Three robust hits at pocket #1
The headline result
After all controls, three FDA-approved drugs show robust mutation-selective binding to PACS1-R203W at pocket #1:
Drug | Drug class | Δ (kcal/mol) | exhaust=8 | exhaust=16 | exhaust=32 | Reliability |
PHENYTOIN | Antiepileptic (hydantoin) | −0.61 | −0.58 | −0.58 | −0.61 | 85% |
CLONAZEPAM | AED (benzo, 7-NO₂/2′-Cl) | −0.53 | −0.53 | −0.53 | — | 75% |
OXAZEPAM | Anxiolytic (benzo, 7-Cl/3-OH) | −0.54 | −0.54 | −0.54 | — | 70% |
The true negative control: CARBAMAZEPINE (antiepileptic, dibenzazepine) gave Δ = +0.09 — no selectivity, confirmed.
Why this is a non-trivial finding
Among 19 FDA-approved antiepileptic drugs tested at pocket #1, only phenytoin and clonazepam show robust selectivity. Lamotrigine, levetiracetam, valproate, carbamazepine, oxcarbazepine, lacosamide, tiagabine, topiramate, gabapentin, pregabalin, phenobarbital, primidone, eslicarbazepine, felbamate, rufinamide, zonisamide, brivaracetam and clobazam all show Δ ≥ −0.20 kcal/mol — no R203W-selective binding.
This is the opposite of what would happen if pocket #1 binding were unspecific. Among 7 randomly-chosen FDA drugs at pocket #1, 0/7 show robust selectivity. The discrimination is genuine.
What kind of binder is each hit
- PHENYTOIN (5,5-diphenylhydantoin) — the founding hit. Δ increases (more selective) as docking exhaustiveness increases. FDA-approved since 1953. Already used empirically in many PACS1 patients.
- CLONAZEPAM (7-NO₂, 2′-Cl benzodiazepine) — a benzodiazepine antiepileptic. Δ stable across exhaustiveness. Also used empirically in pediatric epilepsy.
- OXAZEPAM (7-Cl, 3-OH benzodiazepine) — an anxiolytic benzodiazepine, not classically used as a first-line antiepileptic but with sedative/anxiolytic properties potentially useful in PACS1 behavioral symptoms.
The three drugs share a precise chemical pattern (see page 7).
What this does NOT tell us
- We do NOT claim phenytoin/clonazepam/oxazepam will be clinically effective in PACS1 — we claim they bind preferentially in our in silico model, which is a hypothesis to test.
- We do NOT claim the binding is via the canonical mechanism (Na⁺ channel block, GABAergic potentiation). The structural prediction is for an allosteric pocket distal from the mutation site.
- We do NOT rule out other drugs not in our 1,397-drug ChEMBL FDA subset.
Where the binding happens
Pocket #1 is a 9-residue cavity (K126, K127, E168, T169, E170, A229, K249, I250, Y251) located ~25–30 Å from the R203 mutation residue. The mutation R→W modulates this distant pocket via long-range conformational propagation — an allosteric effect. The chemistry of pocket #1 (three lysines basic, two glutamates acidic, one tyrosine aromatic) matches the pharmacophore of the three robust hits (page 7).