Role of the MDR Efflux Pump AcrAB in Epithelial Cell Invasion by Shigella flexneri

The tripartite complex AcrAB-TolC is the major RND pump in Escherichia coli and other Enterobacteriaceae, including Shigella, the etiological agent of bacillary dysentery. In addition to conferring resistance to many classes of antibiotics, AcrAB plays a role in the pathogenesis and virulence of several bacterial pathogens. Here, we report data demonstrating that AcrAB specifically contributes to Shigella flexneri invasion of epithelial cells. We found that deletion of both acrA and acrB genes causes reduced survival of S. flexneri M90T strain within Caco-2 epithelial cells and prevents cell-to-cell spread of the bacteria. Infections with single deletion mutant strains indicate that both AcrA and AcrB favor the viability of the intracellular bacteria. Finally, we were able to further confirm the requirement of the AcrB transporter activity for intraepithelial survival by using a specific EP inhibitor. Overall, the data from the present study expand the role of the AcrAB pump to an important human pathogen, such as Shigella, and add insights into the mechanism governing the Shigella infection process.


Introduction
Multidrug-resistance (MDR) efflux pumps (EPs) are transmembrane transporters capable of extruding a wide range of toxic compounds, including antibiotics [1,2]. Genes encoding MDR EPs are generally located on the chromosome and are highly conserved across species. Currently, genes encoding MDR EPs are considered not only the simple result of recent evolution favored by the intense use of antibiotics but ancient genes encoding protein complexes deeply involved in the physiology of the bacterial cell [3,4]. This is further underlined by the fact that, in addition to antibiotics, MDR EPs are able to expel a large range of different molecules from the cell, including bacterial metabolites, siderophores, heavy metals, quorum-sensing molecules and virulence factors [2,5,6]. This capability makes MDR EPs important players in the maintenance of cellular homeostasis, in the interplay between bacteria, in bacteria-host cell interactions, in biofilm formation and in the virulence [7][8][9][10][11].
Bacterial MDR EPs are commonly present as single-component transporters embedded in the inner membrane. They can also form tripartite complexes that cross the double membranes in Gram-negative bacteria [12]. MDR EPs mainly belong to the following families: the ATP binding cassette (ABC) superfamily, the major facilitator superfamily (MFS), the multidrug and toxic compound extrusion (MATE) family, the small multidrug resistance (SMR) family and the resistance nodulation division (RND) family [2].
Members of the RND family play a critical role in the emergence of multidrug resistance in many Gram-negative bacterial pathogens [1,13,14]. Six MDR EPs of the RND family are usually present in E. coli K12, namely AcrAB, AcrD, AcrEF, MdtABC, MdtEF, M90T, gene disruption was performed using the one-step inactivation method of chromosomal genes [35]. In particular, the kanamycin resistance gene was amplified via PCR using pKD4 as template and the oligo pairs ShAF/ShBR for the acrAB deletion, ShAF/ShAR for the acrA deletion and ShBF/ShBR for the acrB deletion (Table S1). The resulting PCR products were used to transform M90T recipient strain harboring the pKD46 plasmid expressing the Red recombinase. In order to maintain unaltered transcription and translation of the downstream acrB gene, the Km resistance cassette in the mutant M90T ∆acrA was eliminated through the flippase encoded by the pCP20 plasmid using flippase/flippase recognition target (Flp/FRT) recombination [35]. To construct the M90T ∆acrD mutant, P1 phage isolated from the donor strain JW2454-1 ∆acrD of the Keio Collection [36] was used to infect M90T wild-type strain and transduction of the mutated locus was verified using PCR. All plasmids used in this study are listed in Table 1. Plasmid pGSfacrB was obtained by cloning acrB coding sequence downstream the tac promoter of pGIP7 plasmid and then used as template for site directed mutagenesis. In particular, as previously described [28], the D408A substitution in AcrB encoded by the pSfacrB D408A plasmid was obtained via a Site-Directed Mutagenesis System (GENEART ® Invitrogen-Thermo Fisher Scientific, Waltham, MA, USA), using the oligo pair pSfacrB D408A F/pSfacrB D408A R (Table S1), that allowed to replace the A at position 1223 (from the ATG of acrB) with C, generating a codon change from GAC to GCC (D408A). The presence of the correct mutation on pSfacrB D408A plasmid has been verified via DNA sequencing (Biofab, Rome, Italy). The obtained plasmid was then transformed in the M90T ∆acrB strain. Trypticase soy agar to monitor the expression of the virulence phenotype prior to infection assays. Antibiotics were used at the following concentrations: ampicillin 50 µg/mL; cloramphenicol 25 µg/mL; kanamycin 30 µg/mL; streptomycin 10 µg/mL. Growth kinetics of M90T and its derivatives were measured using a CLARIOstar plate reader (BMG LABTECH, Offenburg, Germany).

Antimicrobial Susceptibility
In order to determine MIC of erythromycin, tetracycline and streptomycin, M90T wild-type strain and its derivatives M90T ∆acrAB, M90T ∆acrA, M90T ∆acrB pSfacrB D408A , and M90T ∆acrD, were inoculated into LB and grown at 37 • C by shaking for 16 h. Cultures were then diluted to OD 600 0.02 in LB and 100 µL aliquots were transferred to 96-well plate, each well containing 100 µL of 2-fold serial dilutions of the compounds to be tested (erythromycin 0.001 mg/mL to 2 mg/mL, tetracycline 0.0078 µg/mL to 16 µg/mL, and streptomycin 0.0625 µg/mL to 128 µg/mL). After 16 h incubation at 37 • C, the lowest concentration of antibiotic inhibiting bacterial growth was estimated, as previously described [40]. At least three biological replicates were performed. Bile salts (B8756, Sigma-Aldrich, St. Louis, MO, USA), consisting of an approximate 1:1 mixture of cholate and deoxycholate, were used at 0.25% (w/v).

Live and Death Assay and Viable Bacterial Count
To collect intracellular bacteria at each infection time point, infected cells were washed twice with 1× PBS and lysed by adding 1% Triton X-100 (Sigma-Aldrich, St. Louis, MO, USA)) for 5 min. To evaluate the percentage of intracellular dead bacteria, recovered bacteria were pelleted, washed with 1× PBS and suspended in 1× PBS containing 10 µg/mL DAPI to stain the entire population and 15 µM Propidium Iodide (PI) to label dead bacteria. After 20 min of incubation at room temperature in the dark, bacteria were washed with 1× PBS and resuspended in 5 µL 1× PBS containing 50% glycerol. The entire sample was spotted on the glass slide and overlaid with the coverslip for immediate observation and counting under the fluorescence microscopy. Samples were examined using a Leica DMRE fluorescence microscope equipped with a 100× lens.
To define the number of viable intracellular bacteria, infection of epithelial cells was carried out as describe above. At the indicated time points, the cell lysate containing intracellular bacteria was collected, washed and resuspended in 1× PBS. Serial dilutions of the bacterial suspensions were plated on LB agar plates to calculate the CFU/mL.

Plaque Assay
To determine the ability of the M90T wild-type strain and its derivatives to spread intercellularly a plaque assay was performed [42]. The 5 × 10 6 Caco-2 cells were seeded in 60 mm plates in DF10. Once reached confluency (usually after 24 h), cells were serumstarved overnight in DF0.5. DF0.5 medium was replaced with DMEM containing only L-glutamine 2 h before the infection. The infection was carried out at MOI 0.001, plates were centrifuged at 750 g for 15 min and subsequently incubated at 37 • C in a 5% CO 2 atmosphere for 45 min. Extracellular bacteria were then removed by washing the plates three times with 1× PBS. An agarose overlay containing DMEM, gentamicin (100 µg/mL), FBS (5%) and agarose (0.5%) was added to each plate. Infected cells were incubated for 72 h, then the agarose overlay was carefully removed, and cells were ethanol fixed and Giemsa stained.

LDH Cytotoxicity Assay
CyQUANT™ LDH Cytotoxicity Assay kit (Invitrogen, Thermo Fisher Scientific, Waltham, MA, USA) was used to measure cytotoxicity and verify that the NMP inhibitor does not affect cell viability. Caco-2 cells were plated in DMEM with or without NMP (100 µg/mL) in 35 mm plates (Falcon, Thermo Fisher Scientific, Waltham, MA, USA) at 0.8 × 10 6 cells/well. LDH activity was determined after 2 and 4 h of treatment by measuring absorbance at 490 nm and 680 nm with a CLARIOstar plate reader (BMG LABTECH, Offenburg, Germany). Percentage of cytotoxicity was calculated according to the manufacturer's instructions.

Statistical Analysis
Statistically significant differences in viable bacterial counts and live and death assays were identified using a two-tailed student's t-test.

Lack of AcrAB Impairs the S. flexneri M90T Infection of Epithelial Cells
AcrAB is one of the most important MDR EP of E. coli and is expressed at high levels even under laboratory conditions [24]. Despite the intense gene decay of the housekeeping genes [20], the acrA and acrB encoding genes are conserved in all Shigella spp. Additionally, they share a high homology with those of E. coli [4]. As for E. coli, acrAB operon is also highly expressed in S. flexneri M90T. During the infection of U937-derived macrophages and Caco-2 epithelial cells the mRNA level of acrA gene, measured to monitor the behavior of the entire acrAB operon, further increases very early upon Shigella entry into host cells, the transcriptional response being more evident in M90T infecting Caco-2 cells [11]. Based on these notions and on the role this MDR EP plays in other pathogenic bacteria, we wondered whether AcrAB can be part of the mechanisms ensuring a successful Shigella infection process. To this end, an M90T derivative lacking the entire acrAB operon was generated via the one-step method of gene inactivation [35] (Table 1). Lack of AcrAB does not have an impact on M90T growth properties in laboratory conditions ( Figure S1), while, as expected, it causes a higher susceptibility to drugs as compared to the M90T wild-type strain ( Figure 1A). Silencing of acrAB also affects M90T resistance to bile salts at physiological concentration (0.25% w/v) ( Figure 1B), which is in agreement with previous observations [34].
while, as expected, it causes a higher susceptibility to drugs as compared to the M90T wild-type strain ( Figure 1A). Silencing of acrAB also affects M90T resistance to bile salts at physiological concentration (0.25% w/v) ( Figure 1B), which is in agreement with previous observations [34]. To test the response to the host cell environment in the absence of AcrAB, both THP-1-derived macrophages and Caco-2 epithelial cells were infected with the wild-type M90T strain and the ΔacrAB derivative. We first assessed Shigella intracellular survival via DAPI/PI double staining of bacteria harvested from infected cells at different time points post-infection (p.i.). Figure 2A shows that lack of AcrAB EP poorly affects the viability of M90T when inside macrophage environment as, by and large, the proportion of dead bacteria detected at each time point was very similar either for the wild-type or mutant strain. Conversely, the absence of AcrAB is detrimental to M90T infecting epithelial cells. Indeed, as shown in Figure 2B, while a certain fraction of dead wild-type bacteria was observed to constantly increase throughout the infection period analyzed, a high amount of PI-positive M90T ΔacrAB was found already at a very early stage of infection (T0) of Caco-2 cells. This proportion further increases at two hours p.i., keeping higher than that of the parental strain till the last time point analyzed.  To test the response to the host cell environment in the absence of AcrAB, both THP-1derived macrophages and Caco-2 epithelial cells were infected with the wild-type M90T strain and the ∆acrAB derivative. We first assessed Shigella intracellular survival via DAPI/PI double staining of bacteria harvested from infected cells at different time points post-infection (p.i.). Figure 2A shows that lack of AcrAB EP poorly affects the viability of M90T when inside macrophage environment as, by and large, the proportion of dead bacteria detected at each time point was very similar either for the wild-type or mutant strain. Conversely, the absence of AcrAB is detrimental to M90T infecting epithelial cells. Indeed, as shown in Figure 2B, while a certain fraction of dead wild-type bacteria was observed to constantly increase throughout the infection period analyzed, a high amount of PI-positive M90T ∆acrAB was found already at a very early stage of infection (T0) of Caco-2 cells. This proportion further increases at two hours p.i., keeping higher than that of the parental strain till the last time point analyzed.
while, as expected, it causes a higher susceptibility to drugs as compared to the M90T wild-type strain ( Figure 1A). Silencing of acrAB also affects M90T resistance to bile salts at physiological concentration (0.25% w/v) ( Figure 1B), which is in agreement with previous observations [34]. To test the response to the host cell environment in the absence of AcrAB, both THP-1-derived macrophages and Caco-2 epithelial cells were infected with the wild-type M90T strain and the ΔacrAB derivative. We first assessed Shigella intracellular survival via DAPI/PI double staining of bacteria harvested from infected cells at different time points post-infection (p.i.). Figure 2A shows that lack of AcrAB EP poorly affects the viability of M90T when inside macrophage environment as, by and large, the proportion of dead bacteria detected at each time point was very similar either for the wild-type or mutant strain. Conversely, the absence of AcrAB is detrimental to M90T infecting epithelial cells. Indeed, as shown in Figure 2B, while a certain fraction of dead wild-type bacteria was observed to constantly increase throughout the infection period analyzed, a high amount of PI-positive M90T ΔacrAB was found already at a very early stage of infection (T0) of Caco-2 cells. This proportion further increases at two hours p.i., keeping higher than that of the parental strain till the last time point analyzed.  This observation prompted us to explore in more details the behavior of ∆acrAB derivative during the infection of epithelial cells. We determined the viability of bacteria recovered from infected epithelial cells using the CFU assay. We found that, as expected, the number of colonies formed by intracellular mutant bacteria harvested at the various time points is much lower compared to M90T. However, the CFU/mL produced by the ∆acrAB derivative increases over the infection period following a kinetics very similar to that of the parental strain ( Figure 3A) suggesting that, although much more ∆acrAB cells succumb to the host attacks, the surviving intracellular bacteria do not lose the ability to multiply. Full invasion process relies on Shigella capability to disseminate to neighboring cells in an epithelial layer without further extracellular step [31]. The cell-to-cell spreading capacity of viable ∆acrAB was investigated using plaque assay. Figure 3B shows that lack of AcrAB dramatically impairs the spreading of M90T to adjacent Caco-2 cells as a very low number of plaques formed by the acrAB deletion mutant was observed, representing less than 3% of the parental strain. This observation prompted us to explore in more details the behavior of ΔacrAB derivative during the infection of epithelial cells. We determined the viability of bacteria recovered from infected epithelial cells using the CFU assay. We found that, as expected, the number of colonies formed by intracellular mutant bacteria harvested at the various time points is much lower compared to M90T. However, the CFU/mL produced by the ΔacrAB derivative increases over the infection period following a kinetics very similar to that of the parental strain ( Figure 3A) suggesting that, although much more ΔacrAB cells succumb to the host a acks, the surviving intracellular bacteria do not lose the ability to multiply. Full invasion process relies on Shigella capability to disseminate to neighboring cells in an epithelial layer without further extracellular step [31]. The cell-to-cell spreading capacity of viable ΔacrAB was investigated using plaque assay. Figure 3B shows that lack of AcrAB dramatically impairs the spreading of M90T to adjacent Caco-2 cells as a very low number of plaques formed by the acrAB deletion mutant was observed, representing less than 3% of the parental strain.

Both AcrB and AcrA Components Contribute to the Survival of S. Flexneri M90T Inside Epithelial Cells
It is well acknowledged that each of the components of the AcrAB-TolC MDR EP has specific activities ensuring the proper exporter function, as the inner membrane transport protein AcrB is critical for the substrate specificity and the periplasmic adapter protein AcrA is important for connecting AcrB to the TolC channel [12]. Based on the results obtained with the M90T ΔacrAB strain we asked whether both AcrB and AcrA are required for the survival of Shigella inside epithelial cells. To this end, single acrA and acrB deletion mutants were generated via site-directed mutagenesis (Table 1). Since AcrB is a very abundant inner membrane protein, the M90T ΔacrB strain was complemented with an AcrB protein lacking the efflux activity (AcrB D408A) (Table 1), in order to generate a defective strain for AcrB-associated transporter function without altering the membrane protein composition. Moreover, based on the knowledge that the AcrA periplasmic adapter is also used by AcrD [30], the other RND EP conserved in M90T, to form a functional transporter complex, a ΔacrD strain has been also constructed ( Table 1). M90T ΔacrA, M90T ΔacrB pSfacrBD408A and M90T ΔacrD displayed growth properties overlapping those of M90T wild-type strain ( Figure S1). Both M90T ΔacrA andM90T ΔacrB pSfacrBD408A were much

Both AcrB and AcrA Components Contribute to the Survival of S. Flexneri M90T Inside Epithelial Cells
It is well acknowledged that each of the components of the AcrAB-TolC MDR EP has specific activities ensuring the proper exporter function, as the inner membrane transport protein AcrB is critical for the substrate specificity and the periplasmic adapter protein AcrA is important for connecting AcrB to the TolC channel [12]. Based on the results obtained with the M90T ∆acrAB strain we asked whether both AcrB and AcrA are required for the survival of Shigella inside epithelial cells. To this end, single acrA and acrB deletion mutants were generated via site-directed mutagenesis (Table 1). Since AcrB is a very abundant inner membrane protein, the M90T ∆acrB strain was complemented with an AcrB protein lacking the efflux activity (AcrB D408A) (Table 1), in order to generate a defective strain for AcrB-associated transporter function without altering the membrane protein composition. Moreover, based on the knowledge that the AcrA periplasmic adapter is also used by AcrD [30], the other RND EP conserved in M90T, to form a functional transporter complex, a ∆acrD strain has been also constructed ( Table 1). M90T ∆acrA, M90T ∆acrB pSfacrB D408A and M90T ∆acrD displayed growth properties overlapping those of M90T wild-type strain ( Figure S1). Both M90T ∆acrA andM90T ∆acrB pSfacrB D408A were much more susceptible than M90T to the antibiotics tested, while lack of AcrD did not change the MIC profile of M90T ( Figure 1A). The presence of bile salts in the LB medium affected the growth curve of all the three M90T derivatives, with milder effects on the multiplication rate of M90T ∆acrB pSfacrB D408A and M90T ∆acrD ( Figure 1B). Parallel Caco-2 epithelial cell infections were carried out with the three mutant strains along with wild-type M90T and the viability of intracellular bacteria was assessed at different time points p.i. using DAPI/PI double staining. As shown in Figure 4A, the percentage of M90T ∆acrB pSfacrB D408A PI-positive bacteria was very similar to the wild-type at very early stage of the infection (T0), while it significantly increased in the following hours of infection. The defective phenotype conferred by the absence of AcrA was much more severe, chiefly soon after bacterial entry into the host cells (T0). Indeed, at this time point, we found that more than 20% of intracellular M90T ∆acrA bacteria were not viable compared to 2% observed for the intracellular wild-type strain. The proportion of dead M90T ∆acrA bacteria further increases as the infection proceeds. Data obtained for the M90T strain lacking AcrD indicate that this transporter is dispensable for Shigella survival inside the epithelial cells. Indeed, the proportion of PI-positive M90T ∆acrD bacteria was very similar to that of the parental strain at each time point analyzed ( Figure 4A). Overall, the data obtained using deletion mutants in single components of the two RND EPs point out the main role of the AcrAB EP during the Shigella infection of the epithelial cells.
Biomolecules 2023, 13, x FOR PEER REVIEW 8 of 13 more susceptible than M90T to the antibiotics tested, while lack of AcrD did not change the MIC profile of M90T ( Figure 1A). The presence of bile salts in the LB medium affected the growth curve of all the three M90T derivatives, with milder effects on the multiplication rate of M90T ΔacrB pSfacrBD408A and M90T ΔacrD ( Figure 1B). Parallel Caco-2 epithelial cell infections were carried out with the three mutant strains along with wild-type M90T and the viability of intracellular bacteria was assessed at different time points p.i. using DAPI/PI double staining. As shown in Figure 4A, the percentage of M90T ΔacrB pSfacrBD408A PI-positive bacteria was very similar to the wild-type at very early stage of the infection (T0), while it significantly increased in the following hours of infection. The defective phenotype conferred by the absence of AcrA was much more severe, chiefly soon after bacterial entry into the host cells (T0). Indeed, at this time point, we found that more than 20% of intracellular M90T ΔacrA bacteria were not viable compared to 2% observed for the intracellular wild-type strain. The proportion of dead M90T ΔacrA bacteria further increases as the infection proceeds. Data obtained for the M90T strain lacking AcrD indicate that this transporter is dispensable for Shigella survival inside the epithelial cells. Indeed, the proportion of PI-positive M90T ΔacrD bacteria was very similar to that of the parental strain at each time point analyzed ( Figure 4A). Overall, the data obtained using deletion mutants in single components of the two RND EPs point out the main role of the AcrAB EP during the Shigella infection of the epithelial cells. AcrAB inhibitors exist, mainly targeting the transporter function of the AcrB protein [1]. Considering the results we obtained, it is reasonable to regard such compounds as potential anti-Shigella virulence. Indeed, by inhibiting AcrB-mediated extrusion, survival, and thus dissemination of the pathogen should be significantly impaired. To test this hypothesis, we carried out infection experiments in the presence of 1-(1-naphthyl-methyl)piperazine (NMP), a well characterized and active AcrB inhibitor [1,43]. This inhibitor belongs to the family of arylpiperazines and it has been suggested to block the normal substrate extrusion by acting as an EP substrate [1]. NMP neither affects M90T viability/growth properties nor has toxic effects on Caco-2 cells, the la er assessed by measuring LDH activity (Lactate dehydrogenase) present in the cell supernatant (data not shown). Caco-2 cells were infected with the M90T wild-type strain in the presence or absence of AcrAB inhibitors exist, mainly targeting the transporter function of the AcrB protein [1]. Considering the results we obtained, it is reasonable to regard such compounds as potential anti-Shigella virulence. Indeed, by inhibiting AcrB-mediated extrusion, survival, and thus dissemination of the pathogen should be significantly impaired. To test this hypothesis, we carried out infection experiments in the presence of 1-(1-naphthyl-methyl)-piperazine (NMP), a well characterized and active AcrB inhibitor [1,43]. This inhibitor belongs to the family of arylpiperazines and it has been suggested to block the normal substrate extrusion by acting as an EP substrate [1]. NMP neither affects M90T viability/growth properties nor has toxic effects on Caco-2 cells, the latter assessed by measuring LDH activity (Lactate dehydrogenase) present in the cell supernatant (data not shown). Caco-2 cells were infected with the M90T wild-type strain in the presence or absence of NMP. The drug (100 µg/mL) was added to the bacteria in DMEM medium, just before exposing them to the epithelial cells and the effect of the treatment on viability was monitored soon after bacterial entry into the host cells (T0) and after two hours of infection. As shown in Figure 4B, drug treatment mildly affects the survival of intracellular bacteria at a very early stage of infection, while it causes a significant decrease in intracellular M90T viability at two hours p.i. Interestingly, the M90T phenotype induced by the AcrB inhibitor overlaps with that observed with the intracellular ∆acrB pacrB D408A derivative at the same time points, highlighting, on one hand, the inhibitor specificity and confirming, on the other hand, the important involvement of AcrB transporter in assisting Shigella during the infection of epithelial cells.

Discussion
In this work, we report evidence demonstrating that the MDR EP AcrAB contributes to the virulence of S. flexneri by favoring the bacterial survival within the epithelial cells. Analysis of mutants lacking AcrA or AcrB clearly indicates that both components are required for the bacterial invasion of and survival within the Caco-2 epithelial cells. Moreover, the loss of AcrAB hampers the ability of S. flexneri to spread within the epithelial monolayer.
The hallmark of Shigella pathogenicity is its ability to penetrate the colonic epithelium, escape macrophages by inducing cell death, and, subsequently, invade colonocytes from the basolateral side and propagate infection through cell-to-cell spread, eventually causing the destruction of the intestinal barrier function [31]. The whole infection process is mostly built on the action of several factors encoded by the virulence plasmid that Shigella acquired during the evolution [18,19]. On the road of evolution from commensal E. coli ancestor, Shigella has also undergone a patho-adaptation process following an intense gene decay involving the loss of detrimental or unnecessary functions for intracellular lifestyle [18,[20][21][22][23]. Gene silencing also affected some MDR EPs belonging to the RND family, as, of those present in E. coli K12 MG1655, only the acrAB and acrD genes were conserved as fully functional in the S. flexneri M90T genome [11]. Among these two, AcrAB undoubtedly stands out for its important contribution to the pathogenicity of several bacteria [27,[44][45][46][47]. More recently, the pivotal role of AcrAB in supporting survival of AIEC LF82 within the macrophages has been also demonstrated [28], further expanding the range of bacterial pathogens exploiting AcrAB function to facilitate the infection process. Regarding Shigella, the involvement of AcrAB in the invasion of host cells has never been explored. The data obtained from the present study clearly indicate that the AcrAB function protects Shigella during epithelial cell infection, mainly at the early stages. Indeed, lack of the entire acrAB operon results in a larger proportion of intracellular bacteria undergoing cell death, compared to the wild-type strain, soon after cell entry, picking at two hours p.i. We also show that surviving intracellular M90T ∆acrAB bacteria appear perfectly viable and multiply with a kinetics similar to that of the parental M90T. However, when the ∆acrAB derivative was challenged for the ability to disseminate within the Caco-2 monolayer, most of the viable bacteria failed to form visible plaques, at least in the time taken by the wild-type M90T. Indeed, as can be inferred by comparing the CFU/mL with the number of plaques (Figure 3), the viable intracellular M90T ∆acrAB bacteria forming colonies on LB agar are about one-seventh of the parental strain, while those forming plaques are less than the thirtieth part of the wild-type M90T. This observation suggests that the absence of AcrAB, in addition to causing accumulation of deadly toxic metabolites, might also interfere with the expression of functions involved in the actin-based intra/intercellular movement of Shigella. In this regard, it is interesting to consider transcriptomic studies carried out in Salmonella demonstrating that, in the acrAB-defective background several genes essential for invasion, encoded by the Salmonella pathogenicity island (SPI), are downregulated [27,48]. Moreover, it has been recently suggested that the enrichment of certain natural AcrB substrates in the exometabolome might be important for Salmonella virulence, some of these being involved in the regulation of virulence factor expression [49]. Assessing whether modulation of virulence factors also occurs in Shigella AcrAB-defective mutant will deserve a much deeper investigation in the future.
It is well known that AcrA and AcrB provide the EP with specific functions contributing to its full activity. The evidence presented here indicates that the inactivation of AcrA alone leads to a marked decrease in bacterial viability in epithelial cells comparable to that observed with the M90T double mutant. As discussed above, among the MDR EPs belonging to the RND family, S. flexneri M90T has also kept functional AcrD transporter, which is orphaned of its own periplasmic protein and shares AcrA with the AcrB transporter [11,30]. The evaluation of the impact of each transporter indicates that the activity associated with AcrB is strongly required by Shigella to better survive within epithelial cells, while AcrD appears to be dispensable, at least for the intracellular viability. The use of a M90T ∆acrB derivative complemented with the unfunctional AcrB D408A protein in the infection experiments of epithelial cells allowed us to avoid profound alteration of the inner membrane due to the lack of an abundant constituent, such as AcrB and to specifically pinpoint the involvement of the efflux activity associated with AcrB.
To date, this is the second study emphasizing the role that EPs play in supporting Shigella to overcome the hostile environment met inside host cells. Depending on the host cell context the activity of different EPs is exploited. It has been recently demonstrated that the MFS EmrKY is specifically activated in macrophages in response to the cytoplasmic acidic pH induced by Shigella infection, improving the bacterial fitness [11]. Here, we show that the highly expressed AcrAB works to promote invasion of and spread within epithelial cells and that both AcrA and AcrB components are required. Overall, these reports point out EPs as important components of the Shigella pathogenicity mechanism, and, together with the data we obtained with the AcrB specific inhibitor, lead to envisage the efflux activity of specific transporters and/or the expression of specific EPs as valuable targets to attenuate Shigella virulence.