Maturation of the infant rhesus macaque gut microbiome and its role in the development of diarrheal disease

Abstract

Background: Diarrhea is the second leading cause of death in children under 5 years of age. Enhanced understanding of causal pathways, pathogenesis, and sequelae of diarrhea is urgently needed. Although the gut microbiota is believed to play a role in susceptibility to diarrheal diseases, our understanding of this association remains incomplete. Infant rhesus macaques (Macaca mulatta) are susceptible to diarrhea making them an ideal model to address this question.

Results: The maturation of the infant rhesus macaque gut microbiome throughout the first 8 months of life occurs in a similar pattern as that described for human infants. Moreover, the microbiome of the captive reared infant rhesus macaque more closely resembles that of human infants in the developing world than in the western world. Importantly, prior to disease onset, the gut microbiome of infants that later develop diarrhea is enriched in pathways of immunomodulatory metabolite synthesis, while those of infants that remain asymptomatic are enriched in pathways for short-chain fatty acid production. We identify Prevotella strains that are more abundant at 1 month in infants that later develop diarrhea. At 8 months, the microbiomes of animals that experience diarrhea show increased abundance of Campylobacter and a reduction in Helicobacter macacae.

Conclusion: The composition of the microbial community could provide a phenotypic marker of an infant's susceptibility to diarrheal disease. Given the significant physiological and immunological similarities between human and nonhuman primates, these findings provide potential markers of susceptibility to diarrhea that could be modulated to improve infant health, especially in the developing world.

Fig. 1

Maturation of the rhesus gut microbiome throughout the first 8 months of life. a Rectal swabs were collected from 80 dams 1 month after giving birth (40 at ONPRC and 40 at CNPRC) as well as their infants. Half of the infants (20/site) were then followed longitudinally with additional swabs collected at the 3- and 6-month time points. Finally, swabs from all 80 infants were obtained at the 8-month time point. b Principal coordinate analysis (PcoA) of unweighted UniFrac distances between microbial communities at different ages and locations. c The contribution of age, location, and individual to the total variance in the weighted and unweighted UniFrac dissimilarity matrices measured using PERMANOVA (Adonis with 10,000 permutations). d Bar graphs illustrating average UniFrac distances between infants at different ages and dams (top) and within each age group (bottom) (separate one-way ANOVA for both within group and vs. dams’ p < 0.001, with Holm-Sidak’s multiple comparison test, *p < 0.05, **p < 0.01, ***p < 0.001, dams were significantly different from all infant time points). e Violin plot of measured phylogenetic diversity at each time point each point represent an individual sample with solid lines indicating the median value for that age group (one-way ANOVA p < 0.001, with Holm-Sidak’s multiple comparison test, **p < 0.01, ***p < 0.001)

Fig. 2

Similarity of the infant macaque gut microbiome to human children. a Principal coordinate analysis (PcoA) of Bray-Curtis distances between gut microbial communities of pre-weaned 1-month-old infant rhesus macaque and human infants between 6 months and 2 years of age from the USA (western), Malawi (developing), and Amerindians (developing) at the genus (L6) level. b Bar graphs illustrating the average Bray-Curtis distances between 1-month-old infant macaques and human (6 months–2 years) from western (USA) and developing (Malawi, Amerindians) countries (one-way ANOVA p < 0.001, with Holm-Sidak’s multiple comparison test, ****p < 0.0001, error bars = SEM). c Principal coordinate analysis (PcoA) of Bray-Curtis distances between gut microbial communities of post-weaned 8-month-old infant rhesus macaque and human infants between 2 and 6 years of age from the USA (western), Italy (western), Malawi (developing), Amerindians (developing), and Burkina Faso (developing) at the genus (L6) level. d Bar graphs illustrating the average Bray-Curtis distances between 8-month-old infant macaques and human (2–6 years old) from western (USA and Italy) and developing (Malawi, Amerindians, and Burkina Faso) countries (one-way ANOVA p < 0.001, with Holm-Sidak’s multiple comparison test, ****p < 0.0001, error bars = SEM)

Fig. 3

Age-related changes in taxa in the rhesus macaque gut microbiome. a Phyla plot organized by host age. All phyla below 1% average abundance grouped into “Other”. Bars represent the average for the indicated time point. b Line graph indicating longitudinal changes in the relative abundance of the Actinobacteria and Spirochetes phyla in infant macaque gut microbiome (two-way ANOVA p < 0.0001, Bonferroni multiple comparison test *p < 0.05, ***p < 0.001). c Density plot of 12 abundant taxa to illustrate host age-dependent phylogenetic shifts

Fig. 4

Impact of diarrhea on the taxonomy of the rhesus gut microbiome. a Growth trajectory of asymptomatic monkeys, and those that experienced diarrhea (unpaired t-test at each time point, **p < 0.01, ***p < 0.001). b PcoA of unweighted UniFrac distances at the 1-month time point (prior to diarrhea) and at the 8-month time point (after diarrhea). c The contribution of host status to the total variance in the weighted and unweighted UniFrac dissimilarity matrices within each time point measured using PERMANOVA (Adonis with 10,000 permutations). d UniFrac distances illustrating inter-group variation at the 1-month time point (prior to diarrhea) and at the 8-month time point (after diarrhea) (one-way ANOVA p < 0.001, with Holm-Sidak’s multiple comparison test, **p < 0.01, ****p < 0.0001). e LEfSe (Log₁₀ LDA score > 2) illustrating taxa that are significantly different between infants that remained asymptomatic and those that had diarrhea at the 8-month time point. f Violin plot of the relative abundance of Campylobacter and Helicobacter at each time point, each point represents an individual sample with solid lines indicating the median value for that age group (one-way ANOVA p < 0.001, with Holm-Sidak’s multiple comparison test, **p < 0.01, ***p < 0.001)

Fig. 5

The functional potential of the gut microbiome of infant that experienced diarrhea or remained asymptomatic at 1 and 8 months of age. a PcoA Bray-Curtis dissimilarity built on the abundance of all functional genes annotated using the Uniref50 database. b The contribution of host status to the total variance in the weighted and Bray-Curtis dissimilarity matrices within each time point measured using PERMANOVA (Adonis with 10,000 permutations). c, d Select MetaCyc pathways that are enriched in animals that experienced diarrhea or remained asymptomatic at 1 (c) and 8 (d) months of age (LEfSe, Log₁₀ LDA score > 2)

Fig. 6

Assembled Prevotella and Campylobacter genomes show diarrhea-related trends. a Prevotella core genome phylogram built on the alignment of all protein coding genes common to all members of the tree (15 assembled genomes, 3 isolate genomes, 4 previously publish metagenomic assembled genomes) with exception of the out-group Bacteroides fragilis. Five genomes were placed in the diarrhea-associated Prevotella group due to their distance from other assembled genomes. b Percentage of metagenomic reads that align to the five diarrhea-associated Prevotella genomes; each point represents an individual sample; mean and standard error of the mean are shown (one-way ANOVA p < 0.001, with Holm-Sidak’s multiple comparison test, **p < 0.01, ***p < 0.001). c Campylobacter core genome phylogram built on the alignment of all protein coding genes common to all members of the tree (3 assembled genomes, 4 human isolate genomes, 4 rhesus macaque clinical isolate genomes) with exception of the outgroup H. macacae. d Percentage of metagenomic reads that align to assembled Campylobacter genomes for both asymptomatic monkeys and those that had diarrhea; each point represents an individual sample; mean and standard error of the mean are shown (one-way ANOVA p < 0.001, with Holm-Sidak’s multiple comparison test, **p < 0.01, ***p < 0.001)