Salmonella enterica Serovar Typhimurium Temporally Modulates the Enteric Microbiota and Host Responses To Overcome Colonization Resistance in Swine

Limited information is available on host and enteric microbiota responses incited by Salmonella enterica serovar Typhimurium in swine and on possible mechanisms by which the bacterium overcomes colonization resistance to incite salmonellosis. Temporal characterization of a variety of host metrics in piglets (e.g., physiological, histopathological, and immunological) showed the importance of studying the progression of salmonellosis. A number of host responses integrally associated with disease development were identified. Utilization of next-generation sequence analysis to characterize the enteric microbiota was found to lack sufficient resolution; however, culture-dependent and -independent methods in combination identified taxon- and location-specific changes to bacterial communities in infected piglets. The study identified bacterial and host responses associated with salmonellosis, which will be beneficial in understanding colonization resistance and in the development of effective alternatives to antibiotics to mitigate salmonellosis.

ABSTRACT

Salmonella enterica serovar Typhimurium is a prevalent incitant of enteritis in human beings and nonhuman animals. It has been proposed that host defense responses incited by Salmonella allow the bacterium to overcome colonization resistance. Piglets (n = 24) were orally inoculated with S. enterica serovar Typhimurium DT104 or buffer alone, and the host and microbial responses were temporally examined at the acute (2 days postinoculation [dpi]), subacute (6 dpi), and recovery (10 dpi) stages of salmonellosis. At the acute stage of disease, body temperatures were elevated, and feed consumption and weight gain were reduced. The densities of Salmonella associated with the gut mucosa decreased over time, with higher densities of the bacterium in the ileum and the large intestine. Moreover, substantive histopathological changes were observed as a function of time, with prominent epithelial injury and neutrophil infiltration observed at 2 dpi. Correspondingly, a variety of host metrics were temporally affected in piglets with salmonellosis (e.g., TNFα, IFNγ, PR39, βD2, iNOS, IL8, REGIIIγ). The enteric microbiota was characterized using culture-independent and -dependent methods in concert, and taxon- and location-specific changes to the microbiota were observed in infected piglets. Bacteroides spp. (e.g., Bacteroides uniformis, Bacteroides fragilis), Streptococcus spp. (e.g., Streptococcus gallolyticus), and various Gammaproteobacteria were highly associated with inflamed tissues, while bacteria within the Ruminococcaceae and Veillonellaceae families were mainly associated with healthy mucosae. In conclusion, the study findings showed that S. Typhimurium incited temporal and spatial modifications to the swine autochthonous microbiota, and to host defense responses, that were consistent with overcoming colonization resistance to incite salmonellosis in swine.

IMPORTANCE Limited information is available on host and enteric microbiota responses incited by Salmonella enterica serovar Typhimurium in swine and on possible mechanisms by which the bacterium overcomes colonization resistance to incite salmonellosis. Temporal characterization of a variety of host metrics in piglets (e.g., physiological, histopathological, and immunological) showed the importance of studying the progression of salmonellosis. A number of host responses integrally associated with disease development were identified. Utilization of next-generation sequence analysis to characterize the enteric microbiota was found to lack sufficient resolution; however, culture-dependent and -independent methods in combination identified taxon- and location-specific changes to bacterial communities in infected piglets. The study identified bacterial and host responses associated with salmonellosis, which will be beneficial in understanding colonization resistance and in the development of effective alternatives to antibiotics to mitigate salmonellosis.

INTRODUCTION

The gastrointestinal tract (GIT) of mammals is inhabited by commensal and mutualistic microorganisms. Bacterial densities in the GIT system differ within sites, ranging from 10³ cells per ml in the stomach to 10¹¹ cells per g in the colon (1). These bacterial communities have been shown to play an essential role in nutrient acquisition, the development of gut-associated lymphoid tissue, defense against pathogens, and maturation of the intestine (2, 3). Changes in the numbers and diversity of bacteria that comprise the normal microbiota can be triggered by diverse factors potentially leading to health consequences for the host. Dramatic changes in the intestinal microbiota of swine have been observed after enteropathogenic infections (4), as a result of nutritional and dietary additives (5), and due to the administration of antibiotics (6). Furthermore, the presence of an inflammatory response triggered by infections with enteric pathogens such as Salmonella enterica and Campylobacter jejuni have been associated with alterations to the composition of the intestinal microbiota (4, 7).

Salmonella enterica serovar Typhimurium is an important zoonotic pathogen that successfully infects the intestinal tracts of swine. A number of studies to elucidate mechanisms of pathogenesis (8, 9), host immune responses (10, 11), and the epidemiology and control of salmonellosis have been conducted in pigs (8). High-throughput sequencing methods have been used to profile the enteric microbiota of swine under the influence of different feed efficiencies (12) and different levels of fatness (13). A similar culture-independent approach was employed to investigate changes in the ileal microbiota of swine subjected to an early S. Typhimurium infection (14). However, the majority of studies conducted to date have focused on the evaluation of differences in the fecal microbiota of pigs, e.g., evaluation of the fecal microbiota during weaning transition (15) or after antibiotic administration (16), and comparison between high and low Salmonella shedders (17). Importantly, the evidence indicates that the fecal microbiota is not representative of the intestinal microbiota (e.g., mucosa-associated communities) (18). Differences in protocols among laboratories, specifically in DNA extraction, variation of sequencing coverage to detect minority populations (19), differences within methods for bioinformatics analyses, the incapacity to differentiate live from dead bacteria, and relatively poor taxonomic resolution are some of the salient limitations of the culture-independent strategies. Moreover, bacteria are not recovered for additional analysis (e.g., for empirical function determinations). Thus, comprehensive analyses that include culturomics together with sequence-based microbiome analysis approaches are necessary to properly characterize the intestinal microbiota.

Arguello et al. (14) and Bearson et al. (17) examined changes in the composition of the enteric microbiota of pigs infected by S. Typhimurium. However, they did not characterize the composition and structure of the microbiota as a function of inflammation intensity; in pigs, salmonellosis follows a temporal progression developing from acute infection, characterized by severe diarrhea, vomiting, and fever, to subclinical infection, represented by intermittent shedding of the pathogen (20). Additionally, those investigators used high-throughput sequencing techniques exclusively to characterize the microbiota, and they did not characterize communities associated with the mucosa. It has been observed that the composition of bacterial communities in association with the mucosa can differ from those in digesta within the intestinal lumen (21). Thus, a primary objective of the current study was to comprehensively characterize the microbiota in the ilea, ceca, and colons of pigs at different stages of salmonellosis experimentally incited by S. Typhimurium DT104 (i.e., the acute, subacute, and recovery phases) by using both culture-based and culture-independent methods. Moreover, we collated temporal changes to the microbiota with a variety of host metrics to glean information on how the pathogen overcomes colonization resistance. Our study provides a comprehensive and extensive understanding of temporal interactions between the host and the intestinal microbiota and of the impact that enteric inflammation incited by S. Typhimurium has on this interplay.

RESULTS

Infection by Salmonella serovar Typhimurium induced temporal changes in health status.

All piglets inoculated with S. Typhimurium (ST⁺) showed behavioral evidence of disease in a temporal manner. Inoculated piglets rapidly developed transient watery diarrhea (some with mucus discharge). This was particularly conspicuous within 2 to 3 days postinoculation (dpi). Moreover, all inoculated piglets were modestly depressed and lacked appetite. By 6 dpi, diarrhea had largely abated, and by 10 dpi, all piglets showed distinct evidence of recovery (e.g., semisolid stools) and restoration of normal food consumption. ST⁺ animals exhibited an increase (P = 0.028) in body temperature over time (P = 0.007) (Fig. 1A); the body temperatures of inoculated piglets were elevated (P ≤ 0.050) 1, 2, and 3 dpi. The consumption of feed was suppressed (P ≤ 0.001) in infected animals 2 to 5 dpi (Fig. 1B). ST⁺ piglets lost weight for 4 days and weighed less (P = 0.003) than control piglets (ST^–) at all experimental end points (Fig. 1C).

FIG 1.

Temporal changes in host parameters of piglets orally inoculated with Salmonella enterica serovar Typhimurium (ST⁺) or medium alone (ST^–). (A) Rectal temperature; (B) daily feed consumption; (C) body weight. Error bars represent standard errors of the means. Asterisks indicate differences (P < 0.050) between treatments at a particular time point.

Gross pathological changes induced by Salmonella Typhimurium.

At 2 and 6 dpi, all ST⁺ piglets showed gross evidence of enteritis in the cecum, ascending colon, and spiral colon and, to a lesser extent, in the ileum. This included evidence of hyperemia, excessive gas, and accumulation of liquefied digesta. Fibrinous mucosal necrosis with casts was frequently observed in the spiral colons and ascending colons of ST⁺ piglets. All ST⁺ piglets displayed enlarged lymph nodes, and some inoculated animals exhibited splenomegaly. In no instance were conspicuous gross pathological changes observed in the duodenum or jejunum.

Salmonella Typhimurium induced temporal histopathological alterations in the distal small intestine and large intestine.

Substantive histopathological changes to the intestine were observed in ST⁺ relative to ST^– piglets. No differences were observed in histological changes within infected animals between the two cecal locations (P = 0.408) or the two spiral colon locations (P = 1.000) examined; therefore, information from the two locations within the cecum or spiral colon was grouped. Higher (P ≤ 0.041) total histopathological scores (i.e., epithelial injury, neutrophil infiltration, fibrosis, villus fusion) were observed in the distal small intestine (distal jejunum and ileum) and throughout the large intestine in ST⁺ piglets at 2, 6, and 10 dpi (Fig. 2). The degree of neutrophil infiltration (mucosa to submucosa) in ST⁺ piglets was higher than that in ST⁻ piglets (P ≤ 0.001) in the distal small intestine and large intestine at 2 dpi and progressively subsided by 10 dpi (see Fig. S1 in the supplemental material). Changes were particularly evident in these intestinal sites; there was marked transmural neutrophilic inflammation with multifocal to coalescing areas of epithelial erosion (Fig. S2). In contrast, fibrosis in the large intestines of ST⁺ pigs was more prominent (P ≤ 0.001) at 10 dpi (Fig. S3).

FIG 2.

Total histopathological scores in piglets orally inoculated with Salmonella enterica serovar Typhimurium (ST⁺) or medium alone (ST^–). Scores at 2 dpi (A), 6 dpi (B), and 10 dpi (C) are shown. Intestinal locations are as follows: 1, duodenum; 2, proximal jejunum; 3, midjejunum; 4, distal jejunum; 5, ileum; 6, cecum; 7, ascending colon; 8, spiral colon; 9, descending colon. Error bars represent standard errors of the means. Asterisks indicate differences (*, P < 0.050; **, P < 0.010; ***, P < 0.001) between treatments at a particular intestinal location.

Infection by Salmonella Typhimurium affected total white cell densities in blood.

There was no difference (P = 0.627) in the numbers or types of immune cells collected from the portal vein or systemic venous blood. The densities of total white blood cells and the percentage of granulocytes in the blood of ST⁺ piglets were elevated (P ≤ 0.001) only at 10 dpi (Fig. S4A and D). Moreover, the percentage of lymphocytes circulating in the blood of ST⁺ piglets was lower (P ≤ 0.005) at 10 dpi (Fig. S4B). The percentage of monocytes in blood was lower (P ≤ 0.020) at 10 dpi (Fig. S4C). Inoculation of pigs with S. Typhimurium had no effect (P > 0.500) on blood chemistry, including alanine aminotransferase, alkaline phosphatase, creatinine, glucose, total protein, and blood urea nitrogen levels (data not presented).

Higher densities of Salmonella Typhimurium were observed in infected piglets at 2 days postinoculation.

Salmonella Typhimurium was not isolated from the feces of any of the piglets upon arrival at the Lethbridge Research and Development Centre (LeRDC), nor was the bacterium isolated from ST^– animals during the experimental period. In contrast, S. Typhimurium was isolated from the feces of ST⁺ animals at high densities throughout the study period; the densities of the bacterium shed in feces peaked at 2 and 4 dpi (Fig. S5). High densities of S. Typhimurium DNA were also observed in digesta and associated with mucosae throughout the intestinal tract (Fig. 3). The densities of the pathogen associated with the mucosa and, to a lesser extent, in digesta were highest (P ≤ 0.039) in the distal small intestine and large intestine and decreased (P ≤ 0.017) over time. Salmonella Typhimurium was frequently isolated from the jejunal and ileal-cecal lymph nodes and from the liver and spleen. The bacterium was isolated from the brain (frontal cortex and midbrain) of one piglet 2 dpi and from systemic blood of an additional animal 10 dpi. Genotyping of arbitrarily selected strains of S. Typhimurium isolated from piglets showed that recovered isolates possessed the same pulsed-field gel electrophoresis (PFGE) fingerprint as S. Typhimurium SA970934 (data not presented).

FIG 3.

Salmonella densities from piglets at 2, 6, and 10 days postinoculation with Salmonella enterica serovar Typhimurium. Locations are as follows: 1, duodenum; 2, proximal jejunum; 3, midjejunum; 4, distal jejunum; 5, ileum; 6, cecum; 7, ascending colon; 8, spiral colon; 9, descending colon. Error bars represent standard errors of the means. Different letters above bars indicate significant differences (P < 0.050). (A) Digesta at 2 dpi; (B) mucosa-associated bacteria at 2 dpi; (C) digesta at 6 dpi; (D) mucosa-associated bacteria at 6 dpi; (E) digesta at 10 dpi; (F) mucosa-associated bacteria at 10 dpi. No Salmonella was detected in digesta or associated with mucosae from piglets orally administered medium alone.

The pH of meat was affected in piglets infected with Salmonella Typhimurium.

The pH of the longissimus dorsi muscle from ST⁺ piglets was lower (P < 0.001) than that for ST^– piglets (Fig. S6). Moreover, the pH of the longissimus dorsi muscle was lower (P < 0.001) in pigs at 6 and 10 dpi regardless of the Salmonella treatment. S. Typhimurium infection had no effect (P = 0.824) on drip moisture loss from the longissimus dorsi muscle harvested from piglets (data not presented).

The immune response was temporally modulated in infected piglets.

Differential expression of mRNA for immune genes was not observed in the duodenum or jejunum (data not presented). In contrast, the differential regulation of a variety of immune genes was observed in the ileum, cecum, and spiral colon (Fig. S7). In the ileum, the genes encoding beta-defensin 2 (βD2) (P = 0.011), mucin 4 (MUC4) (P = 0.038), and regenerating islet-derived protein 3γ (REGIIIγ) (P = 0.011) were upregulated at 6 dpi and/or at 2 dpi. In the spiral colon, βD2 (P = 0.029) and MUC1 (P = 0.024) were downregulated in ST⁺ piglets at 2 dpi and/or 6 dpi. At 2 dpi, the genes encoding tumor necrosis factor alpha (TNFα) (P < 0.001), interferon gamma (IFNγ) (P = 0.006), interleukin 17 (IL17) (P = 0.014), IL10 (P = 0.006), proline-rich antimicrobial peptide 39 (PR39) (P = 0.002), IL1β (P < 0.001), IL8 (P = 0.016), inducible nitric oxide synthase (iNOS) (P = 0.006), toll-like receptor 4 (TLR4) (P = 0.033), and transforming growth factor beta (TGFβ) (P = 0.030) were upregulated in the ceca of ST⁺ piglets relative to ST^– animals (Fig. 4). At 6 dpi, TNFα (P = 0.043), IL8 (P = 0.008), IL17 (P = 0.016), and IL1β (P = 0.045) were upregulated in the ceca of ST⁺ piglets relative to ST^– animals. Additionally, at 6 dpi, MUC1 (P = 0.028) was downregulated in the ceca of ST⁺ piglets relative to ST^– animals (Fig. S7).

FIG 4.

Relative gene expression in cecal tissue of piglets orally inoculated with Salmonella enterica serovar Typhimurium (ST⁺) or medium alone (ST^–) at 2 days postinoculation. Error bars represent standard errors of the means. Asterisks above bars indicate differences (*, P < 0.050; **, P < 0.010; ***, P < 0.001) between the two treatments.

Bacterial communities characterized by next-generation sequence analysis in the mucosa of the spiral colon did not differ in infected piglets.

To determine if bacterial communities observed in digesta corresponded with those associated with the mucosa, we examined the composition of the microbiota associated with the mucosa of the spiral colon. Bacterial communities associated with the mucosa were similar in composition to those within digesta, and no significant differences were observed at the phylum, family, or genus level (Fig. S8). Additionally, no significant differences in microbiota composition were observed between infected and control animals (Fig. S8).

Next-generation sequence analysis showed taxon- and location-specific changes in bacterial communities in piglets infected with Salmonella Typhimurium.

Amplicon sequence variant (ASV) counts were lower (P = 0.008) in ST⁺ animals than in ST^– animals. At the phylum level, infected piglets showed a decrease in the relative abundance of Firmicutes in digesta of the cecum and spiral colon (25 to 14% and 29 to 12%, respectively) at 2 dpi. In contrast, infected animals showed an increase in the relative abundance of Bacteroidetes in the ileum, cecum, and spiral colon (4 to 7.5%, 72 to 79%, and 68 to 85%) at 2 dpi. Piglets infected with S. Typhimurium showed an increase in the relative abundance of Prevotellaceae in digesta within the cecum at 2 and 6 dpi (71 to 79% and 50 to 56%) (Fig. 5). A higher abundance of Enterobacteriaceae was observed in infected animals at 6 dpi and, to a lesser extent, at 2 dpi in the ileum. Although the levels of some genera, such as members of the Ruminococcaceae and Veillonellaceae, tended to decrease in the ceca and spiral colons of ST⁺ piglets, no significant differences in relative abundance were observed in ST⁺ piglets.

FIG 5.

Spatial characterization of the main taxa in digesta from the ilea, ceca, and spiral colons of piglets inoculated with Salmonella enterica serovar Typhimurium (ST⁺) or medium alone (ST^–). Samples were obtained from piglets 2, 6, and 10 days postinoculation. Relative abundances (expressed as percentages) are represented at different taxonomic levels. (A) Phyla; (B) families.

A lower (P = 0.048) alpha diversity of bacteria was observed in the ceca of ST⁺ piglets than in those of ST^– piglets at 2 dpi, but not at 6 and 10 dpi (Fig. S9). In contrast, alpha diversity was equivalent within the ilea and spiral colons of ST⁺ and ST^– piglets (Fig. S9).

The culturable bacteria differed in piglets infected with Salmonella Typhimurium.

A total of 1,526 bacterial isolates were recovered and identified; of these, 419, 331, and 484 were collected from the ileum, cecum, and spiral colon, respectively. The compositions of the culturable and culture-independent bacterial communities differed (Fig. 6). For example, no Bacteroides or Parabacteroides organisms were detected by next-generation sequence (NGS) analysis, whereas 12 species of these two genera were isolated (Bacteroides caccae, Bacteroides denticanum, Bacteroides eggerthii, Bacteroides fragilis, Bacteroides pyogenes, Bacteroides stercoris, Bacteroides uniformis, Bacteroides vulgatus, Bacteroides xylanisolvens, Bacteroides heparinolyticus, Parabacteroides distasonis, and Parabacteroides merdae) (Fig. S10). Moreover, Streptococcus species were not detected in the cecum and spiral colon by NGS analysis, whereas Streptococcus gallolyticus was commonly isolated from the ileum, cecum, and spiral colon (Fig. 7). Analysis of taxa revealed that the abundances of the taxa isolated differed between infected and control pigs (Fig. 8). For example, S. gallolyticus was isolated only from the intestines of ST⁺ piglets. Additionally, higher abundances of Gammaproteobacteria (Escherichia/Shigella, Proteus, and Salmonella) were observed in ST⁺ piglets at 6 dpi. Bacteria that were more common in ST^– animals included unclassified members of the Ruminococcaceae family. These bacteria included putative members of the Intestinimonas genus, as well as some bacteria that are most closely related to “Candidatus Soleaferrea massiliensis,” among other previously undescribed bacteria. At 10 dpi, the composition of the culturable microbiota from ST⁺ animals did not differ from that for control piglets with the exception of Gammaproteobacteria, whose abundance remained higher in ST⁺ animals. In the ileum, members of the Actinobacteria order (e.g., Bifidobacterium pseudolongum) were more commonly isolated from ST^– piglets (Fig. 7). In contrast, Gammaproteobacteria (Escherichia/Shigella and S. enterica) and Fusobacterium species (Fusobacterium varium and Fusobacterium gastrosuis) were commonly isolated from the ilea of ST⁺ animals. In the ceca and spiral colons of ST⁺ piglets, members of the class Bacteroidia (B. uniformis, B. fragilis, and B. heparinolyticus) were more abundant. Bacteria within the Firmicutes phylum (i.e., Bacilli, Clostridia, Erysipelotrichia, and Negativicutes) were recovered from all locations and from both ST⁺ and ST^– piglets. However, members of the Bacilli (i.e., Lactobacillales and Bacillales) were more prevalent in ST^– piglets. Although Clostridia bacteria were more commonly associated with samples collected from the spiral colons of ST⁺ piglets, a higher abundance of the Clostridiaceae family was observed in the ceca of ST^– piglets.

FIG 6.

Spatial characterization of the main families of bacteria in digesta from the ilea, ceca, and spiral colons of piglets inoculated with Salmonella enterica serovar Typhimurium (ST⁺) or medium alone (ST^–). (A) Next-generation sequence analysis; (B) culturomics. Data were combined across sample times (i.e., 2, 6, and 10 days postinoculation).

FIG 7.

Cladogram illustrating the abundances of bacterial species isolated from the ilea, ceca, and spiral colons of piglets inoculated with Salmonella enterica serovar Typhimurium (ST⁺) or medium alone (ST^–). The cladogram background is color coded to illustrate relative changes in the abundances of isolated bacteria between treatments and intestinal locations. Moreover, 21 bacterial species that were differentially abundant in ST⁺ and ST^– piglets are indicated by colored circles and are labeled “a” through “u.” Gold circles represent bacteria in which no difference in abundance due to infection by S. Typhimurium was observed.

FIG 8.

Abundances of bacteria isolated from the ilea, ceca, ascending colons, and spiral colons of piglets infected with Salmonella Typhimurium or inoculated with buffer alone (control) at 2, 6, and 10 days postinoculation. Cells are colored according to the number of bacteria recovered, and the distribution and color scheme of these counts are indicated in the “normalized counts” plot above the heat map. The heat map was generated using the heatmap.2 function contained in the gplots package of R (93).

qPCR confirmed that the densities of bacterial genera differed by intestinal location and in piglets infected by Salmonella Typhimurium.

Lower quantities of DNA for Intestinimonas spp., Prevotella spp., and Ruminococcus spp. were observed in the ileum than in the cecum (P < 0.001) and spiral colon (P < 0.001) (Fig. S11). Although the concentration of Clostridium cluster I sp. DNA in the ileum was higher than those of the other genera examined, concentrations were still lower (P < 0.007) in the ileum than in the cecum (Fig. S11C). Evaluation of the specific species B. uniformis and S. gallolyticus showed lower quantities of DNA in the ileum than in the cecum (P < 0.001) and spiral colon (P < 0.001) by quantitative PCR (qPCR). DNA of these species was detected in the digesta and in association with mucosae (data not presented) of the ileum, cecum, and spiral colon, albeit at low densities. In agreement with the frequency of isolation, the densities of B. uniformis were higher (P ≤ 0.058) in the digesta of infected animals in the cecum at 10 dpi and in the spiral colon at 6 dpi (Fig. 9). Similarly, a trend for higher (P ≤ 0.124) densities of S. gallolyticus in the digesta of infected animals was observed in the cecum and spiral colon at both 6 and 10 dpi.

FIG 9.

Densities of Bacteroides uniformis and Streptococcus gallolyticus in digesta of piglets inoculated with Salmonella Typhimurium (ST⁺) or medium alone (ST^–) as determined by quantitative PCR. (A) B. uniformis in the cecum; (B) S. gallolyticus in the cecum; (C) B. uniformis in the spiral colon; (D) S. gallolyticus in the spiral colon. Error bars represent standard errors of the means. Asterisks above bars indicate differences (*, P ≤ 0.124) between the ST⁺ and ST^– treatments at 2, 6, and 10 days postinoculation.

Biochemical pathways, KEGG numbers, and EC numbers did not differ between Salmonella Typhimurium-infected and control animals.

Biochemical pathway inference was carried out with 16S rRNA phylogenies obtained from mucosa samples of the spiral colon. No differences in biochemical pathways, Kyoto Encyclopedia of Genes and Genomes (KEGG) numbers, or Enzyme Commission (EC) numbers were detected (P ≥ 0.873) between ST^– and ST⁺ animals at 2, 6, and 10 dpi.

Interaction networks showed lower ASV associations in Salmonella Typhimurium-infected animals.

Interaction networks were generated based on Bray-Curtis distances of the taxonomically classified ASVs. Modestly lower ASV interactions were observed in ST⁺ than in ST^– animals (Fig. S12). Network associations tended to decrease between the Campylobacter and Prevotella genera in ST⁺ animals.

DISCUSSION

The composition of the intestinal microbiota of piglets has been studied by a number of research groups (5, 15, 16, 22, 23). However, the impact of infection and ensuing inflammation on the enteric microbiota is poorly understood at present. In our study, we temporally (2, 6, and 10 dpi) and spatially (small to large intestine) characterized host responses to S. Typhimurium infection in piglets. In this regard, we concomitantly measured immune responses, histopathological alterations, and changes to the structure of the enteric microbiota at the acute, subacute, and chronic stages of salmonellosis. The current study comprehensively characterized the microbiota of piglets infected with S. Typhimurium by using both culture-dependent and culture-independent methods. The highest degree of inflammation was observed in the cecum. NGS analysis showed that the primary changes to the microbiota happened in the cecum at the earliest stages of disease (i.e., 2 dpi), which correlated with the highest level of inflammation. Substantially lower Shannon diversity, an increase in the relative abundance of Proteobacteria, and a decrease in the relative abundance of Clostridiaceae and Lachnospiraceae were observed in the ceca of infected animals. Additionally, culture-based methods allowed us to examine bacteria at higher resolution than NGS analysis and demonstrated higher association of some species with inflamed tissues (e.g., S. gallolyticus and B. uniformis). Our study provides novel information on the coordinated influence of S. Typhimurium on the enteric microbiota and host responses.

Salmonellosis is a concern to the swine sector for two primary reasons. First, the disease causes significant economic losses, and second, S. Typhimurium is an important foodborne pathogen responsible for significant morbidity of humans (20), resulting in additional losses to the swine sector (e.g., meat recalls). Swine infected with S. Typhimurium exhibit a disease progression, but the temporal characteristics of acute salmonellosis in pigs are not well defined. Similarly to Scherer et al. (20), we observed evidence of enteritis (24, 25) in piglets 1 to 2 dpi, including elevated body temperature, inappetence, body weight loss, and diarrhea. To temporally characterize the impacts of salmonellosis on the host, samples were recovered and analyzed at 2, 6, and 10 dpi. Substantive histopathological changes were observed throughout the intestines of infected piglets at all three time points. In this regard, pathogen densities in digesta and mucosa samples, and total histopathological scores, including epithelial injury and neutrophil infiltration, were higher at 2 dpi in piglets infected with S. Typhimurium. We observed that the densities of S. Typhimurium within digesta and associated with the mucosa decreased over time. There was also an attenuation of intestinal injury that returned to near-normal levels by 10 dpi. However, fibrosis is considered to be a sign of tissue repair, and higher levels of fibrosis were observed in the ceca, ascending colons, spiral colons, and descending colons of piglets infected with S. Typhimurium at 10 dpi. The three sample times at which piglets were evaluated in the current study (i.e., 2, 6, and 10 dpi), represented the acute, subacute, and recovery stages of salmonellosis, respectively. Spatially, histopathological changes were most pronounced in the distal small intestine and within the cecum and spiral colon, findings consistent with those of previous studies (26, 27). It is suggested that the locations of tissue injury may be attributed to the lower concentration of bile salts and beta-defensins present at these sites (28).

We examined the modulation of immune markers during the progression of salmonellosis in swine. Immune responses triggered after infection with S. Typhimurium in swine have been described previously (10, 29, 30). However, the immune response within the entire GIT has not been evaluated previously. In the current study, we observed that S. Typhimurium did not affect the expression of target genes in the duodenum, jejunum, and proximal ileum. In contrast, targeted genes were upregulated in the distal ilea, ceca, and colons of S. Typhimurium-infected piglets, which corresponded to the histopathological changes observed. In these locations, differential regulation of gene expression occurred primarily at 2 dpi (acute stage), with modest changes at 6 dpi, and no difference between infected and uninfected piglets at 10 dpi (recovery). The cecum was the most affected intestinal site. In the cecum, we observed elevated expression of IL17 and IFNγ, corresponding to the findings of other studies (31, 32). Secretion of IL-17 and contact between the pathogen and enterocytes have been shown to enhance the secretion of IL-8, attracting neutrophils from the microvasculature of the intestinal mucosa (33). In the current study, greater neutrophilic infiltration was observed in the cecum at 2 dpi, corresponding with the elevated expression of IL8 observed at this time point. We observed elevated expression of the gene encoding iNOS in the cecum at 2 dpi. Since iNOS secretion has been associated with elimination of S. Typhimurium and also with epithelial injury (34), this is consistent with the mucosal damage that we observed in the cecum at 2 dpi. Induction of a proinflammatory response by pathogens is often accompanied by an anti-inflammatory response triggered by the host (35). We observed that S. Typhimurium-infected piglets, which exhibited a higher proinflammatory response in the cecum at 2 dpi, concurrently presented higher expression of IL10 and TGFβ. Both of these anti-inflammatory cytokines have been shown previously to incite a regulatory response to attenuate inflammation within the intestine (35); therefore, the elevated expression of these cytokines is likely directed to control the inflammatory response triggered by S. Typhimurium.

One of the main mechanisms of innate defense that the host utilizes to eliminate pathogens and to shape the microbiota is the secretion of host defense peptides (HDPs) (36). In the current study, we examined the expression of genes encoding beta-defensins (e.g., βD2), C-type lectins (e.g., REGIIIγ), and cathelicidins (e.g., PR39). We observed that these three types of HDPs were upregulated after Salmonella infection. The secretion of beta-defensins, which have bactericidal activity, is induced after pathogens have been detected by the host (37). In piglets, the secretion of βD2 has been shown previously to inhibit the growth of S. Typhimurium (38). Higher expression of REGIIIγ, which exerts its bactericidal action by pore-forming activity, has also been observed following infection of pigs with enterotoxigenic Escherichia coli (39). A primary group of HDPs are the cathelicidins (40). In pigs, PR39, a homologue of LL37 and mCRAMP in humans and mice, respectively, has been studied due to its broad antimicrobial spectrum, but not extensively in pigs with salmonellosis (40, 41). The release of cathelicidins by neutrophils and their expression on mucosal surfaces have been associated with local host defense. Their antimicrobial activity is not limited to their potent, rapid, and broad-spectrum bactericidal function but also involves a capacity to bind and neutralize endotoxins (42). PR39 has been observed to inhibit bacterial DNA and protein synthesis (41). Its immunomodulatory effects include leukocyte chemotaxis, modulation of cytokine production, and stimulation of wound healing (43) and angiogenesis (44). In the current study, HDP genes were the only genes upregulated not only in the ceca of S. Typhimurium-infected piglets but also in their ilea. The differential regulation of βD2 and REGIIIγ in the ileum relative to the cecum was expected, since higher expression of both of these immune markers in the ilea of infected pigs has been reported previously (38, 39). Since PR39 is mainly secreted by neutrophils, the higher expression in the cecum that we observed in the current study could likely be related to the level of neutrophil infiltration observed at this intestinal site. Although the roles of cathelicidins in modulating the enteric microbiota and salmonellosis are enigmatic at present, the changes in expression that we observed could be intimately related to their bactericidal activity directed against S. Typhimurium (45).

A primary goal of the current study was to describe how the commensal microbiota is temporally modified during the induction and progression of intestinal disease. We comparatively characterized bacterial communities in the ilea, ceca, and spiral colons of S. Typhimurium-infected and control piglets by using a combination of culture-independent and culture-dependent methods. Overall, we observed taxon-specific changes in the composition of bacterial communities within digesta between S. Typhimurium-infected and control piglets. The most pronounced changes between the two treatment groups occurred at 2 dpi, which corresponded to the highest degree of intestinal damage (i.e., acute salmonellosis). Communities in cecal digesta were more affected by inflammation than those in the spiral colon. In this regard, a decrease in alpha diversity was observed in the cecal digesta of S. Typhimurium-infected piglets at 2 dpi, corresponding to a similar observation in chickens infected with S. Typhimurium (46). Higher abundances of Proteobacteria (e.g., Enterobacteriaceae) were observed in both the ceca and spiral colons of S. Typhimurium-infected piglets, and this occurrence was most pronounced in the cecum at 2 dpi, as has been observed previously (47). An increase in bacterial proportions within this phylum is a hallmark of dysbiosis and epithelial injury (46, 47), and since the phylum Proteobacteria contains many human and porcine pathogens, including Salmonella, the increase in abundance of pathogenic organisms within inflamed tissues is expected. Correspondingly, obligate anaerobes, such as Clostridiaceae 1 and Lachnospiraceae, decreased in abundance within the ceca and spiral colons of S. Typhimurium-infected animals at 2 dpi but recovered to normal levels by 10 dpi. Changes in the relative abundances of Enterobacteriaceae, Clostridiaceae, and Lachnospiraceae are influenced by the levels of oxygen in the lumen of infected piglets (47, 48). When inflammation is triggered within the GIT, higher levels of oxygen are released into the lumen (48). The increase in luminal oxygen could be due to greater blood supply to the GIT during chronic inflammation, with extravasation of hemoglobin carrying oxygen into the lumen (48). A change in the oxygen level could also be the result of inflammation itself, as an oxidative burst triggered by neutrophils releases reactive oxygen species into the lumen, benefiting facultative anaerobes (49). Moreover, a negative correlation between Enterobacteriaceae and Prevotellaceae, Ruminococcaceae, and Lachnospiraceae has been reported previously in other hosts infected with Salmonella (46).

The composition of the microbiota varies in the cross-sectional axis of the intestine. Unique niches for enteric bacteria (23) are provided by higher concentrations of oxygen, proximity to the mucosa, and the abundance of mucin glycoproteins and other nutrient sources. We observed higher densities of S. Typhimurium in association with the mucosa than in the luminal digesta in the cecum and spiral colon. This is consistent with the lowered densities of Prevotellaceae, Lachnospiraceae, and Ruminococcaceae previously observed in association with the cecal mucosa (23). We observed that communities associated with the mucosa were not significantly different in composition from those in the digesta of the spiral colon (i.e., via Illumina sequencing of the community). Although we observed a trend for higher abundances of bacteria within the Enterobacteriaceae and Campylobacteraceae in association with the mucosa (i.e., families that contain species known to associate with the mucosa), these differences were not statistically significant. Studying mucosa-associated bacteria is challenging. For example, removal of residual digesta from mucosal surfaces without disrupting the integrity of the loosely adherent mucus is exceptionally difficult. Furthermore, the predominance of host DNA can lead to its preferential amplification; to circumvent this possibility, we used a DNA enrichment method that allowed the capture and elimination of methylated host DNA while conserving and enriching bacterial DNA. Although some studies have reported a unique bacterial community associated with the mucosa (21), conclusions regarding differential composition of mucosa-associated and luminal communities without supporting evidence (e.g., fluorescent in situ microscopy) should be interpreted carefully. Network interaction analyses carried out in the mucosa samples did not show significant effects; however, a higher number of microbial associations was observed in controls than in Salmonella-infected animals. The decreased number of network interactions observed between the Campylobacter and Prevotella genera in infected animals could be strictly related to the lower diversity observed under infection. The lower number of ASVs from the genus Prevotella associated with the mucosae of infected animals could result from the higher levels of inflammation and damage detected in the spiral colons of ST⁺ animals, which is in line with previous studies (23).

Despite the current emphasis on NGS technologies to characterize the intestinal microbiota of mammals, there are inherent limitations to sequence-based methods (e.g., poor taxonomic resolution, taxon overrepresentation, poor sensitivity, and incapacity to differentiate live from dead bacteria) (50). Thus, we also employed culture-based methods to characterize enteric communities in the intestines of S. Typhimurium-infected and control piglets. To isolate bacteria, we applied a variety of media and strict anaerobic methods (e.g., differential killing of vegetative cells, induction of endospore germination, direct plating, long-term enrichments, and an isolation chip (Ichip) method modified to isolate enteric bacteria) (51–53). We observed that no single isolation method was comprehensive and that a combination of methods was required. Using a single method, Russell (54) isolated and characterized bacteria from the intestines of healthy pigs. He recovered 46 presumptive taxa that were identified using physiological characteristics. In contrast, we recovered a minimum of 173 different bacterial species, representing seven phyla, using a combination of methods in concert. Our findings of prominent taxa broadly correspond with those of Russell (54) in that Eubacterium, Lactobacillus, and Gram-positive cocci were frequently recovered. Additionally, the targeting of bacteria associated with inflamed tissues in piglets, coupled with the use of a combination of culturomic methods, likely contributed to recovering a higher diversity of bacteria.

Both our culturomic and our NGS analyses of enteric bacteria in piglets with salmonellosis showed an elevated abundance of Gammaproteobacteria (Escherichia/Shigella, Proteus, and Salmonella), which has been reported previously (55). However, in other instances, the compositions of the bacterial communities as determined by culturomics and NGS analysis differed conspicuously. In this regard, analysis of culturomics data showed that a number of taxa were differentially associated with inflamed tissues. For instance, S. gallolyticus was isolated solely from S. Typhimurium-infected piglets, and B. fragilis, B. heparinolyticus, B. uniformis, and Acidaminococcus fermentans were more commonly isolated from pigs with salmonellosis; these three genera were not detected using NGS analysis. The fact that we isolated S. gallolyticus at a conspicuously higher frequency from inflamed tissues of piglets infected with S. Typhimurium suggests that S. gallolyticus is favored by this condition/habitat. An association between S. gallolyticus and colon cancer tissues in human beings has been reported recently (56). Therefore, the presence of this bacterium in S. Typhimurium-infected piglets may result from a propensity of this bacterium to colonize inflamed tissue sites (57, 58). A characteristic of S. gallolyticus that may favor colonization in animals with inflammation is its ability to form biofilms on exposed collagen in inflamed tissues (56). Another possibility is that S. gallolyticus is able to evade the immune response mounted by the host (e.g., HDPs), as has been observed for other taxa effective at overcoming colonization resistance in the intestine. Of the 10 species of Bacteroides that we isolated, only B. fragilis, B. uniformis, and B. heparinolyticus were more commonly isolated from piglets with salmonellosis. An association of enterotoxigenic B. fragilis with inflamed tissues in piglets and human beings has been reported previously (59, 60). This bacterium is known to colonize intestinal crypts (61). Moreover, the formation of capsules provides resistance to the complement system, as well as to phagocytic uptake and killing (62). The utilization of a comprehensive isolation strategy revealed a number of bacteria that are differentially associated with infected piglets and demonstrated the value of using culture-dependent methods in concert with NGS analysis of communities to characterize bacteria potentially associated with salmonellosis. Ancillary examination of cell densities by taxon-specific quantitative PCR (qPCR) showed a tendency of S. gallolyticus and B. uniformis to increase in the ceca and spiral colons of S. Typhimurium-infected piglets. However, the low level of detection of these taxa associated with the mucosa and the limitation of real-time qPCR in differentiating between nucleic acids from live and dead cells emphasize the requirement of conducting in vivo analyses to confirm the propensity of these taxa to colonize inflamed tissue.

The “commensal” microbiota participates in host nutrition, the development of the gastrointestinal immune system, maturation of the intestine, and defense of the GIT (2, 3). Crucially, commensal microorganisms impede pathogens from effectively colonizing the GIT via a variety of mechanisms, thereby attenuating or preventing disease (63). The production of short-chain fatty acids (SCFAs) by the large intestinal microbiota can influence colonization resistance, providing beneficial impacts for intestinal health (64). Therefore, SCFAs can enhance intestinal health by reduction of the pH, stimulation of mucin secretion, improvement of tight-junction integrity, and induction of T-regulatory cell differentiation to impair pathogen colonization (64). Although we did not measure SCFA production, we observed a negative correlation between SCFA-producing taxa (e.g., Ruminococcaceae, Prevotellaceae, and Lachnospiraceae) and bacteria within the Enterobacteriaceae, which could be intimately related to their excretion of SCFAs and thus their mechanisms of defense (65). Additionally, the upregulation of MUC4 that we observed in ST⁺ animals at 2 dpi could be associated with microbially stimulated enhancement of barrier function, which has been reported after secretion of butyric acid (66). Stimulation of host immune defenses by the commensal flora, mainly the release of HDPs, has been described as an indirect mechanism of colonization resistance (63). In the current study, the elevated expression of HDPs (e.g., PR39, REGIIIγ, and βD2) that was observed in ST⁺ piglets at 2 dpi may be a key step in restoring homeostasis. However, pathogens have developed various strategies to compete with the microbiota for nutrients and binding sites in order to enhance their ability to colonize the GIT. In this regard, S. Typhimurium has been shown to induce inflammation in an attempt to outcompete the microbiota (26, 67, 68). Therefore, the higher degree of inflammation, the higher abundance of S. Typhimurium, and the lower diversity of commensals observed in the ceca of ST⁺ animals at 2 dpi could also be an indication of the importance of the normal microbiota in maintaining equilibrium in the GIT.

In conclusion, we evaluated the progression of salmonellosis in pigs by characterizing histopathological changes in the GIT, host immune responses, and alterations to the enteric microbiota. To characterize the microbiota, we applied a combination of culture-dependent and culture-independent methods, and we found that reliance on NGS analysis was insufficient to detect taxon-specific changes associated with inflammation. Furthermore, we recovered a number of taxa that were differentially abundant in pigs with salmonellosis (e.g., S. gallolyticus), and the acquisition of these bacteria will facilitate functional assessments. A detailed evaluation of the host immune responses pointed to upregulation of HDPs as an important mechanism modulating disease. It is currently believed that these peptides execute their bactericidal function in a prescribed manner; however, the mechanisms by which HDPs trigger an immune response and also modify the microbiota are not well understood. Knockout mice lacking HDP genes could be used to elucidate the role that these peptides play in salmonellosis. Moreover, delivery of HDPs to sites of enteric inflammation may represent a novel and nonantibiotic strategy to mitigate enteritis. These are areas of current research by our team.

MATERIALS AND METHODS

Ethics.

The project was approved by the LeRDC Animal Care Committee (Animal Use Protocol Review 1512) and the LeRDC Biosafety and Biosecurity Committee before commencement. Since infection by S. Typhimurium in pigs is reportable in Alberta, approval to conduct Salmonella inoculations in piglets was also obtained from the Head Provincial Veterinarian, Gerald Hauer, Alberta Agriculture and Forestry, Edmonton, Alberta, Canada.

Experimental design.

The experiment was arranged as a completely randomized design with three levels of sample time (2, 6, and 10 dpi) and two levels of immunological stress (with or without S. Typhimurium) (Fig. S13). The goal was to obtain samples that corresponded temporally to the acute, subacute, and recovery phases of salmonellosis in swine. To ensure humane animal care, piglets were housed in pairs within individual animal rooms. The limited number of animal rooms available in the Livestock Containment Unit (LCU) at the LeRDC necessitated conducting the experiment on separate occasions (i.e., two replicates at time 1, and two replicates at time 2); the separate times were treated as a random effect in the statistical model.

Animal maintenance.

Castrated male piglets at 6 weeks of age were used in the experiment. Piglets were vaccinated for circovirus, ileitis, and erysipelothrix. Neither the sow (during pregnancy or postpartum) nor piglets were administered antibiotics. Piglets were group housed for 3 days in the LCU under a 14-h-dark, 10-h-light cycle. After the adaptation period, arbitrarily selected animals were assigned to individual pens, with two animals per room. Piglets were provided a minipellet ration diet that was free of antibiotics (Proform Pig Starter 2; Hi-Pro Feeds, Okotoks, Alberta, Canada). Feed was provided daily, and piglets were permitted to eat and drink ad libitum. Straw was used for bedding, and toys were provided for environmental enrichment. Animals were monitored daily for activity level, and behavioral signs of pain and stress (i.e., diarrhea) were recorded. Bedding and residual food and water were replaced daily. Initial body weights were recorded upon the assignment of animals to individual cages, and every other day thereafter. Food consumption was determined daily.

Inoculation.

Piglets were orally inoculated with S. enterica serovar Typhimurium DT104 (strain SA970934) (69). The bacterium was grown aerobically on MacConkey’s agar (MA) (Difco BD, Mississauga, Ontario, Canada) at 37°C for 16 to 24 h. Biomass was removed from the surface of the agar and transferred to Columbia broth (CB) (Difco BD, Mississauga, Ontario, Canada). Cultures were maintained for 180 to 210 min at 37°C, with shaking at 150 rpm, until an optical density at 600 nm of >1.2 was obtained. Cultures were centrifuged at 4,000 × g for 15 min, supernatants were removed to a volume of 20 ml, and the density was adjusted to a target of 3.0 × 10⁹ cells ml⁻¹. To confirm the densities of viable cells, the inoculum was diluted in a 10-fold dilution series, 100 μl of each dilution was spread in duplicate onto MA, cultures were incubated at 37°C, and S. Typhimurium colonies were counted at the dilution yielding 30 to 300 CFU after 24 h. Individual piglets were gavaged on 2 consecutive days with S. Typhimurium cells in CB (1.0 ml) (ST⁺) or with CB alone (1.0 ml) (ST^–). Each piglet was administered the two treatments in 1-ml aliquots using sterile 3.0-ml syringes. The animals were individually restrained, the syringe was placed in the rear of the mouth, and the liquid slowly dispensed, taking care to avoid aspiration.

Body temperature, feed consumption, weight gain, and feces collection.

Rectal temperature and food consumption were measured daily. In addition, body weights were measured at 2-day intervals using a Model 75 scale (Reliable Scale Corporation, Calgary, Alberta, Canada). Fresh feces were collected from the pen floor immediately before inoculation and at 2-day intervals thereafter, including just before humane euthanization. Fecal samples were transported to the laboratory within 30 min of collection for analysis. Samples of fresh feces were processed for the presence of S. Typhimurium via dilution plating. In addition, aliquots of the feces were weighed and placed at –80°C for qPCR.

Intestinal tissue collection.

At 2, 6, and 10 dpi, randomly designated animals were anesthetized for sample collection. Piglets were sedated with a cocktail of ketamine (Ketaset; Pfizer, Kirkland, Quebec, Canada) and xylazine (Xylamax; Bimeda, Cambridge, Ontario, Canada) at doses of 22 and 2.2 mg kg of body weight⁻¹, respectively. An intramuscular injection was administered. Animals were placed in dorsal recumbency on a v-trough surgical table and were intubated, and general anesthesia was established with isoflurane (Abbott Laboratories, Chicago, IL). The abdomen was sanitized with chlorhexidine (Stanhexidine; Omega Laboratories Ltd., Montreal, Quebec, Canada), 70% ethanol, and poloxamer-iodine (Prepodyne; West Penetone Inc., Ville D’Anjou, Quebec, Canada).

With animals under general anesthesia, segments of intestine (≈10 cm long) were collected from the duodenum, proximal jejunum, midjejunum, distal jejunum, ileum, cecum (two segments located at the free end and adjacent to the ileal-cecal junction), ascending colon, spiral colon (two segments located at the central flexure and ≈100 cm distal from the flexure), and descending colon. To ensure the integrity of the intestinal segment and to minimize the release of ingesta, double ligatures were established at the two ends of the segment, and the segment was excised from the intestine by cutting between the two ligatures. Care was taken to ligate mesentery blood vessels immediately prior to intestinal segment removal and to ensure the maintenance of blood flow to adjacent intestinal tissue. All intestinal samples were processed within ca. 5 to 10 min of their removal from live animals. To prevent the introduction of air, additional ligatures were established on segments from which anaerobic bacteria were to be isolated (i.e., ileum, cecum, and spiral colon). Ligated subsegments for culturomics were removed and immediately placed in an anaerobic jar; the ambient atmosphere was removed by vacuum and replaced with nitrogen (N₂), and the segments were transported to the laboratory. With the exception of intestinal segments used to isolate anaerobic bacteria, segments were longitudinally incised in ambient atmosphere, and luminal contents (i.e., digesta) were aseptically collected and weighed for DNA extraction (e.g., for quantitation of S. Typhimurium and selected commensal bacterial taxa and for characterization of bacterial communities). Sections of the intestine were collected for RNA extraction (i.e., for gene expression), and DNA extraction (e.g., for quantitation of mucosa-associated S. Typhimurium and select commensal bacterial taxa, histology, and characterization of bacterial communities). Samples for characterization of mucosa-associated S. Typhimurium and microbial communities were gently flooded with sterile phosphate-buffered saline (PBS) (≈3 ml) to remove residual ingesta with minimal disruption to the adherent mucus. Samples for RNA extraction (three subsamples per segment) were placed in RNAprotect (Qiagen Inc., Toronto, Ontario, Canada) and maintained at −80°C until processing. For histopathological scoring, intestinal segments were placed in 10% neutral buffered formalin (Surgipath Canada, Inc., Winnipeg, Manitoba, Canada) for a minimum of 24 h.

Blood collection and animal euthanization.

Blood (≈20 ml) was collected from the portal vein (draining the intestine) and from the right atrium of the heart or vena cava (systemic blood) using an 18-gauge needle. Blood was collected in BD Vacutainer tubes with K₂EDTA (BD, Franklin Lakes, NJ) for complete blood count (CBC) analysis and detection of S. Typhimurium. Immediately after blood removal, the animal was euthanized under general anesthesia.

Accessory tissue collection.

Within 5 to 10 min of death, ileal-cecal lymph nodes, jejunal lymph nodes (one to two per animal), the liver, and the spleen were removed, and samples excised from these tissues were placed in RNAprotect (Qiagen Inc.) and maintained at –80°C until processing. In addition, samples from the above tissues, and from the frontal cortex, midbrain, and brain stem, were collected for isolation of Salmonella.

Histopathology.

Tissues for hematoxylin and eosin (H&E) staining were prepared by our standard procedure (70). Tissues from the duodenum, proximal jejunum, midjejunum, distal jejunum, ileum, cecum, spiral colon, ascending colon, and distal colon in formalin were first dehydrated using a tissue processor (Leica TP 1020; Leica Biosystems) and then embedded in paraffin (Fisherfinest Histoplast PE; Thermo Fisher Scientific, Edmonton, Alberta, Canada) using a Shandon Histocentre 3 system (Thermo Fisher Scientific). Five-micrometer sections were transferred to positively charged slides (Fisherbrand Superfrost Plus Gold; Thermo Fisher Scientific) and allowed to dry prior to being deparaffinized with xylene. Slides were then rehydrated in ethanol and stained with H&E using a standard protocol. Histopathological changes were assessed by a veterinary pathologist (R.R.E.U.) who was blinded to the treatment protocol. The scoring system used was a modification of previous scoring protocols developed by Boyer et al. (2015) (71), and Garner et al. (2009) (72). The tissues were scored for villus blunting, villus fusion, lymphoid depletion, neutrophil infiltration, epithelial injury, and fibrosis. Villus blunting (i.e., crypt-to-villus ratio) scores were as follows: 0, normal (no changes in villus height); 1, <25% reduction in villus height; 2, 26 to 50% reduction in villus height; 3, 51 to 75% reduction in villus height; 4, complete villus loss. Villus fusion scores: 0, normal (with no increases in the presence of villus fusion); 1, mild (with small increases in the numbers of fused villi); 2, moderate (with prominent increases in the number of fused villi); and 3, severe (with substantive increases in the numbers of fused villi). Lymphoid depletion scores: 0, none; 1, mild (with a small reduction in lymphoid cells); 2, moderate (with a prominent reduction in lymphoid cells); and 3, severe (with a marked reduction in lymphoid cells). Neutrophil infiltration was examined in both the lamina propria (designated D1) and the lamina muscularis and serosa (designated D3). Neutrophil infiltration scores were as follows: 0, none; 1, rarely observed in tissue; 2, a few scattered neutrophils within the tissue; 3, many foci with collections of a few neutrophils; 4, large numbers of neutrophils present within the tissue. Epithelial injury scores: 0, none; 1, rare (shedding of <10 surface epithelial cells); 2, mild (focal epithelial erosions; shedding of 11 to 50 surface epithelial cells); 3, moderate (multifocal surface epithelium erosions; shedding of <50 surface epithelial cells); 4, severe (multifocal to coalescing areas of surface epithelium erosions; shedding of >50 surface epithelial cells). Fibrosis (desmoplasia) scores: 0, normal; 1, rare (with small foci of collections of reactive fibroblast or small areas with increased amounts of mature collagen); 2, small focal to multifocal areas of collections of reactive fibroblast or small focal to multifocal areas with increased amounts of mature collagen; 3, large areas of collections of reactive fibroblast or large areas with increased amounts of mature collagen. The total histopathological score was the sum of all individual tissue measurements (maximum score, 21).

Blood analysis.

Complete blood counts were performed on a HemaTrue blood analyzer (Heska, Des Moines, IA) within 45 min of collection. Blood chemistry was analyzed on a VetTest blood analyzer (Idexx Laboratories, Westbrook, ME) using a preanesthetic blood panel. The preanesthetic blood panel included tests for alanine aminotransferase, alkaline phosphatase, creatinine, glucose, total protein, and blood urea nitrogen.

Isolation and genotyping of Salmonella.

Salmonella was isolated and identified from the feces, systemic blood, ileal-cecal lymph nodes, jejunal lymph nodes, liver, frontal cortex, midbrain, and brain stem as described previously (73). Briefly, for feces, ≈1 g of fecal matter was weighed, diluted 1:10 in buffered peptone water (Oxoid Inc., Nepean, Ontario, Canada), and suspended in the liquid by vortexing. For blood, 1 ml of systemic blood was added to 9 ml of buffered peptone water and vortexed. For tissues, any bacteria on the outside of the samples were killed by immersion in boiling water for 10 s; this method does not kill Salmonella internalized in the tissues (74). As with feces, ≈1 g of tissue was weighed and diluted 1:10 in buffered peptone water. Samples were homogenized using a Tissue-Tearor (model 985370; Biospec Products Inc., Bartlesville, OK). All suspensions were incubated at 37°C for 16 to 24 h, and 50 μl of the suspension was transferred to 5 ml of Rappaport-Vassiliadis enrichment broth (Oxoid Inc.) and incubated at 42°C for 16 to 24 h. A 10-μl inoculation loop was used to transfer a subsample of the Rappaport-Vassiliadis enrichment broth to both Brilliant Green agar (BGA; BD Difco, Mississauga, Ontario, Canada) and modified lysine iron agar (MLIA; Oxoid Inc.). The BGA and MLIA cultures were incubated for 48 h at 37°C to allow time for H₂S production to manifest. Representative red colonies on BGA were transferred to triple sugar iron agar (TSIA; BD Difco) slants. Two representative black colonies from MLIA were transferred to TSIA slants. The TSIA slant cultures were incubated at 37°C for 16 to 24 h. Representative colonies from Salmonella-positive TSIA slants were transferred to MacConkey agar (BD Difco) and incubated at 37°C for 16 to 24 h. Colorless colonies on MacConkey agar were considered Salmonella positive and were stored in 30% glycerol in broth at −80°C.

To identify presumptive Salmonella-positive colonies, genomic DNA was extracted using an automated DNA extraction robot (model Autogen 740; Autogen, Inc., Holliston, MA) according to the manufacturer’s recommendations. Taxon-specific PCR was conducted using primers F-(Sal) and R-(Sal), which target the invA gene of S. enterica (75). The reactions were run on an Eppendorf Mastercycler Pro S thermocycler (VWR International, Mississauga, Ontario, Canada) at 95°C for 5 min, followed by 35 cycles of 94°C for 1 min, 64°C for 1 min, and 72°C for 1 min, and 1 cycle at 72°C for 10 min. DNA obtained from S. Typhimurium SA970934 was used as a positive control, and Optima water (Fisher Scientific, Edmonton, Alberta, Canada) was used as a template negative control. To genotype Salmonella isolates, the PFGE protocol used by PulseNet (PNL05) was applied to representative Salmonella isolates recovered from piglets relative to S. Typhimurium SA970934.

Meat characteristics.

Exsanguinated animals were hung in a 4°C cooler, and the pH of the right longissimus dorsi muscle (≈2 to 3 cm posterior to the last rib) was determined at 45 min and 24 h postmortem using a portable meat pH meter (HI99163; Hanna Instruments, Laval, Quebec, Canada). At 24 h postmortem, the right and left longissimus dorsi muscles were removed, and a moisture drip loss test was conducted. Briefly, the longissimus dorsi muscle from each side was treated as an observation (n = 2). The individual muscles were trimmed to equal dimensions, cut in two, weighed, placed in a porous nylon bag, suspended within a liquid impervious bag for 48 h at 4°C, and reweighed.

RNA extraction.

To quantify the mRNAs of targets of interest, RNA was extracted from samples (≈0.5 by 0.5 cm) from the intestines (duodenum, jejunum, ileum, cecum, ascending colon, spiral colon, and descending colon) and spleen stored in RNAprotect (Qiagen Inc.) by using an RNeasy minikit (Qiagen Inc.), with a DNase step added to eliminate residual genomic DNA (Qiagen Inc.). RNA quantity and quality were determined using a Bioanalyzer 2100 instrument (Agilent Technologies Canada Inc., Mississauga, Ontario, Canada), and 1,000 ng of RNA was transcribed to cDNA by following the manufacturer’s protocol (Qiagen Inc.). Reactions were run on a 384-well plate, and each reaction mixture contained 5.0 μl QuantiTect SYBR green master mix (Qiagen Inc.), 0.5 μl of each primer (10 μM) (Table 1), 3.0 μl of RNase-free water, and 1.0 μl of cDNA. Quantitative PCR was performed using an ABI 7900HT thermocycler (Applied Biosystems, Carlsbad, CA) with the following cycle conditions: 95°C for 15 min; 40 cycles of 95°C for 15 s, the primer annealing temperature for 30 s, and 72°C for 30 s; and melt curve analysis from 55 to 95°C. Some primer sequences specific to gene targets were generated using NCBI Primer-BLAST to generate an amplicon between 75 and 200 bp. Reactions were run in triplicate, and average threshold cycle (C_T) values were used to calculate expression relative to the expression of the peptidylprolyl isomerase A (PPIA), hypoxanthine-guanine phosphoribosyltransferase (HPRT), and beta-glucuronidase (GUSβ) reference genes. These genes were selected using the geNorm algorithm in qbase+ (Biogazelle, Zwijnaarde, Belgium), which identifies the stability of expression among samples (76).

TABLE 1.

Sequences and annealing temperatures for primers used to quantify bacteria

Target organism	Primer	Sequence (5′ to 3′)	Ta (°C)a	Source
Prevotella spp.	PrevF	CACCAAGGCGACGATCA	58	94
	PrevR	GGATAACGCCYGGACCT
Ruminococcus	Rflbr730F	GGCGGCYTRCTGGGCTTT	58	95
	Clep866mR	ACCTTCCTCCGTTTTGTCAAC
Clostridium cluster I	CI-F1	TACCHRAGGAGGAAGCCAC	55	96
	CI-R2	GTTCTTCCTAATCTCTACGCAT
Intestinimonas strain AF211	PFF590F	AAAACTATGGGCTCAACCCA	58	97
	PFF702R	GTCAGTTAATGTCCAGCAGG
Total bacteria	F-Tot	GCAGGCCTAACACATGCAAGTC	56	98
	R-Tot	CTGCTGCCTCCCGTAGGAGT
Bacteroides uniformis	BaUNI-F	TACCCGATGGCATAGTTCTT	55	99
	BaUNI-R	GGACCGTGTCTCAGTTCCAA
Streptococcus gallolyticus	Sg-F	TGACGTACGATTGATATCATCAAC	60	100
	Sg-R	CGCTTAACACATTTTTAGCTAATACG

^aT_a, annealing temperature.

Bacterial genomic DNA extraction from digesta and tissue samples.

For the quantification of S. Typhimurium and select commensal bacteria associated with the mucosa by qPCR, DNA was extracted from the ileal, cecal, and spiral colonic samples using the Qiagen blood and tissue kit (Qiagen Inc.) Gram-positive protocol. For the characterization of mucosa-associated bacteria by NGS analysis, DNA extracted from intestinal tissue of the spiral colon was enriched using the NEBNext Microbiome DNA enrichment kit (New England Biolabs, Ipswich, MA). For the quantification of S. Typhimurium and select commensal bacterial taxa, and the characterization of bacterial communities, in feces and digesta from the ileum, cecum, and spiral colon, DNA was extracted using the QIAamp Fast DNA stool minikit (Qiagen Inc.). A bead homogenization step using 5.0-mm-diameter stainless steel beads and following the Qiagen protocol for isolation of bacterial DNA using a TissueLyser LT instrument (Qiagen Inc.) at 30 Hz was included to ensure comprehensive extraction of genomic DNA. The bead homogenization step (30 s) was conducted three times per sample.

Quantification of Salmonella.

To enumerate S. Typhimurium bacteria by qPCR, duplicate reaction mixtures (20 μl) were prepared as follows: 10 μl of QuantiTect SYBR green master mix (Qiagen Inc.), 1 μl of each primer (0.5 μM) (IDT, San Diego, CA), 2 μl bovine serum albumin (BSA) (0.1 μg μl⁻¹) (Promega, Madison, WI), 2 μl DNA, and 4 μl DNase-free water (Qiagen Inc.). The primers used were F-(Sal) and R-(Sal) (75). Data were collected using an Mx3005p real-time PCR instrument (Agilent Technologies Canada Inc., Mississauga, Ontario, Canada). The cycle conditions used were 95°C for 5 min, followed by 40 cycles of 94°C for 15 s, 64°C for 30 s, and 72°C for 30 s. Serial dilutions of genomic DNA containing 1.5 × 10⁶ copies μl⁻¹ were used to prepare a standard curve; the concentrations of Salmonella DNA in the sample were determined based on standard curve C_T values; and the number of copies per square centimeter was calculated. A dissociation curve was included with each run to verify amplicon specificity.

Characterization of bacterial communities by culturomics.

(i) Bacteriological media. The basic protocol described by Moote et al. (77) was used. All media were reduced before use (77). Media were prepared without the addition of cysteine and were autoclaved for 5 min. Once autoclaved, warmed media were immediately transferred to a chamber containing a nitrogen (N₂)-predominant atmosphere consisting of N₂-CO₂-H₂ at 85%:10%:5% or a carbon dioxide (CO₂)-predominant atmosphere consisting of CO₂-H₂ at 90%:10% and were vigorously agitated to displace oxygen from the media. For media used in the CO₂ atmosphere, 40 ml liter⁻¹ of 8% sodium carbonate (Sigma-Aldrich, Ottawa, Ontario, Canada) was added to prevent acidification. When cooled, media were decanted into separate bottles containing the desired l-cysteine monohydrate content to remove any residual oxygen; the bottles were sealed with a screw cap containing a rubber stopper, removed from the chamber, and autoclaved for 30 min at 121°C and 15 kPa. The pHs of media were tested using pH paper (Micro Essential Laboratory, Brooklyn, NY).

Agar (1.5% agar; BD Difco) and resazurin as an oxygen indicator (25 μg ml⁻¹; Chemicon International, Inc., Temecula, CA) were added to media before autoclaving for 30 min. Media were dispensed into petri dishes and were maintained in the N₂ and CO₂ atmosphere chambers for 24 h before use. For enrichments, resazurin sodium salt (25 μg ml⁻¹) was added; media (10 ml) were dispensed into 15-ml glass Hungate tubes (Kimble-Chase, Vineland, NJ); tubes were sealed with a screw cap fitted with a black butyl rubber stopper (Bellco Glass Inc., Vineland, NJ), autoclaved for 30 min, and then transferred to the N₂ and CO₂ atmosphere chambers.

(ii) Liberation of bacteria from digesta and the mucosal surface. Ligated intestinal samples were transferred to an anaerobic chamber (Forma Scientific, Inc., Marietta, OH) containing the N₂ atmosphere. The anoxic status of anaerobic chambers was routinely monitored using a resazurin anaerobic indicator (Oxoid Inc.). Once the samples were in the chamber, ligations were aseptically removed, and the intestinal segment was incised to expose the mucosal surface and digesta. A 1-cm² sample of the intestinal wall with digesta was removed and transferred into 5 ml of reduced CB (HiMedia Laboratories LLC, West Chester, PA) in a 50-ml Falcon tube, where it was gently mixed by rocking side to side for 30 s. To isolate mucosa-associated bacteria, individual washed tissue segments were transferred to a new tube containing 5 ml of CB. The remaining suspension in the initial tube was used to isolate digesta-associated bacteria. Tubes were then vortexed vigorously (high setting for 0.5 min), and the resultant suspensions were used to isolate bacteria. Half of the tubes were retained within the N₂ atmosphere chamber, and half of the tubes were transferred to a CO₂ atmosphere chamber.

(iii) Direct plating. Bacteria suspended in CB (10 μl) were streaked onto each agar medium within the N₂ and CO₂ atmosphere chambers, and cultures were maintained for 7 days at 37°C. Where possible, cells from a minimum of five morphologically distinct colonies per culture were transferred to fresh reduced Columbia agar supplemented with 10% whole sheep blood (CBA) and were streaked for purity as warranted.

(iv) Enrichment. Bacteria suspended in CB (10 μl) were added to enrichment media, and cultures were maintained at 37°C for 12 weeks in the N₂ and CO₂ atmosphere chambers. After the incubation period, bacteria were isolated by streaking the culture onto CBA as described above for direct plating.

(v) Ichip. A modified version of the original Ichip method (53) was used. The Ichip includes multiple small diffusion chambers, to the bottom of which a membrane containing 0.2-μm pores (Sterlitech Corporation, Kent, WA) was attached using Silicone II sealant (General Electric Company, Fairfield, CT). The assembled Ichip was sterilized by autoclaving and maintained in the N₂ and CO₂ atmosphere chambers for 24 h before use. Warm (37°C) reduced PBS (0.1 M; pH 7.2) containing 0.5% agarose was used to dilute bacteria liberated from digesta and mucosal surfaces in CB. Bacterial cells in suspension (20 μl) were stained with trypan blue (20 μl) (Sigma-Aldrich) and enumerated using a Petroff-Hausser chamber at ×100 magnification. Bacterial suspensions were diluted in the reduced PBS-agarose medium to a target density of 1 cell in 200 μl, and 200-μl aliquots were dispensed into individual cells within the Ichip apparatus. The top of the Ichip was sealed with a nonporous adhesive membrane (Bio-Rad Laboratories Inc., Hercules, CA). Another salient modification from previous reported applications of the Ichip was the continuous submergence of the apparatus in freshly collected rumen fluid, which provided nutrients including vitamins and cofactors; rumen fluid was collected from fistulated beef cattle (LeRDC Animal Use Protocol Review 1614). Ichips were maintained for 12 weeks, and rumen fluid was replaced with fresh fluid at 2-week intervals. After the incubation period, the Ichips were removed and left on sterile paper towels within anaerobic chambers to dry. The Ichip top seal was then carefully removed, taking care to prevent cross-contamination among wells, and a 10-μl aliquot was removed from each well and streaked onto reduced CBA as above.

(vi) Endospore-forming taxa. Two strategies were applied to kill vegetative bacteria and to facilitate the isolation of endospore-forming taxa. These were the basic ethanol killing method described by Browne et al. (52) and tyndallization. For the former, an equal volume of ethanol (70%) was added to 750 μl of the bacterial cells suspended in CB in 2-ml sample vials, vigorously vortexed, and maintained at 37°C for 4 h. Subsamples of the CB-ethanol mixture (10 μl) were streaked onto CBA containing 0.1% taurocholic acid (Millipore, Burlington, MA). In addition, 10 μl of the suspensions was transferred to 10-ml Hungate tubes containing reduced Dehority’s medium amended with 0.5% xylan (Sigma-Aldrich) or 0.5% porcine mucin (Sigma-Aldrich), as well as CB with 5% sheep blood for enrichment. Enrichments were maintained for 12 weeks at 37°C, and 10-μl aliquots were streaked onto CBA as above.

For tyndallization, ≈500 μl of the bacterial suspension was transferred to a sterile 1.5-ml glass tube and then sealed within the anaerobic chambers. The samples were transferred to an oven, which was maintained at 100°C for 30 min to kill vegetative cells and to stimulate the germination of endospores. After the heat treatment, the tubes were returned to the anaerobic chambers and processed as for the ethanol killing method.

(vii) Recovery and storage of bacteria. Based on colony morphology, five representative colonies were selected per culture (1,523 isolates in total), and cells from the colonies were streaked for purity on CBA. After ≥7 days, bacterial biomass was suspended in 1.5 ml of CB containing 40% glycerol, and the tubes were sealed, snap-frozen on dry ice, and transferred to −80°C for medium-term storage.

(viii) Bacterial identification. Bacteria were rejuvenated from glycerol stocks on CBA at 37°C in the atmospheres from which they were originally isolated until sufficient biomass was produced. Biomass was scraped from the agar surface, placed in CB, and sedimented by centrifugation (13,200 × g for 10 min). Genomic DNA was extracted using an automated DNA extraction robot (Autogen, Inc.) according to the manufacturer’s recommendations. Amplicons of the 16S rRNA gene were generated using primers 27F and 1492R (78). Amplicons were purified using a QIAquick PCR purification kit (Qiagen N.V., Hilden, Germany) and sequenced by Eurofins Genomics (Toronto, Ontario, Canada) using primer 27F. Sequence chromatograms were visualized, assessed for quality, and trimmed using Geneious (Biomatters, Inc., San Diego, CA).

(ix) Analysis of isolated bacteria. Bacterial taxonomy was assigned using the Seqmatch tool of the Ribosomal Database Project (RDP) with the following settings selected: strain, both (type and nontype); source, isolates; size, both (>1,200 and <1,200); quality, good; taxonomy, nomenclature; KNN matches, 1 (79). Diversity metrics were made using functions of the vegan package in R (version 3.4.3), and phylogenetic relationships were analyzed using the LefSe tool on the Galaxy Instance of the Huttenhower lab (http://huttenhower.sph.harvard.edu/galaxy/). Phylogenetic trees were generated, and evolutionary analyses were conducted, from the 16S rRNA gene sequences using the MEGA-X software package (80). Briefly, a multiple sequence alignment was made using MUSCLE software, and a tree was generated using the unweighted pair group method with arithmetic means (UPGMA) (81). A bootstrap test for phylogeny was used with the number of replications set to 1,000. A tree figure was generated with FigTree, version 1.4.4, using the resulting Newick file (82, 83). The evolutionary distances were computed using the maximum composite likelihood method (84) and were expressed as the number of base substitutions per site. All ambiguous positions were removed for each sequence pair (pairwise deletion option); there were a total of 1,258 positions in the final data set (80).

Bacterial community characterization by NGS analysis.

DNA extracted from digesta and mucosa was processed with an Illumina protocol for creating 16S rRNA sequencing libraries. Extracted genomic DNA was amplified with Illumina indexed adaptor primers (V4 Schloss primers [85]). The PCR mixture contained 5 μl of PCR buffer, 1 μl of 10 mM deoxynucleoside triphosphates (dNTPs), 1 μl of 25 mM MgCl₂, 2.5 of μl of each primer, 0.25 μl of Hot Start Taq DNA polymerase (Qiagen Inc.), 32.8 μl of molecular-grade water, and 5 μl of bacterial community DNA. Amplicons were purified with a QIAquick PCR purification kit (Qiagen Inc.) according to the manufacturer’s recommendations. The effectiveness of the cleanup was checked by agarose gel electrophoresis, followed by quantification with a Qubit (Fisher Scientific, Ottawa, Ontario, Canada). Indexed DNA libraries were normalized to 1.5 ng μl⁻¹ and were pooled. A PhiX control (10%) was run with the normalized DNA library, and both were denatured and diluted to 8 pM prior to loading into the MiSeq reagent kit, v2 (500 cycles) (Illumina, San Diego, CA). An average of 5,673 16S rRNA gene amplicon reads were obtained per sample, with 94% of the reads passing the Q30 score. QIIME2 (86) was used to classify bacterial reads for digesta- and mucosa-associated communities. This analysis was done using the Tourmaline Snakemake reproducible workflow to automate QIIME2 (version 2019.7) analyses (https://github.com/ropolomx/tourmaline). Briefly, raw reads were denoised with DADA 2 (87), and representative ASVs were generated. A phylogenetic tree of ASV sequences was generated, and the taxonomy of each ASV was identified by using a naïve Bayes classifier pretrained with the reference SILVA 132 database (silva-132-99-515-806-nb-classifier.qza). Alpha diversity metrics, including the number of taxa observed, the Shannon index of diversity, and the inverse Simpson index, were calculated. The phyloseq package (version 1.28.0) of R, version 3.6.1, was used to evaluate beta diversity with a principal coordinate analysis (PCoA) of the calculated unweighted UniFrac distances, generating an ordination plot. Differential abundance between tissues was detected with the analysis of composition of microbiomes (ANCOM) in QIIME2 (88).

Biochemical pathway inference.

Biochemical pathways inferred from the 16S rRNA gene phylogenies of the mucosal data sets were analyzed with Phylogenetic Investigation of Communities by Reconstruction of Unobserved States (PICRUST2) (89). The following outputs were generated with PICRUST2: biochemical pathways, KEGG numbers, and EC numbers. Differential abundance of each of the PICRUST outputs in ST⁺ versus ST^– treatment samples was completed for individual tissues for each of the days postinoculation. Differential abundances of biochemical pathways, KEGG numbers, and EC numbers were tested with the analysis of variance (ANOVA)-like differential gene expression analysis method (ALDEx2 package, version 1.20.0; R, version 4.0.0) (90, 91).

Network analysis.

Interaction networks were generated based on Bray-Curtis distances of the taxonomically classified ASVs using the phyloseq package (version 1.32.0) (92) in R (version 4.0.0).

Quantification of commensal bacterial taxa by quantitative PCR.

To enumerate commensal taxa of interest (e.g., B. uniformis and S. gallolyticus) by qPCR, duplicate reaction mixtures (20 μl) consisting of 10 μl of QuantiTect SYBR green master mix (Qiagen Inc.), 1 μl of each primer (0.5 μM) (Table 2), 2 μl of community DNA, and 6 μl of DNase-free water (Qiagen Inc.) were prepared. Data were collected using an Mx3005p real-time PCR instrument (Agilent Technologies Canada Inc.). Cycle conditions were 95°C for 15 min, followed by 40 cycles at 94°C for 15 s, at the individual annealing temperatures (Table 2) for 30 s, and at 72°C for 30 s. Serial dilutions of genomic DNA containing 1.0 × 10⁶ copies μl⁻¹ were used to prepare a standard curve. Concentrations of bacterial DNA in the sample were determined based on standard curve C_T values, and copies per milligram or gram were calculated. A dissociation curve was included with each run to verify amplicon specificity.

TABLE 2.

Sequences and annealing temperatures for primers used for gene expression

Target	Primer	Sequence (5′ to 3′)	Ta (°C)a	Source
PPIA	Ppia-F	GACAGCAGAAAACTTCCGTG	58	This study
	Ppia-R	ACCACCCTGGCACATAAATC
HPRT	pHPRT-F	AGGCTATGCCCTTGACTACA	58	This study
	pHPRT-R	GGCTTTGTATTTTGCCTTTCCA
GUSβ	pGusB-F	CATGAGGCCTACCAGAAACC	58	This study
	pGusB-R	GAGGTGGATCCTCGTGAAAC
βD2	pBD2-F	AGCTGGCTGCAGGTATTAAC	58	This study
	pBD2-R	TCAATCCTGTTGAAGAGCGG
MUC4	pMuc4-R	GTCCCCTGGGTGTTTCTGAG	58	This study
	pMuc4-R	CATAGTGTTTCCACCCAGGAC
REGIIIγ	REG3g-F	AGCCTGTCAAGAAACACAGGAT	58	This study
	REG3g-R	ATCCAATCTCATCTAGCCCTTG
MUC1	pMuc1-F	ACCCCTATGAGCAGGTTTCT	58	This study
	pMuc1-R	CCCCTACAAGTTGGCAGAAG
TNFα	pTnf-a-F	CCACGTTGTAGCCAATGTCA	58	This study
	pTnf-a-R	GTTGTCTTTCAGCTTCACGC
IFNγ	pIFN-g-F	AGAATTGGAAAGAGGAGAGTG	58	This study
	pIFN-g-R	ACTCAGTTTCCCAGAGCTACCA
IL17	IL17a-F	CCAGACGGCCCTCAGATTAC	65	29
	IL17a-R	CACTTGGCCTCCCAGATCAC
IL10	pIL-10-F	CTGGAAGACGTAATGCCGAA	58	This study
	pIL-10-R	CAGAAATTGATGACAGCGCC
PR39	PR39-F	TAATCTCTACCGCCTCCTGG	62	29
	PR39-R	CCCGTTCTCCTTGAAGTCAC
IL1β	pIL1b-F	CCCATCATCCTTGAAACGTG	58	This study
	pIL1b-R	CTCATGCAGAACACCACTTC
IL8	IL8-F	TCCTGCTTTCTGCAGCTCTC	62	29
	IL8-R	GGGTGGAAAGGTGTGGAATG
iNOS	iNOS-F	GAGAGGCAGAGGCTTGAGAC	62	This study
	iNOS-R	TGGAGGAGCTGATGGAGTAG
TLR4	pTLR4-F	CAGCCATGGCCTTTCTCTC	58	This study
	pTLR4-R	ATGTTAGGAACCACCTGCAC
TGFβ	pTGF-B1-F	CCGGAACCTGTATTGCTCTC	58	This study
	pTGF-B1-R	TGACATCAAAGGACAGCCAC

^aT_a, annealing temperature.

Statistical analyses.

The experiment was arranged as a two (with or without Salmonella)-by-three (2, 6, and 10 dpi) factorial experiment with four replicates. These two factors and the interaction between them were treated as fixed effects. Given that the experiment was run on two separate occasions (i.e., runs), a run was treated as a random effect in the statistical model. Analyses for gene expression, histopathological measurements, Salmonella quantification, bacterial quantification, physiopathology changes, and meat pH were performed using Statistical Analysis Software (SAS; SAS Institute Inc., Cary, NC). Normality was checked and analyzed in continuous data using the MIXED procedure of SAS. In the event of a main treatment event effect (P ≤ 0.050), the least-squares means test was used to compare treatments within factors. Histopathological measurement data were analyzed using the pairwise Fisher exact test in SAS. Quantification data for commensal bacteria were not normally distributed, and the data were analyzed using the Kruskal-Wallis test in SAS. Data are expressed as means ± standard errors of the means.

Data availability.

The raw sequencing reads were submitted to the Sequencing Read Archive of NCBI under BioProject accession number PRJNA612572.