Induction of a chemoattractant transcriptional response by a Campylobacter jejuni boiled cell extract in colonocytes

Background Campylobacter jejuni, the commonest cause of bacterial diarrhoea worldwide, can also induce colonic inflammation. To understand how a previously identified heat stable component contributes to pro-inflammatory responses we used microarray and real-time quantitative PCR to investigate the transcriptional response to a boiled cell extract of Campylobacter jejuni NCTC 11168. Results RNA was extracted from the human colonocyte line HCA-7 (clone 29) after incubation for 6 hours with Campylobacter jejuni boiled cell extract and was used to probe the Affymetrix Human Genome U133A array. Genes differentially affected by Campylobacter jejuni boiled cell extract were identified using the Significance Score algorithm of the Bioconductor software suite and further analyzed using the Ingenuity Pathway Analysis program. The chemokines CCL20, CXCL3, CXCL2, Interleukin 8, CXCL1 and CXCL6 comprised 6 of the 10 most highly up-regulated genes, all with Significance Scores ≥ 10. Members of the Tumor Necrosis Factor α/Nuclear Factor-κB super-family were also significantly up-regulated and involved in the most significantly regulated signalling pathways (Death receptor, Interleukin 6, Interleukin 10, Toll like receptor, Peroxisome Proliferator Activated Receptor-γ and apoptosis). Ingenuity Pathway Analysis also identified the most affected functional gene networks such as cell movement, gene expression and cell death. In contrast, down-regulated genes were predominantly concerned with structural and metabolic functions. Conclusion A boiled cell extract of Campylobacter jejuni has components that can directly switch the phenotype of colonic epithelial cells from one of resting metabolism to a pro-inflammatory one, particularly characterized by increased expression of genes for leukocyte chemoattractant molecules.


Background
Campylobacter jejuni (C. jejuni) is a gram-negative microaerophilic bacterium responsible for the majority of human bacterial enteric infections worldwide [1,2]. C. jejuni is commonly found as a commensal organism in the intestinal tracts of a wide range of wild and domestic animals, including commercial poultry [3]. Cross-contamination from raw poultry or insufficient cooking of poultry meat are common sources of infection. Enteric infections by this pathogen are often associated with a potent localized inflammatory response. Symptoms arising from infection include watery or bloody diarrhoea with abdominal cramping and fever. In addition, C. jejuni can be invasive and is associated with septicaemia, meningitis, Guillain-Barré syndrome [4] and more recently with immuno-proliferative disease [5].
C. jejuni virulence factors for human disease include flagella based chemotaxis, adhesin-based cellular adherence, host cell invasion and the elaboration of a heat labile cytolethal distending toxin (CLDT) [2,6,7] In previous studies we have additionally shown that a heat stable C. jejuni boiled cell extract (BCE) is able to activate the transcription factor NF-B (nuclear factor kappa-light-chainenhancer of activated B cells) [8]. This signalling molecule is responsible for inducing the expression of a number of genes involved in inflammation and cell mediated immunity [9], including chemokines capable of attracting leukocytes, resulting in inflammation. NF-B is held inactive in the cytoplasm of a cell, whilst its nuclear localization domain is masked by inhibitory IB proteins. If IB is phosphorylated, leading to ubiquitin-mediated proteolysis, then NF-B is released to transport to the nucleus of the cell, where it affects transcription of B-responsive promoters. Therefore products that activate NF-B can be presumed to have a strong role in triggering inflammation. Previous work has shown that live C. jejuni and a BCE can induce both NF-B, and the synthesis and release of the chemokine interleukin-8 [8].
In order to identify a wider range of genes affected by C. jejuni products and assess the relative importance of the NF-B response we used microarray technologies to identify genes that were both up and down-regulated in HCA-7 cells after exposure to a C. jejuni BCE [8,10]. Use of the Ingenuity Pathway Analysis (IPA) program suite enabled us to group co-regulated genes in order to identify the cellular signalling pathways activated in HCA-7 cells in response to C. jejuni BCE. The transcriptomic data were confirmed by real time quantitative PCR (RQ-PCR).

C. jejuni culture and preparation of BCE
The type strain C. jejuni National Collection of Type Cultures (NCTC) 11168 was used throughout these experi-ments, since it was originally isolated from a patient with diarrhoea, its genome sequence is available and it has a well-characterized pathological phenotype [11]. It was incubated on blood-agar plates (Blood Agar Base CM0271 from Oxoid, Basingstoke, UK with 5%, v/v defibrinated horse blood) under micro-aerobic conditions for 24 h. and used to inoculate Nutrient Broth no. 2 (Oxoid CM0067, 600 ml in 1000 ml flask). Inoculated flasks were shaken at 140 rpm at 42°C for 16 h. under micro-aerobic conditions. Culture purity was determined by plating samples from each overnight culture onto blood plates and incubating for 24 h., 42°C in micro-aerobic conditions. Bacteria were collected by centrifugation at 10,000 g for 15 min. The cell pellet was washed three times in Phosphate Buffered Saline (PBS), weighed and re-suspended in PBS to achieve a 10% (w/v) suspension, which was boiled for 10 min., cooled on ice for 5 min. before being centrifuged at 10, 000 g for 10 min. The supernatant was collected, passed through a 0.2 m filter to remove residual bacteria and stored at -20°C until required.

HCA-7 cell culture and treatment with C. jejuni BCE
The human colonocyte line HCA-7 [10], clone 29, was grown to confluence in a 5% CO 2 atmosphere in monolayer cultures on monolayer dishes in Dulbecco's Modified Eagle's Medium supplemented (DMEM) with 100 g/ml penicillin, 100 g/ml streptomycin and fetal calf serum at 10% (v/v, Fisher Scientific, Loughborough, UK) at 37°C. Twenty-four hours prior to induction by BCE, HCA-7 cells were transferred to serum-free DMEM. HCA-7 cells were then incubated for 6 h. with 25 l BCE or PBS control in a total volume of 1 ml of DMEM. The BCE preparation was determined in parallel to induce NF-B 300-fold using a reporter cell assay [8]. At 6 h. post induction total RNAs were extracted using RNAeasy columns (Qiagen, West Sussex, UK). Total RNA yields and purity were determined using an Agilent 2100 Bioanalyzer (Agilent Technologies UK Limited, Stockport, UK).

cDNA synthesis
Approximately 10 g of total RNA was reverse transcribed at 42°C for 1 h. to generate first strand DNA using 100 pmol oligo dT (24) primer containing a 5'-T7 RNA polymerase promoter sequence (5'-GCCAGTGAATTGTAATAC-GACTCACTATAGGGAGGCGG-(dT) 24 4 , 0.15 mM -NAD + and 10 mM dNTPs. 10 units of T4 DNA polymerase were added and the reaction allowed to proceed for a fur-ther 5 min. before termination with 0.5 M EDTA. Double stranded cDNA products were purified using the Gene-Chip Sample Cleanup Module (Affymetrix, Santa Clara, CA, USA).

cRNA synthesis
The synthetic cDNAs were in vitro transcribed using T7 RNA polymerase (ENZO BioArray High Yield RNA Transcript Labeling Kit, Affymetrix, Santa Clara, CA, USA) with biotinylated ribonucleotides to generated biotinylated complementary RNAs (cRNAs). The cRNAs were purified using the GeneChip Sample Cleanup Module before random fragmentation at 94°C for 35 min. in a buffer containing 40 mM Tris-acetate (pH 8.1), 100 mM potassium acetate and 30 mM magnesium acetate to generate molecules of approximately 35 to 200 bases long.

Array hybridization
Changes in gene transcription were analyzed by hybridization to Affymetrix Human Genome U133A array (HG-U133A) which contains probes for over 22,000 transcripts, including representation of the RefSeq database sequences and probe sets http://www.affymetrix.com/ products_services/arrays/specific/hgu133.affx. The fragmented cRNAs were mixed with 0.1 mg/ml of sonicated herring sperm DNA in a hybridization buffer containing 100 mM 2-N-morpholino-ethane-sulfonic acid (MES), 1 M NaCl, 20 mM EDTA and 10% Tween 20 to make the hybridization mixture. The hybridization mixture containing the fragmented cRNA was denatured at 99°C for 5 min. and equilibrated for a further 5 min. at 45°C before centrifugation at 10,000 g for 5 min. to remove any insoluble material from the hybridization mixture. The hybridization mix was transferred to the ATH1-121501 genome array (Affymetrix, Santa Clara, CA, USA) cartridge and hybridized at 45°C for 16 h. on a rotisserie at 60 rpm.
After a 16 h. hybridization period the arrays were washed and stained in a Fluidics station (Affymetrix, Santa Clara, USA). The arrays were initially washed in a low stringency buffer A (6 × SSPE [0.9 M NaCl, 0.06 M NaH 2 PO 4 , 0.006 M EDTA], 10% Tween 20) at 25°C for 10 min. and then incubated with a high stringency buffer B (100 mM MES, 0.1 M NaCl, 10% Tween 20) at 50°C for 20 min. and stained with 10 mg/ml of streptavidin phycoerythrin (SAPE), in stain buffer containing 100 mM MES, 1 M NaCl, 0.05% Tween 20 and 2 mg/ml BSA at 25°C for 10 min. After a further wash in wash buffer A at 25°C for 20 min. they were stained with biotinylated anti-streptavidin antibody at 25°C for 10 min. After antibody staining the arrays were stained again with SAPE for signal amplification and washed with buffer A at 30°C for 30 min. The arrays were finally scanned and the intensities averaged with the Agilent GeneArray Scanner (Agilent Technology UK, West Lothian, UK).

Statistical analysis of Array data and Generation of Networks and Canonical Pathways
In order to identify genes of interest we used the S Score (Significance Score) algorithm as implemented in the Bioconductor software suite http://www.bioconduc tor.org [12] based on the R package http://www.rproject.org [13] that takes advantage of the fact that most genes are unchanged and calculates an S score (SD from the mean). The S score threshold of +/-2.5 and an alpha value of P = 0.005 was used to define gene changes of interest. Data listing all genes that satisfied these criteria were analyzed by Ingenuity Pathway Analysis, Ingenuity ® Systems, http://www.ingenuity.com. This generated functional networks and canonical pathways that connect the differentially expressed genes, using the IPA Knowledge base, where the interactions are supported by peer reviewed publications and which contains over 1.4 million interactions between genes, proteins, and drugs. Scores were assigned allowing ranking of the networks, using a Fisher's right tailed exact test.

Analysis of microarray data by real time quantitative PCR
To confirm microarray results, extracted HCA-7 total RNA was amplified by oligo dT (15) primers according to the Im-Prom II Kit (Promega UK, Southampton UK) methodology. Representative samples of genes from a number of the major functional groups and gene networks identified by IPA program were selected to confirm the array data using RQ-PCR analysis (Tables 1, 2 and 4) under appropriate conditions for an ABI Prism 7700. Primer and probe design utilized Primer Express software (Applied Biosystems, Warrington, UK). The primers were validated for gene specificity by agarose gel electrophoresis. Reporter dye-labelled probes were used with FAM (6-carboxyfluorescein) at the 5'-end and TAMRA (6-carboxytetramethyl-rhodamine) at the 3'-end. Reactions were set up in a final volume of 25 l containing 12.5 l of 2 × Taqman Universal PCR Mastermix (Applied Biosystems, Warrington, UK): 0.75 l of each primer (10 pmol/l), 0.5 l of probe (10 pmol/l), 2 l of cDNA (equivalent to 5 ng total RNA/l) and 8.5 l of water. Samples were analyzed in triplicate and the emission released reporter dye was monitored by an ABI Prism 7700 Sequence Detector (Applied Biosystems, Warrington, UK) using the default PCR program of 2 min at 50°C and 10 min at 95°C; each cycle included denaturing at 95°C for 15 s and annealing at 60°C for 1 min. Analysis of the data was via the Sequence Detection System (SDS) software (Applied Biosystems, Warrington, UK). A no template control was included in each analysis and did not give any signal with any of the primer/probe combinations. RQ-PCR data were normalized using primers to -actin based on the considerations outlined by Hugget et al. [14].

Chemokine and cytokine analyses
Cultured cells were prepared and induced as described above. After 6 h. incubation, the media was removed and stored at -20°C until examined using a Coulter-Alter Flow Cytometer in conjunction with a BD cytometric bead array human inflammation kit according to manufacturer's instructions (BD Biosciences, Oxford, UK). IL8 and CCL20 (MIP-3) were specifically measured using a sandwich ELISA, by capture with a murine anti-human IL8 or CCL20 and detected using biotinylated goat anti-human IL8 using streptavidin-coupled horseradish-peroxidase, according to the manufacturer's instructions (R&D Systems, Minneapolis, MN, USA).

Results
The Bioconductor and IPA programs identified 356 genes that changed with a positive or negative S score of 2.5 or greater (maximum 13.54). Three hundred were up-regulated and 56 were down-regulated (Additional file 1). Table 2 shows 48 genes that were up-regulated with an S score of 5 or greater. These were grouped by class and ordered by the highest S score in each class. Chemokines dominate the most highly up-regulated genes with six of the ten highest S scores. Members of the TNF-NF-B super family were also highly up-regulated (Table 2). Other highly up-regulated genes were those involved in apoptosis and ubiquitination, extra-cellular matrix proteins, the folate receptor, superoxide dismutase, thioredoxin reductase, Intercellular Adhesion Molecule (ICAM) 1 and cytokines or their receptors (Colony Stimulating Factor [CSF] 2 and interferon- receptor 1).

Down-regulated genes
Fewer genes were down-regulated than those that were up-regulated and negative S scores were less pronounced than those for the up-regulated genes. For comparative purposes Table 3 shows down-regulated genes that were selected on the basis of a more permissive S score of -2.6 or less to yield a similar number (46). These genes were grouped by class and ordered by the highest negatively regulated (lowest value) S score in each class. The pattern of down-regulated gene classes differ markedly to those that were up-regulated. Most prominent were genes concerned with the maintenance of normal cell cycle, DNA replication and cell structure. The down-regulated group feature specific genes encoding components involved in membrane transport, mitosis, nucleotide synthesis, transcription, protein synthesis and export, membrane transport and energy metabolism.

Signalling pathways
IPA identified a number of canonical signalling pathways that were most significantly affected ( Figure 1). Figure 2 shows a simplified composite of all genes identified by  IPA as being part of specific signalling pathways that are most significantly regulated, together with their individual S scores. Here the central mediator is the NF-B signalling pathway that is clearly contributory in affecting the signalling through the Death Receptor, IL6, IL10, Toll-like receptor and PPAR pathways (also see Gene Networks section below and Figure 3 which also features NF-B). In addition, several other canonical signalling pathways, some of which do not feature NF-B, were also identified as significantly affected. The

Predicted functional effects
The IPA program can determine if groups of significantly changed genes have related cellular and molecular functions ( Figure 4). Here IPA identified 16 functional categories that were significantly affected by the C. jejuni BCE. The most prominent functions implicated were cellular movement (reflecting changes in chemokines, adhesion receptors and molecules affecting cytokinesis), cell growth and proliferation and cell death.

Gene networks
The IPA program constructed 16 interconnected gene networks that were significantly altered as a result of treatment of HCA-7 cells with C. jejuni BCE, all with network scores of  8. The network score is the probability that a network would be assembled by chance where a level of > 3 is statistically significant, at p < 0.001. In the four most significantly regulated all 35 focus genes of the network were affected, all giving an identical score of 52 (P < 1E-52).
The first network ( Figure 3) contains genes concerned with cellular movement, particularly chemotaxis. NF-B occupies a central position in the network and includes a

Validation of Microarray data
Changes in gene expression identified by microarray were confirmed by RQ-PCR (Table 4). However, hierarchical differences are apparent between the RQ-PCR values normalized against -actin compared with the S score associated with the significantly regulated genes as indicated by differential hybridization of the cRNA preparations to the microarray.

Chemokine and cytokine responses
To further validate the gene transcriptional changes using microarray and RQ-PCR methods, we measured the levels of secretory immunomodulatory proteins in parallel cell supernatants of HCA-7 cells pre-and post-induction with C. jejuni BCE. Table 5 presents the chemokine and Canonical Signalling Pathways identified by IPA software as significantly regulated by C. jejuni BCE Figure 1 Canonical Signalling Pathways identified by IPA software as significantly regulated by C. jejuni BCE. A Fisher's exact test was used to calculate a p-value (Bars) determining the probability that the association between the genes in the dataset and the canonical pathway can be explained by chance alone. Threshold refers to the cut off for p < 0.05.
cytokine levels of pro-and anti-inflammatory secretory proteins. Consistent with the microarray observations the pro-inflammatory chemokine CCL20 showed a 12.6-fold increase in levels 6 h. post treatment. IL8 levels were also found to increase, but far more dramatically than CCL20 with a 460-fold induction. HCA-7 colonocytes are particularly IL8 responsive with post-induction levels of 18.4 ng/ml, an observation that is consistent with previous reports with this cell line [8]. The pro-inflammatory cytokine IL1 showed a weak response consistent with the transcriptional response recorded in the microarray study. Pro-inflammatory cytokine IL6 showed a 5-fold increase, whereas the anti-inflammatory cytokine IL10 remained static. The transcriptional response of the genes encoding IL6 and IL10 did not show marked transcriptional changes but the pathways associated with these immunomodulatory proteins were recognized by IPA and are responsive to NF-B.

Discussion
Understanding the pathogenesis of C. jejuni enteric disease is important both because C. jejuni is a major cause of diarrhoeal illness worldwide and because it may serve as a model for ulcerative colitis, the pathology of which it closely resembles [15]. Previous work has shown that direct interaction between C. jejuni and epithelial cells is capable of inducing pro-inflammatory and pro-secretory processes [8,16]. These are associated with cellular invasion [17] and secretion of IL8 by CLDT dependent and independent mechanisms [16,18]. Direct use of a BCE has Regulated molecules in conical signalling pathways identified by IPA allowed us to use a reductionist approach to investigate effects of C. jejuni that are not dominated by these linked processes of cellular invasion by live bacteria and by toxin based cell lysis. BCE has been determined to contain polysaccharide and protein components of the cell. As demonstrated previously the NF-B inducing activity of C. jejuni BCE is relatively insensitive to digestion by protease K [8]. However the protein content has been determined using tryptic digests of SDS-polyacryamide extracted protein bands using MALDI-TOF mass spectrometry as flagellin (Cj1339c), trigger factor (Cj0193c), lipoprotein (Cj0983), major outer membrane protein (Cj0599), cytochrome-c peroxidase (Cj0358), bacterioferritin (Cj1534c), cell binding factor PEB4A (Cj0496), hypothetical protein (Cj0706), periplasmic protein (Cj0772c), fibronectin binding protein (Cj1478c), non-heme iron protein (Cj0012c), periplasmic protein (Cj1380), periplasmic protein (Cj0420), periplasmic protein (Cj0998c), DNA-binding protein HU (Cj0913c), periplasmic cytochrome C (Cj1153) and thioredoxin (Cj0147c) [11]. The polysaccharide component features -glucan oligomers. The C. jejuni extract is notably devoid of the dominating heat-labile effects of the CLDT. C. jejuni BCE, like infection with live C. jejuni, has been shown to be a potent inducer of NF-B using either luciferase based reporter assays, western blots with antibodies against IB or electrophoretic mobility shift assays in epithelial cells [8] but, unlike treatment with live C. jejuni, this does not lead to host cell lysis. These observations are consistent with the hypothesis that a heat stable component plays a significant role in the pro-inflammatory response upon exposure to C. jejuni.
IPA generated NF-B-centred gene network Figure 3 IPA generated NF-B-centred gene network. Network contains nodes (gene/gene product) and edges (indicating a relationship between the nodes) showing the cellular/subcellular location as indicated. An asterisk indicates that duplicates were identified in each dataset. Function classes of nodes indicated by shape to represent functional class, a plus sign indicates node is contained in other networks. All 35 focused genes are significantly up-regulated. Genes with an S score of  7 are shown in red and those with an S score of between 2.5-7 are shown pink. Explanation of edge types and shapes is indicated.
We hypothesize that NF-B modulation is central to the response of enterocytes to C. jejuni BCE; to study this we determined the global changes in gene expression induced by C. jejuni BCE treatment of the well-differentiated human colonocyte line HCA-7, clone 29. In order to ensure the relevance of our results we have adopted strin-gent criteria for the identification of significantly affected genes and used the IPA program to determine the functional links between these gene products, identify the signalling pathways and networks to which they belong. These changes were validated by showing similar affects Functional Molecular and Cellular pathways significantly affected by C. jejuni BCE Figure 4 Functional Molecular and Cellular pathways significantly affected by C. jejuni BCE. A Fisher's exact test was used to calculate a p-value (Bars) determining the probability that the association between the genes in the dataset and the canonical pathway can be explained by chance alone. Threshold refers to the cut off for p < 0.05.  (CXYA2) as well as IL1 and CSF2, all of which were also significantly up-regulated in our study [19]. The NFKB1, NFKB2 and RELB components of NF-B are also similarly up-regulated in our study. Other changes that are likely to be of functional importance and are the up-regulation of COX2 (PTGS2), TNIP2, MYC, SOD2, ELF3 and ICAM1 (Additional file 1), where all of these processes are also downstream targets of NF-B [20] and mediators of feedback inhibition of NF-B activation such as NFKBIA (IB) [9], TNIP1 [21] and TNIP2 ( Figure 3) [22]. A central role for NF-B is also supported by data using the monocytic cell line THP-1 [23]. Studies in which Caco-2 cells were incubated with live bacteria resulted in expression of many genes similar to those reported here, including chemokines, but additionally, the NF-B inhibitor NFK-BIZ [24]. This difference may reflect the ability of live bacteria to invade cells and/or elaborate a CLDT with DNase activity [6].
The pattern of significantly down-regulated genes ( Table  3) is remarkably different with a reduction in expression in constitutively expressed genes concerned with nucleotide synthesis, transcription, DNA replication, mitosis, structural protein synthesis, membrane transport and energy metabolism. These changes likely reflect the reprioritization of cellular metabolism in response to proinflammatory products.
Whether the changes caused by the C. jejuni BCE would lead to increased or reduced apoptosis is difficult to predict, especially as HCA-7 lack a functional TP53 protein, although these cells are capable of apoptosis given the appropriate signal [25]. Invasive C. jejuni infection can cause cell death in HCA-7 cells [16], although we did not see this with the addition of BCE [8]. Increased expression of members of the death receptor pathway, the TNF superfamily and their receptors, but also of TNF agonists may imply regulated activation of pro-apoptotic activity [26][27][28][29][30]. Up-regulation of TRAIL, DR5, and FAS ligand acting via FADD, the universal adaptor protein known domain-containing members of the TNF receptor superfamily, would successively activate caspases 8, 10 and 3 as well as possible G1-S cell cycle progression [27]. However, the antagonists TNFAIP3, FLIP and cIAP, which respectively inhibit apoptosis via TRAF6, caspases 8, 9, 10 and TRAF-2 directly or indirectly are also prominent amongst the up-regulated genes [29][30][31][32].
Moreover, several other key proteins for the cell cycle and apoptosis are affected. Thus CDKN1A (P21, WAF, WAF1 or CIP1) which plays a pivotal role in inhibiting cell cycle progression at several points in response to DNA damage [33], is up-regulated, as are FOXO1A and SMAD 2 (Additional file 1) and 3 (Additional file 2), which act together to increase CDKN1A activity [34,35]. Conversely, other genes that inhibit cell cycle progression are down-regulated. These include SKP2, the F-box receptor that interacts with p19 and the CDK2/cyclin A to prevent entry into G1 [36] and SFN (stratifin or 14-3-3) a key target of the tumour suppressor gene TP53 which acts to cause G2 arrest [37].
Five other changes of potential functional importance are of note. Firstly, a number of potentially antibacterial agents are highly induced, including LCN2 (lipocalin-2) [38,39] and PI3 (peptidase inhibitor 3, aka ELAFIN) [40], whilst MMP7 is thought to activate defensins [41]. Secondly, five key molecules involved in antigen processing and presentation (Figure 1, 2) [42] were also up-regulated and could play a role in the development of immune responses to C. jejuni. Thirdly, alterations in matrix metalloproteinases and leukocyte receptors would influence the inflammatory response, with MMP9 acting to facilitate neutrophil transfer by activating interleukin-8 [43] and MMP7 acting to localize them to sites of tissue damage [44]. Fourthly, the ephrin pathway (Figure 2), including Ephrin A2 and B2 receptors (EPHA2, EPHB2) and Ephrin A1 (EFNA1, Figure 3), rho kinase (ROCK2), Rac, ARP2/3, CDC42 and WASP appeared to be strongly up-regulated. This pathway is concerned with activation of cytokinetic changes that may potentially play a role in rapid restitution [45,46]. Finally, up-regulation of the folate receptor (FOLR1) may reflect preparation for reparative nucleotide synthesis dependent upon one-carbon transfer activity [47].

Conclusion
The data we have generated using a BCE of C. jejuni represents a reductionist approach to determine some of the cellular responses associated with C. jejuni infection.
However, because C. jejuni BCE represents a robust NF-B inducing activity that is not only heat-stable but resistant to protease and acidic pH (pH 3) [8], these may indeed be of clinical significance if these products are shed upon C. jejuni infection or co-delivered through the diet. C. jejuni has been detected in many commercially available chicken portions [2] and clinical cases of Campylobacter enterocolitis are frequently associated with ingestion of partially cooked poultry meat [48].
Changes in host gene expression following C. jejuni BCE interestingly reflects some of the changes that are known to occur in inflammatory bowel diseases (IBD) such as ulcerative colitis, for which C. jejuni colitis can be considered a model, and may therefore indicate other potential targets for investigation of epithelial-derived mediators of inflammation in ulcerative colitis/IBD. Up-regulation of NF-B is well recognized and considered a possible target of mesalazine [49,50]. Genes up-regulated by C. jejuni that have been associated with active ulcerative colitis/ IBD include chemokines [51], such as IL8 and CCL20 (macrophage inflammatory protein 3) [52][53][54] cytokines, including TNF [55], eicosanoids [53] and elafin [56]. IL23, IL32 [57][58][59] and receptors such as interferon- receptor, and TLR2 [60] have all been demonstrated to be altered here ( Table 2, Additional file 1). Activation of pro-apoptotic pathways involving the TNF superfamily and death domain signalling pathway have been reported to be up-regulated in colonic enterocytes isolated from patients with ulcerative colitis, from which C-IAP2 (BIRC3) has been proposed as a disease marker [61], whilst the leukocytes serine anti-proteinase elafin has recently been identified as a candidate biomarker for ulcerative colitis but with attenuated induction in Crohn's disease [56]. Thus, the data we report here include a number of pathways and mediators that may be realistic anti-inflammatory therapeutic targets to prevent or reduce the activity of C. jejuni colitis or ulcerative colitis. These targets include mechanisms for chemoattraction of inflammatory cells, cellular processes associated with repair and the processes associated with apoptosis, as well as NF-B itself, the utilization of which can be investigated by intervention studies in model systems and humans.