PhoPQ Regulates Quinolone and Cephalosporin Resistance Formation in Salmonella Enteritidis at the Transcriptional Level

ABSTRACT The two-component system (TCS) PhoPQ has been demonstrated to be crucial for the formation of resistance to quinolones and cephalosporins in Salmonella Enteritidis (S. Enteritidis). However, the mechanism underlying PhoPQ-mediated antibiotic resistance formation remains poorly understood. Here, it was shown that PhoP transcriptionally regulated an assortment of genes associated with envelope homeostasis, the osmotic stress response, and the redox balance to confer resistance to quinolones and cephalosporins in S. Enteritidis. Specifically, cells lacking the PhoP regulator, under nalidixic acid and ceftazidime stress, bore a severely compromised membrane on the aspects of integrity, fluidity, and permeability, with deficiency to withstand osmolarity stress, an increased accumulation of intracellular reactive oxygen species, and dysregulated redox homeostasis, which are unfavorable for bacterial survival. The phosphorylated PhoP elicited transcriptional alterations of resistance-associated genes, including the outer membrane porin ompF and the aconitate hydratase acnA, by directly binding to their promoters, leading to a limited influx of antibiotics and a well-maintained intracellular metabolism. Importantly, it was demonstrated that the cavity of the PhoQ sensor domain bound to and sensed quinolones/cephalosporins via the crucial surrounding residues, as their mutations abrogated the binding and PhoQ autophosphorylation. This recognition mode promoted signal transduction that activated PhoP, thereby modulating the transcription of downstream genes to accommodate cells to antibiotic stress. These findings have revealed how bacteria employ a specific TCS to sense antibiotics and combat them, suggesting PhoPQ as a potential drug target with which to surmount S. Enteritidis.

Salmonella infections (3). However, the prevalence of S. Enteritidis isolates with resistance to multiple antimicrobials, including quinolones and cephalosporins, makes this serotype a serious public health concern (4-7). Resistance to quinolones is commonly mediated by mutations that reduce the affinity of quinolones to topoisomerase (8). Cephalosporin resistance is usually achieved by b-lactamases that degrade and inactivate cephalosporins (9). In addition, the genes encoding the multidrug efflux pump and the outer membrane (OM) porin also contribute to resistance by controlling the intracellular levels of drugs (9)(10)(11). Despite a set of resistance determinants being identified, it is still unclear how these determinants are regulated by Salmonella to deal with the stress posed by quinolones and cephalosporins.
Two-component systems (TCSs) are among the key elements through which bacteria sense and respond to environmental changes (12). A canonical TCS consists of an inner membrane imbedded histidine sensor kinase (HK) and a cytosolic response regulator (RR). When HK is activated by environmental stresses, it auto-phosphorylates and then transfers phosphoryl groups to its cognate RR for activation (13). The activated RR binds specifically to the promoters of target genes to regulate their transcription, thereby initiating cellular responses (14). A series of TCSs from various species have been reported to be associated with antibiotic resistance, including aminoglycoside and b-lactam resistance-associated CpxAR (15), drug efflux-associated EvgSA (16), b-lactam resistance-associated VbrKR and CreBC (17,18), as well as polymyxin B resistance-associated PmrAB and PhoPQ (19). Collectively, TCSs represent a mechanism that is favorable for bacteria to survive under adverse environments.
PhoPQ is a conserved TCS that is involved in low-magnesium adaptation, virulence expression, and polymyxin resistance in Gram-negative pathogens (20). PhoQ, the dimeric HK of this system, contains a periplasmic sensor domain, a transmembrane (TM) domain, and three cytosolic domains. The dimeric sensor domain and (or) the TM domain adopt different conformations after sensing stimuli, such as low divalent cations (21), an acidic pH (22), an osmotic upshift (23), or the presence of cationic antimicrobial peptides (CAMPs) (24), to yield an autophosphorylated PhoQ. The phosphorylated PhoQ leads to a phosphorylated RR PhoP that binds to the promoters of target genes at the heptanucleotide sequence (G/T)GTTTA(A/T), thereby enabling coordinated changes in global gene expression profiles (25,26). The regulons triggered by PhoPQ vary in a species-dependent manner among Salmonella Typhimurium (S. Typhimurium), E. coli, Pseudomonas aeruginosa (P. aeruginosa), Shigella flexneri, Yersinia pestis, and Mycobacterium tuberculosis (27)(28)(29)(30)(31)(32)(33)(34). In our previous work, functional defects in PhoPQ in various S. Enteritidis isolates led to a frequent 2 to 8-fold decrease in minimal inhibitory concentration (MIC) values against quinolones and cephalosporins (35). However, the regulatory mechanism underlying the PhoPQ-mediated antibiotic resistance in S. Enteritidis remains poorly understood.
Here, we reported a previously unidentified regulatory role of the PhoPQ system that allows for the recognition of antibiotics and governs the quinolone and cephalosporin resistance pathways in S. Enteritidis. Via a strand-specific RNA-sequencing (RNA-seq) analysis, we assessed the impact of PhoPQ on the global transcription profile of S. Enteritidis treated by nalidixic acid and ceftazidime, which are a representative quinolone and cephalosporin, respectively. Notably, we identified the critical genes and progress pathways regulated by PhoPQ and demonstrated that nalidixic acid and ceftazidime molecules could be sensed by directly binding to PhoQ. This work will contribute to a better understanding of antibiotic resistance mechanisms as well as provide a basis upon which to combat antibiotic resistance in S. Enteritidis.

RESULTS
A comparative transcriptomic analysis reveals the differentially expressed genes regulated by PhoP in S. Enteritidis. To test the regulatory roles of S. Enteritidis PhoPQ in its global gene transcription to combat nalidixic acid (NAL) and ceftazidime (CAZ), the differentially expressed genes (DEGs) in a phoP deletion mutant, compared with the wild-type (WT), were identified via RNA-seq analysis. It was found that a total of 90 upregulated and 97 downregulated DEGs were identified under NAL pressure, PhoPQ Regulates Quinolone and Cephalosporin Resistance mBio whereas 645 upregulated and 659 downregulated DEGs were identified under CAZ pressure (Fig. 1). The RNA-seq data were verified via reverse transcription-quantitative PCR (RT-qPCR) assays (Fig. S1), which showed that the consistency levels between the RT-qPCR and RNA-seq results were 100% and 93.75% under NAL and CAZ conditions, respectively (Fig. S1).
To gain insights into the metabolic pathways that are influenced by PhoP, the identified DEGs were assigned to functional classes and COG categories. It was demonstrated that 187 DEGs under NAL pressure were related to the metabolism, membrane, transcription, genomic islands, innate immunity/virulence, and conjugal transfer categories (Fig. 1A). In comparison, the CAZ pressure led to an increased number of DEGs in association with the metabolism and membrane categories as well as those associated with translation and motility, which were not observed under NAL pressure (Fig. 1B). These DEGs were generalized into three strikingly regulated aspects: (i) innate immunity defense, (ii) substrate transport and metabolism progress, and (iii) cell envelope homeostasis.
A KEGG cluster analysis ( Fig. 1C and D) revealed an array of significantly changed pathways under NAL pressure, including amino acid/carbohydrate/cofactor and vitamin metabolism, two-component systems, bacterial secretion systems, CAMP resistance, quorum sensing, and lipopolysaccharide (LPS) biosynthesis. In addition, CAZ pressure resulted in significant changes in the pathways related to carbohydrate metabolism, signal transduction, drug resistance, lipid metabolism, etc. Intriguingly, it was found that some previously unidentified pathways and genes were involved in DNA replication and repair, ABC transporters, peptidoglycan synthesis, and membrane transport. Here, we integrated the correlative pathways and genes under NAL/CAZ pressures, and we analyzed the largely affected DEGs in depth. These DEGs were responsible for envelope integrity, the osmotic stress response, and the redox balance.
PhoP regulates the envelope integrity in S. Enteritidis. The expression profiles of DEGs related to bacterial envelope integrity, including LPS modification, fatty acid synthesis, and degradation pathways, peptidoglycan synthesis, cross-linking and degradation, and membrane-related innate proteins were analyzed in this study. The transcription of the genes involved in LPS modification in DphoP was virtually downregulated under both antibiotic pressures ( Fig. 2A). This is exemplified by the UDP-glucose 6-dehydrogenase gene (ugd) and the lipid A hydroxylase gene (lpxO), the products of which modify lipid A phosphates with 4-amino-4-deoxy-L-arabinose or hydroxylation acyl chains. In addition, the transcription of the PmrAB-targeted arn operon (aminoarabinose), eptA (phosphoethanolamine), and pagP (palmitoyltransferase) that function to stabilize the membrane by producing more negatively charged LPS was also downregulated, implying interrupted LPS in DphoP. The results from a silver staining assay of membrane LPS extracts supported the RNA-seq data, meaning that the deletion of phoP resulted in a decrease in production of LPS with fewer and shallower stained ladders and that this deficiency was exacerbated under the pressure of NAL or CAZ (Fig. 3A). These data suggested positive regulatory impacts of PhoPQ on LPS production and modification.
Under CAZ pressure, the transcription of the genes involved in fatty acid synthesis (yciA, fabA, ybgC, and acpP, etc.) and phospholipid synthesis (plsC) was significantly upregulated (1.04 to 3.71-fold), whereas those related to cyclopropane fatty acyl phospholipid synthesis (cfa) and fatty acid degradation (adhP, adhE, and frmA) were significantly downregulated (1.08 to 1.99-fold) (Fig. 2B). The consistent variations in the expression of these genes, albeit to a less significant degree, were also observed under NAL pressure (Fig. 2B). These identified DEGs hinted at a fluctuant composition of membrane fatty acids in DphoP, and this was supported by the fatty acids analysis, in which DphoP exhibited significant a decrease in the saturated fatty acid (SFAs) C18:0 and an increase in the unsaturated fatty acids (UFAs) C14:1n5 and C17:1n7, with a significantly higher ratio of UFAs to SFAs (UFAs/SFAs) (51.49% to 60.41%) than that observed in the WT (39.92% to 45.12%) ( Fig. 3B and C). The UFAs/SFAs ratio was a predominant parameter with which to evaluate membrane fluidity in bacteria, and bacterial cells with lower membrane fluidity usually exhibited higher resistance to environmental stresses (36). Hence, the altered fatty acids with a higher UFAs/SFAs ratio in DphoP might, to some degree, explain the increased susceptibility to quinolones and cephalosporins. The deletion of phoP led to the downregulation of the genes involved in peptidoglycan (PG) synthesis and cross-linking (Fig. 2C), including the glycogen synthase operon glg, which is responsible for the elongation of glycogen, the D, D-transpeptidase gene vanX for the primary cross-linking of PG with D-Ala 4 -meso-DAP 3 , the L, D-transpeptidase genes ldtA, ldtD, and ynhG for the meso-DAP 3 -meso-DAP 3 cross-links (37)(38)(39), and the glycosyltransferase gene mgtA, which is involved in PG polymerization. Additionally, the PG hydrolysis-related mltD, mltF, mepS, ampG, yafK, and DXN21-RS00390 genes were upregulated (1.08 to 1.55-fold) under CAZ pressure, with the corresponding variations, albeit to a lesser degree, being observed under NAL pressure (Fig. 2C). These results indicated that the deletion of phoP interferes with PG synthesis and exacerbates PG hydrolysis, which might lead to a poorly cross-linked PG layer with weakened mechanical strength. The scanning electron microscopic (SEM) results showed that a considerable number of lysed cells were found in DphoP in the presence of NAL or CAZ (Fig. 3D). Of note, the WT cells retained their characteristic rod shape, with an average length of 1.6 mm (ranging from 1.35 to 2.25 mm for 20 cells examined), under both treatments, whereas the DphoP cells were largely elongated (with the length being hard to examine) under the CAZ treatment, suggesting uncoordinated elongation and division. Given the role of coordinated division with PG synthesis-mediated elongation in cell shape determination (40), the downregulated cell division-associated genes zapA, murG, ftsL, ftsW, and lpoB, as well as the upregulated Tol-Pal system (which is required for proper PG remodeling at the division site) genes ybgC and tolQ, might be responsible for the abnormal cell morphologies (Fig. S2).
In DphoP, cell membrane-related genes, including the glucan transportation gene bcsC, the envelope stress response gene nlpE, the OM-peptidoglycan cross-linking gene lpp, and the efflux pump genes arcA, macB, and mdt operon were downregulated, whereas the OM porin genes ompF, ompD, and ompL were upregulated (Fig. 2D), implying more admission to drugs through OM porins, accompanied with inefficiencies in expelling them via efflux pumps. The NPN (1-n-phenylnapthylamine) uptake assays also supported the role of PhoPQ in the limitation of envelope permeability under antibiotic pressure, as the deletion of phoP rendered the cells exposed to CAZ/NAL with a stronger NPN fluorescence intensity (Fig. 3E). Altogether, the PhoP regulator functions in the maintenance and modification of the bacterial envelope by transcriptionally regulating multiple associated genes.
PhoP regulates the osmotic stress response genes in S. Enteritidis. The RNA-seq data revealed that the ABC transporter pathway is regulated by PhoP. Under NAL/CAZ pressure, the deletion of phoP led to the transcriptional downregulation of the genes that encode K 1 transport (kdpA), osmoprotectants (proP), and ABC-type osmotically inducible proteins (osmV, osmX, osmY, and osmW) as well as the upregulation of the genes that encode Na 1 -Ca 21 /H 1 antiporters (nhaB, gntU, and DXN21_RS00200) (Fig. 2E), of which the upregulated antiporters would cause a low intracellular accumulation of metal ions. In addition, under CAZ pressure, the genes that encoded osmotically inducible proteins (osmB and osmE) and were involved in trehalose synthesis and utilization (otsA, otsB, treA, treF, and DXN21_RS05875) were also downregulated ( Fig. 2E). Gram-negative bacteria usually accumulate K 1 , nonpolar and facultative ionic solutes, trehalose, and glycine betaine to tackle osmotic stress (41). Therefore, PhoP activation, mediated by exposure to antibiotics, may improve the transport of substrates via the regulation of the genes that are related to osmotic stress responses.
PhoP regulates the redox balance in S. Enteritidis. The genes that are responsible for the redox balance were shown to be differentially expressed in DphoP and WT under both NAL and CAZ pressure, including reactive oxygen species (ROSs) formation and scavenging, quinone synthase, Fe-S cluster assembly, the iron ion ABC transporter, and the NADH-related electron transport chain ( Fig. 2F and G). In the DphoP cells, the genes of nadB and fumA were upregulated, whereas the genes related to quinone oxidoreductase (wrbA), menaquinone synthase (menC), ubiquinone synthase (ubiC), and cytochrome bd oxidase (cydABX) were downregulated (Fig. 2F), implying the accumulation of H 2 O 2 in cells lacking PhoP. The genes responsible for the Fe-S cluster assembly (sufABCDES) were downregulated, whereas the iron ion ABC transporter genes (fepBCDG and fhuACF) were upregulated, implying an increase in free iron that was readily oxidized by H 2 O 2 to trigger the Fenton reaction (Fig. 2F). This was supported by the significantly higher ROS levels that were observed in the DphoP cells that were exposed to both antibiotics (Fig. 3F).
The decreased transcription of the catalase (CAT) genes katE and katG, the superoxide dismutase (SOD) genes DXN21_RS23620, sodB, and sodC, which function to scavenge intracellular H 2 O 2 or superoxide, as well as the decreased transcription of dps, whose product acts to mineralize and store iron ions in a bioavailable and nontoxic form and protect DNA during oxidative stress, were observed in the DphoP cells that were exposed to both antibiotics (Fig. 2G). These DEGs indicated that PhoP has a role in antioxidation pathways, and this result was supported by lower CAT activities and SOD enzymes being observed in the DphoP cells that were exposed to antibiotics ( Fig. 3G and H). Although the DphoP cells that were exposed to NAL had higher SOD activities, this was presumably due to the complementary effects of the total SOD enzymes, other than SodB and SodC.
An equivalent cofactor NADH that functions as a distinct electron carrier can be oxidized to NAD 1 to promote the redox balance. In this study, the upregulated NAD 1 biosynthesis genes nadB and pncB as well as the downregulated ammonia-dependent NAD 1 synthetase gene nadE under CAZ pressure suggested a role of PhoP in NAD 1 production, even though the transcriptional variations in these genes were not significant when exposed to NAL (Fig. 2G). The role of PhoP was supported by the significantly decreased NAD 1 /NADH ratio in the DphoP cells that were exposed to antibiotics (Fig. 3I), indicating that the deletion of phoP disturbed the cellular redox balance. Additionally, the NADH dehydrogenase II-encoding ndh that functions to convert NADH to NAD 1 was also downregulated, which might lead to the accumulation of NADH and was partially responsible for the decreased NAD 1 /NADH ratio. Taken together, these data suggested that PhoP regulates cellular redox homeostasis and limits the accumulation of ROS in response to antibiotic stress, which helps S. Enteritidis combat the antibiotics.
PhoP directly regulates the ompF and acnA genes in S. Enteritidis. To identify the genes directly regulated by PhoP in S. Enteritidis, the putative PhoP-binding motifs Typhimurium and E. coli were mapped to the promoter regions of 1,481 DEGs (1,304 and 187 for CAZ and NAL, respectively). As a result, 68 PhoP-targeted candidates were screened (Fig. S3). Eight of them (vanX, mgtA, mgtC, slyB, pagP, sseL, pgtE, and yoaE) were reported to be PhoP-targeted genes in previous studies (25,33,42,43). To figure out the PhoPQ-targeted genes that were responsible for antibiotic resistance, genes involved in envelope homeostasis (osmY, cfa, ompF, yoaE, and lpp) and carbon/nucleotide metabolism (csrA, acnA, and rbsK) were selected for the following tests. The His-tagged PhoP protein was expressed in E. coli (Fig. S4), and its ability to bind to the promoters of the candidate genes was tested via electrophoretic mobility shift assays (EMSAs). As a result, the addition of phosphorylated PhoP (PhoP-P) blocked the mobility shift of the fragments upstream of ompF and acnA in a concentration-dependent manner ( Fig. 4A and B). These fragments were a 221-bp FAM-labeled ompF promoter probe (P ompF -FAM) and a 231-bp acnA promoter probe (P acnA -FAM), suggesting a capacity of PhoP-P to bind to the promoter of ompF and acnA. Additionally, this binding was specific, as it was completely abolished by a 50Â unlabeled ompF or acnA promoter probe (P ompF and P acnA ) ( Fig. 4A and B, lane 7).
To assess the regulatory effects of PhoP on ompF and acnA, we constructed transcriptional fusions by combining the native promoter of each candidate with a lacZ (encoding b-galactosidase) reporter. The resulting pLACZ-P phoP , pLACZ-P ompF , or pLACZ-P acnA was transformed into E. coli DH5a with an arabinose inducible expression of the phoP vector (pBAD33-phoP) or an empty vector (pBAD33  (Fig. 4C). The expression of phoP induced by arabinose significantly decreased the b-galactosidase activity of the cells with a [pLACZ-P ompF ] reporter (P , 0.01) and insignificantly increased that of the cells with [pBAD33-phoP]:[pLACZ-P acnA ] (Fig. 4C), indicating that: (i) PhoP can regulate the transcription of ompF and acnA via their own promoters and (ii) PhoP plays reverse roles in the regulation of ompF and acnA. To explain the insignificant regulatory function of PhoP on acnA, we compared the transcription level of the acnA gene in the WT with those of the DphoP cells under different conditions. RT-qPCR results suggested that the presence of PhoP did not lead to an increased expression of acnA in cells under no antibiotic pressure but did under the pressure of NAL or CAZ (Fig. 4D). Therefore, other cofactors (e.g., sigma factors, c-di-GMP, ppGpp) might be required for PhoP to regulate the transcription of acnA. A comparative analysis on the acnA promoter region in several species revealed that this identified PhoP binding site "TGTTTGTGTTATCTTTA" was highly conserved across Salmonella, including Salmonella bongori (S. bongori), S. Enteritidis, S. Typhimurium, S. Newport, S. Dublin, S. Indiana, S. Birkenhead, and S. Berta, as well as E. coli, Shigella flexneri, and Citrobacter koseri, indicating that this regulatory mechanism was conserved across these species (Fig. 4E). Surprisingly, the ompF promoter region was not conserved in S. bongori or in E. coli, Citrobacter koseri, and Shigella flexneri, implying that the identified PhoPQ regulatory mechanism on the ompF gene might be specific to S. enterica (Fig. 4F). These results provided clues regarding how bacteria evolve regulons to control gene transcription in the presence of antibiotics.
Quinolones and cephalosporins lead to phosphorylated PhoQ. To further investigate whether PhoQ is sufficient to respond to quinolones and cephalosporins, we isolated membrane extracts from an E. coli DH5A strain expressing His-tagged PhoQ. The membrane extracts were treated with both antibiotics and ATP in vitro, and PhoQ phosphorylation was assessed via a Phos-tag assay. A shifted band of phosphorylated PhoQ (PhoQ-P) was observed when the membrane extract was treated with quinolones (nalidixic acid, ciprofloxacin, and ofloxacin), cephalosporins (ceftazidime, ceftriaxone, and cefepime) or ATP; the intensity of such a shifted band was much lower when treated with azithromycin or when without antibiotics (Fig. 5A, lanes 1 and 5). In vivo, we grew an S. Enteritidis DphoQ mutant (Fig. S5), and then the DphoQ-hisphoQ compensatory strain was constructed for a Phos-tag assay. In the presence of NAL or CAZ, a shifted band of PhoQ-P was detected in the whole-cell lysate of DphoQ-hisphoQ (Fig. 5B), suggesting that PhoQ was phosphorylated. Taken together, these results indicated that the presence of quinolones or cephalosporins led to phosphorylated PhoQ.
Quinolones and cephalosporins directly bind to the extracellular sensor domain of PhoQ. We hypothesized that antibiotics trigger the phosphorylation of PhoQ by binding to the periplasmic sensor domain of PhoQ (PhoQ SD ). To test this, the truncated PhoQ SD was expressed in E. coli BL21 (Fig. S6), where PhoQ SD was found in a dimer form, according to the native-PAGE (data not shown). Differential scanning calorimetry (DSC) and microscale thermophoresis (MST) assays were performed to evaluate the interactions between the PhoQ SD dimer and antibiotics.
For the DSC measurements, PhoQ SD was adjusted to 20 mM with or without antibiotics (100 mM) in 10 mM sodium phosphate buffer (pH 7.4 or 5.5). A sharp exothermic peak was detected to conform to the thermal denaturation temperature (T m ) melting curves of PhoQ SD , PhoQ SD -NAL, and PhoQ SD -CAZ at 52.29°C, 51.00°C, and 50.54°C under neutral conditions, respectively (Fig. 5C). Similarly, under slightly acidic conditions (pH 5.5), PhoQ SD melted at a T m of 55.79°C (Fig. 5D), which was higher than that observed in the presence of NAL or CAZ (53.88°C and 54.27°C, respectively). These data suggested that PhoQ SD could bind to the antibiotics and further destabilize their tertiary structures.
For the MST tests, the recombinant PhoQ SD was titrated with various concentrations of antibiotics, and the raw data of the fluorescence time trace in 16 capillaries are shown in Fig. S7. As a positive control, we measured the binding affinity of colistin with PhoQ SD . As the concentration of colistin increased, PhoQ SD displayed an enhanced binding affinity to colistin, which formed a thermophoretic amplitude with a dissociation constant (K d ) value of 6.51 6 0.35 mM (Fig. 5E) and a well-fitted fluorescence curve. Notably, when the same procedure was applied for NAL, CAZ, or ofloxacin, we also obtained well-fitted curves of fluorescence changes in the thermophoretic amplitude, with K d values of 24.73 6 4.30 mM, 25.63 6 13.42 mM, and 33.08 6 28.28 mM, respectively. The interactions between PhoQ SD and sulfamethoxazole/azithromycin displayed poor affinity with the dispersed fluorescence. Interestingly, the fluorescence of PhoQ SD binding to CAZ decreased as the antibiotic concentration increased, in contrast to what was observed with antibiotics. This might be attributable to the different binding sites in PhoQ SD . These data revealed the specific binding affinity of PhoQ to NAL, CAZ, and ofloxacin.
The internal cavity in the PhoQ sensor domain is critical for the recognition of antibiotics. To clarify the spatial position in which the interaction between NAL/CAZ and PhoQ SD occurred, molecular docking was used to model and verify their binding at the atomic level. The evolutionary conservation of the PhoQ sensor was estimated, based on the homologous sequence of phylogeny, using Consurf. The sequences of PhoQ SD (residues 45 to 188) from S. Enteritidis, S. Typhimurium, S. bongori, E. coli, Klebsiella pneumoniae (K. pneumoniae), Vibrio parahaemolyticus (V. parahaemolyticus), and P. aeruginosa were aligned, suggesting relatively conserved primary and secondary structures of PhoQ among various species (Fig. 6A). S. Enteritidis PhoQ SD was composed of a mixed a/b-structure with a central, seven-stranded b sheet core, and it shared an amino acid sequence identity of .81.08% with other species. In addition, the homologous modeling of the three-dimensional structure of S. Enteritidis PhoQ SD was carried out using the crystal structures of E. coli and S. Typhimurium PhoQ (PDB ID: 3BQ8 and 1YAX) as the templates. The two monomer structures of PhoQ SD were highly conserved and formed a cavity pocket in the N terminus and in the C terminus.
Given that PhoQ was anchored to the intracellular membrane in a dimeric form and that the dimeric crystal structures of the PhoQ sensor domain of E. coli and S. Typhimurium show different conformations, we constructed two models of the S. Enteritidis PhoQ SD dimer for molecular autodocking. When NAL was docked into the PhoQ SD dimer (using 3BQ8 as the template), seven residues (Lys46, Arg50, Leu51, Lys142, Ile181, Lys186, and Ser188) were found to be located at a distance of within 5 Å to NAL, among which Lys46, Arg50, Lys142, and Ser188 formed hydrogen bonds with the NAL molecule (Fig. 6C). Besides that, CAZ was docked into 8 residues (Phe44, Asp45, Thr48, Ser158, Asp179, Ile183, Lys186, and Arg187) of PhoQ SD by binding to Thr48 and Ser188 via hydrogen bonds (Fig. 6D). In contrast to NAL, which recognized the two N termini and the one C terminus of the PhoQ SD dimer, CAZ was prone to binding in the horizontal line of the two N-terminus and C terminus central positions. These residues were almost located in the cavity pocket of PhoQ SD , and this binding model would provide a vacant space facing the cell membrane. The lowest binding energy that was observed in the PhoQ SD dimer with NAL or CAZ was 26.60 kcal/mol or 24.55 kcal/mol, respectively (Table 1). Notably, the crystal structure of S. Typhimurium PhoQ SD (PDB ID: 1YAX) was in a Ca 21 -bound state, and the monomer was in close proximity to the inner membrane, leading to PhoQ repression. Therefore, NAL and CAZ were only bound to the PhoQ SD monomolecular when S. Typhimurium PhoQ SD was used as the template, with the lowest binding energies of 26.10 and 24.48 kcal/mol, respectively ( Fig. 6E and F) ( Table 1).
The virtual mutagenesis of the screened amino acids located in the PhoQ SD cavity, including 5 basic residues (Lys46, Thr48, Arg50, Arg187, and Lys186), an acidic residue (Asp45), and a polar residue (Ser188) was performed. The docking results suggested that the PhoQ SD variants with each mutation displayed enhanced binding energy with antibiotics, particularly those with double and triple mutations (Fig. 7A). Furthermore, these residues were replaced with small, nonpolar residues of alanine, glutamine, or leucine to obtain the DphoQ-hisphoQ mutant strains. All of the strains were cultured in LB with NAL/CAZ, and the cell lysates were obtained for a phosphorylation analysis. A shifted band of PhoQ was observed in the DphoQ-hisphoQ cell lysate, whereas the intensity of such a shifted band almost disappeared in the variants of DphoQ-hisphoQ D45A/K46Q , DphoQ-hisphoQ T48A/R50L , and DphoQ-hisphoQ K186Q/R187L/S188A (Fig. 7B,  lanes 3 to 9), strongly suggesting that these residues were crucial for PhoQ autophosphorylation. Therefore, the PhoQ cavity was crucial for the recognition of antibiotics, which allowed for PhoQ autophosphorylation and signal transduction to PhoP.

DISCUSSION
Gram-negative bacterial cells are encased by a trilaminar envelope that is constituted by a thin peptidoglycan layer that is sandwiched between two membranes (the outer membrane and the inner membrane). This structure serves as a mechanistic barrier that defends against antibiotics of various actions. Typically, hydrophobic quinolones reach their intracellular targets by diffusion through the lipid bilayer of the outer membrane, whereas hydrophilic cephalosporins penetrate the outer membrane through porins or selective channels (44). This prompted us to wonder whether the PhoPQ-mediated resistance of S. Enteritidis to these antibiotics was achieved via the rearrangement of envelope entities so as to control the intracellular accumulation of antibiotics. The RNA-seq results revealed that the envelope-associated events were largely affected by PhoPQ, including the modification of the LPS surface charge, the alteration of the fatty acid composition, and the interruption of peptidoglycan synthesis. The phenotypic observations caused by the loss of PhoP supported the RNA-seq results and indicated that cells lacking the PhoP regulator, when exposed to quinolones and cephalosporins, underwent envelope perturbation that was favorable to the killing of antibiotics. A prominent example was the expressional alteration of the LPS modification genes, which led to an unstable membrane with a smaller amount of negatively charged LPS, which was prone to an influx of massive hydrophobic NAL or hydrophilic CAZ. Previous studies have also demonstrated the contribution of PhoPQ to LPS homeostasis in S. Typhimurium and E. coli, where PhoPQ modulates the remodeling of the lipid A domain of LPS in the presence of CAMPs or low concentrations of Mg 21 (25,45). The observed upregulation of the outer membrane porin genes (Fig. 2D), combined with the nature of porins to promote the entry of antibiotics, could, to some extent, explain the increased sensitivity of the cells lacking PhoP. In addition, cells with membranes of high fluidity are vulnerable to environmental stress (46). Given the contribution of UFAs to membrane fluidity, the higher level of expression in UFAs (Fig. 3C) might also explain the envelope perturbation in the DphoP cells. Moreover, the peptidoglycan layer is another critical barrier that undertakes the guard of cells, with a meshwork formed by cross-linked glycan strands (38,39,47). The downregulated genes related to peptidoglycan biosynthesis and cross-linking in DphoP cells (Fig. 2C) suggested that PhoPQ acts to augment peptidoglycan synthesis to compensate for the loss due to cell wall hydrolysis by CAZ. These data collectively revealed a role of PhoPQ in the coordination of envelope components to limit the intracellular accumulation of antibiotics, which might be also available against other antibiotics, as envelope-mediated resistance is not specific to one or several classes of antibiotics.
Osmolarity is known as one of the host environmental factors to bacteria, and PhoPQ perceives an osmotic upshift and induces the activation of virulence factors (23). Consistently, our RNA-seq results revealed that the absence of PhoP did reduce the expression of pagN as well as the SPI-2 type III secretion system-related operons (ssa, ssc, and sse) (Fig. S2), which are the major factors that are responsible for the Salmonella adhesion and invasion of host cells (48)(49)(50). These findings might provide the basis for an explanation of the synchronized occurrence of resistance and pathogenicity. The reason for a hyperosmotic response in a nonhyperosmotic environment (LB broth) in this study might be explained by the proton-leaving. Proton antiporters are responsible for export of cations, which leads to a decreased cytoplasmic osmotic pressure, but this descending osmotic  The amino acids that are unmarked are located in the A chain of the PhoQ SD dimer. The amino acids that are underlined were located in the B chain of the PhoQ SD dimer. NAL, nalidixic acid; CAZ, ceftazidime.
PhoPQ Regulates Quinolone and Cephalosporin Resistance mBio pressure could be offset by the osmoprotectant transport system, which was functional with cotransport protons and required energy to be supplied (51,52). In this work, for the first time, we observed the upregulation of the MotA/TolQ/ExbB proton channel family genes and the downregulation of the osmotic stress-related genes in DphoP cells (Fig. 2E). In addition, in DphoP cells, the expression of energy-producing systems, operons, and genes (dms, cyd, asr, pta, frdC, etc.) was also dramatically downregulated (Fig. S2C). Therefore, we assumed that the deficiency in PhoPQ causes proton-leaving and an osmotic pressure imbalance under the pressure of antibiotics (Fig. 8), and this ultimately retarded cell viability under NAL and CAZ. The production of ROS, including H 2 O 2 , the hydroxyl radical (OH), and the superoxide anion radical (O 2 -), is among the bactericidal pathways that are shared by drugs of various actions (53,54). Our findings suggested that NAL treatments and CAZ treatments remarkably increased ROS production and decreased the NAD 1 /NADH ratio in DphoP cells. Our findings also demonstrated the transcription of the genes that are responsible for ROS formation and iron transport as well as the decreased transcription of the genes that are responsible for ROS scavenging, Fe-S cluster assembly, and the PhoPQ Regulates Quinolone and Cephalosporin Resistance mBio NADH-related electron transport chain (55)(56)(57)(58)(59)(60)(61)(62), which resulted in an ROS-generating Fenton reaction and an intracellular redox imbalance (Fig. 8). On the other hand, it was reported that the antibiotic treatments could stimulate ROS production by hyperactivating the bacterial central metabolism (e.g., the tricarboxylic acid cycle) (63,64). The downregulated expression of the tricarboxylic acid cycle-related genes acnA, fumC, and frdC in DphoP cells (Fig. S2C) suggested a positive regulation of PhoPQ in the central metabolism. These data indicated that the regulatory pathways of PhoPQ were complex but closely linked so that they could sustain a rigid envelope and cytoplasm homeostasis. The RNA-seq data, combined with phenotypic validations, allowed us to understand which downstream genes are influenced by the PhoP regulator. Based on the PhoP box motif sequence, we further discriminated 68 putative gene candidates that were directly targeted by PhoP from all of the DEGs in the DphoP cells, of which the negative regulon ompF and the positive regulon acnA were identified for the first time via EMSA and lacZ reporter assays. These direct targets might reflect the emergency measurements that are immediately taken by S. Enteritidis once stressed by antibiotics. The negatively regulated ompF encodes an outer membrane porin that prohibits the diffusion and entry of a variety of small molecules, including quinolones and b-lactams, into the cell. The positively regulated acnA encodes a citric acid cycle enzyme aconitase that responds to the superoxide anion and hydrogen peroxide to facilitate energy production and prevent the liberation of reactive iron from oxidative damage (65). This implies that the reduced expression of OM porins as well as the maintenance of intracellular basic metabolism and the redox balance might serve as the "first aid" in S. Enteritidis when it is confronted with antibiotics.
Having gained information about how PhoP regulates the genes to deal with the stresses from antibiotics, we explored how this TCS exploits the "sentry" PhoQ to report their presence to PhoP. Although TCSs are widely distributed across bacterial species, there are limited details regarding the signal transduction and the intact PhoPQ Regulates Quinolone and Cephalosporin Resistance mBio crystal structure for histidine kinases (66)(67)(68). The PhoQ periplasmic sensor domain was reported to respond to low divalent cations (21,69), an acidic pH (22), CAMPs, and the small regulator protein SafA (70,71). Here, we surprisingly found that quinolones and cephalosporins could directly interact with the cavity of PhoQ sensor domain and that this binding mode provided a vacant space for conformational changes of PhoQ for activity control and autophosphorylation. It was previously shown that SafA directly interacts with the cavity structure of PhoQ for activation, and the mutagenesis of the residues surrounding the cavity attenuated the SafA-mediated activation of PhoQ (70,71). In this study, the mutagenesis of the residues (D45, K46, T48, R50L, K186, R187, and S188) in the cavity of PhoQ additionally enhanced its binding energy with antibiotics and abrogated its phosphorylation, suggesting that the antibiotics of quinolone and cephalosporin could activate PhoQ in a similar way as SafA. Therefore, we concluded that this binding mode enables the activation of PhoQ via a conformational change and the delivery of a signal to PhoP. Altogether, S. Enteritidis PhoQ directly senses quinolones or cephalosporins to activate PhoPQ regulation in a rapid way without damage to the integrity of the cell envelope, which may represent a novel mechanism by which to defend against quinolones or cephalosporins. Conclusively, we have demonstrated the transcriptional regulation of PhoPQ on bacterial envelope integrity, the osmotic stress response, and the redox balance, which enabled resistance to quinolones and cephalosporins in S. Enteritidis. The histidine kinase PhoQ is an antibiotic sensor that initiates signal transduction, leading to a phosphorylated PhoP that activates or represses the transcription of target genes (ompF and acnA) in response to antibiotics. Therefore, we have proposed a schematic model illustrating how the membrane-associated histidine kinase PhoQ signals the presence of antibiotics and dispatches PhoP to trigger the expression of overall genes to adapt to antibiotic-caused stress (Fig. 8).
In future work, it would be feasible to develop PhoPQ inhibitors, based on the novel target genes identified in our work, to potentiate the efficacy of quinolones or cephalosporins in treating infections caused by S. Enteritidis, which may provide a promising means by which to inhibit antibiotic resistance.

MATERIALS AND METHODS
Strains and culture conditions. S. Enteritidis SJTUF12367 (GenBank assembly accession no. GCA _004323915.1; https://www.ncbi.nlm.nih.gov/assembly/GCF_004323915.1/) and its derivatives were used in this study (35). The complete list of bacterial strains, vectors, and primers was shown in Table S1. Bacterial routine culture was performed in lysogeny broth (LB) medium at 37°C. Unless otherwise specified, the subinhibitory concentration antibiotic of 32 mg/mL nalidixic acid or 2 mg/mL ceftazidime was added to the medium treatment for 8 h. All strains could grow, and the DphoP mutant showed a susceptible phenotype.
RNA-seq and RT-qPCR. Bacterial cultures in the exponential or log phase in LB were inoculated to antibiotics and taken for RNA extraction at 8 h. The total RNA was extracted using the TRIzol reagent (Invitrogen), referencing the method described by Huang et al. (72). The strand-specific transcriptome library construction and Illumina RNA-seq were performed by Majorbio (Shanghai, China). Three independent, biologically repeated experiments were performed. The integrity of the RNA was confirmed via electrophoresis (Fig. S1A). The transcripts per kilobase per million mapped reads incorporate normalization steps to ensure that the expression levels for different genes and transcripts can be compared across runs. The DESeq2 method was used to calculate the DEGs. Genes with a fold change of $2 and a Bonferroni-corrected P value (P adj ) of ,0.05 in at least one sample were determined to be DEGs. COG and KEGG annotation analyses of the DEGs were carried out with the free, online Majorbio Cloud Platform. The RT-qPCR was run in triplicate and amplified using the TB green Premix Ex Taq II reagent (TaKaRa), and the amplification efficiencies of the oligonucleotide primers were listed in Table S2. The relative gene transcriptional levels were determined using the 2 -DDCT threshold cycle (CT) method and were converted into log 2 values (72). The expression of 16S rDNA was used as an internal reference.
Data analysis. The data analysis was conducted using a two-way analysis of variance (ANOVA) and the Holm-Sidak multiple-comparison test via the Prism8.0.1 program.
Data availability. The Gene expression data have been deposited into the NCBI Gene Expression Omnibus (GEO) under the accession number GSE217589. The detailed protocols for the phenotypic analysis, bacterial genetic manipulations, EMSA, DSC, MST, molecular docking, and in vitro and in vivo phosphorylation assays are provided in the "Text S1" file.

SUPPLEMENTAL MATERIAL
Supplemental material is available online only. TEXT S1, DOC file, 0.1 MB. We declare no conflict of interest.