On the humanization of VHHs: Prospective case studies, experimental and computational characterization of structural determinants for functionality

Abstract

The humanization of camelid-derived variable domain heavy chain antibodies (VHHs) poses challenges including immunogenicity, stability, and potential reduction of affinity. Critical to this process are complementarity-determining regions (CDRs), Vernier and Hallmark residues, shaping the three-dimensional fold and influencing VHH structure and function. Additionally, the presence of non-canonical disulfide bonds further contributes to conformational stability and antigen binding. In this study, we systematically humanized two camelid-derived VHHs targeting the natural cytotoxicity receptor NKp30. Key structural positions in Vernier and Hallmark regions were exchanged with residues from the most similar human germline sequences. The resulting variants were characterized for binding affinities, yield, and purity. Structural binding modes were elucidated through crystal structure determination and AlphaFold2 predictions, providing insights into differences in binding affinity. Comparative structural and molecular dynamics characterizations of selected variants were performed to rationalize their functional properties and elucidate the role of specific sequence motifs in antigen binding. Furthermore, systematic analyses of next-generation sequencing (NGS) and Protein Data Bank (PDB) data was conducted, shedding light on the functional significance of Hallmark motifs and non-canonical disulfide bonds in VHHs in general. Overall, this study provides valuable insights into the structural determinants governing the functional properties of VHHs, offering a roadmap for their rational design, humanization, and optimization for therapeutic applications.

Keywords: Hallmark; NKp30; VHHs; antibody engineering; framework residues; humanization; molecular dynamics; natural killer cells; single domain antibody.

FIGURE 1

(a) VHH1 and (b) VHH2 sequence annotations and alignment to the most similar germline sequences. The first three rows indicate CDR1‐3 (cyan, magenta, brown) residues according to IMGT, Chothia and Kabat nomenclature in green. The fourth row shows the sequence of parental VHHs and the following four rows the alignment to the most similar germline sequences. Amino acid differences to the most similar germline sequence are indicated in gray. The last row shows residues in different key structural positions: Vernier positions (indicated in blue), Hallmark positions (orange), one position known to be essential for binding to preADAs (green) and non‐canonical cysteines (yellow). Sequence optimization: Designed variants towards increased human‐likeness in the framework regions. Mutations that have been introduced in key structural positions are colored accordingly. For straightforward sequence activity relationship (SAR) analysis, the KD values (Table 1) are added at the end of each designed sequence and are complemented with a green to red coloring based on the affinity reduction compared to the parental sequence.

FIGURE 2

Crystal structures and AlphaFold2 generated binding models of NKp30 with different ligands. (a) X‐ray structure of apo‐VHH1 (shown in red, PDB code 9FXF), superimposed in the AlphaFold2 generated VHH1‐NKp30 complex, reveals that the VHH1 apo structure matches the VHH1 conformation of the AlpaFold2 generated VHH1‐NKp30 complex. (b) Binding mode of the B7‐H6–NKp30 complex (PDB code 3PV6). (c) X‐ray structure of the VHH2‐NKp30 complex (PDB code 9FWW). (d) Superimposition of VHH1‐, VHH2‐, and B7‐H6‐binding modes against NKp30 reveals that the epitopes of B7‐H6 and VHH1 are overlapping, while VHH2 is binding to a distinct epitope.

FIGURE 3

(a) AlphaFold2 predicted VHH1‐NKp30 complex. (b) Visualization of VHH1 paratope residues (within 4.5 Å of NKp30). (c) Depiction of the non‐canonical disulfide bond that covalently links CDR3 to Cys55. (d) 3D‐alignment of AlphaFold2 predicted binding modes of parental VHH1‐v1.0 and humanized VHH1‐v1.9. (e) 3D‐alignment of parental VHH1‐v1.0 and fully framework‐humanized VHH1‐v1.10 indicates a possible steric clash of humanized Hallmark residue Trp52 with NKp30. Key residues that were subjected to humanization are indicated in the same color coding as in Figure 1.

FIGURE 4

Non‐canonical disulfide bridge substantially stabilizes the CDR3 loop in the binding competent conformation. (a) Free energy landscape of the CDR3 loop for VHH1‐v1.0 with and without disulfide bridge shows a substantial increase in conformational diversity. (b) Root mean square fluctuation (RMSF) for CDR3 residues for the MD simulations of VHH1‐v1.1 with and without the disulfide bridge between residues Cys55 and Cys107 (highlighted in yellow). Structural visualization of the conformational ensembles obtained from the clustering analysis using the same RMSD distance cut‐off criterion (2.5 Å) for VHH1‐v1.1 (c) with and (d) without disulfide bond between non‐canonical cysteines Cys55 and Cys107, aligned to the x‐ray structure of VHH1‐v1.0 (PDB code 9FXF), illustrating that the formation of the cysteine bridge stabilizes the binding competent CDR3 conformation.

FIGURE 5

(a) X‐ray structure of the VHH2‐NKp30 complex (PDB code 9FWW). (b) Visualization of VHH2 paratope residues (within 4.5 Å of NKp30). (c) Depiction of an intramolecular interaction between Hallmark residue Phe42 and CDR3 residue Tyr115 that seems essential for positioning CDR3 in its bioactive conformation. (d, e) visualize a network of hydrophobic residues between Val25, Ile39 and Val87 that seem to be essential to position CDR2 in the bioactive conformation. Key residues that were subjected to humanization are indicated in the same color coding as in Figure 1.

FIGURE 6

Structural and dynamic characterization of Hallmark mutations suggesting conformational entropy as critical determinant for antigen‐recognition. (a) Mutation F42V destabilizes the CDR3 loop, resulting in a bigger conformational space and a higher number of low populated states. The higher conformational diversity of the CDR3 loop is also reflected in the broader conformational ensemble. (b) In addition to directly stabilizing the CDR loops, the Hallmark residue mutation V87L reveals a conformational rearrangement of the CDR2 and CDR1 loops to accommodate the bulkier leucine sidechain. (c, d) Free energy surface of the paratope for VHH2‐v1.0 compared with VHH2‐v1.10 and VHH2‐v1.18. Both VHH2‐v1.10 and VHH2‐v1.18 are non‐binders and reveal a substantially increased conformational diversity reflected in a broader conformational ensemble, accompanied by a substantial population shift towards three equally lower populated states in contrast to one dominant state corresponding to the binding competent state for VHH2‐v1.0.

FIGURE 7

Sequence alignment of most prevalent V‐genes from llamas or alpacas. For comparison, the sequences of human IGHV3‐23*01 as generally recommended template for most VHH humanization campaigns (Sulea, 2022), is provided. Key residues are indicated in the same color coding as in Figure 1.

FIGURE 8

Visualization of (a) sequence and (b) structure diversity of VHHs as observed in NGS and PDB data. (a) Normalized Shannon entropy of residues along the VHH sequence (top row) and sequence logos of the entire VHH NGS data set (ALL) and for specific subsets, together with their percentages observed in the dataset. Positions colored in red indicate that the most prevalent residue in this position is different from the most prevalent residue in the entire dataset (ALL). (b) 3D‐alignment of PDB structures, based on the same subset definitions as for the NGS subsets. For clarity, CDR3 is hidden in the upper row and shown in the lower row. (c) CDR3 length distribution over the entire NGS data set and for the specific subsets. Key residues are indicated in the same color coding as in Figure 1. CDR and FR definitions are based on the IMGT nomenclature.

FIGURE 9

(a) NMI matrix based on normalized Shannon entropies along the VHH sequence (in IMGT numbering). (b) Rearranged NMI matrix, including the dendrogram associated with the hierarchical clustering and the assigned cluster IDs. Cluster 12 is indicated in the matrix. (c) NMI matrix for residues within Cluster 12 that indicate the dependencies between the pairs of amino acid positions. Residues of the classical Hallmark signature are indicated. (d) 3D‐alignment of six PDB structures (PDB codes 1BZQ, 1F2X, 1IEH, 1KXV, 1QD0, 1SJX) with different Hallmark signatures. Key residues are indicated in the same color coding as in Figure 1 and position 40 is indicated in yellow.