Table 1.
Fig 1.
Genus-level taxonomic composition reveals extensive short-term donor microbial engraftment with FMT, followed by long-term reduction in donor similarity relative to donor self-similarity over a similar time period.
Genera present in the donor appear in (a) Subject A and (b) Subject B recipients with inconsistent long-term residence. For visual clarity, genera represented by >5% of assigned reads from at least one timepoint are shown. (c) Whole-community Bray-Curtis donor similarity for Subjects A and B, and an additional four patients (Subjects C, D, E and F) for which long-term and donor samples were obtained, is shown for genus-level composition. Donor similarity is calculated for each recipient timepoint with the corresponding donor sample used for FMT. As a point of reference, we show multiple timepoints for Donor #29, where similarity is measured against the first timepoint. Microbial sequencing reads were classified using Kraken [29] in conjunction with a sequence database collected from NCBI Refseq and Genbank microbial genome references. (d) Whole-community donor Chao similarity [33] in gene content is shown. Gene abundances were measured by alignment to the Uniref50 functionally annotated protein sequence database [32].
Fig 2.
Dominant Roseburia strain(s) are different in the short versus long term timepoint after FMT.
(a) A high number of single nucleotide variants (SNVs) distinguish the genus Roseburia strains present in the donor FMT sample (day 33 in donor time series) and Subject A recipient shortly after FMT from those in Subject A after one year following FMT. (b) Roseburia strains demonstrate stable, low-level diversity in the donor across eight time points taken over ~8 months, contrasting with the divergence observed in the recipient. Sequences were assembled from short read data obtained at the long-term time-point, and resultant contigs were classified taxonomically (see Methods). Contigs belonging to genus Roseburia were used as a reference for alignment of read data from all timepoints, followed by SNV calling and variable site selection. All polymorphisms distinguishing the donor from the long-term sample were selected, and presence of these discriminant polymorphisms in intermediate timepoints was recorded (adapted from a previously described approach [17]).
Fig 3.
The total gene complement of E. coli in Subject A experiences large-scale remodeling in association with FMT, resulting in broad reductions in genes significantly enriched for virulence factors.
PanPhlAn [37] was used to identify the presence of individual genes within the pan-genome of E. coli. Reads were aligned to a nonredundant collection of genes present in phylogenetically diverse E. coli isolate genome sequences. Those genes occurring at an abundance consistent with E. coli-specific occurrence were chosen and designated present or absent, as described previously [37]. Gene occurrence profiles were hierarchically clustered and divided into 10 groups with similar occurrence over the time series. Cumulative gene counts across groups are indicated on the left margin. Genes associated with virulence were identified by alignment to the Virulence Factor Database [39]. Over- or underrepresentation of virulence genes in cluster groups was determined using Fisher’s exact test to detect significant departure from random occurrence. Significant (p < 0.05) underrepresentation was found in group 1, and significant overrepresentation was found in groups 5 and 9. KEGG annotations of genes within each group are given in S1 File. Virulence factors found within each group are given in S2 File.
Fig 4.
Antibiotic resistome profile of Subject A is rapidly remodeled immediately after FMT.
Antibiotic resistance gene abundances were measured by alignment to the CARD antibiotic resistance database [19] and normalizing by per-sample coverage. Individual gene abundances are aggregated by antibiotic class. Genes conferring multiple antibiotic class resistance phenotypes are counted toward each antibiotic class.