C57BL/6 (B6) mice were purchased from Charles River Laboratories (Orient Bio Inc., Sungnam, Korea). All mice were maintained under specific pathogen-free conditions in the animal research center at the International Vaccine Institute (Seoul, Korea) where they received sterilized food and water ad libitum. All animal experiments described were approved by Institutional Animal Care and Use committees (IACUC PN2011-002). S. flexneri 5a (M90T), IpaB4 deletion mutant S. flexneri 5a (M90TΔIpaB4), wild-type Salmonella typhimurium (UK-1), and non-invasive Shigella strains (BS176) were used.

For oral infection, each mouse was orally administered 5×10⁹ bacteria. For ileal loop infection, mice were anesthetized and abdomens were surgically opened for about 10 mm. The ileum was looped with a suture about 40-mm long and then 5×10⁹ bacteria and/or 200µM sphingosine-1-phosphate (S1P; Avanti Polar Lipids, Inc., Alabaster, AL) were injected into lumen in the loop. The looped tissues were harvested and analyzed for histological changes.

Intestinal tissue was washed with PBS containing gentamicin and fixed in 4% formaldehyde for 1 hour at 4℃. The tissues were dehydrated by gradually soaking them in alcohol and xylene and then embedded in paraffin. The paraffin-embedded specimens were cut into 5-µm sections, stained with H&E or Alcian blue periodic acid-Schiff (PAS; Merck, Nottingham, UK), and viewed with a digital light microscope (Olympus, Tokyo, Japan). To detect green fluorescent protein (GFP)-expressing bacteria in the intestinal tissue, the frozen sections of ileal or colon tissues were prepared and viewed under a confocal scanning laser microscope (Carl Zeiss, Göttingen, Germany).

Histological changes of small intestine were assessed using H&E-stained samples of ileum after bacterial infection as previously reported (13). The category and parameters include host cell death (crypt, epithelium, lamina propria, and muscles), epithelium shedding, barrier integrity, inflammation, and goblet cell hyperplasia.

To assess the numbers of bacteria from intestinal tissues of non-infected and Shigella-infected mice, tissues were extensively washed in PBS with gentamicin (50µg/ml) to remove luminal and simply attached bacteria. Tissues were homogenized and plated onto TSB agar plates containing streptomycin since the M90T strain is streptomycin resistant. After overnight culture at 37℃, colonies were counted. For in vitro colony-forming unit (CFU) counts, a Caco-2 cell line originated from human intestinal epithelium was seeded at 5×10⁵/well in a 6-well plate. The next day Shigella strains were infected at 100 MOI (multiplicity of infection) in antibiotic-free culture medium for 1 hour. After extensively washing with culture media containing antibiotics to remove non-infected bacteria, Caco-2 cells were harvested at the indicated time point to evaluate CFU.

For real time PCR analysis, total RNA was extracted using an RNeasy kit (Qiagen, Hilden, Germany) and cDNA was synthesized by Superscript II reverse transcriptase with oligo(dT) primer (Invitrogen). Then, gene expression quantification was performed using an ABI PRISM sequence detection system (Applied Biosystems, Foster, CA). The levels of mRNA expression were displayed as the expression units of each target gene relative to the expression units of GAPDH. Murine Ppap2a (208 bp, forward: CTTACGACCCATGCTCCAGT, reverse: CGTGTGTGGATCCTCTTCCT), murine Sgpl1 (169 bp, forward: CTGATGACCGTACCATGTGC, reverse: GAATCCCTGAGAAGGGGAAG), human S1P lyase (HS1PL, 244 bp, forward: CTGGAGCACCCATTTGATTT, reverse: TCTCACCGAAGTGCATCAAG), human S1P phosphatase 2 (hS1PP2, 164 bp, TCCTCTTGGTTCGTCAGCTT, reverse: CACAAAGGTTGTAGCGCAGA), human S1P phosphatase 1 (hS1PP1, 188 bp, forward: TATTGCGGATCCTCATAGGG, reverse: TGTGATGGAGAAACCAACCA), human sphingosine kinase 2 (hSK2, 200 bp, forward: TGGCAGTGGTGTAAGAACCA, reverse: CAGTCAGGGCGATCTAGGAG), human sphingosine kinase 1 (hSK1, 227 bp, forward: ACCCATGAACCTGCTGTCTC, reverse: CAGGTGTCTTGGAACCCACT). For gene chip analysis, total RNA was amplified and purified using the Ambion Illumina RNA amplification kit (Ambion, Austin, TX) to yield biotinylated cRNA according to the manufacturer's instructions. In total, 750 ng of labeled cRNA samples were hybridized to MouseRef-8 V2.0 expression bead array for 16~18 hours at 58℃, according to the manufacturer's instructions (Illumina, San Diego, CA). Array signal detection was carried out using Amersham fluorolink streptavidin-Cy3 (GE Healthcare Bio-Sciences, Little Chalfont, UK) as described in the bead array manual. Arrays were scanned with an Illumina bead array reader confocal scanner according to the manufacturer's instructions.

To investigate host responses in adult mice, 6~8-week-old B6 mice were challenged orally with M90T (5×10⁹ cfu), a wild type invasive strain of S. flexneri 5a. Consistent with results of a previous study (13), acute epithelial cell death followed by epithelial barrier breakage could be observed 1 hour after M90T infection in the terminal ileum of the small intestine. The damaged epithelium had mostly recovered 24 hours later (Fig. 1A). There was no inflammatory sign despite loss of barrier functions and entry of shigellae and commensal bacteria in the lumen. Infection with T3SS mutant strain M90TΔipaB4 produced no histological changes. Degranulation of secretory granules in the Paneth cells of crypt was seen from 1 to 3 hours after infection (Fig. 1B). To ensure that M90T infected by the route colonized, and not simply adhered to the small intestine, we tracked down GFP-expressing M90T in the small intestine. Of note, M90T-GFP bacteria were found on the epithelium and even deep within the crypts (Fig. 1C). We observed similar findings in the colon. There was only mild epithelial tissue damage 3 hours after oral M90T infection in the ascending and descending colon but no inflammation (Fig. 2A). Bacterial colonization was successful as indicated by detection of M90T-GFP in the epithelium and lamina propria of the colon (Fig. 2B). shigellae were detected in the colon and cecum 3 hours after infection and were completely cleared 24 hours later (Fig. 2C). These results indicate that shigellae successfully invade and induce host responses in the murine gut only during the acute phase and do not cause inflammation.

To investigate the reason why Shigella infection results in loss of gut epithelial barrier but does not induce inflammation, we compared gene expression profiles of non-infected and M90T-infected ileum of the small intestine. In contrast with the uninfected mice and those infected with the T3SS mutant strain M90TΔipaB4, many genes were dramatically changed (Fig. 3A). S1P-related genes such as sphingosine phosphate lyase 1 (Sgpl1) and phosphatidic acid phosphatase 2a (Ppap2a) were up-regulated from the M90T-infected group (Fig. 3B). These findings were confirmed by real time RT-PCR of M90T-infected ileal homogenates. To interpret Shigella-specific modulation of host gene expression, wild-type S. typhimurium (UK-1) was used. Spgl1 expression was up-regulated in both Shigella- and Salmonella-infected tissues; however, Ppap2a was up-regulated only in M90T-infected tissues (Fig. 3C). These results suggested that oral Shigella infection could modulate host gene expression, especially the S1P-related genes.

S1P is well known to induce immune responses by inducing lymphocyte egress from lymphoid organs (15). We hypothesized that shigellae may have the ability to modulate host S1P levels to suppress inflammation. To investigate this hypothesis, we used the Caco-2 cell line, a human colorectal epithelial cell line, and infected the cells with non-invasive or wild type Shigella strains. To validate bacterial infection, bacterial CFU were measured. At 2 hours after infection, Shigella organisms were amplified within the host cells (Fig. 4A). S1P lyase (S1PL) and S1P phosphatase (S1PP) break down the phosphate of S1P, while sphingosine kinase (SK) produces S1P from sphingosine (Fig. 4B). The mRNA expression of S1PL, S1PP1, and S1PP2 increased in Shigella-infected Caco-2 cells 3 hours after infection while SK2 levels decreased. There was no significant change of SK1 (Fig. 4C). This suggests that Shigella organisms down-regulate host S1P levels, leading to suppression of inflammation.

To confirm that down-regulation of S1P by shigellae suppressed inflammation, we examined murine ileum following M90T infection with co-administration of exogenous S1P. To maintain local high concentrations of S1P, murine ileum were looped and then injected with M90T and S1P. When injected with M90T only, there was massive cell death but no inflammation. However, when M90T were injected together with S1P, severe inflammation resulted with shortened villi, loss of crypt, thickened serosa, leucocyte infiltration, and loss of tissue integrity observed 1 hour after injection (Fig. 5A and 5B). Similar results were seen 24 hours after infection (Fig. 5A and 5C), suggesting that Shigella organisms actively regulate host S1P levels to inhibit inflammatory responses.