Abstract
Bacteria commonly live in dense and diverse communities, known as biofilms. As the major mode of microbial life, biofilms have been widely recognized for their impact on global biogeochemical cycling and the health of higher living organisms. Commonly used assays to study biofilm probe biofilm formation and behavior using ensemble averaged data. However, to better understand how the individual behaviors of biofilm dwelling cells contribute to the emergent macroscopic properties of biofilms, cellular level information needs to be extracted from densely packed bacterial biofilms. In this work, we integrated lattice light sheet microscopy (LLSM) and microfluidic systems for non-invasive, high-resolution time-lapse imaging of live bacterial communities under precisely controlled physical and chemical conditions. With this combination, we successfully imaged the colonization of glass surfaces by S. oneidensis MR-1 biofilms, a well-studied biofilm formation species, under media flow over a time period of three days, visualizing the evolution of single surface-attached cells into a dense 3D biofilm. To quantitatively analyze biofilm-dwelling cells, we developed Bacterial Cell Morphometry 3D (BCM3D), an integrated image analysis package that combines deep learning with conventional image analysis, and its novel extension version BCM3D 2.0, which enables measurement of cellular phenotypes such as cell size and distance to the nearest neighboring cell. With this quantitative analysis ability, we demonstrated that the presence of bile salts leads to aggregation of S. flexneri, an intracellular pathogen that causes watery or bloody diarrhea, at a cellular level, which had previously only been shown at an ensemble level. This cellular level imaging and analysis ability enables us to study the emergent properties of bacterial biofilms in terms of the fully-resolved behavioral phenotypes of individual cells, which provides a more complete understanding of bacterial biofilms.
