The dense microbial ecosystem in the gut is intimately connected to numerous facets of human biology, and manipulation of the gut microbiota has broad implications for human health. In the absence of profound perturbation, the bacterial strains that reside within an individual are mostly stable over time1. By contrast, the fate of exogenous commensal and probiotic strains applied to an established microbiota is variable, generally unpredictable and greatly influenced by the background microbiota2,3. Therefore, analysis of the factors that govern strain engraftment and abundance is of critical importance to the emerging field of microbiome reprogramming. Here we generate an exclusive metabolic niche in mice via administration of a marine polysaccharide, porphyran, and an exogenous Bacteroides strain harbouring a rare gene cluster for porphyran utilization. Privileged nutrient access enables reliable engraftment of the exogenous strain at predictable abundances in mice harbouring diverse communities of gut microbes. This targeted dietary support is sufficient to overcome priority exclusion by an isogenic strain4, and enables strain replacement. We demonstrate transfer of the 60-kb porphyran utilization locus into a naive strain of Bacteroides, and show finely tuned control of strain abundance in the mouse gut across multiple orders of magnitude by varying porphyran dosage. Finally, we show that this system enables the introduction of a new strain into the colonic crypt ecosystem. These data highlight the influence of nutrient availability in shaping microbiota membership, expand the ability to perform a broad spectrum of investigations in the context of a complex microbiota, and have implications for cell-based therapeutic strategies in the gut.
Publisher’s note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
We thank N. Ratnayeke for early experimental assistance, S. Higginbottom for gnotobiotic assistance, and K. Ng, Z. Russ and W. Van Treuren for analytical assistance; the Amieva and Huang laboratories for use of their microscopy resources, and D. Shepherd for valuable discussions; N. Pudlo and E. Martens for the protocol on porphyran extraction, and E. Sonnenburg, T. Fukami and M. Fischbach for commenting on this manuscript. This material is based upon work supported by the National Science Foundation under grant number 1648230, the NIDDK (R01-DK085025 to J.L.S.) and an NSF Graduate Fellowship (DGE-114747 to E.S.S.).Reviewer information
Nature thanks D. Bolam and the other anonymous reviewer(s) for their contribution to the peer review of this work.
Extended data figures and tables
Extended Data Fig. 1 Three model background communities of gut microbes are distinct from each other.
a, Principal coordinates (PC) analysis of weighted UniFrac distance for 16 S rRNA gene amplicons from faeces from the three background community groups from Fig. 1 before diet switch for conventional (RF) or humanized (Hum-1 and Hum-2) mice, n = 10. b, Comparison of weighted UniFrac distances within each group (Intra) or across groups (A × B, A × C, B × C). One-way ANOVA, P < 0.0001.
Extended Data Fig. 2 NB001 can utilize both inulin and porphyran as the only carbon source for growth.
a, NB001 demonstrates growth in minimal medium with either glucose (blue, doubling time = 157 min), or inulin (orange, doubling time = 127 min), as the only provided carbon source. b, Schematic of porphyran PUL from NB001 based on alignment to the previously published B. plebeius PUL, based on data from whole-genome sequencing. Grey bar, the region deleted via homologous recombination to abolish the ability to utilize porphyran. c, NB001 has the ability to grow on porphyran (doubling time = 98 min) as the only carbon source (WT), and growth is abrogated when genes required for porphyran utilization are knocked out (KO).
Extended Data Fig. 3 Porphyran does not significantly impact the gut microbiota in the absence of a known utilizer.
Weighted UniFrac analysis was performed on faecal 16 S rRNA data for conventional mice colonized with a porphyran utilization knockout (as in Fig. 2c) before (Pre, n = 8) or after (Post, n = 9) addition of porphyran. a, Principal coordinates analysis. b, Weighted UniFrac analysis. Unpaired two-tailed t-test, P = 0.25 (n.s., not significant).
Extended Data Fig. 4 Primary colonizer displacement is robust and contingent upon access to porphyran.
a, Conventional mice (n = 7), which were fed a MAC-rich diet and colonized with NB001 (PUL− 1) containing an eight-gene deletion abrogating its ability to utilize porphyran (Extended Data Fig. 2b, c), demonstrated resistance to subsequent challenge with an isogenic knockout strain (PUL− 2) in the presence of 1% porphyran in the drinking water. Notably, our conventionally raised mice were permissive to colonization by this strain and other tested species of Bacteroides (B. thetaiotaomicron, B. fragilis, B. uniformis, B. vulgatus, stable colonization range of 8 × 105–3 × 108 c.f.u. per ml faeces), which differs from reports of tests on other conventionally raised mice, potentially reflecting inter-colony microbiota differences. b, Mice from Fig. 2e were challenged with the originally colonizing porphyran utilization knockout (PUL−) that was displaced by the utilizer (PUL+) and demonstrated colonization resistance to the previously displaced knockout strain. Data are mean ± s.d. The grey-shaded boxes represent the limit of detection. Source Data
Extended Data Fig. 5 Minimal porphyran utilization PULs were constructed via PCR and yeast assembly.
Schematic representing construction of designed porphyran PULs. On the basis of the alignment to the previously published B. plebeius porphyran utilization PUL, three regions were targeted for minimal PUL assembly and amplified via PCR from the NB001 genome. The PCR fragments were assembled with digests of both a custom and commercially available vector in yeast (see Methods), after which colonies carrying correctly assembled plasmids were lysed and directly added to E. coli for electroporation.
Extended Data Fig. 6 B. stercoris and B. thetaiotaomicron demonstrate different abilities to grow in minimal medium.
a, b, Wild-type B. stercoris (a) and B. thetaiotaomicron (b) grown in SMM with glucose as the only carbon source demonstrate different maximum optical densities reached (B. stercoris maximum OD = 0.363, B. thetaiotaomicron maximum OD = 0.453). This suggests a possible explanation for why both species with the 21-gene PUL grow to different maximum optical densities as well (Fig. 3b).
Extended Data Fig. 7 Abundance of B. stercoris with the 34-gene PUL can be controlled in the context of a conventional mouse microbiota.
Conventional mice (n = 5) that were fed a MAC-rich diet were colonized with a strain of B. stercoris harbouring the designed 34-gene porphyran PUL and density of the engineered strain was tracked in the faeces. Upon administration of 1% porphyran in the drinking water (green-shaded box), density of B. stercoris increased, and subsequently decreased upon removal of porphyran. Data are mean ± s.d. The grey-shaded box represents the limit of detection. Source Data
This file contains Supplementary Table 1 (lists all oligos used for PCR and construction of fragments for yeast assembly), Supplementary Table 2 (lists the strategies for generating all PCR products used in creating PUL plasmids) and Supplementary Table 3 (lists the digests of commercial and custom vectors used in yeast assembly of PUL plasmids).