Differential responses of cecal microbiota to fishmeal, Eimeria and Clostridium perfringens in a necrotic enteritis challenge model in chickens

Abstract

Clostridium perfringens causes enteric diseases in animals and humans. In poultry, avian-specific C. perfringens strains cause necrotic enteritis, an economically significant poultry disease that costs the global industry over $2 billion annually in losses and control measures. With removal of antibiotic growth promoters in some countries this disease appears to be on the rise. In experimental conditions used to study disease pathogenesis and potential control measures, reproduction of the disease relies on the use of predisposing factors such as Eimeria infection and the use of high protein diets, indicating complex mechanisms involved in the onset of necrotic enteritis. The mechanisms by which the predisposing factors contribute to disease progression are not well understood but it has been suggested that they may cause perturbations in the microbiota within the gastrointestinal tract. We inspected changes in cecal microbiota and short chain fatty acids (SCFA) induced by Eimeria and fishmeal, in birds challenged or not challenged with C. perfringens. C. perfringens challenge in the absence of predisposing factors did not cause significant changes in either the alpha or beta diversity of the microbiota nor in concentrations of SCFA. Moreover, there was no C. perfringens detected in the cecal microbiota 2 days post-challenge without the presence of predisposing factors. In contrast, both fishmeal and Eimeria caused significant changes in microbiota, seen in both alpha and beta diversity and also enabled C. perfringens to establish itself post challenge. Eimeria had its strongest influence on intestinal microbiota and SCFA when combined with fishmeal. Out of 6 SCFAs measured, including butyric acid, none were significantly influenced by C. perfringens, but their levels were strongly modified following the use of both predisposing factors. There was little overlap in the changes caused following Eimeria and fishmeal treatments, possibly indicating multiple routes for progressing towards clinical symptoms of necrotic enteritis.

Figure 1. Diagram of the experimental design used in the animal trial.

“C.p” was used to abbreviate C. perfringens.

Figure 2. Barcharts showing lesion scores (A) and mortality (B) in groups challenged with C. perfringens.

The groups not challenged with C.perfringens had no notable lesions or mortality. Lesions were scored as described in . The bars show the standard errors. Missing bars indicate no lesions or mortality.

Figure 3. UnweightedUniFracPCA plot.

Panel A is coloured by presence (red) and absence (blue) of C. perfringens in samples; in panel B samples are coloured according to presence (red) or absence (blue) of Eimeria and in panel C the same samples are coloured by presence (red) or absence (blue) of fishmeal diet.

Figure 4. PCA plot generated using the Ade4 R phylogenetic package .

Principal Component Analysis combined with duality diagram functions (dudi) was used to perform Between-Class Analysis (bca) with respect to sample assignment to treatment groups. Monte Carlo test based on 999 permutations demonstrates that groups are significantly (Monte Carlo P-value = 0.003) different. The graph shows sample and group relationship with respect to microbiota structure. The OTU table used for the analysis was 100 times rarefied (normalised) and OTUs with abundance lower than 0.001 were removed.

Figure 5. Duality diagram function and between class analyses presented in the form of a 3D PCA plot and selected significant OTUs in form of boxplots.

The 3D graph (A) shows the relationship between sample groups and OTUs driving the sample differences. OTUs significantly different between groups were selected in Ade4 using the relative abundance table and confidence level of 95%; only OTUs with total abundance higher than 0.001 were retained in the analysis. The large circles represent treatment groups while small black circles represent OTUs. Only OTUs that are significantly associated with differences between groups are enlarged and coloured in the same colour as the sample they are distinguishing from others. Some of the significant OTUs are given in views B, C and D while all OTUs selected by Ade4 analysis are given in Figure S3. The boxes represent the limits of the second and third quartiles; the whiskers indicate the data within 1.5 times the interquartile range and the dots are outliers.

Figure 6. Table of TukeyHSD corrected p-values given in form of a heatmap.

ANOVA was calculated for differences between groups for each of the SCFAs, pH and cultured bacterial counts (columns of the heatmap). ANOVA was followed by TukeyHSD honesty test to calculate p-values (shown in each cell of the heatmap) for each to each group-to-group comparison (rows). The heatmap was coloured by p-values according to the key below the graph in order to emphasise significant differences (in dark red). For example the lower right corner cell shows that groups E+FM+Cp and CTRL group (labelled E+FM+Cp – CTRL on the graph) are showing slight differences (TukeyHSD corrected P-value = 0.054184) in concentration of propionic acid. The clustering of heatmap rows and columns was done to identify similarities in SCFA changes between conditions and was based on Euclidean distance and complete agglomeration method in R package ggplot. This identified a hot-spot in the lower right corner of the heatmap of sample comparisons showing significant differences in isobutyric, isovaleric and propionic acid.

Figure 7. Spring embedded network showing Pearson correlations between SCFA and OTUs (analysed in Qiime).

Edge length in the network is proportional to Pearson correlation r value; blue edge is used for negative and red edge for positive correlation. SCFA are represented with big red circles and shortened as AA = acetic Acid, SA = Succinic Acid, IVA = isovaleric Acid, BA = butyric Acid, IVA = IsoValeric Acid and PA = propionic Acid. OTUs are represented with green circles.