Mesoscale computational protocols for the design of highly cooperative bivalent macromolecules

ABSTRACT The last decade has witnessed a swiftly increasing interest in the design and production of novel multivalent molecules as powerful alternatives for conventional antibodies in the fight against cancer and infectious diseases. However, while it is widely accepted that large-scale flexibility (10 − 100 nm) and free/constrained dynamics (100 ns − µ s) control the activity of such novel molecules, computational strategies at the mesoscale still lag behind experiments in optimizing the design of crucial features, such as the binding cooperativity (a.k.a. avidity). In this study, we introduced different coarse-grained models of a polymer-linked, two-nanobody composite molecule, with the aim of laying down the physical bases of a thorough computational drug design protocol at the mesoscale. We show that the calculation of suitable potentials of mean force allows one to apprehend the nature, range and strength of the thermodynamic forces that govern the motion of free and wall-tethered molecules. Furthermore, we develop a simple computational strategy to quantify the encounter/dissociation dynamics between the free end of a wall-tethered molecule and the surface, at the roots of binding cooperativity. This procedure allows one to pinpoint the role of internal flexibility and weak non-specific interactions on the kinetic constants of the NB-wall encounter and dissociation. Finally, we quantify the role and weight of rare events, which are expected to play a major role in real-life situations, such as in the immune synapse, where the binding kinetics is likely dominated by fluctuations. SIGNIFICANCE Multivalent and multispecific molecules composed of polymer-linked nanobodies have gained interest as engineered alternatives to conventional antibodies. These therapeutic molecules have a larger reach due to their smaller size and promise substantial and tunable gains in avidity. This paper studies a model diabody to lay the bases of a multi-scale computational design of the structural and dynamical determinants of binding cooperativity, rooted in a blend of atomistic and coarse-grained MD simulations and concepts from statistical mechanics.