucidated (Bertani et al., 2003;Muriel-Millán et al., 2017). In this work, we focus on the role of RssB regulation of RpoS protein levels in two representative species of these genera: Azotobacter vinelandii and Pseudomonas aeruginosa.
Azotobacter vinelandii is a nitrogen-fixing soil bacterium that produces polyβ-hydroxybutyrate (PHB), a biodegradable thermoplastic of industrial interest. In this bacterium, the synthesis of PHB is carried out during the stationary phase of growth from acetyl-CoA and NADPH, and the enzymes responsible for such biosynthesis are encoded in the phbBAC operon (Segura et al., 2003). The expression of the phbBAC biosynthetic operon is under the control of the transcriptional activator PhbR and RpoS; therefore, mutational inactivation of phbR or rpoS genes significantly reduces PHB synthesis (Hernandez-Eligio et al., 2011;Peralta-Gil et al., 2002). In addition, in A. vinelandii, RpoS is involved in the resistance to oxidative stress and nutrient starvation (Sandercock & Page, 2008) and it is necessary for alkylresorcinols synthesis and encystment (Cocotl-Yañez et al., 2011).
Pseudomonas aeruginosa is an environmental bacterium (Moradali et al., 2017) that has been considered to be ubiquitous (Crone et al., 2020), but that represents an important health hazard due to its production of virulence factors (Diggle & Whiteley, 2020;Gellatly & Hancock, 2013;Jurado-Martín et al., 2021) and high antibiotic resistance (Pang et al., 2019). P. aeruginosa resistant to carbapenem has been defined by the World Health Organization as a critical priority to search for alternative therapeutic treatment (Tacconelli et al., 2018). One of the most important P. aeruginosa virulence factor is the phenazine pyocyanin, which possesses redox-active properties and generates reactive oxygen species that damage host cells (Hall et al., 2016). The production of this virulence factor is negatively regulated by RpoS (Suh et al., 1999) by a yet undefined mechanism. Besides, RpoS positively affects the quorum sensing systems Las and Rhl (Schuster et al., 2004) and negatively the biofilm formation (Whiteley et al., 2001).
The three-dimensional structure of P. aeruginosa protein PA2798, that is, an RssB-like protein sharing 23% identity with E. coli RssB, has been reported (Protein Data Bank [PDB] ID 3EQ2). However, it has been proposed that P. aeruginosa RssB is not functionally equivalent to RssB of E. coli, because the former conserves the Mg 2+ -binding amino acids for phosphatase activity and exhibits a rigid linker between the C-TD and N-TD, which could interfere with the action mechanism described for E. coli RssB (Battesti et al., 2013;Dorich et al., 2019). In addition, no experimental evidence has been reported supporting a relation between RssB and RpoS degradation in P. aeruginosa.
The P. aeruginosa PA2798 (rssB) gene is found next to the PA2797 gene (here named rssC), which encodes a protein annotated as an antianti-sigma factor. These proteins seem to be translationally coupled, as the genes encoding them are likely organized in an operon (Winsor et al., 2015). In A. vinelandii, genes Avin32720 and Avin32710 encode homologs of P. aeruginosa PA2798 and PA2797, respectively (Setubal et al., 2009). In this study, we investigated the role of RssB and RssC in the regulation of RpoS protein levels in A. vinelandii and P. aeruginosa. We found that in these species, RssB and RssC are necessary for the degradation of RpoS during the exponential phase of growth.
The regulatory function of RssB/RssC affected RpoS-dependent phenotypes such as PHB and pyocyanin in A. vinelandii and P. aeruginosa, respectively. In addition, we identified an in vivo interaction between RssB and RpoS only in the presence of RssC. Our results reveal the mechanism by which the protein levels of the sigma factor RpoS are controlled during the exponential phase of growth in these two species belonging to the Pseudomonadaceae family.

| In silico analysis of RssB and RssC of A. vinelandii and P. aeruginosa
As mentioned above, rssB is immediately upstream of Avin43710 and PA2797 genes (rssC) in A. vinelandii and P. aeruginosa, respectively. In both species, the TGA stop codon of rssB overlaps the rssC ATG start codon, indicating that they are transcriptionally organized as an rssBC bicistronic operon (Figure 1a).
RssB proteins of A. vinelandii and P. aeruginosa are composed of 394 amino acid residues, share 79% identity and are 57 amino acids larger than E. coli RssB. The N-terminal region contains a response regulator receiver domain (REC) (Pfam: PF00072), whereas the Cterminal holds a domain present in proteins having phosphatase activity (PP2C, Pfam: PF00481), including the conserved Asp210, His288 and Asp336 residues (Galperin, 2006). These domains are connected by a coiled-coil linker of 83 residues (Figure 1b). In the case of RssC, this protein has 160 residues in both species and shares 89% identity. RssC proteins have a STAS (Sulfate Transport and Anti-Sigma factor antagonist) domain (Pfam: PF01740), commonly found in bacterial proteins that inhibit the activity of anti-sigma factors such as RsbV of Bacillus subtilis (Dufour & Haldenwang, 1994). Therefore, it seems that RssB and RssC are co-expressed and thus might be involved in the same function in A. vinelandii and P. aeruginosa.

| Inactivation of rssB and rssC genes affects RpoS-dependent phenotypes in A. vinelandii and P. aeruginosa
To investigate the role of RssB and RssC proteins in these two Pseudomonadaceae family species, we constructed mutant derivatives of A. vinelandii UW136 and P. aeruginosa PAO1 strains carrying mutations in rssB, rssC and rssBC genes (see Materials and methods).
First, we determined whether these mutations affect the growth in both species. The A. vinelandii UWrssB, UWrssC and UWrssBC strains showed a longer lag phase than the wild-type strain, although the protein concentration was similar among all strains after 48 h of incubation ( Figure S1a). The P. aeruginosa mutants PArssB, PArssC and PArssB* (an rssB mutant with a polar effect on rssC expression) also exhibited a delay at the start of the exponential growth phase. However, after 6 h of incubation, the three mutants showed a similar optical density (OD 600nm ) to that of the PAO1 strain ( Figure S1b). These data suggest that the absence of proteins RssB or RssC is detrimental during the exponential phase of growth in A. vinelandii and P. aeruginosa.
In A. vinelandii, the PHB synthesis depends on RpoS (Hernandez-Eligio et al., 2011;Peralta-Gil et al., 2002); therefore, we tested the effect of the rssB, rssC and rssBC mutations on the accumulation of this polymer. As shown in Figure 2a, the inactivation of rssB and rssC caused a white phenotype in A. vinelandii, compatible with the increase in PHB accumulation, in contrast to that observed in the UWrpoS mutant. Quantification of the PHB content revealed an increase of 25%, 66% and 59% in the three strains carrying the rssB, the rssC and rssBC mutations, respectively, compared with the wild-type UW136 ( Figure 2a). To confirm the negative role of RssB and RssC proteins on polymer synthesis, we complemented the rssB and rssC mutants by integrating a wild-type copy of the respective genes into the chromosome. As expected, the levels of PHB in the complemented strains were reduced to those observed in the wild-type strain ( Figure 2a).
In P. aeruginosa, pyocyanin synthesis is under RpoS negative regulation. Figure 2b shows that the strains carrying rssB, rssC or rssB* mutations showed a less blue-green pigmentation phenotype relative to the PAO1 strain, compatible with a reduction in pyocyanin synthesis.

P. aeruginosa.
Previous studies reported that in these species, the proteolytic complex ClpXP degrades RpoS during the exponential phase (Bertani et al., 2003;Muriel-Millán et al., 2017). Because in E. coli, the adaptor protein RssB is necessary for the recognition and degradation of RpoS by ClpXP, we hypothesized that in A. vinelandii and P. aeruginosa, RssB and RssC could affect the protein levels of RpoS.
Indeed, Western blot assays revealed that the rssB, rssC and rssBC mutations significantly increased the sigma factor levels during the exponential phase of growth compared with the wild-type A. vinelandii UW136 and P. aeruginosa PAO1 strains (Figure 3a Phenotype and levels of pyocyanin in P. aeruginosa PAO1, mutant derivatives PArssB, PArssC, PArssB* (an rssB strain with a polar effect on rssC), PAOrpoS and PArssB complemented strain. The strains were grown in LB liquid medium for 24 h at 37°C. For (a, b), the data are the mean of three independent experiments. Error bars: standard deviation (SD).

F I G U R E 3 RssB and RssC affect the protein levels of RpoS during the exponential phase. (a) Detection of RpoS by Western blot in
Azotobacter vinelandii strains. The cultures were grown in a PY liquid medium for 8 h at 30°C. About 20 μg of total protein was used for each sample. An unspecific interaction of antiserum is shown as the loading control (*). (b) Detection of RpoS in Pseudomonas aeruginosa strains grown in LB liquid medium at 37°C at OD 600nm 0.2-0.3. For each sample, about 40 μg of total protein was used. For (a) and (b), an antiserum anti-RpoS of A. vinelandii was used. RpoS levels in wild-type strains (UW136 and PAO1) were assumed to be 100%. Data are the mean of three independent experiments, and SD is shown in parenthesis (±SD).
supporting the hypothesis that RssB and RssC are involved in RpoS recognition by ClpXP.
We also determined the effect of RssB and RssC on the in vivo stability of RpoS during the exponential phase of growth after blocking protein synthesis. In the A. vinelandii UW136 strain, RpoS showed a half-life of <30 min, while in UWrssB, UWrssC and UWrssBC mutants, this was >120 min, which is similar to the clpP strain ( Figure 4a,b). In the P. aeruginosa wild-type strain, the half-life of RpoS also was <30 min, and it was increased by mutations in rssB or rssC or both to >45 min (Figure 4c,d). These results demonstrate that RssB and RssC are necessary for the degradation of RpoS by the proteolytic complex ClpXP in A. vinelandii and P. aeruginosa during the exponential growth phase.

| P. aeruginosa PAO1 RssB restores the RpoS levels in the A. vinelandii rssB strain
To further demonstrate the role of RssB in the degradation of RpoS by ClpXP, we carried out heterologous complementation of the A.
vinelandii UWrssB mutant with the rssB gene of P. aeruginosa. For this, we used plasmid pSEVA:ParssB, which encodes a P. aeruginosa rssB copy under the control of an IPTG-inducible promoter. This plasmid and the empty vector (used as a negative control) were used to transform the wild-type UW136 and rssB strains. As shown in Figure 5, the protein levels of RpoS in the A. vinelandii UWrssB mutant were reduced in the presence of P. aeruginosa rssB, but not when this strain was transformed with the empty pSEVA vector. In the case of the UW136 strain, the expression of P. aeruginosa rssB gene did not affect the RpoS protein levels ( Figure 5), suggesting that increased levels of RssB in wild-type conditions do not promote the RpoS degradation. These results confirm the role of RssB in the proteolysis of RpoS in these species of the Pseudomonaceae family.

| RssB interacts with RpoS only in the presence of RssC
The above results demonstrate that both RssB and RssC proteins are necessary for the degradation of RpoS by ClpXP. A possible mechanism that explains how the sigma factor is presented as a substrate of the proteolytic complex is a direct protein-protein  For (a, b), an antiserum anti-RpoS of A. vinelandii was used. The density at time zero was assumed to be 100%. Data are the mean of three independent experiments, and SD is shown in parenthesis (±SD).
interaction, as occurs between RssB and RpoS in E. coli. Therefore, we used the two-hybrid system based on the functional complementation of the adenylate cyclase activity in E. coli (BACTH) (Karimova et al., 1998) (Figures 3 and 4), we hypothesized that the interaction between RssB and RpoS may require the presence of RssC. Therefore, we adapted a bacterial three-hybrid system by cloning the A. vinelandii rssBC operon into the pUT18C and pKT25 vectors, which allows the expression of a hybrid RssB (fused to T18 or T25 domains) and a non-hybrid RssC. The vectors were co-transformed with the respective plasmid harbouring the A. vinelandii rpoS gene. In this case, we detected interaction between RpoS and RssB in the co-transformant T25-AvRssBC/T18-AvRpoS but not in the other combination (T25-AvRpoS/T18-AvrssBC). Detection of protein interactions only in one of the two possible combinations using BACTH is not rare and may be related to the different copy numbers of T25 and T18 plasmids (Battesti & Bouveret, 2012;Koch et al., 2020). For P. aeruginosa proteins, interactions between RpoS and RssB in the presence of RssC were detected using the two possible combinations (T25-PaRpoS/ T18-ParssBC and T25-ParrsBC/T18-PaRpoS) (Figure 7). These results suggest that RssB may interact with RpoS only when RssC is present, which could explain the role of RssB and RssC in the degradation of RpoS by ClpXP.

| RssB but not RssC interacts with ClpX
Low-affinity interactions between RssB and ClpX in the absence of RpoS have been documented in past studies by in vitro F I G U R E 5 Pseudomonas aeruginosa RssB restores the wild-type RpoS protein levels to the Azotobacter vinelandii UWrssB mutant. The strains were grown in PY medium for 8 h at 30°C. For UW136 and UWrssB strains transformed with plasmids pSEVAParssB and pSEVA424 (negative control), 0.5 mM IPTG (final concentration) was added to cultures. Experimental procedures for detecting RpoS were the same as those for Figure 3.

F I G U R E 6
In vivo interaction by the bacterial two-hybrid system between RssB and RssC proteins of (a) Azotobacter vinelandii and (b) Pseudomonas aeruginosa . For (a,b), the co-transformed Escherichia coli BTH101 strains were spotted onto LB and M63-maltose agar plates supplemented with ampicillin (100 μg/mL), kanamycin (50 μg/mL), X-Gal (40 μg/mL) and IPTG (0.5 mM). The plates were incubated for 24-48 h at 30°C. Development of blue colour and growth capacity (only for M63 plates) indicate the interaction between tested proteins. β-galactosidase activity was measured from LB liquid cultures incubated for 24 h at 30°C. Values are relative to the enzymatic activity detected for negative control (empty pKT25 and pUT18C plasmids) and represent the mean of three independent determinations. Error bars; SD. experiments with purified E. coli proteins (Zhou et al., 2001). In addition, two-hybrid approaches revealed that RssB (MviA) of Salmonella Typhimurium interacts with ClpX (Moreno et al., 2000). Therefore, we wanted to determine whether A. vinelandii RssB and RssC can interact with the ClpX chaperone. Two-hybrid assays revealed that ClpX interacted with RssB, whereas no interaction between ClpX and RssC was detected in any of the two combinations ( Figure 8).
These results indicate that, opposite to the interaction between RpoS and RssB, RssC is not required for the RssB-ClpX interaction.
Furthermore, these data also support that ClpX of the E. coli reporter strain is not acting as a 'bridge' in the interaction detected between RssB and RssC ( Figure 6) as has been reported for Caulobacter crescentus CpdR, an adaptor that interacts in a natural manner with ClpX rather than with some of its substrates (Lau et al., 2015). In P. aeruginosa, rssBC is flanked by tal and a gene coding a conserved hypothetical protein, followed by vacJ (Figure 9). This genomic organization is also present in most Pseudomonas species (P. stutzeri, P. protegens, P. chlororaphis, P. syringae and P. aestusnigri) except for P. putida, in which downstream of rssBC, lies the PP_2167 gene that encodes a PhoD-like phosphatase. In Azomonas and Entomomonas genera, rssBC is also conserved but flanked by other genes different from those in P. aeruginosa and A. vinelandii ( Figure 9). We also determined the amino acid sequence identity of RssB and RssC of P. aeruginosa and A. vinelandii with those from the other species. Table S3 shows that all the proteins share >70% sequence identity, except for RssB of P. aestusnigri and E. moraniae, whose identity with P.aeruginosa/A. vinelandii RssB was 60/61 and 63/62%, respectively, and RssC of E. moraniae, which showed 55%/53% identity. These results show that RssB and RssC are conserved throughout Pseudomonadaceae species, suggesting that the regulatory mechanism of RpoS stability by these proteins is common in this family.

| DISCUSS ION
RpoS is one of the most important alternative sigma factors in Gramnegative bacteria. Besides its protecting role under stress conditions, RpoS plays pivotal functions in other cellular processes such as virulence factor synthesis, biopolymer production and biofilm formation (Duan et al., 2021;Hernandez-Eligio et al., 2011;Suh et al., 1999).
Therefore, characterization of the regulatory mechanisms of RpoS expression and activity is of main interest in understanding bacterial physiology.
In E. coli, the regulation of RpoS protein stability is governed by the RssB adaptor, an atypical two-domain member of the response regulator (RR) family, which binds RpoS and delivers it to the chaperone-protease complex ClpXP (Muffler et al., 1996). As in E. coli, in A. vinelandii and P. aeruginosa the RpoS sigma factor is also degraded by ClpXP in exponentially growing cells. However, how F I G U R E 7 In vivo interaction by the bacterial three-hybrid system between RpoS and RssB in the presence of RssC. Plasmids that express hybrid RssB and non-hybrid RssC of both Azotobacter vinelandii (pKT25AvrssBC and pUT18CAvrssBC) and Pseudomonas aeruginosa (pKT25ParssBC and pUT18CParssBC) were paired with the respective vectors and co-transformed into the Escherichia coli BTH101 strain. The growth conditions and experimental procedures were the same as those for two-hybrid assays in Figure 6.

F I G U R E 8
In vivo interaction by the bacterial two-hybrid system between RssB and ClpX from Azotobacter vinelandii. Growth conditions of co-transformed Escherichia coli BTH101 strains and experimental procedures were the same as those for two-hybrid assays in Figure 6. the protease recognizes RpoS was not understood in these species (Bertani et al., 2003;Muriel-Millán et al., 2017). In this work, we describe the mechanism by which RpoS is presented as a proteolytic substrate in A. vinelandii and P. aeruginosa.
Pseudomonas aeruginosa RssB is structurally similar to RssB of E. coli (Dorich et al., 2019). However, relevant differences between these proteins (e.g. at the interdomain linker and active site of the PP2C domain) and that in P. aeruginosa, RssB is translationally coupled with a protein annotated as an anti-sigma antagonist (RssC), led to propose that RssB-dependent regulation of RpoS in P. aeruginosa differs from that in E. coli and possibly resembles a classical partner switching system instead of proteolytic inactivation (Battesti et al., 2013;Dorich et al., 2019;Gottesman, 2019). However, our results demonstrate that in both A. vinelandii and P. aeruginosa, the proteolysis of RpoS involves both RssB and RssC, as mutations in either rssB or rssC genes increased the stability of RpoS protein (Figure 4).
Indeed, the RssB-like effect of RssC on RpoS levels and stability supports that RssC does not act as an anti-anti-sigma protein; instead, it is functionally associated with RssB for RpoS degradation.
In Acinetobacter baumannii AB5075, the RssB and RssC homologs, GigA and GigB, respectively, constitute a partner switching system, where GigA dephosphorylates GigB, and the unphosphorylated GigB acts as an anti-anti-sigma factor sequestering NPr protein from the sigma factor RpoE, which is necessary for the transcription of stress response genes since A. baumannii does not present an RpoS homologue (Gebhardt & Shuman, 2017;Robinson et al., 2010).
Several experimental evidence supports that the sigma factor regulation mechanism of GigA/GigB differs from this of RssB/RssC. For instance, mutations in either gigA or gigB result in a sensitive phenotype to antibiotic stress in A. baumannii because RpoE activity is reduced due to its sequestration by NPr (Gebhardt & Shuman, 2017).
In contrast, in A. vinelandii and P. aeruginosa, the inactivation of rssB or rssC resulted in phenotypes compatible with increased activity of RpoS (Figure 2). In addition, in A. vinelandii, a mutation in ptsO, the gene encoding NPr, promotes degradation of RpoS by ClpAP but not ClpXP, producing a negative PHB phenotype (Muriel-Millán et al., 2017), whereas in A. baumannii, inactivation of ptsO increases the resistance to Km as a consequence of an increment of RpoE function (Gebhardt & Shuman, 2017). Therefore, it seems that the A. baumannii GigA/GigB system is a canonical partner-switching system and does not involve sigma factor proteolysis as RssB/RssC. Based on the action mechanism of E. coli RssB, we hypothesized that RssB could interact with RpoS in both A. vinelandii and P. aeruginosa. However, bacterial two-hybrid assays detected interaction between RssB and RssC but not of RpoS with any of these proteins ( Figure 6). In contrast, using a bacterial three-hybrid system, we detected interaction of RssB with RpoS only in the presence of RssC (Figure 7), suggesting that interaction between RssB and RssC enables RssB to recognize RpoS ( Figure 10). Structural studies revealed that the interdomain linker of RssB of P. aeruginosa is devoid of a glutamate cluster and is more rigid than that of E. coli RssB, which would impede the RpoS binding and its subsequent degradation as is carried out in E. coli (Dorich et al., 2019). Therefore, it is tempting to speculate that in A. vinelandii and P. aeruginosa, the RssC-RssB interaction could structurally modify RssB, allowing its interaction with RpoS for its degradation by ClpXP protease. On the other hand, homooligomerization of RssB and RssC detected here ( Figure 6) could play key roles in the induction of RpoS degradation, as has been proposed for E. coli RssB (Battesti et al., 2013). However, because additional work is needed to understand the biological function of such oligomers, we do not incorporate them in our proposed model ( Figure 10).
Regulation of RssB activity in E. coli is carried out by IraP, IraM and IraD; three small proteins called anti-adaptors, induced by phosphate and magnesium starvation and DNA damage, respectively (Bougdour et al., 2006;Bougdour et al., 2008). These proteins use disting binding modes to interact with RssB, resulting in an impediment of the adaptor to deliver RpoS to ClpXP protease (Battesti et al., 2013;Micevski et al., 2015). The genomes of both A. vinelandii F I G U R E 9 Genomic context of rssBC in several species belonging to the Pseudomonadaceae family. Grey arrows represent genes that encode conserved hypothetical proteins.

F I G U R E 1 0 Proposed model for the RpoS degradation by ClpXP
in Azotobacter vinelandii and Pseudomonas aeruginosa. During the exponential phase of growth, the interaction between RssB and RssC allows RssB interacts with RpoS and presents it to ClpXP for its degradation, maintaining low levels of the sigma factor. In addition, RssB may interact with ClpX during the recognition of RpoS for proteolysis. RssB and RssC are shown as monomers, but they could be as homodimers as well. and P. aeruginosa do not have homologs of E. coli RssB anti-adaptors, suggesting that other regulatory mechanisms of RssB/RssC function are present in these species. The A. baumannii GigB protein is phosphorylated at serine-59 residue (Gebhardt & Shuman, 2017), which is conserved in its RssC homolog. Therefore, we hypothesize that both RssB and RssC can be phosphorylated, and that the phosphorylation state of these proteins could be related to their function in RpoS degradation. This hypothesis will be explored in the future.
In summary, we propose a new model for the regulation of RpoS degradation in A. vinelandii and P. aeruginosa during the exponential growth phase, where RssC and RssB form a complex that allows the RssB adaptor to interact with RpoS. In turn, RpoS within this complex is recognized and degraded by ClpXP (Figure 10). The presence of the rssBC operon in other Pseudomonadaceae species (Figure 9) strongly suggests that the new regulatory pathway proposed here is conserved throughout the Pseudomonadaceae family.

| Strains, media and culture conditions
Strains used in this study are listed in Table S1

| Construction of plasmids and mutant strains
Oligonucleotides and plasmids used in this work are listed in Tables S1 and S2, respectively. For the construction of A. vinelandii strains, total DNA from UW136 strain and oligonucleotides FragFwRssB and FragRv-RssB were used to amplify a 2315 bp fragment containing the rssB and rssC genes. This fragment was cloned into plasmid pJET1.2 (Thermo Scientific). E. coli DH5a cells were transformed with the resultant plasmid pJETFragRssB-RssC.
A gentamicin resistance cassette from plasmids pBSL141 (Alexeyev et al., 1995) was inserted in the same direction of rssB transcription (non-polar orientation) into the unique SalI site located within rssB to produce plasmid pJETrssB::Gm (pJETrssB). Plasmid linearization was carried out to avoid the co-integration of the whole vector into the bacterial chromosome by single recombination events (Moreno et al., 2019).
For the construction of P. aeruginosa PArssB* (an rssB strain with a polar effect on rssC expression) and PArssC mutants, disrupting plasmids pEXrssB and pEXrssC were constructed as follows: a 545 bp DNA fragment upstream of rssB was amplified using the oligonucleotides 1PA2798Fw and 2PA2798Rv. A second DNA fragment of 537 pb downstream of rssB was amplified using 5PA2798Fw and 6PA2798Rv oligonucleotides. The Apr resistance gene was amplified from pIJ773 plasmid using the oligonucleotides ApraFw and ApraRv. The three PCR products were purified from gel and used as templates in a nested PCR using the oligonucleotides 1PA2798Fw and 6PA2798Rv. The PCR product was gel purified, digested with HindIII and cloned into the HindIIIdigested plasmid pEX18-Strep (García-Reyes et al., 2021), resulting in the pEXrssB plasmid. The same strategy was used to construct the pEXrssC vector, using oligonucleotide pairs 1PA2797 and 2PA2797Rv and 5PA2797Fw and 6PA2797Rv to amplify DNA fragments upstream (514pb) and downstream (510 pb) of rssC, respectively. The pEXrssB and pEXrssC plasmids were used to transform electrocompetent PAO1 cells, and Apr-resistant, Smsensitive transformants, generated by double recombination events, were isolated.
For the non-polar PArssB mutant construction, the rssB* strain was transformed with plasmid pFLP2 to excise the Apr marker by the Flp recombinase (Hoang et al., 1998). This excision leaves an Flp recombinase target (FRT) scar of 85 bp, which does not have a polar effect on the downstream rssC gene (Choi & Schweizer, 2005). All mutant strains constructed in this study were confirmed by PCR analysis to carry the respective mutations.
Complementation of the PArssB strain was carried out by amplifying the rssB gene with oligonucleotides PA2798BamHFw and PA2798XbaRv and cloning it into plasmid pSEVA424, under the control of an IPTG-inducible promoter (pSEVA424PArssB).

| Western-blot assays
Detection of RpoS levels in A. vinelandii was carried out as pre-

| Detection of protein-protein interactions
Bacterial two-hybrid assays were performed using the system based on the functional complementation of the adenylate cyclase activity in E. coli (BACTH) (Karimova et al., 1998) (Euromedex Cat#   EUK001). The rssB, rssC, rssBC and rpoS genes from A. vinelandii and P. aeruginosa and clpX from A. vinelandii were cloned into the pKT25 and pUT18C vectors to construct in-frame fusions to T25 and T18 domains of B. pertussis adenylate cyclase, respectively (Table S1).
All the constructed plasmids were verified by DNA sequencing.
The respective plasmid pairs (including negative and positive controls) were used to co-transform competent E. coli BTH101 cells, and co-transformants were isolated in LB plates added with ampicillin (Amp) (100 μg/mL) and km (50 μg/mL) incubated at 30°C for 24 h. Then, colonies were picked and grown in Amp-Km LB plates at 30°C for 24 h. For spot assays and β-galactosidase activity quantification, co-transformants were grown overnight at 30°C in LB broth added with 0.5 mM IPTG, Km and Amp. Then, a 10 μL aliquot of each inoculum was spotted on the surface of LB and M63-maltose medium plates supplemented with 40 μg/mL 5-bromo-4-chloro-3indolylβ-d-galactoside (X-Gal), 0.5 mM IPTG, Km, Amp. The plates were incubated at 30°C for 24-48 h. β-galactosidase activity was measured from cultures in LB incubated for 24 h at 30°C using standard protocols (Miller, 1972).

| Analytical methods
PHB content was determined using the protocol described by Law and Slepecky (Law & Slepecky, 1961) from A. vinelandii strains grown in PY medium at 30°C for 48 h. Pyocyanin production in P. aeruginosa was determined as previously reported (Essar et al., 1990)   COV is a doctoral student of MDCBQ-UNAM and receives a scholarship from CONACyT (CVU 823305).

CO N FLI C T O F I NTE R E S T S TATE M E NT
The authors declare that they have no conflict of interest.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are available from the corresponding author upon reasonable request.