Contributors: Chester
Structural validation
AlphaFold3-predicted GH10/Cex domain validated against the Cex crystal structure in complex with a xylobiose-derived imidazole inhibitor (PDB: 1FHD, Notenboom et al. 2000), via Cα superimposition in PyMOL.
Result: RMSD = 0.293 Å over 1,958 aligned atoms — near-identical backbone geometry, confirming the AlphaFold3 model faithfully represents the experimental fold.
This falls within the 0.3–0.4 Å range Notenboom et al. (2000) reported for Cα deviations between inhibitor-bound and wild-type Cex structures — i.e., the AlphaFold3 model differs from the crystal structure by no more than the crystal structure differs from itself upon ligand binding. This validates docking conducted on this model as structurally credible.
Active site detail
GLU127 (acid/base) and GLU233 (nucleophile) operate via a retaining double-displacement mechanism typical of GH10 enzymes:
- In 1FHD, the inhibitor’s exocyclic nitrogen forms a 2.6 Å hydrogen bond with GLU127
- Inhibitor C1 approaches GLU233 at 3.0–3.2 Å (Notenboom et al. 2000)
- GLN87 and TRP281 are steric barrier residues whose side chains must rearrange to accommodate the C-5 hydroxymethyl group of glucoside substrates (Notenboom et al. 1998)
GLN87 re-interpretation from our docking
In our cellobiose docking, GLN87 appears as an active hydrogen bond contributor rather than a displaced steric barrier — suggesting the pose captured the post-rearrangement conformation, where GLN87 has moved to accommodate cellobiose’s C-5 hydroxymethyl group and formed a new ligand interaction.
This is consistent with Notenboom et al. (1998)‘s finding that the Gln87Met mutation failed to improve glucoside specificity: methionine retained enough flexibility to rearrange similarly to glutamine, so the mutation carried a dual cost — losing both steric barrier removal and an active hydrogen bond. This dual cost may help explain GH10’s intrinsically lower beta-glucosidase activity relative to GH1.
[Attachment: GH10.cif — AlphaFold3 model used for alignment]