Computational design and engineering of self-assembling multivalent microproteins with therapeutic potential against SARS-CoV-2

Abstract

Multivalent drugs targeting homo-oligomeric viral surface proteins, such as the SARS-CoV-2 trimeric spike (S) protein, have the potential to elicit more potent and broad-spectrum therapeutic responses than monovalent drugs by synergistically engaging multiple binding sites on viral targets. However, rational design and engineering of nanoscale multivalent protein drugs are still lacking. Here, we developed a computational approach to engineer self-assembling trivalent microproteins that simultaneously bind to the three receptor binding domains (RBDs) of the S protein. This approach involves four steps: structure-guided linker design, molecular simulation evaluation of self-assembly, experimental validation of self-assembly state, and functional testing. Using this approach, we first designed trivalent constructs of the microprotein miniACE2 (MP) with different trimerization scaffolds and linkers, and found that one of the constructs (MP-5ff) showed high trimerization efficiency, good conformational homogeneity, and strong antiviral neutralizing activity. With its trimerization unit (5ff), we then engineered a trivalent nanobody (Tr67) that exhibited potent and broad neutralizing activity against the dominant Omicron variants, including XBB.1 and XBB.1.5. Cryo-EM complex structure confirmed that Tr67 stably binds to all three RBDs of the Omicron S protein in a synergistic form, locking them in the "3-RBD-up" conformation that could block human receptor (ACE2) binding and potentially facilitate immune clearance. Therefore, our approach provides an effective strategy for engineering potent protein drugs against SARS-CoV-2 and other deadly coronaviruses.

Keywords: Computational design; Cryo-EM; Microprotein; Nanobody; Protein therapeutics; SARS-CoV-2.

Fig. 1

Workflow for computational design of trivalent anti-SARS-CoV-2 microproteins. a Structure-guided computational design of trivalent microproteins to geometrically match the three binding sites of the trimeric S protein. The trimerization scaffold, linker, and monovalent binder are shown in blue, green, and gray, respectively. RosettaRemodel was used to design linkers connecting the C-terminus of the monovalent binder and the N-terminus of the trimerization scaffold. b Molecular dynamics (MD) evaluation of the trivalent constructs. Binding free energies of the monomers were estimated to assess trimerization tendency using the MM/GBSA method. Free energy landscapes were constructed to study the possible distributions of the trimer conformations. c Experimental verification of trivalent constructs using size-exclusion chromatography and Native-PAGE. d Functional tests of the top-ranked constructs. Binding affinity was measured by BLI, and pseudovirus neutralization assays were performed to determine the neutralizing activity against SARS-CoV-2

Fig. 2

Structure-guided linker design. a Two trimerization scaffolds used in this study. b The left panel: The trivalent construct designed to match the geometry of the three binding sites on RBDs (orange) of the S protein (silver). The monovalent binder and the trimerization scaffold are shown in blue and green, respectively. The right panel: The calculated minimum distances required between the binder and scaffold. c Schematic of linker models generated by RosettaRemodel. For each model, the distance from the C-terminus of the binder to the N-terminus of the scaffold was determined. d Distributions of the mentioned distances of the designed linkers of different lengths. The dotted black lines indicate the minimum distances required for the constructs

Fig. 3

Molecular simulation assessment of the trimerization tendency of trivalent constructs. a Schematic diagram for the binding process to calculate the binding free energy of a given trivalent construct. b Relative binding free energies of the trivalent constructs calculated by MM/GBSA. For F-scaffold trimers, MP-5rf was used as the reference to calculate the relative values of other constructs. For C-scaffold trimers, the reference is MP-3rc. Please see the MM/GBSA raw data in (Additional file 1: Table S2

Fig. 4

Free energy landscapes (FELs) for the MD conformations of the trivalent constructs. a FELs of conformational projections onto the first and the second principal components (PC1 and PC2). b FELs of conformational projections onto two alternative reaction coordinates: root mean square deviation (RMSD) and radius of gyration (Rg)

Fig. 5

Experimental verifications of the trivalent constructs of miniACE2. a SEC profiles of the purified proteins of the constructs. b Native-PAGE analysis of the protein trimer fractions isolated from SEC. c BLI measurements of the binding kinetics of the monovalent miniACE2 and the trivalent construct, MP-5ff, to the immobilized RBD of SARS-CoV-2 (Wuhan-Hu-1). Red traces represent the raw data and the kinetic fits are shown in gray. d Neutralizing activity of miniACE2 and MP-5ff against SARS-CoV-2 pseudovirus (Wuhan-Hu-1)

Fig. 6

Design and experimental characterization of Tr67. a Schematic diagram illustrating the trimerization of Nb67 nanobody fused with the optimal trimerization unit 5ff (see its amino-acid sequence of the fusion monomer of Nb67 with 5ff in (Additional file 1: Table S1). b SEC and Native-PAGE analysis of Tr67, showing high trimerization tendency and conformational homogeneity. c BLI measurement of the binding kinetics of the monovalent Nb67 and Tr67 to the immobilized RBD of SARS-CoV-2 pseudovirus (Omicron BA.1). d Neutralizing activities of Nb67 and Tr67 against SARS-CoV-2 pseudovirus (Omicron BA.1). Three independent experiments were performed

Fig. 7

Broad-spectrum neutralization potential of Tr67 against the dominant SARS-CoV-2 Omicron variants. a—i Neutralizing activities against SARS-CoV-2 pseudoviruses Omicron BA.2, BA.2.75, BA.2.12.1, BA.3, BA.5, BF.7, BQ.1.1, XBB.1, and XBB.1.5 variants, respectively. Three independent experiments were performed for each variant

Fig. 8

Cryo-EM structures of Tr67 in complex with the SARS-CoV-2 (Omicron BA.1) spike protein. a EM density map of Tr67-spike complex in “3-RBD-up” conformation at the overall resolution of 9 Å. b Fitting of the atomic model of the designed Tr67-spike complex into the EM density map. The EM density is shown as a transparent gray surface and the spike protein is rendered in blue. c Cross-section view of the stem region of the spike protein. d An open-like, “3-RBD-up” conformation with the SARS-CoV-2 (Wuhan-Hu-1) spike protein induced by three separated Nb67. e The binding interfaces of trivalent (left panel) and monovalent (right panel) Nb67 binding with the RBDs (shown as gray surface) are shown in pink, and the contact residues on Nb67 are shown as sticks