Indispensable role of AcrEF in modulating Salmonella virulence and disrupting host tight junctions to facilitate paracellular entry and invasion

Abstract

AbstractSection Background

See More

The Salmonella stress response regulator CpxR controls the AcrEF drug efflux pump, which is crucial for key virulence traits such as invasiveness and disruption of the host paracellular pathway. Understanding the role of acrEF and its regulation by CpxR can provide new insights into Salmonella pathogenesis and future therapeutic development.

AbstractSection Methods

We performed differential gene expression analysis using Salmonella enterica serovar Typhimurium (ST) wild-type and cpxR mutants to identify virulence-associated genes affected by the stress response system. The candidate genes acrE and acrF were deleted and used to evaluate virulence-related phenotypes, including adhesion, invasion, and epithelial barrier disruption both in vitro and in vivo, in wild-type and mutant strains.

AbstractSection Results

Deletion of acrEF in both ST wild-type and cpxR mutants significantly reduced Salmonella adhesion, invasion of multiple epithelial cell lines, and expression of virulence genes. It also led to enhanced tight junction integrity in epithelial cells, potentially via upregulation of genes like ZO-1, suggesting a novel invasion mechanism. The loss of acrEF function impaired the bacteria’s ability to breach host cell tight junctions, which directly correlated with attenuated invasion and survival in vivo. These effects were similarly observed in both wild-type and cpxR mutants, indicating a central role for acrEF in Salmonella virulence.

AbstractSection Conclusion

The AcrEF efflux pump plays a key role in regulating Salmonella virulence, particularly in modulating tight junction disruption and epithelial invasion. Although CpxR may regulate acrEF expression, the loss of acrEF function independently results in significant attenuation of virulence. These findings reveal a critical pathway of Salmonella epithelial invasion mediated by the AcrEF system and regulated, in part, by the CpxR stress response regulator.

AbstractSection Clinical trial number

Not applicable.

AbstractSection Graphical Abstract

Background

Salmonella species are notable pathogens that cause acute and chronic infections in various animals, including humans [1]. Salmonella enterica serovar Typhimurium (ST), a member of the non-typhoidal Salmonella (NTS) serovar, is known to cause systemic infections in mice, and self-limiting gastroenteritis in humans and pigs [2,3,4]. In the recent past, ST has been implicated in bloodstream infections resulting in approximately six hundred thousand lives globally [5, 6]. The Salmonella infection may be facilitated by in vivo circulating bacterial components, which create a favorable immunological landscape for its survival [7]. Occasionally, Salmonella causes infection in the central nervous system, where it has to breach multiple host barriers. These conditions are common in neonates, infants, immunocompromised individuals, and elderly people [8,9,10,11,12]. The mortality rates for these infections can be as high as 50% in some cases [13, 14]. It is also noted that the incidence of infections caused by Salmonella is increasing, partly due to certain Salmonella pathovars acquiring multi-drug resistance (MDR) in clinical settings [15, 16].

The modulation of Salmonella virulence genes occurs in a spatially and temporally regulated manner, which begins with Salmonella sensing its surrounding environment. Environmental sensing is done via various sensory regulation systems. One of the major envelope response-regulator systems is CpxA-CpxR, which is a two-component sensor kinase signal transduction system. It efficiently senses stresses in the periplasmic environment and controls the expression of bacterial genes associated with virulence [17]. The function of cpxA and cpxR can be functionally independent under certain circumstances; however, generally, cpxA senses stress conditions and deploys cpxR to interact with promoter regions of virulence genes and activate their specific functions [18]. The virulence targets for cpxR are very versatile, and include genes belonging to Salmonella pathogenicity island I (SPI-1) and SPI-2 [19]. During general circumstances, cpxR functions as a negative regulator of virulence genes, which could be an essential feature in survival without triggering an alarm signal by revealing Salmonella’s presence or by overexposing it within the host. Thereby, Salmonella gains the ability to hide from the host immune system and acquire virulence phenotypes such as biofilm formation and drug resistance. Unlike infections in the peripheral organs, infections in deeper organs, such as the central nervous system, require specialized genetic capability allowing Salmonella to penetrate multiple cellular barriers. Virulence modulation and proteins involved in drug resistance pathways are recognized to perform a pivotal role in this respect. Our understanding of cpxA/R could be further expanded by investigating candidate genes that are regulated in response to cpxA/R activity.

By conducting differential gene expression analysis between the Salmonella Typhimurium wild type (ST-WT) JOL 401 strain and the cpxR mutant JOL 910 strain, we observed that the acrEF/envCD operon transcriptional regulator could be affected by the absence of cpxR. By further analysis, we selected acrE and acrF as plausible candidates for further virulence study, which are also components of the AcrAB-TolC family multidrug efflux pump owing to their connection to the acrEF/envCD operon transcriptional regulator [20]. The regulatory interconnection between cpxR and acrEF is less understood. The genetic machinery of Salmonella’s antibiotic resistance involves a tripartite protein structure spanning the double membrane, which exports antibiotics from the intracellular environment to the extracellular environment [21,22,23]. The tripartite assembly consists of an outer membrane protein (OMP), an inner membrane protein (IMP) that belongs to the Resistance-Nodulation-Cell Division (RND) family, and a periplasmic membrane protein that spans by connecting two proteins above [23,24,25,26]. To determine the selectivity of drugs, the IMP plays a crucial role by catalyzing drug/H + antiporter systems [27]. The activity of efflux-related genes is regulated by local or global regulatory elements [28], one such element is the AcrAB-TolC efflux system (RND pump). It is known to regulate the pathogenesis of ST in chicken [29], however, many underlying mechanisms are yet to be uncovered. Efflux pump genes of Salmonella are often located in operons, where a regulator gene controls the expression of member genes. This arrangement ensures that Multiple genes involved in antibiotic resistance can be co-regulated, allowing for a robust response to antibiotics in the environment [30]. The interaction between stress response regulators and efflux pumps may drive the coordinated expression of virulence-related genes, enabling adaptation to various changing environments and complex host defense mechanisms, which ensures the successful establishment of Salmonella in host tissues. The evidence from the current study reveals a close interaction between cpxR, a regulator, and the acrEF proteins, which are components of drug efflux pumps, during Salmonella’s invasion of epithelial cells. We highlight a distinct difference in avrE and acrF expression during normal physiological growth conditions and the process of active invasion into epithelial cells. An increased expression of both acrE and acrF in cpxR mutant Salmonella during infection demarcates that there is a complex interconnection among these protein entities. Dramatic deprivation of both adhesion and invasion into several epithelial cells by acrEF mutant Salmonella points that there could be a more profound role of acrEF other than its already known function. This also highlights an avenue of therapeutic control of Salmonella by targeting Salmonella’s acrEF as a primary therapeutic target that can intervene in Salmonella infection irrespective of its antibiotic resistance profile.

Materials and methods

Bacterial strains, plasmids, growth conditions, and growth curve analyses

The bacterial strains and plasmids used in this study are listed in Table 1. Salmonella strains were maintained on Luria broth (LB; Franklin Lakes, NJ, USA) at 37 °C with constant agitation at 225 rpm. The medium was supplemented with 50 µg/mL chloramphenicol and/or ampicillin 100 µg/mL whenever required. All the strains were grown to the mid-log phase (0.6 OD₆₀₀) before in vitro and in vivo infection. The asd mutant strains were grown by supplementing 50 µg/mL diaminopimelic acid (DAP: Sigma, St. Louis, MO, USA). Comparisons of growth curves for the ST-WT strain, mutants, and complemented strains were performed by using a 1% inoculum collected from an overnight culture in LB media. Hundred-milliliter flasks containing 20 mL of culture were incubated with vigorous agitation, and 1 mL samples were taken hourly over 12 h for absorbance measurements. All Salmonella strains used in this study, including the wild-type strain and mutants, were derived from the JOL 401 parent strain, isolated from a chicken specimen, and confirmed by biochemical and genetic characterization.

Table 1 List of bacterial strains and plasmids used in this study

RNA sequencing and differential gene expression analysis

Salmonella wild-type (ST-WT) strain JOL 401 and mutant JOL 910 (ΔcpxR) were grown overnight in LB broth. A 1:100 dilution of the overnight-grown culture was used to inoculate 10 mL of the same medium and grown to the mid-log phase (0.6 OD₆₀₀). The bacterial cells were harvested by centrifugation at 2630 x g for 10 min. The total RNA was extracted using a RNeasy mini kit (Qiagen, Hilden, Germany), and the integrity of the RNA samples was measured by BioAnalyzer 2100 (Agilent Technologies, Baltimore, USA). The samples with an RNA integrity number of over 8.0 were used for the subsequent step. For ribosomal RNA depletion, 5 µg of the total RNA was processed by Ribo-Zero rRNA Removal Kit (MRZMB126, Illumina Inc., USA). Sequencing libraries for RNA-Seq were constructed using TruSeq Stranded Total RNA Library Prep Kit (RS-122-2201, Illumina, San Diego, CA, USA), following the manufacturer’s instructions. Sequencing was performed at Macrogen NGS service lab, South Korea, using NovaSeq 6000 System instrument (Illumina Inc), following the manufacturer’s protocol, which generated 200,400 bp paired-end reads for each sample. The sequencing adapter removal and quality-based trimming for the raw data were performed using Trimmomatic v. 0.3657 [30] with TruSeq adapter sequences. The trimmed reads were mapped to the reference genome with bowtie2 [31]. For counting the reads mapped to each coding DNA sequence, feature Counts59 was used. Finally, Fragments Per Kilobase of transcript per Million mapped reads or Reads Per Kilobase of transcript per Million mapped reads is used as a normalization value (Supplementary data set).

Construction of acrEF Salmonella mutants

The acrEF-deleted Salmonella mutants were constructed using the lambda red recombination method [32]. Briefly, the parental ST-WT strain JOL 401 and ST JOL 910 (ΔcpxR) were electroporated with pKD46. Cultures were grown at 30 °C with ampicillin and used for electro-competent cell preparation. The FRT (FLP recognition target)-flanked chloramphenicol resistance gene was PCR amplified from a pKD3 plasmid using primers carrying sequences homologous to the flanking region of the target acrE gene. The resultant PCR products were gel-purified, treated with DpnI (New England Biolabs, Ipswich, MA, USA) restriction enzyme to eliminate plasmid contaminations, and electroporated into pkD46-competent cells. Transformants were then grown at 37 °C. Successful transformants were screened on chloramphenicol agar, and the deletions were confirmed by PCR (Table S1). The extinction of pKD46 was confirmed by assessing ampicillin resistance, and the subsequent removal of the chloramphenicol-resistant cassette was achieved by transforming the pCP20 plasmid. The deletion of acrF in JOL 401 ∆acrE (JOL 2512) and JOL 910 ∆acrE (JOL 2527) was achieved by the same method, and after eliminating the antibiotic-resistance gene, the final mutants were designated as JOL 2539 (JOL 401 ∆acrEF) and JOL 2555 (ΔcpxR ∆acrEF). Complementary strains were created by cloning cpxR, acrE, and acrF into a low-copy-number plasmid, pWSK29 (Table 1).

Antibiotic sensitivity

To assess the antibiotic sensitivity of the ST wild-type strain JOL 401 and mutants JOL 910 (∆cpxR), JOL 2539 (∆acrEF), and JOL 2555 (∆cpxR ∆acrEF), a disc diffusion assay was conducted. Bacterial strains grown overnight were used to prepare a 1% fresh inoculum in LB medium. All strains grown to the mid-log phase were mixed into 4 mL of 0.4% soft agar and used to overlay onto premade LB agar plates. After solidification, antibiotic discs were placed on the bacterial lawn at a sufficient distance. Plates were incubated overnight and pictured for analysis of the diffusion circles surrounding the antibiotic discs. The diameters of the inhibition zones were measured using ImageJ Fiji software [33]. Disk diffusion assay was conducted twice, and the averages were taken for calculations.

In vitro bacterial interaction with host cells and macrophage survival

The bEND.3 (CRL-2299), Caco2 (HTB-37), Neuro2A (CCL-131), and RAW 264.7 (TIB-71) were obtained from American Type Culture Collection (ATCC), and maintained in Dulbecco’s Modified Eagle Medium (DMEM; Lonza, Walkerville, MD, USA), with 10% fetal bovine serum (FBS; Serana, Pessin, Germany) and 1% broad spectrum antibiotics. The cells were seeded in 12-well plates at a seeding density of 2མ 10⁵ cells/well and maintained in a 5% CO₂ atmosphere at 37 °C. At confluency, cells were used for interaction studies with Salmonella strains at a multiplicity of infection (MOI) of 20 in triplicate [34]. The ability of bacterial strains to adhere and invade the cells was determined using a gentamicin protection assay [35]. For the adhesion assay, bacterial cells were interacted with monolayers for 45 min, and infected cells were washed with PBS three times and lysed with 0.05% Triton-X 100 prepared in PBS. The number of adhered cells was enumerated by plating on LB agar [36]. For invasion assays, bacteria were allowed to interact with cells for 2 h, followed by washing with PBS three times and treatment with gentamycin (100 µg/mL) for 2.5 h to eliminate extracellular bacteria. After incubation, cells were washed with PBS three times and lysed using 0.5 mL of 0.05% Triton-X 100 PBS. The lysate was collected and serially diluted, and plated on LB agar for enumeration. To assess bacterial survival within phagocytic cells, a macrophage survival assay was conducted. Murine macrophage cell line RAW 264.7 cells were grown in 12-well plates at a seeding density of 2 × 10⁵ cells/mL for 24 h. When the confluence is > 80%, cells were infected with ST wild type JOL 401, and mutants JOL 910 (∆cpxR), JOL 2539 (∆acrEF), and JOL 2555 (∆cpxR ∆acrEF) for 2 h at 40 MOI. After incubation, gentamycin treatment was conducted, and cells were replenished with fresh media containing 10 µg/mL gentamycin and further incubated for 12 h. After incubation, surviving bacterial cell numbers were enumerated by plating of serial dilutions.

Actin polymerization assay

The ability of the bacterial mutants to induce cytoskeletal rearrangement in the bEND.3, Caco2, Neuro2A, and RAW 264.7 cells were determined using the Alexa Fluor^®-phalloidin method as previously described with minor modifications [34]. Briefly, cells were seeded at a density of 1 × 10⁶ cells/well on sterile glass coverslips placed in 24-well plates. Cells were incubated overnight at 37 °C in a 5% CO₂ environment and infected with the Salmonella WT or mutant strains at 20 MOI. Incubation was conducted for 1 h, and cells were washed three times with PBS to remove non-adherent bacteria. Cells were fixed with 4% paraformaldehyde for 15 min at room temperature and permeabilized with 0.01% Triton X-100 in PBS for 10 min at room temperature. Cells were subsequently incubated with Alexa Fluor 488-conjugated phalloidin (Invitrogen, Carlsbad, CA, USA) in PBS containing 1% bovine serum albumin (BSA) for 30 min in the dark. Coverslips were then processed and fixed on mount media, and the degree of actin polymerization and membrane ruffling induced by each strain was visually evaluated under fluorescence microscopy (Leica, Wetzlar, Germany) and compared. The fluorescent intensity of each image was quantified using ImageJ Fiji software. The green fluorescent intensity was normalized against the background intensity of each image.

Quantitative RT-PCR analyses

The transcriptional level of Salmonella SPI-1 bacterial genes avrA, hilD, invF, spiA, sopB, and ompA, in adherent bacteria, was evaluated. Briefly, bEND.3 cells grown in 12-well plates were infected with ST-WT JOL 401 and mutant strains, JOL 910 (∆cpxR), JOL 2539 (∆acrEF), and JOL 2555 (∆cpxR ∆acrEF) at 20 MOI in triplicate. Thirty minutes post-infection, cell monolayers were lysed, and bacterial cells were collected by centrifugation. The total RNA was isolated (GeneAll, Daejeon, South Korea) and reverse transcribed using a PrimeScript™ 1st strand cDNA Synthesis Kit (Takara, Tokyo, Japan). To evaluate gene expression in selected host cell-associated genes, bEND.3 cells were similarly infected with bacteria and treated with gentamycin (100 µg/mL) for 2.5 h. After incubation, cells were washed and replaced with complete medium containing gentamycin 10 µg/mL. The incubation was conducted for 12 h, and the total RNA was isolated, and cDNA was prepared. Quantitative real-time PCR was conducted for pro-inflammatory cytokine markers, TNF-α, IL-1β, INF-γ, IL-8 and host apoptotic factors, ENOPH1, ADI1, Caspase-3, and the expression levels of host cell tight junction-associated genes ZO-1,-2, -3, OCLN, CLDN-1, 3, 5, and 12 were evaluated. Expression values were normalized using housekeeping genes, rrsG for genes with prokaryotic origin and GAPDH for genes with eukaryotic origin. Results were presented as the 2^−ΔΔCT method (Livak and Schmittgen, 2001).

Cytotoxicity assay

The induction of cytotoxicity by ST-WT, and mutants was assayed by the propidium iodide staining method using bEND.3 cells. Cells were grown in 12-well plates in triplicate until confluence and used in the infection process with ST-WT strain JOL 401 and mutants, JOL 910 (∆cpxR), JOL 2539 (∆acrEF), and JOL 2555 (∆cpxR ∆acrEF). After 2 h post-infection (hpi), the cells were washed three times with PBS and treated with medium containing gentamycin 100 µg/mL for 2.5 h to eliminate extracellular bacteria. Cells were further incubated for 12 h under 10 µg/mL gentamycin to suppress bacterial growth. For imaging, the cell medium was supplemented with propidium iodide 5 µg/ mL and observed using the IncuCyte live cell imaging system (Göttingen, Germany). Images collected after two hours of incubation were taken for comparison [37].

Transepithelial electrical resistance (TEER) measurement

Transendothelial/transepithelial electrical resistance (TEER) was measured in monolayers of bEND.3 and Caco-2 cells grown in Transwell^® plates (Corning, Steuben County, NY, USA). Briefly, the cells were grown in the upper chamber (12 mm insert) of the Transwell^® system equipped with 0.4 μm pore polycarbonate filters (Corning, New York, USA) for 14 days and treated with 5 mM EGTA, WT ST, mutant ST, or culture medium (pH 7.5). The TEER measurements were obtained using a Millicell^®-ERS meter (Millipore, Inc., Billerica, MA, USA) at 2, 4, and 6 hpi. For each condition, three wells were employed. Each time, the probe was inserted and remained until the instrument reading was stabilized. Once stabilized, the reading was recorded.

Transmission electron microscopy (TEM) to evaluate TJ structural integrity

The degree of TJ disruption of Salmonella-infected bEND.3 cells were assessed by transmission electron microscopy [38]. The confluent monolayer was infected with the ST-WT strain or ST mutants, JOL 910 (∆cpxR), JOL 2539 (∆acrEF), and JOL 2555 (∆cpxR ∆acrEF) or EGTA at a 5 mM/ mL concentration. Three hours post-infection, the cells were fixed in fixation buffer (2.5% glutaraldehyde, 0.08 M sodium cacodylate, 5 mM CaC1₂ at 10% v/v, pH 7.4) at 4 °C overnight. Cells were post-fixed with 1.5% osmium tetroxide phosphate buffer (pH 7.4) for 2 h. Cells were dehydrated in an ethanol gradient and negatively stained with 1.5% uranyl acetate. Cells were embedded and subjected to sectioning at 0.5 μm thick slices using a microtome. Cells were observed with a Hitachi H-7650 transmission electron microscope using an acceleration voltage of 100 kV (Tokyo, Japan).

Infection of mice

Five-week-old specific pathogen-free (SPI) female BALB/c mice (N = 20, n = 5) were purchased (Koatech, Pyeongtaek, Korea) and randomly allocated to four groups. Mice were inoculated intraperitoneally with 1 × 10⁵ CFU/100 µL of ST-WT strain JOL 401, or ST mutants, JOL 2539 (∆acrEF), JOL 910 (∆cpxR), and JOL 2555 (∆cpxR ∆acrEF) suspended in 100 µL of PBS [39]. Mice were housed in the animal research facility at the College of Veterinary Medicine, Jeonbuk National University, and provided with sterile food and water ad libitum. Twelve-hour day and night cycles were maintained throughout the study period. Mice were euthanized by exposure to chloroform at 5 dpi, following the safety precautions. Briefly, a transparent, sealable glass jar was kept in a fume hood, and an absorbent pad containing 1.0 mL/100 g body weight was placed inside. Mice were confined to the jar and observed until the termination of reflexes. Then the brain, spleen, and liver were aseptically isolated. Half of each organ sample was fixed in 10% formalin and processed for histopathological analyses. Bacterial load was determined in the remaining tissues (CFU/g). All animal experiments were performed according to the methods approved by the Chonbuk National University Animal Ethics Committee (NON2022-024-002). Images of dissected brain tissues were obtained by using the FOBI live in vivo imaging system (Yuseong-gu, Daejeon, South Korea). To investigate the appearance of infected mice brains, a separate set of mice (N = 20, n = 4) was intraperitoneally infected using the same dose conditions. At 5 dpi, mice were euthanized, and their brains were extracted for imaging using the FOBI imaging system.

Histopathological analysis of organ samples

Tissue samples collected from the brain, liver, and spleen of test mice were fixed in neutral buffered formalin for 4 days at 4 °C. Tissues from infected and uninfected mice were subjected to a histopathological process to investigate bacterial colonization and bacteria-associated tissue damage. In brief, the samples were deparaffinized using xylene, rehydrated through a decreasing concentration of ethanol, and finally immersed in distilled water for 5 min before different staining. Haematoxylin-eosin staining was performed to determine any observable bacterial infection-mediated morphological and pathological changes, such as inflammation, hemorrhages, and possible tissue necrosis within the tissue Sections [40].

Immunofluorescence assay

The immunofluorescence assay (IFA) was performed to examine the presence of Salmonella in the brain of infected mice. For antigen retrieval, slides were placed in antigen retrieval buffer (10 mM sodium citrate, pH 6.0) and then heated at 95 °C for 15 min using a water bath. Washed with PBS, and the endogenous peroxidase was blocked by hydrogen peroxide (3% in methanol). After blocking with 5% skim milk for 1 h, the sections were incubated overnight with a primary rabbit antibody (1:100) against Salmonella at 4 °C. Afterward, the sections were stained with Alexa Fluor 488-conjugated donkey anti-rabbit IgG (1:5000) (Invitrogen, USA) as a secondary antibody, and the sections were mounted with MM 24 Mounting Media (Leica Biosystems, Wetzlar, Germany) for microscopy. Fluorescence quantification was done using ImageJ Fiji software.

Protein structure modeling and protein docking

Protein modeling of AcrE and AcrF from Salmonella enterica serovar Typhimurium LT2 was performed using the Phyre2 protein fold recognition server [68]. The structure of the zonula occludens 1 (ZO-1; also termed the tight junction 1) protein of mice was modeled using sequences from the NCBI database (accession number NM_001163574.1). Protein-protein docking was conducted in the ClusPro web server [41]. Based on the analysis result, we relied on the top-ranked clusters, which represent the most frequently sampled confirmations, which are generally regarded as most reliable binding orientations [42].

Bacterial two-hybrid (B2H) assays for AcrEF and TJ zona occludens (ZO-1)

The interaction between bacterial proteins and host proteins was studied using a bacterial two-hybrid assay (B2H), as previously described[70]. Briefly, pUT18 and pKT25 plasmids containing the genes of interest (Table S1) were electroporated together into Escherichia coli BTH101(cya). Then, the strains were allowed to grow overnight at 30 °C in LB supplemented with ampicillin (100 µg/mL) and kanamycin (50 µg/mL). Then, 2 µL of cells were spotted onto LB agar containing 1 mM IPTG, 100 µg/mL ampicillin, 50 µg/mL kanamycin, and 20 µg/mL X-Gal, followed by incubation at 30 °C for 40 h. Further, a detailed quantitative analysis was performed using β-galactosidase assays.

Preparation of antibodies

Anti-AcrE and AcrF polyclonal antibodies (pAbs) were raised in New Zealand white rabbits (N = 2) as previously described [43]. The concentration of the pAbs was determined by the BCA method, and their final concentration was adjusted to 20 mg/mL. The titer of the antigen-specific IgG antibodies in each pAb was determined using an ELISA. The ELISA microtiter plates (Greiner Bio-One GmbH, Frickenhausen, Germany) were coated with AcrE or AcrF recombinant protein (250 ng/well). The antibody titer was defined as the highest dilution with an absorbance value (OD₄₉₂) of more than three times the negative control.

In vitro protection efficacy of anti-AcrEF antibodies

The protection efficacy of the anti-AcrEF pAbs was evaluated in vitro using bEND.3 cells [44]. Briefly, confluent monolayers of bEND.3 cells grown on 12-well plates were washed and replenished with DMEM containing diluted anti-AcrEF pAbs mixed with 10⁷ ST-WT JOL 401 or JOL 910 (∆cpxR). The plates were incubated at 37 °C for 2 h, a period sufficient for the bEND.3 cells to internalize the ST. A standard invasion assay was conducted to determine the neutralizing efficacy of anti-AcrEF polyclonal antibodies. Similarly treated wells were fixed with 4% PFA and used for IFA detection of internalized bacterial cells.

Bacterial challenge and salvage using anti-acrEF antibodies

Five-week-old female BALB/c mice (N = 20, n = 5) were randomly divided into four groups. The groupings were included ST-WT strain JOL 401, JOL 401 + polyclonal antibody (pAb), JOL 910 (∆cpxR), JOL 910 + pAb. Two hours post-infection, mice were intraperitoneally administered with 100 µL of v/v combined Abs (2 mg) around the same inoculation location. The mice were observed for any observable changes. On the day 5 post-infection, all mice were euthanized to harvest brain, liver, and spleen for bacterial enumeration and IFA.

acrEF and tolC expression in intracellular Salmonella

bEND.3 and RAW 264.7 cells were seeded in a 12-well tissue culture plate at 2 × 10⁵ cells/well. When the cells reached > 80% confluence, Salmonella infection using ST-WT strain JOL 401, JOL 910 (∆cpxR), JOL 2539 (∆acrEF), or JOL 2555 (∆cpxR ∆acrEF) was carried out at a multiplicity of infection (MOI) of 100 for 2 h. Subsequently, cell monolayers were washed three times in PBS and incubated in prewarmed DMEM with 100 µg/mL gentamicin for 1 h to eliminate extracellular bacteria. The culture medium was changed to DMEM with 10 µg/mL gentamicin. At 1 and 6 hpi, cells were washed three times with phosphate-buffered saline (PBS) and then lysed. Cell lysates were first centrifuged at 600 × g for 5 min to remove nuclei and cell debris. The resulting fractions were further spun at 4,000 × g for 20 min to pellet intracellular bacteria. Total RNA from the intracellular bacteria was extracted, followed by 1st strand cDNA synthesis. The results were normalized against rrsG using the 2^−ΔΔCT method [45].

Fluorescent microscopy

To investigate the intracellular expression and localization of Salmonella-expressed AcrE protein in Caco2 epithelial cells, the acrE open reading frame coupled to a His-tag sequence was cloned into an expression vector (pMPP65), accompanied bla secretion signal. The plasmid was then transformed into an attenuated Salmonella strain, JOL 2800 (∆lon∆sifA∆asd). As a negative control, the Salmonella WT strain lacking acrEF, JOL 2539 (∆acrEF), was employed. After bacteria-mediated transfection at 40 MOI, cells were incubated for 6 h, and then monolayers were washed and treated with gentamycin to eliminate non-infected Salmonella. Cells were washed and fixed with 4% paraformaldehyde and permeabilized with 0.05% Triton X-100 for 10 min at room temperature. After washing, cells were subjected to antibody treatment with mouse anti-His-tag antibody (1:3000) followed by staining and Goat-anti-mouse-Alexa Flour 488 (1:4000) and DAPI at 5 µg/mL (Sigma, St. Louis, MO, USA), for a confocal assay. Images were captured and examined for green fluorescence signals.

Statistical analysis

All statistical analyses were performed using GraphPad Prism 6.0 software (GraphPad, Franklin Street, Boston, MA, USA). A p-value < 0.05 was considered significant. Data in graphs are presented as the mean ± SEM with *p < .05; **p < .01; ***p < .001. ANOVA (Post-hoc test) was conducted to compare the statistical differences between groups. The false-discovery rates (FDR) for genes were controlled at < 1%. Only the genes that were detected in ≥ 2 individual samples from either JOL 401 or JO 910 (∆cpxR) were selected as quantified genes. Genes with q values of < 0.05 and ratios (JOL 910/JOL 401) of > 2.0 or < 0.5 were considered differentially expressed.

Results

The effect of cpxR deletion on acrEF expression

Transcriptomic changes of ST in the absence of the cpxR gene were investigated by total RNA sequencing by comparing the cpxR mutant (JOL 910) and the ST wild-type strain (JOL 401) under ordinary environmental conditions. About 86% of the total genes of Salmonella were expressed, accounting for 4,311 ST genes (Fig. 1Aa). Herein, the sequencing results of JOL 910 the cpxR mutant against ST-WT, revealed a total of 1837 genes were upregulated, while 2471 were downregulated. Among these genes, 322 genes were found to be differentially expressed. The false-discovery-rate [FDR [46]] value of < 0.05 in JOL 910 compared to ST-WT JOL 401. The volcano plot generated by quantitative genomic data revealed that out of 322 differentially expressed genes, 265 genes as up-regulated and only 57 genes were down-regulated (Fig. 1Ab). Up-regulated and down-regulated genes are demonstrated as red and blue spots, respectively. We observed that deleting the cpxR gene from the ST-WT strain did not cause a significant expression change either in acrE or acrF. However, the acrEF/envCD operon transcriptional regulator was among the upregulated genes (log 3.02 with FDR 1.47). The RNA isolation was done for RNA sequencing during normal physiological states of bacterial cells (Fig. 1Aa and Ab; green spot). However, under monolayer infection conditions, though the acrEF/envCD operon remains non significantly different between the ST-WT strain JOL 401 and the cpxR mutant JOL 910 (Fig. 1B), we noted that the expression of individual acrE and acrF was up-regulated > 4-fold in the cpxR mutant JOL 910 strain as compared to the WT strain, suggesting that cpxR may suppress acrEF expression during epithelial infection. Similarly, the transcriptional levels of the outer membrane channel tolC of the acrEF-tolC increased by 6-fold in ∆cpxR (Fig. 1B). This observation compelled us to further investigate acrE and acrF on Salmonella virulence during infection of cell monolayers (Fig. 1B).

Fig. 1

Genomic landscape of ST lacking cpxR. (Aa) A volcano plot of bacterial genes detected by RNA sequencing in JOL 910 (∆cpxR). The logarithmic ratios of average fold changes are reported on the x-axis. The y-axis data plot negative logarithmic false-discovery-rate (q) values. Upregulated genes were marked in red, and downregulated genes were marked in blue. The total number of genes was 4311, where 1837 were upregulated and 2471 genes were downregulated. Green dot- acrEF/envCD operon transcriptional regulator (Ab). The volcano plot shows the differentially regulated 322 genes. Up and downregulated proteins are denoted by the red and blue dots, respectively. Green dot- acrEF/envCD operon transcriptional regulator. Among the total 322 differentially expressed genes, 265 was upregulated and 57 was downregulated. (B) The qRT-PCR for validation of RNAseq results was expressed as the mean fold change in transcription of acrEF/envCD operon transcriptional regulator. Also, the mRNA levels of acrE, acrF, and tolC genes compared to rrsG in ST-WT JOL 401 and JOL 910 (∆cpxR). The data are represented as the mean relative fold change ± the SD from at least three independent experiments performed in duplicate. Means were compared using Tukey’s multiple comparison test. Significant differences (ns- not significant, **p < .01, ***p < .005) were assessed compared to the parental control. acr = Acriflavin

The effect of acrEF on Salmonella’s antibiotic sensitivity

The assessment of the sensitivity of ST-WT strain JOL 401 and mutant strains towards eight classes of common antibiotics was investigated by disc diffusion assay. Overall, the wild-type strain (JOL 401) exhibited moderate inhibition zones across most antibiotics, reflecting the baseline multidrug resistance characteristic of Salmonella Typhimurium parent strain. The ΔcpxR mutant (JOL 910) showed inhibition profiles largely similar to the wild type, suggesting that loss of cpxR alone does not markedly affect antibiotic susceptibility. In contrast, the ΔacrEF mutant (JOL 2539) displayed significantly larger inhibition zones, particularly against several β-lactams (aztreonam, amoxicillin, ceftazidime, cephazolin) and tetracyclines (doxycycline, tetracycline), highlighting the critical role of the AcrEF efflux pump in resistance to these drug classes. The double mutant ΔcpxR ΔacrEF (JOL 2555) mirrored the ΔacrEF phenotype, with strong increases in susceptibility to β-lactams and tetracyclines, indicating that the loss of acrEF function is the dominant factor driving this phenotype. Antibiotic class-wise analysis also showed that aminoglycoside resistance (amikacin, gentamycin) remained mostly unchanged across all strains, carbapenems (meropenem, imipenem) were ineffective against all strains, and fluoroquinolone (ciprofloxacin) susceptibility was consistent, indicating limited impact of acrEF on these agents. Similarly, all strains remained resistant to macrolides (erythromycin), linezolid, trimethoprim, and vancomycin, consistent with intrinsic resistance mechanisms. In summary, deletion of acrEF, either alone or in combination with cpxR, significantly increased susceptibility of S. Typhimurium to β-lactams and tetracyclines, whereas cpxR deletion alone had little effect. These findings emphasize the pivotal role of the AcrEF efflux system in maintaining multidrug resistance in Salmonella (Fig. 2A and B).

Fig. 2

Assessment of antibiotic sensitivity. Salmonella wild-type strain JOL 401 and mutants JOL 910 (∆cpxR), JOL 2539 ((∆acrEF), and JOL 2555 (∆cpxR ∆acrEF) were assessed via disc diffusion assay for 18 individual antibiotic strains belonging to eight classes. Sensitivity was compared by considering the diameter of the halo zone surrounding the disc area. (A) Relative resistance (by inhibition area) normalized against the ST-WT strain JOL 401 was demonstrated. (B) The appearance of each inhibition zone is demonstrated by comparing the ST-WT JOL 401 strain against JOL 910 (∆cpxR), JOL 2359 (∆acrEF), and JOL 2555 (∆cpxR ∆acrEF) mutant strains

Adhesion and invasion assay of acrEF mutants

Adhesion and invasion assays were conducted using bEND.3, Caco-2 and Neuro2A cells revealed that the deletion of the acrEF gene resulted in a significant reduction in adhesion and almost abolished ST’s ability to invade bEND.3 and Caco-2 cell lines. The invasion into Neuro2A cells is retained but with significantly lower levels compared to ST-WT JOL 401 and JOL 910 (∆cpxR). The single mutant of the cpxR gene (JOL 910) did not confer a reduction either in adhesion or in invasion. Additionally, we observed that the incorporation of cpxR deletion into acrEF deletion in Salmonella (JOL 2555) did not exhibit a synergistic effect (Fig. 3A). Unlike adhesion, we observed a complete abortion of invasion of acrEF mutants, JOL 2539 and JOL255 (∆cpxR ∆acrEF), into bEND.3 and Caco-2 epithelial cells, demonstrating its essential role for epithelial invasion, irrespective of the epithelial origin. Complementation assays could completely re-establish invasiveness, proving functional involvement of these genes for the invasion process (Figure S3A, B, and C). Earlier studies have indicated that the disruption of the acrB gene also caused a compromise in the adhesion and invasion ability of ST strains [46]. Thus, it is evident that acrEF plays a crucial role in Salmonella adhesion and invasion into cells such as bEND.3 and Caco2, which can be further investigated to reveal the mechanistic aspects of Salmonella infection into epithelial cells. Corresponding to a lack of membrane stress tolerance and adaptability due to cpxR and acrEF deletions, Salmonella compromised their ability to reside in murine macrophages, though their impact is marginal on macrophage survival (Fig. 3B).

Fig. 3

In vitro Adhesion, invasion assays, macrophage survival, Differential expression of SPI-1 genes, and actin polymerization. (A) Adhesion and invasion of bEND.3, Caco-2, and Neuro2A monolayers were observed post-infection with ST-WT JOL 401, and ST mutants JOL 910 (∆cpxR), JOL 2539 (∆acrEF), and JOL 2555 (∆cpxR∆acrEF) at a multiplicity of infection 20 (MOI = 20). The data represent the adhesion and invasion (Log₁₀ CFU/mL) ± the SD from at least three independent experiments performed in duplicate. P values of < 0.05 were considered statistically significant. Means were compared using Tukey’s multiple comparison test. Significant differences (ns- not significant, **p < .01) were assessed compared to the parental control. Letters “a” and “b” indicate the absence of colonies. (B) The macrophage survival assay was conducted on RAW 264.7 cells after infection with the ST-WT strain and mutants for 12 h. Bacterial survival was compared by lysing cells, followed by plating serial dilutions on agar plates. (C) The bacteria (ST-WT and mutants) adhering to bEND.3 cells were isolated at 0.5 hpi, and the transcription of avrA, hilD, invF, sipA, sopB, and ompA was evaluated. The mRNA levels for the SPI-1 genes were compared in infected and uninfected cells and normalized to the expression of the Salmonella housekeeping gene, rrsG. The data are presented as the mean relative fold change ± the SD from at least three independent experiments performed in duplicate. Means were compared using Tukey’s multiple comparison test. Significant difference (ns- not significant, *p < .01, **p < .0001) was assessed compared to the wild-type control. (D) The ability of bacteria to alter the host cytoskeleton is essential for the invasion of host cells. The image depicts actin polymerization in bEND.3 (endothelial) and Neuro2A (nerve) cell lines. The infected cells were stained with phalloidin (for F-actin) and DAPI (nuclear stain). White arrow- actin arrangement; yellow arrow- actin condensation. Mean fluorescent intensity of each image was quantified using the ImageJ Fiji software and indicated at the bottom of each figure, along with SD

The effect of acrEF on Salmonella virulence regulators

The Salmonella adhesion and invasion process is largely regulated by various effector proteins located in the Salmonella pathogenicity island, SPI-1 [47, 48]. Once inside the cell, chronic survival is governed by the association of SPI-2 genes. During the infection of bEND.3 monolayers, the ST-WT strain JOL401 strongly induced the SPI-1 regulators hilD and invF as well as downstream effectors sipA and sopB, while avrA was repressed. In contrast, the acrEF mutant (JOL910) and cpxR mutant (JOL2539) showed significantly reduced expression of hilD and its regulon, with the effect being most pronounced in the double mutant (JOL2555), which exhibited an additive reduction in SPI-1 gene induction. These findings indicate that both the acrEF efflux system and the CpxR regulator contribute to the optimal activation of the hilD-invF regulatory cascade, thereby promoting transcription of invasion-associated SPI-1 effectors during endothelial cell infection (Fig. 3C). Furthermore, several outer membrane proteins, such as OmpA also down-regulated in these mutants, potentially illustrating their inability to cause efficient adhesion and invasion into epithelial cells. The internalization and infection process requires extensive reorganization of the actin cytoskeleton, which facilitates ST invasion into the cell [49]. Here, we noticed that the absence of acrEF demonstrated a significant absence of such actin modulation in bEND.3 cells compared to the ST-WT JOL401 and JHL 910 (∆acrEF). It was also notable that such an effect was not apparent in Neuro2A cells receiving the same treatments (Fig. 3D). Fluorescent intensity measurements further revealed that the impact of ST-WT and mutants’ invasion into mice macrophage RAW164.7 cells minimally changed the actin reorganization, possibly due to their phagocytic nature, leading to active engulfing of bacteria.

Effect of acrEF deletion on pro-inflammatory cytokine expression

Upon internalization of ST, it triggers the release of pro-inflammatory cytokines, which may be a reason for the breakdown of the BBB [50]. In this study, we observed that the infection of JOL 401 and JOL 910 (∆cpxR) caused significant upregulation of pro-inflammatory cytokines at 6 h post-infection (hpi) over 5-fold and continued to stimulate their production up to 24 hpi over 80-fold. TNF-α levels were the highest, followed by IL-1β, IL-8, and IFN-γ. In contrast, JOL 2539 (ΔacrEF) and JOL 2555 (ΔcpxRΔacrEF) induced cytokine production by less than 2-fold at 6 hpi and caused no further changes thereafter. Additionally, intracellular bacterial replication activates the host apoptotic cascade. Enolase-phosphatase 1 (ENOPH1) in the BBB is known to regulate endothelial cell death and apoptosis [51]. In this study, bEND.3 cells infected with JOL 401 exhibited a more than 10-fold increase in mRNA levels of ENOPH1, as well as expression of its downstream protein aci-reductone dioxygenase 1 (ADI1) at 12 hpi. A similar increase was observed in the expression of Caspase-3, a key apoptosis-associated protein, in bEND.3 cells infected with JOL 401 (Fig. 4A). Cells infected with JOL 910 showed more than a 6-fold increase in the expression of apoptotic factors. On the other hand, JOL 2539 (ΔacrEF) and JOL 2555 (ΔcpxRΔacrEF) caused less than a 2-fold increase in host apoptotic factor expression. These results were further validated by live cell imaging using the Incucyte system (Fig. 4B), demonstrating that the mutants are lacking in cytotoxic responses. This effect was similar in acrEF mutants, irrespective of the presence or absence of cpxR gene function. Further, the trans-complementation of acrEF function restored Salmonella virulence and cytotoxic responses (Figure S3), confirming their pivotal role in virulence modulation.

Fig. 4

acrEF alters the expression of host pro-inflammatory cytokines, apoptotic factors, and cell lysis. (A) The mean fold change in the transcriptional levels of cytokines (TNF-α, IL-1β, IL-8, or IFN-γ) at 6, 12, and 24 hpi and host apoptotic factors (ENOPH1, ADI1, and Caspase-3) at 12 hpi in bEND.3 cells. The data are represented as the mean relative fold change ± the SD from at least three independent experiments performed in duplicate. Means were compared using Tukey’s multiple comparison test. Significant differences (ns-not significant, *p < .005, **p < .01, ***p < .001) were assessed compared to the control. (B) Propidium iodide staining was conducted using bEND.3 cells upon ST-WT JOL 401, the cpxR mutant JOL 910, the acrEF mutant JOL 2539, and the cpxR and acrEF double mutant. Representative images were taken after 2 h of gentamycin treatment. Yellow color arrows point out places of extensive cell cytotoxicity

The effect of deleting acrEF on host tight junctions

Salmonella utilizes multifaceted mechanisms to breach cell tight junctions by utilizing specific effector proteins and secretion systems that facilitate its invasion through the epithelial barrier. Salmonella can be involved in the modification of tight junctions by altering the prejunctional actomyosin ring during invasion and infection [52]. Gene expression analysis revealed that the bEND.3 cells infected with JOL 2539 (ΔacrEF) and JOL 2555 (ΔcpxRΔacrEF) exhibited maintenance of expression of ZO-1, -2, and − 3 more than 4-fold (Fig. 5A) compared to the ST-WT strain JOL 401 and the cpxR mutant JOL 910. We also noted increased expression of leaky protein occludin (OCLN) and claudins (CLDN-1, 3, 5, and 12) in cells treated with acrEF mutant strains JOL 2539 and JOL 2555 (Fig. 5B). The expression of these genes was significantly low in ST-WT strain JOL 401 and cpxR mutant JOL 910-treated cells. The distortion of tight junction-related genes in ST-WT JOL 401 and JOL 910 (∆cpxR) was associated with the permeability of endothelial monolayers [53] of bEND.3 and Caoco2 cells, which could be quantified by measurements of trans-epithelial electrical resistance (TEER) measurements. A distinct drop in TEER values was observed for JOL 401 and JOL 910 after 2 h of infection in epithelial cells of bEND.3 and Caco2 grown in trans-well plates, and didn’t reverse the effect until the 6 h time points, indicating pronounced epithelial damage caused by ST during the invasion process. However, these observations were absent in ΔacrEF mutants (Fig. 5C and D). This observation could be associated with impaired invasion into bEND.3 cells and due to the acrEF effect on other virulence-related genes, which are associated with tight junction modulation, such as sopB.

Fig. 5

acrEF alters the expression of host TJ protein and disrupts the ultrastructure of the paracellular barrier using bEND.3 cells. (A) The mean fold change in the transcriptional levels of Zona occludens (ZO-1, -2, and 3) genes. (B) The mean fold change in the transcription of occludens (OCLN) and claudin (CLDN-1, -3, -5 and 12) genes. Means were compared using Tukey’s multiple comparison test. Significant differences (ns-not significant, *p < .005, **p < .01, ***p < .001) were assessed compared to the control. (C) TEER measurements in bEND.3 cells measured throughout 6 hpi with ST strains were compared with the TEER of cells treated with EGTA and uninfected control cells. (D) Representative transmission electron microscopy (EM) images show bacterial interaction-mediated disrupted TJs in EGTA-treated cells, cells infected with JOL 401, JOL 910, JOL 2539, or JOL 2555. The changes in the TJ ultrastructure in infected cells were compared with TJs in healthy cells. Purple color triangles indicate the extensively damaged tight junctions, while green colors indicate the preservation of tight junctions on bEND.3 cells infected with acrEF mutants

Transmission electron microscopy observation of the tight junction

To further confirm TEER measurements across epithelial cell lines, we conducted transmission electron microscopy to examine the state of tight junctions upon bacterial interaction with epithelial cells. Results revealed, in comparison to the EGTA-treated positive control, the ST wild-type strain JOL 401 and JOL 910 (∆cpxR) caused significant enlargement of the tight junction gap (Fig. 5C and D). However, JOL 2539 (∆acrEF) and JOL 2555 (∆cpxR ∆acrEF) mutant strains significantly preserved tight junction integrity, as almost comparable to that of healthy cells. These results indicate that the acrEF confers a significant role in Salmonella’s effect on epithelial breach and damage to the tight junction integrity. Here, too, we are unable to distinguish any additive effect of acrEF deletion on cpxR mutants regarding tight junction damage. It is also possible to hypothesize that the deprived internalization effect of acrEF mutants may prevent Salmonella’s impact on cellular damage, while superficial interaction may not cause an effect on the cellular tight junctions. To precisely elucidate these mechanisms, further elaborations are essential.

In vivo infection of Salmonella mutants in mice

In vivo, bacterial colonization in BALB/c mice revealed that ST-WT JOL 401 and JOL 910 (∆cpxR) showed significantly higher colonization of the mouse brain (Fig. 6A and B). However, deletion of acrEF in the ST-WT strain and cpxR mutants resulted in no detectable bacterial colonization in cerebral tissue by 5 days post-infection (dpi) (Fig. 6A and B). The external appearance of extracted brains also reveals fewer hemorrhagic signs in acrEF mutants; however, it was high in the ST-WT strain and the cpxR mutant strain. No significant differences were observed in bacterial colonization in the liver across the groups (Fig. 6B), however the spleen colonization was significantly low for acrEF mutants, JOL 2539 and JOL 2555. Pathological evaluation revealed that brains infected with ST-WT strain cpxR mutant (JOL 910) exhibited severe neutrophil infiltration and hemorrhagic lesions, while no notable pathological changes were observed in brains infected with JOL 2539 (∆acrEF) or JOL 2555 (∆cpxR ∆acrEF) (Fig. 6C). Immunofluorescence confirmed bacterial localization in brain tissues following infection with ST-WT (JOL 401) or cpxR mutant JOL 910, but no detectable bacterial presence was observed in cerebral tissues infected with JOL 2539 (∆acrEF) or JOL 2555 (∆cpxR ∆acrEF). All groups showed evidence of hepatomegaly and splenomegaly, indicating the ability of all strains to colonize the spleen and liver. Infected spleens exhibited diffuse non-follicular enlargement of the white pulp, reduction of the red pulp, and a distorted splenic architecture. Similarly, liver samples showed dilated sinusoids and hepatocyte degeneration.

Fig. 6

In vivo colonization of ST-WT JOL 401 and mutants JOL 910 (∆cpxR), JOL 2539 (∆acrEF), and JOL 2555 (∆cpxR ∆acrEF) in mice. (A) A group of mice (N = 20, n = 4) was intraperitoneally infected for brain morphology assessment. On 5th dpi, brains were extracted and imaged for comparison. Figure demonstrates the external appearance of infected mice brains with each ST strain, and (B) the colonization of the brain, spleen, and liver by bacterial mutants was evaluated and compared with the invading ability of WT Salmonella. The bacteria in the infected mice’s brains were enumerated by plating the serially diluted homogenized brain on LB agar plates (n = 4 mice/group). Letters “a” and “b” indicate the absence of colonies observed in the brains of JOL 2539 (∆acrEF) and JOL 2555 (∆cpxR ∆acrEF) infected mice, respectively. Representative image of bacterial recovery from the infected mouse brain. The data are presented as log CFU/g ± the SD from at least three independent experiments performed in duplicates. Means were compared using Tukey’s multiple comparison tests. Significant differences (ns-not significant, *p < .005) were assessed compared to the healthy control. An infected brain manifests pathological lesions associated with bacterial internalization. The mice (n = 5/group) were administered 10⁵ bacteria (100 µL volume) intraperitoneally. Five days after infection, the samples were embedded and processed for immunohistological analysis of bacterial-infection-associated cerebral tissue damage. (C) The microscopic alterations in the brain cytoarchitecture were evaluated by H and E staining. The degree of bacterial localization in the brain tissue was evaluated using an immunofluorescence assay. Black arrowheads indicate vascular congestion, green arrowhead indicates hyperchromatic cells, and yellow arrowheads indicate bacterial localization in brain tissue. Mean fluorescent intensities obtained by ImgeJ Fiji software were indicated at the bottom of each fluorescent image with standard deviation

Potential interaction of acrE and acrF with tight junction genes

Even though AcrE and AcrF proteins are localized in the inner and periplasmic space, these proteins contain the possibility to interact with the outer environment. Portions of proteins can protrude out from the outer membrane and occasionally shed into the external environment, leading to immune modulation. In the present context, we evaluated the outcome of AcrE and F proteins with host cell tight junction protein ZO-1. In silico modeling analyses revealed that AcrE and AcrF can interact with the PDZ domain of ZO-1. We hypothesize that the distorted TJ-associated delocalization of ZO-1 may augment the interaction between AcrEF and the PDZ of ZO-1. An energy score of − 702.4 kcal/Mol was derived for the interaction between ZO-1 and AcrE (Fig. 7A) and of − 1060.0 kcal/Mol for the interaction between ZO-1 and AcrF (Fig. 7B). The interactions were validated by a bacterial two-hybrid assay by expressing the fusion of the C-terminal cyaA-T18 fragment to the acrE or acrF genes, and of the N-terminal fusion of the cyaA-T25 fragment to the PDZ domain of the ZO-1 gene in an E. coli lacking CyaA adenylate cyclase. The cells were then spotted onto an LB plate containing X-gal and IPTG. A plausible heterodimerization between ZO-1 and the bacterial proteins leads to functional complementation between T25 and T18 fragments. The intensity of blue color produced correlated with the degree of interaction, indicating a stronger interaction between ZO-1 and AcrF compared to ZO-1 and AcrE (Fig. 7C). These results suggest the enhanced affinity between AcrF and PDZ of ZO-1 may be due to the AcrE–AcrF stoichiometry of 6:3 [39, 40]. Further, the efficiency of this interaction was measured by assaying the β-galactosidase enzymatic activities in bacterial extracts, wherein a higher β-galactosidase activity was detected in AcrF-T18: T25-ZO-1 compared to AcrE-T18: T25-ZO-1 (Fig. 7D). Despite these proteins demonstrating a firm interaction with ZO-1, which is a host protein, it is unclear whether the acrEF-deleted Salmonella could reach the tight junctions intracellularly in certain cell types, such as bEND.3 cells, where acrEF deletion conferred a near complete abolishment of invasion.

Fig. 7

The AcrEF interaction with zona occludens 1 (ZO-1) prediction. The in silico analysis performed using ClusPro revealed positive interactions between (A) ZO-1 and AcrE and (B) ZO-1 and AcrF. (C) Bacterial two-hybrid assay between ZO-1 and AcrEF. Escherichia coli BTH101 strains harboring two plasmids (pUT18 and pKT25 derivatives expressing a C-terminal fusion of the cyaA T18 fragment to the acrE or acrF coding region and N-terminal fusions of the cyaA T25 fragment to the ZO-1 or the pKT25 empty vector [negative]) were spotted onto LB plates containing 20 µg/mL X-Gal and 1 mM IPTG and incubated at 30 °C for 40 h. Blue-colored colonies indicate a positive interaction. (D) The average β-galactosidase activities were represented in Miller’s units ± the SD from at least three independent experiments performed in duplicate. Means were compared using Tukey’s multiple comparison test. Significant differences (***p < .001) were assessed in comparison with the negative control. (E) The pAb against AcrEF altered the bacterial invasion of bEND.3 cells in vitro. The antibody halted the internalization of both JOL 401 and JOL 910 when present at 30 µg/mL. The number of localized bacteria in bEND.3 cells. (F) Bacterial colony counting for bEND.3 cells demonstrated a significant reduction of intracellular bacterial load upon treatment with polyclonal antibodies. (G) For the in vivo study, the mice (n = 6/group) were infected intraperitoneally with a lethal dose of Salmonella (JOL 401 or JOL 910) and treated with anti-AcrEF pAb at 2 hpi. The clinical signs were compared in the pAb-treated infected, untreated infected, and healthy control mice. At five dpi, the mice were sacrificed, and bacterial colonization in the brain was evaluated by plating

Furthermore, polyclonal antibodies raised against AcrE and F proteins (Figure S4A, B) could abrogate Salmonella’s ability to cause infection in bEND.3 cells. Fluorescent object counting revealed a clear reduction in ST-WT strain JOL 401 and cpxR mutant strain JOL 910 upon polyclonal antibody treatment. Though these observations are preliminary, acrEF could be a promising therapeutic target for further development. The Salmonella neutralization capability of external exposure to polyclonal antibodies also indicates that these proteins could have surface-exposed portions, despite their known localization within the periplasmic space (Fig. 7E and F), Furthermore, we observed a significant reduction in bacterial loads in challenged my upon salvation treatment using peritoneal injection of anti-acrEF polyclonal antibodies (Fig. 7G). Though these observations require further experimental support, it is clear that the presence of acrEF function is vital for Salmonella survival and virulence.

Localization of AcrE within host cells

The fluorescent images indicate that the AcrE protein potentially occupies the cell periphery, indicating the possibility of interacting with components near tight junctions. If they are near the tight junctions, there is a possibility that these proteins may physically interact with the proteins assembling the tight junctions (Figure S5). However, previous literature supports that Salmonella initiates tight junction disruption at much earlier stages by secreting effector proteins such as SopA, SopB, and SopE, which can interact with proteins such as ZO-1, claudins, and occludin. These observations suggest that Salmonella could influence cellular tight junctions in several different pathways, cumulatively promoting their efficient invasion into the in vivo environment. However, without acrEF functions, these properties are significantly reduced from Salmonella, leaving them unable to enter host-epithelial cells.

Discussion

The CpxA/R system plays a crucial role in Salmonella survival and pathogenicity by sensing stress and regulating gene expression in response to external stimuli [54]. It is involved in multidrug resistance [46] and plays an indispensable role in ST epithelial infection and the successful establishment of infection. Salmonella infection through mucosal surfaces, into the deeper organs requires the crossing of multiple epithelial layers and tolerance against various chemical agents, such as antimicrobial components present in the host cellular environment, and against various antibiotics. During infection, close orchestration of sensory components such as cpxA/R and their gene targets occurs in more complex patterns. Previous research findings have demonstrated that cpxR acts as a negative regulator for SPI-1 and SPI-2 virulence genes that must be regulated in a delicate balance [19, 55]. The absence of cpxR causes a profound increase in virulence genes, and in turn increasingly vulnerable to host defenses, such as susceptibility to oxidative stress. This phenomenon is also evident during infection, where we observed that both acrE and acrF were upregulated in the absence of cpxR function in Salmonella. Interestingly, under normal physiological conditions, cpxR presence or absence did not affect significant expression changes in acrEF protein, yet their influence became prominent during infection (Fig. 1). Since acrEF is a component of the Salmonella drug efflux system, deletion of acrEF, either alone or along with cpxR, significantly increased their susceptibility to β-lactams and tetracyclines, whereas cpxR deletion alone had little effect (Fig. 2A and B).

Comparison of adhesion and invasion in three different cell lines, bEND.3, Caco-2 and Neuro2A revealed an interesting feature, where the lack of cpxR did not cause any significant alteration of adhesion or invasion compared to the wild type strain (Fig. 3A). The addition of acrEF into ST-WT strain JOL 401 or cpxR mutant JOL 910 revealed a marked reduction in adhesion capacity and almost complete impairment of invasion potential into bEND.3 and Caco-2 cells. In Neuro2A cells, both adhesion and invasion remained, yet significantly lower than both the ST-WT strain and the cpxR mutant. These observations suggest the possibility of different entry mechanisms, such as trigger vs. zipper and alternative receptor-based entry features present in Salmonella that are differentially affected by acrEF absence. Herein, we can point out the presence of alternative invasion pathways, for instance, EGFR-driven invasion or binding to heparan-sulfate proteoglycans could be plausible reasons. The Neuro2A cells are also known to express EGFR at high levels, allowing mutants to invade in even absence of cpxR or acrE and acrF gene functions. In turn, Salmonella’s fitness to resist inside macrophage cell lines declined in mutants, justifying their inability to withstand host-mediated defenses (Fig. 2B). These observations suggest a significant role of acrEF in Salmonella virulence and epithelial invasion, other than the known functions of acrEF [56]. To further elaborate on the effect of cpxR and acrEF deletion on virulence gene expression, we evaluated a set of selected genes, namely, avrA, hilD, invF, sipA, sopB, and ompA. The gene expression pattern of all selected marker genes demonstrates both ST-WT strain JOL 401 mirrors the expression pattern demonstrated by JOL 910, the cpxR mutant. The acrEF mutant JOL 2539 mirrors the expression pattern demonstrated by JOL 2555, which is the cpxR and acrEF double mutant. Except for the avrA gene, there was no additive effect observed between JOL 2539 (∆acrEF) and JOL 2555 (∆cpxR ∆acrEF). However, the expression pattern of the selected genes indicates a countering effect between the gene functions of cpxR and acrEF in the Salmonella genome, which could be a reason why these genes were highly expressed in cpxR mutants, except for avrA. These observations highlight that the essential coordination between cpxR and acrEF is an essential factor for the proper expression of SPI-1 virulence genes in Salmonella. By considering the general function of acrEF in Salmonella, we can hypothesize that a dramatic reduction in virulence gene expression could be a result of an accumulation of virulence factors, being unable to secrete out of the bacterial cell due to a lack of efflux functions, causing dysregulation of virulence gene expression. During the early phase of infection, Salmonella injects various effector proteins belonging to SPI-1 and 2, to successfully enter and survive in the host cell. Cytoskeleton remodeling, which occurs during entry stages, demarcates two key mechanisms: the trigger and zipper mechanisms. Herein, the trigger mechanism is largely dependent upon T3SS-1, while the zipper mechanism utilizes an alternative pathway modulated by invasins (Ex, RcK of Salmonella). The Salmonella utilizes either trigger or zipper mechanisms contextually dependent upon the cell type, even in the absence of cpxR and acrEF functions (Fig. 3D).

However, the lack of invasiveness in acrEF mutants should be coming from their inability to modulate SPI-1 and 2 genes as a response to the aforementioned dysregulation. The particular effect was evident in different types of epithelial cells, including Caco-2, Neuro2A, and bEND.3. These observations were somewhat different in macrophage cells, as they have mechanisms to actively engulf bacterial pathogens (Fig. 3D). In addition to cytoskeleton modulation, the lack of invasiveness had affected the induction of inflammatory responses by acrEF mutants and failed to induce apoptotic markers and subsequent cytotoxicity (Fig. 4A and B). These observations were profound in acrEF mutants irrespective of the presence or absence of cpxR, highlighting their significant role in Salmonella virulence, especially associated with invasiveness.

Apart from acrEF’s direct influence on Salmonella virulence gene expression and virulence modulation via SPI-1 and 2 genes, mutants significantly lack their ability to breach epithelial monolayers. Compared to the acrEF mutants, both ST-WT JOL 401 and JOL 910 (∆cpxR) strains actively damaged the tight junction connections between cells (Fig. 5A and B, and C) demarcated by ZO-1 and other tight junction-related protein expression and TEER measurements. Cells treated with acrEF mutants demonstrated increased expression of ZO-1, ZO-2, ZO-3, and some selected OCLN proteins. This could be due to many reasons, for example, a lack of effector proteins (SopA, SopB, and SopE) in acrEF mutants, unable to modulate tight junctions as much as WT ST does (Fig. 5D). Another possibility is the reduced invasiveness of acrEF mutants, unable to enter the host cells for a subsequent virulence outcome. Superficial interaction may be insufficient for significant damage to tight junctions. This inability to invade and breach host cell tight junctions comes at a cost, as acrEF mutants are significantly unable to cause a successful infection in the host. Mice inoculation studies revealed that the ST-WT JOL 401 and JOL 910 (∆cpxR) strains were strong enough to penetrate the blood-brain barrier and caused cerebral infection when inoculated via the IP route. That means the pathogens are capable of breaching multiple barriers and ultimately reaching deeper organs, such as the brain, that are protected by meningeal membranes (Fig. 6A). However, the acrEF mutants failed to confer brain infection, possibly due to their lack of invasiveness, compared to the wild-type strain. These observations were visualized by colony counting using IFA for brain tissues (Fig. 5B). These observations were true for other peripheral organs, too, such as the spleen and liver, which are common infection targets for Salmonella during an infection.

Could bacterial proteins such as acrE or acrF can interact with host cells ' counterparts is a question of concern. In silico predictions (Fig. 7A) and in vitro hybridization studies (Fig. 7C and D) indicate that acrE and acrF can interact with host protein ZO-1. Some bacterial proteins that are located in the periplasmic space can still interact with external proteins by many different mechanisms. For example, certain protein portions can extend beyond the periplasmic space and extrude out of the bacterial cell. Also, proteins can shed their components into the extracellular environment during various cellular processes and can be released upon destruction of the bacterial cell. Under any of these circumstances, periplasmic proteins such as acrE or F can engage and interact with the host cellular targets, such as ZO-1, in the current scenario. We could predict and demonstrate that AcrE and AcrF potentially can interact with ZO-1, which is a major protein in cell tight junctions. The second important concern is that the acrEF mutants significantly lack invasiveness, while retaining a moderate ability to adhere to various cells. This raises the argument that, are these mutants sufficiently internalize into host epithelial cells to cause a firm interaction with host protein targets such as ZO-1. Since the ST-WT strain sufficiently accumulates in host epithelial cells, its acrEF could be capable of interacting with the target protein ZO-1. To obtain an answer for the current question, we developed polyclonal antibodies against ST acrE and acrF proteins and let them interact with ST-WT strain JOL 401 and cpxR mutant JOL 910 before their interaction with bEND.3 cells. A dramatic reduction in internalized bacterial cells indicates that the external blocking of acrEF proteins can interfere Salmonella infection process (Fig. 7E and F). Though this observation requires further research evidence, it was clear that the acrEF could be a potential therapeutic target to fight against Salmonella, especially multidrug-resistant serotypes. These speculations could be justified by the salvage potential of polyclonal antibodies raised against acrE and acrF. Injection into mice challenged with the ST-WT strain JOL 401 and cpxR mutant JOL 910 could significantly lower bacterial numbers reaching the brains of infected mice (Fig. 7G). The ability of the intracellularly expressed acrE protein to localize in the periphery also supports the possibility of these proteins interacting with tight junction-associated proteins.

The observations gathered from this study suggest that acrEF may be partially regulated by the cpxA/R system; however, it has consisted much broader effect on Salmonella’s virulence modulation, particularly associated with invasion of certain epithelial cells. The absence of acrEF dramatically affected invasion into cell types such as bEND.3 and Caco2, while alternative invasion strategies were evident on cell types such as Neuro2A. Due to acrEF’s crucial role in Salmonella virulence, there could be enormous importance for therapeutic developments targeting disruption of Salmonella’s efflux system, making them completely lacking their invasiveness and liable to rapid destruction by the host immune system. Such heights could be achieved by antibody or similar therapeutic interventions and thus could be a commendable avenue for treatment against multidrug-resistant Salmonella and other related species. Hence, the insight gained by this study could be used as a benchmark for further investigations relying on Salmonella’s acrEF axis of virulence.