Role of OB-Fold Protein YdeI in Stress Response and Virulence of Salmonella enterica Serovar Enteritidis

S. Enteritidis during its life cycle encounters diverse stress factors inside the host. These intracellular conditions allow activation of specialized secretion systems to cause infection. We report a conserved membrane protein, YdeI, and elucidate its role in protection against various intracellular stress conditions. A key aspect of the study of a pathogen’s stress response mechanism is its clinical relevance during host-pathogen interaction. Bacterial adaptation to stress plays a vital role in evolution of a pathogen’s resistance to therapeutic agents. Therefore, investigation of the role of YdeI is vital for understanding the molecular basis of regulation of Salmonella pathogenesis. In conclusion, our findings may contribute to finding potential targets to develop new intervention strategies for treatment and prevention of enteric diseases.

ABSTRACT

An essential feature of the pathogenesis of the Salmonella enterica serovar Enteritidis wild type (WT) is its ability to survive under diverse microenvironmental stress conditions, such as encountering antimicrobial peptides (AMPs) or glucose and micronutrient starvation. These stress factors trigger virulence genes carried on Salmonella pathogenicity islands (SPIs) and determine the efficiency of enteric infection. Although the oligosaccharide/oligonucleotide binding-fold (OB-fold) family of proteins has been identified as an important stress response and virulence determinant, functional information on members of this family is currently limited. In this study, we decipher the role of YdeI, which belongs to OB-fold family of proteins, in stress response and virulence of S. Enteritidis. When ydeI was deleted, the ΔydeI mutant showed reduced survival during exposure to AMPs or glucose and Mg²⁺ starvation stress compared to the WT. Green fluorescent protein (GFP) reporter and quantitative real-time PCR (qRT-PCR) assays showed ydeI was transcriptionally regulated by PhoP, which is a major regulator of stress and virulence. Furthermore, the ΔydeI mutant displayed ∼89% reduced invasion into HCT116 cells, ∼15-fold-reduced intramacrophage survival, and downregulation of several SPI-1 and SPI-2 genes encoding the type 3 secretion system apparatus and effector proteins. The mutant showed attenuated virulence compared to the WT, confirmed by its reduced bacterial counts in feces, mesenteric lymph node (mLN), spleen, and liver of C57BL/6 mice. qRT-PCR analyses of the ΔydeI mutant displayed differential expression of 45 PhoP-regulated genes, which were majorly involved in metabolism, transport, membrane remodeling, and drug resistance under different stress conditions. YdeI is, therefore, an important protein that modulates S. Enteritidis virulence and adaptation to stress during infection.

IMPORTANCE S. Enteritidis during its life cycle encounters diverse stress factors inside the host. These intracellular conditions allow activation of specialized secretion systems to cause infection. We report a conserved membrane protein, YdeI, and elucidate its role in protection against various intracellular stress conditions. A key aspect of the study of a pathogen’s stress response mechanism is its clinical relevance during host-pathogen interaction. Bacterial adaptation to stress plays a vital role in evolution of a pathogen’s resistance to therapeutic agents. Therefore, investigation of the role of YdeI is vital for understanding the molecular basis of regulation of Salmonella pathogenesis. In conclusion, our findings may contribute to finding potential targets to develop new intervention strategies for treatment and prevention of enteric diseases.

INTRODUCTION

Salmonella enterica serovar Enteritidis is a leading causative pathogen of foodborne gastroenteritis. Annual disease burden estimates of up to 93 million infections and 155,000 deaths worldwide pose severe clinical and economic concern (1). During the course of infection, Salmonella enterica serovars encounter a wide range of stress factors inside the host. Following ingestion, the pathogen experiences several stress conditions, such as the action of antimicrobial peptides (AMPs), low pH, bile salts, nutrient scarcity, like carbon starvation, micronutrient (Mg²⁺, Mn²⁺, and Ca²⁺) limitation, high osmolarity, and oxidative and nitrosative stresses (2). To survive these hostile conditions, the pathogen triggers stress response and virulence genes, which also determines the efficiency of infection (2).

The major virulence factors of Salmonella are expressed by the type III secretion system (T3SS) encoded on Salmonella pathogenicity islands (SPIs) (2, 3). SPI-1 is essential for pathogenic invasion into host intestinal epithelial cells, and SPI-2 facilitates phagocytic survival and systemic dissemination in the host (4). Salmonella pathogenesis has been extensively studied in Salmonella enterica serovar Typhimurium (S. Typhimurium), while pathogenesis of S. Enteritidis remains to be explored (3). Of the two serovars, S. Enteritidis is the most prevalent in clinical settings (1, 5). Therefore, identification and functional elucidation of factors that contribute to S. Enteritidis adaptation to stress and virulence during infection are vital for better understanding of nontyphoidal salmonellosis.

One of such bacterial stress response determinants is the oligosaccharide/oligonucleotide binding-fold (OB-fold) family of proteins. The members of the OB-fold family of proteins are periplasmic and are reported to contain an oligosaccharide/oligonucleotide binding-fold domain (COG3111) (6). They are comprised of five antiparallel β-sheets that form a closed or partially opened barrel-like structure. This structural conformation allows it to function as a scaffold or supports binding of ligands like nucleotides, oligosaccharides, and proteins (6). However, due to lack of nucleotide binding residues, the members are known to bind specifically to oligosaccharides, proteins, and bacterial enterotoxins (6).

PhoP is one of the major transcriptional regulators of bacterial stress response and virulence. It controls the transcription of a large number of genes directly or indirectly in response to vacuolar stress signals such as AMPs, low pH, free radicals, and low Mg²⁺ ions, which promotes pathogen survival and systemic infection inside the host (7–12). A periplasmic protein, YdeI, belonging to the OB-fold superfamily was previously reported to interact with porin OmpD and conferred protection against AMPs under regulation of RcsBCD and PhoPQ in S. Typhimurium strain SL1344 (13). Nonetheless, the functional role of YdeI in bacterial stress responses and pathogenesis remains largely unexplored, especially in S. Enteritidis.

In the present study, we have identified an ortholog of YdeI of S. Typhimurium in S. Enteritidis through BLASTp and synteny analyses. Interestingly, deletion (ΔydeI) and complementation (cΔydeI) analyses of ydeI under exposure to AMPs, glucose or Mg²⁺ starvation, and bile stress indicated that YdeI plays a crucial role in stress response of S. Enteritidis, unlike its ortholog in S. Typhimurium (13). Our study showed that ydeI is transcriptionally regulated by PhoP under different stress conditions through differential activation of its promoter (P_ydeI). We also observed significant changes in expression of several PhoP-regulated genes involved in metabolism, transport, membrane remodeling, and drug resistance in the ΔydeI mutant versus the WT. Computational docking analysis supported YdeI binding interaction with OmpD in S. Enteritidis, which corroborated with the previous experimental evidence on YdeI-OmpD interaction and its role in AMP resistance in S. Typhimurium (13). Overall, YdeI protein was found to be critical in promoting S. Enteritidis survival under various stress conditions, invasion, and intramacrophage survival in vitro and systemic infection in the C57BL/6 mouse infection model.

RESULTS

YdeI confers resistance to multiple stress conditions in S. Enteritidis.

The protein sequence of YdeI from Salmonella enterica serovar Typhimurium strain SL1344 (GenBank accession no. HAD6491983.1) was used as the BLAST query to find its ortholog in Salmonella enterica serovar Enteritidis strain P125109 (WT) (GenBank accession no. CAR33115.1). Sequences from both serovars exhibited sequence identity and sequence similarity of 100% (see Fig. S1B in the supplemental material). A synteny analysis of the neighboring genomic region of YdeI revealed conservation of the neighboring genes across the two serovars. Conserved domain analysis of YdeI revealed that the protein belongs to the OB-fold family of proteins (Fig. S1A), corroborating the previous report in S. Typhimurium (13). The protein is 130 amino acid (aa) residues in length, with a predicted signal peptide at the position aa 1 to 19 and one conserved OB-fold domain from aa 24 to 130 (Fig. S1A). Moreover, YdeI is prevalent across many important Gram-negative pathogens, like Klebsiella pneumoniae, Escherichia coli, Shigella dysenteriae, and several serovars of Salmonella enterica (Fig. S1C) (13).

YgiW protein, which is a member of OB-fold family of proteins, was previously reported to promote S. Typhimurium survival against vacuolar stress, including AMPs, low pH, and low Mg²⁺ ions, as well as oxidative stress and Cd²⁺ toxic conditions (14). Since YdeI is an OB-fold protein and was previously reported to be regulated by PhoP in S. Typhimurium (13), we investigated YdeI’s role in several stress conditions of S. Enteritidis. The WT, ΔydeI deletion mutant, and cΔydeI complemented strain (Table 1) were treated with AMPs: 1 μg/ml polymyxin B (PMB) and 10 μg/ml LL-37. The ΔydeI strain had a survival rate of 9.12% (P < 0.0001) after treatment with PMB compared to the WT (80.39%). The mutant had 20.95% survival (P < 0.0001) following treatment with LL-37 in comparison to the WT (87%) (Fig. 1A). The cΔydeI strain showed restoration of survival to 70.56% (P < 0.05) and 69.6% (P < 0.05) after treatment with PMB and LL-37, respectively, in comparison to the WT (Fig. 1A). The present findings corroborated the previously reported role of YdeI in AMP resistance of S. Typhimurium (13). The ΔydeI strain was further investigated for its survival under glucose starvation. The ΔydeI strain showed ∼2-fold-decreased survival (P < 0.01) compared to the WT (Fig. 1B). The cΔydeI strain showed ∼1.5-fold-increased survival compared to its mutant (P < 0.05), and its growth was similar to that of the WT (Fig. 1B). The ΔydeI mutant showed no difference in its survival compared to the WT under pH stress (data not shown). Furthermore, the mutant displayed 2-fold-decreased survival (P < 0.01) compared to the WT during Mg²⁺ starvation. The cΔydeI strain showed an ∼1.5-fold increase in survival over the mutant (P < 0.01), partially restoring its survival to WT levels under Mg²⁺ starvation (Fig. 1C).

TABLE 1.

Strains and plasmids used in this study

Strain or plasmid	Relevant genotype and/or phenotypea	Background	Resistanceb	Reference
Strains
WT	Salmonella enterica serovar Enteritidis strain P125109	WT	Strr (naturally resistant)	50
Mutant
ΔinvC strain	TTSS-1− (invC::aphT)	WT	Strr Kanr	50
ΔssaV strain	TTSS-2− (ssaV::aphT)	WT	Strr Kanr	58
ΔydeI strain	YdeI− (ydeI::aphT)	WT	Strr Kanr	This study
ΔompD strain	OmpD− (ompD::cat)	WT	Strr Cmr	This study
ΔydeI ΔompD strain	YdeI− OmpD− (ydeI::aphT ompD::cat)	ΔydeI	Strr Kanr Cmr	This study
Complemented
cΔydeI strain	ΔydeI mutant complemented with ydeI through pCH2189	ΔydeI	Strr Kanr Ampr	This study
cΔompD strain	ΔompD mutant complemented with ompD through pCH1086	ΔompD	Strr Cmr Ampr	This study
cΔydeI ΔompD strain	ΔydeI ΔompD mutant complemented with ydeI through pCH2189	ΔydeI ΔompD	Strr Kanr Cmr Ampr	This study
Plasmids
pKD4	bla FRT aphT FRT PS1 PS2 oriR6K		Kanr	52
pKD3	bla FRT cat FRT PS1 PS2 oriR6K		Cmr	52
pKD46	bla PBAD gam bet exo pSC101 oriTS		Ampr	52
pCJLA	Low-copy-no. plasmid; aphT gfp-mut3 cat in a derivative of pACYC177; orip15A		Kanr Cmr	58
pCH112	Low-copy-no. plasmid; hilA ORF cloned into PBAD/myc-His; oripBR322		Ampr	58
pM968	Low-copy-no. plasmid; bla promoterless gfp-mut2 in a derivative of pBAD24 vector		Ampr	56
pM2155	Low-copy-no. plasmid; bla PSEN1140 gfp-mut2 plasmid; oripMB1		Ampr	50
pCH2189	Low-copy-no. plasmid; ydeI ORF with 1,000-bp upstream region cloned into pCH112 between NcoI and XbaI by replacing the ORF of hilA; oripBR322		Ampr	This study
pCH1086	Low-copy-no. plasmid; ompD ORF with 1,000-bp upstream region cloned into pCH112 between NcoI and XbaI by replacing the ORF of hilA; oripBR322		Ampr	This study
pM3968	Low-copy-no. plasmid; bla promoter of ydeI (PydeI) cloned into pM968 between XbaI and PstI; oripMB1		Ampr	This study

^aFRT, FLP recombination target.

^bStr^r, streptomycin resistance; Kan^r, kanamycin resistance; Amp^r, ampicillin resistance; Cm^r, chloramphenicol resistance.

FIG 1.

YdeI protects S. Enteritidis from multiple stresses. The survival of log-phase cultures of the ΔydeI mutant and the cΔydeI complemented strain was assessed in comparison to the WT. (A) Antimicrobial peptide (AMP) stress assay. (B) Glucose starvation stress assay. (C) Mg²⁺ starvation stress assay. (D) Bile stress assay. (E) Toxicity of Cd²⁺ ions. AMP stress survival was represented as a percentage of that of the WT. The bacterial counts were calculated in at the indicated time points. Relative survival of the ΔydeI and cΔydeI strains was assessed against that of the WT. All data represent three independent experiments, and error bars indicate mean ± SD. Statistical significance by two-way ANOVA: *, P < 0.05; ****, P < 0.0001.

Furthermore, the ΔydeI mutant was assessed for its viability following treatment with small toxic molecules such as bile salts and CdCl_2. The results indicated the ΔydeI strain had an ∼2.5-fold decrease in viability (P < 0.001) compared to the WT strain (Fig. 1D). The ΔydeI strain was ∼3 times more susceptible to toxicity from 4 mg/ml CdCl₂ salts than the WT strain (P < 0.001) (Fig. 1E). Concurrently, the cΔydeI complemented strain showed increased survival similar to WT levels (Fig. 1E). These data, therefore, suggested that YdeI was important for resistance against several microenvironmental stress conditions.

PhoP regulates the expression of ydeI.

Previous reports have shown that ydeI was regulated by PhoP in S. Typhimurium (11, 13). So, we investigated the role of PhoP in regulation of ydeI in S. Enteritidis through quantitative real-time PCR (qRT-PCR) analysis. ydeI showed significantly reduced expression in a ΔphoP mutant compared to the WT cultured in LB medium (2.2-fold in Fig. 2A) and M9 minimal medium (6.5-fold in Fig. 2B). The gene showed ∼45- and ∼68-fold-decreased expression compared to the WT following treatment with the AMPs PMB and LL-37, respectively (Fig. 2C). Furthermore, ydeI was ∼100-fold downregulated in the ΔphoP mutant under glucose starvation (Fig. 2D). Compared to the WT, the gene showed ∼42-fold-decreased expression in the ΔphoP strain under Mg²⁺ starvation (Fig. 2E). Furthermore, ydeI was ∼6- and ∼15-fold downregulated following treatment with bile salts (Fig. 2F) and CdCl₂ (Fig. 2G), respectively.

FIG 2.

PhoP regulates the expression of ydeI. (A to G) Expression of ydeI was assessed in the ΔphoP mutant versus the WT under different conditions through qRT-PCR. (A) LB medium. (B) M9 minimal medium. (C) Antimicrobial peptide stress. (D) Glucose starvation stress. (E) Mg²⁺ starvation stress. (F) Bile stress. (G) Toxicity of Cd²⁺ ions. The 16S rRNA gene was taken as the housekeeping gene for qRT-PCR. Error bars indicate the mean ± SD. Statistical significance by Student’s t test: **, P < 0.01; ***, P < 0.001; ****, P < 0.0001. (H to N) The putative promoter of ydeI (P_ydeI) was predicted and cloned into pM968 vector to generate the GFP-expressing pM3968 construct. The construct was transformed into the WT and ΔphoP strains to generate the WT/pM3968 and Δpho/pM3968 strains for flow cytometry-based GFP reporter assays. Representative histogram plots depict the mean fluorescence intensity (MFI [AU]) of the WT/pM3968 and ΔphoP/pM3968 strains cultured under different stress conditions. The WT/pM968 strain (harboring non-GFP-expressing plasmid pM968) and the WT/pM2155 strain served as the negative and positive controls, respectively. The experiments were conducted three times using the BD FACScanto II cytometer (Becton Dickinson, Erembodegem, Belgium), and data were analyzed by FlowJo v. 10.4.2.

A putative promoter of ydeI (P_ydeI) was identified using BPROM software. P_ydeI was located 420 bp upstream of the transcriptional start site of the target gene at position 1636172 to 1636381. The genetic orientation of ydeI and putative promoter region is shown in Fig. S2A in the supplemental material. Although, we could identify the −35 and −10 sequences of the predicted promoter (Fig. S2A), the transcriptional factor binding sites (TFBSs) of PhoP (the PhoP box) were not identified in the upstream regions of ydeI. The PhoP box is typically characterized by a tandem direct repeat of the heptanucleotide sequence (G/T)GTTTA(A/T) (15). The absence of the PhoP box in the upstream portion of ydeI implied ydeI was possibly indirectly regulated by PhoP.

P_ydeI was subsequently cloned into promoterless gfp-mut2 pM968 vector (Table 1) to further validate the role of P_ydeI and PhoP regulation of ydeI by flow cytometric green fluorescent protein (GFP) reporter assays. The cloned construct pM3968 (Table 1) expressed GFP, which suggested P_ydeI is the potential promoter of ydeI. The WT strain transformed with pM968 and pM2155 (Table 1) served as the experimental control for the study. Both the WT and ΔphoP mutant were transformed with pM3968 (WT/pM3968 and ΔphoP/pM3968) and cultured under different growth and stress conditions. Net change in the activity of P_ydeI was calculated by measuring the mean GFP fluorescence (mean fluorescence intensity [MFI] measured in arbitrary units [AU]) of the strains. The ΔphoP/pM3968 strain showed ∼5.7-fold-reduced expression compared to the WT/pM3968 strain following culturing in LB and M9 medium (Fig. 2H and I; Fig. S2B and S2C). Furthermore, the ΔphoP/pM3968 strain showed ∼5-fold (Fig. 2J; Fig. S2D), ∼6.5-fold (Fig. 2K; Fig. S2E), ∼5.7-fold (Fig. 2L; Fig. S2F), ∼5.1-fold (Fig. 2M; Fig. S2G), and ∼7.2-fold (Fig. 2N; Fig. S2H)-decreased MFI compared to the WT/pM3968 strain following treatment with different stress conditions.

In conclusion, the reduced activation of P_ydeI in the ΔphoP/pM3968 strain corroborated with downregulation of ydeI in the ΔphoP mutant. This suggested a pertinent role of PhoP in regulation of ydeI under various growth conditions.

Deletion of ydeI differentially regulates expression of several PhoP-regulated genes.

The phenotypic characteristics of the ΔydeI strain were similar to the previously reported phenotypes of phoP mutants (9, 15, 16). According to the current and previously reported findings, ydeI was regulated by PhoP. So, we hypothesized that deletion of ydeI could be altering the expression of several PhoP-regulated genes associated with stress resistance and virulence. A total of 45 PhoP-regulated genes involved in metabolism, antimicrobial resistance, transporters, and membrane remodeling were selected for this study that were previously reported to play a role in Salmonella stress response and virulence (2, 7, 9, 11, 17–19). The levels of expression of the genes were assessed in the ΔydeI mutant compared to the WT by qRT-PCR analysis, following culture in LB and under key intracellular stress conditions: PMB and glucose and Mg²⁺ starvation. These genes were differentially expressed at P < 0.05 in the ΔydeI mutant compared to the WT (Fig. 3). A complete list of the genes and associated functional information, fold change values, and statistical analysis is given in Data Set S1.

FIG 3.

Deletion of ydeI differentially regulates expression of PhoP-regulated genes. RNA was isolated from the WT and ΔydeI mutant cultured under different growth and stress conditions in biological duplicates for qRT-PCR analysis. Shown are expression profiles of differentially expressing genes in the ΔydeI strain compared to the WT under different conditions. (A) LB medium. (B) Antimicrobial peptide stress. (C) Glucose starvation stress. (D) Mg²⁺ starvation stress. The 16S rRNA gene was used as the housekeeping gene for the experiments. The fold change values are indicated by the color scale at the right of each panel. Genes with a log₂ fold change of ≤−1.5 were downregulated, and those with a log₂ fold change of ≥1.5 were upregulated. Statistical significance by t test: P < 0.05.

The ΔydeI mutant shows reduced invasion into HCT116 colon epithelial cells.

S. Enteritidis adheres to and then invades intestinal epithelial cells to cause infection. We observed no significant difference in the adhesion of the ΔydeI mutant compared to WT cells in HCT116 colon epithelial cells (Fig. 4A). However, the ΔydeI strain had ∼89% reduced invasion compared to the WT. The cΔydeI complemented strain showed restoration in invasion potential (86.63%) compared to the WT (Fig. 4B). In addition, the WT, ΔydeI, and cΔydeI strains were transformed with GFP expression plasmid pCJLA (Table 1) to generate the WT/pCJLA, ΔydeI/pCJLA, and cΔydeI/pCJLA strains for flow cytometry analysis. The ΔydeI/pCJLA mutant showed 1.40% invasion compared to 6.89% invasion of the WT/pCJLA strain (Fig. 4E). The cΔydeI/pCJLA complemented strain showed partial restoration in its invasion potential to 3.71% compared to the WT/pCJLA strain. We also checked the expression of ydeI in the WT after invasion into HCT116 cell line at indicated time points (Fig. 4C). ydeI showed 2-fold- and 4-fold-increased expression after 1 and 2 h of invasion into HCT116 cells, respectively (Fig. 4C).

FIG 4.

The ΔydeI mutant shows reduced invasion into HCT116 colon epithelial cells. (A and B) The HCT116 colon epithelial cell line was infected with log-phase cultures of the WT, ΔydeI mutant, and cΔydeI and ΔinvC complemented strains at an MOI of 10 to perform the adhesion assay (A) and invasion assay (B). The ΔinvC strain served as the experimental control. Data are represented as percentages of adhesion and invasion for the ΔydeI and cΔydeI strains in comparison to the WT value normalized to 100%. (C) Expression of ydeI in HCT116 after infection with the WT at the indicated time points through qRT-PCR. (D) Expression of SPI-1 genes in the ΔydeI strain compared to the WT through qRT-PCR. The 16S rRNA gene was taken as the housekeeping gene for qRT-PCR. Error bars indicate the mean ± SD from three independent experiments. (E) The HCT116 cell line was infected at an MOI of 50 with GFP-expressing strains: the WT/pCJLA strain, ΔydeI/pCJLA mutant, and cΔydeI/pCJLA complemented strain. Invasion of strains was measured by the percentage of mean fluorescence intensity (MFI) of the GFP-positive population. Data were acquired using a BD FACScanto II cytometer (Becton Dickinson, Erembodegem, Belgium) and analyzed by using FlowJo v. 10.4.2. Statistical significance by one-way ANOVA (A and B) and Student's t test (C and D): *, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001; ns, not significant (P ≥ 0.05).

Due to the reduced invasion of the ΔydeI mutant in HCT116, expression of SPI-1 genes was assessed in the mutant compared to the WT cultured in SPI-1 inducing medium (see Materials and Methods) (Fig. 4D). In SPI-1, carrying the T3SS-1 assembly genes prgH (5.17-fold), prgK (2.68-fold), and prgJ (5.17-fold), expression of these genes was downregulated in the ΔydeI strain compared to the WT. The T3SS-1 effector genes sopD (3.82-fold), sopE2 (12.25-fold), and sopB (3.82-fold) were downregulated in the ΔydeI strain compared to the WT (Fig. 4D). Additionally, positive regulators of T3SS-1 hilA and invF showed no significant change in their expression in the ΔydeI strain. However, the negative regulators of T3SS-1 rcsB (22.97-fold), rcsA (1.8-fold), and phoP (3.95-fold) showed significant upregulation in the mutant. These findings supported the role of YdeI in facilitating invasion of S. Enteritidis.

The ΔydeI mutant displays attenuated survival inside RAW 264.7 murine macrophage cells.

The ΔydeI mutant exhibited ∼45.3% reduced uptake compared to the WT (Fig. 5A) and ∼15-fold-reduced intramacrophage survival compared to WT survival at 24 h postinfection (p.i.) in the murine macrophage RAW 264.7 cell line (Fig. 5B). The cΔydeI complemented strain efficiently survived inside macrophages, displaying ∼14.4-fold survival compared to WT survival of 19-fold. In addition, flow cytometric analysis showed the ΔydeI/pCJLA strain had decreased uptake (2.20%) in comparison to the WT/pCJLA strain (23.9%) (Fig. 5E). The macrophage replication assay via flow cytometry showed reduced survival of the ΔydeI/pCJLA strain (1.54%) compared to WT/pCJLA survival of 13.9% at 24 h p.i. (Fig. 5E). Similarly, the cΔydeI/pCJLA strain showed restoration in uptake (24.3%) and survival (12.9%) compared to the WT/pCJLA strain (Fig. 5E). Concomitantly, ydeI had 3.23-fold-increased expression in the WT at 12 and 24 h p.i. in RAW 264.7 cells (Fig. 5C).

FIG 5.

The ΔydeI mutant displays attenuated survival inside RAW 264.7 murine macrophage cells. (A and B) RAW 264.7 cells were infected with the WT, ΔydeI mutant, and cΔydeI complemented strain at an MOI of 10 for the macrophage uptake assay (A) and macrophage replication and survival assay (B). Bacterial counts were calculated by lysing of RAW 264.7 cells, serial dilution, and plating at 2 and 24 h of infection. Percentages of uptake for the ΔydeI and cΔydeI strains were compared to WT uptake normalized to 100%. Intracellular survival at 24 h of infection is represented as fold replication and compared to the WT. The ΔssaV mutant serves as the experimental control. (C) Expression of ydeI in RAW 264.7 cells after infection with the WT at indicated time points through qRT-PCR. (D) Expression of SPI-2 genes in the ΔydeI mutant compared to the WT through qRT-PCR. The 16S rRNA gene was taken as the housekeeping gene in qRT-PCR experiments. Error bars indicate the mean ± SD from three independent experiments. (E) RAW 264.7 cells were infected at an MOI of 50 with GFP-expressing strains: the WT/pCJLA strain, the ΔydeI/pCJLA mutant, and the cΔydeI/pCJLA complemented strain. Experiments were conducted using a BD FACScanto II cytometer (Becton Dickinson, Erembodegem, Belgium), and data were analyzed by using FlowJo v. 10.4.2. Uptake and replication survival were measured by acquiring the percentage of MFI of the GFP-positive population. Statistical significance by one-way ANOVA (A and B) and Student's t test (C and D): *, P < 0.05; **, P < 0.01;***, P < 0.001; ****, P < 0.0001; ns, not significant (P ≥ 0.05).

Due to the attenuated intramacrophage survival of the ΔydeI mutant, expression of T3SS-2 assembly and effector genes carried in SPI-2 was evaluated in the mutant compared to the WT in SPI-2 inducing medium (Fig. 5D). T3SS-2 assembly genes ssaD, ssaE, and ssaB were downregulated by 2.61-, 2.83-, and 1.77-fold, respectively, in the ΔydeI mutant compared to the WT. T3SS-2 effector genes sseG and sseJ were downregulated by 112.10- and 10.03-fold, respectively. One of the critical findings in this study is downregulation of the mgtA, mgtB, and mgtC genes under different stress conditions (Fig. 3). These genes encode Mg²⁺ transporters and activate PhoPQ and T3SS-2 genes for phagocytic survival (20, 21). mgtA, mgtB, and mgtC also showed significant downregulation under the SPI-2 inducing condition by ∼14.2-, 429-, and 6-fold, respectively, in the ΔydeI strain (Fig. 5D). Furthermore, T3SS-2 regulators phoP (1.3-fold) and ssrB (1.2-fold) showed no significant change in their expression, while ompR was 7.15-fold downregulated in the mutant. These findings collectively indicated the potential role of YdeI in promoting S. Enteritidis survival inside macrophages.

Deletion of ydeI attenuates survival in C57BL/6 mice and reduces inflammation in vivo.

The streptomycin-pretreated C57BL/6 murine model was used to investigate the virulence of the ΔydeI mutant (n = 5 mice) compared to the WT. The ΔydeI strain displayed ∼2-fold (Fig. 6A) and 1.5-fold (Fig. 6B)-reduced fecal colonization compared to the WT at 24 and 48 h p.i., respectively. In comparison to the ΔydeI strain, the cΔydeI complemented strain showed 1-fold-increased fecal colonization at 24 h and 48 h p.i. (Fig. 6A and B). Furthermore, the bacterial load was evaluated from mesenteric lymph node (mLN), spleen, liver, and cecum at 72 h p.i. The results showed that relative to the WT, the ΔydeI strain had 2-fold-reduced survival in mLN (Fig. 6C) and spleen (Fig. 6D). The cΔydeI strain showed improved colonization with 1.5-fold-increased survival in mLN (Fig. 6C) and spleen (Fig. 6D). The mutant had ∼3-fold-reduced colonization in liver compared to the WT, while the cΔydeI strain showed 1-fold-increased colonization over the ΔydeI strain (Fig. 6E). Histopathological evaluation of cecum infected with the WT, ΔydeI, and cΔydeI strains was done through hematoxylin and eosin (H&E) staining. The ΔydeI strain had significantly lower cecal inflammation (Fig. 6I), with an average pathoscore of 4.4 (Fig. 6H) compared to the WT pathoscore of 12.84 (Fig. 6H and I). Mice infected with the cΔydeI strain displayed significantly induced inflammation in the cecum (Fig. 6I), to which an average pathoscore of 10.5 (Fig. 6H) was assigned. The cecal inflammation by the strains was further validated by estimation of levels of fecal lipocalin-2, which serves as a biomarker of intestinal inflammation during Salmonella pathogenesis (22, 23). ΔydeI strain infection in mice elicited 2.5-fold-decreased lipocalin-2 levels compared to WT infection (Fig. 6G). cΔydeI strain infection in mice elicited an ∼1.5-fold-higher lipocalin-2 response than the mutant, which partially rescued the inflammatory phenotype (Fig. 6G). Salmonella-induced inflammation was also validated by enzyme-linked immunosorbent assay (ELISA) estimation of serum levels of the inflammatory cytokines tumor necrosis factor alpha (TNF-α), interleukin-6 (IL-6), and IL-1β and the two chemokines KC and macrophage inflammatory protein 2 (MIP-2). ΔydeI strain infection elicited significantly reduced levels of these cytokines and chemokines compared to WT infection (see Fig. S3 in the supplemental material). The reduced inflammation in mice infected with the ΔydeI strain was in agreement with the attenuation of mutant virulence in the gut lumen and systemic organs of mice. In conclusion, YdeI was important for efficient systemic colonization and inflammation by S. Enteritidis inside the host.

FIG 6.

Deletion of ydeI attenuates systemic survival of S. Enteritidis in C57BL/6 mice and reduces inflammation in vivo. Streptomycin-treated C57BL/6 mice (n = 5) were infected intragastrically with ∼10⁷ CFU of the WT, ΔydeI mutant, cΔydeI complemented strain, and PBS (negative control). (A and B) Bacterial load in the feces was obtained at 24 h p.i. (A) and 48 h p.i. (B). (C to F) Mouse groups (n = 5) were euthanized, and the bacterial loads of the WT, ΔydeI, and cΔydeI strains in mesenteric lymph node (mLN) (C), spleen (D), liver (E) and cecal content (F) were determined at 72 h p.i. (G) Lipocalin-2 concentrations from feces supernatant of mice were monitored by enzyme-linked immunosorbent assay (ELISA). The horizontal bars indicate the median. (H) The cecal pathoscore was obtained by examining the hematoxylin and eosin (H&E)-stained cecal tissue sections from each mouse of all groups. a.u., absorbance units. Data are represented as mean ± SD. (I) H&E-stained representative tissue sections of cecum (size, 5 μm) from each group of mice showing induced cecal inflammation from infection with the WT, ΔydeI, and cΔydeI strains at 72 h p.i. (left to right). L, lumen; LP, lamina propria; S, submucosal edema. Bars, 200 μm. Statistical significance by Mann-Whitney U test (A to G) and one-way ANOVA (H): *, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001; ns, not significant (P ≥ 0.05).

YdeI interacts with OmpD in S. Enteritidis.

The three-dimensional (3D) structure of YdeI was predicted using Modeller 9 v. 12 software. The modeled YdeI protein (Fig. 7A) exhibits the conserved oligosaccharide/oligonucleotide binding-fold (OB-fold) domain, containing five strands of antiparallel β-sheet arranged in a closed β-barrel structure, which is capped by α-helix (24). The barrel-helix structure is an essential feature that is evolutionarily conserved across all members of the OB-fold family of proteins (25). Likewise, the structure of OmpD protein was modeled using Modeller 9 v. 12 (Fig. 7B). It has been shown that proteins spanning the outer membrane of Gram-negative bacteria fold into the conserved structural motif known as the transmembrane β-barrel (TMBB) (26). The predicted topology of the OmpD protein consists of a contiguous sequence of 17 antiparallel β-strands which transverse the membrane in a barrel-like conformation. The length of the transmembrane β-strands varies from 6 to 16 aa. These β-strands were found to be connected by short peptide segments that are often referred to as turns with lengths varying from 2 to 12 aa. The qualities of the predicted 3D models were checked with the validation tool PROCHECK. The PROCHECK results for OmpD (see Fig. S4A in the supplemental material) and YdeI (Fig. S4B) indicated the in silico structures had a consistent stereochemical quality of more than 90.9% (OmpD) and 79.1% (YdeI) of the residues in the allowed region of the Ramachandran plot. These results concluded that the modeled YdeI and OmpD structures were stereochemically favorable for further studies.

FIG 7.

In silico docking analysis of YdeI with OmpD of S. Enteritidis. (A and B) Ribbon models of YdeI (A) and OmpD (B). The modeled structures were created using Modeller v. 9.21 through homology modeling. (C and D) Stick representation displaying the interaction between YdeI and OmpD. The interacting residues of OmpD are shown in blue, and those of Ydel are in red. The electrostatic contacts are in yellow, the conventional H bonds are in green, the C-H bonds are in light green, and other nonbonded hydrophobic contacts are in pink. (E) Interaction of YdeI-OmpD after molecular dynamics (MD) simulation using GROMACS for 100 ns. (F) Root mean square deviation plot of YdeI and YdeI-OmpD after MD simulation up to 100 ns.

Following structure determination, molecular docking of YdeI and OmpD was performed to explore their potential binding interfaces (Fig. 7C). The binding interaction of the YdeI-OmpD docked complex is shown in Fig. 7D and E. The interacting H bond distances were observed within the range of 1.7 to 3.2 Å. The residue Lys₄₉ of YdeI protein made two H bonds with Asp₁₀₇ and Tyr₁₀₈ of OmpD. The remaining residues, like Tyr₁₀₈, Tyr₁₁₃, Tyr₁₅₅, Lys₂₉, Thr₄₄, Asn₁₀₉, and Arg₁₂₇ of YdeI, interacted with Ala₄₇, Asp₄₅, Tyr₄₁, Tyr₁₇₃, Tyr₁₅₅, Tyr₁₀₈, and Ser₁₁₀ of OmpD protein, respectively, through single H bonds (Fig. 7D). Molecular dynamics (MD) simulation was performed for 100 ns to evaluate the stabilities of the modeled YdeI protein and the docked YdeI-OmpD complex (Fig. 7F). After an initial steep rise in the root mean square deviation (RMSD) for the first ∼4 ns, the simulation systems YdeI and the docked YdeI-OmpD converged at a similar time to final stable RMSDs of 3.7 and 3.9 Å, respectively (Fig. 7F). As shown in Fig. 7E, four interactions, viz., Lys₄₉-Asp₁₀₇, Tyr₁₀₈-Ala₄₇, Lys₂₉-Tyr₁₇₃, and Arg₁₂₇-Ser₁₁₀, conserved their H bonds over the course of the 100-ns simulations. This demonstrated that these residues contributed significantly toward maintaining a stable interaction between YdeI and OmpD. In conclusion, the in silico data supported YdeI interaction with OmpD, which corroborated with the previously reported YdeI-OmpD interaction in S. Typhimurium (13).

YdeI exhibits phenotypes independent of OmpD during stress and virulence.

A phenotypic assessment of the ΔompD and ΔydeI ΔompD deletion mutants (Table 1) was conducted to evaluate the influence of YdeI-OmpD interaction on the stress response and virulence of S. Enteritidis. Being a general porin, OmpD primarily forms an outer membrane channel to transport cations and export small molecules like AMPs (13). Both ΔompD and ΔydeI ΔompD strains were sensitive to the action of PMBs, with average survival rates of 0.42 and 0.48% (P < 0.0001) (Fig. 8A), respectively, in comparison to the WT (72.5%). Similarly, on treatment with LL-37, the ΔompD and ΔydeI ΔompD strains had survival rates of 0.35 and 1.02% (P < 0.0001) (Fig. 8A), respectively, in comparison to the WT (72.3%). The cΔompD and cΔydeI ΔompD complemented strains showed significant restoration in survival compared to their corresponding mutants (P < 0.01). The ΔompD and ΔydeI ΔompD strains showed similar levels of decreased viability in response to Mg²⁺ starvation stress. The mutants displayed at least ∼2- to 3-fold (P < 0.0001)-reduced survival compared to the WT. The complementation of ΔompD and ΔydeI ΔompD mutants partially restored their phenotype, and their survival rates were similar to WT levels (P > 0.05) (Fig. 8C). Since the single and double deletion ΔompD and ΔydeI ΔompD strains were similarly sensitive to action of AMPs and Mg²⁺ starvation stress, this could be because YdeI and OmpD shared the same pathway(s) for conferring resistance to AMPs and Mg²⁺ starvation.

FIG 8.

Independent YdeI and OmpD phenotypes during stress and virulence in vitro. (A to D) Shown is the survival of log-phase cultures of the WT, ΔompD and ΔydeI ΔompD mutants, and cΔompD and cΔydeI ΔompD complemented strains during antimicrobial peptide stress (A), glucose starvation stress (B), Mg²⁺ starvation stress (C) and bile stress (D). (E) Toxicity of Cd²⁺ ions. The bacterial counts were calculated at the indicated time points. Relative survival of mutant and complemented strains was assessed against the WT. AMP stress survival is represented as a percentage. (F and G) HCT116 cell lines were infected with log-phase cultures of the WT, ΔompD, ΔydeI ΔompD, cΔydeI ΔompD, and ΔinvC strains at an MOI of 10 for the (F) adhesion assay and (G) invasion assay. Results are represented as percentages of adhesion and invasion for the ΔompD, ΔydeI ΔompD, and cΔydeI ΔompD strains in comparison to the WT value normalized to 100%. (H and I) RAW 264.7 cells were infected with the WT, ΔompD, ΔydeI ΔompD, and cΔydeI ΔompD strains at an MOI of 10. (H) Macrophage uptake assay. (I) Macrophage replication and survival assay. Bacterial counts were calculated by lysing of RAW 264.7 cells, serial dilution, and plating at 2 and 24 h of infection. Percentages of uptake for the ΔompD, ΔydeI ΔompD, and cΔydeI ΔompD strains were compared to WT uptake normalized to 100%. Intracellular survival of the ΔompD, ΔydeI ΔompD, and cΔydeI ΔompD strains at 24 h of infection is represented as fold replication and compared to the WT. Error bars indicate the mean ± SD from three independent experiments. Statistical significance by two-way ANOVA (A to E) and one-way ANOVA (F to I): **, P < 0.01; ****, P < 0.0001; ns, not significant (P ≥ 0.05).

YdeI had an independent role in promoting S. Enteritidis survival during glucose starvation, bile stress, and Cd²⁺ toxic conditions. The ΔompD mutant showed similar levels of viability to the WT under these stress conditions (Fig. 8B, D, and E). On the contrary, the ΔydeI ΔompD mutant displayed ∼1.7-fold (Fig. 8B) and ∼1.9-fold-decreased survival under glucose starvation stress and bile stress, respectively (Fig. 8D). The ΔydeI ΔompD strain showed 3-fold-decreased survival following treatment with 4 mg/ml CdCl₂ salts compared to the WT (Fig. 8E). Furthermore, there was no difference in the growth rate of mutants compared to the WT cultured in LB and M9 minimal medium (see Fig. S5 in the supplemental material). Thus, all the observed phenotypes were not due to growth defects in the strains. The above findings provided evidence that YdeI was an important determinant that conferred resistance to multiple stresses in S. Enteritidis. The phenotypes of YdeI and OmpD in S. Typhimurium and S. Enteritidis under different stress conditions are summarized in Table 2.

TABLE 2.

Phenotypic roles of YdeI and OmpD in S. Typhimurium and S. Enteritidis

Stress factor	Result fora :
	S. Typhimurium		S. Enteritidis
	YdeI	OmpD	YdeI	OmpD
Antimicrobial peptides	+*	+*	+*	+*
Glucose starvation	−	−	+	−
Mg2+ starvation	−	−	+*	+*
Bile salts	−	−	+	−
Cd2+ toxicity	−	−	+	−

^a+, promotes survival during stress; −, no role in stress response; *, phenotype possibly due to YdeI-OmpD interaction.

For in vitro assessment of virulence, the ΔompD mutant displayed no significant change in adhesion (Fig. 8F) and invasion (Fig. 8G) compared to the WT. However, the ΔydeI ΔompD mutant had ∼78% reduced invasion into HCT116 cells (Fig. 8G) and displayed no difference in its adhesion (Fig. 8F) compared to the WT. Additionally, the ΔompD strain displayed no significant difference in its uptake (Fig. 8H) and survival (Fig. 8I) at 24 h p.i. in RAW 264.7 cells, which is in agreement with the previously reported role of OmpD in virulence (27, 28). However, the ΔydeI ΔompD strain had ∼55% reduced uptake (Fig. 8H) and 4-fold attenuated intramacrophage survival (Fig. 8I) in comparison to the WT. Furthermore, the flow cytometric assessment of invasion (see Fig. S6A in the supplemental material) and intramacrophage survival (Fig. S6B) of mutants transformed with pCJLA corroborated well with the CFU assessment of virulence (Fig. 8G to I). From the above findings, it could be inferred that OmpD played no role in virulence and YdeI was essential in facilitating the virulence of S. Enteritidis.

DISCUSSION

The OB-fold family of proteins is ubiquitously conserved across Gram-negative pathogens, but their functions remain unclear. Several members of the OB-fold family have been acquired through horizontal gene transfer and encoded as prophage inserts. This association disposes the OB-fold family of proteins to have undergone rapid evolution in order to promote bacterial adaptation to different environments and pathogenesis (6). Recent studies have implicated critical roles for members of the OB-fold family like YgiW (VisP) protein in adaptation to vacuolar stress conditions and virulence, via interaction with LpxO and peptidoglycan subunits of S. Typhimurium (14). Interestingly, YdeI was previously reported to only confer AMP resistance via interaction with outer membrane porin OmpD but played no role in S. Typhimurium adaptation to other stresses (13). On the contrary, our findings revealed that YdeI conferred resistance to multiple stress factors, such as AMPs, glucose and Mg²⁺ starvation, and toxicity from bile salts and CdCl₂, in S. Enteritidis. The genomic differences between S. Typhimurium and S. Enteritidis could possibly explain these discrepant phenotypes across the two serovars.

Although the genetic data and docking analysis suggested that YdeI and OmpD may function together to confer protection against cationic AMPs and Mg²⁺ starvation in S. Enteritidis, it emerged that YdeI conferred resistance to other stresses via unknown mechanisms. It is noteworthy that the phenotypes of the ΔydeI mutant were similar to the previously reported observations in phoP mutants (29, 30). Additionally, our qRT-PCR and GFP reporter assays also showed ydeI was regulated by PhoP through activation of P_ydeI under different stress conditions. Moreover, there were significant changes in the expression of various PhoP-regulated genes in the ydeI mutant cultured under different stress conditions. This could be due to an epistatic interaction with PhoP or a pleiotropic effect due to genomic deletion of ydeI.

One of the critical findings of this study is downregulation of genes encoding the MgtBAC transporter, ethanolamine utilization (Eut) proteins, and the AcrAB efflux pump system in the ΔydeI mutant cultured under normal culture and stress conditions. Mg²⁺ transporters maintain cellular homeostasis of Mg²⁺ ions that serve as a checkpoint to maintain membrane integrity in response to several stresses and signal the transcriptional activation of virulence genes during infection (8, 31, 32). Their mutants have been previously reported to be attenuated for phagosomal survival, show increased susceptibility to AMPs, and display severe virulence defects in mice (8, 33–35). The Eut operon catabolizes ethanolamine, which serves as an important carbon and energy source for Salmonella survival in the anaerobic environments of the gut (36). Eut protein mutants were previously reported to have reduced survival in macrophages and were avirulent in mice (37, 38). Previous reports also showed that EutR directly activated SPI-2 gene expression to promote intramacrophage survival of Salmonella (37). Furthermore, AcrA and AcrB mutants with genes that encode RND efflux pumps were previously reported to be susceptible to sodium deoxycholate and β-lactam antibiotics (39, 40). Their overproduction contributed to β-lactam antibiotic resistance and adaptation to various environmental conditions in Salmonella (41). Downregulation of these genes in our study corroborated with the previously reported role of the MgtBAC transporter, Eut proteins, and AcrAB efflux pump in Salmonella stress response and virulence. Based on the above observations, we propose that YdeI could be regulating or interacting with either of these molecules besides OmpD to facilitate host-specific or niche-specific adaptation during Salmonella pathogenesis.

The reduced invasion into HCT116 cells and attenuated survival in RAW 264.7 cells of the ΔydeI mutant could be attributed to (i) downregulation of several T3SS-1 and T3SS-2 genes and (ii) increased susceptibility to the intracellular stress factors, such as AMPs, low Mg²⁺ ions, glucose limitation, and bile salts, that are likely to be encountered by Salmonella in the gut. Mutants with mutations of prg, hilA, sopE, sopD, and other T3SS-1 genes were previously reported to display reduced invasion and virulence in vivo (42, 43). Mutants with mutations of ssaE, ssaD, sseJ, and sseG and other T3SS-2 genes were also previously reported to be attenuated for phagocytic survival (44–46). Hence, a loss or downregulation of these previously reported T3SS-1 and T3SS-2 genes in the ΔydeI mutant may lead to reduced virulence of the ΔydeI strain.

The attenuation of ΔydeI mutant virulence in C57BL/6 mice may be due to reduced survival of the mutant in the gut lumen, resulting from the major clearance of the mutant load in the oral cavity, intestinal epithelium or Peyer’s patches enriched with CRAMPs (cathelicidin-related antimicrobial peptides; orthologs of human LL-37), and tissue-resident macrophages. Mutants with mutations of yejABEF and sapABCDF that encode ATP binding cassette transporters are sensitive to PMB and were reported to be avirulent in mice (47, 48). A similar outcome was observed in a mutant with mutation of a PhoP-regulated operon of pmrHFIJKLM, which is responsible for lipid A modification, which was sensitive to PMB in S. Typhimurium (49). Additionally, the mutant’s sensitivity to other tissue-specific stress factors and a replication defect inside macrophages could collectively contribute to the ΔydeI strain’s reduced colonization counts in the gut.

Our study explores some unanswered areas about the underlying mechanistic roles of YdeI and its potential partners to influence S. Enteritidis survival under various host stress conditions. It is well known that cell wall is an excellent target for development of new antibacterial targets because of its significant role in antibiotic resistance. YdeI as a periplasmic protein and its proposed association with crucial membrane components imply it is a promising target for antimicrobial intervention strategies. This may be achieved by either exploiting such targets to resensitize cells in the host system or by other approaches, such as disruption of outer membrane assembly or enhanced permeability to antimicrobial agents. The therapeutic relevance of OB-fold proteins has not been investigated and is a field worthy of more focused future studies.

MATERIALS AND METHODS

Conserved domain analysis and multiple sequence alignment of YdeI.

The YdeI protein (GenBank accession no. CAR33115.1) of Salmonella enterica serovar Enteritidis strain P125109 was identified as an ortholog of YdeI protein (GenBank accession no. HAD6491983.1) of Salmonella enterica serovar Typhimurium strain SL1344 through BLASTp analysis at NCBI (National Center for Biotechnology Information; https://www.ncbi.nlm.nih.gov). Pairwise sequence alignment between the two proteins was done using MAFFT v. 7 (http://mafft.cbrc.jp/alignment/server/). Synteny analysis was done using standard gene visualization tools at NCBI.

For multiple sequence alignment, homologous sequences to YdeI protein (GenBank accession no. CAR33115.1) from Salmonella enterica serovar Enteritidis strain P125109 were found using BLAST analysis against the NCBI nonredundant database (E value of 0.01). Domain conservation was validated through analysis of the sequence in the Pfam database (http://pfam.sanger.ac.uk/). Multiple alignment analysis was done using MAFFT v. 7 (http://mafft.cbrc.jp/alignment/server/). The alignments obtained were manually analyzed and edited by BioEdit software v. 7.1.

Bacteria, plasmids, and growth conditions.

Bacterial strains (Table 1) were grown under different culture conditions for subsequent experiments. M9 minimal medium (20% 5× M9 salts [211.3 mM Na₂HPO₄, 110 mM KH₂PO₄, 42.77 mM NaCl, 93.4 mM NH₄Cl], 2 mM MgSO₄, 0.1 mM CaCl₂, 0.4% glucose, 0.1% Casamino Acids) was used for bacterial stress response assays (16). For the invasion assay and SPI-1 gene expression, bacterial strains were grown in LB medium containing 0.3 M NaCl at 37°C under a static condition overnight (SPI-1 inducing medium) (50). Overnight cultures were then subcultured at a 1:20 ratio in SPI-1 inducing medium until an optical density at 600 nm (OD₆₀₀) of 0.6 was achieved. For the macrophage uptake assay, macrophage replication assay, and study of SPI-2 gene expression, bacterial strains were cultured in SPI-2 inducing medium [5 mM KCl, 7.5 mM (NH₄)₂SO₄, 80 mM MES, 38 mM glycerol, 0.1% Casamino Acids, 24 mM MgCl₂, 337 mM PO₄³⁻] (51) at 37°C and 150 rpm overnight.

For growth curve analysis, overnight cultures of bacterial strains were subcultured in LB and M9 minimal medium at a ratio of 1:100 at 37°C and 150 rpm. To obtain bacterial counts and OD, cultures were collected at hourly intervals from 0 to 10 h. Bacterial counts were obtained through serial dilution and plating on LB agar. Data are represented in CFU per milliliter, and the OD at 600 nm was obtained using an Agilent Cary 60 UV-visible light (UV-Vis) spectrophotometer.

The mutants used in this study were cultured with the antibiotics listed in Table 1, and the complemented strains were cultured with antibiotics: 50 μg/ml streptomycin (Str) and 100 μg/ml ampicillin (Amp). The strains transformed with plasmid pCJLA were cultured with antibiotics: 50 μg/ml Str and 20 μg/ml chloramphenicol (Cm). Lastly, the strains used for GFP reporter assays were cultured with antibiotics: 50 μg/ml Str and 100 μg/ml Amp.

Construction of isogenic mutant and complementation.

Chromosomal deletion of the genes ydeI and ompD from the genome of Salmonella enterica serovar Enteritidis strain P125109 (S. Enteritidis WT) was done through λ-red recombinase mutagenesis (52). Template plasmid pKD4 (aphT; kanamycin resistance cassette) was used for amplification of the product to chromosomally delete ydeI. Plasmid pKD3 (cat; chloramphenicol resistance cassette) was used to generate the amplicon to delete ompD and phoP. The ydeI and ompD double deletion mutant was achieved by transducing P22 phage lysate containing ompD::cat into the ΔydeI recipient strain. Primers used for construction and confirmation of the ΔydeI, ΔompD, ΔphoP, and ΔydeI ΔompD nonpolar deletion mutants are listed in Table S1 in the supplemental material. The mutants were confirmed by PCR using a reverse primer annealing to an antibiotic cassette and a forward confirmatory primer (Table S1) annealing 200 bp of the upstream region of the target gene. Mutants were confirmed by generating 1,200- and 1,000-bp amplicons for the kanamycin resistance cassette and chloramphenicol resistance cassette, respectively.

Plasmids pCH2189 (Table 1) and pCH1086 (Table 1) were constructed for complementation of mutants by manipulation of the pCH112 vector (50). Genes ydeI (size, 393 bp) and ompD (size, 1,089 bp), along with the 1,000-bp upstream sequence (native promoter), were amplified by PCR using the primers listed in Table S1. The amplicons and vector pCH112 were digested using restriction enzymes NcoI and XbaI (New England Biolabs [NEB], USA) according to the manufacturer’s instructions. The digested inserts of ydeI and ompD were cloned into the pCH112 vector, replacing the open reading frame (ORF) of hilA with the ORF of the target gene and its promoter sequence. The cloned constructs pCH2189 and pCH1086 were then transformed into the ΔydeI and ΔompD mutants, generating the cΔydeI and cΔompD complemented strains. The ΔydeI ΔompD mutant was transformed with pCH2189 to generate the cΔydeI ΔompD complemented strain.

Stress response assays. (i) Antimicrobial peptide sensitivity assay.

The AMP sensitivity assays were performed as previously described (14). Briefly, overnight cultures of the WT strain, ΔydeI, ΔompD, and ΔydeI ΔompD mutants, and the cΔydeI, cΔompD, and cΔydeI ΔompD complemented strains were subcultured at a 1:100 ratio in LB medium at 37°C and 150 rpm until an OD₆₀₀ of 0.4 (log phase) was achieved. Strains were diluted with 1× phosphate-buffered saline (PBS) to obtain 10⁸ CFU/ml. A total of 10⁸ CFU/ml of each strain suspension were treated with the AMPs PMB (1 μg/ml; Himedia, India) and LL-37 (10 μg/ml; Sigma) and incubated at 37°C for 1 h. CFU of strains were obtained by serial dilutions and plating on LB agar. Percentages of survival of strains were compared to WT survival normalized to 100%.

(ii) Glucose starvation assay.

The WT strain, ΔydeI, ΔompD, and ΔydeI ΔompD mutants, and cΔydeI, cΔompD, and cΔydeI ΔompD complemented strains were grown overnight in M9 minimal medium supplemented with 0.4% glucose and 0.1% Casamino Acids at 37°C and 150 rpm. The overnight cultures were subcultured at a ratio of 1:100 in M9 minimal medium at 37°C and 150 rpm until log phase. The log-phase cultures were then starved in M9 minimal medium containing 0.03% glucose for up to 4 h at 37°C and 150 rpm. Bacterial counts were enumerated through serial dilutions and plating on LB agar for up to 4 h (53). Survival of the mutants and complemented strains was calculated in log CFU per milliliter and compared against that of the WT at the indicated time points.

(iii) Magnesium ion starvation assay.

The magnesium starvation assay was performed as previously described (14). Briefly, the log-phase cultures of the WT strain, ΔydeI, ΔompD, and ΔydeI ΔompD mutant, and cΔydeI, cΔompD, and cΔydeI ΔompD complemented strains were transferred into a fresh M9 minimal medium containing 20 μM MgSO₄ (Mg²⁺ starvation condition). The strains were also grown in M9 minimal medium containing 200 μM MgSO₄, which served as the experimental control. Bacterial counts of strains after magnesium starvation were enumerated through serial dilutions and plating on LB agar for up to 4 h. Survival of the mutants and complemented strains was calculated in log CFU per milliliter in comparison to that of the WT at the indicated time points.

(iv) Bile stress assay.

Bile stress assays were performed as previously described (54). Briefly, the log-phase cultures of the WT strain, ΔydeI, ΔompD, and ΔydeI ΔompD mutants, and cΔydeI, cΔompD, and cΔydeI ΔompD complemented strains were incubated at 37°C for up to 4 h with bile salts mixture (15%; Himedia, India). Surviving bacterial counts were obtained by serial dilution and plating on LB agar for up to 4 h. Survival of the mutants and complemented strains was calculated in log CFU per milliliter and compared to that of the WT at the indicated time points.

(v) Cadmium ion toxicity.

Determination of cadmium salt (CdCl₂) toxicity was performed as previously described (14). Briefly, overnight cultures of the WT strain, ΔydeI, ΔompD, and ΔydeI ΔompD mutants, and cΔydeI, cΔompD, and cΔydeI ΔompD complemented strains were subcultured at a ratio of 1:100 until log phase. One milliliter of each culture was then incubated with 4 mg/ml CdCl₂ (Himedia, India) for up to 4 h. Bacterial counts after treatment with CdCl₂ were obtained by serial dilutions and plating on LB agar for up to 4 h. Survival of the mutants and complemented strains was calculated in log CFU per milliliter and compared to that of the WT at the indicated time points.

In silico identification of the ydeI promoter.

For identification of the promoter, a sequence 500 bp upstream of the transcriptional start site of ydeI was retrieved from the Salmonella enterica serovar Enteritidis strain P125109 genome (GenBank accession no. AM933172.1) and was analyzed using BPROM software (http://www.softberry.com/berry.phtml?topic=bprom&group=programs&subgroup=gfindb) (55). The predicted promoter of ydeI (P_ydeI) was subsequently cloned into pM968 between XbaI and PstI restriction sites to create the construct pM3968. pM3968 was subsequently transformed into the WT and ΔphoP mutant to generate the green fluorescent protein (GFP)-expressing WT/pM3968 and ΔphoP/pM3968 strains.

GFP reporter assay.

The WT/pM3968 and ΔphoP/pM3968 strains were cultured under normal and stress conditions as described above. Briefly, log-phase cultures of the WT/pM3968 and ΔphoP/pM3968 strains were subjected to AMP stress, glucose and Mg²⁺ starvation, bile salts, and CdCl₂ toxicity for 30 min. The bacterial cells were then centrifuged at 2,000 rpm for 5 min (Eppendorf 5424 R) and washed twice with 1× PBS to remove supernatant medium. The cells were then resuspended in BD fluorescence-activated cell sorter (FACS) sheath fluid for flow cytometry analysis using BD FACScanto II cytometer (Becton Dickinson, Erembodegem, Belgium).

Bacterial cells were gated using appropriate photomultiplier tube voltage settings for forward scatter (FSC) and side scatter (SSC). The GFP intensity of the WT/pM3968 and ΔphoP/pM3968 strains was acquired using an excitation wavelength of 460 nm and emission filter of 510 nm. The WT/pM968 strain (WT harboring non-GFP-expressing pM968) (56) was used as a negative control, while the WT/pM2155 strain (50) was used as the positive control for the study. All experiments were performed in biological triplicates and analyzed using Flowjo v. 10.4.2.

In vitro virulence assays. (i) Adhesion and invasion assays.

Adhesion and invasion assays were conducted as previously described (57). Briefly, the HCT116 colon epithelial cells were grown in Dulbecco’s modified Eagle’s medium (DMEM; Gibco, Germany), with 10% fetal bovine serum (FBS; Gibco) and antibiotics (antibiotic solution 100× liquid; Himedia catalog no. A001A) at 37°C and 5% CO₂. Twenty-four-well plates were seeded with HCT116 cells at a density of 2 × 10⁵, and the cells proliferated until they achieved confluence of 80%. The cells were incubated in DMEM without antibiotic before infection. The WT strain, ΔydeI, ΔompD, and ΔydeI ΔompD mutants, cΔydeI, and cΔydeI ΔompD complemented strains were cultured in SPI-1 inducing medium for infection, and bacterial counts were obtained by serial dilution and plating on LB agar to obtain the preinoculum density.

Adhesion assays were performed by incubating the bacterial inoculums and HCT116 cells on ice for 30 min. HCT116 cells were then infected with the WT, mutant, and complemented strains at a multiplicity of infection (MOI) of 10, followed by incubation on ice for 30 min. The cells were lysed with 0.1% sodium deoxycholate in PBS to release the adhered bacteria. Bacterial counts were obtained after serial dilutions and plating on LB agar. The percentage of adhesion was obtained by dividing the CFU recovered by preinoculum density, followed by multiplying the result by 100. The percentages of adhesion of mutants and complemented strains were compared to the WT value normalized to 100%.

For invasion assays, HCT116 cells were infected at an MOI of 10 and incubated for 50 min at 37°C and 5% CO₂. After 50 min, the HCT116 cells were washed twice with DMEM without antibiotic and incubated in DMEM containing gentamicin (100 μg/ml) for 2 h. Cells were washed twice with 1× PBS after gentamicin protection and lysed in PBS containing 0.1% sodium deoxycholate to release internalized bacteria. The number of invaded bacteria was obtained by serial dilutions and plating of the lysed cells on LB agar. The percentages of invasion of mutants and complemented strains were compared to that of the WT normalized to 100%.

(ii) Macrophage uptake and survival assays.

Macrophage uptake and survival assays were carried out as previously described (58). Briefly, 24-well plates were seeded with RAW 264.7 cells at a density of 2 × 10⁵. The cells were maintained in DMEM supplemented with 10% FBS and without antibiotic at 37°C and 5% CO₂ before infection. Bacterial strains were cultured in SPI-2 inducing medium and infected RAW 264.7 cells at an MOI of 10 for 50 min. The RAW 264.7 cells were then washed twice with DMEM without antibiotics and incubated in DMEM containing gentamicin (100 μg/ml) for 2 h. Bacterial uptake by macrophages was calculated by lysing cells with 0.1% Triton X-100 in PBS after 2 h of gentamicin protection. Bacterial counts were obtained by serial dilutions and plating on LB agar. The percentage of uptake was obtained by dividing the CFU obtained at 2 h by the number of bacterium-infected RAW 264.7 cells, followed by multiplying the result by 100. Uptake of the strains was compared to that of the WT normalized to 100%. For the macrophage survival assay, RAW 264.7 cells after 2 h of incubation in DMEM containing 100 μg/ml gentamicin were washed twice and incubated in DMEM containing 10 μg/ml gentamicin for 24 h. Survival after 24 h was obtained by lysing RAW 264.7 cells and plating onto LB agar after serial dilutions. Macrophage survival is represented as fold replication in comparison to WT survival at 24 h p.i.

(iii) Flow cytometer analysis.

The WT strain, ΔydeI, ΔompD, and ΔydeI ΔompD mutants, and cΔydeI, cΔompD, and cΔydeI ΔompD complemented strains were transformed with GFP-expressing plasmid pCJLA (Table 1) (53). The resulting WT/pCJLA, ΔydeI/pCJLA, ΔompD/pCJLA, ΔydeI ΔompD/pCJLA, cΔydeI/pCJLA, cΔompD/pCJLA, and cΔydeI ΔompD/pCJLA strains were assessed for their virulence in the HCT116 and RAW 264.7 cell lines by flow cytometric analysis as previously described (53). Briefly, HCT116 and RAW 264.7 cells were infected with the WT/pCJLA, ΔydeI/pCJLA, ΔompD/pCJLA, ΔydeI ΔompD/pCJLA, cΔydeI/pCJLA, cΔompD/pCJLA, and cΔydeI ΔompD/pCJLA strains at an MOI of 50. Infection was carried out by following the procedure of the standard gentamicin protection assay. Cell lines not infected with GFP-expressing bacterial strains were taken as uninfected and nonfluorescent controls for the experiments. The GFP-positive population was gated on the fluorescein isothiocyanate A (FITC-A) channel and is represented by a heat map plotting GFP intensity. The color scale for the GFP-positive population denoted various intensities of green fluorescence at the single-cell level. The mean intensity of GFP is represented as heat map FACS plots. Invasion, macrophage uptake, and bacterial replication data were obtained in terms of the percentage of mean fluorescence intensity (MFI) compared to cells infected with the WT/pCJLA control strain.

Isolation of bacterial RNA.

To study the role of PhoP in ydeI regulation, log-phase cultures of the WT and ΔphoP strains were treated with AMP stress, glucose and magnesium starvation stress, bile stress, and Cd²⁺ toxicity conditions. After 30 min of exposure to stress, bacterial cultures were centrifuged at 13,000 rpm for 1 min, and RNA was isolated from the obtained pellet of WT and ΔphoP cells using QIAzol lysis reagent (Qiagen, Germany). The isolated RNA was resuspended in nuclease-free water (Qiagen, Germany) and then treated with RNase-free DNase I (Thermo Scientific, catalog no. EN0521) to obtain pure RNA.

To study the expression of PhoP-regulated genes in the ΔydeI strain, the log-phase cultures of ΔydeI and WT cells were treated with AMP, glucose and magnesium starvation, and bile stress. After 30 min of exposure to these stresses, the cultures were centrifuged at 13,000 rpm for 1 min, and RNA was isolated from the bacterial pellet using QIAzol lysis reagent. The isolated RNA was resuspended in nuclease-free water and treated with RNase-free DNase I to remove DNA contaminant. The pure RNA was used for subsequent cDNA synthesis and qRT-PCR analysis. These experiments were performed in biological duplicates, and statistical analyses (t test) were performed to validate the fold expressions of the genes.

To study the effect of ydeI deletion on SPI-1 and SPI-2 gene expression, the ΔydeI and WT strains were cultured in SPI-1 and SPI-2 inducing media, and RNA was isolated using QIAzol lysis reagent as described above.

To study the expression of ydeI in the WT during infection, the HCT116 and RAW 264.7 cell lines were infected with the WT at an MOI of 50. DMEM was removed after infection, and the cells were washed with 1× PBS to remove noninternalized bacteria. The cells were recovered at the indicated time points (Fig. 4 and 5) and lysed with QIAzol lysis reagent to isolate RNA. The isolated RNAs were resuspended in nuclease-free water and treated with RNase-free DNase I to obtain pure RNA.

qRT-PCR analysis.

RNA was quantified using a Colibri microvolume spectrometer (Titertek Berthold) and normalized for cDNA synthesis. cDNA synthesis was done using the Verso cDNA synthesis kit (Thermo Scientific, catalog no. AB-1453/A). qRT-PCR was performed using the synthesized cDNA as the template and 2× DyNAmo ColorFlash SYBR green qPCR kit (Thermo Scientific, catalog no. F416-L) in a Realplex4 epgradient Mastercycler (Eppendorf). The primers used for qRT-PCR analysis are listed in Table S1. The 16S rRNA gene was used as the housekeeping gene in this study. Relative gene expression was calculated by the threshold cycle (ΔΔC_T) method. Genes with a log₂ fold change of ≥1.5 were considered upregulated, and those with a log₂ fold change of ≤−1.5 were considered downregulated, with P < 0.05 considered statistically significant.

Gene ontology analysis.

Gene ontology (GO) annotation of genes screened in the qRT-PCR study was performed using UniProt (http://www.uniprot.org/) based on biological process, molecular function, and cellular components (Data Set S1).

Ethics statement.

All of the mice were maintained and the infection experiment was carried out following the guidelines of the Institutional Animal Ethics Committee (IAEC), Kalinga Institute of Industrial Technology (KIIT) University, under license no. KSBT/IAEC/2017/MEET-1/A5.

C57BL/6 mouse infection experiment.

Six- to 8-week-old C57BL/6 specific-pathogen (SPF) free mice were housed in a ventilated cage facility at the School of Biotechnology, KIIT. Mouse groups (n = 5) were streptomycin pretreated for 3 days at a dosage of 50 mg through the orogastric route, as previously described (50), to clear resident microbiota for promotion of colonization of Salmonella and development of Salmonella-induced colitis in mice (59). The WT, ΔydeI, and cΔydeI strains were grown overnight in LB and subcultured at a ratio of 1:20 until an OD₆₀₀ of 0.6 was attained. Bacterial inoculum dosages of ∼10⁷ CFU were prepared to infect each mouse group (n = 5) through the orogastric route. Colonization counts were obtained from feces by serial dilutions and plating at 24 and 48 h p.i. on MacConkey agar plates. Feces were collected at 72 h p.i. from all mouse groups for lipocalin-2 estimation. Mice were euthanized at 72 h p.i. for examination of their organs: mLN, spleen, liver, and cecum. Bacterial loads in the systemic organs were obtained by serial dilutions and plating of the organ homogenates on MacConkey agar with the required antibiotics. Tissue segments of ileum, cecum, and colon were fixed and cryo-embedded in OCT (Sakura Finetek, Inc.) by snap-freezing in liquid nitrogen and were stored at −80°C.

Histopathological evaluation.

The cryo-embedded cecal segments were sectioned to a size of 5 μm at −30°C. The sections were obtained on glass slides and dried for at least 2 h at room temperature before staining with H&E stain. H&E images were acquired using a Zeiss Apotome.2 at a scale bar of 200 μm. The stained cecal sections were evaluated according to the previously described scoring system for analysis of cecal inflammation (50, 60). The H&E-stained sections from all mouse groups were assessed and scored based on the previously reported criteria (50).

Lipocalin-2 estimation.

Lipocalin-2 was measured using a commercially available Thermo Scientific Pierce mouse lipocalin-2 (LCN2) ELISA kit from fecal supernatants of mice infected with the WT, ΔydeI, and cΔydeI strains at 72 h p.i. Briefly, fecal samples were collected at 72 h p.i. and homogenized in 500 μl PBS. These samples were centrifuged at 7,000 rpm for 15 min, and the supernatants were collected. Fecal supernatants and standards were diluted according to the manufacturer’s instructions. ELISA was performed according to the manufacturer’s instruction. The final absorbance was acquired through a Thermo Scientific Multiskan GO Microplate spectrophotometer at λ_max values of 450 and 550 nm. The concentration of lipocalin-2 was obtained by interpolating the values from the standard plot and are expressed as log ng/g of feces.

Serum cytokine analysis.

A serum cytokine assay was performed on the blood serum of the mice (n = 5) infected with the WT or ΔydeI strain and PBS at 72 h p.i. The cytokine assay was conducted using a MILLIPLEX MAP mouse cytokine/chemokine magnetic bead panel-premixed 32 Plex immunology multiplex assay. The experiment was performed according to the manufacturer’s instructions, and data were acquired using the Bio-Rad Bioplex 200 system.

Prediction of structures of YdeI and OmpD.

The tertiary structure of YdeI protein was generated using the Modeller v. 9.21 program (61). For template selection, a BLASTp search was performed against the PDB database in addition to the iTASSER tool (62) to find X-ray crystallographic structures with the maximum identity and lowest E value. The crystal structure of YgiW from Escherichia coli was selected (PDB ID 1NNX_A; resolution, 1.45 Å) as the template for modeling with a query coverage of 84%, E value of 3e−15, and sequence identity of 35%. Two hundred distinct models were generated using the template in Modeller v. 9.21, which were subsequently evaluated based on their lowest discrete optimized protein energy (DOPE) score for quality. The best fit model with the lowest DOPE score was further subjected to loop and side chain refinement using GalaxySite (63) and What-IF (64). Similarly, the tertiary structure of OmpD protein was also generated using the Modeller v. 9.21 program (61). The template structure for modeling of OmpD was selected using BLASTp and DELTA-BLAST search against the PDB. Due to the presence of multiple templates for OmpD with high sequence coverage and less bias, four templates—1osm_A, 2xe5_A, 1pho_A, and 3hw9_A from PDB (see Table S2 in the supplemental material)—were selected for generation of the 3D model of OmpD. The best modeled structure was chosen based on the lowest DOPE score and lowest Cα-root mean square deviation (RMSD) with respect to the templates and subsequently refined using GalaxySite and What-IF tools. The YdeI model aligned well with 1NNX, with a root mean square (RMS) value of 0.53 Å. The OmpD model aligned well with the template structures 1OSM, 2XE5, 1PHO, and 3HW9 at RMS values of 0.65, 0.45, 0.53, and 0.59 Å, respectively.

Docking of Ydel and OmpD.

The modeled YdeI and OmpD proteins were docked using the ClusPro server (https://cluspro.bu.edu/login.php). Clusters with a maximum number of structures with least energy were selected as the top-ranked pose for further optimization using HADDOCK v. 2.2 (65). HADDOCK clustered 200 structures into a single cluster representing 100.0% of the water-refined models. The final docked protein complex was visualized using the PyMOL tool (http://www.pymol.org/) for analysis of the intermolecular interactions.

Molecular dynamics simulations.

To refine and validate the accuracy of the docked complexes, MD simulations were performed using GROMACS v. 2018.2 (66). All polar hydrogen atoms were added to the proteins, and the topologies of the proteins were prepared employing the GROMOS54A7 force field. Each system was solvated in a cubic box with a simple point charge water model. The distance between the solute and the box was maintained at 10 Å. The simulation systems were electro-neutralized with Na⁺ and Cl⁻ ions as well as by replacement of random water molecules. The system was subjected to equilibration post-energy minimization through the NVT and NPT ensemble for 1 ns each using the Leapfrog algorithm. After the equilibration stage, a production run of 100 ns was conducted with an integration step of 0.2 ns at constant pressure (Parrinello-Rahman method) and temperature (V-rescale method). All bond lengths were constrained using the LINCS algorithm (67), and the electrostatic interactions were calculated by using the particle mesh Ewald method (68). The GROMACS topologies for the ligands were generated using the Dundee PRODRG server (69). Trajectories obtained from various simulations were analyzed using the grace (http://plasma-gate.weiz-mann.ac.il/Grace/) and VMD (70) programs. Analysis of hydrogen bonds from the simulation trajectory was performed using the GROMACS g hbond utility with a cutoff distance of 3.5 Å. The RMSD of backbone atoms with respect to the initial conformation was calculated as a function of time to assess the conformational stability of the proteins during simulation studies.

Statistical analysis.

All experiments were carried out in triplicate, unless otherwise specified. Data are represented as the mean ± standard deviation (SD). One- and two-way analysis of variance (ANOVA), Student's t test, and the Mann-Whitney U test were used to determine statistical difference. All statistical analyses were performed using GraphPad Prism v. 7.0.