4-AP docking reframed: signal consistent with an axonal Kv1.1/Kv1.2 blocker, regeneration story not structurally supported
A correction to our earlier 4-AP framing
Two earlier internal framings of 4-aminopyridine were too broad. Additional DiffDock runs support a narrower, mechanism-consistent reading.
- Earlier framing (retracted): 4-AP as a 'multi-target recovery agent' engaging CORO1C, NCALD, SMN2, SMN1, UBA1 and driving motor-neuron regeneration. We no longer support this; CORO1C in particular is not a motor-neuron drug target.
- Earlier framing (retracted): 4-AP as a 'complementary recovery agent' acting through BDNF/TrkB/STAT3/PTEN/mTOR. The docking data do not support these as 4-AP binding partners.
- Current reading: the docking signal is consistent with 4-AP acting as an axonal Kv1.1/Kv1.2 blocker, i.e. the established dalfampridine/Ampyra mechanism. The structural data give no support to a regeneration mechanism.
All scores below are DiffDock rank-1 confidence values. They are computational ranking signals, not binding affinities or experimental results.
Regeneration target panel (5 targets)
BDNF (1BND), TrkB (4ASZ), STAT3 (1BG1), PTEN (1D5R), mTOR FRB (4DRI) all returned rank-1 confidences between -0.16 and -0.38. The slightly higher BDNF value (-2.96) is attributable to BDNF being a small homodimer with broad, promiscuous pockets. None of these is supported as a 4-AP binding partner.
Kv1 family panel
| Target | DiffDock conf. |
|---|---|
| Kv1.1 (6EBK) | -0.05 |
| Kv1.3 (3OC3) | -0.78 |
| Kv1.5 (7SIT) | -0.10 |
Taken with the earlier Kv1.2 result, the ranking is most favorable at the axonal Kv1.1/Kv1.2 pair, with modest separation from cardiac Kv1.5 and lymphoid Kv1.3. This is directionally consistent with the known Ampyra mechanism, though docking confidence values cannot establish subtype selectivity on their own.
Head-to-head vs 3,4-diaminopyridine (Firdapse)
| Target | 4-AP | 3,4-DAP |
|---|---|---|
| Kv1.1 | -0.05 | -0.17 |
| Kv1.3 | -0.78 | -1.17 |
| Kv1.5 | -0.10 | -0.30 |
| TrkB | -0.19 | -0.69 |
The two compounds rank differently across subtypes. We do not draw a clinical-preference conclusion from docking scores.
Why publish the correction
Narrowing an over-broad hypothesis back to what the structural data actually support is the same discipline as publishing a negative result. The corrected reading is a single-mechanism computational hypothesis (axonal Kv1.1/Kv1.2 block) and remains pending experimental validation.
Data: https://github.com/Bryzant-Labs/sma-research/tree/main/campaigns/4-AP/2026-04-10_updates/diffdock_extensions and Dropbox/SMA/open_data/4ap_extensions_2026-04-10/. CC-BY-4.0.