The multidrug efflux pump regulator AcrR directly represses motility in Escherichia coli

ABSTRACT Efflux and motility are two key biological functions in bacteria. Recent findings have shown that efflux impacts flagellum biosynthesis and motility in Escherichia coli and other bacteria. AcrR is known to be the major transcriptional repressor of AcrAB-TolC, the main multidrug efflux pump in E. coli and other Enterobacteriaceae. However, the underlying molecular mechanisms of how efflux and motility are co-regulated remain poorly understood. Here, we have studied the role of AcrR in direct regulation of motility in E. coli. By combining bioinformatics, electrophoretic mobility shift assays (EMSAs), gene expression, and motility experiments, we have found that AcrR represses motility in E. coli by directly repressing transcription of the flhDC operon, but not the other flagellum genes/operons tested. flhDC encodes the master regulator of flagellum biosynthesis and motility genes. We found that such regulation primarily occurs by direct binding of AcrR to the flhDC promoter region containing the first of the two predicted AcrR-binding sites identified in this promoter. This is the first report of direct regulation by AcrR of genes unrelated to efflux or detoxification. Moreover, we report that overexpression of AcrR restores to parental levels the increased swimming motility previously observed in E. coli strains without a functional AcrAB-TolC pump, and that such effect by AcrR is prevented by the AcrR ligand and AcrAB-TolC substrate ethidium bromide. Based on these and prior findings, we provide a novel model in which AcrR senses efflux and then co-regulates efflux and motility in E. coli to maintain homeostasis and escape hazards. IMPORTANCE Efflux and motility play a major role in bacterial growth, colonization, and survival. In Escherichia coli, the transcriptional repressor AcrR is known to directly repress efflux and was later found to also repress flagellum biosynthesis and motility by Kim et al. (J Microbiol Biotechnol 26:1824–1828, 2016, doi: 10.4014/jmb.1607.07058). However, it remained unknown whether AcrR represses flagellum biosynthesis and motility directly and through which target genes, or indirectly because of altering the amount of efflux. This study reveals that AcrR represses flagellum biosynthesis and motility by directly repressing the expression of the flhDC master regulator of flagellum biosynthesis and motility genes, but not the other flagellum genes tested. We also show that the antimicrobial, efflux pump substrate, and AcrR ligand ethidium bromide regulates motility via AcrR. Overall, these findings support a novel model of direct co-regulation of efflux and motility mediated by AcrR in response to stress in E. coli.

and its homologs can efflux most classes of antibiotics and other toxic compounds entering the cells from the outside, acting synergistically with the permeability barrier provided by the outer membrane of Gram-negatives to prevent the accumulation of antibiotics in these bacteria (3,5).
In E. coli, the impact of efflux on motility was first discovered in a mutant deleted for the acrB gene (9), which encodes for the substrate recognition and binding component of the AcrAB-TolC pump (3)(4)(5)(25)(26)(27).Deletion of acrB produced broad changes in gene expression, especially a very strong upregulation of nearly all flagellum biosynthesis and motility genes (9).Accordingly, this mutant showed an increase in swimming motility compared to the wild-type strain (9).This finding raised the question of how AcrAB-TolC, which is located in the cell envelope, can regulate the expression of flagellum biosyn thesis, motility, and other genes.A later study in E. coli showed that deletion of the acrAB main transcriptional repressor acrR (9,28) produced a similar upregulation of flagellum biosynthesis and motility genes (which was accompanied by increased flagella production), and a similar increase in swimming motility (29) to those we observed in the ΔacrB mutant (9).Changes in motility caused by efflux pumps or their regulators, either similar or opposite to those found in E. coli, have also been found in other bacteria such as Salmonella enterica, Serratia marcescens, or Acinetobacter baumanii (8,16,(30)(31)(32).For example, while AcrR has been found to repress both efflux (9,28) and motility (29) in E. coli, Thota and Chubiz (32) found that, in S. enterica, the transcriptional regulators MarA, SoxS, Rob, and RamA, known to activate the expression of efflux genes, instead repress flagellar gene expression and motility.These findings indicate that the interplay between efflux and motility is both important and specific for the biology of a broad number of bacteria.However, the molecular mechanisms of how such interplay occurs are not well-understood yet.
Here, we have examined the role of AcrR in regulating swimming motility in E. coli.AcrR is a TetR-family transcriptional regulator composed of a C-terminal ligand-bind ing domain and an N-terminal DNA-binding domain (33).The ligands that control the activity of AcrR have been found to be both exogenous antimicrobials known to be AcrAB-TolC substrates, i.e., ethidium bromide, as well as cellular metabolites, i.e., polyamines (9,34,35).Regarding its target genes, AcrR was initially identified as the local transcriptional repressor of the acrAB operon (28), but was later found to also directly regulate the expression of the acrAB activators SoxS and MarA (36), as well as the putrescine degradation gene puuA and the spermidine efflux operon mdtUJI (35).However, it remained unknown whether AcrR also directly regulates flagellum biosynthesis genes and motility in E. coli, and which motility genes are direct targets of this regulator, or whether AcrR indirectly regulates flagellum genes and motility by regulating efflux or other target genes.Here, we report that AcrR is a direct regulator of the flhDC operon, which encodes for the master regulator of flagellum biosynthesis and motility genes, but does not directly regulate the fliE, fliLMNOPQR or fliDST genes/ operons.We also show that such direct regulation of swimming motility by AcrR allows E. coli to synergistically respond to ethidium bromide not only by increasing efflux [activating acrAB expression (35)], but also by increasing swimming motility.These findings support a model in which AcrR senses the accumulation of antimicrobials or cellular metabolites normally effluxed by the AcrAB-TolC multidrug efflux pump and then directly co-regulates efflux and motility to increase efflux and to escape from these compounds.

Bioinformatics analyses suggest that the AcrR transcriptional repressor directly regulates flagellum biosynthesis and motility genes
Prior findings have shown that inactivation of the transcriptional repressor AcrR (ΔacrR::kan mutant) results in a strong upregulation of about 50 flagellum biosynthesis and motility genes and increased flagella production, alongside with an increase in swimming (0.3% agar) but not swarming (0.6%) motility (29).These findings alongside with the identification in the flhDC promoter of three 10 bp fragments with partial overlap with the known 24 bp AcrR-binding site in the acrAB promoter (34), led to the suggestion that the effects of AcrR on motility might be mediated by flhDC (29).The FlhD 4 C 2 complex is the known master regulator that activates the expression of flagellum biosynthesis and motility genes (37,38).
To further investigate the role of AcrR as a direct regulator of motility in E. coli, we performed a whole-genome bioinformatics search of promoters regions with predicted AcrR-binding sites using the known 24 bp AcrR-binding site in the acrAB promoter (34) and the Colibri (http://genolist.pasteur.fr/Colibri/)search tool.We have recently found that AcrR directly regulates the putrescine degradation gene puuA and the spermidine efflux operon mdtUJI, whose promoters contain one predicted AcrR-binding site with 9 mismatches and two sites with 10 mismatches, respectively, compared to the AcrR-bind ing site in the acrAB promoter (35).Therefore, to maximize the finding of potential AcrR target genes, we focused our analysis on hits with promoter regions that contained at least one predicted AcrR-binding site with 11 or less mismatches compared to the known AcrR site in the acrAB promoter.We found that four motility genes/operons fulfilled these criteria: flhDC (motility master regulator; class I), fliE (component of the hook-basal body complex; class II), fliLMNOPQR (components of the flagellar motor switch and flagellar export apparatus; class II), and fliDST (FliD, flagellar filament capping protein; FliST: chaperones of the flagellar export system; class III).
We then computationally analyzed more in-depth the promoter regions of the four flagellum biosynthesis and motility genes identified as potential AcrR direct targets (Fig. 1).For the flhDC promoter region, in addition to the three 10 bp partial AcrR sites previously identified (29), which were found upstream of the −35 element, overlapping the −35 element, and downstream of the flhD translation start codon, respectively, we found two full-size (24 bp) predicted AcrR-binding sites (Fig. 1A).Site 1 was located downstream of the transcriptional start site and contained 10 mismatches compared to the AcrR site in the acrAB promoter.Site 2 had 11 mismatches compared to the AcrR site in the acrAB promoter and was located downstream of the flhD translation start codon and overlapping with one of the 10 bp partial sites identified by Kim et al. (29) (Fig. 1A).For the fliE promoter region (Fig. 1B), we identified three full-size predicted AcrR-binding sites with 11, 12, and 12 mismatches, respectively, compared to the AcrR site in the acrAB promoter.Site 3 overlapped the σ 28 −35 element, and the other two sites were located significantly upstream.For the fliLMNOPQR promoter (Fig. 1C), we found two full-size predicted AcrR-binding sites with 12 and 11 mismatches, respectively, both located significantly upstream of the σ 70 and σ 28 −35 elements.For the fliDST promoter (Fig. 1D), we identified two full-size predicted AcrR-binding sites with 12 and 11 mismatches, respectively, with site 2 overlapping both the σ 70 −35 (fully) and −10 (partially) elements, and site 1 being located upstream.

AcrR represses motility by directly repressing the flhDC operon, which encodes for the master regulator of flagellum biosynthesis and motility genes, in vitro and in vivo, and does not directly regulate fliE, fliLMNOPQR, or fliDST in vitro
We next used EMSAs to determine whether purified AcrR directly binds to the promoter regions of the four flagellum and motility genes/operons identified as potential direct targets of AcrR (Fig. 2A).For each of these four genes/operons, we used promoter fragments that included all their predicted AcrR-binding sites (the primer regions used to amplify each promoter are indicated in green boxes in Fig. 1).In agreement with previous findings (35), AcrR bound to the acrAB promoter (positive control), but not the gapA promoter (negative control).Importantly, we found that AcrR also binds to the promoter identified in this study, and the three partial AcrR-binding sites in the flhDC promoter identified by Kim et al. (29).For each full or partial predicted site, mismatches compared to the AcrR-binding site in the acrAB promoter are indicated in lowercase letters.Greater overall conservation, and especially in the second 10-bp inverted repeat, can be observed for the full predicted AcrR-binding site 1 in flhDCp compared to the full predicted AcrR-binding site 2.This difference may explain why AcrR was found to bind to the promoter fragment with only site 1 (flhDCp S1 ), but not to the promoter fragment with only site 2 (flhDCp S2 ), in the above EMSA (A).
flhDC promoter region (Fig. 2A).This finding is consistent with the increased expression of flagellum biosynthesis genes and increased motility previously found in the ΔacrR::kan mutant (29), and with the presence of the two aforementioned predicted binding sites for AcrR in the flhDC promoter (Fig. 1A).We also found that AcrR does not directly bind to the fliE, fliLMNOPQR or fliDST promoters (Fig. 2A).Overall, these findings indicate that AcrR regulates motility in E. coli by directly regulating transcription of the flhDC master regulator of flagellum biosynthesis and motility.However, we cannot discard that AcrR might also directly regulate other flagellum and motility genes not identified here as potential AcrR targets.The AcrR site in the acrAB promoter contains two 10-bp inverted repeats (Fig. 2B, in bold) connected by a 4-bp spacer (34).AcrR has been suggested to bind to this site as a dimer of dimers, with each dimmer binding opposite to each other on the double-stranded DNA site, and each monomer in a dimer binding to each one of the two 10-bp inverted repeats (33,34).Of the two full predicted AcrR-binding sites identified here in the flhDC promoter, site 1 was the most conserved.In fact, site 1 contains only 10 mismatches compared to the AcrR site in the acrAB promoter, and those mismatches are evenly distributed leaving two inverted repeats each with 60% conservation and thus likely remain functional (Fig. 2B).On the contrary, site 2 contains 11 mismatches, which results in a 70% conserved left inverted repeat but a 30% poorly conserved right inverted repeat (Fig. 2B).We hypothesize that this lack of conservation on the right inverted repeat of site 2 may hinder binding of its corresponding AcrR monomer.
Interestingly, when two different fragments of the flhDC promoter region were tested, we found that AcrR binds to the fragment that contained the full predicted AcrR site 1 identified here [plus two of the partial sites identified by Kim et al. (29)], but not to the fragment that contained only the full predicted AcrR site 2 (Fig. 2A).These findings are in agreement with the greater overall conservation, including two likely functional inverted repeats, described above for site 1 compared to site 2.However, it is possible that both sites 1 and 2 [and perhaps the additional partial AcrR-binding sites identified by Kim et al. (29)] are important for regulation of flhDC by AcrR in vivo, or that prior binding of AcrR to site 1 contributes to binding of AcrR to site 2.
After finding that AcrR directly binds to the flhDC promoter, we next studied the effect of AcrR on the expression of the flhDC operon using real-time quantitative PCR (RT-qPCR) assays (Fig. 3).The expression of flhDC increased by 2.3-fold in the ΔacrR mutant compared to the parental strain (Fig. 3), in agreement with our EMSA results, and the 2.3-to 2.6-fold increase found for this operon in the ΔacrR::kan mutant by Kim et al. ( 29) using a microarray.As expected, complementation of the ΔacrR mutant by expressing acrR from the pBAD18-Kan-acrR plasmid reduced the expression of the flhDC operon in this mutant to a level close to that in the parental strain (Fig. 3), although there was a residually higher, but non-statistically significant, flhD expression in the complemented strain that we speculate was caused by acrR expression from the pBAD18-Kan-acrR plasmid having only been induced for one hour.This hypothesis is supported by the fact that full complementation of the ΔacrR mutant was observed in the motility experiments described below (Fig. 4A), which involve much longer induction times.Overall, our gene expression findings further support our bioinformatics and EMSA results, and thus the role of AcrR as a direct transcriptional repressor of flhDC.

AcrR represses swimming motility in E. coli at both 37°C and 30°C
We next studied whether AcrR regulates swimming motility in E. coli at 37°C (human host temperature), as well as at 30°C (environmental temperature) for comparison with prior motility assays performed at 30-33°C (9, 29) (Fig. 4A and B).We found that deletion of acrR significantly increased swimming motility in E. coli at both 37°C and 30°C, by 1.8-fold and 1.7-fold, respectively.Overall, our motility results with the ΔacrR mutant at both 37°C and 30°C are in agreement with the increased swimming motility found at 33°C in a ΔacrR::kan mutant by Kim et al. (29), confirming that such increased motility was caused by the lack of acrR, and not by the kan cassette.Moreover, complementation of the ΔacrR mutant by expressing acrR from the pBAD18-Kan-acrR plasmid significantly decreased motility in this mutant, completely restoring its motility to the parental levels at both temperatures (Fig. 4A).These complementation experiments, not included in the study of Kim et al. (29), further confirm the specific effect of acrR in repressing motility in E. coli.In addition, our motility results are in agreement with our findings that AcrR directly binds to the promoter of the flagellum biosynthesis and motility regulator flhDC operon and represses its expression (Fig. 1 to 3), further supporting the role of AcrR as a direct repressor of swimming motility.
Finally, we tested whether AcrR played a role in the increased swimming motility previously found by Ruiz and Levy (9) when the AcrAB-TolC efflux pump was inactivated by deletion of the acrB gene (Fig. 4B and C).Given the role of AcrR in directly regulating flhDC and swimming motility reported here (Fig. 1 to 4), and that cellular metabolites that accumulate in the ΔacrB mutant have been suggested to bind to and inactivate AcrR (9,17), we hypothesized that inactivation of AcrR by cellular metabolites contributes to the increased motility previously found in the ΔacrB mutant.Thus, we also hypothesized that overexpression of AcrR would compensate such metabolic inactivation and restore motility in the ΔacrB mutant.We found that the ΔacrB mutant showed an increase in swimming motility of 1.8-fold at 37°C and of 1.3-fold at 30°C compared to the parental strain (Fig. 4B and C), in agreement with our prior findings at 30°C for this mutant (9).Interestingly, overexpression of acrR using the pBAD18-Kan-acrR plasmid restored motility in the ΔacrB mutant at both temperatures (Fig. 4C).This finding strongly supports our hypothesis that inactivation of AcrR by cellular metabolites contributes to the increase in motility found when AcrAB-TolC efflux pump is inactivated, and supports the model summarized in Fig. 5 and discussed below.To further test the hypothesis that ligands that bind to and inactivate AcrR contribute to the increased motility found in the ΔacrB mutant, we studied the effect of adding ethidium bromide to the ΔacrB mutant containing pBAD18-Kan-acrR plasmid (Fig. 4D).Ethidium bromide is an exogenous molecule not produced or metabolized by E. coli, and has recently been found to bind to and inactivate AcrR (35).We used low ethidium bromide concentrations to minimize its toxic effects in the ΔacrB mutant.Consistent with our hypothesis, addition of ethidium bromide to the ΔacrB+pBAD18-Kan-acrR mutant made this mutant hypermotile again, FIG 5 Model of the role of AcrR in co-regulation of efflux and motility in E. coli.We hypothesize that the two main functions of the AcrAB-TolC pump are to efflux antimicrobials such as ethidium bromide (1-7) and to efflux cellular metabolites (9,10,17,18), whereas the transcriptional repressor AcrR acts as the main sensor and gene-expression effector of this pump.In normal conditions (A), efflux would prevent the accumulation of antimicrobials or cellular metabolites, and AcrR would keep at basal levels the expression of acrAB (9), its transcriptional activators MarA and SoxS (9), the puuA and mdtJI genes for polyamine detoxification and efflux (35), and the flhDC motility master regulator (Fig. 1 to 3).(B) When efflux is insufficient or the AcrAB-TolC pump is deleted, cells would accumulate antimicrobials such as ethidium bromide and/or cellular metabolites such as polyamines that bind to and inactivate AcrR (35).These cellular metabolites (9, 10, 17, 18), which would be themselves AcrAB-TolC substrates, or intermediates, end-products or by-products of these substrates, may either function as siderophores, signaling molecules, be toxic, or cause cellular stress because their accumulation would disrupt the normal metabolic flow of cells.Thus, inactivation of AcrR by antimicrobials or cellular metabolites would derepress acrAB, marA, and soxS, as previously observed (9), to increase the production of the AcrAB-TolC pump and thus facilitate the efflux of these compounds.Inactivation of AcrR would also depress genes involved in metabolism and other functions such as the polyamine detoxification and efflux genes puuA and mdtJI (35) to contribute to maintaining homeostasis.Finally, inactivation of AcrR would also lead to increased expression of flhDC (Fig. 1 to 3), which in turn would increase the expression of flagellum biosynthesis and motility genes.Overexpression of these genes would ultimately increase motility (Fig. 4) and facilitate the escape of E. coli cells from antimicrobials and/or the accumulated cellular metabolites by moving to a different environment.
thus preventing the effects of having additional AcrR copies in the motility of this mutant (Fig. 4D).
Interestingly, a later study by Kim et al. (29) revealed that a similar upregulation of flagellum biosynthesis and motility genes, and a similar increase in swimming motil ity occurs, along with increased flagella production, when the acrAB transcriptional repressor acrR gene was deleted (ΔacrR::kan mutant).Such findings, together with the identification in the flhDC promoter of three putative 10-bp fragments with partial overlap with the known 24-bp AcrR binding site, whereas no predicted AcrR sites were identified in the fliAZ promoter, suggested that the effects of AcrR on motility might be mediated by flhDC (29).The FlhD 4 C 2 complex is the known master regulator that activates the expression of flagellum biosynthesis and motility genes, whereas fliA encodes for the σ 28 factor that controls the expression of a subset of motility genes and is downstream of the FlhD 4 C 2 master regulator in the regulatory cascade of flagellum biosynthesis and motility genes (37,38).However, it was still unknown whether AcrR directly regulates the expression of flhDC and/or other flagellum biosynthesis and motility genes in E. coli, or whether AcrR indirectly regulates flagellum genes and motility by regulating efflux and/or other target genes.To address this gap in knowledge, we have examined the role of AcrR as a direct regulator of flagellum biosynthesis and motility genes, as well as its role in the increased motility previously found in both ΔacrB and ΔacrR mutants.
First, we performed a genome-wide bioinformatics search of motility genes with promoters that contained predicted AcrR-binding sites using the known 24-bp AcrRbinding site in the acrAB promoter (34).We identified four flagellum biosynthesis and motility genes/operons-flhDC, fliE, fliDST, and fliLMNOPQR-whose promoter regions contained at least one full (24 bp) predicted AcrR-binding site (Fig. 1).We next tested for direct binding of purified AcrR to these four promoter regions by EMSA and found that AcrR only binds to the flhDC promoter region (Fig. 2A).This promoter region contains two predicted AcrR-binding sites with 10 and 11 mismatches, respectively, compared to the AcrR binding site in the acrAB promoter.In contrast, the fliE, fliDST, and fliLMNOPQR promoter regions each contain one predicted site with 11 mismatches plus one or two predicted sites with 12 mismatches (Fig. 1).Overall, our bioinformatics and EMSA findings indicate that the broad overexpression of about 50 flagellum biosynthesis genes and motility genes and increased motility previously found in the ΔacrR mutant (29) is the result of AcrR being a direct repressor of the motility master regulator flhDC operon.However, we do not discard that, in addition to its role as a flhDC regulator, AcrR might also directly regulate other motility genes/operons not identified here as potential AcrR targets.
To further investigate the role of AcrR in repressing motility by directly repressing flhDC expression, we combined bioinformatics, in vitro, in vivo gene expression and motility assays (Fig. 1 to 4).We first investigated the two potential full-size (24 bp) AcrR-binding sites identified in the flhDC promoter region.Site 1, located downstream of the transcriptional start site, has 10 mismatches compared to the known AcrR site in the acrAB promoter and maintains two 60% conserved inverted repeats.Site 2 has 11 mismatches compared to the known AcrR site in the acrAB promoter, including a 70% conserved left inverted repeat and a 30% poorly conserved right inverted repeat (Fig. 1A and 2B).Using EMSA to test for direct binding of purified AcrR protein to two different fragments of the flhDC promoter region, we found that besides AcrR binding to and shifting the full flhDC promoter region, AcrR was also able to bind to and shift the promoter fragment containing site 1 but not site 2 (Fig. 2A).On the contrary, no binding of AcrR was found for the promoter fragment containing only site 2 (Fig. 2A).Combined, these findings strongly support the role of AcrR as a direct regulator of the flhDC operon, and suggest that the more conserved site 1 might be the primary binding site for AcrR.
To further test the hypothesis that AcrR directly represses the flhDC operon, we performed in vivo gene expression and swimming motility assays comparing a parental strain, with the ΔacrR mutant, and the ΔacrR mutant complemented with the acrR gene cloned in an inducible plasmid.Deletion of acrR significantly increased the expression of the flhDC operon and motility at both 37°C and 30°C (Fig. 3, 4A and B), which is consistent with previous findings (29).In addition, we show for the first time that complementation of the ΔacrR mutant by overexpression of acrR from a plasmid reduces flhDC expression and motility at both temperatures down to the parental levels.These results are in agreement with our findings that the flhDC promoter region contains two predicted full AcrR-binding sites and that purified AcrR directly binds to this promoter region.Overall, these findings indicate that AcrR regulates motility by acting as a direct transcriptional repressor of the flhDC master regulator of flagellum biosynthesis and motility genes.
Finally, we examined the role of AcrR in the interplay between the AcrAB-TolC multidrug efflux pump and swimming motility in E. coli.Considering that AcrR represses acrAB transcription, earlier findings that deletion of acrB or acrR produced a similar overexpression of flagellum biosynthesis and motility genes and a similar increase in swimming motility (9, 29) might seem counterintuitive.However, several factors may contribute in explaining these findings.Both the AcrAB-TolC pump and flagella are powered by the proton motive force (PMF).PMF consumption by AcrAB-TolC has been suggested to impact its fitness contributions (39), and might also impact flagella function given that the speed of the flagellar motor varies with the PMF (40).Thus, the increase in swimming motility found in the ΔacrB mutant might be the result of an increase in PMF available for flagella rotation in this mutant.However, given that deletion of the acrR repressor leads to overexpression of the AcrAB-TolC pump and thus less available PMF, it would be expected that motility would decrease in the ΔacrR mutant, which is the opposite of what Kim et al. (29) and this study (Fig. 4A and B) have observed.Moreover, changes in PMF would not explain why the ΔacrB and ΔacrR mutants both show a similar strong overexpression of flagellum biosynthesis and motility genes.Instead, based on our results here showing that AcrR is a direct repressor of flhDC, we hypothesize that the similar flagellum gene expression and motility changes observed in both ΔacrB and ΔacrR mutants may occur because cellular metabolites that accumulate in the ΔacrB mutant can function as ligands that bind to and inactivate AcrR.Such ligand inactivation of AcrR in the ΔacrB mutant would derepress the expression of flhDC, which would explain why this mutant behaves as the ΔacrR mutant.Consistent with this hypothesis, we indeed found that overexpression of acrR from a plasmid reduced motility in the ΔacrB mutant down to parental levels at both 37°C and 30°C (Fig. 4C); whereas the addition of the AcrR ligand ethidium bromide prevented such effect by AcrR in the ΔacrB mutant (Fig. 4D).However, we do not discard that, in addition of flhDC upregulation caused by ligand-mediated inactivation of AcrR, changes in PMF may also contribute to the increased motility of the ΔacrB mutant.
Overall, these and prior findings portray a broader and more complex role of AcrR beyond being the local repressor of acrAB, and suggest that AcrR plays a major role in sensing stress caused by the accumulation of AcrAB-TolC substrates and coordinating efflux, metabolism and motility in response to such stress.This model is detailed in Fig. 5 and is based on three major premises: The first premise is that besides its known role in removing exogenous antimicrobials such as antibiotics, bile salts or ethidium bromide (1-7), the AcrAB-TolC pump also plays a role in effluxing cellular metabolites.These metabolites would be effluxed because they themselves, or their intermediates, end-products or by-products, may function as siderophores, signaling molecules, be toxic, and/or disrupt the normal metabolic flow of cells when they accumulate.A metabolic role of AcrAB-TolC is supported by its role in exporting enterobactin (18); the increased acrAB expression previously found in metabolic mutants (9,10); the altered expression of many metabolic genes found in ΔacrB and ΔacrR mutants (9,29); and the global changes in the intracellular and extracellular metabolite profile we found in the ΔacrB and ΔacrR mutants by untargeted metabolomics (17).Among other changes, these included a strong accumulation of amino acids (e.g., lysine) and tricarboxylic acid cycle intermediates in the ΔacrB mutant (17).
The second premise is that AcrR senses the accumulation of antimicrobials and cellular metabolites that are AcrAB-TolC substrates, or their metabolic precursors or derivatives, because some of these compounds can function as ligands that bind to and inactivate AcrR (Fig. 5).This premise is supported by the findings that AcrR was required for the increased acrAB expression found in the ΔacrB mutant and that this role was dependent on changes in its activity, not its expression (9); the finding that three exogenous compounds with antimicrobial activity and known to be effluxed by AcrAB-TolC (ethidium bromide, rhodamine 6G, and proflavine), bind in vitro to both AcrR and AcrB with a similar dissociation constant (34,41); and our recent findings that ethidium bromide and three cellular metabolites (polyamines such as the lysinederivative cadaverine) directly bind to and inactivate AcrR, thus derepressing the acrAB promoter (35), and presumably other AcrR-regulated genes.
The third premise is that AcrR is a direct regulator that, once it has sensed the accumulation of AcrAB-TolC substrates such as ethidium bromide, or cellular metabolites that may disrupt normal metabolic flow or be toxic, directly coordinates a broad cell response that involves efflux, metabolic and motility changes to cope with the accumu lation of these substrates and/or their precursors or derivatives (Fig. 5).This premise is supported by the well-known role of AcrR as a direct repressor of the acrAB operon (3-6, 28, 35); its role as a regulator SoxS and MarA expression (36), which are direct activators of the expression of acrAB and other genes involved in coping with antibiotics and other toxic molecules (4,5,42,43); the recent finding that AcrR directly regulates polyamine detoxification and efflux genes (35); and the findings reported here that AcrR is a direct repressor of the master motility regulator flhDC.
In conclusion, this manuscript combined bioinformatics, EMSA, gene expression, and motility assays to reveal that AcrR regulates motility in E. coli by acting as direct transcriptional repressor of the flhDC operon, which encodes for the master regulator of flagellum biosynthesis and motility genes.To our knowledge, this is the first report of AcrR directly regulating genes unrelated to efflux or detoxification, which contributes to potentially redefine AcrR as a central regulator of a global stress response regulon.The results reported here and prior findings support a model in which AcrR senses the accumulation of antimicrobials or cellular metabolites effluxed by the AcrAB-TolC multidrug efflux pump, and then co-regulates efflux, metabolism, and motility to synergistically contribute to maintaining homeostasis and adapting to environmental hazards.
Plasmid pBAD18-Kan-acrR was constructed as follows.First, the acrR gene from the parental E. coli BW25113 strain was amplified by PCR using the Thermo Fisher Scien tific (Waltham, MA, USA) DreamTaq polymerase as recommended by the manufacturer, a T m of 60°C, and primers acrRclF (5′-GATCGAGCTCAGGAGGCGAACATATGGCACGAAA; SacI site underlined, translation start codon of acrR in bold) and acrRclR (5′-GATCCTG CAGGTCAGATTCAGGGTTATTCG, PstI site underlined, complement sequence of the acrR translation stop codon in bold).After amplification, the PCR product was column-puri fied, digested with SacI and PstI (New England Biolabs, Ipswich, MA, USA), and ligated into pBAD18-Kan linearized with the same enzymes and gel-purified to generate plasmid pBAD18-Kan-acrR.Correct cloning of acrR was confirmed by plasmid isolation using of the Plasmid Miniprep kit from Zymo Research (Irvine, CA, USA) followed by Sanger sequencing at Laragen Inc (Culver City, CA, USA).Plasmids pBAD18-Kan or pBAD18-Kan-acrR were then electroporated into the parental strain E. coli BW25113 or its ΔacrR or ΔacrB mutant derivatives to generate the strains listed in Table 1 used for gene expression and motility complementation experiments.

Bioinformatics analysis to identify potential AcrR-binding sites in the flhDC promoter and other flagellum biosynthesis and motility genes
The sequence of the known 24-bp AcrR-binding site in the acrAB promoter (acrABp) [5′-TACATACATTTGTGAATGTATGTA (34)] and search tool of Colibri (http://genolist.pasteur.fr/Colibri/) were used to perform an initial whole-genome search of genes in E. coli that contain potential AcrR-binding sites and thus might be directly regulated by AcrR, focusing on flagellum biosynthesis and motility genes.The promoter regions of the four flagellum/motility genes/operons identified as candidate direct targets of AcrR (flhDCp, fliEp, fliMNOPQR, and fliDSTp) were then computationally analyzed more in-depth using MEGA X v.11 software (46) to identify all potential AcrR-binding sites, the major promoter features, and design primers to test for AcrR binding to these promoters in the EMSA experiments described below.

Gene expression experiments
The expression of the flhDC operon, measured as flhD mRNA, was determined by reverse transcription followed RT-qPCR as previously described (9,47) with the following modifications.First, strains BC7307 (parental + pBAD18-Kan), JM7312 (ΔacrR + pBAD18-Kan), and JM7314 (ΔacrR + pBAD18-Kan-acrR), were grown for 20 h in LB-Kan medium at 37°C with agitation at 200 rpm.Next, cultures were subcultured 1:1,000 in fresh LB-Kan and incubated until they reached an OD 600 nm of 0.1, before adding arabinose at a 0.4% final concentration to induce the expression of acrR from the pBAD18-Kan-acrR plasmid.Cultures were then incubated for approximately 1 h until they reached mid-exponential phase (OD 600 nm = 0.4).Next, cultures were treated with RNAprotect Bacteria reagent (Qiagen, Valencia, CA, USA) to stabilize their RNA, followed by RNA extraction using the Qiagen RNAeasy Mini Kit and the Qiagen on-column RNAse-free DNAse kit.The purity and concentration of the extracted RNAs were then determined using a Nanodrop 2000 (Thermo Fisher Scientific).
RNAs were then diluted to a concentration of 50 µg/mL in RNase-free water before reverse transcribing 200 ng of each RNA using the Thermo Fisher Scientific RT Invitrogen SuperScript IV First-Strand Synthesis System Kit and random hexamers, with RNAse H treatment, according to the manufacturer's specifications.RT minus reactions with water instead of reverse transcriptase were used as controls to confirm the lack of DNA contamination in the purified RNA samples.
After reverse transcription, flhD and gapA levels were quantified using 5 ng of total cDNA, 300 nM each of gene-specific forward and reverse primer, and the Applied Biosystems SYBR Green PowerUp Master Mix and QuanStudio 3 thermocycler (Thermo Fisher Scientific) as recommended by the manufacturer, using gene-specific standard plots for absolute quantification.qPCR gene-specific primers for flhD were: RTflhDF 5′-GCTATGTTTCGTCTCGGCATAA and RTflhDR 5′-CGGAAGTGACAAACCAGTTGA (T m = 63°C), and were designed using the PrimerQuest tool (qPCR with intercalating dyes parameters) from Integrated DNA Technologies (https://www.idtdna.com/Primerquest).Primers for gapA, which was used as our control gene because it is known not regulated by AcrR (29,35), are described elsewhere (48).All experiments were conducted using three to five biological replicates (cultures/RNA extractions), each with two (RT reactions) and three (qPCR reactions) technical replicates.

Motility experiments
Swimming motility experiments were performed as previously described by Ruiz and Levy (9) with the following modifications.First, strains BC7307 (parental + pBAD18-Kan), JM7312 (ΔacrR + pBAD18-Kan), JM7314 (ΔacrR + pBAD18-Kan-acrR), JM7317 (ΔacrB + pBAD18-Kan), and JM7316 (ΔacrB + pBAD18-Kan-acrR) were grown for 20 h in LB-Kan agar plates at 37°C.Next, a single representative colony of each strain was stabbed using a sterile toothpick onto semi-solid LB-Kan plates containing 0.35% agar (which made the semi-solid medium in plates incubated at 37°C less fragile than using 0.3%) and 0.4% arabinose to induce the expression of acrR from the pBAD18-Kan-acrR plasmid.Inoculated plates were incubated for 18 h (or 24 h in motility assays performed in the presence of ethidium bromide) at 37°C, or 24 h at 30°C, before measuring the diameter of the zone of migration in mm.All experiments were conducted using five to six biological replicates.

Statistical analysis
For gene expression and swimming motility experiments, statistically significant differences between strains or treatments were determined by t test (two independ ent samples with equal variance, two-tailed distribution) using Microsoft Excel 2021 software.

FIG 1 4 FIG 2
FIG 1 Predicted AcrR-binding sites in the four flagellum biosynthesis and motility genes/operons identified as direct target candidates of the AcrR repressor.(A) Sequence of the σ 70 -dependent promoter region of the E. coli master regulator of flagellar biosynthesis and motility flhDC operon (flhDCp).The −35 and −10 sequences, the transcriptional start site (TSS), and the flhD translational start site, are highlighted in blue lettering.The two predicted AcrR-binding sites in the flhDC promoter identified in this study, which had 10 and 11 mismatches, respectively, compared to the 24 bp AcrR-binding site in the acrAB promoter [5′-TACATACATTTGTGAATGTATGTA (34)], are indicated in red lettering, with mismatches shown as lowercase letters.The three partial (10 bp) AcrR-binding sites identified by Kim et al. (29) are indicated as underlined lettering.The forward primers and reverse complementary (rc) sequences for the reverse primers used to amplify the flhDC full promoter and promoter fragments used in EMSA (Fig. 2A) are indicated as green boxes.(B−D) Sequences of the fliE, fliLMNOPQR, and fliDST promoter regions (fliEp, fliLMNOPQRp, and fliDSTp), respectively.Predicted AcrR-binding sites, promoter elements, and primers used for EMSA are labeled in the same manner as described above for flhDC.Predicted AcrR-binding sites labeled as (rc) denote that the sequence and mismatches shown correspond to the reverse complementary sequence (opposite strand) of the AcrR-binding site (5′-TACATACATTCACAAATG TATGTA).fliEp contained three predicted AcrR-binding sites with 11, 12, and 12 mismatches, respectively, compared to the AcrR-binding site in the acrAB promoter.fliMNOPQRp and fliDSTp each contained two predicted AcrR-binding sites with 12 and 11 mismatches, for sites 1 and 2 in each promoter, respectively.

FIG 3
FIG3 AcrR regulates the expression of the flhDC operon in vivo.The expression of flhD and the control gene gapA were measured by RT-qPCR using RNA extracted from mid-exponential phase cultures in LB with 50 µg/mL kanamycin that were induced for 1 h with 0.4% arabinose.Experiments were performed using three to five biological replicates each with two RT and three qPCR technical replicates.The data are presented as average ±SEM (n = 3-5) and are shown as the n-fold change in the expression of flhD or gapA of each strain normalized to that of the parental strain.No statistically significant differences were found between any of the strains for the control gene gapA (statistics not shown in the figure for clarity).For flhD, statistically significant differences between strains are indicated as ** (P < 0.001) or * (P < 0.05); and lack of statistically significant differences is indicated as N.S.The expression of flhD was significantly increased in the ΔacrR mutant compared to the parental strain, whereas complementation of this mutant using the pBAD18-Kan-acrR plasmid reduced flhD expression to a level similar to that in the parental.

FIG 4
FIG4 AcrR represses swimming motility in E. coli at 37°C and 30°C.(A-C) Motility was measured as the diameter of the zone of migration in mm using LB 0.35% agar plates supplemented with 0.4% arabinose and 50 µg/mL kanamycin after incubation at 37°C for 18 h (blue bars) or 30°C for 24 h (red bars).All experiments were performed using four to six biological replicates.The data are presented as average ±SEM (n = 4-6) and are shown as the n-fold change in the diameter of the zone of migration of each strain normalized to that of the parental strain, which was 5.3 ± 0.5 mm at 37°C for 18 h, and 8.4 ± 0.4 mm at 30°C for 24 h.Statistically significant differences between the ΔacrR or ΔacrB mutants compared to the parental strain (all containing the empty pBAD18-Kan plasmid), or between each mutant containing the pBAD18-Kan plasmid and the same mutant with the pBAD18-Kan-acrR plasmid, are indicated as ** (P < 0.001) or * (P < 0.05).No statistically significant differences were found between the parental-pBAD18-Kan and the ΔacrR or ΔacrB mutant strains containing the pBAD18-Kan-acrR plasmid (statistics not shown in the figure for clarity).(A) The ΔacrR mutant showed significant increased motility at both temperatures compared to the parental strain.Complementation of the ΔacrR mutation using the pBAD18-Kan-acrR plasmid restored the motility of this mutant to parental levels.(B) Representative pictures of the swimming motility results obtained for the strains tested in panels (A and C).(C) The ΔacrB mutant showed significant increased motility at both temperatures compared to the parental.Overexpression of the acrR gene using the pBAD18-Kan-acrR plasmid restored the motility of this mutant to the parental level at both 37°C and 30°C.(D) Addition of increasing concentrations of the AcrR ligand ethidium bromide (EtBr) prevents the swimming motility reduction caused by overexpressing acrR in the ΔacrB mutant.Statistically significant differences between the untreated (0 µM) and ethidium bromide treatments are indicated as ** (P < 0.01) or * (P < 0.05).

TABLE 1
Bacterial strains and plasmids used in this study