Transcriptomic profiles of single-copy marker genes enable predicting bacterial growth states in microbial communities

Abstract

Studying microbial community dynamics is fundamental to better understand ecosystem stability, resilience and environmental change. Community composition changes with the growth of individual members, yet current methods to estimate microbial growth in communities face substantial limitations. For example, genome sequence-based estimates of maximum growth rates may not reflect growth patterns in the natural environment well, and metagenomic in situ growth prediction requires the availability of reference genomes and shows limited accuracy for slow-growing bacteria. Gene expression data provide an information-rich readout of community activity that could reflect growth, however, cross-species comparisons in community settings remain challenging. An approach using expression signatures of universal, single-copy marker genes provides independence from reference genomes and may thereby enable comparability across species. Here, we present a transcriptomic, marker gene-based growth classifier that predicts the growth states of bacterial strains from different phyla cultivated in diverse conditions. We demonstrate its application in vivo in gnotobiotic mice carrying the same bacterial strains, and in a more complex synthetic community, where predicted growth states align with reported growth inhibition induced by systemic inflammatory response. This approach offers a new method for predicting bacterial growth states across species, with potential for broad application in the study of microbial growth dynamics at the whole community level.