Antibiotics and Host-Tailored Probiotics Similarly Modulate Effects on the Developing Avian Microbiome, Mycobiome, and Host Gene Expression

Abstract

The microbiome is important to all animals, including poultry, playing a critical role in health and performance. Low-dose antibiotics have historically been used to modulate food production animals and their microbiome. Identifying alternatives to antibiotics conferring similar modulatory properties has been elusive. The purpose of this study was to determine if a host-tailored probiotic could recapitulate effects of a low-dose antibiotic on host response and the developing microbiome. Over 13 days of life, turkey poults were supplemented continuously with a low-dose antibiotic or oral supplementation of a prebiotic with or without two different probiotics (8 cage units, n = 80 per group). Gastrointestinal bacterial and fungal communities of poults were characterized by 16S rRNA gene and ITS2 amplicon sequencing. Localized and systemic host gene expression was assessed using transcriptome sequencing (RNA-Seq), kinase activity was assessed by avian-specific kinome peptide arrays, and performance parameters were assessed. We found that development of the early-life microbiome of turkey poults was tightly ordered in a tissue- and time-specific manner. Low-dose antibiotic and turkey-tailored probiotic supplementation, but not nontailored probiotic supplementation, elicited similar shifts in overall microbiome composition during development compared to controls. Treatment-induced bacterial changes were accompanied by parallel shifts in the fungal community and host gene expression and enhanced performance metrics. These results were validated in pen trials that identified further additive effects of the turkey-tailored probiotic combined with different prebiotics. Alternative approaches to low-dose antibiotic use in poultry are feasible and can be optimized utilizing the indigenous poultry microbiome. Similar approaches may also be beneficial for humans.IMPORTANCE Alternative approaches are greatly needed to reduce the need for antibiotic use in food animal production. This study utilized a pipeline for the development of a host-tailored probiotic to enhance performance in commercial turkeys and modulate their microbiota, similar to the effects of low-dose antibiotic administration. We determined that a host-tailored probiotic, developed in the context of the commercial turkey gut microbiome, was more effective at modulating these parameters than a nontailored probiotic cocktail. Furthermore, the host-tailored probiotic mimicked many of the effects of a low-dose antibiotic growth promoter. Surprisingly, the effects of the antibiotic growth promoter and host-tailored probiotic were observed across kingdoms, illustrating the coordinated interkingdom effects of these approaches. This work suggests that tailored approaches to probiotic development hold promise for modulating the avian host and its microbiota.

Keywords: antibiotic; bacteria; fungi; host; microbiota; poultry; probiotics.

FIG 1

Enhanced weight gain achieved through single-strain probiotic supplementation. Each treatment represents 30 birds, with standard deviation shown. Weight differences at day 14 were tested with ANOVA and Tukey’s HSD. Bars with different letters are significantly different.

FIG 2

The young turkey microbiome develops over time. (A) Relative abundance of the most predominant taxa at the genus, or lowest taxonomic level denoted, per sample. “Other” represents taxa comprising less than 10% of the total relative abundance per sample. (B) Principal-coordinate analysis of unweighted UniFrac distances, colored by age of the bird. (C) OTUs significantly different in centered log-ratio-transformed relative abundance from one time point to the next. Stars denote the OTUs significantly different from that time point compared to day 6 (P < 0.05). OTUs are labeled as their most specific taxonomic identifier available. For panels A to C, turkeys were fed only control diet. Ileum: n = 14, day 3; n = 16, day 6; n = 15, day 13. Cecum: n = 16, day 3; n = 16, day 6; n = 16, day 13.

FIG 3

Antibiotics disrupt the turkey ileum microbiome. (A) Principal-coordinate analysis of unweighted UniFrac distances of the turkey ileum microbiome, colored by treatment (teal, control; orange, antibiotic treatment [BMD]). (B) Alpha diversity (Shannon index) for control and BMD-treated turkeys with P values reported. (C) OTUs significantly different in centered log-ratio-transformed relative abundance (CLR RA) in BMD versus control birds, by time point. Stars denote P < 0.05. OTUs are labeled as their most specific taxonomic identifier available. For panels A to C, numbers were as follows: day 3, n = 15, BMD; n = 14, control; day 6, n = 15, BMD; n = 16, control; day 13, n = 15, BMD; n = 15, control.

FIG 4

Antibiotics and probiotics similarly alter the turkey microbiome. Principal-coordinate analysis of unweighted UniFrac distances of the turkey ileum (A) and cecum (B) microbiome, colored by treatment: control, antibiotic (BMD), turkey-tailored probiotic (T-Pbx), commercial probiotic (FM-B11), and prebiotic (Prebx). Differences in centroids by treatment (denoted by a diamond) were tested by PERMANOVA, with R² and P values reported. Pairwise PERMANOVA was also performed on each treatment pair, with insignificant differences in centroids (P > 0.05) denoted by N.S. Full pairwise comparison results are listed in Data Set S1. (C) Centered log-ratio-transformed relative abundances of minor contributing OTUs in the ileum (less than 10%).

FIG 5

Antibiotics and probiotics similarly alter the turkey mycobiome on day 6. (A) The relative abundance of the most predominant fungal taxa at the species, or lowest taxonomy level denoted, per sample. “Other” represents taxa comprising less than 20% of the total relative abundance per sample. (B) Principal-coordinate analysis of Bray-Curtis distances, colored by treatment. Differences in centroids by treatment (denoted by a diamond) were tested by PERMANOVA, with R² and P values reported. Pairwise PERMANOVA was also performed on each treatment pair, with insignificant differences in centroids (P > 0.05) denoted by N.S. Full pairwise comparison results are listed in Data Set S1. (C) OTUs significantly different in centered log-ratio-transformed relative abundance (CLR RA) from one time point to the next. Stars denote OTUs significantly different in that treatment compared to its control treatment (P < 0.05). OTUs are labeled as their most specific taxonomic identifier available. For panels A to C, numbers were as follows: n = 14, control; n = 15, BMD; n = 16, T-Pbx; n = 12, FM-B11; n = 15, Prebx.

FIG 6

Antibiotics and probiotics can modulate ileal gene expression on day 6 of life. (A) Distribution of significant differentially expressed genes (P < 0.05) in the turkey at |log₂FC| > 1.0 (left) and |log₂FC| > 2.0 (right) according to pairwise tests with false-discovery rate correction, faceted by day of life. (B) Shared and unique differentially expressed genes in the turkey ileum on day 6 at |log₂FC| > 2.0. Circle size is proportional to the number of genes, and direction of expression change (↑ or ↓) is given.

FIG 7

Host phenotypes are correlated with microbiome shifts and treatment. (A) Procrustes analysis of principal coordinates from bacterial unweighted UniFrac distances and reactome pathway Euclidean distances. Samples are colored by treatment: control, antibiotic (BMD), turkey-tailored probiotic (T-Pbx), commercial probiotic (FM-B11), and prebiotic (Prebx). M² and P values from n = 999 permutations are reported. (B) Change in total body weights from day 3 to day 13 for turkeys separated by treatment and tested via Student’s t test with P values reported. Numbers per treatment: control, n = 37; BMD, n = 38; T-Pbx, n = 36; FM-B11, n = 36; Prebx, n = 34. (C) Weights for each turkey on days 3, 6, and 13 of life, separated by treatment. Differences in weight by treatment were tested for by using a Student t test. N.S. represents P values of <0.05. (D) Villus height/crypt depth ratios averaged by treatment group in the ileum across treatment groups. Letters denote statistical significance (P < 0.05).

FIG 8

(A) Peptide phosphorylation kinome profiles display clustering patterns by treatment, and turkey weight is loosely associated with treatment. Phosphorylation data for each peptide target site are presented, relative to control, on a heat map; red is increased phosphorylation, and green is decreased phosphorylation. Each condition is shown on the x axis. The connecting lines above the heat map show the relative similarity clustering of the different conditions described by the length of the lines. (B) Unique and shared effects of T-Pbx and BMD at the peptide signal (top) and KEGG pathway (bottom) levels, compared to their respective controls.