Autotransporter folding avoids a kinetic trap during vectorial translocation across the bacterial outer membrane
TL;DR
This work introduces BEAM, a multiscale framework that learns slow collective variables from coarse-grained simulations to guide all-atom enhanced sampling to explain how vectorial secretion accelerates pertactin folding by excluding an off-pathway kinetic trap.
Abstract
Autotransporter proteins are major virulence factors in Gram-negative pathogens, yet how they fold during secretion remains incompletely understood. A longstanding puzzle is why pertactin folds and is secreted in vivo within minutes but refolds in vitro over hours to days. We introduce BEAM, a multiscale framework that learns slow collective variables from coarse-grained simulations to guide all-atom enhanced sampling. Applied to a C-terminal segment of the pertactin passenger domain from Bordetella pertussis, BEAM achieved four- to six-fold greater conformational coverage than traditional collective-variable-guided adaptive sampling or unbiased molecular dynamics. The resulting free-energy landscape revealed a compact, non-native intermediate accessible in bulk solution but geometrically incompatible with vectorial translocation across the outer membrane. Kinetic simulations show that access to this intermediate slows folding, whereas excluding it produces rapid, in vivo-like kinetics. Together, these results explain how vectorial secretion accelerates pertactin folding by excluding an off-pathway kinetic trap. More broadly, BEAM provides a multiscale strategy for revealing hidden conformational states at atomic resolution.
