Abstract
AbstractThe peptidoglycan (PG) layer ofEscherichia coliis a single, interconnected gigaDalton molecule that is the largest in the cell. Experimental studies have established a number of the PG’s properties, and previous computational studies have simulated aspects of its behavior on sub-cellular scales, but none have fully modeled the PG’s compositional heterogeneity and no models have yet been constructed on the whole-cell scale. Here we use a combination of computational modeling approaches to construct whole-cell PG models at a resolution of one coarse-grained (CG) bead per glycan that are consistent with a wide variety of available experimental data. In particular, we derive plausible glycan strand length distributions for the polar and cylindrical regions of the cell that cover the full range of possible strand lengths and that are consistent with all available experimental data. In addition, we develop stochastic simulation code that explicitly models a cross-linking experiment from the literature that has a direct bearing on the extent to which Braun’s lipoprotein (Lpp) is partitioned between periplasmic and surface-exposed locations. We then use all of these data as inputs to a new computer code,PG_maker, which builds CG models of the PG on a whole-cell scale in under an hour. Finally, we use the resulting 3D models as a basis for: (a) estimating pore size distributions – which, despite the idealized nature of the models, are shown to be in surprisingly good agreement with experimental estimates – and (b) calculating the effects of the large numbers of periplasmic Lpps on the ability of freely diffusing proteins to access the compartment that lies between the PG and the outer membrane. The ability to combine a wide range of experimental data into structural models that are physically realizable in 3D helps to set the stage for performing simulations of the PG on the whole-cell scale in the near future.
Publisher
Cold Spring Harbor Laboratory