Phenotypic and genetic variation in the response of chickens to Eimeria tenella induced coccidiosis

Abstract

Background: Coccidiosis is a major contributor to losses in poultry production. With emerging constraints on the use of in-feed prophylactic anticoccidial drugs and the relatively high costs of effective vaccines, there are commercial incentives to breed chickens with greater resistance to this important production disease. To identify phenotypic biomarkers that are associated with the production impacts of coccidiosis, and to assess their covariance and heritability, 942 Cobb500 commercial broilers were subjected to a defined challenge with Eimeria tenella (Houghton). Three traits were measured: weight gain (WG) during the period of infection, caecal lesion score (CLS) post mortem, and the level of a serum biomarker of intestinal inflammation, i.e. circulating interleukin 10 (IL-10), measured at the height of the infection.

Results: Phenotypic analysis of the challenged chicken cohort revealed a significant positive correlation between CLS and IL-10, with significant negative correlations of both these traits with WG. Eigenanalysis of phenotypic covariances between measured traits revealed three distinct eigenvectors. Trait weightings of the first eigenvector, (EV1, eigenvalue = 59%), were biologically interpreted as representing a response of birds that were susceptible to infection, with low WG, high CLS and high IL-10. Similarly, the second eigenvector represented infection resilience/resistance (EV2, 22%; high WG, low CLS and high IL-10), and the third eigenvector tolerance (EV3, 19%; high WG, high CLS and low IL-10), respectively. Genome-wide association studies (GWAS) identified two SNPs that were associated with WG at the suggestive level.

Conclusions: Eigenanalysis separated the phenotypic impact of a defined challenge with E. tenella on WG, caecal inflammation/pathology, and production of IL-10 into three major eigenvectors, indicating that the susceptibility-resistance axis is not a single continuous quantitative trait. The SNPs identified by the GWAS for body weight were located in close proximity to two genes that are involved in innate immunity (FAM96B and RRAD).

Fig. 1

Definitions of immune categories used

Fig. 2

Experimental design

Fig. 3

Mean weight (g) of control (n = 200, dashed-line) and infected (n = 942, solid-line) birds throughout the trial period, with standard error bars. See also (Additional file 2: Table S2) for full details

Fig. 4

Box plot of body weight for a control and infected birds and b the two sexes at d6.5pi

Fig. 5

Distribution of a caecal lesion score and b IL-10 for male (M) and female (F) infected birds

Fig. 6

Visualisation of the relationship between caecal lesion score and IL-10 (pg/ml) in infected birds. *P < 0.05; **P < 0.001

Fig. 7

Eigenvector decomposition of the phenotypic covariance matrix. Vector loadings for the three measured traits [weight gain (WG), caecal lesion score (CLS) and serum interleukin-10 (IL-10)] are presented in standard deviation units on the y-axis. The percentage of variance explained by each eigen vector (EV) is in brackets on the x-axis

Fig. 8

Manhattan and corresponding QQ plots from the GWAS for: a weight gain; b caecal lesion score; c serum IL-10. The log₁₀ P value is plotted for each SNP on the relevant chromosome (x-axis). Bonferroni corrected thresholds were set as P ≤ 1.1×10⁻⁶ and P ≤ 2.1×10⁻⁵ for genome-wide (P ≤ 0.05) and suggestive (i.e. one false discovery per genome-scan) levels, corresponding to −log₁₀ P values of 5.97 (blue) and 4.67 (red) lines