Overview of current KCNQ2-related

epilepsy research at BCM

Kristen Park, MD

Colorado Children’s

Univ. of Colorado Univ.

John Millichap, MD

Lurie Children’s

Northwestern Univ.

Ed Cooper, MD, PhD

Associate Professor

Neurology, Neuroscience,

Molecular & Human Genetics

BCM, Houston TX

NINDS,

AES

CURE,

Jack Pribaz

GSK

KCNQ2-related epilepsy: paradigm for

translating a well-established mechanism into

effective personalized epilepsy care

I. “Fill in” mechanistic understanding

II. “Reach out” across disciplines

III. “Fit in” to broader questions and medical

needs

“Filling in” mechanistic understanding

1. What are KCNQ2 (and KCNQ3)?

2. What do KCNQ2 mutations do?

3. Are there mechanism-based approaches to

therapy?

Li Li Mingxuan

Xu

Baouyen

TranZhigang Ji

Maarten Kole

What are KCNQ2 and KCNQ3?

1. Two of ~100 genes encoding potassium (K+)

channel subunits

2. The KCNQ2 and 3 names are used both for the

gene and the protein product, but Kv7.2 (3) are

alternative names for the protein products

3. KCNQ1 mutations cause the largest subset of

cases of inherited long QT. KCNQ4 mutations

causes a form of inherited childhood onset

deafness. KCNQ5 (no diseases known).

KCNQ subunits form “M-channels” Wang, Science (1998) 282,1890

• very slowly voltage-gated K channels, suppress repetitive firing

• Inhibited by Ach and many neurotransmitters (but enhanced by others)

• Inhibition causes increase in cell excitability Normal function: restrain firing, especially repetitive firing

musc

control

control Muscarine

current

clamp

voltage

clamp

conducta

nce

-40-50

Mod from Delmas

Brown NatRevNeuro 2005

KCNQ2 and KCNQ3: two subunit

compositions are important in neurons

Q2 homotetramers

Q2Q2

Q2Q2

Heterotetramers

Q3

Q2

Q2

Q3

Wang et al. Science

1998

Hadley et al. J

Neurosci 2003

Martire et al. J

Neurosci 2004

KCNQ2 and 2/3 channels have low “peak

open probability” in absence of therapeutic

drugs

when channel gates are maximally opened by strong

membrane depolarization, they are still closed most of the

time

single channel recording

Li and Shapiro

Journal of Neuroscience

2004

KCNQ2, KCNQ3, Na channels, and ankyrin-G are all

colocalized at axon initial segments (AIS) and nodes

Cortex layer 5 pyramidal cell

Q2 AnkG

pial

callosal

CA1 pyramidal cells

NaV NaV

sciatic nerve node

Pan et al.

2006

Battefeld et al.

2014

JNeurosci

KCNQ2/KCNQ3 at AISs are functional M-channels

Battefeld et al. 2014 JNeurosci

Bao

Tran

Houston

Maarten

Kole

Amsterdam

Initial segment and node of Ranvier KCNQ2/3

channels control the local RMP and AP firing

Maarten Kole

(Battefeld et al.

J Neurosci, April 2014)

wash

out

4 min

Action potentials normally activate only a

very small percentage of axonal KCNQ2/3

channels

1% maximal KCNQ

observed peak

conductance1 msec

Bao Tran and Maarten Kole

(Battefeld et al., 2014)

4% open at rest

-77 mV

AISnode of Ranvier

7% open after AP

Why do some KCNQ2 mutations cause mild,

but others cause severe disease?

Severe mutations are heterozygous and

missense–substitution of one aa in one of 2

KCNQ2 genes

to understand these, must think about how

channel primary sequence is assembled into a

3 dimensional structure (4 subunits x 872

aa/subunit)

“Macro”anatomy of a voltage-gated ion

channel

outside

inside

sensor

+

Resting potential: closed

gate

--

+

open

-70 mV -30 mV-

- -

++

- -

+

K+

81 published BFNS mutations: randomly

distributed, 2/3 likely no protein

Gene deletion 6

Small deletion 3

frameshift 20

no start codon 2

splice site 15

stop 7

missense 28 (35%)

total 81

The first few encephalopathy mutations suggested

a mechanism: 3 functional “Achilles’ heels”

Weckhuysen... Berkovic , Scheffer, de Jonghe, 2012; Millichap and Cooper, 2012

T274M

The first few encephalopathy mutations suggested

a mechanism: 3 functional “Achilles’ heels”

Millichap and Cooper 2012

ER

golgi

plasma

membrane

calmodulin

X

8/2014: 97 EE mutations: 96 missense (one

single aa deletion), 80 in 4 functional hot

spots

Millichap, Park et al., in preparation

Binomial distribution predicts the maximal

suppression by “channel-poisoning”

heterozygous mutant KCNQ2 subunits

15/16 or 92.5% have one mutant 75%

1 : 4 : 6 : 4 : 1 1 : 2 : 1

T274M introduces a bulkier side-chain near the pore

Threonine (T) medium sizeMethionine (M) Slightly bigger

T274M

Rationale for drug treatment (summary)

1. Due to low peak open probability and slow voltage-

dependent gating, neurons use only small fraction of

.3 to 3% of KCNQ2 and KCNQ2/3 maximal capacity

2. With a efficacious enough drug, should be able to

increase activity considerably, even if only 1/16 of the

channels are capable of responding

3. 8-fold difference between worst predicted mutant

suppression and BFNS, a mild transient condition.

Retigabine/ezogabine and ICA-069673

promote channel opening through effects at

distinct sites

ICA binding site

1:1 T274M/wt channels are responsive to 10 uM

retigabine/ezogabine, though less than wt alone

Li Li

Homomericeffect on

midpoint

voltage for

activation,

not on peak

conductance

Q2 wt

Q2+T274M 1:1

Bringing lab insights back to the clinic

1. What are KCNQ2 (and KCNQ3)?“natural anti-seizure function”

very slightly activated under physiological conditions

2. What do KCNQ2 mutations do?strongly suppress currents

some residual WT channels remain

3. Are there mechanism based approaches to

therapy?2 drugs potently and synergistically enhance currents

many other drugs in preclinical stages

no good data: pediatric safety, dosing, efficacy

The RIKEE project:

1. -Rational Intervention for KCNQ2 Epileptic

Encephalopathy: our efforts are focused on

developing and testing therapies

2. Completed: retrospective physician survey-

based study (23 patients/11 centers)

3. Just launched: IRB approved program of

prospective research including a patient

registry and website

Why a website, why a registry, and how are

the two different?

RIKEE Website

1. Information on all known

KCNQ2 and KCNQ3

variants

2. No personally

identifiable information

3. Extensive links to

scientific literature and

other resources

4. Curated by expert team

for use by all

5. “Complementary” model

KCNQ2 Registry

1. Information on patients

registered after informed

consent

2. At entry, permission to

recontact may be given

3. Clinical information

coded so can be

accessed for IRB

approved research but

personal information not

generally available

Building the registry and website: many

steps

1. Many documents require review and approval: protocol

summary, letters to MDs and patients, data entry forms,

advertisements, webpages...

2. In every case, family must initiate contact (MDs can

recommend to pt, not to us)

3. Informed consent requires direct conversation with our

team before enrollment

4. Procedures for receipt of information, encryption,

computer hardware/software

All designed to preserve patient autonomy, privacy

RIKEE clinical registry/website team

1. Cooper Lab

• Nishtha Joshi, MPH – coordinator

• Mia Cooper, BA – project intern

2. ICTR staff

• Alicia Brown, MPH, CCRP - regulatory

• Uma Ramamurthy, PhD - database

• Wren Pratt, BA, CCRC - coordination

www.rikee.org TOUR

www.rikee.org TOUR

www.rikee.org TOUR

www.rikee.org TOURAll published variants

Unpublished variants

if permission granted

www.rikee.org TOUR

Continues...

www.rikee.org TOUR

link

www.rikee.org TOUR

www.rikee.org TOUR

www.rikee.org TOUR

Reaching out: across disciplines & beyond

1. Clinical pediatric epileptologists neonatal

neurologists, neonatologists, others

2. Clinical Geneticists have a distinctive important

role

• Need for validated classification criteria

(ClinGen/NCBI ClinVar), rapid sequencing

test

3. Parents: network has rapidly globalized

4. Industry

5. Media

Finding the right industry partners

UV light: Ezogabine SF0034

SciFluor Bioscience

SF0034:

More potent than retigabine/

ezogabine in NINDS (Univ.

Utah) in vivo AED screenings

More potent in vitro

Reduced UV absorbance

reduced skin discoloration

risk?

Small company, interested in

KCNQ2 encephalopathy

indication

F

Ezogabine/

Retigabine

SF0034

FDA warning: Ezogabine

Linking up/Fitting in

KCNQ2-related

Epilepsy

Symptomatic

neonatal seizures

(Yogendra Raol,

Col Children’s)

Other “Severe Early

Life Epilepsies”--

Ohtahara, EMEGeneralized epilepsy

(Jackie French NYU,

Atul Maheshwari,

BCM)

Adult partial

epilepsy

Non-neonatal

syndromes w/

distinctive biologies

Summary

1. KCNQ2 encephalopathy is a new subtype of

neonatal-onset epilepsy with moderate to

profound global delay: de novo, missense

variants

2. First handful of mutations: dominant-negative

fitting mathematical/structural model

3. In vitro, drugs seem to work. In vivo, no

controlled studies but some promise

4. Infrastructure for longitudinal observational

studies and trials being built

5. Case number is still low for trials, requires

improved ascertainment

AcknowledgementsKCNQ2 families and supporters

BCM

Mingxuan Xu, Bao Tran, Li Li, Zhigang Ji

RIKEE Network

Lionel Carment, Universite de Montreal Marc Patterson, Mayo

Eric Marsh, Xilma Ortiz-Gonzalez, Emily Robbins Children’s Hospital of Philadelphia Bruria Ben Ze’ev Tel Aviv Molly Tracy Hasbro Children’s, Brown Univ.

Tammy Tsuchida, Phil Pearl (now BCH), Children’s National (DC)

John Millichap, Lurie Children’s (Northwestern U.)

Kristen Park, Paul Levisohn, Colorado Children’s (Aurora, CO)

Brenda Porter, Packard Children’s (Stanford)

Other Collaborations

UPENN

Zongming Pan Steve Scherer Amy Brooks-Kayal

Tingching Kao Steve Cranstoun Yogi Raol

Van Bennett (Duke) Jurgen Schwarz (Hamburg) Hugh Bostock (UCL) Ryuji Kaji(Tokushima) Yasushi Okamura, Atsuo Nishino (NIPS) Maarten Kole (Amsterdam, NIN) David Brown (UCL), Mala Shah (UCL)

Funding: NINDS, Miles Family Fund, AES/EF, Jack Pribaz Foundation, CURE, GSK