Protein homodimers have been classified as three-state or two-state dimers depending
on whether a folded monomer forms before association, but the details of the folding–binding
mechanisms are poorly understood. Kinetic transition networks of conformational states
have provided insight into the folding mechanisms of monomeric proteins, but extending
such a network to two protein chains is challenging as all the relative positions
and orientations of the chains need to be included, greatly increasing the number
of degrees of freedom. Here, we present a simplification of the problem by grouping
all states of the two chains into two layers: a dissociated and an associated layer.
We combined our two-layer approach with the Wako–Saito–Muñoz–Eaton method and used
Transition Path Theory to investigate the dimer formation kinetics of eight homodimers.
The analysis reveals a remarkable diversity of dimer formation mechanisms. Induced
folding, conformational selection, and rigid docking are often simultaneously at work,
and their contribution depends on the protein concentration. Pre-folded structural
elements are always present at the moment of association, and asymmetric binding mechanisms
are common. Our two-layer network approach can be combined with various methods that
generate discrete states, yielding new insights into the kinetics and pathways of
flexible binding processes.
