Porcine intestinal epithelial cell line IPEC-1, derived from the small intestine of a newborn unsuckled piglet, was cultured in Dulbecco's Modified Eagle Medium (DMEM)/Ham's F-12 (1:1) medium (Life Technologies, Carlsbad, CA, USA) supplemented with 5% fetal bovine serum (FBS, PAA Laboratories GmbH, Pasching, Austria), epidermal growth factor (5 μg/mL) (Sigma–Aldrich, St. Louis, MO, USA), insulin (10 μg/mL), transferrin (5 μg/mL), sodium selenite (5 ng/mL) (ITS Premix, Sigma) and 2 mM L-glutamine (Life Technologies). Human intestinal epithelial cells INT-407 (human embryonic intestine, ATCC CCL-6) were cultured in RPMI-1640 medium (Lonza, Basel, Switzerland) and supplemented with 10% FBS (PAA Laboratories GmbH) and 2 mM L-glutamine (Life Technologies). All cell lines were seeded in multi-well tissue culture plates (Thermo Fischer Scientific, Waltham, USA) the day before the assay, and allowed to reach confluence for the in vitro infection. The cells were maintained in an atmosphere of 5% CO₂ at 37°C.

A confirmed isolate of C. jejuni from chicken feces was recovered from stocks kept at −80°C by plating on Columbia sheep blood agar (Oxoid, Basinstoke, Hampshire, UK) for 48 h at 37°C under microaerobic conditions (AnaeroGen system, Oxoid). The bacteria were harvested from plate and resuspended in fresh cell culture media. The optical density (OD₆₀₀) was adjusted to 1 to achieve 10⁸ CFU/ml for their straight inoculation into INT-407 and IPEC-1 cells at a multiplicity of infection (MOI) of 100/1. All in vitro cell infections with bacteria were performed in triplicate as previously described (Aguilar et al., 2014) using two points in the study time course: 3 h (early time, T3h) and 24 h (late time, T24h). The same time course was used with the uninfected controls.

The cells were centrifuged at 10,000 × g for 3 min, and cell pellets were frozen in liquid nitrogen until used for protein extraction. Approximately 10⁷ cells were pooled from each time point and homogenized with a needle (27G) in 500 μl lysis buffer [1% Triton X-100 supplemented with Complete protease inhibitor cocktail (Roche, Basel, Switzerland)]. The samples were sonicated for 1 min in an ultrasonic cooled bath, followed by 10 s of vortexing. After 3 cycles of sonication-vortexing, total cell extracts were centrifuged at 200 × g for 5 min to remove cell debris. The supernatants were collected and protein concentration was determined using the Bradford Protein Assay (Bio-Rad, Hercules, CA, USA) with bovine serum albumin (BSA) as standard. Protein extracts (150 μg) were on-gel concentrated by SDS-PAGE as previously described (Villar et al., 2014). The unseparated protein bands were visualized by staining with GelCode Blue Stain Reagent (Thermo Scientific, Waltham, MA, USA), excised, cut into 2 × 2 mm cubes and digested overnight at 37°C with 60 ng/ml of sequencing grade trypsin (Promega, Madison, WI, USA) at 5:1 protein:trypsin (w/w) ratio in 50 mM ammonium bicarbonate, pH 8.8 containing 10% (v/v) acetonitrile (Shevchenko et al., 2006). The resulting tryptic peptides were extracted by 30 min-incubation in 12 mM ammonium bicarbonate, pH 8.8. Trifluoroacetic acid was added to a final concentration of 1% and the peptides were finally desalted onto OMIX Pipette tips C18 (Agilent Technologies, Santa Clara, CA, USA), dried-down and stored at –20°C until mass spectrometry analysis.

The desalted protein digest was resuspended in 5% acetonitrile with 0.1% formic acid and analyzed by reverse phase liquid chromatography coupled online with mass spectrometry (RP-LC-MS/MS) using an Ekspert nLC 415 system coupled to a 6,600 TripleTOF mass spectrometer (AB SCIEX, Framingham, US) through Information-Dependent Acquisition (IDA) followed by SWATH (Sequential Windowed data independent Acquisition of the Total High-resolution Mass Spectra). Approximately 4 μg of each protein digest from each of the replicate samples were pooled together as a mixed sample for each condition (control and infected cells) and each cellular type. Pooled mixed samples were then used for the generation of the reference spectral ion library as part of SWATH-MS analysis. The peptides were concentrated using a 0.1 × 20 mm C18 RP precolumn (Thermo Scientific), and then separated using a 0.075 × 250 mm C18 RP column (New Objetive, Woburn, MA, USA) operating at 0.3 ml/min. Peptides were eluted using a 120-min gradient from 5 to 40% solvent B followed by 15-min gradient from 40 to 60% solvent B (Solvent A: 0.1% formic acid in water, solvent B: 0,1% formic acid in acetonitrile) and directly injected into the mass spectrometer for analysis. Three technical replicates of each mixed sample were analyzed. For IDA experiments, the mass spectrometer was set to scanning full spectra (350–1,400 m/z) using 250 ms accumulation time per spectrum, followed by up to 50 MS/MS scans (100–1,500 m/z). Candidate ions with a charge state between +2 and +5, and counts per second above a minimum threshold of 100, were isolated for fragmentation. One MS/MS spectrum was collected for 100 ms, before adding those precursor ions to the exclusion list for 15 s (mass spectrometer operated by Analyst® TF 1.6, AB SCIEX). Dynamic background subtraction was turned off. MS/MS analyses were recorded in high sensitivity mode with rolling collision energy on and a collision energy spread of 5. For SWATH quantitative analysis, 8 μg of each mixed sample were subjected in triplicate to the cyclic data independent acquisition (DIA) of mass spectra using the SWATH variable windows calculator (V 1.0, AB SCIEX) and the SWATH acquisition method editor (AB SCIEX), following previously established methods (Gillet et al., 2012). A set of 50 overlapping windows was constructed (containing 1 m/z for the window overlap), covering the precursor mass range of 400–1,250 m/z. For these experiments, a 50 ms survey scan (350–1,400 m/z) was acquired at the beginning of each cycle, and SWATH-MS/MS spectra were collected from 100 to 1,500 m/z for 70 ms at high sensitivity mode, resulting in a cycle time of 3.6 s. Collision energy for each window was determined according to the calculation for a charge +2 ion-centered upon the window with a collision energy spread of 15.

To create a spectral library with those peptides that have been detected and identified in the experimental time course, the IDA MS raw files were combined and subjected to database searches in unison using ProteinPilot software v. 5.0.1 (AB SCIEX) with the Paragon algorithm. Spectra identification was performed by searching against the Homo sapiens proteome for INT samples, Sus scrofa taxa for IPEC-1 samples and the C. jejuni taxa for INT and IPEC infected samples (Uniprot Databases: 70,826, 34,409, and 41,050 entries respectively, in December 2015) with the following parameters: iodoacetamide cysteine alkylation, trypsin digestion, gel-based ID as special factor, identification focus on biological modification and evolutionary variants and thorough ID as search effort. The detected protein threshold was set at 0.05. An independent False Discovery Rate (FDR) analysis, using the target- decoy approach provided by ProteinPilot, was used to assess the quality of identifications. Positive identifications were considered when identified proteins reached a 1% global FDR. For SWATH processing, up to 10 peptides with seven transitions per protein were automatically selected by the SWATH Acquisition MicroApp2.0 in the PeakView2.2 software (AB SCIEX) with the following parameters: 15 ppm ion library tolerance, 5 min XIC extraction window, 0.01 Da XIC width, and considering only peptides with at least 99% confidence and excluding those which were shared or contained modifications. However, to ensure reliable quantitation, only proteins with 3 or more peptides available for quantitation were selected for XIC peak area extraction and exported for analysis in the MarkerView 1.3 software (AB SCIEX). Global normalization was performed according to the Total Area Sums of all detected proteins in the samples. A Student's T-test was used to perform two-sample comparisons between the averaged area sums of all the transitions derived for each protein across the three replicate runs for each sample under comparison, in order to identify proteins that were significantly differentially represented between infected and uninfected samples.

Ingenuity Pathway Analysis (IPA) web-based application (Ingenuity Systems Inc., Redwood City, CA, USA) was used to assess the biological meaning in the host proteome datasets. IPA retrieves biological information from the literature and then integrates the differentially expressed proteins into functions and pathways with biological meaning and significance (p < 0.05). Functional association networks of the potentially interacting proteins were generated using STRING (Search Tool for the Retrieval of Interacting Genes, v.10 web server, http://stringdb.org/), a database of known and predicted protein interactions (Szklarczyk et al., 2011). Virulence factors expressed in C. jejuni-infected INT-407 cells were used as dataset framework for mapping functional pathways. The model was enriched up to 100 partners with different sorts of associations to the proteins included in the model.

Differentially expressed (DE) proteins were identified both in human and porcine intestinal epithelial cells after C. jejuni infection (Figure 1, Table S1). A total of 366 and 485 DE proteins were identified at T3h and T24h time points of infection, respectively, in INT-407 cell line. We then used Ingenuity Pathway Analysis (IPA) to examine the predicted biological effects of the protein expression differences seen between control and infected cells. The results of this analysis (Table S2) predicted a significant activation of the inflammatory response, increased at T3h compared to T24h, and associated to the differential expression of many proteins involved in cell migration and infiltration, cell movement, cytoskeleton organization and promotion of reactive oxygen species (ROS) (Figure 2A). IPA analysis also predicted that infection of cells, molecular transport and synthesis of proteins were functions activated at T3h and reduced at T24h. Similarly, INT-407 infected cells showed increased endocytosis, proliferation and viability at T3h compared to T24h. Cell death as well as gene expression, mainly represented by mRNA expression and translation process, were significantly decreased at early and late time points (Figure 2A, Table S2). Additionally, we found changes in the host energy metabolism at T24h characterized by upregulation of key glycolytic enzymes and transporters for glucose uptake, and downregulation of enzymes participating in the pyruvate metabolism and oxidative phosphorylation (Figure 2A, Table S2). The canonical signaling and metabolic pathways of all differentially expressed proteins in C. jejuni-infected cells at 3 and 24 h are shown in Table S3. Table 1 lists the top 10 most significantly enriched pathways. As shown in Table 1, diverse and complex signals related to cell immune/inflammatory transduction pathways, cell death, cell growth and proliferation, and metabolic pathways are involved in the intestinal mucosa after C. jejuni infection. The most significant canonical pathway both at early and late time post-infection was eukaryotic initiation factor-2 (eIF2) signaling, which is involved in eukaryotic protein synthesis and plays a central role in antibacterial response. Other pathways involved in translational regulation such as regulation of eIF4 and p70S6K signaling and mTOR signaling, as well as protein ubiquitination and mitochondrial dysfunction were also induced at both times of infection in human cells. Also in INT-407 cells, Ras related nuclear protein (RAN) signaling, caveolar-mediated endocytosis, tRNA charging and granzyme B signaling were induced only at earlier stages of infection, while remodeling of epithelial adherens junction and phagosome maturation were induced only at later times.

In the porcine epithelial cells, we found 355 upregulated and 209 downregulated proteins shortly after C. jejuni infection (Figure 1). Later, we observed upregulation of 280 molecules and downregulation of 322. The response proteins DISP2 and NAA15 were among the most upregulated molecules, finding higher levels of expression at later stages of the infection (Table S1). When exploring the biological functions altered (Figure 2B, Table S2), we observed a predicted inhibition of apoptosis signaling during early infection. This decreased apoptosis and subsequent increase in cell proliferation was mainly IL-8 and PI3K/AKT mediated (Table S3). In both human and porcine epithelial cell lines, cell survival was activated in early stages, but tended to be inhibited at later stages of infection, as apoptosis increased.

Infection of this porcine epithelial cell line was demonstrated by upregulation of proteasome subunits PSMA1, PSMA2, PSMA5, PSMC5, and PSMD12, as well as other molecules such as CLTA, BSG, IDH1, KPNB1, MAP4, PCBP2, PGM1, RPL10A, RPL12, SF3A1, SF3B2, SNRPD3, TAGLN2, and TFRC (Table S1). However, we did not find upregulation of canonical inflammatory pathways at any time point after infection, and acute phase response was downregulated (Table S3). Actin polymerization was activated at early but inhibited in late stages of infection (Table S3). The endocytic process (membrane ruffling) mediated by beta integrins (e.g., CDC42 and RAC1) was mildly increased at T3h, but inhibited T24h after infection. In IPEC-1 cells, eIF2 signaling pathway was predictably inhibited at early infection, and activated later (Table 1, Table S3). Similarly, cholesterol biosynthesis metabolic process was downregulated at early stages, while it was increased at later stages, thus inhibiting NFκB-mediated transcription of inflammatory mediators. Glycolisis and gluconeogenesis were upregulated 24 h after infection.

The label-free quantification by SWATH-MS allowed the identification of 53 and 31 C. jejuni proteins upon infection of INT-407 (Table S4) and IPEC-1 cells (Table S5). Some of these proteins are related to virulence and others are involved in bacteria survival, metabolism, DNA replication and genetic information processing. The virulence factors expressed by C. jejuni in human and porcine intestinal epithelial cells are shown in Table 2. Eight virulence proteins were exclusively detected in INT-407 cells, including two type VI secretion system effectors (pebA and hcP), two adaptive response proteins (the chaperone dnaK and the histidine kinase cheA), the p19 iron acquisition protein, the thiol-peroxidases Tpx that plays a role in oxidative stress protection, and the two-component regulator Cj0355c. Ten C. jejuni virulence proteins associated to adherence, adaptive response, oxidative stress and iron acquisition were expressed both in human and porcine intestinal epithelial cell lines. Interestingly the expression of some of these proteins in each cell type was markedly different. For example, at early infection, the flagellin protein flaA, the methyl accepting chemotaxis protein cstIII (p < 0.01) and the chaperonin groL (p < 0.01) expression was upregulated in INT-407 cells (Table S4). In contrast, flaA and groL expression was downregulated in porcine IPEC-1 cells after C. jejuni infection whereas cstIII (p = 0.33) remained unchanged (Table S5). Iron captation proteins such as iron deficiency induced protein cfbpA and the ferric enterobactin receptor cfrA expression resulted upregulated at T24h in both cell lines compared to T3h (p < 0.001).

A relevant number of C. jejuni proteins identified in this study were involved in metabolism. Thus, 33 proteins associated to metabolic processes were detected in INT-407 infected cells, 16 of these proteins showed significantly (p < 0.05) reduced levels at T24h compared to T3h and only three proteins showed a moderate increase in their levels at 24 h post-infection. This C. jejuni metabolic downshift at T24h of infection was not observed in IPEC-1, where from 12 C. jejuni identified proteins associated to metabolism, only hisF (p < 0.01) and acnB (p < 0.05) resulted significantly downregulated after 24 h of infection.

By using STRING tool we built an enriched model to predict associations among observed C. jejuni proteins and functional partners (Figure 3, Table S6). The predicted model organized proteins in different clusters. The main cluster was built around cheA chemotaxis protein, surrounded by 15 functional proteins in total. This “chemotaxis cluster” was closely linked to the “motility cluster,” where the adhesion protein flaB was associated to flagellar proteins such as fliD, fliY, or motA as well as the protein pseE, involved in the glycosylation of the flagella (Figure 3). The heat-shock proteins dnaK and htpG were clustered together and with the chemotaxis protein groL. Finally, two “iron uptake clusters” were conformed. The siderophor cbpA and iron captation protein p19 were joined to the oxidative stress protein ahpC and other functional partners, which include the permease cfbpB, the lipotrotein cj0176c, and the tioredoxin trxB. On the other hand, iron captation protein cfrA was included in other iron uptake cluster, among its partners are included the protein exB1 associated to iron uptake from the host and the protein atpD, involved in the synthesis of ATP, which was detected in our study, although not included as virulence factor.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.