Binding pattern analysis of (GPO)₅, (GPO)₃, and Toolkit-peptide III-30. (A) Graph showing the relative contribution to GPVI binding energy by collagen–peptide residues calculated using the FoldX SequenceDetail function as in the supplemental Methods.³⁶ Binding energy values were averaged for residues occupying equivalent positions in the 4 crystallographically resolved complexes; values for individual complexes are provided in supplemental Figure 7A-D; collagen residue numbering here conforms to Figure 2C. The inset table lists cumulative binding energies for different length sequence patterns and shown as percentage of the total energy. ^§Exists as Pro instead of Hyp in the second chain of the carboxy-terminal binding site of (GPO)₃. (B) Surface rendering of (GPO)₅ (top panel) and schematic representation of the chain stagger (bottom panel) displaying the calculated binding energy contributions per residue from panel A on a blue–white color scale. (C) Binding of WT GPVI-Fc (green) and control mutants R38A (red) and E40A (blue) to alanine-scanned variants of Toolkit peptide III-30 and control peptides III-30, CRP, (GPP)₁₀, and bovine serum albumin (BSA). For easy reference, the amino acid sequence of III-30 is shown below, highlighting the presence of an OGPOGP motif (orange) and two AGPOGP motifs (underlined). Data measured as A₄₅₀ is normalized to binding of WT GPVI to CRP. All data points represent the mean ± SD of at least three independent experiments. See also supplemental Figure 11. (D) Modeled binding of GPVI to the 9-AGPOGP-14 region of III-30. Relative to the (GPO)₅ and (GPO)₃ complexes, Ala instead of Hyp at position 9 disrupts interactions with GPVI Gln71 and Leu36 (left panel), whereas Glu instead of Hyp at position 15 just outside the motif enables the formation of an additional hydrogen bond to Gln82 (right panel). The predicted differences in interactions are in line with the divergent effect of GPVI mutations Gln71Ala, Leu36Ala, and Gln82Ala on binding to CRP and III-30 (see Figure 3A-C).