Design of Novel Lectins by Computer‐Guided Directed Evolution

Abstract

Changes in the level and type of glycosylation are implicated in mediating countless physiological and disease processes, yet the glycome is poorly defined. To overcome this barrier, we integrate state of the art computational methods with guided directed evolution for the design of selective glycan binding reagents based on a natural lectin, cyanovirin.Cyanovirin‐N (CV‐N) is a small (11 kDa) lectin that displays potent antiviral activity by binding to the oligomannosides of gp120. The solution structure shows two quasi‐symmetric domains, defined as A (residues 1‐38/90‐101) and B (residues 39–89), which are connected on each side by a short helical linker. Each domain contains a carbohydrate binding pocket, which binds selectively Mana(1®2)Mana termini in branched high‐mannose oligosaccharides. We assessed the binding affinity of WT CV‐N through ITC titrations with dimannose revealing a high‐affinity binding site (Kd=15.3 mM) and a low‐affinity site (Kd=0.400 mM). To increase the binding affinity for glycosylated gp120, we exploited avidity effects by designing a dimeric CV‐N in which a CV‐N sequence is spliced into another. The construct contains four glycan binding domains in two sets (A, A′, B, B′), and binds dimannose with Kd=24.5 mM and 0.95 mM respectively. Four indistinguishable binding events, with Kd=1.1 mM, are observed with Man9.We recently developed computational tools to predicted protein‐glycan interactions accurately by using multiple sequence alignment (MSA) and flexibility analysis to identify co‐evolved positions, and coupling this analysis to our flexible docking tool (BP‐Dock). These tools were validated by assessing the energetic contribution of polar residues within the binding pocket, docking dimannose to single‐point CVN mutant models. We found that the E41A/G and T57A mutations led to a significant decrease in binding energy scores due to rearrangements of the hydrogen‐bond network that reverberated throughout the binding cavity. N42A decreased binding by affecting the integrity of the local protein structure. In contrast, N53S resulted in a high binding energy score. Experimental characterization of the five mutants by NMR and ITC confirmed the binding affinity pattern predicted computationally.Computational analysis also yielded a set of mutations at several positions beyond the immediate binding pocket that impact glycan recognition. These predictions were validated by directed evolution in yeast display; screening resulted in mutants with increased affinity and selectivity for dimannose. We are currently using the same library to identify mutants with high affinity and selectivity towards glycan biomarkers.Support or Funding InformationThis work is supported by NIH‐IR21AI129895 grant.This abstract is from the Experimental Biology 2018 Meeting. There is no full text article associated with this abstract published in The FASEB Journal.