Why are so many enteric pathogen infections asymptomatic? Pathogen and gut microbiome characteristics associated with diarrhea symptoms and carriage of diarrheagenic E. coli in northern Ecuador

Figures & data

Figure 1. Summary of study design (a) and sample groups (b). All participant stool samples were PCR-screened for DEC pathotype genes and positive isolates were whole-genome sequenced. A random subset of age and sex-matched participant stool samples were selected for 16S rRNA gene amplicon gut microbiome sequencing, and a further subset were selected for shotgun metagenome sequencing so that there would be an approximately equal number of shotgun metagenomes for diarrheal case and control samples and DEC-positive and negative samples. Symbols in panel b indicate diarrhea case (+) or control (-) status and the presence (+) or absence (-) of DEC in participant stools. All samples with a DEC isolate genome are from infections because by-definition there were no DEC isolates detected in the uninfected stool samples. See text and Supplementary Table 2 for additional sample selection and filtering criteria.

Figure 2. DEC isolate metagenome abundances. Dashed lines indicate mean abundance for case versus control sample groups. Data are also shown as an inset box plot.

Figure 3. Comparison of the numbers of virulence genes from the virulence factor database (VFDB) in symptomatic versus asymptomatic DEC infections and in uninfected cases versus controls. Data are shown for virulence genes in the VFDB annotated as any taxa (a) and as E. coli (b).

Figure 4. Mean relative abundances of virulence genes that were significantly differentially abundant by DEC infection and diarrhea case/control status (4-way Kruskal-Wallis adjusted p-values < 0.05).

Figure 5. Comparisons of coverage-based estimates of within-sample alpha diversity for symptomatic versus asymptomatic DEC infections and for uninfected cases versus controls. Shannon, Simpson, and observed alpha diversity were calculated using 16S rRNA gene amplicon data (a-c); Nonpareil alpha diversity was calculated using shotgun metagenome data (d).

Figure 6. Mean relative abundances of 16S rRNA gene amplicon family-level bacterial taxa that were significantly associated with diarrhea and DEC infection status (corncob and/or LEfSe analyses; adjusted p < 0.05, LEfSe LDA threshold < 3). Values are square root transformed to improve visualization of taxa with low relative abundances.

Figure 7. Mean relative abundances of genes with secondary KEGG annotations that were significantly differentially abundant by DEC infection and diarrhea status (4-way Kruskal-Wallis adjusted p-values < 0.05).

Figure 8. Core genome phylogenetic tree of DEC strains from symptomatic and asymptomatic individuals; there was no phylogenetic signal for case/control status (δ = 0.59, p = 0.92).

Table 1. Results summary for analyses comparing pathogen abundance, gut microbiome characteristics, and pathogen characteristics for symptomatic versus asymptomatic DEC infections.

Factor	Figure	Analysis description	Primary result	Supporting result(s)
Pathogen abundance	Figure 2. E. coli relative abundance in the gut	Compared the relative abundance of DEC isolate sequences in the corresponding shotgun metagenome to determine the amount of E. coli in the gut for symptomatic versus asymptomatic DEC infections.	E. coli relative abundance was highly variable but higher on-average (p=0.0049) in symptomatic compared to asymptomatic DEC infections.	There was a strong correlation between the metagenome and qPCR-based E. coli abundance estimates (Figure S1). Results were similar for a sub-analysis of participants aged <5 years.
	Figure 3. Number of virulence genes in the gut	Compared the number of total and E. coli-annotated bacterial virulence genes in the gut.	There were more total (p=0.0075) and E. coli-annotated (p=0.0051) virulence genes in symptomatic than asymptomatic DEC infections.	There was no significant difference in the number of virulence genes in uninfected cases versus controls. There were no Salmonella or Campylobacter-annotated virulence genes that passed our filtering criteria of presence in > 10% of metagenomes, and the number of Klebsiella-annotated virulence genes was not significantly different by infection or diarrhea status. Results were similar for a sub-analysis of participants aged <5 years.
	Figure 4. Relative abundance of differentially abundant virulence genes in the gut	Compared the mean relative abundance of bacterial virulence genes that were significantly differentially abundant between sample groups.	Virulence genes that were significantly different (adjusted p < 0.05) between groups were most abundant in symptomatic DEC infections.	Abundances were next-highest in uninfected cases, followed by asymptomatic DEC infections. Those without diarrhea or a DEC infection (uninfected controls) had very low virulence factor abundances. Most genes were annotated as E. coli or closely related Shigella spp. Gene functions included pilus formation, nutrient acquisition, secretion systems, adhesion, and other classical virulence functions. Results were similar for a sub-analysis of participants aged <5 years.
(ii) Gut microbiome characteristics	Figure 5. Gut microbiome alpha diversity	Compared within-sample alpha diversity metrics calculated using 16S rRNA gene amplicon (Shannon, Simpson, and Observed ASVs) and shotgun metagenome (Nonpareil) gut microbiome data.	Alpha diversity was higher in asymptomatic than symptomatic DEC infections, though the comparison was only significant (p<0.05) for Observed ASVs.	A similar trend was observed for uninfected cases versus controls, where individuals without diarrhea had higher alpha diversity. Results were similar after filtering E. coli from the analyses.
	Figure 6. Differential abundance of gut microbiome taxa	Tested for differentially abundant family-level taxa in the gut microbiome using 16S rRNA gene amplicon data.	Bacterial families belonging to the Proteobacteria, including the Enterobacteriaceae and Pasteurellaceae were most abundant in symptomatic DEC infections, while asymptomatic infections were associated with high relative abundances of short chain fatty acid producers and strict anaerobes, including the Bifidobacteriace, Bacteroidaceae, and Verrucomicrobiaceae.	There were also differences in the relative abundance of gut microbiome taxa associated with case/control status, even in uninfected individuals; for example, individuals without DEC infections had high relative abundance of Paraprevotellaceae, regardless of diarrhea symptoms. Results were similar, though with fewer significant taxa, with differential relative abundance for the 16S rRNA gene amplicon sub-analysis of participants aged <5 years and the shotgun metagenome analysis.
	Figure 7. Relative abundance of genes associated with high-level cellular functions	Compared the mean relative abundances of bacterial functional genes that were significantly differentially abundant by DEC infection and diarrhea status.	Functional genes that were significantly different (p<0.05) between groups were most abundant in symptomatic DEC infections and spanned a wide array of high-level molecular functions.	Gene functions that were enriched in symptomatic DEC infections included those important for bacterial infection (capsule, pilus formation, etc.), cell maintenance, and the biosynthesis and metabolism of macromolecules.
(iii) Pathogen characteristics	Figure 8. Core genome phylogenetic tree	Tested for trait-associated phylogenetic signal using a core genome phylogeny of DEC isolates from symptomatic versus asymptomatic infections.	The core genome DEC isolate tree was mixed, with no phylogenetic signal based on isolation from an symptomatic versus asymptomatic infection.	Despite the lack of overall structure in the tree, we did observe small, tight clusters of closely related DEC strains that were all or majority from symptomatic or asymptomatic infections. There were no isolate virulence or accessory gene sequences associated with isolation from symptomatic versus asymptomatic infections.

Data availability statement

The data that support the findings of this study are openly available under NCBI BioProject PRJNA486009. R code is available on Github: https://github.com/kjojess/EcoZUR-symptomatic-asymptomatic-manuscript.