Avian resistance to Campylobacter jejuni colonization is associated with an intestinal immunogene expression signature identified by mRNA sequencing

Abstract

Campylobacter jejuni is the most common cause of human bacterial gastroenteritis and is associated with several post-infectious manifestations, including onset of the autoimmune neuropathy Guillain-Barré syndrome, causing significant morbidity and mortality. Poorly-cooked chicken meat is the most frequent source of infection as C. jejuni colonizes the avian intestine in a commensal relationship. However, not all chickens are equally colonized and resistance seems to be genetically determined. We hypothesize that differences in immune response may contribute to variation in colonization levels between susceptible and resistant birds. Using high-throughput sequencing in an avian infection model, we investigate gene expression associated with resistance or susceptibility to colonization of the gastrointestinal tract with C. jejuni and find that gut related immune mechanisms are critical for regulating colonization. Amongst a single population of 300 4-week old chickens, there was clear segregation in levels of C. jejuni colonization 48 hours post-exposure. RNAseq analysis of caecal tissue from 14 C. jejuni-susceptible and 14 C. jejuni-resistant birds generated over 363 million short mRNA sequences which were investigated to identify 219 differentially expressed genes. Significantly higher expression of genes involved in the innate immune response, cytokine signaling, B cell and T cell activation and immunoglobulin production, as well as the renin-angiotensin system was observed in resistant birds, suggesting an early active immune response to C. jejuni. Lower expression of these genes in colonized birds suggests suppression or inhibition of a clearing immune response thus facilitating commensal colonization and generating vectors for zoonotic transmission. This study describes biological processes regulating C. jejuni colonization of the avian intestine and gives insight into the differential immune mechanisms incited in response to commensal bacteria in general within vertebrate populations. The results reported here illustrate how an exaggerated immune response may be elicited in a subset of the population, which alters host-microbe interactions and inhibits the commensal state, therefore having wider relevance with regard to inflammatory and autoimmune disease.

Figure 1. Infection model optimization and distribution of colonization levels.

(A) To determine the optimum dose where differential colonization is observed, three groups of fifteen birds were initially inoculated with low (i, 3.5 ×10), medium (ii, 3.5×10) and high (iii, 3.5×10) doses of C. jejuni, and the colonization status of their caeca estimated after 48 hours. No colonization was detected in any bird after infection with 3.5×10 CFU. All birds were colonized after high dosage. Maximum differentiation is achieved after inoculation with the medium dose of 3.5×10 CFU C. jejuni. (B) 255 birds were challenged with 3.5×10 C. jejuni and their caecal colonization status determined after 48 hours. No C. jejuni colonization could be detected in 38 birds. Colonization levels of the remaining birds varied from 2×10– 4 ×10 CFU/g with the majority of this group having very high caecal C. jejuni levels (10CFU/g C. jejuni).

Figure 2. RNAseq read mapping percentage breakdown.

175.31 million (48.2%) of the reads generated from mRNA sequencing could be mapped uniquely to known gene models and therefore could be used in estimation of gene expression levels. 101.28 million reads (27.8%) were mapped successfully to the genome but could not be mapped to NCBI gene models (unknown transcripts, 3.65 Gbp) indicating a necessity for more comprehensive annotation of the chicken transcriptome. Relatively few repeat reads (20.49 million reads, 5.6%) were observed consistent with the low repeat density of the chicken genome .

Figure 3. Detection of differentially expressed genes between high-colonized and nil-colonized birds.

Log2 fold change is plotted versus mean count numbers reflecting expression level. The R package DESeq was used to compare expression levels between the two groups of high-colonized and nil-colonized chickens and to identify genes displaying significant differential expression using a negative binomial model of the count data. 221 genes exhibiting significant differential expression with p-value 0.01 are highlighted red.

Figure 4. Differential expression as a function of read count.

The proportion of genes differentially expressed is plotted against the total number of reads for each gene. This plot investigates whether differential expression is more likely to be detected in genes with a higher number of read counts. The green line is the probability weighting function fitted by Goseq. Flat  =  no bias present and no correction. This illustrates that the method to detect differential expression was robust against bias introduced by differing count numbers.