σE controlled regulation of porin OmpU in Vibrio cholerae.

Bile resistance is essential for enteric pathogens, as exemplified by Vibrio cholerae, the causative agent of cholera. The outer membrane porin OmpU confers bacterial survival and colonization advantages in the presence of host-derived antimicrobial peptides as well as bile. Expression of ompU is controlled by the virulence regulator ToxR. rpoE knockouts are accompanied by suppressor mutations causing ompU downregulation. Therefore, OmpU constitutes an intersection of the ToxR regulon and the σE -pathway in V. cholerae. To understand the mechanism by which the sigma factor σE regulates OmpU synthesis, we performed transcription studies using ompU reporter fusions and immunoblot analysis. Our data revealed an increase in ompU promoter activity in ΔrpoE strains, as well as in a ΔompU background, indicating a negative feedback regulation circuit of ompU expression. This regulation seems necessary, since elevated lethality rates of ΔrpoE strains occur upon ompU overexpression. Manipulation of OmpU's C-terminal portion revealed its relevance for protein stability and potency of σE release. Furthermore, ΔrpoE strains are still capable of elevating OmpU levels under membrane stress conditions triggered by the bile salt sodium deoxycholate. This study provides new details about the impact of σE on ompU regulation, which is critical to the pathogen's intestinal survival.

INTRODUCTION

Vibrio cholerae, a Gram‐negative, facultative anaerobic bacterium, is an infectious agent that causes the human disease cholera (Cava, 2017). This pathogen is endemic in India, Bangladesh, Southeast Asia, and South America (Faruque et al., 1998). Recent outbreaks in Haiti and Yemen have shown its high spreading potential (Ramamurthy et al., 2019). The estimated number of cholera cases worldwide reaches 2.9 million, with 95.000 deaths every year (Legros, 2018). The bacterium is able to transit between two environmental habitats, an aquatic reservoir, where it persists by forming biofilms on chitinous surfaces of crustaceans, zooplankton, and phytoplankton, and the human gastrointestinal tract (Colwell, 2004). Upon oral ingestion of contaminated water or food, V. cholerae colonizes the small intestine after passage of the gastric acid barrier and the penetration of the intestinal epithelial mucus lining. During the passage through the stomach to its primary site of infection, the pathogen rapidly adapts its physiology by altering gene expression, protein biosynthesis and stability, posttranscriptional control, and surface exchange in order to counteract host‐specific stress responses. Eventually, virulence gene expression is induced, enforcing full colonization fitness (Cakar et al., 2018; DiRita et al., 1991; Herrington et al., 1988; Pennetzdorfer et al., 2019; Zingl et al., 2020). Stressors attacking the bacterial outer membrane (OM) interfere with V. cholerae survival in the intestine. Bile salts, organic acids, factors of the innate immune system, for example complement factors secreted by intestinal epithelial cells (Andoh et al., 1998) and defensins produced by Paneth cells (Mallow et al., 1996), are ranked among such membrane‐disruptive agents.

The pathogenesis of cholera is primarily based on the activity of secreted cholera toxin (CT) and the production of toxin‐coregulated pili (TCP), eventually resulting in a highly severe form of secretory diarrhea (Childers and Klose, 2007). This virulence gene cascade termed the ToxR regulon, consists of several transcription factors, for example the transcriptional complex ToxRS, which is integrated into the inner membrane (Peterson and Gellings, 2018). In addition to the initiation of the ToxR regulon, ToxRS inversely regulate the expression of virulence‐associated outer membrane porins (OMPs) encoded by ompU and ompT. Upon host entry, the activation of ompU transcription and ompT repression by ToxR, as well as the exchange of OmpT for OmpU, are crucial for the rapid establishment of bile salt resistance, facilitating V. cholerae to colonize and survive within the human intestine (Crawford et al., 1998; Li et al., 2000; Provenzano and Klose, 2000; Zingl et al., 2020). Promoter regions of both genes harbor AT‐rich ToxR binding elements. This 5′‐TNAAA‐N5‐TNAAA‐3′ element (or its reverse complement) occurs once upstream of the ompT coding sequence (CDS) and three times at the ompU operator site (Goss et al., 2013). Concordantly, the bile salt sodium deoxycholate (Na‐DC) modulates OMP regulation by activating the ToxRS complex in both O1 biotypes El Tor and classical strains (Lembke et al., 2018; Provenzano and Klose, 2000; Provenzano et al., 2000). Additionally, increased ToxR levels achieved in response to supplementation with the amino acids asparagine, arginine, glutamate, and serine (NRES) were shown to positively affect ompU activation and ompT repression (Mey et al., 2012), whereas higher temperatures cause an opposite regulatory effect (Parsot and Mekalanos, 1990).

Electrophysical channel characterization revealed various differences between OmpT and OmpU. In contrast to OmpT, OmpU possesses higher cation selectivity, leading to an influx restriction of negatively charged components across the OM, and consequently preventing cytoplasmic membrane destruction (Pathania et al., 2018; Simonet et al., 2003). Although bile resistance acquired by OmpU might be indirect, ompU or toxR deletion strains are more susceptible to bile salt damage than strains lacking OmpT in their OMs. Since bile salt uptake across the OM by OmpU is lower than that by OmpT, ompT repression is more important than ompU expression (Wibbenmeyer et al. 2001).

Moreover, initiation of regulated intramembrane proteolysis (RIP) of ToxR under stress conditions is inhibited in the presence of Na‐DC, ensuring ToxR‐operator binding by enhanced homo‐ and heterodimer formation between ToxR and ToxS (Lembke et al., 2018; Lembke et al., 2020; Midgett et al., 2017).

Additionally, OmpU was identified to act as a signal transducer in the σE pathway (Mathur et al., 2007). In general, this system facilitates the response of Gram‐negative bacteria to misfolded periplasmic proteins and perturbation of the OM (Mescas et al., 1993; Ruiz and Silhavy, 2005). The alternative sigma factor σE (encoded by rpoE) is sequestered to the inner side of the cytoplasmic membrane by the membrane‐embedded anti‐sigma factor RseA to keep it inactive in unstressed cells (De Las Penas et al., 1997; Missiakas et al., 1997). Exposure of C‐terminal signal peptides of misfolded OMP to the periplasm results in PDZ domain‐dependent activation of the site‐1 protease DegS and, in turn, site‐2 protease RseP/YaeL, initiating RIP of RseA (Alba et al., 2002; Griorova et al., 2004; Kanehara et al., 2002; Walsh et al., 2003; Wilken et al., 2004). Thus, σE associated with residual RseA protein is released into the cytoplasm, where RseA remnants are degraded by ClpXP (Flynn et al., 2004). Eventually, σE can interact with RNA polymerase core enzyme to initiate transcription from σE‐dependent promoters (De Las Penas et al., 1997; Missiakas et al., 1997). Genes belonging to the σE regulon encode for rpoE itself and for periplasmic chaperones and proteases, for example degP, which supports refolding or acts as a protease (Dartigalongue et al., 2001; Kabir et al., 2005; Kovacikova and Skorupski, 2002; Krojer et al., 2002; Rhodius et al., 2006; Spiess et al., 1999).

In V. cholerae, the σE RIP system is linked to the ToxR regulon, as exemplified by proteolysis of the membrane‐bound transcription factors TcpP and ToxR affecting virulence gene expression (Pennetzdorfer et al., 2019). Based on current models, OmpU activated by ToxR acts as a sensor component, and comprises a C‐terminal signal motif that can be recognized by the DegS PDZ domain upon exposure to human defensin P2 or polymyxin B (Mathur et al., 2007). High‐throughput sequence analysis of rpoE deletion strains revealed suppressor mutations within the ompU transcriptional control region, resulting in reduced ompU expression (Davis and Waldor, 2009).

Intriguingly, OmpU seems to constitute an intersection point of the ToxR regulon and σE pathway, as evidenced by suppressor mutations originated within the ompU promoter upstream region. By using transcriptional reporter fusions, we demonstrate that σE release depends on OmpU, but proper ompU expression also requires σE and OmpU itself. Remarkably, overexpression of ompU is detrimental to ΔrpoE strains, resulting in significant survival deficiency. Additionally, our data highlight the importance of the OmpU C‐terminal peptide in σE activation and OmpU stability. In this context, we provide evidence for the molecular mechanisms of a negative feedback loop in order to maintain OmpU at physiologically operating levels in V. cholerae.

RESULTS

ΔompU strains tolerate rpoE deletions

The alternative sigma factor σE is well known in Escherichia coli, and it is strictly linked to OM stress induced by high temperatures or OMP overproduction (Erickson and Gross, 1989; Mecsas et al., 1993). Previous work reported that the homologue of rpoE in V. cholerae is not essential for survival (Ding et al., 2004; Kovacikova and Skorupski, 2002). Later findings revealed suppressor mutations in the upstream region of the ompU CDS in a rpoE deletion background (Davis and Waldor, 2009). Therefore, we hypothesized that ΔompU strains tolerate rpoE knockouts, and sequenced the entire operator and promoter regions of ompU, including the 5′ untranslated region (UTR) of WT, ΔompUΔrpoE, several ΔrpoE, and ΔdegS strains generated under the same selective growth conditions (Figure 1). The isolated strains, either lacking the entire rpoE or degS gene, each revealed a single nucleotide deletion within the spacer region of the two ToxR boxes in operator O1 compared to the WT. In contrast, deletion of rpoE in the ΔompU strain resulted in an unaffected ompU operator sequence. It is worth noting that DegS is also required for counteracting membrane stress in V. cholerae when challenged with cationic antimicrobial peptides (Mathur et al., 2007). Since degS mutants cannot liberate σE from the membrane, they cause similar consequences as rpoE knockout strains. Taken together, rpoE and degS deletion strains show suppressor mutations in the ompU expression control region, whereas no suppressor mutations were observed in ΔompU strains.

FIGURE 1

Position of suppressor mutations within the ToxR binding box. Multiple DNA‐sequence alignment in ClustalW format (5′–3′) of ToxR binding box (O1) −184 to −150 bp upstream to the transcription start site +1 of ompU in WT, ΔompUΔrpoE, ΔdegS and ΔrpoE. Single nucleotide deletions in ΔdegS and ΔrpoE strains are marked by a hyphen; identical nucleotides are highlighted by asterisks. ToxR binding boxes are displayed in bold and underlined as described in Crawford et al. (1998)

ΔrpoE strains can still react to bile salts by elevating OmpU levels

Since the upstream region of the ompU promoter is a target for suppressor mutations, we further characterized the ompU promoter control region, focusing on ToxR boxes/operators (Figure 2a). Therefore, and in accordance with previous reports (Davis and Waldor, 2009), we additionally characterized mutations within the ompU operator region in independently generated ΔrpoE strains, all of which featured either insertions or deletions of single base pairs within ToxR box O1 or O2 (Figure 2b). ompU transcription activation directly relies on the membrane located regulator ToxR (Crawford et al., 1998). Three operator sites are located upstream of the transcription start site of ompU for ToxR‐specific binding (Crawford et al., 1998). In the presence of Na‐DC, ToxR activates ompU expression, resulting in bile resistance of V. cholerae (Lembke et al., 2018; Provenzano and Klose, 2000).

FIGURE 2

Analysis of bile salt‐dependent changes in OmpU‐ and ToxR levels in ompU operator mutants. (a) Immunoblot analysis of WCLs obtained from WT, ΔrpoE, ompUO1*, ompUΔO1, ompUO2*, ompUΔO2, ompUO3*, ompUΔO3 and ΔompU utilizing α‐OmpU antibodies grown in LB supplemented without or with 0.1% Na‐DC ON. (b). Operator regions with respective ToxR binding boxes are displayed in bold each with the respective spacing sequences of 5 bp. Promoter elements are drawn to scale to the transcription start site (+1) and are highlighted in dark grey. The 5′UTR is shown as an elliptical black shape, followed by the ompU CDS displayed as a light grey arrow, corresponding to the sequence published in (Sperandio et al., 1996). Light grey boxed sequences represent the loci of suppressor mutations: operator O1 was affected either by an insertion or a deletion of a single adenine within its spacing region found in four independently generated ΔrpoE strains. Furthermore, we characterized an insertion of 5′CAC AAC AA3′ at position +46 of ompU 5′UTR that corresponds to a duplication. To understand the relevance of each single operator, artificial point mutations and deletions of respective operator regions were inserted into the chromosome which were marked by an asterisk or by curved parenthesis, respectively

We therefore asked next whether supplementation with Na‐DC affects OmpU levels in ΔrpoE or ompUO1* strains carrying a single nucleotide deletion that resulted in space interference between the two ToxR boxes. Equal amounts of protein of whole cell lysates (WCLs) obtained from WT, ΔrpoE, and ompUO1* cultures incubated in LB supplemented with or without Na‐DC for immunoblot analysis were separated by polyacrylamide gel electrophoresis (Figure S1a,b), and analyzed using α‐OmpU (Figure 2a) or α‐ToxR (Figure S1c) antibodies (Fan et al., 2014; Salem et al., 2015). ToxR levels were similar in all mutant strains cultivated in media containing Na‐DC. On the contrary, strains harboring ompUO1* mutations still displayed lower OmpU amounts in LB compared to the WT. In contrast, ΔrpoE and ompUO1* produced OmpU equally well as the WT, when incubated in LB supplemented with Na‐DC. Additionally, the ΔrpoE strain revealed a similar survival decrease upon incubation with 0.1% Na‐DC as the WT (Figure S1d).

We further asked whether the operators had distinct roles in bile‐dependent ompU transcription initiation. To address this question, we constructed artificial deletions, either by complete removal of the spacer region between ToxR boxes within one operator site, or by creating spacing interference between two ToxR boxes by single nucleotide deletion. Subsequently, we monitored OmpU (Figure 2a) and ToxR levels (Figure S1c) by immunoblot analysis of WCLs of these strains after incubation in LB with or without Na‐DC, and also performed loading controls (Figure S1a,b). ToxR levels remained at comparable levels regardless of Na‐DC addition. Moreover, ompUΔO1 and ompUO2* were still able to mediate a response to Na‐DC with proportionally higher OmpU levels compared to the respective untreated samples. Furthermore, these mutants also displayed reduced OmpU levels compared to WT‐derived OmpU levels produced upon bile salt exposure. In addition, ompUΔO2 revealed a massive attenuation in bile‐dependent OmpU production. Interestingly, both mutations in operator O3 resulted in an entire loss of OmpU at any condition tested. Thus, strains with an ompUO1* genotype are still capable of elevating OmpU levels upon bile exposure. In contrast, other mutated operator variants are limited in OmpU production, highlighting operator O3, which is unable to initiate ompU transcription, even though ToxR is present.

σE‐mediated transcriptional control of degP depends on OmpU

DegP is a multimeric periplasmic protein, which can act as either a protease or chaperone (Spiess et al., 1999). Each DegP subunit provides a PDZ domain, similar to the membrane‐embedded site‐1 protease DegS, inducing its protease activity upon interaction with the protease active center (Kolmar et al., 1996; Krojer et al., 2008).

A microarray study indicated that degP expression is dependent on σE release in V. cholerae (Ding et al., 2004). Consequently, we used degP transcription as a readout for σE activation to further characterize the link between OmpU and the σE pathway. To validate differences in degP regulation in an OmpU‐dependent manner, we measured alkaline phosphatase (PhoA) activities in degP‐phoA reporter strains harboring a transcriptional fusion of the promoterless phoA gene to intact degP (Figure 3a). Upon deletion of ompU, a significant decrease in PhoA activity was observed, which was similar to that observed in the ΔrpoE and ΔompUΔrpoE strains.

FIGURE 3

OmpU dependency of degP expression. (a) Alkaline phosphatase activities (Miller units) were quantified in WT, ΔompU, ΔompUΔrpoE and ΔrpoE harboring a degP‐phoA transcriptional fusion grown in LB ON. Data represent mean values ± standard deviation of six biological replicates and a total sample size of 24. Significant differences between the WT and respective deletion strains are indicated by an asterisk (one‐way ANOVA followed by a Dunnett’s multiple comparisons test, *p < .05). (b) Alkaline phosphatase activities (Miller units) of degP‐phoA transcriptional fusion were measured in strain backgrounds of WT pBAD, ΔompU pBAD, ΔompU pBADompU, ΔompUΔrpoE pBAD, ΔompUΔrpoEpBADompU, ΔrpoE pBAD and ΔrpoE pBADrpoE (synonym for pBAD18‐Kanbla‐rpoE‐FLAGrseABC), see Table 1. After 2 h of induction with 0.05% arabinose in LB in mid‐log phase, cell samples were harvested and measured. Data represent mean values ± standard deviation of six biological replicates and a total sample size of 12. Significant differences are indicated by an asterisk (one‐way ANOVA followed by a Sidak’s multiple comparisons test, *p < .05)

TABLE 1

Strains and plasmids used in this study

Strain or plasmid	Genotype, description, resistance	References
E. coli
DH5αλpir	F‐Φ80ΔlacZΔM15Δ(argF lac)U169 deoR recA1 endA1 hsdR17 (rK ‐mK +) supE44 thi‐1 gyrA69 relA1, λpirR6K	Hanahan (1983)
SM10λpir	thi thr leu tonA lacY supE recA::RP4‐2‐Tc:Mu λpirR6K, Kmr	Miller and Mekalanos (1988)
V. cholerae
WT	P27459‐S, O1 Inaba, El Tor, clinical isolate, Bangladesh 1976, spontaneously Smr	Pearson et al. (1993)
ΔtoxR	P27459‐S ΔtoxR, deletion in toxR, Smr	Fengler et al. (2012)
ΔompU	P27459‐S ΔompU, 825 bp in‐frame deletion in ompU, Smr	Provenzano and Klose (2000)
ΔdegS	P27459‐S ΔdegS::cat, degS replaced by cat cassette, single nucleotide deletion in ompU operator O1, Smr, Cmr	Lembke et al. (2018)
ΔdegP	P27459‐S ΔdegP::cat, degP replaced by cat cassette , Smr	Lembke et al. (2018)
ΔrpoE	P27459‐S ΔrpoE::cat, rpoE replaced by cat cassette, single nucleotide deletion in ompU operator O1 ompUO1*, Smr, Cmr	This study
ΔompU ΔrpoE	P27459‐S ΔompU ΔrpoE::cat, rpoE replaced by cat cassette in ΔompU background, Smr, Cmr	This study
ompU ΔYDF	P27459‐S ompUΔYDF, in‐frame deletion of the C‐terminal amino acids Tyr, Asp, Phe (AA position 348 ‐ 350) of ompU, Smr	This study
ompU∆O1	P27459‐S ompU∆O1, deletion of ompU operator O1, Smr	This study
ompUO1*	P27459‐S ompUO1*, single nucleotide deletion between the two ToxR boxes in ompU operator O1, Smr	This study
ompU∆O2	P27459‐S ompU∆O2, deletion of ompU operator O2, Smr	This study
ompUO2*	P27459‐S ompUO2*, single nucleotide deletion between the two ToxR boxes in ompU operator O2, Smr	This study
ompU∆O3	P27459‐S ompU∆O3, deletion of ompU operator O3, Smr	This study
ompUO3*	P27459‐S ompUO3*, single nucleotide deletion between the two ToxR boxes in ompU operator O3, Smr	This study
ompU‐gfp	P27459‐S ompU‐gfp, 15 N‐terminal AA of ompU fused in frame to gfp, ompU‐ Smr	This study
ompU‐gfp ΔrpoE	P27459‐S ompU‐gfp ΔrpoE, rpoE replaced by cat cassette in ompU‐gfp background, 15 N‐terminal AA of ompU fused in frame to gfp, ompU‐ Smr, Cmr	This study
WT degP‐phoA	P27459‐S with insertion of pGP704phoA downstream of degP, Smr, Apr	Lembke et al. (2018)
ΔompU degP‐phoA	P27459‐S ΔompU with insertion of pGP704phoA downstream of degP, Smr, Apr,	This study
ΔrpoE::cat degP‐phoA	P27459‐S ΔrpoE::cat with insertion of pGP704phoA downstream of degP, Smr, Apr, Cmr	This study
ΔompU ΔrpoE::cat degP‐phoA	P27459‐S ΔompU ΔrpoE::cat with insertion of pGP704phoA downstream of degP, Smr, Apr, Cmr	This study
ΔompU ompU‐phoA	P27459‐S ΔompU with insertion of pGP704phoA downstream of ompU, Smr, Apr	This study
ΔompU ΔrpoE::cat ompU‐phoA	P27459‐S ΔompU ΔrpoE::cat with insertion of pGP704phoA downstream of ompU, Smr, Apr, Cmr	This study
Plasmids
pCVD442	Suicide vector, oriR6K , mobRP4, sacB, Apr	Donnenberg and Kaper (1991)
pBAD18‐Kanbla‐	Expression vector, oriColE1 , arabinose inducible, deletion of bla sequence, Kmr	Guzman et al. (1995)
pMMB67EH	Expression vector, RSF1010 oriV, RSF1010 oriT, lacI, RSF1010 repABC, IPTG inducible, Apr	Morales et al. (1991)
pACYC184	Cloning vector, orip15A , Tetr, Cmr	Rose (1988)
pGP704phoA	Cloning vector, oriR6K , mobRP4, promoterless phoA of SM10λpir, Apr	Berg et al. (2007)
pTAC3575	pBR322 ori, promoterless phoA and lacZ, Apr	Atlung et al. (1991)
pCVD442rpoE::cat	pCVD442 encoding for up and down fragments of rpoE from P27459‐S flanking a cat cassette, Apr, Cmr	This study
pCVD442ompUΔO1	pCVD442 encoding for up and down fragments of ompU O1 from P27459‐S, Apr	This study
pCVD442ompUO1*	pCVD442 encoding for up and down fragments of ompU O1 from P27459‐S ΔrpoE::cat to achieve a single nucleotide deletion between the two respective ToxR boxes, Apr	This study
pCVD442ompUΔO2	pCVD442 encoding for up and down fragments of ompU O2 from P27459‐S, Apr	This study
pCVD442ompUO2*	pCVD442 encoding for up and down fragments of ompU O2 from P27459‐S to achieve a single nucleotide deletion between the two respective ToxR boxes, Apr	This study
pCVD442ompUΔO3	pCVD442 encoding for up and down fragments of ompU O3 from P27459‐S, Apr	This study
pCVD442ompUO3*	pCVD442 encoding for up and down fragments of ompU O3 from P27459‐S to achieve a single nucleotide deletion between the two respective ToxR boxes, Apr	This study
pCVD442ompUΔYDF	pCVD442 encoding for up and down fragments of ompU from P27459‐S to achieve a deletion of the C‐terminal AA Tyr, Asp and Phe (AA position 348–350)	This study
pCVD442ompU‐gfp	pCVD442 encoding for up and down fragments of ompU from P27459‐S comprising the ompU 5′UTR, RBS and the first 15 AA fused in frame to gfp to achieve a translational fusion	This study
pBAD18‐Kanbla‐ rpoE‐FLAGrseABC	rpoErseABC cloned from P27459‐S on pBAD18‐Kanbla‐ with an in‐frame FLAG‐tag fused to C‐terminal rpoE, Kmr	This study
pBAD18‐Kanbla‐ompU	ompU from P27459‐S on pBAD18‐Kanbla‐, Kmr	This study
pMMB67EHdegP	degP of P27459‐S in pMMB67EH, Apr	Lembke et al. (2018)
pGPdegPphoA	pGP704phoA with degP gene fragment from P27459‐S, Apr	Lembke et al. (2018)
pGPompUphoA	pGP704phoA with ompU gene fragment from P27459‐S, Apr	Lembke et al. (2018)
pTAC3575ompU	Operator and promoter region of ompU from P27459‐S cloned in the same orientation as promoterless lacZ on pTAC3575, Apr	This study
pTAC3575ompUO1*	Operator and promoter region of ompU harboring a single nucleotide deletion in operator O1 from P27459‐S ΔrpoE::cat cloned in the same orientation as promoterless lacZ on pTAC3575, Apr	This study

To further elucidate the role of OmpU in σE pathway activation, we performed complementation studies of either ompU or rpoE using ΔompU, ΔompUΔrpoE, and ΔrpoE strains, respectively. In these strains, we monitored degP expression upon plasmid‐derived induction of ompU expression for 2 h in LB medium in mid‐log phase compared to the vector control cultivated under the same conditions (Figure 3b).

A significant elevation in PhoA activity occurred upon ompU expression in trans compared to the empty vector control with ΔompU background. On the contrary, PhoA activities remained at low level in ΔompUΔrpoE cells upon expression of ompU in trans, indicating that degP expression in such strain background shows co‐dependency on RpoE and OmpU. Moreover, rpoE overexpression in the ΔrpoE strain led to significantly elevated PhoA activities compared to the respective empty vector control, indicating complementation. Furthermore, significantly reduced PhoA activities were also observed in ΔdegS grown in LB (Figure S2). Thus, degP expression and, therefore, the activation of the σE pathway directly correlate with the absence or presence of OmpU.

Lethality caused by ompU overexpression in ΔrpoE strains can be rescued by DegP

Viable ΔrpoE strains are only obtainable with accompanying suppressor mutations in the ompU expression control regions. Such mutations lead to artificial downregulation of ompU expression phenotypes. To characterize such strains for an ompU dependent growth deficiency, we compared the survival fitness of ΔompU to ΔompUΔrpoE upon induction of plasmid‐derived overexpression of ompU (Figure 4a, S3a). Similar survival rates were observed in all strains tested at 0 h. After 2 h of induction of plasmid‐derived ompU expression in the ΔompUΔrpoE strain, cell viability significantly decreased compared to the complemented strain ΔompU pBADompU. After 24 h, survival rates of the ompU‐expressing double knockout strain dropped significantly, highlighting a disadvantage for the rpoE deletion mutant. Strains carrying empty vector controls displayed no significant survival impairment at any of the time points analyzed. Thus, ΔrpoE strains are incapable of tolerating high OmpU levels, resulting in severe lethality.

FIGURE 4

OmpU toxicity in ΔrpoE and rescue by DegP. (a). Survival plating of ΔompU and ΔompUΔrpoE carrying pBAD or pBADompU, respectively, was performed after induction of plasmid‐derived expression with 0.05% arabinose in mid‐log phase at time points 0, 2, and 24 h. (B). Shown are survival plating of strains ΔompUΔrpoE carrying pBAD or pBADompU under similar conditions as used in A) except that Na‐DC (0.1%) was added. Data in Figure A and B represent median values ± interquartile range of four biological replicates and a total sample size of at least 10. Significant differences (p < .05) between data sets at a given time point are marked by a symbol for the following comparisons: in panel A * ΔompUΔrpoE pBAD w/o pBAD versus ΔompUΔrpoE pBADompU; in panel B * ΔompUΔrpoE pBAD w/o versus ΔompUΔrpoE pBADompU w/o, + ΔompUΔrpoE pBAD w/o versus ΔompUΔrpoE pBAD Na‐DC, # ΔompUΔrpoE pBADompU w/o versus ΔompUΔrpoE pBADompU Na‐DC (Kruskal–Wallis test followed by Dunn’s multiple comparisons test; for a representative loading and quality control, see Kang‐stained gel in Figures S3A). C. log10 fold change presentation of survival plating of time point 2 h compared to time point 0 h obtained from ΔompUΔrpoE harboring pBAD and pMMB, pBADompU and pMMB, or pBADompU and pMMBdegP, respectively, was conducted after induction of plasmid‐derived expression with 0.05% arabinose and 1 mM IPTG in mid‐log phase. Data represent median with interquartile range of four biological replicates and a total sample size of 4. Significant differences are indicated by an asterisk (Kurskal–Wallis multiple comparison, *p < .05; for a representative loading and quality control, see Kang‐stained polyacrylamide gel in Figures S4A)

We also tested stress and nonstress conditions using Na‐DC in the growth media. This was done to characterize whether a general lethality phenotype of ompU overexpression or an inadequately bile response was exposed. Consistent with Figure 4a, we found a survival deficiency under nonstress conditions upon ompU expression (Figure 4b, compare ΔompUΔrpoE mutant with pBAD w/o and pBADompU w/o). In presence of Na‐DC ΔompUΔrpoE pBAD revealed a significant defect at 24 h, consistent with literature (Simonet et al., 2003; Zingl et al., 2020) since OmpU acts as a resistance factor, for example, versus Na‐DC. Furthermore, in presence of Na‐DC OmpU toxicity upon overexpression in ΔompUΔrpoE pBADompU strain, was compensated due to the positive effect of OmpU activity in bile salt resistance. Thus, toxicity of OmpU in ΔompUΔrpoE is a conditional defect as it can be negated upon presence of an antimicrobial compound like bile salts, which required OmpU. Next, we asked whether the severe toxicity of high OmpU levels in ΔrpoE cells was related to the inability to produce DegP or other related proteases such as by chance DegS or Tsp (Teoh et al., 2015). Accordingly, degP was expressed simultaneously to ompU or the respective empty vector control in trans in the ΔompUΔrpoE strain. After 2 h of induction of both plasmid‐derived promoter systems, OmpU‐induced lethality of ΔompUΔrpoE cells was significantly reduced compared to the strain exclusively expressing ompU (Figures 4c, S3b). However, in similar experiments performed, we found no evidence for lethality reduction in the presence of degS or tsp encoding plasmids (Figure S3d). We conclude that high OmpU levels in the ΔrpoE strain result in severe lethality, which is influenced by the inability to produce periplasmic chaperones/proteases, for example DegP, as this phenotype can be partially rescued by degP overexpression.

ΔrpoE causes elevated ompU expression

Based on these results, we asked why the ΔrpoE strain could not tolerate high OmpU levels and appeared to be unable to regulate ompU expression accordingly. To overcome the limitation of ompU expression, we aimed to establish a reporter system to quantify and analyze ompU expression. Therefore, we cloned the entire ompU operator, promoter, and 5′ UTR regions into the promoter‐probe plasmid pTAC3575 encoding for promoterless lacZ. To prove differential ompU expression regulation in WT and ΔrpoE strains, we quantified β‐galactosidase (LacZ) activities (Figure 5a). A significant increase in LacZ activities were observed for pTACompU carrying ΔrpoE cells compared to the WT, indicating a σE‐dependent negative transcriptional effect on ompU expression. Furthermore, as previously shown, suppressor mutations in the ompU promoter control region of ΔrpoE cells impair ompU promoter‐activated LacZ reporter activity (Davis and Waldor, 2009). In line with these findings, the WT and ΔrpoE strain, the latter inherently encodes a ompUO1* mutation and carries pTACompUO1* lacZ‐reporter plasmid, exhibited a significant decrease in LacZ activities, regardless of the presence or absence of σE. These data indicate that suppressor mutations emerging upon rpoE deletion provoke a pronounced downregulation of ompU transcription, counteracting the loss of σE‐dependent repression.

FIGURE 5

σE‐dependency of ompU expression at consistent ToxR levels. (A). β‐galactosidase activities (Miller units) were quantified in WT and ΔrpoE harboring pTACompU and pTACompUO1*, respectively, obtained from cultures grown in LB ON. Data were normalized to pTAC, respectively, represent mean values ± standard deviation of six biological replicates and a total sample size of 24. Significant differences between the WT and respective deletion strains are indicated by an asterisk (Student’s t test, *p < .05). (b) Immunoblot analysis of WCLs derived from WT, ompUO1*, ΔrpoE, ΔompU and ΔtoxR utilizing α‐OmpU antibodies grown in LB ON. (●) represents a nonspecific cross‐reacting background band. (c). Immunoblot analysis of WCLs obtained from WT, ompUO1*, ΔrpoE, ΔompU, and ΔtoxR utilizing α‐ToxR antibodies grown in LB ON (For a representative loading and quality control, see Kang‐stained polyacrylamide gel in Figures S4A)

To elucidate the amount of OmpU, WCLs of WT, ompUO1*, ΔrpoE, ΔompU, and ΔtoxR strains were used and whole cell lysates (WCLs) were prepared. Subsequently, immunoblot analysis using α‐OmpU antibody was performed (Figure 5b). Both strains harboring a single nucleotide deletion in operator O1, that is, ompUO1* and cellΔrpoE, displayed decreased OmpU levels compared to the WT (Figures 2and 5b). The observation of reduced OmpU levels in ompUO1* strains by immunoblot analysis was in line with the ompU expression quantification profile.

In order to exclude the possibility of variable ToxR levels in WT and ΔrpoE strains, we also performed immunoblot analysis using α‐ToxR antibody on equal amounts of WCLs (Figures 5a and S4a). Interestingly, ToxR stayed at similar levels displaying minor band intensity variations derived from all tested strains, except for the ΔtoxR strain, indicating a ToxR‐independent mechanism for downregulation of ompU. It is to note that other mechanisms cannot be excluded at this point which may lead to changes of ToxR activity, such as ToxR‐ToxS interaction or changes of disulfide bond formation (Lembke et al., 2020). Additionally, the toxR promoter region in ΔrpoE cells did not exhibit any mutations (Figure S4b). Hence, these data provide evidence for a σE‐mediated transcriptional mechanism for adjusting ompU expression to a physiologically appropriate level.

ompU is autoregulated

To link the observed impairment of σE activation of degP‐phoA in the ΔompU strain (Figure 3) with the subsequent elevation of ompU expression under such conditions (Figure 5a), we attempted to determine whether OmpU had an autoregulative effect on its own expression. To test this hypothesis, we quantified LacZ activities in WT and ΔompU strains carrying pTACompU, reflecting ompU promoter activity (Figure 6a). A significant increase in LacZ activity was measured upon deletion of ompU compared to WT cells in mid‐log phase, and this difference was even more pronounced after ON incubation.

FIGURE 6

OmpU feedback response on ompU expression. (a). β‐galactosidase activities (Miller units) were measured in WT and ΔompU harboring pTACompU grown in LB to mid‐log phase and ON. Data were normalized to pTAC, respectively, represent mean values ± standard deviation of four biological replicates and a total sample size of 8. Significant differences between the WT and respective deletion strains are indicated by an asterisk (one‐way ANOVA followed by a Sidak’s multiple comparisons test, *p < .05). (b). GFP fluorescence quantifications (RFU/OD600) were performed in ompU‐gfp carrying pBAD or pBADompU after 2 h of induction of plasmid‐derived expression with 0.03% arabinose in mid‐log phase in LB. Data represent mean values ± standard deviation of eight biological replicates. Significant differences are indicated by an asterisk (Student’s t test, *p < .05). (c). Alkaline phosphatase activities (Miller units) were quantified in WT, ΔompU and ΔompUΔrpoE harboring an ompU‐phoA transcriptional fusion grown in LB ON. Data represent mean values ± standard deviation of six biological replicates and a total sample size of 12. Significant differences are indicated by an asterisk (one‐way ANOVA followed by a Sidak’s multiple comparisons test, *p < .05). (d) GFP fluorescence quantifications (RFU/OD600) were conducted in ompU‐gfp and ompU‐gfp ΔrpoE incubated overnight in LB. Data represent mean values ± standard deviation of three biological replicates and a total sample size of 9. In the case of ompU‐gfp ΔrpoE, eight different clones were tested for each three biological and technical replicates were performed, resulting in a total sample size of 72. Significant differences are indicated by an asterisk (Student’s t test, *p < .05)

These results were also confirmed by monitoring ompU expression in a GFP fluorescence assay using ompU‐gfp, in which GFP was translationally fused to the first 15 amino acids of OmpU. There (Figure 6b) overexpression of ompU in trans resulted in a significant reduction by approximately 50% of relative fluorescence units per OD600 (RFU/OD600).

Moreover, since ΔompU features a partial deletion of the ompU CDS resulting in an OmpU‐deficient strain (Figure 5b), it could still be used for chromosomal ompU‐phoA fusion. To confirm differential ompU regulation in an OmpU‐dependent manner, we measured PhoA activities in ompU‐phoA reporter strains harboring a transcriptional fusion of the promoterless phoA to residual ompU (Figure 6c). In line with the results described above, upon deletion of ompU, a high increase in PhoA activity was monitored, which was even greater in a ΔompUΔrpoE background. These findings were further supported by quantifying ompU expression utilizing the translational ompU‐gfp fusion in an ΔrpoE strain (Figure 6d), revealing a significant elevation of RFU/OD600 in this deletion background. Taken together, these results highlight the existence of an OmpU‐responsive negative feedback autoregulation of ompU expression and additionally provide strong evidence for a σE‐dependent negative effect on ompU transcription.

OmpU C‐terminal YDF motif is not an exclusive trigger for σE activation

Previous work indicated that OmpU acts as a sensor for the σE‐mediated periplasmic stress response triggered by antimicrobial peptides, in which the C‐terminal YDF motif of OmpU appears to be the signal that activates the site‐1 protease DegS in V. cholerae (Mathur et al., 2007). These findings are based on overexpression studies with the 50 C‐terminal amino acids of OmpC fused to cytochrome b52, which is anchored to the inner membrane (Mathur and Waldor, 2004; Mathur et al., 2007; Walsh et al., 2003). OmpU is structurally built up as a homotrimer. The C‐terminal YDF motif is localized at the interaction surface between the monomeric subunits buried in the OM (Li et al., 2018). It is unlikely that this signal motif is recognized by a PDZ domain if OmpU is maintained in its intact, folded and assembled state. Besides binding C‐terminal portions of misfolded proteins, PDZ domains are also known to interact with internal peptide sequences (Liu and Fuentes, 2019). V. cholerae OmpU comprises five YXF motifs, four of which are conserved in ompU among other Vibrio species (Figure S6). We therefore, revisited the impact of the C’‐terminal located YDF motif and characterized a chromosomal deletion mutant of the C‐terminal YDF motif in order to study its effects on the σE pathway. We assessed differences in σE activation by quantifying degP mRNA levels in WT, ΔrpoE, and ompU ΔYDF strains harvested from cultures grown in LB until mid‐log phase (Figure 7a). The degP transcript quantity was 10‐fold reduced in ΔrpoE cells, but, unexpectedly, significantly increased in ompU ΔYDF cells compared to the WT. Additionally, OmpUΔYDF could not be detected by either immunoblot analysis or Kang staining of polyacrylamide gels. To further elucidate whether ompU ΔYDF was insufficiently expressed or served as target of rapid proteolysis, we overexpressed either ompU or ompU ΔYDF in ΔompU, ΔompUΔdegP, and ΔompUΔrpoE strains and prepared WCLs for SDS‐PAGE and immunoblot analysis using α‐OmpU antibodies. OmpU was detected in all strains tested. OmpUΔYDF remained undetectable in ΔompU cells, whereas low amounts were detected in ΔompUΔdegP, and higher OmpU levels in the ΔompUΔrpoE strain as observe in Kang‐stained SDS‐PAGE and immunoblot (Figures S5 and 7b). These data suggest that OmpUΔYDF may occur in an unstable state, and thereby activate σE release, resulting in high protease production and activation rates, for example of degP.

FIGURE 7

Molecular characterization of OmpU terminal YDF motif. (a) Relative expression levels of degP in WT, ΔrpoE and ompU ΔYDF were determined by quantitative real time PCR during growth in LB normalized to rpoB levels. Data represent mean values ± standard deviation of a minimum of three to nine biological replicates and a total sample size of 9 to 27. Significant differences are indicated by an asterisk (one‐way ANOVA followed by a Dunnett’s multiple comparisons test, *p < .05). (b) Immunoblot analysis of WCLs derived from ΔompU, ΔompUΔdegP and ΔompUΔrpoE carrying pBAD, pBADompU or pBADompUΔYDF, respectively, utilizing α‐OmpU antibodies. WCLs were prepared after 2 h of plasmid‐derived expression induction with 0.05% arabinose in LB in mid‐log phase (For a representative loading and quality control, see Kang‐stained polyacrylamide gel in Figures S5)

DISCUSSION

The transition of V. cholerae from an aquatic environment to the human host requires a quick change in its expression profile. Adaption to antimicrobial effectors, for example bile salts, is crucial for the pathogen to fully activate colonization and virulence factor expression. This process involves changes in the OM profile, including the removal of OmpT by outer membrane vesicle production, as well as by transcriptional repression, and the simultaneous activation of OmpU biosynthesis (Provenzano and Klose, 2000; Zingl et al., 2020). This inversion of OMP production is ensured by activation of the main virulence regulator ToxR upon host entry, for example by bile salts (Lembke et al., 2018). ToxR recognizes AT‐rich DNA elements located upstream of a particular CDS with its winged helix‐turn‐helix motif (Goss et al., 2013; Morgan et al., 2019). These specific ToxR boxes are each separated by five base pairs. Moreover, σE is known to be essential for stress responses in E. coli and other Gram‐negative bacteria, and disruption of rpoE leads to suppressor mutations (Button et al., 2007; De Las Penas et al., 1997; Heusipp et al., 2003). Studies of σE in V. cholerae show its requirement for intestinal colonization, in RIP of ToxR, as well as in cell wall damage, which may be provoked by antimicrobial peptides or β‐lactam antibiotics (Almagro‐Moreno et al., 2015; Ding et al., 2004; Kovacikova and Skorupski, 2002; Mathur et al., 2007; Weaver et al., 2018). Several different suppressor mutations within the 5′ upstream regulatory and noncoding regions of ompU were identified upon rpoE depletion, causing reduced OmpU levels (Davis and Waldor, 2009).

In this work, we characterized the survival ability of V. cholerae with respect to the σE pathway and σE‐controlled ompU expression. We present comprehensive data on the relationship between the σE pathway, suppressor genotypes in rpoE knockout mutants, and the expression status of ompU. The ΔrpoE strains analyzed herein exhibited single nucleotide deletions or insertions within the five base pairs stated, separating the two ToxR boxes of an operator, in turn leading to decreased ompU expression, similar to what has been reported earlier (Davis and Waldor, 2009). Remarkably, comparable results were obtained for the ΔdegS strain, which is deficient in the site‐1 protease responsible for RIP of the anti‐sigma factor complex RseA, and liberation of σE (Kim, 2015). This indicates that either the lack of rpoE or the inability to liberate σE from RseA leads to segregation of similar suppressor mutations in the upstream region of ompU, causing a downregulation of ompU expression. In silico analysis of the classical strain O395 revealed an intact rpoE sequence and a similar organized ompU upstream region. Interestingly, the spacer region between ToxR boxes O1 and O2 harbors one additional adenine, similar as found herein and by Davis and Waldor (2009) in El Tor rpoE knockout backgrounds. In our opinion, this finding may not indicate evidence for a correlation between rpoE mutants and changed ompU upstream regions per se, and deeper analysis of expression studies of ompU, in combinations with rpoE knockout mutants would be needed for clarification for the classical strain. For the in herein characterized ΔompU ElTor strain the role of OmpU itself was characterized, revealing that rpoE depletion in ΔompU strains was obtained without any mutations in the upstream ompU region. Since the deletion mutations described above were identified in ToxR box O1, it seems likely that this may have caused a conflict, resulting in inefficient DNA accessibility for ToxR or ToxR‐ToxR interaction (Lembke et al., 2020), and eventually reduced transcription initiation of ompU.

Additional data show, as discussed below, that suppressor mutations in the ompU expression control region were selected for rescue of cell growth, as the levels of OmpU in ΔrpoE strains play a critical role. As established earlier, the ompU promoter control region consists of three ToxR binding sites, O1‐O3, each separated by a spacer region (Goss et al., 2013; Morgan et al., 2019). O3 is located immediately upstream of the core promoter, followed by O2 and O1, respectively. Comprehensive analysis of a single base pair deletion within and complete removal of the entire spacer region from O1 to O3 revealed successive loss of ompU expression strength. Therefore, the spacer region of operator O3 was identified to be essential. Interestingly, bile salt‐dependent activation of ompU expression was activated in a ΔrpoE background, as well as in the O1* and O2* constructed strains. In contrast, deletions of the respective spacer regions always resulted in a decrease in bile‐dependent ompU expression. Therefore, we conclude that by genetic evidence all three operator regions are important in order to fully respond to bile salts and disabled operator‐ToxR interaction is compensated in O1* and O2* strains. Interestingly, we recently published that in a bacterial two hybrid system ToxR‐ToxR interactions have the highest activity in the presence of operator sites and bile salts (Lembke et al., 2020). This may indicate that bile influences ToxR interaction in a way to better fit to operators even if the spacing is reduced. Furthermore, if bile salts are present, V. cholerae cells tolerate higher levels of ompU expression, even in a ΔrpoE deletion background. These data suggest that another physiological response is activated in the presence of bile salts, providing high levels of OmpU in WT and ΔrpoE strains, not leading to a growth conflict.

The above discussed results indicate that ompU expression itself causes a condition in favor of selection of suppressor mutations in strains lacking rpoE or degS. Therefore, we characterized the impact of ompU expression and its role in the σE pathway response by monitoring the expression of degP. Obtained data showed a strong decrease in degP expression in ΔrpoE, ΔompU, and ΔompUΔrpoE strains. Functional complementation reversed this phenotype in both single deletion strains, whereas ompU complementation in the double knockout strain had no effect on degP expression. Thus, these data indicate that while OmpU positively controls σE activity, which in turn acts negatively on ompU expression, yields in the conclusion that deletion of rpoE or ompU acts negative on degP expression control. Additionally, our data confirm and identify OmpU as a trigger for σE pathway activation, as suggested previously (Mathur et al., 2007).

Further experiments addressed the role of OmpU in growth and survival. Overexpression of ompU was associated with growth defects in rpoE‐deficient strains. Moreover, toxicity of the ompU overexpression in ΔompUΔrpoE is negated in presence of bile/Na‐DC. Interestingly, survival deficiency was rescued by DegP, indicating that loss of DegP‐controlled OmpU synthesis or assembly leads to harmful conditions in the periplasm or OM. Hence, it is tempting to speculate whether a σE‐controlled downstream factor, for example DegP, is involved in the feedback regulation of OmpU.

Such feedback response adjusts the regulation of ompU expression according to cell stress and survival function. Thereby, we asked whether this control mechanism links ompU expression to the σE pathway. The underlying mechanism may involve factors that are under σE control acting negatively on ompU expression. Based on the fact that ΔrpoE strains exhibited suppressor mutations causing ompU downregulation, characterization of ompU expression in the rpoE knockout strain was limited. Therefore, we reconstructed one suppressor mutation (O1*) on the chromosome in a rpoE background. Consistent with the original rpoE suppressor mutant, the reconstructed strain (O1*) identified this mutation as an ompU promoter‐down mutation exhibiting lower OmpU protein levels in an rpoE background. No toxR mutation or change in ToxR levels was observed in these strains; therefore, we found no evidence for any involvement of ToxR. Furthermore, a plasmid‐based quantification method was used to measure ompU transcription under WT and O1* promoter activity in a promoter‐probe plasmid in WT and rpoE knockout strains (Atlung et al., 1991). The results showed that WT ompU promoter activity tested in the rpoE knockout strain was accompanied by a strong increase in ompU expression, whereas the O1* promoter mutation caused its downregulation, independent of rpoE. These data strongly suggest the existence of a yet unknown negative feedback regulation circuit between the σE pathway and ompU expression. Recently, σE‐dependent expression of the small inhibitory regulative RNA MicV was found to affect ompU expression (Peschek et al., 2019). However, a micV knockout mutant did not interfere with ompU expression under the conditions tested (data not shown), suggesting an involvement of additional components.

To gain evidence in support of the involvement of OmpU itself within a negative regulatory pathway, we tested transcriptional and translational reporter fusions in WT and ΔompU strains. For all reporter systems used, we identified autoregulatory effects, showing increased ompU promoter activity for the plasmid‐encoded ompU promoter‐lacZ fusion, and for chromosomally encoded ompU‐phoA reporter systems in the ompU background. Additionally, similar effects were observed in an ompU‐gfp reporter strain upon ompU overexpression. Hence, OmpU also has a negative effect on its own expression, operating in a negative feedback mode.

Since OmpU seems central to the activation of the σE system, a more detailed characterization of OmpU and its role as a sensor was performed. Our findings indicate that DegS’s PDZ domain also recognizes internal YXF motifs of OmpU, eventually resulting in σE release, and causing elevated degP expression in ompU ΔYDF cells compared to the WT. Moreover, degP is under σE control, comprises PDZ domains, and is involved in the periplasmic stress response by acting either as a multimeric chaperone or protease on misfolded periplasmic proteins, for example OmpU (Kolmar et al., 1996, Krojer et al., 2008). Plasmid‐derived OmpUΔYDF protein was hardly detectable in ΔompU cells. Remarkably, OmpUΔYDF was regained successively in the ΔdegP and ΔrpoE strains, respectively, indicating σE‐dependent proteolysis, for example by DegP.

In summary, the present work states multiple levels of OmpU regulation that ultimately influence the effectiveness of V. cholerae in establishing a σE pathway response. The ability to modulate the OM, and eventually the presence of OmpU in the OM, affects its colonization fitness, bile resistance, and intestinal survival. Therefore, fine‐adjusted orchestration of properly folded OmpU under unstressed conditions, accurate expression of ompU at a physiologically relevant level, and feedback responses to membrane stress are highlighted by various experiments reported herein. It is appealing to speculate that OmpU levels and folding states are crucial for proper physiological counteraction of membrane stress conditions. The results obtained in this study point to the possibility of the existence of a σE‐regulated repressor element that is responsible for transcriptional downregulation of ompU, as suggested by our model (Figure 8). Thus, negative feedback regulation of ompU expression and stability by σE‐dependent factors could represent a fine‐tuned mechanism on multiple levels relevant for proper OmpU function and modulation of the signal transducer for membrane stress.

FIGURE 8

A regulatory circuit indicates the interplay between the major outer membrane porin OmpU, the σE response pathway and periplasmic stress stimuli in V.cholerae. This study reveals a negative feedback regulation resulting in fine‐tuning of the ompU expression level, balancing stress response and survival. In (a), OmpU serves to trigger DegS proteolytic attack (first scissor in the periplasm scheme) at the anti‐sigma factor RseA followed by attack through RseP (second scissor intra membrane scheme), which in consequence lead to a release of RpoE/ σE to act on gene regulation, for example, degP and others not shown. Also shown as dark gray arrow, a yet unknown regulatory factor mediate negative regulation on ompU expression. (b) in case of rpoE knockout mutation no negative regulation takes place, subsequently ompU promoter activity increases, as shown herein. As a consequence, rpoE deletion mutants can only be obtained if point mutations arise which decrease ompU expression. (c) for WT the presence of bile leads to increased ompU expression, either indicating bona fide ToxRS activation, or through inhibition of RpoE negative feedback regulation. Whether bile activation mechanism is separated from RpoE pathway or acts in concert with ToxRS must be further characterized

EXPERIMENTAL PROCEDURES

Bacteria, plasmids and growth conditions

Bacterial strains and plasmids used in this study are listed in Table 1; oligonucleotides are listed in Table 2. V. cholerae P27459‐S, a spontaneous streptomycin (Sm)‐resistant mutant of the clinical isolate O1 El Tor was used as wild type (WT) (Pearson et al., 1993). Escherichia coli strains DH5αλpir and SM10λpir were used for genetic manipulation (Hanahan, 1983; Kolter et al., 1978; Miller and Mekalanos, 1988). The latter strain was used to deliver plasmids to V. cholerae by conjugation. Unless stated otherwise, strains were cultured in lysogeny broth (LB) (1% tryptone, 1% NaCl, and 0.5% yeast extract) either shaking with 180 rpm or in agar plates with aeration at 37°C. V. cholerae was usually incubated overnight (ON) or inoculated with a starting OD600 of 0.1 and grown to mid‐log phase followed by the induction of plasmid‐derived expression. If required, antibiotics or other supplements were used in the following final concentrations: streptomycin (Sm) 100 µg/ml, kanamycin (Km) 50 µg/ml, chloramphenicol (Cm) 2 or 5 µg/ml, ampicillin (Ap) 100 µg/ml or 50 µg/ml in combination with other antibiotics, D‐glucose (Gluc) 0.2%, L‐arabinose (Ara) 0.05%, sodium‐deoxycholate (Na‐DC, Sigma‐Aldrich) 0.1%.

TABLE 2

Oligonucleotides used in this study

Primer name	Sequence (5′ to 3′)†
F1_rpoE‐KO_XbaI_fw	ATATCTAGAGCGGCCCTTTGCTTAATA
F1_rpoE‐KO_HindIII_rv	ATAAAGCTTTCGAGCGGTCACTCCTAT
cat‐pACYC184_HindIII_fw	ATAAAG CTTAGCACCTCAAAAACACCATC
cat‐pACYC184_BamHI_rv	TTAGGATCCCACCAGGCGTTTAAGGGCA
F2_rpoE‐KO_BamHI_fw	ATAGGATCCCGCAAATTCCGTAATGAC
F2_rpoE‐KO_SacI_rv	ATAGAGCTCTCATTCCACAACCGATTC
F1_ompU‐KO_SacI_fw	ATAGAGCTCAAATCTTTATGCTTTATGGTTGAA
F1_ompU‐KO_NdeI_rv	AATCATATGATTAAGTCCTAATTTA
F2_ompU‐KO_NdeI_fw	AATCATATGGCGTTTACTGCGACAT
F2_ompU‐KO_SphI_rv	ATAGCATGCAGTATAACCGAGATAGAAAAACGCCT
F1_ompU_XbaI_fw	AATCTAGAACAAAGGCCAATTTGGTG
F2_ompU_SacI_rev	ATTGAGCTCATCTGCGGTGGTCAG
F1_ompUΔYDF_SOE_rv	CAATTAACGTAGACCGATAGCCAG
F2_ompUΔYDF_SOE_fw	CTACGTTAATTGTTGACTTCAGGTCAC
F1_ompU‐op_XbaI_fw	AATTCTAGATGGTATTCCGCATTCTCTTTC
F2_ompU‐op_SacI_rv	AATGAGCTCTTGAACCTGCTTTGTCACCGC
F1_ompU‐ΔO1_SOE_rv	TGTTGTATTTCGAGATTGAGCAAAATGCACGCAATC
F2_ompU‐ΔO1_SOE_fw	CAATCTCGAAATACAACAAATTAAATTAAAAAAAACACTTAC
F1_ompU‐ΔO2_SOE_rv	AGTTATTGTATGTTTGTTGGTAAGTGTTTTTTTTAA
F2_ompU‐ΔO2_SOE_fw	AACAAACATACAATAACTTGATAAATTTTTACCAAC
F1_ompU‐O2*_SOE_rv	GTTTTTAAAGATTTTAATGTTTGTTGGTAAGTGTTT
F2_ompU‐O2*_SOE_fw	ATTAAAATCTTTAAAAACAATAACTTGATAAATTTT
F1_ompU‐op_SacI_fw	AATGAGCTCTGGTATTCCGCATTCTCTTTC
F2_ompU‐op_SphI_rv	AATGCATGCCAGAGTACTTGGCAGCGT
F1_ompU‐ΔO3_SOE_rv	ATCAGTTAGTCCAACTTATGAACACTGTTTTATTGT
F2_ompU‐ΔO3_SOE_fw	TAAGTTGGACTAACTGATAGCGGAACTTTGGGAGTA
F1_ompU‐O3*_SOE_rv	AGTAAAATGTATAAATCCAACTTATGAACACTGTTT
F2_ompU‐O3*_SOE_fw	GGATTTATACATTTTACTAACTGATAGCGGAACTTT
F1_ompU‐gfp_XbaI_fw	ATTTCTAGACATCATTTACCTCTGCGCCAAAG
F1_ompU‐gfp_SOE_rv	CTTTACTCACTGCAGCAGCTGATACAG
F2_ompU‐gfp_SOE_fw	GCAGTGAGTAAAGGAGAAGAACTTTTCACTG
F2_ompU‐gfp_SOE_rv	TCAACAACTATTTGTATAGTTCATCCATGCC
F3_ompU‐gfp_SOE_fw	CAAATAGTTGTTGACTTCAGGTCACACG
F3_ompU‐gfp_SacI_rv	ATTGAGCTCGTCACGCTGATGGCTTGC
rpoE‐FLAG_KpnI_fw	ATAGGTACCGAAATTGTCGAAAAATTC
rpoE‐FLAG_XbaI_rv	ATATCTAGATTACTTGTCATCGTCGTCCTTGTAGTCCAGAAGAGGTTTGATTTTCTT
ompU_SacI_fw	AATGAGCTCCGTGGCTTACGTCGCACA
ompU_SphI_rv	AATGCATGCTTAGAAGTCGTAACGTAG
ompU‐prom_BamHI_fw	ATTGGATCCTTACCTCTGCGCCAAAGT
ompU‐prom_XbaI_rv	ATTTCTAGATTTGTGCGACGTAAGCCA
rpoB_fw	CTG TCT CAA GCC GGT TAC AA
rpoB_rv	TTT CTA CCA GTG CAG AGA TGC
degP_fw	GCT TCC TCT CTC AGT CAA T
degP_rv	CAG ACG CTG TCT TGA AAC T

Restriction sites are underlined; site directed mutations are in bold.

Construction of in‐frame mutants, translational gfp fusions, expression plasmids and promoter fusions

The isolation of chromosomal DNA, plasmids or PCR products, and the construction of chromosomal transcriptional phoA fusions, constructs and in‐frame deletion mutants were carried out as described previously (Seper et al., 2011). Plasmids, digested plasmids and PCR products were purified using the QIAquick® PCR purification, the QIAquick® gel extraction, or the QIAprep® Spin Mini Kit (Qiagen). PCRs for subcloning and sequencing were performed using Q5® DNA polymerase (New England Biolabs); for all other PCRs Taq DNA polymerase (New England Biolabs) was used.

Deletion strains were constructed by using derivatives of the suicide vector pCVD442 (Donnenberg and Kaper, 1991). Start to stop deletion of rpoE, ompU operator point mutations, ompU operator space region deletions (Figure S1), ompU truncations and translational gfp fusions in V. cholerae P27459‐S were generated via SOE‐PCR (splicing by overlap extension, (Horton et al., 1989)). A cat cassette from pACYC184 was obtained to improve the selection of rpoE deletion strains. The generation of overlapping regions of two PCR fragments allowed the annealing of the two PCR products in another PCR. Respective suicide vectors were generated by PCR amplification of 600–1,000 bp up‐ and downstream of the locus of interest using the corresponding oligonucleotide pairs (Table 2). The resulting fragments were digested with the appropriate restriction enzyme indicated by the name of the oligonucleotide, and finally ligated into the identically digested suicide vector backbone of pCVD442. Subsequently, the cloned suicide plasmids were first transferred into electro competent E. coli DH5αλpir. Positive clones were selected by colony PCR and Sanger‐sequencing (data not shown) and afterward transferred into electro competent SM10λpir in order to transfer the construct to V. cholerae by conjugation. Cells were grown on agar plates containing Sm and Ap to select for clones which integrated the deletion construct into the chromosome by homologous recombination, followed by the incubation on sucrose (10%) agar plates to obtain ApS clones in which in turn an excision of the suicide vector from the chromosome occurred. Correct gene deletions were confirmed by colony PCR and Sanger‐sequencing (data not shown).

pBAD18‐Kanbla‐ expression constructs were generated by PCR or SOE‐PCR of the respective coding region including a ribosome binding site (RBS). The resulting fragments were digested with the appropriate restriction enzyme indicated by the name of the oligonucleotide (Table 2) and eventually ligated into the identically digested pBAD18‐Kanbla‐ vector backbone. The obtained constructs were transferred into electro competent DH5αλpir and KmR colonies were screened by colony PCR for the existing cloned constructs which were additionally verified by Sanger‐sequencing and by immunoblotting for correct expression (data not shown). Subsequently, plasmids were isolated and transferred by electroporation into V. cholerae strains, which were also tested for expression by immunoblotting before being used in other experiments like survival plating and the analysis of membrane stress response.

Derivatives of pTAC3575 were constructed to obtain plasmid‐encoded transcriptional fusions of lacZ to respective operator and/or promoter regions. Respective noncoding regions of ompU and degP, including the annotated transcriptional start site and parts of the 5′UTR, were amplified by PCR or SOE‐PCR using the appropriate oligonucleotide pairs (Table 2). Amplicons were digested by BamHI and XbaI, ligated into identically digested pTAC3575, transferred by electroporation into DH5αλpir and ApR colonies were determined by colony PCR for the existing cloned constructs which were additionally verified by Sanger‐sequencing (data not shown). Subsequently, plasmids were isolated and transferred by electroporation into V. cholerae strains, which were used for the determination of β‐galactosidase activity depending on the power of the cloned promoters in the respective deletion strains.

Alkaline phosphatase (PhoA) assay

To determine the enzymatic activities of transcriptional phoA fusions, alkaline phosphatase assays were performed as described previously (Miller, 1992). Respective V. cholerae strains were grown in LB at 37°C, 180 rpm, ON. Cell disruption was performed with sodium dodecyl sulfate (SDS) and chloroform. Experiments were performed with a minimum of six biological replicates and technical duplicates. PhoA activities were expressed in Miller units, calculated as following: MU = (OD405 × 1000) / (OD600 × tmin × v ml × 0.96).

β‐galactosidase (LacZ) assay

To determine the enzymatic activities of transcriptional lacZ fusions, β‐galactosidase assays were performed as described previously (Miller, 1992). Respective V. cholerae strains were grown in LB at 37°C, 180 rpm, ON. Cell disruption was performed with SDS and chloroform. Experiments were performed with a minimum of six biological replicates and technical duplicates. LacZ activities were expressed in Miller units, calculated as following: MU = (OD405 × 1000) / (OD600 × t min × v ml × 0.486).

Analysis of ompU translation by GFP quantification

To quantify ompU translation in V. cholerae, respective gfp fusion strains were incubated in LB at 37°C, 180 rpm, ON, cultures were pelleted and 1 OD600 unit was resuspended in 1 ml saline. 1:10 dilutions were measured in 96 well fluorescence plates (Costar) using the FLUOstar® plate reader (Omega). The software determined fluorescence with 485 nm excitation and 520 nm emission wavelengths as well as the OD600. Relative fluorescence units (RFU) were normalized to the respective OD600.

Preparation of whole cell lysates (WCLs)

To obtain WCLs of E. coli and V. cholerae, cultures were grown in LB ON or inoculated in fresh LB to an OD600 of 0.1, grown to mid‐log phase followed by induction with 0.05% arabinose or as specified of plasmid‐derived expression for at least 2 h with aeration at 37°C, and 180 rpm shaking. Cell equivalents of 1.5 OD600 units were harvested by centrifugation in an Eppendorf centrifuge (5 min at 6,500 ×g), resuspended in 60 µl Laemmli buffer (Laemmli, 1970), boiled for 30 min at 100°C and used for standard sodium dodecyl sulfate‐polyacrylamide gel electrophoresis (SDS‐PAGE). WCLs were normalized to similar protein amounts by SDS‐PAGE Kang staining (Kang et al., 2002) before being used for immunoblot analysis.

SDS‐PAGE and immunoblot analysis

The standard SDS‐PAGE procedure was performed using 15% polyacrylamide gels (Mini‐PROTEAN® Tetra cell, BIO‐RAD) in combination with the PageRulerTM Prestained Protein Ladder (10–180 kDa, Thermo Fisher Scientific) as a molecular mass standard for separated proteins. Protein bands were stained according to Kang et al. (2002) or transferred to an AmershamTM ProtanTM 0.45 µm nitrocellulose membrane (GE Healthcare Life Sciences) for further immunoblot analysis. After blocking in tris‐buffered saline (TBS) with 10% skim milk at 4°C ON, the membranes were incubated with the respective primary antibody at RT for 2 h: α‐ToxR rabbit serum diluted 1:1,000 in TBS 10% skim milk (Fan et al., 2014), kindly supplied by Jun Zhu, University of Pennsylvania, USA; α‐OmpU mouse serum diluted 1:5,000 in TBS 10% skim milk (Salem et al., 2015). After several washing steps as previously described by Fengler et al. (2012), the membranes were incubated with the respective secondary antibody at RT for 1 h: horseradish peroxidase conjugated goat anti‐rabbit (Dianova GmbH) diluted 1:10,000 in TBS 10% skim milk and horseradish peroxidase conjugated goat anti‐mouse (Dianova GmbH) diluted 1:7,500 in TBS 10% skim milk. The membranes were incubated in ECL solution for 5 min (ClarityTM Western ECL Blotting Substrates, BIO‐RAD) for visualization of reactive protein bands by chemiluminescence using a Molecular Imager ChemiDocTM XRS system (BIO‐RAD). To adjust the protein content of different WCLs, final SDS‐PAGE was performed also using the same sample volumes for immunoblot analysis.

qRT‐PCR analyses

Oligonucleotides used for qRT‐PCR are listed in Table 2 and are labeled by x_fw or x_rv, whereas x represents the respective tested gene. qRT‐PCR was performed as described previously (Schild et al., 2007). RNA was isolated from a minimum of three independent cultures grown in LB medium until mid‐log phase (OD600 = 0.5) using QIAGENTM RNAprotect kit (Qiagen). To remove chromosomal DNA, the samples were treated with RQ1 RNase‐Free DNase (Promega). By using iScriptTM Select cDNA Synthesis Kit (Bio‐Rad), cDNA was synthesized from 200 ng RNA, also including controls without reverse transcriptase. Quantification of cDNA was performed with SYBR GreenERTM qPCR SuperMix for ABI PRISM® instrument (Invitrogen) using a Rotor‐GeneTM 600 and Rotor‐GeneTM 600 Series Software 1.7 (GenXpress). Each reaction contained 400 nM oligonucleotides and 10 ng cDNA template. Each independent sample was tested in triplicates. Subsequently, the mean cycle threshold of the test transcript was normalized to the reference gene, rpoB and to one selected rpoB WT reference. Values above or below 1 indicate an up‐ or downregulation in the mutant, respectively, compared to the WT.

Statistical analysis

Normally distributed data were analyzed by the unpaired Student’s t test in case of single comparison. Ordinary one‐way ANOVA followed by either Dunnett’s or Sidak’s multiple comparison tests were conducted, respectively, for multiple comparisons.

In case of non‐Gaussian distributed data Kruskal–Wallis followed by Dunn’s multiple comparison tests were performed. Differences were considered significant for p values of < .05. GraphPad Prism version 7 was used for all statistical analysis.

Conflict of Interest

The authors declare that the research was conducted without any commercial or financial relationships that could be constructed as a potential conflict of interest.

AUTHOR CONTRIBUTIONS

Designed the study: Pennetzdorfer, Höfler, and Reid. Performed the experiments and/or the analysis: Pennetzdorfer, Höfler, Wölflingseder, and Tutz. Contributed to the discussion and the data evaluation: Pennetzdorfer, Höfler, and Reid. Performed data analysis: Schild, Pennetzdorfer, and Wrote the manuscript: Reid and Pennetzdorfer and Reid.

Supplementary Material

Click here for additional data file.