Recently, nanobody-based therapeutics have emerged as a highly effective strategy for COVID-19 treatment. However, camelid-derived nanobodies often require humanization engineering to reduce immunogenicity in clinical applications while simultaneously preserving their target-binding affinities. Here, we employed a computational and engineering approach to optimize the binding affinities of complementarity-determining region (CDR)-grafted humanized variants of the camelid-derived nanobody Nb2-67, which exhibits potent SARS-CoV-2 neutralization. By grafting the three CDR loops of Nb2-67 onto the humanized scaffold of the approved therapeutic nanobody Caplacizumab and refining the target-binding interface, we generated five nanobody variants with improved computational humanness scores. Three of these variants (Nb491, Nb273, and Nb1052) retained neutralizing activity. To further enhance their potency, we fused these variants to a self-assembling scaffold, generating three multivalent constructs with higher humanness scores. Pseudovirus assays showed that all the trivalent nanobodies exhibited picomolar neutralizing potency comparable to the original trivalent Nb2-67. Our study presents a novel computational and multivalent engineering strategy that effectively restores the antiviral efficacy of humanized CDR-grafted nanobody variants, offering a valuable approach for developing nanobody-based therapeutics against COVID-19 and other diseases.

MOTIVATION: Single-particle cryo-electron microscopy (cryo-EM) has become a powerful technique for determining 3D structures of biological macromolecules at near-atomic resolution. However, this approach requires picking huge numbers of macromolecular particle images from thousands of low-contrast, high-noisy electron micrographs. Although machine-learning methods were developed to get rid of this bottleneck, it still lacks universal methods that could automatically picking the noisy cryo-EM particles of various macromolecules. RESULTS: Here, we present a deep-learning segmentation model that employs fully convolutional networks trained with synthetic data of known 3D structures, called PARSED (PARticle SEgmentation Detector). Without using any experimental information, PARSED could automatically segment the cryo-EM particles in a whole micrograph at a time, enabling faster particle picking than previous template/feature-matching and particle-classification methods. Applications to six large public cryo-EM datasets clearly validated its universal ability to pick macromolecular particles of various sizes. Thus, our deep-learning method could break the particle-picking bottleneck in the single-particle analysis, and thereby accelerates the high-resolution structure determination by cryo-EM. AVAILABILITY AND IMPLEMENTATION: The PARSED package and user manual for noncommercial use are available as Supplementary Material (in the compressed file: parsed_v1.zip). SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.