Structural basis of human LRG1 recognition by Magacizumab, a humanized monoclonal antibody with therapeutic potential

Figures

• 

  Figure 1. Crystal packing and peptide arrangement. (a) The asymmetric unit consists of two peptide molecules (labeled Pep and Pep*; purple and pink, respectively) packed between two MagaFab molecules (MagaFab molecule 1 light chains labeled VH/VL; MagaFab molecule 2 light chains labeled VH*/VL*). A third neighboring VL chain (orange, not labeled) from a different asymmetric unit establishes weak interactions with one of the peptides. (b) Electrostatic surface showing polar pockets for the peptide interaction and important structural waters. Red indicates a negatively charged surface, while blue indicates positive charge.

• 

  Figure 2. Main MagaFab–peptide interactions. (a) The bidentate interaction between the amides of Gln5Pep and Gln50VH, as well as the salt bridges between Lys3Pep and Asp55VH and Asp57VH, are the key interactions that determine the recognition of the peptide by the Fab. Additional weaker interactions and those involving structural waters are also represented by yellow dashed lines. The VH chain is shown in green, the VL chain in orange, the main peptide in purple and the symmetric peptide in pink. (b) The same as (a) but from a different orientation. (c) Stabilization of the nonpolar section of the peptide. The short 310-helix of the peptide is flanked by Trp96VL, while Leu7Pep is located near Val103VH. The carbonyl group of Val6Pep is stabilized by a water located between the hydroxyl group of Ser95VL and the Tyr54VL and Thr102VH carbonyl groups. An interaction with the symmetric peptide is established between Asp10Pep and Asn2Pep*. (d) Stabilization of the peptide 310-helix moiety by symmetry molecules. Lys9Pep, Asp10Pep, Leu11Pep and Gln15Pep are stabilized by Ser206VL, Gly1Pep*, Asn2Pep* and Thr101VH from symmetry molecules (represented in light orange, light pink and light green,…

• 

  Figure 3. Structural alignment of murine hybridoma 15C4 homology models and the MagaFab crystal structure. (a) All residues interacting with the peptide and its surroundings are conserved. MagaFab is shown in orange. Murine 15C4 heavy and light variable domains are shown in green and blue, respectively. (b) The same as in (a) but rotated by 90°. (c) Potential flexibility of the VH loop. Murine 15C4 variable heavy domain (VH, green) shows a different conformation of the VH loop compared with the same loop in MagaFab (orange) that partially stabilizes the peptide. While the MagaFab variable light domain (VL, orange) contains a leucine at position 50, the murine homologue (blue) features a phenylalanine. These observations together could point to an important mutation location since the residue at position 50 could interact with the loop and alter its conformation, therefore modifying the affinity of MagaFab for the peptide and possibly for LRG1.

• 

  Figure 4. LRG1 models. (a) Superposition of a homology model of LRG1 (red, predicted with SWISS-MODEL) and NetrinG ligand 2 (NGL2; PDB entry 3zyi, chain A) used as the template. The arrow shows the location of the epitope recognized by MagaFab (black, LRR7). Leucine-rich repeats 1, 7 and 9 are labeled as a reference. (b) Superposition of the LRG1 homology model (red), the best threading model (blue) and the ab initio model predicted by AlphaFold (yellow). Leucine-rich repeats 1, 7 and 9 are labeled as a reference and the signal peptide is labeled SP.

• 

  Figure 5. MagaFab–LRG1 complex. (a) Superposition of the LRG1 model (green) with the crystal structure of the peptide (violet) and MagaFab (orange), taking the Cα atoms of Gln225 and Val226 as a reference, including a 90° side view. A flexible and exposed loop of MagaFab that is potentially involved in LRG1 recognition is labeled with a star. (b) The superposition of LRG1 LRR7 with the peptide of the crystal structure shows significant conformational differences in the main chain, probably not only due to the already described crystal contacts with the peptide but also due to inaccuracies in the LRG1 homology model.