All experiments were conducted according to guidelines established by the United States Department of Agriculture (USDA) animal care and use committee, which operates in accordance with established principles (14). Broiler chickens from a single genetic lineage were obtained from a commercial hatchery. At hatch, straight-run (mixed sexes) chicks were placed in floor pens (4 m × 4 m) containing wood shavings, supplemental heat, water and a balanced, un-medicated corn, and soybean meal-based chick starter diet ad libitum. The feed contained 23% protein and 3,200 kcal of metabolizable energy/kg of diet, and all other nutrient levels met or exceeded established requirements (15). The birds were not vaccinated or given any medications during the course of the study.

A poultry isolate of Salmonella enterica serovar Enteritidis (S. Enteritidis) was obtained from the National Veterinary Services Laboratory (Ames, IA, USA), and was selected for resistance to nalidixic acid and novobiocin and maintained in tryptic soy broth (Difco Laboratories, Sparks, MD, USA) containing antibiotics (20 µg/mL nalidixic acid and 25 µg/mL novobiocin; Sigma Chemical Co., St. Louis, MO, USA). A stock culture was prepared in sterile phosphate buffered saline (PBS) and adjusted to a concentration of 1 × 10⁹ colony forming units (cfu)/mL as previously described (16). The challenges were then diluted from the 1 × 10⁹ cfu/mL stock culture to the desired concentration. The viable cell concentration of the challenge dose for each experiment was determined by colony counts on XLT4 agar base plates with XLT4 supplement (Difco) and nalidixic acid and novobiocin (XLT-NN).

Four-day-old broiler chicks were challenged orally with S. Enteritidis (0.5 mL; 4.8 × 10⁵ cfu/chick) while controls were administered sterile PBS; 0.5 mL. Cloacal swabs were collected 3 days post challenge to confirm the controls were not infected and that all birds that were challenged were culture positive for S. Enteritidis. Briefly, a sterile cotton swab was gently inserted into the cloaca and a fecal sample was collected. The entire swab and sample for each bird was then placed into a separate tube containing tetrathionate enrichment broth (10 mL, Difco) and incubated overnight at 41°C. Following enrichment, 10 µL were streaked onto XLT-NN plates, incubated 24 h at 41°C, then the plates examined for non-lactose fermenting NN-resistant Salmonella colonies. Representative colonies were confirmed positive by plate agglutination using specific Group D₁ antisera (Difco).

One-day-old broiler chickens were randomly distributed into two experimental groups: non-infected control and infected (n = 50). Early samples were collected between 4 and 7 days post-infection (dpi), middle samples were collected between 10 and 17 dpi, and late samples were collected between 24 and 37 dpi. The experiments were conducted on two separate occasions.

Control and infected chickens (n = 10) were euthanized by cervical dislocation and necropsied at three timeframes (early, middle, and late) over the course of a 42-day grow-out. The cecum from each chicken was removed aseptically, and the contents (0.25 g) were serially diluted to 1:100, 1:1,000, or 1:10,000 and spread onto XLT-NN plates to enumerate S. Enteritidis. The plates were incubated at 41°C for 24 h, and the number of NN-resistant S. Enteritidis cells per g of cecal contents determined. A piece of jejunum (100 mg) was collected and rinsed with PBS to remove content and then placed into a cryovial containing 1.5 mL RNAlater RNA stabilization reagent (Qiagen, Valencia, CA, USA) and stored at −20°C until tissue homogenization and RNA isolation was performed for quantitative real-time Reverse Transcriptase-PCR (qRT-PCR). Additionally, a section of jejunum (100 mg) was collected from each bird, rinsed with PBS to remove content, and then immediately flash frozen in liquid nitrogen to preserve kinase enzymatic activity for the array. Samples were taken from liquid nitrogen and transferred to a −80°C freezer until further experimental procedures were conducted. Following microbiological analysis of the cecal contents (described previously), the jejunum tissues from four birds with the highest and four birds with the lowest levels of recoverable S. Enteritidis (out of the 10 birds per time) were used for the peptide arrays.

PepStar peptide microarrays were obtained from JPT Peptide Technologies GmbH (Berlin, Germany), and the peptide array protocol was carried out as previously described (6) with the following modifications (8, 17). Jejunum tissue samples were weighed to obtain a consistent 40 mg sample for the array protocol. Samples were homogenized by a hand-held TissueRuptor (Qiagen, Valencia, CA, USA) in 100 µL of lysis buffer (20 mM Tris–HCl pH 7.5, 150 mM NaCl, 1 mM Ethylenediaminetetraacetic acid, 1 mM ethylene glycol tetraacetic acid, 1% Triton X-100, 2.5 mM sodium pyrophosphate, 1 mM Na₃VO₄, 1 mM NaF, 1 µg/mL leupeptin, 1 g/mL aprotinin, and 1 mM Phenylmethylsulphonyl fluoride). All chemicals purchased from Sigma-Aldrich, Co. (St. Louis, MO, USA) unless otherwise indicated.

The Phospho Explorer Antibody Array kit (catalog PEX100; Full Moon BioSystems, Sunnyvale, CA, USA) consists of over 1,300 antibodies from over 30 signaling pathways and is an alternative approach to procuring phosphor-specific antibodies individually and performing numerous western blot assays. The protocol was carried out as per manufacturer’s instructions with the exception that the tissue was homogenized using a hand-held TissueRuptor (Qiagen, Valencia, CA, USA) instead of the bead and vortex method suggested in the kit.

Data normalization and analysis was performed for both the kinome and antibody microarrays as described (17). Images were gridded using GenePix Pro software, and the spot intensity signal was collected as the mean of pixel intensity using local feature background intensity calculation with the default scanner saturation level. The data was then analyzed using the Platform for Intelligent Integrated Kinome Analysis (PIIKA2) peptide array analysis software (http://saphire.usask.ca/saphire/piika/index.html). Briefly, the resulting data points were normalized to eliminate variance due to technical variation, for example, random variation in staining intensity between arrays or between array blocks within an array. Variance stabilization and normalization was performed. Note: as the arrays were printed with triplicate peptide blocks there are three values for each peptide. Using the normalized data set comparisons between treatment and control groups were performed, calculating fold-change and a significance P-value. The P-value is calculated by conducting a one-sided paired t-test between treatment and control values for a given peptide.

This consistent analysis method facilitates a more direct comparison between the two distinct array datasets and allows for a statistically robust analysis of the phosphorylation events being measured. Gene ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis was performed by uploading the statistically significant peptide lists to the Search Tool for the Retrieval of Interacting Genes (STRING¹) (18).

Tissue homogenization was performed using a BeadBug microtube homogenizer (Benchmark Scientific, Edison, NJ, USA). Briefly, a piece of tissue (30–40 mg) was removed from RNAlater and placed in a 2 mL prefilled tube containing 1.5 mm high impact zirconium beads (TriplePure M-Bio Grade; Benchmark Scientific). Lysis buffer (350 µL; RNeasy Mini Kit; Qiagen) was added and the sample was homogenized in the BeadBug for 2 min on the maximum speed. Total RNA was then isolated from the homogenized samples according to the manufacturer’s instructions, eluted with 50 µL RNase-free water, and stored at −80°C until qRT-PCR analyses performed.

Interleukin 6 and CXCLi2 mRNA expression was quantified using a well-described method. Primers and probes for cytokines, chemokines, and 28S RNA-specific amplification has been previously described (19–21). The qRT-PCR was performed using the TaqMan one-step RT-PCR master mix reagents (Applied Biosystems, Branchburg, NJ). Amplification and detection of specific products were performed using the Applied Biosystems 7500 Fast Real-Time PCR System with the following cycle profile: one cycle of 48°C for 30 min, 95°C for 20 s, and 40 cycles of 95°C for 3 s and 60°C for 30 s. Quantification was based on the increased fluorescence detected by the 7500 Fast Sequence Detection System due to hydrolysis of the target-specific probes by the 5′ nuclease activity of the rTth DNA polymerase during PCR amplification. To correct for differences in RNA levels between samples within the experiment, the correction factor for each sample was calculated by dividing the mean threshold cycle (C_t) value for 28S rRNA-specific product for each sample, by the overall mean C_t value for the 28S rRNA-specific product from all samples. The corrected cytokine mean is calculated: (Average of each replicate × cytokine slope)/28s slope × 28 s correction factor. The data shown are corrected 40-C_t values.

The mean and SEM for each cytokine/chemokine were calculated at each time and statistical analyses performed (Student’s t-test); comparisons were made between a single timeframe comparing the birds with high and low levels of S. Enteritidis cecal colonization. No statistical analysis was performed for the differences in bacterial load. Details for the array analysis are provided in the Data Analysis: Kinome and Antibody Arrays section described above. For all analyses, significance was considered if P ≤ 0.05.

Cloacal swabs were collected on all birds in the study three days post challenge to confirm the controls were not infected and that all challenged birds were culture positive for the challenge strain of S. Enteritidis. All birds administered the challenge were positive for S. Enteritidis while all of the controls were negative (data not shown).

Within this single line of birds, varying levels of S. Enteritidis cecal colonization were observed and, therefore, studying birds with relatively high and low numbers of recoverable S. Enteritidis was pursued. The levels of S. Enteritidis colonization for the high and low groups at each timeframe are summarized in Figure 1. Chickens in the early time point and designated to the low group had 3.0 ± 0.9 Log₁₀ cfu of recoverable S. Enteritidis while the birds in the high group had 5.9 ± 0.3 Log₁₀ cfu. The values for the middle group were 0.8 ± 0.4 and 4.4 ± 1.1 Log₁₀ cfu S. Enteritidis for the low and high groups, respectively. The recoverable S. Enteritidis continued to decline by the late timeframe and were 0.3 ± 0.2 and 2.2 ± 0.5 Log₁₀ cfu S. Enteritidis for the low and high groups, respectively.

The GO Consortium assigns defined terms representing gene product properties and is broken into three categories: cellular component, molecular function, and BP. The BP terms include operations or sets of molecular events with a defined beginning and end that pertain to the functioning of the integrated living units.² Using STRING functionality, GO results for BP were generated for each dataset. The total number of BPs associated with each time and bacterial load were: early low = 1,218; early high = 617; middle low = 1,084; middle high = 1,001; late low = 942; and late high = 1,049 (data not shown). The most significant (based on P-value) immunologically relevant GO BP were selected for further analysis. Analysis of the kinome data showed distinct differences in the observed BP between the loads of bacteria, and some of the central differences are provided in Table 1. Each term listed had a false discovery rate (FDR) P ≤ 0.01. The BP with high numbers of peptide phosphorylation events associated with low levels of S. Enteritidis colonization regardless of time (early, middle, or late) are clustered around immune responses and subsequent signaling pathways and include: cell surface receptor signaling pathway, positive regulation of cellular processes, defense response, innate immune response, regulation of immune response, immune system process, and regulation of signaling. Activation of the BP were observed at the early, middle, and late infection times, but the numbers of differentially phosphorylated peptides between the low/high loads of S. Enteritidis were greatest at the early and middle points. By late in the infection, the numbers of peptides was similar between the birds with low and high loads of S. Enteritidis as might be expected based on the overall lower numbers of bacteria recovered (Figure 1). These data point to the early immune mechanisms that aid in controlling S. Enteritidis.

Using STRING functionality, KEGG pathway results were generated for each dataset. To ensure that changes in phosphorylation were a direct result of the infection, the results were corrected using the appropriate age-matched controls. The KEGG pathway results showed numerous pathways that were significantly different between the birds with high and low loads of S. Enteritidis at each time of the infection (P ≤ 0.05 FDR). In order to be included, a pathway had to be significant for each bird within a group and time. Pathways that were not significant for each bird at a specific time and bacterial load were excluded. Additionally, the numbers for a subset of the significantly different peptides within each of the KEGG pathways are also provided. The numbers shown are a small fraction of the total number of significant peptides within a specific pathway; however, as based on our criteria to be included in the dataset, a peptide had to be statistically significantly different from control for every bird in a given group (i.e., every early/low bird, early/high bird, middle/low bird, middle/high bird, late/low bird, or late/high birds).

The significant KEGG pathways observed at the early time are shown in Table 2, and the pathways for the middle and late times are provided in Tables 3 and 4, respectively. Pathways listed in bold showed statistically significant changes at all time points in birds with either low or high loads of S. Enteritidis. The common pathways observed in chickens with low loads of S. Enteritidis included: chemokine signaling pathway, Fc ε RI signaling pathway, focal adhesion, insulin signaling pathway, Jak–Stat signaling pathway, mitogen-activated protein kinase (MAPK) signaling pathway, neurotrophin signaling pathway, pathways in cancer, T cell receptor signaling pathway, and Tuberculosis. The specific proteins associated with each of the pathways, including those specifically related to cancer, would also be pivotal in determining the hosts’ immunological response against a challenge and/or disease. The only pathway that was significantly different at each time in birds with high levels of S. Enteritidis colonization was the MAPK signaling pathway.

All peptides that were statistically different (P ≤ 0.05) for each time were input into the STRING database (22) and diagrams depicting the protein–protein interactions were generated (Figure 2). The most striking difference in the magnitude of the protein–protein interactions was observed between the birds with low or high loads of S. Enteritidis colonization at the early time (Figures 2A,B, respectively). The protein–protein interactions for middle low and middle high birds are shown in Figures 2C,D, respectively, and the late low and late high interactions are shown in Figures 2E,F, respectively.

To validate the kinome peptide array data, an antibody array containing both pan-specific and phospho-specific antibodies was employed instead of performing individual western blot analyses (8). Analysis of antibody array data via the STRING database showed similar GO BP and KEGG pathway activity at the early and middle times as observed with the kinome array; representative samples are shown in Table 5. Though outside the scope of this manuscript, but for validation purposes only, a small number of individual peptides associated with KEGG pathways identified with the kinome array (Tables 2–4) were selected and evaluated to demonstrate consistency between the kinome and antibody arrays. Phospholipase C gamma 1 (PLCG1), protein tyrosine kinase 2 (Pyk2), Raf1, and SMAD2 shared similar phosphorylation/de-phosphorylation patterns and were, in general, in agreement between the two arrays, thus further validating the kinome array results (Table 6). The antibody array was not utilized on the late samples since consistency was demonstrated at the early and middle times. Furthermore, as already described, there were fewer differences between the birds with high and low loads of S. Enteritidis at the late time point lessening the applicability of a comparative validation technique.

The expression of IL6 and CXCLi2 mRNA was quantified (40-C_t) in tissue from birds with high and low levels of S. Enteritidis cecal colonization at early, middle, and late times. Birds with lower levels of S. Enteritidis at the early time point had significantly (P ≤ 0.05) higher mRNA expression levels of CXCLi2 than birds with higher loads of S. Enteritidis (14.5 and 13.6, respectively). There were no statistical differences in CXCLi2 mRNA expression at the middle and late times. There were no differences in IL6 between birds with high and low loads of S. Enteritidis colonization compared to one another or the respective controls at the early, middle, or late times (data not shown).