Sixteen crossbred, conventionally reared piglets from three Salmonella-fecal-negative sows were weaned at 12 days of age and shipped to the National Animal Disease Center, Ames, IA, USA. Siblings from each litter were divided and raised in two isolation rooms. Piglets tested fecal-negative for Salmonella spp. twice over a 2-week period using bacteriological culture with selective enrichment (20). At 5 weeks of age, piglets received an intramuscular injection of 10¹⁰ TCID₅₀/pig of a replication-defective human adenovirus (Ad5) engineered to express porcine G-CSF (Ad5-G-CSF, n = 9) (14). As previously described, Ad5-G-CSF was derived by directionally cloning G-CSF cDNA into the AdEasyTM XL System (Stratagene, La Jolla, CA, USA) and propagated in specialized AD-HEK-293 cells. Control pigs received the same Ad5 vector lacking the gene encoding G-CSF (Ad5-empty, n = 7). Four days later, all pigs (n = 16) were intranasally inoculated with 1 × 10⁷ colony forming unit (CFU) of a nalidixic acid-resistant derivative of S. enterica serovar Typhimurium UK1 (21) that had been passaged in swine and isolated from the ileocecal lymph node of a pig (strain name: SB 377). Fecal samples were collected at 0, 1, 2, 3, and 7 days post-inoculation (d.p.i.) for microbiota analysis as well as quantitative and qualitative Salmonella culture analyses (see below). Blood samples were collected from the jugular vein at −4, −2, 0, 1, 2, 3, and 7 d.p.i. for enumeration of circulating blood cells by flow cytometry (see below). At 7 d.p.i., all pigs were euthanized and necropsied to obtain tissue samples from the tonsil and the intestinal tract (ileal Peyer’s patches, ileocecal lymph nodes, and cecum) for quantitative and qualitative Salmonella culture analysis (see below). Procedures involving animals followed humane protocols as approved by the USDA, ARS, NADC Animal Care and Use Committee in strict accordance with the recommendations in the Guide for the Care, and Use of Laboratory Animals of the National Institutes of Health.

For quantitative bacteriology, 1 g of pig feces was combined with 5 ml PBS, vortexed, and 0.1 ml directly plated to XLT-4 medium (Beckton, Dickinson and Co., Sparks, MD, USA) containing 30 μg/ml of nalidixic acid. For tissue samples, 1 g of each tissue was combined with 2 ml of PBS in a whirlpak bag, pounded with a mallet, and homogenized in a Stomacher (Seward, Westbury, NY, USA) for 1 min. One hundred microliters of the resulting solution was aliquoted onto XLT-4 medium containing nalidixic acid. One hundred microliters of a 10-fold dilution of each fecal and tissue sample was also plated, and additional dilutions were performed when CFU reached >300/plate. Following 48 h of incubation at 37°C, black colonies were enumerated and a single colony from each plate was confirmed to be Salmonella by serogroup antiserum agglutination (Beckton, Dickinson and Co., Sparks, MD, USA). The total number of CFU for each quantitative tissue or fecal sample was calculated per gram of sample by obtaining the number of Salmonella per plate and multiplying by the dilution factor.

Qualitative bacteriology of Salmonella was performed as follows: 1 g (fecal) or 0.1 ml (homogenized tissue) samples were inoculated in 10 ml tetrathionate broth (TET; VWR, Rutherford, NJ, USA) for 48 h of growth at 37°C. Following incubation, 0.1 ml of each culture was transferred to 10 ml Rappaport–Vassiliadis medium (RV; Difco) and incubated at 37°C for 18–20 h. Cultures were streaked on XLT-4 medium containing nalidixic acid. Colonies suspicious for Salmonella were confirmed by serogroup antiserum agglutination.

Statistical analysis for Salmonella shedding in feces (CFU/g) was Log10-transformed and analyzed using a mixed linear model for repeated measures (Proc Mixed in SAS for Windows, version 9.2; SAS Institute Inc., Cary, NC, USA). Covariance structures within pigs across time were tested and modeled using the REPEATED statement to determine the optimal covariance structure. Linear combinations of the least-squares mean estimates were used in a priori contrasts after testing for a significant (P < 0.05) treatment group effect. Comparisons were made between each group at each time point, using a 5% level of significance (P < 0.05) to assess statistical differences. The endpoint data for bacterial colonization (CFU/g) of tissues collected at necropsy were Log10-transformed and analyzed by analysis of variance using a general linear model for unbalanced data. A 5% level of significance (P < 0.05) was used to assess statistical differences.

White blood cell counts were performed via flow cytometry as previously described (14). Briefly, a 50-μl aliquot of anti-coagulated (EDTA) whole blood was added to a tube containing monoclonal antibody to porcine granulocytes (6D10, Serotech, USA) with appropriate secondary fluorochrome-labeled antibody. After a 20-min incubation, cells were fixed and red blood cells lysed with the addition of 1 ml FACS lyse (BD Biosciences, USA). Microbeads (Spherotech, USA) were added to the tube immediately prior to data acquisition on a flow cytometer (BD LSR II, Becton Dickinson, USA). A gate was drawn around the beads and events were collected on each parameter (neutrophil gate was based on forward and side scatter properties and antibody labeling) until the bead event number was 500. A ratio of total counts to bead counts was used to determine the number of neutrophils per microliter of blood. Calculation of statistical significance of neutrophils per microliter of blood by treatment group was performed using a one-way ANOVA with a Dunnett’s Multiple Comparison Test. A 5% level of significance (P < 0.05) was used to assess statistical differences.

Amplicon libraries of the 16S rRNA gene were generated and sequenced according to Kozich et al. (22), with our primers and procedures described previously (23). Briefly, PCRs contained the following: 17 μl AccuPrime Pfx SuperMix (Life Technologies, Grand Island, NY, USA), 5.0 μM each of the primers i5 + V3 and i7 + V1, and 25 ng of fecal DNA. The following PCR conditions were used: 2 min at 95°C, 22 cycles of (20 s at 95°C, 15 s at 55°C, 5 min at 72°C), 72°C for 10 min. Libraries were normalized using the SequalPrep Normalization Plate Kit (LifeTechnologies) and quantified using both Bioanalyzer (Agilent Technologies, Santa Clara, CA, USA) and Kapa SYBR Fast qPCR (Kapa Biosystems, Wilmington, MA, USA). Normalized pools were sequenced using version 3 (300 × 2) chemistry on the MiSeq instrument (Illumina, San Diego, CA, USA) according to manufacturer’s instructions.

Contig assembly, sequence alignment, chimera removal, and non-bacterial sequence removal were performed in the program mothur (version 1.33.3) (24). Sequences that only occurred once or twice across all samples were removed as potentially spurious. Sequences were rarified to 3,000 sequences, clustered into operational taxonomic units (OTUs) at 97% similarity, and analyzed for community metrics, including richness (25), evenness, and diversity. Analysis of similarity (ANOSIM) and non-metric multidimensional scaling (NMDS) analyses were conducted in PAST (26). Additionally, the OTUs were assigned to bacterial taxonomy using mothur’s implementation of the SILVA database (27). One sample from a pig in the Ad5-G-CSF group at day 7 yielded insufficient sequences to be analyzed. The 16S rRNA gene sequences associated with this study were deposited in Genbank under Bioproject PRJNA339155.

The effects of Ad5-G-CSF administration and S. Typhimurium challenge on circulating neutrophils were determined by enumerating neutrophils in the blood at various days after Ad5-G-CSF administration and S. Typhimurium challenge. S. Typhimurium challenge alone induced a significant approximately threefold increase in circulating neutrophil counts, as values post-challenge were greater when compared to values on the day of challenge (day 0) (Figure 1A). Circulating neutrophils were also enumerated in pigs that were intramuscularly injected with 10¹⁰ TCID₅₀ Ad5-G-CSF prior to Salmonella exposure. As expected on the day of S. Typhimurium challenge (day 0), which was 4 days after Ad5-G-CSF administration, a significant neutrophilia occurred compared to pre-Ad5-G-CSF numbers (day −4, Figure 1B) or compared to Ad5-empty-treated controls on day 0 (day 0, Figure 1A). Following Salmonella challenge of the Ad5-G-CSF group, an additional significant increase in circulating neutrophils was observed at 3 and 7 d.p.i. compared to day 0 (day of Salmonella challenge). Collectively, Ad5-G-CSF administration induced a significant and sustained ~10-fold increase in the number of circulating neutrophils, and Salmonella challenge also induced significant increases in circulating neutrophils.

Salmonella shedding and tissue colonization were compared between Ad5-G-CSF and Ad5-empty-treated pigs. Ad5-G-CSF-treated pigs shed significantly less Salmonella (10³ CFU/g feces) when compared to the Ad5-empy-treated pigs at 2 and 3 d.p.i. (10^4–5 CFU/g feces) (Figure 2). This 1- to 2-log difference between the treatment groups dissipated by 7 d.p.i. as Salmonella shedding in the feces of Ad5-empty-treated pigs declined to the level of the Ad5-G-CSF-treated pigs. Typical for swine, a transient fever was observed in the S. Typhimurium-challenged pigs, peaking at 2 days post-challenge; no significant difference was observed in the elevated body temperatures between treatment groups (data not shown). Gastrointestinal tissues (ileocecal lymph nodes, Peyer’s patch region of the ileum, and cecum) were analyzed at 7 d.p.i., and all tissues were Salmonella positive in both Ad5-G-CSF-treated and Ad5-empty-treated pigs. Of these three tissues, the Peyer’s patch region of the ileum exhibited a significant 0.5-log reduction in Salmonella colonization in the Ad5-G-CSF-treated pigs compared to the Ad5-empty-treated group (Figure 3A). A striking difference in tonsil colonization was observed between treatment groups (Figure 3B). Eight of the nine Ad5-G-CSF-treated pigs harbored no detectable Salmonella in the tonsil, with only one pig being qualitatively positive for Salmonella in the tonsils (i.e., by enrichment). By contrast, all seven Ad5-empty-treated pigs harbored Salmonella in the tonsils at an average of ~10,000 CFU/g. These data suggest that prophylactic administration of Ad5-G-CSF can reduce Salmonella colonization and subsequent fecal shedding, including the tonsils that have been implicated in the carrier-status of Salmonella (28–31).

Fecal 16S rRNA gene sequence data were used to compare the gastrointestinal bacterial communities of the Ad5-G-CSF and Ad5-empty treatment groups following Salmonella challenge. No significant differences in indices for diversity, evenness, or richness were detected among treatments or timepoints. OTU-based analysis of bacterial community structure showed that at 7 days post-Salmonella challenge, the microbiota of pigs that had received Ad5-G-CSF was not significantly different from that of day 2 or 3 (ANOSIM, p > 0.05; R < 0.1), but the microbiota of pigs that received Ad5-empty treatment was significantly different at day 7 compared to all previous time points (ANOSIM, p < 0.05; R > 0.25). However, the difference between the Ad5-G-CSF-treated and Ad5-empty-treated groups at day 7 was insignificant. The dissimilarity of the microbiotas between days 3 and 7 was visualized via an NMDS plot, which showed the disturbed microbiota at day 7 in the Ad5-empty-treated animals compared to Ad5-G-CSF-treated animals (Figure 4). These results demonstrate that Ad5-G-CSF administration slightly decreases the beta-diversity changes in the microbiota that are caused by Salmonella challenge, suggesting that Ad5-G-CSF mitigated the disturbance to the gut microbiota that was caused by Salmonella.