AlphaFold says the structure does not change. AlphaMissense says the variant is pathogenic (0.9016). Both are correct. The paradox resolves when you look at the chemistry, not the geometry.
Stereocilin is an extracellular adhesion protein. It sits on top of OHC stereocilia and physically connects them to the tectorial membrane. The connection is maintained by surface charge, hydrogen bonds, and hydrophilic interactions. Glutamate (E) at position 1659 contributes all three. Alanine (A) contributes none.
This is a surface mutation on an adhesion protein. The fold is irrelevant. The glue is broken.
| Property | Wildtype (Glu) | Mutant (Ala) | Impact |
|---|---|---|---|
| Net charge (pH 7.4) | -1 (deprotonated carboxylate) | 0 (nonpolar methyl) | Lost negative charge at adhesion interface |
| H-bond acceptors | 2 (carboxylate oxygens) | 0 | Two H-bonds broken |
| Side chain volume | 138.4 ų | 88.6 ų | 49.8 ų cavity created |
| Hydrophobicity (K-D) | -3.5 (hydrophilic) | +1.8 (hydrophobic) | +5.3 shift: adhesion surface becomes water-repelling |
Summing individual contributions gives the total destabilization of the stereocilin-TM interaction:
Estimated binding affinity decrease: ~106×
For reference: 1.4 kcal/mol = 10× change in dissociation constant. 8.62 kcal/mol = catastrophic loss of adhesion. The protein folds correctly but cannot stick to the tectorial membrane.
AlphaFold pLDDT at position 1659: 95.69 (very high confidence, well-folded). AF3 overlay of wildtype and E1659A shows virtually identical backbone. This is expected: alanine is small, does not clash with neighbors, and does not disrupt the hydrophobic core.
AlphaMissense score: 0.9016 (likely pathogenic). AlphaMissense was trained on evolutionary conservation and protein fitness, not just structure. It learned that at this position, charge is conserved across mammals. Losing it breaks function.
The pathogenicity is in the chemistry, not the geometry. Identical structure, destroyed adhesion.
E1659A weakens but does not abolish stereocilin function. This explains Misha's moderate (not profound) hearing loss. Truncating variants (frameshifts, nonsense mutations) eliminate stereocilin entirely and cause profound deafness. Missense variants that disrupt surface interactions cause partial loss and moderate-to-severe hearing loss.
The weakened TM attachment means OHC stereocilia gradually decouple from the tectorial membrane. Cochlear amplification decreases. The loss may be progressive as microtrauma accumulates at the weakened interface.