Kcne1

Potassium voltage-gated channel subfamily E member 1 (Delayed rectifier potassium channel subunit IsK) (mISK) (IKs producing slow voltage-gated potassium channel subunit beta Mink) (Minimal potassium

Spatial tier: RESTRICTED · Class: ION_CHANNEL · Top region: VISli4 (Laterointermediate area, layer 4) · Positive regions: 28/554

47.2Discovery Score
33.9Targetability Score
CEvidence grade C
What this means: Kcne1: fine CCF profile is restricted toward VISli4 (top regions: VISli4, VISpor4, AUDpo4, VISam4, AUDp4; 28 positive regions); direct spatial validation is still required.

Function in VISli4: evidence, prediction, innovation

Gene function
Kcne1 encodes minK (IsK), a single-transmembrane beta subunit with no pore of its own. It binds pore-forming Kv alpha subunits and dramatically changes their behaviour: with KCNQ1 it generates the slow cardiac IKs current, and it also modulates Kv12.2, Kv4.3 and hERG. Human mutations cause long-QT syndrome and, with deafness, Jervell and Lange-Nielsen syndrome.
Region function
The laterointermediate area is a small higher visual area in mouse lateral extrastriate cortex; its layer 4 is the main thalamorecipient layer, receiving input from the lateral posterior nucleus and V1, and it contributes to object and motion processing in the ventral-stream-like pathway.
Published in this region
inferred No region-specific literature found. The most relevant neuronal result is Clancy 2009, who showed that KCNE1 and KCNE3 regulate surface expression of Kv12.2 channels and form a tripartite complex in vivo, providing the main evidence that minK has a genuine neuronal partner; otherwise KCNE1's established roles are cardiac (Bett 2006; Abbott 2016) and inner-ear, and the Kcne1-null mouse is known for vestibular and hearing defects (Piccini 1999).
Predicted function
Prediction: in layer 4 neurons of a higher visual area, KCNE1 would act as a tuning subunit for a subthreshold potassium current, most plausibly Kv12.2 or a KCNQ, slowing its activation and raising the threshold for repetitive firing, and thereby sharpening the gain and adaptation of thalamorecipient cells. Kcne1 deletion is predicted to increase intrinsic excitability and reduce spike-frequency adaptation in layer 4, broadening visual responses and weakening adaptation to sustained stimuli; the vestibular and hearing deficits of Kcne1-null mice mean conditional cortical deletion is the cleaner experiment.
Innovation
★★★★★ 5/5 A cardiac and inner-ear channel subunit concentrated in a specific higher visual area is an unexamined and easily testable claim, with knockout mice and layer 4 driver lines both available.
Tools
Kcne1 knockout mice (the classic shaker/waltzer inner-ear phenotype); IKs blockers chromanol 293B and HMR-1556 and the KCNQ1 activator ML277; anti-KCNE1 antibodies; layer 4 accessible with Scnn1a-Tg3-Cre (JAX 009613), Nr5a1-Cre (JAX 006364) and Rorb-IRES2-Cre (JAX 023526).
References

Direct evidence located at the region

No Allen ISH section could be located at this region's centroid for the recorded experiments.

Direct spatial evidence

MERFISH REGION LOCATOR
VISli4 MERFISH region locator
Real single-cell MERFISH coordinates for the VISli4 (Laterointermediate area, layer 4) section -- our own ground-truth data, showing exactly where this region sits, independent of whether the Allen ISH section below is easy to read.

Regional expression figure

Kcne1 regional expression profile

Predicted WMB × fine-CCF profile; not direct full-transcriptome spatial measurement.

Spatial profile

VISli4
0.138
VISpor4
0.138
AUDpo4
0.111
VISam4
0.106
AUDp4
0.105
AUDd4
0.0977
AUDv4
0.0974
VISpl4
0.0926
TEa4
0.0902
VISal4
0.0827
VISl4
0.0807
VISpm4
0.0759
VISrl4
0.0713
SSs4
0.0709
VISC4
0.0696
VISa4
0.067
PL2/3
0.0625
GU4
0.0569
ME
0.0522
AUDpo5
0.0483

Evidence and specificity

Evidence status
PREDICTED_FINE
Direct sources
None; predicted localization only
Exclusive threshold
12 positive regions out of 554
Spatial specificity
0.305
Top-region fraction
0.045
Filter status
PASS

Interpretation limits