MVsim is a toolset for quantifying and designing multivalent interactions

Abstract

Arising through multiple binding elements, multivalency can specify the avidity, duration, cooperativity, and selectivity of biomolecular interactions, but quantitative prediction and design of these properties has remained challenging. Here we present MVsim, an application suite built around a configurational network model of multivalency to facilitate the quantification, design, and mechanistic evaluation of multivalent binding phenomena through a simple graphical user interface. To demonstrate the utility and versatility of MVsim, we first show that both monospecific and multispecific multivalent ligand-receptor interactions, with their noncanonical binding kinetics, can be accurately simulated. Further, to illustrate the conceptual insights into multivalent systems that MVsim can provide, we apply it to quantitatively predict the ultrasensitivity and performance of multivalent-encoded protein logic gates, evaluate the inherent programmability of multispecificity for selective receptor targeting, and extract rate constants of conformational switching for the SARS-CoV-2 spike protein and model its binding to ACE2 as well as multivalent inhibitors of this interaction. MVsim and instructional tutorials are freely available at https://sarkarlab.github.io/MVsim/.

Facile Display of Homomultivalent Proteins for In Vitro Selections

“…We recently reported MVsim, a graphical user-interface-based simulator for predicting multivalent binding kinetics with a comprehensive mechanistic model. , Here, we applied MVsim toward predictions of multivalent binding to motivate our protein designs (Figure A,B). Despite differences in the panning format, by modeling the dissociation kinetics of two ligand-target pairs with significantly different monovalent dissociation rate constants, we can gain insights into expected differences in the dissociation time scales between monovalent and bivalent presentation formats.…”

“…(C) Test selections of SLP76 using our bivalent RD system. Table shows published kinetic rate constants for WT and the single-alanine-substituted mutants P241A and L243A, , and the schematic shows bivalent and monovalent ligand-target binding, with binding-ablating mutations on the second peptide in the monovalent template to minimize other changes (such as translation efficiency or nonspecific binding) that could occur if the peptide length was changed. (D) Two binary libraries containing mRNA templates for WT and P241A (in bivalent or monovalent formats) at a ratio of 10:1 were subjected to panning against bivalent Gads-SH3C with 60 min wash and analyzed by qPCR.…”

“…(C) Test selections of SLP76 using our bivalent RD system. Table shows published kinetic rate constants for WT and the single-alanine-substituted mutants P241A and L243A, 14,15 and the schematic shows bivalent and monovalent ligand-target binding, with binding-ablating mutations on the second peptide in the monovalent template to minimize other changes (such as translation efficiency or nonspecific binding) that could occur if the peptide length was changed. (D) Two binary libraries containing mRNA templates for WT and P241A (in bivalent or monovalent formats) at a ratio of 10:1 were subjected to panning against bivalent Gads-SH3C with 60 min wash and analyzed by qPCR qualitative trends were expected to hold in experiments, differences in the experimental conditions compared with the MVsim simulations, especially the lower temperature required to maintain selection particle stability, likely decreased dissociation rate constants compared with the room-temperature values that were available in the literature to use in MVsim, so we chose to test commensurately longer washing times.…”

“…Finally, we performed proof-of-concept selections with mock libraries to demonstrate improved particle recovery in a bivalent RD system. Using previously characterized single alanine substitutions of SLP-76 with varying dissociation rates, , we created a panel of bivalent and monovalent templates to pan against bivalent Gads-SH3C targets in RD (Figure S2 and Table S1). We tested two of these mutants, P241A and L243A, at two different molar ratios compared with WT (1:1 and 10:1 WT-monovalent:Ala mutant) (Figures C–E and S4).…”

Analysis of emergent bivalent antibody binding identifies the molecular reach as a critical determinant of SARS-CoV-2 neutralisation potency

“…Experimental methods are routinely used to study monovalent kinetics by injecting soluble antigen over immobile antibodies (10) and more recently, methods have been developed to study spatial tolerance by measuring an apparent affinity for precisely spaced antigens (11, 12). Mathematical methods are also available in certain limits (7, 9, 19, 20) but a method that can directly fit binding of antibodies to a random distribution of anchored antigen at a given density is presently unavailable. Here, we developed a fast spatial and stochastic particle-based method to directly fit antibody binding to random antigens enabling estimates of monovalent on/off-rates, the bivalent on-rate, and the molecular reach.…”