Precision culturomics enabled by unlabeled single-cell morphology and Raman spectra

Abstract Selective enrichment of target bacteria from complex communities such as the human microbiome has remained a challenge. Here we report a solution based on morphology, Raman spectrometry and the Laser-Induced Forward Transfer technology, which works at microbial single cells, many generations before they appear as colonies. We develop a machine learning-based framework that enables species-level targeted sorting of single microbial cells from complex microbiome. We illustrate the utility of this approach in selecting for or against specific bacteria in fecal microbiome samples and the potential for quantifying the molecules expressed based on Raman spectra. Analysis of single-cell cultured genomes reveals that brief antibiotic use drives both pre-existing resistance and de novo mutations in the transpeptidase or efflux pumps of gut commensals, along with convergent evolution between different species. Our precision culturomics method should enable detailed morphological, metabolic, and genomic insights into variations in microbial phenotypes at the single-cell level for microbiome studies.