All bacterial strains, plasmids, and primers used in this study are listed in Tables 1–3, respectively. All Salmonella strains used in this work were derived from the Salmonella enterica serovar Typhimurium 4/74. Unless stated otherwise, bacterial cultures were routinely grown at 37°C for 16 h under dynamic or static conditions in Lysogeny Broth (LB) or on agar plates, respectively. According to manufacturer’s recommendations, Biolog Universal Growth agar with 5% sheep blood was used to grow bacteria for the Biolog Phenotype Microarray. Mueller Hinton Broth (MHB) was used to measure antimicrobial activity. When necessary, ampicillin (Amp, 100 μg/mL) or kanamycin (Km, 50 μg/mL) was added. For lac promoter induction, isopropylthio-β-galactoside was added to a final concentration of 0.5 mM. Cell growth was monitored by measuring the optical density (OD) at 600 nm.

The following xenobiotic stock solutions were used: 5,7-dichloro-8-hydroxyquinaldine [55 mg/mL in dimethyl sulfoxide (DMSO)]; bleomycin (2.56 mg/mL in sterile water); carbenicillin (20.5 mg/mL in sterile water); ceftriaxone (4 mg/mL in sterile water); cetylpyridinium chloride (164 mg/mL in sterile water); chlorhexidine acetate (25.6 mg/mL in ethanol); norfloxacin (0.8 mg/mL in DMSO); phosphomycin (40 mg/mL in sterile water); polymyxin B (1 mg/mL in sterile water); spectinomycin [100 mg/mL in DMSO:water (1:1)]; streptomycin (1 mg/mL in sterile water); sulfamonomethoxine (5 mg/mL in ethanol); thioridazine (16 mg/mL in DMSO); tobramycin (43.2 mg/mL in sterile water); umbelliferone (40 mg/mL in ethanol); and vancomycin (100 mg/mL in sterile water). All xenobiotic solutions were sterilized using 0.22 μm membrane filters and diluted in MHB medium to the appropriate concentration.

In the study, a comprehensive array of open-access bioinformatics tools was utilized to investigate the SanA in S. Typhimurium 4/74. The nucleotide and protein sequences of SanA (accession number CP002487.1: 2277943-2278662; protein ID: ADX17941.1) were extracted from the NCBI database in a FASTA format for the analyses. Orthologs of the SanA across various taxonomic groups were identified using the EggNOG tool, enabling the generation of a report on the prevalence of SanA in different taxa (Huerta-Cepas et al., 2019).

To compare the sequence similarity of SanA between Salmonella and Escherichia coli, BLASTN and BLASTP were employed for nucleotide and protein sequence analysis, respectively. Furthermore, the investigation included an in-depth analysis of subcellular protein localization. For this purpose, several tools such as Phobius, SignalP-5.0, PsortB, THMM 2.0, and TMpred were applied (Krogh et al., 2001; Käll et al., 2007; Yu et al., 2010; Petersen et al., 2011; Finn et al., 2014).

Subsequently, Phyre2 was employed to conduct a comparative analysis of SanA against homologous sequences available in the database (Kelley et al., 2015). Additionally, the Panther classification system was utilized to categorize the protein and predict its function, providing insights into its potential biological roles and activities (Thomas et al., 2003).

Salmonella Typhimurium 4/74 with sanA gene knockout was generated using the protocol described by Datsenko and Wanner (2000), with slight modifications (Datsenko and Wanner, 2000). Initially, electrocompetent cells of the wild-type (WT) strain were transformed with a Red recombinase-carrying plasmid pKD46. The positive clones were further transformed with a kanamycin cassette flanked by FRT sites, which was obtained via polymerase chain reaction (PCR) using primers sanA_del_for and sanA_del_rev on the pKD4 template, and selected on LB agar plates containing kanamycin at 37°C. The FRT flippase present on the pCP20 plasmid was then utilized to eliminate the kanamycin cassette. Colony PCR using locus-specific primer pairs, sanA_upstream_for, sanA_upstream_rev, and sanA_internal_rev was performed to confirm the correct integration and removal of the marker cassette. To determine whether the newly-created strain differed in growth rate or morphology from the parental isolate, growth curves were determined, and acridine orange staining was used to examine them using a fluorescence microscope. Furthermore, the absence of any unintended mutations was confirmed using Next Generation Sequencing.

The sanA from S. Typhimurium 4/74 was amplified using sanA_XbaIpWSK_for, sanA_PstIpWSK_rev primers, and Phusion polymerase (Thermo), according to the manufacturer’s protocol. The PCR products were purified using the GeneJET PCR purification kit (Thermo) and the plasmid DNA was isolated using the GeneJET Plasmid Miniprep Kit (Thermo). To insert sanA into the pWSK29 plasmid, the gene was cloned into the XbaI/PstI digestion sites using the classical ligation method. DNA sequence of the insert was confirmed using colony PCR via the use of a specific primer pair sanA_internal_rev and pWSK_T7_up, and Sanger sequencing. For complementing the deletion mutant, electrocompetent S. Typhimurium 4/74 ΔsanA was transformed with a plasmid carrying complementing gene as well as pWSK29 vector plasmid alone (without insert) as a control. All clones were analyzed in positive selection on LB agar with ampicillin.

To determine the growth curves of Salmonella strains, a single bacterial colony of each isolate was inoculated in LB and incubated overnight at 37°C with shaking (180 rpm). The resulting cultures were diluted to OD₆₀₀ = 0.05 using LB and incubated until the early logarithmic growth phase (OD₆₀₀ = 0.5, 37°C, 220 rpm). Each culture was then centrifuged, washed, and suspended in 0.9% NaCl solution. The OD₆₀₀ values were measured, and the cultures were diluted in LB to obtain 5 × 10⁶ CFU/mL bacterial suspensions. For determining the antimicrobial effect of vancomycin and bile salts, the assay was performed in LB or MHB medium, respectively with 0%–15% bile salts and 0–500 μg/mL vancomycin. The samples were then applied to a polystyrene or polypropylene 96-well plate in triplicate and incubated in a spectrophotometer (Tecan) at 37°C with measurements taken at 15-min intervals for 16 h, with shaking before each reading. The experiment was performed in at least three independent biological replicates, and dilution series on LB agar were prepared to verify initial bacterial concentrations.

The susceptibility of mutant and the parental strain to 240 chemical compounds was determined in three independent experiments using the Phenotype MicroArray (PM) PM11-PM20 (Biolog), as described in a previous study (Shea et al., 2012). Briefly, strains were grown overnight on Biolog Universal Growth agar with 5% sheep blood at 37°C, colonies were then picked using a sterile cotton swab and suspended in 15 mL of 1× inoculation fluid (IF-0a GN/GP Base, Biolog). The cell density was adjusted to 85% transmittance (T) using a Biolog turbidimeter. The inoculation fluid for PM11-20 was prepared by mixing 100 mL of IF-10a GN Base (1.2X; Biolog), 1.2 mL of Biolog Redox Dye A (100X; Biolog), 0.6 mL of cell suspension at 85% T, and sterile water to reach a final volume of 120 mL. The mixture was then inoculated in the PM plates (100 μL per well) and color development was monitored every 15 min for 48 h at 37°C using an Omnilog reader (Biolog). The kinetic curves of both strains were compared using Omnilog-PM software to identify the phenotypes. Raw data were obtained for 10 plates, which included 240 antibiotics arranged as a dilution series across four wells (960 wells in total). Data were recorded in the RA format and filtered using differences of average height with standard thresholds to identify statistically significant differences using Student’s t-test (Guard-Bouldin et al., 2007). The reproducibility of our results was ensured by excluding any differences greater than 50 Omnilog units between biological replicates from the analysis (see Supplementary Table S1).

As the manufacturer of PM plates (Biolog) does not disclose the concentrations of compounds in their plates, on the basis of available literature, we selected 8 different concentrations for all compounds by making two-fold dilutions. Polypropylene and polystyrene plates were utilized for cationic and anionic compounds, respectively. Bacterial strains were incubated in LB for 16 h at 37°C, 180 rpm. Further, the OD₆₀₀ was determined, and bacteria were diluted in MHB to get a total density of 10⁶ CFU/mL. The suspensions were aliquoted at 50 μL per well into previously prepared 96-well plates and incubated at 37°C. After 16 h, OD₆₀₀ of each well was measured with using the Tecan microplate reader (Spark®). MHB without xenobiotics serves as a positive control of growth. At least three technical and biological repetitions were performed for each strain. All xenobiotics and tested concentration ranges are listed in Table 4. Antibiotic susceptibility testing results determine the fold change of OD₆₀₀ values between the ΔsanA and WT strain at concentrations showing a significant difference. The standard error of mean (SEM) was calculated using the standard deviation of the sample and the square root of the sample size.

The OM permeability was investigated by utilizing the influx of either the cationic Ethidium Bromide (EB) or the neutral Nile Red dye (NR; Murata et al., 2007; Viau et al., 2011). Overnight bacterial cultures were diluted to OD₆₀₀ = 0.05 using LB, incubated until early stationary growth phase (OD₆₀₀ = 2.0, 37°C, 220 rpm) and rinsed twice with assay buffer (50 mM KH₂PO₄, 137 mM NaCl, pH 7.0). To perform dye uptake assays, the proton motive force inhibitor carbonyl cyanide-m-chlorophenylhydrazone (CCCP) was added at a final concentration of 10 μM. Fluorescence was measured for 30 min at 1-min intervals using a Tecan microplate reader (Spark®) immediately upon mixing cells (final OD₆₀₀ = 0.2) with EB at a final concentration of 6 μM (with excitation at 545 nm and emission at 600 nm) or NR at a final concentration of 2 μM (with excitation at 540 nm and emission at 630 nm). Membrane permeability was measured in at least three independent experiments. According to Murata et al., the dye uptake rates of different strains varied between experiments, but the pattern of dye uptake remained consistent across repetitions (Murata et al., 2007).

Bacterial surface hydrophobicity was determined using the hexadecane adhesion assay (Oguri et al., 2016). Overnight bacterial cultures were diluted to OD₆₀₀ = 0.05 using LB and incubated until early stationary growth phase (OD₆₀₀ = 2.0, 37°C, 220 rpm). Subsequently, the cultures were harvested, washed twice with phosphate buffered saline (PBS), and resuspended in 1 mL of PBS. Following this, 100 μL of cells were diluted 10× in PBS and the OD₆₀₀ was measured (C₀). Next, 900 μL of the cell suspension was mixed with 200 μL of hexadecane (Merck Millipore), vortexed for 1 min, and left undisturbed at room temperature until the phases separated. Cell samples (100 μL) from the lower, aqueous phase were then diluted in 900 μL PBS and OD₆₀₀ was measured (C_H). The percentage of hexadecane adherence was determined in three independent experiments, using the following formula: % hexadecane adherence = [(C₀ – C_H)/C₀] × 100.

The cytochrome c binding assay was performed as described previously, with minor modifications (Kristian et al., 2005). Briefly, overnight bacterial cultures were diluted to OD₆₀₀ = 0.05 using LB and incubated until early stationary growth phase (OD₆₀₀ = 2.0, 37°C, 220 rpm). Bacteria were then collected, washed twice in 3-(N-morpholino)propanesulfonic acid (MOPS) buffer (20 mM, pH 7.4), and adjusted to a final OD₆₀₀ = 7.0 in the same buffer. Next, bacteria were mixed with cytochrome c (Merck Millipore) to a final concentration of 0.5 mg/mL, incubated for 10 min at room temperature, and centrifuged at 18,000 × g for 6 min. Cytochrome c without bacteria in the same buffer was also incubated as a negative control. The cytochrome c contents in the supernatants were measured at the absorption maximum of the prosthetic group (530 nm). The percentage of bound cytochrome c was calculated from three independent experiments, each performed in triplicate.

The isolation of primary bone marrow-derived macrophages (pBMDMs) was performed in accordance with a UK Home Office Project License in a Home Office designated facility, as previously described (Thurston et al., 2016; Bailey et al., 2020). Briefly, bone marrow was obtained from 6 to 8 week-old female C57BL/6 mice (Charles River) by flushing the tibias and femurs. The collected cells were then added to non-tissue culture-treated petri plates at a concentration of 3 × 10⁶ cells per plate in 8 mL Dulbecco’s modified Eagle’s medium (DMEM) with high glucose supplemented with 20% (v/v) L929-MCSF supernatant, 10% (v/v) fetal bovine serum (FBS), 10 mM HEPES, 1 mM sodium pyruvate, 0.05 mM β-mercaptoethanol, and 100 U/mL penicillin/streptomycin. After 3–4 days, 10 mL fresh medium was supplemented and the differentiated BMDMs were harvested on day 7. The macrophages were then seeded into 24-well tissue culture-treated plates at a concentration of 2 × 10⁵ macrophages per well and infected the following day with DMEM media supplemented with the above concentrations of FBS, HEPES, sodium pyruvate, and β-mercaptoethanol but without antibiotics.

The macrophage monolayer was infected with stationary phase bacteria opsonized in mouse serum for 20 min at room temperature at a multiplicity of infection of 10:1. To synchronize the infection, the culture plates were centrifuged for 5 min at 165 × g, followed by a 30-min incubation at 37°C (5% CO₂). Fresh DMEM supplemented with 100 μg/mL gentamicin (Gm) was added to kill extracellular bacteria, and the macrophage monolayers were incubated with added Gm for 90 min (Monack et al., 1996). After washing with DMEM, the monolayers were lysed in 1% Triton X-100 and diluted with PBS. Dilutions of the suspension were then plated on LB agar to quantify the number of viable bacteria. To evaluate intracellular growth, the medium containing 100 μg/mL Gm was replaced with DMEM supplemented with 10 μg/mL Gm, and parallel cell cultures were examined for viable bacteria 24 h following infection (Monack et al., 1996).

Statistical analyses were performed using GraphPad Prism (GraphPad Software, Inc., La Jolla, CA, United States). The Shapiro–Wilk normality test was used to determine data distribution. Depending on the data distribution, either Student’s t-test, two-way ANOVA analysis of variance with Tukey’s correction, or the Kruskal–Wallis test with Dunn’s multiple comparison post-hoc test was used. For each condition, data were collected from at least three independent experiments. p ≤ 0.05 was considered statistically significant. The results were presented as mean ± SEM. The symbols ^*p < 0.05; ^**p < 0.01; ^***p < 0.001; and ^****p < 0.0001 were used to indicate significance levels.

According to our analysis of the eggNOG database, we found that the sanA is present across a diverse range of bacterial taxa. Specifically, the gene was identified in both, gram-negative and gram-positive bacteria with a prevalence of 44.1% in species classified under Gammaproteobacteria, 17.2% of Bacteroidetes, 18.9% of Actinobacteridae, 5.72% of Clostridia, and 3.37% of Spirochaetia. Additionally, all SanA homologs have an unknown domain, DUF218. We performed a BLAST comparison and found that SanA in E. coli and Salmonella share 94% identity, with an estimated 97% amino acids having identical or similar chemical properties, suggesting its high conservative among bacteria.

The predictions for the subcellular localization were inconsistent: Phobius suggested a location outside the cytoplasm, SignalIP-5.0 detected no signal peptide implying a cytoplasmic protein, PsortB indicated a cytoplasmic position, while both THMM 2.0 and TMpred identified a transmembrane domain. Since proteins with similar structures can have similar functions, we elucidated the potential function of SanA by predicting its structure using Phyre2 and comparing it to homologous sequences. Our findings suggested that SanA shares structural similarities with the YdcF protein of E. coli, which is involved in binding S-adenosyl-L-methionine; transferases (5-methyltetrahydrofolate homocysteine s-methyltransferase); OmpA like protein; peptide binding protein; membrane protein, and structural protein. Moreover, the Panther classification system revealed that SanA is an IM protein with potential permease activity and is classified into the transporters group.

As SanA is known as a vancomycin exclusion protein, the first stage of our investigation incorporated analysis of the WT and ΔsanA mutant bacteria growth in the presence of vancomycin or bile salts. Surprisingly, the mutant strain showed higher resistance to vancomycin than the WT (Figure 1A). While the WT grew only up to 125 μg/mL vancomycin, sanA deletion allowed the strain to grow up to a concentration of 250 μg/mL vancomycin. Moreover, a significant difference in the optical density between the two strains was observed in the presence of 62.5 and 125 μg/mL vancomycin, and the highest contrast was visible in the stationary growth phase—after 10 h (Figure 1A). Additionally, the two strains displayed contrasting growth patterns in the presence of bile salts. The deletion mutant strain demonstrated decreased resistance with growth up to only 3.75% bile salts compared to the WT, which grew up to 7.5% bile salts. Significant growth variations were also noted between the two strains at bile salt concentrations of 0.47%–1.88% (Figure 1B). The phenotypic parallels observed between the strain complemented with sanA and the one transformed with the empty pWSK29 plasmid further underscore the function of SanA in these resistance profiles (Supplementary Figure S2).

WT and ΔsanA were further characterized using Biolog (Biolog®) phenotypic arrays to investigate potential gene knockout-induced changes in resistance profiles. The arrays featured various compounds with some known antimicrobials included on plates PM11a to PM20. Prior to testing, no significant differences in growth kinetics were observed among the strains (Supplementary Figure S1).

Our results revealed distinct resistance patterns for more than 20% (49/240) of the analyzed compounds with different mechanisms of action (p < 0.05; Table 5). The ΔsanA strain exhibited improved growth in the presence of approximately 35% (17/49) of these 49 agents, grouped mostly as cell wall-and DNA-associated antibiotics. The same strain demonstrated lower resistance to folate antagonists (sulfonamides), membrane-targeting antibiotics, and DNA and protein-associated antibiotics (fluoroquinolones, glycopeptides, nucleic acid analogs; Table 5).

Considering the known adjustments of the PM plates method with MIC measurements, we cross-checked PM results using a microbroth dilution assay. We focused on 16 compounds with the most significant differences between strains, particularly those targeting the membrane and cell wall (Table 5). The results, showing fold changes at OD₆₀₀ between the ΔsanA and WT at specific concentrations (Supplementary Figure S3), led to further investigation using complemented strains (Figure 2B).

In the presence of 16 different agents, the ΔsanA strain exhibited reduced resistance to ten xenobiotics but displayed increased resistance to six others (Figure 2A). We grouped these agents into the following five categories: (1) membrane, (2) protein synthesis, (3) cell wall and efflux pumps, (4) replication, and (5) transcription (Figure 2A). The sanA knockout resulted in compromised growth with certain membrane-associated xenobiotics such as chlorhexidine acetate (2 μg/mL; p = 0.0003), cetylpyridinium chloride (6.5 μg/mL; p = 0.0093), umbelliferone (500 μg/mL; p = 0.0465), and polymyxin B (0.625 μg/mL; p = 0.0205). Notably, reintroducing sanA restored the resistance pattern on these agents to resemble WT bacteria (Figure 2B).

The ΔsanA exhibited lower resistance to agents targeting protein synthesis, such as tobramycin (0.675 μg/mL; p = 0.0070), streptomycin (100 μg/mL; p = 0.0152), and spectinomycin (31.250 μg/mL; p = 0.0359), as well as those targeting transcription, such as norfloxacin (0.010 μg/mL; p = 0.0052) and 5,7-dichloro-8-hydroxyquinaldine (3.5 μg/mL; p = 0.0065; Figure 2A). Notably, the ΔsanA strain showed a different resistance pattern to protein synthesis agents than that suggested by PM. However, when the mutation was complemented, the resistance pattern was mostly attributed to the sanA deletion, except for tobramycin (0.675 μg/mL; p = 0.9998; Figure 2B).

As anticipated, sanA deletion resulted in greater resistance to cell wall and efflux pumps associated compounds, like ceftriaxone (0.002 μg/mL; p = 0.016), vancomycin (125 μg/mL; p = 0.0044), carbenicillin (3.200 μg/mL; p = 0.0281), and thioridazine (1,600 μg/mL; p = 0.027). Surprisingly, this strain showed reduced resistance to phosphomycin (9.900 μg/mL; p = 0.0483; Figure 2A). Complementation mostly restored WT phenotypes, with the exception of thioridazine (1,600 μg/mL; p = 0.2196). In contrast to the PM data, ΔsanA demonstrated reduced susceptibility to replication agents, such as bleomycin (0.400 μg/mL; p = 0.0104) and sulfamonomethoxine (248 μg/mL; p = 0.0432). This phenotype was further validated by complementing sanA, highlighting its critical role in this phenotype (Figure 2B).

The impact of sanA on resistance to vancomycin along with other antimicrobial agents suggests a general effect on membrane integrity rather than specific vancomycin sensitivity. Thus, OM permeability was determined by measuring influx of the cationic dye, EB or the neutral dye, NR. Dye uptake assays were performed in the presence of CCCP, which enables the inward transport of H⁺ across lipid membranes. Therefore, it prevents the efflux of compound by active pumps, so that only passive permeability is measured. In the experiment without CCCP, a minimal increase in dye uptake was noted, suggesting that the increased retention of EtBr in the ΔsanA is not due to pump inactivation, but increased membrane permeability (Supplementary Figure S4).

When CCCP was present, the ΔsanA bacteria demonstrated a notably higher OM permeability baseline than the WT for both EB and NR, with a remarkably increased rate of dye uptake observed particularly after approximately 10 min of assay initiation (Figure 3; Supplementary Figure S5). Importantly, complementation of the mutation restored the WT phenotype for NR and further reduced the permeability for EB, strongly suggesting that the observed phenotype was primarily due to sanA deletion. Moreover, the permeabilities differed between WT and ΔsanA, as well as between ΔsanA-pWSK29 and its complemented ΔsanA-pWSK29-sanA counterpart, consistently throughout the assay (Figure 3).

Considering the distinct resistance phenotype observed for different groups of xenobiotics, we hypothesized that the surface charges and hydrophobicity of bacterial cells could be contributing factors. To investigate potential alterations in the surface properties of the ΔsanA, we performed the following two assays: (1) determining surface charges by evaluating the binding of the cationic protein cytochrome c to bacterial cells, and (2) determining surface hydrophobicity by measuring the adherence of cells to the hydrophobic solvent hexadecane.

The ΔsanA cells displayed significantly lower affinity to cytochrome c (80%) than that of WT cells (90%; p = 0.0046; Figure 4A). Additionally, approximately 10% of ΔsanA cells adhered to hexadecane, in contrast to approximately 16% of WT cells (p = 0.0256; Figure 4B). This suggests a decrease in the negative charge and hydrophobicity on the cell surface of ΔsanA mutant, respectively. Moreover, all these changes were attributed entirely to sanA, as introduction of sanA to the deletion mutant restored the WT phenotype (p = 0.0199; p = 0.005).

Alterations in bacterial membranes significantly influence antimicrobial efficacy and bacterial replication within phagocytes (Ernst et al., 1999). To further investigate this, we monitored Salmonella replication in primary BMDMs, which provide a relevant physiological context to examine the interactions between Salmonella and host cells. Our results showed that the uptake of S. Typhimurium by BMDMs was similar for the WT and ΔsanA (p = 0.0572; Figure 5A), but the mutant exhibited a significantly increased number of intracellular bacteria 24 h post-infection (p = 0.0051; Figure 5B). Furthermore, we observed a marked difference between ΔsanA + pWSK29 and ΔsanA + pWSK29-sanA, whereby expression of sanA reduced replication (p = 0.0351; Figure 5B).