A Systems Biology Approach using Dynamic Bayesian Networks Reveals Bio-Signatures of Broiler Chicken Bursa Response to Campylobacter jejuni Inoculation

The bacterium Campylobacter jejuni is one of the most commonly reported causes of foodborne illness worldwide. While C. jejuni can cause disease in humans, it acts as a commensal in chickens. Two genetically distinct broiler lines have been previously identified as either resistant or susceptible to C. jejuni colonization in cecal content. We used these two genetic lines in a time-course infection study to investigate the differences in genetic responses to C. jejuni inoculation within and between lines. A total of 128 1-day-old broilers, 64 from each line, were either orally challenged with C. jejuni or inoculated with PBS. Bursas were harvested at 1 h, 4 h, 24 h, and 36 h post-inoculation and RNA isolated. A chicken 44K Agilent gene expression microarray was performed on all 128 samples. Raw microarray data was then analyzed using an integrated systems biology pipeline for multi-conditional analyses, termed BioSignatureDS™. The backbone of this platform is based on Dynamic Bayesian Networks, an advanced form of machine learning and pattern recognition. Using this approach, we integrated gene expression data with gene ontology groups, regulatory gene networks and individual genes to capture the perturbation of the C. jejuni-chicken interactome over time. When compared to non-infected birds, we found that gene networks involving carbohydrate and lipid metabolism were the most significantly perturbed in resistant line A birds, especially in processes involving propanoate and glycan biosynthesis. In susceptible line B birds, networks involving signal transduction and amino acid biosynthesis were found to be most significant, especially for processes utilizing notch signaling and aromatic amino acid production. By integrating the data from the two lines, highly perturbed regulatory signatures between epithelial and immune cell-to-cell communications were observed. These models provide new insights into the early immune response of chicken to C.jejuni and allow in silico hypothesis testing for future functional studies.