Construction and characterization of a genome-scale ordered mutant collection of Bacteroides thetaiotaomicron

Abstract

Background Ordered transposon-insertion collections, in which specific transposon-insertion mutants are stored as monocultures in a genome-scale collection, represent a promising tool for genetic dissection of human gut microbiota members. However, publicly available collections are scarce and the construction methodology remains in early stages of development. Results Here, we describe the assembly of a genome-scale ordered collection of transposon-insertion mutants in the model gut anaerobe Bacteroides thetaiotaomicron VPI-5482 that we created as a resource for the research community. We used flow cytometry to sort single cells from a pooled library, located mutants within this initial progenitor collection by applying a pooling strategy with barcode sequencing, and re-arrayed specific mutants to create a condensed collection with single-insertion strains covering >2500 genes. To demonstrate the potential of the condensed collection for phenotypic screening, we analyzed growth dynamics and cell morphology. We identified both growth defects and altered cell shape in mutants disrupting sphingolipid synthesis and thiamine scavenging. Finally, we analyzed the process of assembling the B. theta condensed collection to identify inefficiencies that limited coverage. We demonstrate as part of this analysis that the process of assembling an ordered collection can be accurately modeled using barcode sequencing data. Conclusion We expect that utilization of this ordered collection will accelerate research into B. theta physiology and that lessons learned while assembling the collection will inform future efforts to assemble ordered mutant collections for an increasing number of gut microbiota members.

A mutant fitness compendium in Bifidobacteria reveals molecular determinants of colonization and host-microbe interactions

“…D) Strain-level statistics of insertions in the progenitor collection. Barcodes were binned according to exact co-occurrence in the progenitor collection, and these bins were treated as effective "strains" 80 Second from bottom: wild type (black, n=6) and insertion strains in Bbr_1847 (shades of green, n=2) were grown in rMRS with 0.5% (w/v) starch. The insertion strains exhibited a growth phenotype, supporting the conclusion that the transporter encoded by Bbr_1847 imports starch.…”

“…We used previously described protocols 21,80 with minor modifications to locate mutant strains within the progenitor collection and choose optimal strains for re-arraying into the condensed collection. Notably, we used a plate-well pooling strategy 80 that generated 268 (172+96) sequencing libraries.…”

“…We used previously described protocols 21,80 with minor modifications to locate mutant strains within the progenitor collection and choose optimal strains for re-arraying into the condensed collection. Notably, we used a plate-well pooling strategy 80 that generated 268 (172+96) sequencing libraries. For inclusion in the condensed collection, we required that the barcode map to only one location in the genome (single-insertion strains) and we prioritized strains that only occurred once within the collection.…”

“…We used a previously developed protocol 80 with minor modifications to re-array select strains from the progenitor collection (172 96-well plates) into the final condensed collection (16 96-well plates). We used an EpMotion 5073 (Eppendorf) running EpBlue v. 40.5.3.10 inside of an anaerobic chamber for re-arraying.…”

High‐throughput functional genomics: A (myco)bacterial perspective

“…It is useful to note, too, that the use in these of a pooled mutant library ensures that very large numbers of mutants can be screened simultaneously (enabling an agnostic approach to candidate identification), by contrast with methods such as PROSPECT (discussed above), which use arrayed libraries of mutants. The disadvantage, of course, is that any mutants of interest must be (re‐)constructed to enable downstream analysis; alternatively, deconvolution strategies can be applied where libraries are of manageable size (Arjes et al, 2022; Baym et al, 2016; Shiver et al, 2021).…”