Engineering of microbial communities in open environments remains challenging. Here we describe a platform used to identify and modify genetically tractable mammalian microbiota by engineering community-wide horizontal gene transfer events in situ. With this approach, we demonstrate that diverse taxa in the mouse gut microbiome can be modified directly with a desired genetic payload. In situ microbiome engineering in living animals allows novel capabilities to be introduced into established communities in their native milieu.
Subscribe to Journal
Get full journal access for 1 year
only $20.17 per issue
All prices are NET prices.
VAT will be added later in the checkout.
Rent or Buy article
Get time limited or full article access on ReadCube.
All prices are NET prices.
The raw data from this study are available from the corresponding author upon request.
Lagier, J. C. et al. Nat. Microbiol. 1, 16203 (2016).
Yaung, S. J., Church, G. M. & Wang, H. H. Methods Mol. Biol. 1151, 3–25 (2014).
Cuív, P. O. et al. Sci. Rep. 5, 13282 (2015).
Mimee, M., Tucker, A. C., Voigt, C. A. & Lu, T. K. Cell Syst. 1, 62–71 (2015).
Tock, M. R. & Dryden, D. T. Curr. Opin. Microbiol. 8, 466–472 (2005).
Marraffini, L. A. Nature 526, 55–61 (2015).
Thomas, C. M. & Nielsen, K. M. Nat. Rev. Microbiol. 3, 711–721 (2005).
Human Microbiome Project Consortium. Nature 486, 207–214 (2012).
Pansegrau, W. et al. J. Mol. Biol. 239, 623–663 (1994).
Hapfelmeier, S. et al. Science 328, 1705–1709 (2010).
Myhal, M. L., Laux, D. C. & Cohen, P. S. Eur. J. Clin. Microbiol. 1, 186–192 (1982).
Kommineni, S. et al. Nature 526, 719–722 (2015).
Saeidi, N. et al. Mol. Syst. Biol. 7, 521 (2011).
Steidler, L. et al. Science 289, 1352–1355 (2000).
Wegmann, U., Horn, N. & Carding, S. R. Appl. Environ. Microbiol. 79, 1980–1989 (2013).
Sheth, R. U., Cabral, V., Chen, S. P. & Wang, H. H. Trends Genet. 32, 189–200 (2016).
Klümper, U. et al. ISME. J. 9, 934–945 (2015).
Dahlberg, C., Bergström, M. & Hermansson, M. Appl. Environ. Microbiol. 64, 2670–2675 (1998).
Bikard, D. et al. Nat. Biotechnol. 32, 1146–1150 (2014).
Brophy, J. A. N. et al. Nat. Microbiol. 3, 1043–1053 (2018).
Simon, R., Priefer, U. & Pühler, A. Nat. Biotechnol. 1, 784–791 (1983).
Datsenko, K. A. & Wanner, B. L. Proc. Natl Acad. Sci. USA 97, 6640–6645 (2000).
Kozich, J. J., Westcott, S. L., Baxter, N. T., Highlander, S. K. & Schloss, P. D. Appl. Environ. Microbiol. 79, 5112–5120 (2013).
Caporaso, J. G. et al. Proc. Natl Acad. Sci. USA 108, 4516–4522 (2011).
Parada, A. E., Needham, D. M. & Fuhrman, J. A. Environ. Microbiol. 18, 1403–1414 (2016).
Apprill, A., McNally, S., Parsons, R. & Weber, L. Aquat. Microb. Ecol. 75, 129–137 (2015).
Edgar, R. C. Nat. Methods 10, 996–998 (2013).
Wang, Q., Garrity, G. M., Tiedje, J. M. & Cole, J. R. Appl. Environ. Microbiol. 73, 5261–5267 (2007).
Nurk, S. et al. J. Comput. Biol. 20, 714–737 (2013).
Carattoli, A. et al. Antimicrob. Agents Chemother. 58, 3895–3903 (2014).
We thank members of the Wang lab for helpful scientific discussions and feedback, especially R. Sheth for helpful insights on NGS library preparation. H.H.W. acknowledges funding from DARPA (W911NF-15-2-0065), NIH (1DP5OD009172, 5R01AI132403, and 1R01DK118044), NSF (MCB‐1453219), ONR (N00014-15-1-2704), and Burroughs Wellcome PATH (1016691). C.R. is supported by a Junior Fellowship of the Simons Society of Fellows (#527896). S.P.C. is supported by an NIDDK F30 fellowship (F30 DK111145-01A1) and an NIH MSTP training grant (NIH T32GM007367). We also thank A. Figueroa at the Columbia University Medical Center Flow Cytometry Core for assistance with FACS studies.
A provisional patent application has been filed by the Trustees of Columbia University in the City of New York based on this work.
Publisher’s note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Integrated supplementary information
Supplementary Figure 1 Overview of metagenomic alteration of gut microbiome by in situ conjugation (MAGIC) and plasmid maps of MAGIC vectors.
(a) In contrast to traditional approaches to cultivate microbes first and then test for genetic accessibility, MAGIC harnesses horizontal gene transfer in the native environment to genetically modify bacteria in situ. Transconjugant bacteria can be detected by FACS or antibiotic selection and further manipulated. (b) Map of Himar transposon integrative vectors (pGT-Ah and pGT-Kh variants found in libraries L2, L4, L5, L6, L7 and L8). (c) Map of replicative vectors with pBBR1 origin of replication (pGT-B variants found in libraries L1, L4, and L6). (d) Map of replicative vectors with RSF1010 origin of replication (pGT-S variants found in library L3). Although this vector backbone contains genes involved in conjugation (black), these vectors are not self-transmissible (J. Bacteriol. 117, 619–630, 1974; Gene 75, 271–288, 1989).
(a) Illustration of FACS enrichment method to isolate transconjugant cells from complex recipient populations. GFP and mCherry fluorescence are used to gate cell populations consisting of E. coli donors and diverse recipients. Quadrants Q1 and Q2 correspond to donor cells (mCh+), and unmanipulated recipients are in quadrant Q3. Quadrant Q4 contains transconjugants that received the GFP gene cargo and are not naturally mCherry-fluorescent (GFP+mCh–). Q4 cells are isolated and further analyzed. This gating was used to analyze fecal samples from each individual mouse in each in situ experiment, as well as every in vitro conjugation in this study by flow cytometry. (b) To validate the FACS enrichment method, we mixed GFP+ E. coli with a natural mouse fecal bacterial community at given levels (1–100% of population) and retrieved the E. coli by FACS. 16S sequencing of the samples showed that the fluorescent E. coli were efficiently and specifically enriched by FACS. Although the raw Q4 population contained some autofluorescent cells, the only remaining OTU in Q4 after application of an enrichment filter (see Methods) was E. coli.
Supplementary Figure 3 pGT vectors were transferred from E. coli donors to representative recipient species during in vitro conjugations.
(a) In vitro conjugation efficiency of replicative vector pGT-B1 from E. coli donor to various recipients, which are plotted by phylogenetic relationships. (b) In vitro conjugation efficiency of vector pG-Ah1 between E. coli donor and various recipients. This vector is replicative only in Proteobacteria (E. coli, S. enterica, V. cholerae, P. aeruginosa) but delivered genetic cargo by transposition into a broader array of bacteria. Asterisks indicate cultures grown in anaerobic conditions; all other cultures were grown aerobically. Conjugation efficiencies were calculated from 2 independent conjugations.
Supplementary Figure 4 pGT vectors were transferred from E. coli donors to mouse fecal bacteria during in vitro conjugations.
(a) In vitro conjugation of pGT vectors from EcGT2 donor strain into fecal bacteria extracted from mouse feces. (b) Aerobic (top) and anaerobic (bottom) conjugations were carried out with EcGT2 strains containing no vector (mock conjugation), a nontransferable vector (pGT-NT), pGT-L3, pGT-L7, and pGT-L8. Aerobic conjugations were plated on selective and nonselective media and grown aerobically at 37 °C for 24 h. Anaerobic conjugations were plated on selective and nonselective media, grown anaerobically at 37 °C for 48 h, and exposed to oxygen at room temperature for 48 h. Red arrows indicate GFP+ CFUs on nonselective plates. (c) Efficiencies of aerobic (top) and anaerobic (bottom) conjugations. Aerobic conjugation efficiencies were calculated from 3 independent conjugations; anaerobic conjugation efficiencies were calculated from 1 conjugation.
Supplementary Figure 5 FACS enriches for GFP+, antibiotic-resistant transconjugant gut bacteria arising from in vitro conjugations.
(a) Implementation of FACS enrichment of in vitro conjugations. (b) Conjugations between EcGT2 harboring vector libraries pGT-L3, pGT-L7, and pGT-L8 and mouse fecal bacteria were performed aerobically overnight. A mock conjugation using EcGT2 with no vector and a negative control conjugation using the pGT-NT nontransferable vector were also performed. 20,000 FACS-sorted events from Q3 (mCherry–GFP–) and Q4 (mCherry–GFP+) populations were plated on selective and nonselective media and grown aerobically to select for transconjugants. Cultivable aerobic transconjugants of pGT-L3 and pGT-L7 vectors were successfully enriched by FACS, although GFP+ CFUs may appear dim against the autofluorescent media. This experiment was performed independently twice, with similar results.
Overlays of bright-field, GFP, and mCherry channels are shown alongside GFP and mCherry channels. Q3 populations from unmodified fecal bacteria are negative for both GFP and mCherry, whereas Q4 populations from aerobic overnight in vitro conjugations of vector libraries pGT-L3 and pGT-L7 show enrichment of GFP-expressing cells, as well as some donor cells (mCherry+GFP–), which were eliminated in downstream sequencing analyses. This experiment was performed independently three times, with similar results.
(a) FACS dot plots of in vitro conjugations of mouse gut bacteria and EcGT2 donors with vector libraries pGT-L1, L3, and L7. This experiment was performed 3 times, with similar results. Green boxes define the sorted GFP+mCherry– transconjugant populations. (b) 16S taxonomic classification of in vitro GFP+mCherry– transconjugants of pGT-L1, L3, and L7 enriched by FACS. Relative abundance of each OTU in the unsorted population is shown in the grayscale heat map, and fold enrichment for transconjugants of each OTU is shown in the orange heat map, with annotated taxonomic identities. Bracketed values indicate confidence of taxonomic assignment by RDP Classifier. Genera with successfully cultivated isolates are denoted by stars. Each column represents FACS-enriched transconjugants from one conjugation. (c) Comparison of OTUs shared between transconjugants arising from each vector library during in vitro conjugations. 18 OTUs were shared between all 3 libraries, with a total of 47 OTUs being shared between at least 2 libraries.
(a) Implementation of MAGIC in a mouse model with fecal bacterial analysis by FACS, antibiotic selection, and sequencing. (b) FACS dot plots of in situ conjugations using EcGT2 donors with vector libraries pGT-L1, L2, and L3. Green boxes define the sorted GFP+mCherry– transconjugant populations. Each plot shows fluorescence expression of bacteria from the combined fecal samples of 3 cohoused mice. The experiment was run 3 independent times, with similar results. (c) 16S taxonomic classification of FACS-enriched transconjugants from in situ mouse experiments using vector libraries pGT-L1, L2, and L3 at 6 h post-gavage. Relative abundance of each OTU in the unsorted population is shown in the grayscale heat map, and fold enrichment for transconjugants of each OTU is shown in the orange heat map, with annotated taxonomic identities. Bracketed values indicate confidence of taxonomic assignment by RDP Classifier. Each column represents data from a separately housed cohort of 3 mice whose fecal samples were combined for analysis. Genera with successfully cultivated isolates are denoted by stars. (d) The pGT-L3 transconjugant population from (b) was further analyzed by comparison of Q4 enriched OTUs against Q3 OTUs, which represent a sample of the GFP– native bacteria population, and by enrichment analysis of Q4 samples that were sorted again for Q4. Enriched GFP+ transconjugants were robust whether compared against the total fecal population or against Q3. 7 out of 11 OTUs enriched in Q4 were present in the double-sorted Q4 population, indicating that Q4 sorting is robust. The OTUs lost upon double-sorting were obligate anaerobes and likely sensitive to prolonged aerobic conditions during double-sorting.
Supplementary Figure 9 Identification of FACS-enriched in situ transconjugants of multi-vector libraries.
(a) Flow cytometric quantification of in situ transconjugants in the total bacterial population after gavage of EcGT2 donors containing pGT-L4 (green; n = 4 mice) or pGT-L5 (blue; n = 4 mice) vector libraries. Control groups gavaged with PBS (black; n = 2 mice) or donors containing a nontransferable pGT-NT vector (red; n = 2 mice) produced no detectable transconjugants. Black bars indicate means. (b) Longitudinal analysis of mouse fecal microbiome by flow cytometry for presence of transconjugants after gavage of EcGT2 donors containing pGT-L4 (green; n = 6 mice) or pGT-L5 (blue; n = 6 mice). Donor cells of these libraries (orange; n = 12 mice) were lost within 48 h, whereas transconjugants were observed up to 72 h post-gavage. The dotted line indicates the detection limit of flow cytometry. Error bars indicate s.d. (c) 16S taxonomic classification of transconjugants (GFP+mCh–) enriched by FACS of pGT-L4 and pGT-L5 recipient groups at 6 h post-gavage. Relative abundance of each OTU in the unsorted population is shown in the grayscale heat map on the left, and fold enrichment for transconjugants of each OTU is shown in the orange heat map on the right, with annotated taxonomic identities. Bracketed values indicate confidence of taxonomic assignment by RDP Classifier. Each column represents data from 6 mice from 2 independent cohorts whose fecal samples were combined for analysis. Genera with successfully cultivated isolates are denoted by stars.
Supplementary Figure 10 Identification of FACS-enriched in situ transconjugants in mice from a different commercial vendor.
(a) FACS dot plots of in situ conjugations using EcGT2 pGT-L3 donors in a cohort of mice from a different vendor (Charles River Laboratories). Green boxes define the sorted GFP+mCherry– transconjugant populations. Flow cytometry was performed 3 times, on fecal samples from individual cohoused mice, with similar results. (b) 16S taxonomic classification of FACS-enriched GFP+mCherry– transconjugants of pGT-L3 at 6 h post-gavage. Relative abundance of each OTU in the unsorted population is shown in the grayscale heat map, and fold enrichment for transconjugants of each OTU is shown in the orange heat map, with annotated taxonomic identities. Bracketed values indicate confidence of taxonomic assignment by RDP Classifier. Each column represents bacteria from one mouse. Genera with successfully cultivated isolates are denoted by stars. (c) Metagenomic 16S rRNA sequencing of mouse fecal samples shows that mice from different vendors have divergent gut microbiomes, with some shared OTUs. (d) In in situ experiments using the same vector library (pGT-L6) in cohorts of 3 mice each from different vendors, 10 transconjugant OTUs were shared between cohorts.
297 PCR-validated isolates from in situ experiments using vector libraries pGT-L3 and pGT-L6 were identified by 16S Sanger sequencing and assigned to a genus using RDP Classifier with assignment confidence > 0.89.
(a) Vector map of pGT-B1. GFP and β-lactamase genes are expressed from separate promoters on a replicative pBBR1 origin plasmid. (b) MGB4, an Escherichia fergusonii isolate containing pGT-B1, lost GFP expression over time when serially passaged without selection for 15 d. Plating was performed for 3 independent serial passages. (c) Quantification of carb-resistant and GFP+ CFUs of MGB4 over time; all CFUs remained carb-resistant as the population lost GFP expression. Center values are the means of 3 serial passages; error bars represent s.d. (d) Colony PCR for the pGT-B1 backbone showed that the plasmid was absent in GFP– CFUs at all time points surveyed. Each lane shows the PCR product for one colony. This PCR was performed once. (e) Vector map of pGT-Ah1, which contains GFP and β-lactamase genes on a transposable cassette. The plasmid backbone contains a chloramphenicol resistance gene for selection. (f) MGB9, an Escherichia fergusonii isolate containing pGT-Ah1, remained 100% GFP+ during serial passaging without selection over 11 d. Plating was performed for 3 independent serial passages. (g) Over time the proportion of MGB9 CFUs expressing the genes on the transposable cassette (GFP+ and carb-resistant) remained at 100%, whereas the chloramphenicol resistance conferred by the pGT-Ah1 backbone was lost in some of the population. Center values are the means of 3 serial passages; error bars represent s.d.
Supplementary Figure 13 Characterization of 3 modifiable gut bacteria (MGB) strains by whole-genome sequencing and in vitro conjugation.
(a) Three distinct MGB strains, isolated from in vitro conjugations between E. coli pGT donors and mouse fecal bacteria, were analyzed by whole-genome sequencing. MGB4 and MGB9 appear to be the same strain isolated from separate experiments with different pGT vectors transferred. Sequencing of (b) MGB4/9 and (c) MGB3 revealed the presence of genes involved in conjugation and genetic transfer. However, only MGB4/9 strains that shared homology with the pECO-fce plasmid were observed to transfer their pGT vectors to E. coli during in vitro conjugations. (d) PCR confirmation of pGT vector transfer from MGB4 to an E. coli recipient following in vitro conjugation. The conjugation was performed 3 times, with similar results; 5 individual transconjugants were assessed by colony PCR.
(a) In vivo gut colonization profiles of MAGIC donors EcGT1 (S17, galK::mCherry), EcGT2 (S17, asd::mCherry), and control E. coli MG1655 in C57BL/6 mice measured by flow cytometry of fecal bacteria after a single gavage of 109 cells. Mean values were calculated using feces from 2 gavaged mice; error bars indicate s.d. (b) Two orally gavaged doses of 109 EcGT1 cells resulted in a longer persistence of this donor in the gut. Mean values were calculated using feces from 2 gavaged mice; error bars indicate s.e.m.
(a) Schematic diagram of experiment: genetically tractable gut microbiota were isolated from the mouse microbiome in vitro and then orally gavaged to recolonize the gut. (b) MGB3, MGB4, and MGB9 strains orally gavaged into mice (n = 4) as a mixture recolonized the GI tract without any antibiotic treatment. MGBs were detectable in fecal samples for at least 15 d post-gavage. (c) MGB strains (namely, MGB4) were present in all sampled locations along the GI tract when the mice (n = 4) were euthanized 15 d post-gavage. Error bars represent s.d. (d) Phylogenetic tree of FACS-sorted GFP+mCherry– transconjugants in fecal samples from mice after 11 d post-gavage of MGB strains. Fecal samples from 4 mice were combined for analysis. Relative abundance of each OTU in the unsorted population is shown in the grayscale heat map, and fold enrichment for transconjugants of each OTU is shown in the orange heat map. Bracketed values indicate confidence of taxonomic assignment by RDP Classifier. The red asterisk denotes the Escherichia/Shigella OTU that shares a genus with the MGB4/9 donors.
About this article
Cite this article
Ronda, C., Chen, S.P., Cabral, V. et al. Metagenomic engineering of the mammalian gut microbiome in situ. Nat Methods 16, 167–170 (2019). https://doi.org/10.1038/s41592-018-0301-y
FEMS Microbiology Reviews (2020)
Synthetic and Systems Biotechnology (2020)
Proceedings of the National Academy of Sciences (2020)
A peek in the micro-sized world: a review of design principles, engineering tools, and applications of engineered microbial community
Biochemical Society Transactions (2020)
The ISME Journal (2020)