H9N2 AIV infection of chickens causes mild clinical signs
In both of our experiments, all birds survived H9N2 AIV challenge. RIR and VALO H9N2 infected chickens showed mild lethargy and diarrhoea especially between day 2 and day 4 post-challenge. There were no significant differences in body weight between control and infected birds for either breed (Additional file 1: Fig. S1). There were significant differences in body weight between the RIR and VALO chickens. At day 0 the RIR control group were on average 54.31 g (± 5.5 g SEM) heavier than the VALO control group (p < 0.0001) whereas at day 14 the RIR control group was 180.3 g (± 38.2 g SEM) heavier than the VALO control group (p = 0.0421) (Additional file 1: Fig. S1).
VALO chickens shed H9N2 virus infection from the buccal cavity, a day longer than RIR chickens
Buccal viral shedding was determined by testing oropharyngeal (OP) swabs at day two, day three, day four, day five and day six post-challenge by plaque assay on MDCK cells (Fig. 1). In experiment 1 where only RIR chickens were infected, all chickens shed virus from the OP cavity on day two with the average titre shed being 6.2 × 104 pfu/ ml (± 12,730 pfu/ml SEM). Virus titre declined on day four and day five, with no virus being recovered from samples taken on day six post infection (Fig. 1a). In experiment 2, both RIR and VALO chickens were challenged with the same dose of the same H9N2 virus. At day two post challenge, all chickens in both lines shed virus from the OP cavity with no statistical difference in titre shed between the chicken lines being observed (average titres being 4.3 × 103 pfu/ ml (± 921 pfu/ ml SEM) for RIR and 2.7 × 103 pfu/ml (± 1086 pfu/ml SEM) for VALO). Similarly, to experiment 1 virus titres in the OP cavity declined on days four and five with no observed virus shedding on day six post challenge for either line. We did see differences in the rate of viral clearance between the RIR and VALO chicken lines from day four post challenge onwards. On day four the mean OP virus shed was 1.6 × 103 pfu/ml (± 1129 pfu/ml SEM) for the RIR compared to 3.6 × 103 pfu/ ml (± 1121 pfu/ml SEM) for the VALO (p value = 0.2465). On day five only 1 out of 8 RIR birds shed virus to titres above the limit of detection for the plaque assay whereas 7 out of 8 VALO birds did (average virus shed was 1.3 × 102 pfu/ml) (Fig. 1b). Cloacal viral shedding of virus by all the birds in both experiments was conducted by qRT-PCR for viral M gene, but only a single RIR bird on day four post challenge had a positive result (data not shown).
Antibody responses to the H9N2 virus prior to challenge and at 10 days (Experiment 1) and 14 days (Experiment 2) post-challenge were measured by Hemagglutination inhibition (HI) assay. All infected birds in both experiments seroconverted with an average HI titre above 8 log2 (Additional file 1: Fig. S2). No significant differences between RIR and VALO infected birds were found.
Batch processing of colon tissue from chickens did not affect the median number of operational taxonomic units (OTUs) obtained from 16S rRNA amplicon sequencing
Following euthanasia of chickens at various time-points colon samples were aseptically sampled at post-mortem examination, homogenised and the microbial DNA extracted. The bacterial DNA was processed to give 16S rRNA amplicons and subjected to Illumina Miseq sequencing. We performed a pre-processing and quality check on the raw reads obtained from the Illumina Miseq platform, which removed about 20% of the sequences. All sequences were then normalized to 62,000 sequences per sample resulting in a total number of 4,094,860 OTUs with a median of 119,082.000 OTUs per sample in Experiment 1 (Batch 1), 5,970,048 of OTUs with a median of 156,783 OTUs per sample in H9N2 infected groups (Batch 2) and 8,313,621 of OTUs with a median of 173,833.5 per sample in control groups in Experiment 2 (Batch 3) (Additional file 1: Fig. S3). This raw data was then used in our described pipelines (“Material and methods” section) to analyse changes in the biodiversity of the chicken colon microbiota following H9N2 infection and whether chicken breed impacted these changes.
Microbiota alpha diversity indices increased over time in healthy chicken colon
A temporal change in diversity measure, number of OTUs, phylogenetic diversity and Shannon diversity, were observed in control groups in both experiments (Figs. 2 and 3 and Additional file 1: Fig. S4). When RIR and VALO chicken colon microbiota composition was analysed weekly (Experiment 2), we identified that statistically significant changes in alpha diversity indices in control groups occur in the time frame of 14 days, especially between day 7 and day 21 of age, suggesting that a 2 weeks’ period was required to see significant maturity changes in healthy colon microbiota (Additional file 2: Table S1). Kruskal Wallis statistical testing showed a statistically significant temporal changes in number of OTUs (the number of bacterial species detected in a sample, species richness) between day 7 and day 21, day 7 and day 28 and day 7 and 32 of age in RIR chickens and between day 7 and day 21, day 7 and day 28 of age in VALO chickens (Additional file 2: Table S1). Similarly, phylogenetic significant changes were seen between day 7 and day 21, day 7 and day 28 and day 7 and 32 of age and day 14 and day 28 in RIR chickens And for VALO chicken between day 7 and day 21 of age (Additional file 2: Table S1). Shannon index, a measure of species distribution in a sample, significantly increased between day 7 and day 21 of age in both chicken breeds suggesting a more even community evolves over time (Additional file 2: Table S1). Spearman correlation coefficient (rs) between time and number of detected OTUs was 0.5968 (p < 0.0001) and time and phylogenetic diversity was 0.6544 (p < 0.0001) in control RIR and VALO groups.
The association between time and microbial diversity was also tested via simple linear regression (Additional file 2: Table S2), with microbial alpha diversity (the number of OTUs and phylogenetic diversity metrices) used as the dependent variable (Additional file 1: Fig. S4). In control groups, the number of OTUs, phylogenetic diversity and Shannon index significantly correlated with time (p > 0.001) (Additional file 1: Fig. S4). However, individual variation within the colon microbiota, especially VALO line at later time points (day 28 and day 32 of age) was greater as compared to age of 7-or 14-days-old.
Colon microbiota alpha diversity indices are significantly lower during the acute phase of H9N2 infection in chickens which is not altered in two different chicken breeds
First alpha diversity was quantified by the total number of observed species (OTUs) in each sample in experiment 1 (H9N2 infected versus sham infected RIR chickens) (Fig. 2a). We observed a significant reduction (p < 0.05 as determined Kruskal Wallis testing), in OTU number in colon samples from chickens at 2 days post influenza virus challenge (Fig. 2a and Additional file 2: Table S3), which corresponds with high levels of viral shedding (Fig. 1a), compared to the sham infected group at the same timepoint. The OTU numbers at day four and ten post challenge were not significantly different to the control group however, we observed higher variability in interquartile range of OTU numbers with the infected groups at days four and ten compared to the corresponding control groups and compared to the day 2 groups. This suggests that the acute reduction in number of OTU on day two occurred more uniformly than the recovery of OTU numbers post infection in individuals (Fig. 2a). Faith`s phylogenetic index of observed species which describes OTU diversity and Shannon diversity index which indicates OTU evenness were also measured (Fig. 2b, c). Kruskal Wallis statistical testing showed statistically significant (p < 0.01) reduced phylogenetic alpha diversity in H9N2 infected RIR chicken at acute phase of infection (day 2 post-challenge) and at day 10 post challenge as compared to control groups at the same time point measured (Fig. 2b and Additional file 2: Table S3). We observed no statistical difference in the Shannon Index between any of the infected groups and associated controls (Fig. 2c and Additional file 2: Table S3). Simple linear regression analysis showed that the Shannon index and the number of OTUs increased over the course of H9N2 infection in infected RIR group, similarly to the control group (Fig. 2d, f and Additional file 2: Table S2). The number of OTUs significantly correlated with time, as determined by linear regression (R2 = 0.666, p < 0.0001, equation Y = 5.914*X + 74.92 for control RIR chickens, and R2 = 0.510, p = 0.0004, equation Y = 5.914*X + 62.44 for H9N2 infected RIR chickens) (Fig. 2d and Additional file 2: Table S2). The Faith`s phylogenetic diversity significantly correlated with time for both the control and infected group but the increase in phylogenetic diversity of the infected group was retarded as compared to the control group, suggesting that diversity development of the chicken colon microbiota was reduced by H9N2 virus infection (R2 = 0.6894, p < 0.000, equation Y = 0.3286*X + 5.220 for control group, and R2 = 0.2114, p = 0.0414, equation Y = 0.1057*X + 4.653 for the H9N2 infected group) (Fig. 2e and Additional file 2: Table S2).
Figure 4 shows the alpha diversity measurements compared for RIR and VALO chicken breeds infected with H9N2 at day zero, day three and day fourteen post challenge. As it was seen in experiment 1, the number of OTUs and phylogenetic diversity dropped during the acute phase of infection (day 3 post infection) following H9N2 AIV challenge in both chicken lines (Fig. 4a, b) however this was not statistically significant (Additional file 2: Table S3). Between the chicken breeds no statistically significant differences in alpha diversity metrices before the challenge (day zero, D0) and during the acute phase of infection (day 3 post-challenge) (Fig. 4) were observed, both lines responded in a similar fashion to infection by H9N2 AIV. Interestingly, an increased Faith`s phylogenetic (Fig. 4b) and Shannon diversity (Fig. 4c) indices were found in VALO chickens as compared to RIR chickens at recovery phase of H9N2 infection (day 14 post-challenge) (p < 0.05) (Additional file 2: Table S3).
Beta diversity gut community changes are associated with H9N2 AIV infection and chicken breed
To compare the beta diversity among the groups at different time points, we performed Principal Coordinates Analysis (PCoA) and Principal component analysis (PCA) using the unweighted Unifrac data of taxonomic composition that includes phylogenetic diversity metrics (Fig. 5). A significant separation in the control groups was observed over the time of birds’ maturity (Fig. 5a). Significant differences in beta diversity within the RIR and VALO breed control groups were found in time-based manner of at least 7 days interval. Analysis of variance have shown significant differences between day 7 and 14, day 21 and 32, day 14 and 21, day 28 and 32 but not between day 28 and 32 in both chicken breed control groups (Additional file 2: Table S4). The only statistically significant difference between RIR and VALO chicken breeds was seen at day 21 of age for the control groups (Fig. 5a and Additional file 2: Table S4).
PCoA plots indicate a significant separation between control and H9N2 RIR infected chickens at all time points tested (day two, day four, and day ten post challenge) for experiment 1 (Fig. 5b). Analysis of variance (PERMANOVA) for measuring beta-diversity showed that the H9N2 RIR infected group had significantly lower diversity as compared to RIR control group at all time points tested (Additional file 2: Table S4). Similarly, significant separation in beta diversity was observed between day zero and day three post challenge for both RIR and VALO H9N2 infected chickens (Fig. 5c). Analysis of variance showed significant lower diversity in RIR and VALO infected groups as compared to sham infected groups at the same time point tested (Additional file 2: Table S4).
Bacterial taxa associated with H9N2 infection
A mean relative abundance of the dominant bacteria at phyla, class, order, and family levels between H9N2 AIV RIR infected and control chickens (Experiment 1) is shown in Additional file 1: Fig. S5 whereas between RIR and VALO H9N2 infected chickens and its corresponding controls (Experiment 2) is shown in Additional file 1: Fig. S6. Analysis of composition of microbiotas (ANCOM) was applied against group and day of infection variables to determine which bacteria were significantly differentiated in relative abundance at genus level (Fig. 6). The ANCOM results showed significant differences between the control and H9N2 infected RIR groups in members of the Furmicutes phylum. Six of Firmicutes phylum, Peptostreptococcaceae (Terrisporobacter), Planococcaceae (Lysinibacillus), Erysipelotrichaceae (Turicibacter), Lachnospiraceae (Cellulosilyticum), Paenibacillacea (Paenibacillus), Clostridiaceae 1 (Clostridium sensu stricto 1), were significantly different between the RIR control and H9N2 infected groups and had a high W-statistics and f-score (Fig. 6a). Detailed significant statistics of ANCOM percentile of different taxa is shown in (Additional file 3: Table S5). Furthermore, we also performed Linear discriminant analysis Effect Size (Lefse) analysis based on OTUs to compare the microbial communities between RIR control and RIR H9N2 infected birds at each time point tested. The LEfse analysis and ANCOM generated similar results (Figs. 6 and 7). LEfse results indicated differences in the phylogenetic distributions of the microbiota of H9N2 infected and control chickens at the OTU level (Fig. 7). The gut microbial communities in H9N2 infected birds were different compared to those in control groups. A histogram of the LDA scores was computed for features that showed differential abundance between H9N2 infected and control chickens (Fig. 7a, b). The LDA scores indicated that the relative abundances of Streptococcaecae (Streptococcus), and Planococcaceae (Lysinibacillus) were much more enriched in H9N2 infected birds versus control at day 2 post-challenge (Fig. 7a) and the most differentially abundant bacteria taxa (LDA score [log 10] > 3). The most differentially abundant bacterial taxon in control birds was characterized by a preponderance of Peptostreptococcaceae (LDA score [log10] > 3) at day 2 post-challenge (Fig. 7a). The differences in the phylogenetic distributions of the microbiotas of H9N2 infected and control chickens at the OTU level were also found at day 4 post-challenge (Fig. 7b). The LDA scores indicated that the relative abundances of Penicibacillacea (Penicibacillus), Planococcaceae (Lysinibacillus), Erysipelotrichaceae (Turicibacter), Clostridiaceae (Clostridium sensu stricto 1) were much more enriched in H9N2 infected birds as compared to control birds at day 4 post-challenge (Fig. 7b). The control birds were characterized mainly by a preponderance of different Clostridia and Bacillaceae (Bacillus) (LDA score [log10] > 3). In addition, we saw differential abundance of bacterial taxa between H9N2 infected RIR and VALO and their relative control groups in our second experiment (Fig. 8). The LDA scores indicated that the relative abundances of Clostridiales (Clostridium sensu stricto 1) and Planococcaceae (Lysinibacillus) (LDA score [log10] > 3.5) were much more enriched in RIR infected birds at day 3 post challenge as compared to the control group, and this correspond to our results obtained in first experiment at day 4 post-challenge (Fig. 7b). In VALO infected chicken, LDA scores indicated that the relative abundances of Aerococcaceae, Paenibacillaceae, Bacillaceae (LDA score [log10] > 3.5) were much more enriched at day 3 post-challenge whereas Clostridiales, Peptostreptococcae and Brevibacteriacea (LDA score [log10] > 3.5) were significantly reduced in the VALO infected group compared to control group at day 3 post challenge (Fig. 8).