Natural and after colon washing fecal samples: the two sides of the coin for investigating the human gut microbiome

Abstract

To date several studies address the important role of gut microbiome and its interplay with the human host in the health and disease status. However, the selection of a universal sampling matrix representative of the microbial biodiversity associated with the gastrointestinal (GI) tract, is still challenging. Here we present a study in which, through a deep metabarcoding analysis of the 16S rRNA gene, we compared two sampling matrices, feces (F) and colon washing feces (CWF), in order to evaluate their relative effectiveness and accuracy in representing the complexity of the human gut microbiome. A cohort of 30 volunteers was recruited and paired F and CWF samples were collected from each subject. Alpha diversity analysis confirmed a slightly higher biodiversity of CWF compared to F matched samples. Likewise, beta diversity analysis proved that paired F and CWF microbiomes were quite similar in the same individual, but remarkable inter-individual variability occurred among the microbiomes of all participants. Taxonomic analysis in matched samples was carried out to investigate the intra and inter individual/s variability. Firmicutes, Bacteroidota, Proteobacteria and Actinobacteriota were the main phyla in both F and CWF samples. At genus level, Bacteirodetes was the most abundant in F and CWF samples, followed by Faecalibacterium, Blautia and Escherichia-Shigella. Our study highlights an inter-individual variability greater than intra-individual variability for paired F and CWF samples. Indeed, an overall higher similarity was observed across matched F and CWF samples, suggesting, as expected, a remarkable overlap between the microbiomes inferred using the matched F and CWF samples. Notably, absolute quantification of total 16S rDNA by droplet digital PCR (ddPCR) revealed comparable overall microbial load between paired F and CWF samples. We report here the first comparative study on fecal and colon washing fecal samples for investigating the human gut microbiome and show that both types of samples may be used equally for the study of the gut microbiome. The presented results suggest that the combined use of both types of sampling matrices could represent a suitable choice to obtain a more complete overview of the human gut microbiota for addressing different biological and clinical questions.

Figure 1

16S absolute quantification of DNA extracted from F and CWF sample matrices. Data are reported as the mean of triplicate ddPCR experiments and expressed as 16S copies/ng of total DNA (A dot-plot; B dumbbell plot).

Figure 2

Dumbbell plot of alpha diversity, measured using the Inverse Simpson’s richness index, for matched F and CWF samples. Alpha diversity scores were calculated by using rarefied data at an equal sampling depth of 68,000 sequences. Box-plot and points represent the overall data distribution and samples, respectively. Samples belonging to the same enrolled subject were connected by using a grey line. The group mean comparison was performed by using the Student’s t-test.

Figure 3

PCoA of unweighted UniFrac distances between paired F and CWF samples for the subjects dataset. Point shape and colors represent the sampling matrix and the enrolled subject, respectively.

Figure 4

Violin plot showing the observed weighted and unweighted UniFrac distances among paired (i.e. samples belonging to the same individual but from different matrices) and random pairs. The resulting p-value is shown for both comparisons.

Figure 5

Bacteria relative abundances at the phylum level. Only taxa with a relative abundance equal or higher to 1% were shown. Rare taxa were collapsed as “Other”.

Figure 6

Venn diagrams depicting relative amounts of unique and shared taxa among F and CWF paired samples at phylum, class, order, family, genus and species levels.

Figure 7

Violin plot showing the observed distances, measured by using both the Bray–Curtis and Jaccard metrics, among paired (i.e. samples belonging to the same individual but from different matrices) and random pairs at family, genus and species ranks. The resulting p-value is shown for each comparison.

Figure 8

F and CWF microbial networks obtained by using the NetCoMi framework. Nodes are sized and coloured according to normalized counts and cluster membership. Edges color reflects the correlation among nodes (green and red for positive and negative correlations, respectively). In particular, cluster sharing at least 5 nodes among the two networks were plotted using the same colour. Details about node name and taxonomic classification are available in Supplementary Table S6.