Metagenomic association analysis of gut symbiont Lactobacillus reuteri without host-specific genome isolation

Lactobacillus reuteri is a model symbiont colonizing the guts of vertebrates used for studies on host adaptation of the gut symbiont. Previous studies investigated host-specific phylogenetic and functional properties by isolating its genomic sequence. This dependency on genome isolation is a significant bottleneck. Here we propose a method to study the association between L. reuteri and its hosts directly from metagenomic reads without strain isolation by using pan-genomes. We characterized the host-specificity of L. reuteri in metagenomic samples not only in the previously studied organisms (mice and pigs) but additionally in dogs. For each sample, two types of profiles were generated: (1) genome-based strain type abundance profiles and (2) gene composition profiles. Our profiles showed host-association of L. reuteri in both phylogenetic and functional aspects without depending on the host-specific genome isolation. We could observe not only the presence of host-specific lineages but also the dominant lineages associated with the different hosts. Furthermore, we show that metagenome-assembled genomes provide detailed insights into the host-specificity of L. reuteri. We could infer evolutionary trajectories of host-associative L. reuteri strains in the metagenomic samples by placing the metagenome-assembled genomes into a phylogenetic tree and identify novel host-specific genes which were unannotated in existing pan-genome databases. Our pan-genomic approach drops the need for time-consuming and expensive host-specific genome isolation while producing consistent results with previous host-association findings in mice and pigs. Additionally, we could predict associations that have not yet been studied in dogs.