Over Feces Microbiota Structure in Laying Hens
Feces samples (n=90) were collected from ten laying hens at nine different time points (06:00-12:00-18:00-00:00-06:00-12:00-18:00-00:00-06:00) to investigate the circadian rhythms of feces microbiota composition using 16S rRNA sequencing with the Illumina HiSeq PE250 high-throughput sequencing platform. Unqualified samples were eliminated for classification, resulting in the analysis of 86 feces samples (9 each from time points 1 (06:00), 6 (12:00), 8 (00:00), and 9 (06:00) and 10 each from the other time points). A total of 6,502,101 high-quality sequences were obtained after quality monitoring, with an average of 75,605 (range 46,014-87,152) high-quality sequences per sample. The rarity curves of 86 samples (the minimum intercept at 97% sequence identity) were basically stable, indicating that the sampling depth was sufficient to describe the microbial community in the feces samples of laying hens.
When all other external conditions were the same, time of defecation could be among the most significant factors to explain the differences among individuals in microbiota community structure. The diversity of the feces microbiota community of laying hens was evaluated by the Observed ASVs and Chao1 and Shannon indexes (Figure 1). According to the results, community diversity (Observed ASVs and Chao1 and Shannon indexes) fluctuated significantly at different time points. The Observed ASVs and Chao1 index of the feces microbiota community of laying hens decreased gradually from 6:00 am on the first day to 12:00 noon on the second day but increased slowly from 12:00 midnight to 6:00 am on the third day (Figures 1A and 1B). Over time, the Shannon index also showed a trend of decreasing initially and then increasing slowly (Figure 1C). In addition, the alpha diversity of the feces microbiota at 6:00 am on the first day was significantly higher than that at other time points. To compare the overall differences in the feces microbiota composition of laying hens at 9 different time points, we conducted PCoA. The PCoA diagram shows that time points 1-4 are clustered together, time point 5 is separated, time points 6, 7 and 8 are further separated and clustered together, and time point 9 is clustered together with time points 1-4 (Figure 1D). Although some samples overlapped among the nine groups, there were moderate differences among the nine different groups in the microbiota. The results showed that the composition of the microbiota oscillated periodically with time.
Feces Microbiota Taxonomic Composition
A GraPhlAn phylogenetic tree shows the relative abundance of taxonomic groups from the phylum to species level with the top 150 features (Figure 2A). There were six phylum-level taxonomic groups with high relative abundance, namely, Firmicutes, Proteobacteria, Bacteroidetes, Acidobacteria, Fusobacteria, and Actinobacteria, and their average relative abundances accounted for 57.88, 14.11, 11.26, 6.23, 3.62, and 1.84% of the total sequences, respectively, which were regarded as the predominant bacterial phyla because their mean relative abundances accounted for greater than 1% of the total sequences. Notably, Firmicutes and Proteobacteria were the most abundant bacteria in the feces microbiota community of laying hens.
A total of 34 identified taxonomic groups were observed in the feces microbiota community at the family level, and 12 of the most predominant bacterial populations were present, including Turicibacteraceae, Streptococcaceae, Enterococcaceae, Lactobacillaceae, Clostridiaceae, Veillonellaceae, Peptostreptococcaceae, Ruminococcaceae, Lachnospiraceae, Bacteroidaceae, Enterobacteriaceae, and Fusobacteriaceae, while their mean relative abundance exceeded 1% of the total sequences, and all these major bacterial families accounted for 43.76% of the total sequences in the feces microbiota of laying hens. The remaining bacterial families, including Erysipelotrichaceae, S24-7, Paraprevotellaceae, Prevotellaceae, Rikenellaceae, Porphyromonadaceae, Burkholderiaceae, Alcaligenaceae, Comamonadaceae, Hydrogenophilaceae, Pasteurellaceae, Moraxellaceae, Bradyrhizobiaceae, Caulobacteraceae, Helicobacteraceae, Syntrophobacteraceae, Koribacteraceae, Bifidobacteriaceae, Actinomycetaceae, Coriobacteriaceae, Nitrospiraceae and Methanobacteriaceae, were considered low-abundance bacterial families, while their sequences all accounted for < 1% of the total sequences, and they accounted for only 24.91% of the total sequences in the feces samples. Lactobacillaceae (7.95%) belongs to the Firmicutes phylum, which was the most abundant classification group in the feces community of laying hens and was the most dominant family in the feces bacterial communities. The family-level taxonomic groups of the Firmicutes phylum were Turicibacteraceae, Streptococcaceae, Enterococcaceae, Lactobacillaceae, Clostridiaceae, Veillonellaceae, Peptostreptococcaceae, Ruminococcaceae, and Lachnospiraceae. Classified taxa in the Proteobacteria, Bacteroidetes, and Fusobacteria phyla at the family level were Bacteroidaceae, Enterobacteriaceae, and Fusobacteriaceae, respectively.
Further analysis of the data was performed to confirm whether the feces microbiota showed a circadian rhythm; therefore, box and line maps were made for eight bacterial phyla of the feces microbiota of laying hens at the phylum level, showing the relative abundances of the predominant phyla. The results showed that the relative abundance of the two most dominant phyla had an obvious circadian rhythm, which oscillated in antiphase. Highly robust circadian fluctuations were found, for instance, in Firmicutes and Proteobacteria, which had opposite fluctuations (Figure 2B and 2C). Between them, the relative abundance of Firmicutes reached its peak while that of Proteobacteria reached its trough at 06:00 the next morning. In addition, the change trend in the relative abundances of the other six predominant phyla did not fluctuate greatly (Additional file 1: Figure S1).
The 16S rRNA data of all feces samples were analyzed using the RF algorithm to determine the most important ASVs, and the top 30 important ASVs were Sutterella, Ruminococcus gnavus, Faecalibacterium, Lachnospiraceae, Ruminococcaceae, Enterococcus cecorum, Enterococcus, Lactobacillus reuteri, Lactobacillus, SMB53, Fusobacteriaceae, Clostridium colinum, Oscillospira, Bradyrhizobiaceae, JG37−AG−70, Burkholderia bryophila, S24−7, Porphyromonadaceae, Clostridium, Clostridiales, Megamonas, Salinispora tropica, Planococcaceae, Enterobacteriaceae, and ABS−6. These bacteria were classified into the phyla AD3, Bacteroidetes, Firmicutes, Fusobacteria, Nitrospirae, and Proteobacteria (Figure 3A). ABS−6 was the taxa related to AD3 in the feces of laying hens with the lowest importance. Two ASVs, S24−7 and Porphyromonadaceae, were related to Bacteroidetes. Firmicutes was the most abundant, including Ruminococcus gnavus, Faecalibacterium, Lachnospiraceae, Ruminococcaceae, Enterococcus cecorum, Enterococcus, Lactobacillus reuteri, Lactobacillus, SMB53, Clostridium colinum, Oscillospira, Clostridium, Clostridiales, Megamonas, and Planococcaceae. Fusobacteria contains Fusobacteriaceae, and Nitrospirae contains JG37−AG−70. Five ASVs, including Sutterella, Bradyrhizobiaceae, Burkholderia bryophila, Salinispora tropica, and Enterobacteriaceae, were related to Proteobacteria, with the importance of Sutterella being the highest. This result was a rather remarkable outcome showing that eight ASVs, Ruminococcus gnavus, Faecalibacterium, Enterococcus cecorum, Ruminococcaceae, Lachnospiraceae, Clostridium, Clostridiales, and Megamonas, which exhibited a circadian rhythm, had displayed a steady rise initially and then a slight decrease in the relative abundance (Figure 3B-I). We found that the relative abundances of these bacteria, including Ruminococcus gnavus, Faecalibacterium, Lachnospiraceae, Clostridiales, and Megamonas, reached their peak at the same time point 4, while the peaks of other bacteria, including Enterococcus cecorum, Ruminococcaceae and Clostridium, were delayed. The remaining ASVs remained in a stable state (Additional file 1: Figure S2). Although the composition of the microbiota was quite different, the common core bacterial community in the feces may play an important role.
Turnover of total, abundant, and rare of ASVs fractions
We calculated βNTI, RCBray to determine the assembly processes driving circadian rhythm of microbiota community composition. Whether for intra-group samples or inter-group samples, the most βNTI values between different samples were <−2 in the total ASVs, except ZT48 time point, suggesting the deterministric process; that is, homogeneous selection played a key role in shaping microbial composition in this study. Conversely, in the abundant and rare ASVs fractions, most βNTI values were between -2 and 2, indicating that the stochastic process was important, except the ZT48 at rare ASVs was >2 (indicates variable selection). The RCBray values of the microbial communities in the abundant ASVs were found to be >0.95, refelcting the dispersal limitation (weak selection) dominantly determined the microbial community. However, RCBray values of the microbial communities in the rare ASVs were found to be <0.95, indicating the “non-dominant” fractions (Figure 4).
Predicted Molecular Functions of Feces Microbiota
We analyzed the composition of enzymes at level 3, and the top 30 important enzymes were selected, which were ranked according to the importance from highest to lowest in the feces samples (Figure 5A). These profiles revealed that all enzymes at level 3 were related to metabolism. In the KEGG database, hydroxydechloroatrazine ethylaminohydrolase was the most predominant enzyme related to xenobiotic biodegradation and metabolism at level 2. At level 2, carbohydrate metabolism was related to eight enzymes: methylaspartate ammonia-lyase, methylaspartate mutase, L-ribulose-5-phosphate 3-epimerase, D-glucosaminate-6-phosphate ammonia lyase, L-xylulokinase, homocitrate synthase, UDP-galactopyranose mutase, and methylmalonyl-CoA carboxytransferase. In addition, N−carbamoylsarcosine amidase, homocitrate synthase and urocanate reductase were related to amino acid metabolism. The five enzymes methylaspartate mutase, urocanate reductase, caffeoyl−CoA O−methyltransferase, D-glucosaminate-6-phosphate ammonia lyase, and methylaspartate ammonia-lyase gained rhythmicity upon metabolic activity of the feces microbiota (Figure 5B and 5C). In addition, there was no obvious circadian rhythmicity in the remaining enzymes (Additional file 1: Figure S3). In further analysis of the data, we selected the related microbiota according to their contribution to these 30 enzymes (Additional file 1: Figure S4). We found that Fusobacteriaceae, Lactobacillaceae and Enterobacteriaceae had a major contribution to these enzymes. In addition, Burkholderiaceae, Clostridiaceae, Bradyrhizobiaceae and Veillonellaceae were also involved in the contribution to these enzymes.
The differential abundance of the feces microbiota gave rise to different functions of the microbiota. To understand the development of the functions of the feces microbiota community over time, Matacyc pathway compositions of the feces microbiota community were predicted using PICRUSt based on 16S rRNA data (Figure 6A). Moreover, the predictable functions were sorted according to the importance from highest to lowest. The 30 most symbolic Matacyc pathways that had been annotated at level 3 were identified in the feces samples, including L−glutamate degradation VIII (to propanoate); creatinine degradation II; methylaspartate cycle; aerobactin biosynthesis; sulfoglycolysis; enterobacterial common antigen biosynthesis; L−glutamate degradation V (via hydroxyglutarate); phospholipases; superpathway of L−tryptophan biosynthesis; superpathway of L−arginine and L−ornithine degradation; superpathway of L−arginine, putrescine, and 4−aminobutanoate degradation; adenosine nucleotides degradation IV; ppGpp biosynthesis; polyisoprenoid biosynthesis (Escherichiacoli); superpathway of sulfolactate degradation; cob(II)yrinate a,c−diamide biosynthesis I (early cobalt insertion); D−arabinose degradation III; superpathway of taurine degradation; superpathway of (Kdo)2−lipid A biosynthesis; reductive acetyl coenzyme A pathway; glutaryl−CoA degradation; L−lysine fermentation to acetate and butanoate; purine nucleobases degradation I (anaerobic); nylon−6 oligomer degradation; polymyxin resistance; superpathway of polyamine biosynthesis II; superpathway of hexuronide and hexuronate degradation; ethylmalonyl−CoA pathway; allantoin degradation IV (anaerobic); and superpathway of pyrimidine ribonucleotides de novo biosynthesis (Figure 6A). Interestingly, some of the Matacyc pathways fluctuated regularly over time. We found pathways involved in cob(II)yrinate a, c−diamide biosynthesis I (early cobalt insertion) to be among the microbiota functions oscillating (Figure 6B). Most remarkable and unexpected, however, were the functionalities that gained rhythmicity along with the microbiota, which included major pathways such as glutaryl−CoA degradation, L−glutamate degradation V (via hydroxyglutarate), L−lysine fermentation to acetate (Figure 6C) and butanoate and L−glutamate degradation VIII (to propanoate) (Figure 6D), as exemplified by cob(II)yrinate a, c−diamide biosynthesis I (early cobalt insertion). In addition, there was no obvious circadian rhythmicity in the remaining pathways (Additional file 1: Figure S5). Taken together, these results suggested that there was an association between functionalities of the feces microbiota and passage of time.
Co-occurrence Networks of Feces Bacteria
To identify the potential interactions among the feces microbiota, cocorrelative network analysis was conducted for a feces bacterial community based on strong and significant correlations (Spearman’s rs< -0.5 or rs> 0.5, P<0.01) (Figure 7). We performed a correlation analysis of the 30 ASVs identified earlier and found that there were significant positive or negative correlations between them. The cocorrelative network of the feces microbiota consisted of 30 nodes (important bacteria). Five clusters (modules) were identified with high credibility in the bacterial cocorrelative network in feces. In this network, Sutterella had positive correlations with Faecalibacterium, Ruminococcus gnavus, Lachnospiraceae, Fusobacteriaceae, Oscillospira, SMB53, Clostridium, Clostridiales, Ruminococcaceae, Megamonas, Clostridium colinum, and JG37-AG-70. However, Sutterella had negative correlations with Enterobacteriaceae, Enterococcus, and Planococcaceae. In addition, Enterococcus had negative correlations with Faecalibacterium, Lachnospiraceae, Fusobacteriaceae, Ruminococcus gnavus, Oscillospira, Sutterella, Ruminococcaceae, Clostridiales, Clostridium colinum, and Megamonas but positive correlations with Enterobacteriaceae and Planococcaceae. The relative abundance of Lactobacillus showed positive correlations with Lactobacillus reuteri but negative correlations with JG37-AG-70 and ABS-6 (Figure 7).