Data on the molecular response of the intestine to the food-borne pathogen Salmonella are derived from in vitro studies, whereas in vivo data are lacking. We performed an oral S. enteritidis infection study in Wistar rats to obtain insight in the in vivo response in time. Expression profiles of ileal mucosa (IM) and Peyer's patches (PP) were generated using DNA microarrays at days 1, 3, and 6 postinfection. An overview of Salmonella-regulated processes was obtained and confirmed by quantitative real-time PCR on pooled and individual samples. Salmonella-induced gene expression responses in vivo are fewer and smaller than observed in vitro, and the response develops over a longer period of time. Few effects are seen at day 1 and mainly occur in IM, suggesting the mucosa as the primary site of invasion. Later, a bigger response is observed, especially in PP. Decreased expression of anti-microbial peptides genes (in IM at day 1) suggests inhibition of this process by Salmonella. Newly identified target processes are carbohydrate transport (increased expression in IM at day 1) and phase I and phase II detoxification (decreased expression at days 3 and 6). Increase of cytokine and chemokine expression occurs at later time points, both in PP and IM. Pancreatitis-associated protein, lipocalin 2, and calprotectin, potential inflammatory marker proteins, showed induced expression from day 3 onward. We conclude that the in vivo gene expression response of the ileum to Salmonella differs to a large extent from the response seen in vitro.
- Salmonella enteritidis
- ileal mucosa
- Peyer's patches
- antimicrobial defense
the enteric pathogen Salmonella enterica serovar enteritidis is one of the main causes of gastrointestinal infection in Europe and the USA (16, 35). Incidence of infection is highest in children, elderly, and immuno-suppressed individuals. In severe cases, illness is the result of translocation from the intestine, mainly the distal ileum, into the bloodstream. Little is known of the response of intestinal cells to the Salmonella infection in vivo. Insight in the in vivo responses is required to assess the relevance of in vitro models and as a starting point to develop strategies to prevent infection, for example by dietary intervention. To obtain an overview of the gene expression response of the ileal mucosa (IM) to Salmonella, we used DNA microarray technology. This technology has been used to analyze the gene transcription response of epithelial cell lines exposed to various bacterial pathogens in vitro, such as Vibrio cholerae, Listeria monocytogenes, Shigella flexneri, and Salmonella (2, 13, 36, 49). These pathogens initiated an immune response in the intestinal cell lines within minutes to hours. However, in vitro studies do not necessarily represent the in vivo situation. The relevant time frame in vivo is days rather than hours. Furthermore, monocultures in vitro lack interaction with other cell types of the epithelial barrier. In vitro cells also lack the microflora and the mucus layer, which are essential components of host-bacterial interaction (19, 38, 48). Some of these limitations are overcome in ligated-ileal-loop models, where Salmonella is directly introduced into the ileal lumen. However, these loop models lack luminal contents, and the natural route of pathogenic delivery is omitted. As far as we know, no in vivo gene expression studies of the intestine have been performed after oral gavage of Salmonella.
Two distinct tissues in the ileum have been identified as targets for Salmonella entry into the host: the ileum mucosa (IM) and the Peyer's patches (PP) (10, 56). One of the major functions of the IM is to absorb nutrients. Its large surface primarily consists of absorptive enterocytes and secretory Paneth and Goblet cells. PP are immune tissues that are part of the gut-associated lymphoid tissue. M cells are located in the follicle-associated epithelium overlaying the PP that have a role in sampling of pathogens (7, 21). The initial site of Salmonella contact and subsequent translocation in the small intestine is not clear (17, 53). Because the primary site of response to Salmonella is not known in rats we decided to study the course of the molecular response of both target tissues to Salmonella in vivo. We chose Wistar rats for our studies, because these have successfully been used in mechanistic infection studies (3, 50). Since diet affects Salmonella colonization and translocation, we used a diet low in calcium and high in fat. This diet results in a low resistance to food-borne bacterial infections (4, 5). As in these studies, we used a dose of Salmonella that causes a self-limiting infection, which is most commonly observed in humans.
To examine time-dependent molecular responses of both IM and PP to oral S. enteritidis infection, intestinal tissue of infected and noninfected controls was collected at different time points postinfection (p.i.) (or sham treatment), and pooled samples were analyzed by microarrays. The microarray analysis provided an overview of Salmonella-targeted processes. In our analysis, we focused on processes represented by at least three genes showing more than a twofold variation from noninfected controls. A selection of genes from every regulated process was analyzed by quantitative real-time PCR of RNA samples from individual rats to obtain information about the interindividual biological variation.
MATERIALS AND METHODS
Animals, diet, and infection.
The experimental protocol was approved by the animal welfare committee of Wageningen University (Wageningen, the Netherlands). Specific pathogen-free male outbred Wistar rats (WU, Harlan, Horst, the Netherlands), 9 wk old, mean body weight 285 g, were housed individually in metabolic cages. All animals were kept in a temperature- (22–24°C) and humidity-controlled (50–60%) room with a 12-h light-dark cycle (lights on from 6 AM to 6 PM). Rats were fed a purified diet during the whole experimental period. Compared with the AIN-93 diet (39), diets were low in calcium (20 mmol CaHPO4·2H2O/kg) and high in fat content (200 g fat/kg) (51) to mimic the composition of a Western human diet. Food and demineralized drinking water were supplied ad libitum. Food intake was recorded every day and body weight every 2 days.
Both the control group and the infected group were comprised of 24 rats. Per section time point, eight rats of each group were killed to collect intestinal samples (described below in more detail). The animals were acclimatized to housing and diet for 11 days, after which they were orally infected with S. enteritidis (clinical isolate, phage type 4 according to international standards; B1214 culture of NIZO food research, Ede, the Netherlands). In the morning, half of the animals (n = 3 × 8) were orally infected by gastric gavage with 1 ml of saline containing 3 × 109 colony forming units (CFU) S. enteritidis. The other half of the animals (n = 3 × 8) were sham treated and received saline only. S. enteritidis was cultured and stored, as described earlier (5). Fresh fecal samples were collected on days 1, 2, 3, and 6 p.i. and analyzed for viable Salmonella by plating 10-fold dilutions in sterile saline on modified brilliant green agar (Oxoid, Basingstoke, UK) and incubating aerobically overnight at 37°C. Sulphamandelate (Oxoid) was added to the agar plates to suppress swarming bacteria, such as Proteus species. The detection limit of this method was 102 CFU/g fecal wet weight. Total 24-h urine samples were collected on the last day before and on 6 consecutive days after infection. Urine was preserved with oxytetracycline and frozen until analyzed for the nitric oxide metabolites nitrite and nitrate (NOx) by a colorimetric method (Nr. 1746081; Roche diagnostics, Mannheim, Germany). At 1, 3, and 6 days p.i., Salmonella-exposed rats and their corresponding controls were killed by carbon dioxide inhalation. The mesenteric lymph nodes (MLN), spleen, and liver were excised aseptically, weighed, homogenized (Ultraturrax Pro200, Pro Scientific Oxford, CT) in sterile saline, serially diluted, and plated to culture for Salmonella, as described above. The detection limit was 102 CFU/g tissue. The distal ileum (defined as the last 12 cm of the small intestine proximal to the cecum) was taken out. The three most distal PP of this intestinal segment were excised and weighed. To obtain IM, the ileum was then longitudinally opened and ileal contents were removed by a quick rinse in cold 154 mM KCl. Subsequently, the mucosa was scraped off using a spatula. The PP and IM were immediately frozen in liquid nitrogen and stored at −80°C for RNA extraction.
PP and IM scrapings were homogenized in liquid N2 using a mortar and pestle cooled with liquid N2 (Fisher Emergo, Landsmeer, The Netherlands). Total RNA was isolated from these homogenates using TRIzol reagent (Invitrogen, San Diego, CA) according to the manufacturer's instructions. Total RNA was purified using RNeasy columns (Qiagen, Chatsworth, CA). Absence of RNA degradation was checked on a 1% Tris-borate-EDTA buffer/agarose gel after a 1-h incubation at 37°C. The purity and concentration were measured with the Nanodrop (Isogen Life Science, Maarssen, The Netherlands). A260/A280 ratios were all between 2.08 and 2.10, indicating good quality of RNA.
For microarray hybridization, mRNA of eight rats per time point per treatment was pooled. Each pool consisted of equal amounts of RNA of IM or PP from each rat. Arrays were performed in duplicate. For this, RNA pools were split and separately reverse transcribed and labeled with Cy5. A standard reference sample, consisting of a pool of all RNA extracted from IM and PP, was labeled with Cy3. For each oligo array, 40 μg of total RNA was used to make Cy5- or Cy3-labeled cDNA. Total RNA was mixed with 4 μg T21 primer, heated at 65°C for 3 min (RNA denaturation), followed by 25°C for 10 min (primer annealing). cDNA was synthesized by adding 5× first-strand buffer (Invitrogen), 10 mM DTT, 0.5 mM dATP, 0.5 mM dGTP, 0.5 mM dTTP, 0.04 mM dCTP, 0.04 mM Cy5-dCTP or Cy3-dCTP, 1.2 U RnaseOUT, and 6 U SuperScript II reverse transcriptase to a total volume of 62.5 μl. The reaction was incubated at 42°C for 2 h. Purification, precipitation, and denaturation of the labeled cDNA were performed as described earlier (54).
Analysis of mRNA expression by oligo arrays.
The rat 10K oligoset (MWG-Biotech, Ebersberg, Germany) used consists of 50-mer rat oligonucleotides representing 9,715 rat genes, 100 replicate oligos (8 different genes, 12–14 replicas each), and 169 control oligos (MWG Biotech). The 10K rat MWG oligoset, together with an additional set of 104 50-mer oligos representing infection-related genes, were printed on Ultra Gaps slides (Corning) using the Microgrid II arrayer (BioRobotics, Cambridge, UK). After printing, microarrays were allowed to dry at room temperature. The microarrays were immobilized by ultraviolet cross-linking (120 mJ of ultraviolet energy). The microarray slides were prehybridized at 42°C for 4 h in prehybridization buffer containing 5× SSC, 0.2% SDS, 5× Denhardt's, 200 μg/ml herring sperm DNA, 50% formamide. After prehybridization, all slides were washed twice in MilliQ and once in isopropanol. After being washed, the microarray slides were dried by centrifugation (2 min, 2,000 rpm). Then the Cy5-labeled cDNAs of the Salmonella-infected and control groups were mixed 1:1 with the Cy3 reference-labeled cDNA (all in duplicate). Hybridization was performed in a Gene frame (Westburg, the Netherlands) in a volume of 150 μl. The microarrays were hybridized overnight at 42°C in a humid hybridization chamber. After hybridization, the arrays were washed and dried as described elsewhere (54). Arrays were scanned using the Scanarray Express HT (Perkin Elmer, Wellesley, MA) at a laserpower of 90% and a photomultiplier tube voltage of 55%.
The software package Array Vision (version 7.0, Imaging Research, Ontario, Canada) was used to extract data from the scanned images. Median density values and background values of each spot were extracted for both the (Cy5) and the reference samples (Cy3). Only the spots with average Cy5 and Cy3 values that were twofold above the background value were included in the data analysis. Of the 9,819 genes present on the array, 6,792 spots fulfilled this criterion. Data normalization was performed with the software package GeneMaths XT (Applied Maths, Sint-Martens-Latem, Belgium) as described elsewhere (37). The microarray data are deposited in ArrayExpress (http://www.ebi.ac.uk/arrayexpress; E-MEXP-636). The microarray data were analyzed using Microsoft Excel (fold change) and GeneMaths XT (principal component analysis, hierarchical clustering). All groups (controls and infected) were hybridized in duplicate. The few genes with a more than twofold difference between technical duplicates were excluded for further analysis. The noninfected control groups of each time point were hybridized separately. Since no differences between the different days were observed, as analyzed by PCA analysis, the controls were averaged as one group per tissue. Fold changes in transcription levels between Salmonella-infected and control samples were calculated from the mean signal values of infected samples of IM or PP on the different time points vs. the mean of the control IM or PP. Genes that changed more than twofold at one of the time points compared with either control IM or PP were selected for pathway analysis.
Processes were identified using statistical overrepresentation in Metacore (GeneGo, St. Joseph, MI), a highly curated web-based application for identification of gene ontology processes in input gene sets (32). The program uses annotation databases and creates a list of gene ontology processes that are ranked according to their P value. To assess whether processes were selected by chance, the Metacore pathway analysis was repeated with 15 random sets of 187 genes, and the average P values of each process, of all 15 sets, were used as a surrogate number for false discovery. Since only 30% of the genes were annotated to gene ontology processes, processes with a P value of <0.01 were manually supplemented with the remaining significant genes using biological databases (BIOCarta, Gene Ontology, GenMAPP, KEGG) and scientific literature. Genes with analogous function or unequivocally being part of the same functional process or pathway were included, whereas far-away members were excluded. To conclude that a biological process was differentially affected in infected vs. control, two criteria were used: 1) initial P value in Metacore had to be smaller than P < 0.01 and 2) at least three genes of that process had to be changed more than twofold on Salmonella infection. The use of two criteria for selection was used to prevent overinterpretation and thus possible misinterpretation.
Analysis of mRNA expression by quantitative real-time PCR.
Quantitative real-time PCR (Q-PCR) on individual samples and pooled samples was performed to confirm differences in mRNA levels. It was considered unnecessary to analyze all (3 × 8) noninfected control animals individually because the array data of the control groups killed at days 1, 3, and 6 were highly similar. Instead, nine control animals were randomly chosen for individual RT verification. One microgram of RNA of all individual samples was used for the cDNA synthesis using the iScript cDNA synthesis kit of Bio-Rad Laboratories (Veenendaal, The Netherlands). Real-time reactions were performed by means of the iQ SYBR Green Supermix of Bio-Rad using the MyIQ single-color real-time PCR detection system (Bio-Rad). Each reaction (25 μl) contained 12.5 μl of iQ SYBR green supermix, 1 μl of forward primer (400 nM), 1 μl of reverse primer (400 nM), 8.5 μl of RNase-free water, and 2 μl of diluted cDNA. The following cycles were performed 1 × 3 min at 95°C, 40 amplification cycles (40 × 10 s 95°C, 45 s 60°C), 1 × 1 min at 95°C, 1 × 1 min at 62°C, and a melting curve (80 × 10 s at 55°C with an increase of 0.5°C per 10 s). A negative control without cDNA template was run with every assay. The optimal melting point of dsDNA (Tm) and the efficiency of the reaction were optimized beforehand. A Tm of 60°C was chosen for all reactions, and a PCR efficiency of 90–110% (3.2 < slope > 3.8) together with a correlation coefficient of >0.99 were accepted. Data were normalized against the reference genes β-actin (Actb) and pleckstrin homology domain-containing family A member 3 (Plekha3). Primers were designed using Beacon designer 4 (Premier Biosoft International, Palo Alto, CA). For sequences, see supplemental table at the Physiological Genomics website. A standard curve for all genes, including reference genes, was generated using serial dilutions of a pooled sample (cDNA from all reactions). mRNA levels were determined from the appropriate standard curve. Samples with mRNA levels below the lowest standard value were given a value not lower than half the value of this lowest standard, corresponding to the detection level. Analysis of all individual samples was performed in duplicate. Statistical analysis of the data was performed in Prism 4 (Prism 4, GraphPad software, San Diego, CA) using Student's t-test. P < 0.05 was considered statistically significant, and P < 0.01 was considered highly significant.
Physiological response to Salmonella.
In agreement with previous studies, food consumption and growth of the Wistar rats were not affected by Salmonella infection. High translocation of Salmonella to MLN, but not the liver and spleen, was seen at day 1 (Table 1). This implies that, at day 1, Salmonella has already crossed the intestinal barrier. Spleen and liver showed colonization by Salmonella at later time points. NOx, as a parameter of systemic infection, was found to be increased from day 3 onward (Table 1).
Gene expression induced by Salmonella in IM and PP.
The oligonucleotide micorarray contained 9,715 genes. After hybridization of RNA from IM and PP, the expression of 6,792 genes was more than twofold above the background in one or both tissues. The expression of 187 genes (98 genes in IM, 128 genes in PP) was altered more than twofold in at least one of the Salmonella-infected groups compared with the uninfected controls (days 1, 3, and 6). For subsequent analyses, the noninfected controls of the three different section days were taken as one group, since their expression pattern fully overlapped. No differentially expressed genes (cutoff ratio of >1.5) could be identified when the three control groups were compared with each other (data not shown). At day 1 p.i., only small effects in Salmonella-induced gene expression were observed. Just five genes were affected more than twofold in the PP. A larger response was observed in IM, where 18 genes showed differential expression compared with controls. At days 3 and 6 p.i., a larger response was seen in both tissues, and, in contrast to day 1, the response in PP was stronger than the response in IM. Most affected genes showed increased expression by Salmonella during the course of infection. However, a notable portion of the genes in IM at day 6 p.i. showed decreased expression (Fig. 1). Some overlap in Salmonella-induced genes was seen in both tissues (Fig. 2). Assessment of epithelial, goblet, and Paneth cell-specific genes (data not shown) and several well known housekeeping genes (see Table 3) indicated that the observed gene expression data did not result from changes in cellular composition of the mucosa.
To further characterize tissue-specific and common responses, the set of 187 genes showing at least twofold increased or decreased expression compared with the noninfected control level were classified in biological processes. Processes with P < 0.01 were inferred to be meaningfully related to the Salmonella response. Several randomly selected sets of 187 genes were also classified into processes; for these sets, the significance for all selected processes, obtained by Metacore pathway analysis, was not significant (P > 0.05). Clearly, the significance of the processes identified using the 187 Salmonella-affected genes was much higher, making identification by chance highly unlikely.
To prevent the occurrence of false positive genes, and overinterpretation of biological processes affected by Salmonella, we focused on biological processes with at least three genes exceeding the cutoff of >2.0. Additionally, we observed that the genes within all functional groups, except for lipid and other transporters, showed a comparable pattern of expression (Table 3). Also, most processes showed a similar pattern of expression in both IM as well as in PP, which strongly indicates that these processes are truly affected by Salmonella.
The majority of the differentially affected genes could be grouped into the following processes: immune response, inflammation, antimicrobial defense, complement cascade, detoxification, transport, and extracellular matrix organization (Table 2). Genes belonging to these processes, but with a differential expression of 1.5- to 2.0-fold, are also included in the tables (Tables 2 and 3).
To confirm the Salmonella-induced effect on biological processes selected, we performed an independent array hybridization of freshly pooled ileal mucosal RNA from the same infection study. The results of this analysis confirmed all selected genes based on the two selection criteria applied: 1) twofold change cutoff; 2) at least three genes changed in a similar biological process. For the present study, this corroborates that the used selection criteria were robust and valid.
Immune activation, inflammation, and antimicrobial defense.
Genes encoding antimicrobial defense proteins defensin 5 (Def5), lysozyme (Lys), and matrilysin (Mmp7) showed a 1.7- to 2.0-fold decreased expression in IM at day 1 p.i. (Table 3). Salmonella did not stimulate the expression of genes related to the innate immune response and inflammation at day 1 p.i. The decreased expression of antimicrobial defense genes had mostly disappeared at days 3 and 6 p.i., whereas expression of those related to immune response and inflammation was clearly induced, with the strongest response observed at day 3. The PP showed a more pronounced immune and inflammatory response than the IM at days 3 and 6 p.i. (Table 3). Lipocalin 2 (Lcn2) and pancreatitis-associated protein 3 (Pap3) encode for inflammatory response proteins whose expression was affected in PP and IM. Furthermore, one activator of the complement cascade, tissue factor (coagulation factor) III, and three inhibitors of this cascade, Decay-accelerating factor (Daf), Cd59 (protectin), and serpinG (C1 inhibitor), showed induced expression on Salmonella infection in both tissues (Table 3).
A second group of affected genes were detoxification genes, which showed lower expression at days 3 and 6 p.i. (Table 3). This group consisted of both phase I and II genes. Phase I genes encoding cytochrome P450 1a1 (Cyp1a1), Cyp 2j4, Cyp 2c24, Cyp 3a9, Cyp 4f1, Cyp 17a1, epoxide hydroxylase 1 (Ephx1), and carboxylesterase 1 (Ces1) showed lower expression, with the exception of Cyp7b1 expression, which was higher. Expression of two phase II genes, encoding UDP-glucuronosyltransferase 1 (Ugt1) and Glutathione S-transferase alpha 2 (Gsta2), was also lower.
A third category of genes affected by Salmonella infection was transporters (Table 3). Three carbohydrate transporters, the apical located Sglt1 and Glut5 and basolateral located Glut2, showed higher expression, primarily at day 1 in both tissues studied. The other transporters that were affected showed diverse expression patterns (Table 3).
Validation of Salmonella-regulated genes in individual animals by Q-PCR.
To verify the Salmonella modulation of certain biological processes, we selected genes from every process for Q-PCR confirmation. Pooled Q-PCR analysis was performed on genes involved in antimicrobial defense (Def 5, Lys, Mmp7), chemotaxis (Groα), inflammation (Lcn2, Pap3), phase I detoxification metabolism (Cyp3a9, Cyp4f1, Cyp1a1, Ephx), and glucose transport (Slc5a1, Slc2a2, Slc2a5). For individual analysis, the genes Def5, Lcn2, Pap3, Cyp4f1, and Slc5a1 were chosen. β-Actin (Actb) and pleckstrin homology domain-containing family A member 3 (Plekha3) were selected as reference genes because they showed constant and treatment-independent expression in the array data (data not shown). All Q-PCR analyses on individual and pooled samples of both IM and PP confirmed the microarray data since relative gene expression changes were similar using both methods (see Table 4). In noninfected control animals, the expression of Groα and Lcn2 was close to background, which may have affected the precise fold change. Analysis of the individual samples by Q-PCR generally revealed large interindividual variation in gene expression within treatment groups (Fig. 3). The gene expression of several genes known to be affected by Salmonella in in vitro studies such as Tlr5 and the nuclear factor-κb (NF-κb) subunits RelA and P105 were not found to be affected based on array data. Q-PCR analysis confirmed the lack of induction in this in vivo study (Fig. 3).
Overall in vivo response.
In this study, we used microarrays to follow the Salmonella enteritidis-induced gene expression changes in the ileum of Wistar rats with time. In two target tissues, the IM and the ileal PP, only a very limited number of genes was changed at 1 day p.i. Altogether, only 0.2–0.9% of the genes in IM show a differential gene expression at day 1 through day 6 p.i. This contrasts with in vitro responses of human intestinal epithelial cells, where 5–35% of the genes were found to be affected within the first 20 h after infection with Salmonella compared with noninfected controls (13). Infection of intestinal cell lines (T84, CaCo-2) with other enteric bacteria for up to 3 h resulted in 4% differentially expressed genes compared with noninfected cells (36, 49). Moreover, not only the number but also the magnitude of most responses in our in vivo study are relatively small compared with gene expression changes in vitro, which may change up to 90-fold. Studies in bovine ileal loops infected with S. typhimurium showed an eightfold increased expression of pro-inflammatory chemokine and cytokine genes within 6 h p.i. (43). In fact, no correlation (r2 = 0.0004) was found if the genes that were found to be more than twofold altered in vitro in HT29 cells 3 h after infection with Salmonella (13) were compared with the in vivo response of the same 30 genes that were present in this study (see supplemental data available online at the Physiological Genomics website).
In vitro models can provide insight in mechanistic aspects of Salmonella-host interaction in vivo. However, the large gene expression responses observed in Salmonella infection studies with HT-29 cells (13) are in contrast with the limited gene expression response observed in Salmonella infected intact mucosa or PP presented in our in vivo study. The overlap in genes observed to change in in vitro studies (13) and our in vivo study is very small. Several chemokines (Mip2/Mip2α, Groα/Il-8) are induced in both type of studies, but even then the time frame (3 h in vitro vs. 3 days in vivo) and magnitude clearly differ (91.5- and 78.0-fold in vitro and 1.4- and 2.5-fold in vivo for Il-8 and Mip2α, respectively).
The in vitro response to Salmonella infection seems unphysiologically reactive, possibly due to the absence of a protective intestinal microflora, mucus layer, and mucosal secretion of antimicrobials. As a result of the absence of these barriers, the number of invasive Salmonella's per cell is likely much higher in vitro than in vivo, and this will affect the response of these cells.
Additionally, whereas in vitro homogeneous cell types are studied, a natural heterogeneous mixture of cell types is studied in vivo. This may affect the type and magnitude of the gene expression response. It should be realized that a response of a limited number of specific cells to Salmonella at early time points of the infection might have been missed in the present study due to possible dilution of these cells in the heterogeneous cell populations present in the ileal mucosal scrapings and PP. Identification of cell type-specific responses of, e.g., dendritic cells or other potential target cells can be addressed using laser microdissection to isolate a specific cell type before RNA isolation.
The absence of gene expression effects of expected chemokines and cytokines at day 1 in the present in vivo study and the larger gene expression effects in vitro and ex vivo are in line with differences in phenotypic, physiological observations. Where in vitro systems show massive cell death at 24 h and ileal loop models show epithelial detachment after 8 h of exposure to Salmonella (63), no inflammatory changes are seen at the first day after oral infection with Salmonella in vivo (44). This late in vivo response is also seen in our study, where NOx, a marker of the a-specific immune response, increased from day 3 onward. It should be noted that at day 1 day p.i. high Salmonella numbers were observed in the MLNs in the present and previous studies (4, 5), demonstrating translocation of Salmonella at this early time point.
That translocation of this pathogen did not provoke extensive early gene expression changes in vivo may indicate that Salmonella infection is a targeted and controlled process. In living organisms, gut epithelial cells are in continuous and intimate contact with gut bacteria. It is known that these host-microbe interactions are important for keeping inflammatory processes in check. The inflammatory response can be repressed by the microflora (9, 19, 24). Also, the surrounding host cells suppress signals, e.g., epithelial cell-derived factors influence dendritic cell responses, which may regulate the generation of an inflammatory response to bacteria (41). Communication and feedback mechanisms between different mucosal cell types help to maintain mucosal homeostasis. Cell lines in vitro miss contact with other cell types and the modulating effect that commensal bacteria may have.
Another possible, but less likely, explanation for the small gene expression changes observed in this in vivo study could be related to host specificity of Salmonella-induced responses. The pathogenicity of Salmonella serovars can be animal species-specific (18, 46, 55), and Salmonella in vitro studies are mostly performed in human cell lines (20). However, many aspects, such as time-course effects of Salmonella colonization and translocation, are largely similar in this rat model compared with humans (3, 5, 15).
Site of Salmonella invasion, mucosal glucose metabolism, and defense.
Based on the number of genes altered by Salmonella, the IM showed an earlier response than the PP. This may indicate that in rat the IM, and not PP, is the first site of interaction or invasion. Within this early response, we newly identified that Salmonella changed genes related to glucose metabolism. These glucose metabolism-related genes, sodium-dependent glucose transporter (Sglt1), fructose transporters Glut5 and Glut2, and sucrase-isomaltase (SI) (fold change 2.9 in IM at day 1 p.i.), are expressed in enterocytes, which implies that the enterocytes are the first contact or entry site of Salmonella infection. The expression of Sglt1, Glut5, and Glut2 in the PP most likely originates from enterocytes overlying the PP (45). The upregulated glucose transport may be triggered by a higher glucose need of infected cells, which has been reported in cells infected by chlamydia and viruses (34, 47). Altered cell metabolism may also explain the differential expression of other transporters (e.g., lipid transports Apob, Apoa1 and 4, E-Fabp, SLC10a2, peptide transporter Pept-1, and organic anion transporter Oat2). Salmonella decreased the expression of genes important for host defense against bacterial intruders [defensin 5 (Def5), lysozyme (Lys), matrilysin (Mmp7), and secretory phospholipase A2 (Pla2g2a)] at the early time point. Reduced expression of α-defensin and lysozyme was also reported in mice inoculated with S. typhimurium (42). Defensin, lysozyme, and matrilysin are expressed by Paneth cells located in the bottom of the mucosal crypts. These early changes suggest that, among the different cell types lining the IM (enterocytes, Paneth cells, goblet cells, and neuro-endocrine cells), the Paneth cells and enterocytes are a target for Salmonella. Based on these results, it seems that the IM is at least as important as the PP as the major site of early Salmonella invasion. This enterocyte-targeted invasion of Salmonella has also been reported in experiments with calves (14), pigs (27), and rabbits (57).
Phase I and phase II detoxification enzymes in both IM and PP showed decreased expression, coinciding with increased expression of inflammatory genes. This was not caused by dilution of epithelial cells, since the expression of I-Fabp, a control for epithelial content (31), showed no significant differential expression. One explanation for the decreased expression of detoxification genes might be that this allows optimal defense by immune cells. Downregulation of cytochrome P450 gene expression is known to be induced by inflammatory mediators such as reactive oxygen species, nitric oxide, IFN, or cytokines (Il-1, transforming growth factor-β) (29, 40) and is also observed in mucosal biopsies from IBD patients (26). Alternatively, the expression may be actively reduced by Salmonella, especially since the Ah receptor nuclear translocator (Arntl), which is the transcription factor regulating Cyp1a1, Ugt1a6, and Gsta2 (61), shows decreased expression at day 1 (fold change −2.1 in IM).
Innate immune response.
Despite clear translocation of Salmonella from the intestinal lumen to the MLN (Table 1), no increased expression of immune response genes was seen at day 1 p.i., neither in the PP nor in the IM (Tables 2 and 3). So far, most of the studies focusing on host gene expression responses on exposure to micro-organisms have been performed in vitro (20). At later time points, we observe in vivo some genes (Il-1α, Il-1β, Groα, Groγ, iNos) that are a confirmation of former in vitro infection studies (20), but we also observe that some genes well known to be upregulated by Salmonella in in vitro studies (62), such as Tlr5 and the NF-κb subunits RelA and P105, were not found to be affected based on array data. To exclude that the absence of differential gene expression was due to a technical issue, we examined the differential expression of Tlr5 by Q-PCR analysis in individual samples and confirmed the absence of differential expression (Fig. 3). Also, the main downstream signaling molecule, NF-κB p105 subunit, was not regulated (Fig. 3). This was also the case for the NF-κB relA subunit (data not shown).
Despite the absence of gene expression changes, this pathway seems to be activated, since we observed differential expression of targets of TLR and NF-κB activation, such as cytokine, chemokine, and inflammatory response genes (Il-6, Groα, β, γ, iNos, Cox2, Sod2) (22, 52), at the later time points. That the immune response genes could not be observed at day 1 may result from induction in a limited number of cells or specific cell types, which are diluted in the heterogeneous cell population of the mucosal scrapings and PP. Another explanation might be modulation at posttranscription level, which escapes detection at transcription level.
Salmonella invasion is characterized by recruitment of monocytes, neutrophils, and dendritic cells to the infected area (60). Indeed, later time points showed a prominent increase in the expression of genes involved in chemotaxis, including several CC chemokines (CCL3, CCL4) and several CXC chemokines (CXCL1, MIP-2, MIP-2B, Cinc-5, Cinc-10, Cinc-11). These genes were among the most highly differentially expressed genes in this study, particularly in the PP. Another prominent group of genes in PP are inflammation-related genes. The stronger induction of genes involved in chemotaxis and inflammation in PP compared with IM suggests either a higher Salmonella invasion of the PP at later time points or a stronger secondary response induced by more recruitment of leukocytes to this tissue.
Protective mechanisms against epithelial barrier disruption show differential expression at later time points. This late induction is most likely stimulated to limit inflammation-induced damage to the mucosa. This is best reflected in the increased expression of Daf and Cd59. These are inhibitors of the complement cascade and aid restoration of blood flow in the microvasculature.
At days 3 and 6 on Salmonella infection, expression of the general inflammatory mediators Cox2, iNOS and Sod2, Pap3, Lcn, and calprotectin (S100A8 and S100A9) was strongly increased. Calprotectin, Pap 3, and Lcn are all highly expressed in the chronically inflamed mucosa of inflammatory bowel disease (IBD) patients and in animal models of this disease (6, 12, 23, 33, 58). Obviously, generic mechanisms are involved in acute inflammation due to Salmonella infection and chronic inflammation in IBD despite different pathologies. Possibly, Pap, Lcn2, and calprotectin could serve as general markers for gastrointestinal inflammation. Since these markers are all secreted into stool, resistant to degradation by intestinal contents, and easily measured (6, 11, 30), they might be very useful to follow the course of an inflammatory period by noninvasive means.
Responses driven by Salmonella.
The present study focused on Salmonella-induced changes in intestinal processes. Some identified processes such as defense and immune response are a confirmation of known effects of pathogens on the host. But others, such as changes in detoxification and transporter genes, are not related to infection before. We cannot discriminate whether these induced changes are directly caused by Salmonella or secondary effects of infection-induced inflammation. Induction by Salmonella itself cannot be excluded, since it is known that microbes are able to induce host gene expression for their own benefit. Salmonella uses its own type III secretion system to alter host cell processes, including apoptosis, cytoskeletal function, and cytokine production (1, 25, 28). Other than pathogens, commensals can also actively induce host gene expression and thus affect important physiological functions (8, 9). Most of these results originate from in vitro studies. The in vivo relevance for (intestinal) host resistance and gut barrier functioning should be addressed in future studies.
Microarray analysis provided an overview of processes in rat ileum that are affected by oral infection of Salmonella. The processes that were obtained were not selected using random sets of 187 genes (data not shown), indicating that these processes are truly affected by Salmonella and not selected by chance. To further confirm the selection of processes, selected genes, representative of various physiological processes, were investigated by Q-PCR in individual rats. The individual genes showed a statistically significant change within the group (n = 8 rats), indicating that these genes are truly affected by Salmonella and not purely by chance. These analyses revealed a high variation in expression among different rats within a treatment group, a finding most likely resulting from the genetic heterogeneity of the outbred Wistar rats used in the present study. Use of inbred animals likely reduces heterogeneity but has the disadvantage that observed effects may be specific for a particular genetic background (59) and thus hamper translation to humans. The large interanimal variation observed has implications for future studies. Using the same setup, it will be difficult to identify differential responses below twofold. Such a twofold differential response may already constitute a relevant and large physiological response, especially if this occurs in several genes in the same pathway simultaneously. Despite the relevance of relatively small changes in gene expression, it will be technically difficult to study, e.g., the preventative effects of dietary components or the therapeutic effects of new drugs on infection-induced processes by current PCR- or array-based methods. To deal with this limitation and to confidently identify smaller effects, it would be best to include more animals per treatment group and to perform array hybridization of individual samples. This allows identification of differentially expressed genes, not only based on magnitude of change but also on statistical power.
In conclusion, the present in vivo study reveals that S. enteritidis induces small gene expression changes in the ileum of Wistar rats. Especially at day 1 p.i., a very limited response in gene expression was observed despite marked translocation of Salmonella to the MLN. Remarkably, the few early changes observed occurred in the IM. This may indicate that IM but not PP is the primary target for Salmonella translocation, at least in rats. The more various gene expression changes at days 3 and 6 p.i. were mainly observed in the PP and were related to immune cell recruitment and inflammation. Infection-induced inflammatory genes overlap with those reported to be upregulated in IBD and may thus be explored as general markers of intestinal inflammation. Finally, we newly identified that mucosal glucose metabolism and detoxification capacity are affected by Salmonella infection in the rat.
This study was funded by TI Food and Nutrition.
The authors thank the biotechnicians at the Small Animal Center of Wageningen University (Wageningen, The Netherlands) for expert assistance. We thank Professor Martijn Katan (Wageningen University) for helpful discussions and critical reading of the manuscript. We also thank our colleagues of the RIKILT Food Bioactives group and NIZO Health & Safety for technical support and fruitful discussions.
Article published online before print. See web site for date of publication (http://physiolgenomics.physiology.org).
Address for reprint requests and other correspondence: J. Keijer, RIKILT Institute of Food Safety, PO Box 230, 6700 AE, Wageningen, The Netherlands (e-mail:).
- Copyright © 2007 the American Physiological Society