Hyperinvasiveness of Salmonella enterica serovar Choleraesuis linked to hyperexpression of type III secretion systems in vitro

Salmonella enterica serovars Choleraesuis and Typhimurium are among the non-typhoid Salmonella serovars that are important zoonotic pathogens. In clinical observation, S. Typhimurium typically causes diarrheal diseases; however, S. Choleraesuis shows high predilection to cause bacteremia. The mechanism why S. Choleraesuis is more invasive to humans remains unknown. In this study, we compared the S. Typhimurium LT2 and S. Choleraesuis SC-B67 proteomes through stable isotope labeling of amino acid in cell culture (SILAC). In SILAC, the expression of many virulence proteins in two type III secretion systems (T3SSs) were significantly higher in S. Choleraesuis than in S. Typhimurium. Similar differences were also found at the transcriptional level. Compared to S. Typhimurium, S. Choleraesuis showed a higher penetration level to Caco-2 (>100-fold) and MDCK (>10-fold) monolayers. In mice after oral challenge, the invasion of spleen and liver was also higher in S. Choleraesuis than in S. Typhimurium. The transcription of hilD in S. Choleraesuis was increased in physiological (1 mM) or high (10 mM) concentrations of Mg2+, but not in low (8 μM) concentration. We conclude that S. Choleraesuis showed hyperinvasiveness in cellular as well as mouse models due to hyperexpression of T3SS genes.

Penetration to epithelial cells. S. Choleraesuis penetrated the MDCK (10-fold) and Caco-2 cell (1000-fold) monolayers at higher efficiency than S. Typhimurium (Fig. 1) at 3 hours post-infection. Even at 6 hours after infection, the number of bacteria penetrating the monolayer was still higher in S. Choleraesuis than in S. Typhimurium, especially in Caco-2 cells (Fig. 1). S. Choleraesuis clinical isolates also showed higher penetration ability than S. Typhimurium clinical isolates (Supplementary Figure 5).
Intra-macrophage survival. We compared the intra-macrophage survival between S. Typhimurium and S. Choleraesuis by using THP-1 cells. The result showed that S. Choleraesuis displayed at least 4-times higher survival than S. Typhimurium inside macrophages (Fig. 2). S. Choleraesuis clinical isolates also showed better intra-macrophage survival ability at 4 hours and 6 hours post-infection than S. Typhimurium clinical isolates (Supplementary Figure 6).

Mouse experiments. S. Choleraesuis showed higher invasion and intracellular survival compared to
S. Typhimurium in cellular experiments. We then compared the pathogenesis of the two serovars using a murine model. One day post infection, S. Choleraesuis was found in spleen of one mouse (1/10) (Fig. 3A). No bacteria (less than 10 CFU/g) was found in spleen and liver of other mice ( Fig. 3A and B). At 3 days post-infection, bacteria was recovered from spleen and liver at least 10 CFU/g in seven mice infected by S. Choleraesuis, but only in three mice infected by S. Typhimurium ( Fig. 3A and B). Both serovars were recovered from the spleen and liver at 5 days post-infection. However, the bacterial numbers recovered were approximately 1000-times higher in S. Choleraesuis than in S. Typhimurium at 5 days post-infection ( Fig. 3A and B). To further confirm the difference, we co-infected the mice with S. Typhimurium and S. Choleraesuis. After 5 days, S. Typhimurium was significantly outnumbered by S. Choleraesuis in liver and spleen (Fig. 3C).
Expression of hilA and hilD in different hilE complemented strains. HilE is a negative regulator protein for SPI-1 T3SS virulence factors 17 . The predicted translation start site of hilE in S. Choleraesuis and S. Typhimurium is different due to the presence of an additional nucleotide between the two start codons in S. Choleraesuis (Supplementary Figure 7). HilE becomes a pseudogene in S. Choleraesuis if it uses the first start codon. The same difference was also found in the clinical S. Choleraesuis isolates (data not shown).We tried to construct the hilE deficient strain in both S. Typhimurium and S. Choleraesuis. However, all the S. Choleraesuis are multiple drug resistant strains, we do not have suitable selection markers to construct mutant strains. To analyze the function of HilE in S. Choleraesuis, we transformed a plasmid harboring hilE sequence from S. Choleraesuis into hilE-deficient strain in SL1344. The expression of hilA and hilD was very low in the deletion mutant carrying the hilE gene from either S. Typhimurium or S. Choleraesuis (Fig. 4). In the intra-macrophage survival assay, both hilE complemented strains showed poor survival (Fig. 5). The results indicated that the hilE gene from either S. Typhimurium or S. Choleraesuis was functional and thus can complement the hilE deletion to suppress the expression of the SPI-1 T3SS.
Hypersensitivity to Mg 2+ in S. Choleraesuis. Environmental signals, like oxygen, Mg 2+ and Fe 3+ , also control the SPI-1 T3SS expression 11,15,18 . We checked whether or not the environmental signals were the cause for differential gene expression between S. Typhimurium and S. Choleraesuis. Under both high and low oxygen conditions, the level of hilD were higher expressed in S. Choleraesuis SC-B67, relative to S. Typhimurium (Supplementary Figure 8). Mg 2+ is a ligand bound to PhoPQ TCS. Under low Mg 2+ condition, PhoPQ TCS is activated to suppress hilA and hilD expression; it is contrary under high Mg 2+ condition 15 . In low (8 μ M) Mg 2+ condition, the expression of hilD was low in both serovars (Fig. 6A). Interestingly, hilD expression was increased in S. Choleraesuis SC-B67 under normal (1 mM) and high (10 mM) Mg 2+ condition (Fig. 6A). PhoPQ TCS in S. Choleraesuis SC-B67 appeared hypersensitive to environmental Mg 2+ , leading to higher expression of HilD.  To support this, we tested expression patterns of the PhoP-activated genes (Pags) under different Mg 2+ concentrations in the two serovars. Under low Mg 2+ condition, the expression of pagC was similar between both serovars, but it was 5.4-fold lower in the S. Choleraesuis than in S. Typhimurium under normal Mg 2+ condition, suggesting S. Choleraesuis was hypersensitive to Mg 2+ to cause HilD activation (Fig. 6B).   Discussion Systemic approaches, like genomic, transcriptomic, and proteomic methods, have been used to compare the difference among different serovars of S. enterica [19][20][21] . In genomic studies, S. Typhimurium and S. Choleraesuis all contained well-defined virulence genes in SPI-1 and SPI-2 19 . It is hard to use sequence comparison to explain why S. Choleraesuis caused more bacteremia than other serovars in humans. Previous studies compared the protein expression pattern between S. Typhimurium and S. Choleraesuis through 2-dimensional SDS-PAGE analysis 21 , indicating that only one enzyme, GabD, showed different expression level between the two serovars 21 . GabD is succinate semialdehyde dehydrogenase I and the expression level of GabD appeared not directly related to invasiveness of S. Choleraesuis. In this study we used proteomic approach, SILAC, to analyze the differential protein expression between S. Typhimurium and S. Choleraesuis. The two serovars shared 4021 proteins in common. A total of 1731 proteins were indentified, and more than 70% of the identified proteins did not express at different level between the two. Many flagella synthesis and chemotaxis response proteins were expressed higher in S. Typhimurium, suggesting that motility of S. Typhimurium may be better than S. Choleraesuis, as previously described 3 . Some proteins from predicted pseudogenes in S. Choleraesuis were also detected in our analysis. This may be because Salmonella pseudogenes could undertake a recoding mechanism 22 . Most of the genes expressed higher in S. Choleraesuis SC-B67 are metabolism-related genes. Interestingly many virulence factors, like those in SPI-1 and SPI-2, were higher expressed in S. Choleraesuis. The result might explain why S. Choleraesuis is more invasive than S. Typhimurium. Salmonella encounter different acid stress environment from gastric acid (pH 2.0~3.0) to phagosome (pH 5.5~6.0) during its infection route 23 . Acid tolerance response is important for their virulence 24 . S. Choleraesuis was more tolerant to the SGF than S. Typhimurium. When S. Choleraesuis achieves the intestine lumen, they have to pass through the intestinal mucosal barrier before they establish systemic infection. In terms of the non-polar or polar epithelial cell invasion, S. Typhimurium and S. Choleraesuis were almost the same. However, all the S. Choleraesuis strains showed better penetration than the S. Typhimurium. Epithelial cell invasion indicates whether bacteria can entry the cells by their own ability. Monolayer cell penetration is more complicated than epithelial cell invasion. Besides entry into epithelial cells, bacteria have to have transcytose or disrupt tight junction to pass through the cellular barrier 25,26 . SPI-1-and SPI-4-encoded genes have been proved to play a vital role in invasion of epithelial cells [27][28][29] . These effectors also can disrupt the structure and function of tight junction, thereby helping Salmonella to penetrate intestine for further disseminatioin 29,30 . In our study, the SPI-1 and SPI-4 genes were shown highly expressed in S. Choleraesuis in SILAC data. High expression of these proteins might contribute to S. Choleraesuis to breach the intestinal barrier to further cause systemic infection.
After penetrating the intestine barrier, survival in the macrophage is an important issue for Salmonella to cause systemic infection. S. Choleraesuis showed better intra-macrophage survival than S. Typhimurium. SPI-2 T3SS and spv genes play a critical role in intracellular survival and systemic infection [31][32][33] . Our results showed that the level of proteins encoded in SPI-2 and spv was expressed at least 5-fold higher in S. Choleraesuis. This may explain why S. Choleraesuis had better intra-macrophage survival than S. Typhimurium.
HilA is a key positive regulator that directly binds the promoter of the structure and effector genes in SPI-1 T3SS 34 . HilA is negatively regulated by HilE which binds to HilD to inhibit the transcription of hilA 17 . In the genomic analysis of S. Choleraesuis, hilE was predicted as a pseudogene. We found in this study that the plasmid-harboring S. Typhimurium or S. Choleraesuis hilE ORF could restore the hilE function in S. Typhimurium hilE deletion strain, meaning that the HilE in S. Choleraesuis could still be functional. Recently Nuccio and Baumler analyzed different Salmonella serovars with their genome sequences available in NCBI 35 . They redefined some pseudogenes to normal genes, including hilE in S. Choleraesuis 35 . Combining the genomic and functional assays, the hyperexpression of SPI-1 T3SS in S. Choleraesuis found in this study appears not related to hilE, which previously was thought to be a pseudogene.
Environmental signals, such as oxygen, Mg 2+ , and osmolarity, also regulate the SPI-1 T3SS expression. High oxygen concentration is a negative regulator for hilA expression. High oxygen inhibits the expression of hilD which down regulates hilA 11 . We tested the sensitivity to oxygen of the two serovars. The expression of hilA and hilD in S. Choleraesuis was higher in high and low oxygen conditions (data not shown). Thus, oxygen appears not the factor that causes SPI-1 hyperexpression in S. Choleraesuis. On the other hand, PhoPQ TCS negatively regulates SPI-1 T3SS after sensing Mg 2+ in the environment. When Salmonella invades into cells, it encounters a low pH and Mg 2+ environment. Low Mg 2+ activates the PhoPQ TCS to suppress SPI-1 T3SS 36 . However, under high Mg 2+ condition, Mg 2+ binds to PhoP, which in turn activates its phosphatase activity to remove the phosphate from autophosphorylated PhoP 37 . The expression of hilA and hilD showed significant difference in the two serovars under low Mg 2+ concentration. Interestingly, hilD in S. Choleraesuis was expressed higher at normal and high Mg 2+ concentrations. Moreover, the downstream gene, pagC, was less expressed in S. Choleraesuis. Taken together, S. Choleraesuis seemed to be more sensitive to Mg 2+ . High concentration of Mg 2+ might cause more PhoP dephosphorylation, which subsequently inhibits the PhoPQ to suppress the hilA and hilD, or to activate the pagC in S. Choleraesuis. In conclusion, this study provides sufficient in vitro evidence to support that S. Choleraesuis is more invasive than S. Typhimuriunm by hyperexpressing T3SS virulence genes. These findings are consistent with the clinical observation that S. Choleraesuis is among the non-typhoid Salmonella serovars more invasive to the host.

Materials and Methods
Bacterial strains, plasmids, and mutagenesis. All Salmonella strains and plasmids used in this study are listed in Table S1. Bacteria were cultured on the Luria-Bartani (LB) agar at 37 °C, with added ampicillin (100 μ g/ml), kanamycin (50 μ g/ml), and tetracycline (12.5 μ g/ml) when appropriate. A λ-red recombinase mutagenesis method was used for Δ hilE mutant construction, as previously described 38 . The primers used to amplify pKD4 kanamycin resistance gene are listed in Table S2. To construct the hilE complement strain, primers to amplify the hilE-contained the putative promoter region from the S. Typhimurium and S. Choleraesuis, respectively. PCR products were digested by BamHI and EcoRI at 37 °C for 1 hour. Digested fragments were ligated into pBR322. The plasmids harbored the hilE from two serovars were then transferred to the Δ hilE mutant strains by electroporation.

SILAC proteome analysis.
To compare the protein expression pattern between S. Typhimuriuma LT2 and S. Choleraesuis SC-B67, we used the SILAC method as previously described 22 . Briefly, S. Typhimurium and S. Choleraesuis were cultured to late log phase in the SILAC medium (Mg 2+ , 0.4 mM) supplement with different isotope form of amino acids. Equal numbers of the two serovars were mixed together for further protein extraction by sonication. Protein concentration was measured by BCA method (Thermo). 40 μ g of protein sample were separated by 2D-SCX/RPLC system (Dionex). Separated samples were analyzed by a LTQ-Orbitrap hybrid mass spectrometer (Thermo). Raw peptide sequences collected from the LTQ-Orbitrap were analyzed by Mascot v2.3 and MaxQuant v1.2 (Matrix Science). The sequences of S. Typhimurium (NC_003197) and S. Choleraesuis (NC_004631) were downloaded from NCBI and used as references for comparison. All experiments of SILAC analysis were repeated twice. qPCR validation. To test that the mRNA expression was consistent with SILAC results, Salmonella strains were cultured in SILAC media to late log-phase. Wild type, Δ hilE, and hilE complement strain of SL1344 were cultured in DMEM (Mg 2+ , 0.8 mM) or RPMI (Mg 2+ , 0.4 mM) media to late log-phase for analyzing the expression of hilA and hilD. To analyze the hilD expression under different oxygen level, SL1344 and SC-B67 were cultured in DMEM with (high oxygen) or without (low oxygen) shaking to late log phase. To compare the hilD and pagC expression level in different Mg 2+ concentration, SL1344 and SC-B67 were inoculated in the N-salt media with 8 μ M (low), 1 mM (normal), and 10 mM (high) of Mg 2+ as previously described 39 . Bacteria RNA were isolated by TRIzol reagent (Invitrogen) according to the phenol-base method. RNA were treated with DNase I (Fermentas) 20 min at 37 °C to remove genomic DNA and then purified using a RNA clean up kit. Before converting RNA to cDNA, 1 μ g of total RNA were used in PCR reaction to make sure that DNA has been removed. 1 μ g of total RNA were reverse transcribed to cDNA by ToolsQuant II Fast RT kit (Tools Biotechnology Co., Ltd.). qPCR experiments were performed with SybrGreen Supermix (Bio-Rad) in an iCycler iQ5 (Bio-Rad) instrument. Expression of each gene was normalized to that of 16S rDNA. All the primers used in the experiments were listed in Table S2. All the qPCR experiments were done in triplicate in each independent experiment. Acid tolerance assay. S. Choleraesuis SC-B67 and S. Typhimurium SL1344 were precultured in LB broth and then 1 ml of the culture was transferred to the tube containing 4 ml of simulated gastric fluid (SGF, pH 2.0). Suspensions were cultured in the 37 °C incubator without shaking and survival was monitored every 5 mins. SGF contained 8.3 g/L proteose-peptone (Sigma), 3.5 g/L D-glucose (Sigma), 2.05 g/L NaCl (Sigma), 0.6 g/L KH 2 PO 4 (Sigma), 0.11 g/L CaCl 2 (Sigma), 0.37 g/L KCl (Sigma), 0.1 g/L lysozyme (Sigma), and 13.3 mg/L pepsin (Sigma) 40 . Final pH was adjusted to 2.0 with sterile 6.0 M HCl (Sigma). The recovered bacteria were counted by serial plating on LB agar at each time point with appropriate dilutions in PBS. Because the incubation period was short, experiment was done one group in each independent experiment. The experiment was repeated three times independently. We combined three independent experimental results for further statistical analysis.
Invasion assay. MDCK cells were used to generate polarized epithelial cell monolayer. 1 × 10 6 cells were cultured in 6-well plate for 5 days 29 . MDCK and HeLa cells, polar and non-polar, respectively, were infected by S. Thphimurium and S. Choleraesuis with MOI 100 as previously described 29 . At 1 hour post-infection, cells were washed with PBS and treated with gentamicin (75 μ g/ml) for 30 mins. Cells were lysed by lysis buffer (0.5% Triton X-100 in PBS pH 7.4) and plated on LB agar with appropriate dilutions. Invasion experiments were done in triplicate in each independent experiment. Penetration assay. To measure the penetration ability of different strains of Salmonella, penetration assays were performed by using MDCK and Caco-2 cell monolayer with MOI 100, as previously described 41 . Penetrated Salmonella organisms were retrieved from the basolateral medium and were plated onto the LB agar media at 1, 3 and 6 hours after infection. E. coli RDEC-1, a non-invasive strain, was used as a negative control in every experiment to ensure the integrity of the cell monolayers 41 . Experiments were done in triplicate in each independent experiment.
Intra-macrophage survival assay. Intra-macrophage survival of Salmonella in THP-1 cells was determined by a gentamicin protection assay, using methods described previously 9 . At 4 and 6 hours post-infection, intracellular bacteria were recovered with a lysis buffer (0.5% Triton X-100 in PBS pH 7.4) and were plated onto LB agar with or without appropriate antibiotics. Colonies were counted on the next day to calculate the intracellular survival rate. All the experiments were done in triplicate in each independent experiment.

Mouse infection model and competitive index assay.
Six-to eight-weeks-old female BALB/c mice were purchased from National Laboratory Animal Center, Taiwan. Mice were monitored daily during the experiments and sacrificed when it showed moribund or pain outcome. All the experiments were approved and followed the national animal care guidelines and the Institutional Animal Care and Use Committee (IACUC) of Chang Gung University (approval No CGU13-112). Salmonella were grown to the late-log phase. Bacteria were then washed twice and resuspensioned in saline. Six to ten mice were grouped and infected orally with S. Typhimurium or S. Choleraesuis (1 × 10 7 cfu/mouse). Organs were harvested at 1-, 3-, and 5-day(s) post-infection and homogenized in saline. Bacteria were cultured on LB agar with appropriate dilutions. In competition assays, six-to eight-weeks-old female BALB/c mice were infected by S. Typhimurium and S. Choleraesuis mixed culture (2 × 10 7 cfu/mouse). At 5 days post-infection, bacteria were recovered from the organs and plated onto LB agar media with or without chlorampnenicol (30μ g/ml) to differentiate between S. Choleraesuis SC-B67 (resistant to chlorampnenicol), and S. Typhimurium SL1344 (susceptible to chlorampnenicol). Competitive Index (CI) is defined as the ratio between S. Choleraesuis and S. Typhimurium within output divided by the ration within the input.

Statistical analysis.
All the experiments were repeated three times. All the quantitative data in this study was performed mean ± standard error. The statistical analysis was calculated by the sigma plot (version 10.0). Unpaired t-test was used to compare the results in the study.