Functional significance of axonal Kv7 channels
in hippocampal pyramidal neurons
Mala M. Shah*†‡, Michele Migliore§, Ignacio Valencia¶, Edward C. Cooper¶, and David A. Brown*
*Department of Pharmacology, University College London, Gower Street, London WC1E 6BT, United Kingdom; †Department of Pharmacology, School of
Pharmacy, University of London, 29-39 Brunswick Square, London WC1N 1AX, United Kingdom; §Institute of Biophysics, National Research Council, via U La
Malfa, 153, 90146 Palermo, Italy; and ¶Department of Neurology, University of Pennsylvania, 3610 Hamilton Walk, Philadelphia, PA 19104
Communicated by Lily Y. Jan, University of California School of Medicine, San Francisco, CA, March 25, 2008 (received for review August 16, 2007)
axon initial segment 兩 CA1 pyramidal neurons 兩 M-current 兩 KCNQ channels
euronal Kv7 (KCNQ) channels form a noninactivating K⫹
current (also known as the M⫺ current); this turns on at
subthreshold potentials and regulates the excitability of a variety
of peripheral and central neurons (1–3). Recent immunohistochemical evidence has shown that the principal subunits forming
native M channels, Kv7.2 and Kv7.3 (3, 4), are concentrated at
the axon initial segment (AIS) and nodes of Ranvier of central
and peripheral principal neurons (5–9), where they colocalize
with Na⫹ channels. Like Na⫹ channels, they contain an ankyrin
G-binding motif that targets them to the AIS (5, 8). They are also
expressed at lower densities at the soma and possibly dendrites
and synaptic terminals (4, 6, 7, 10, 11).
Spontaneous mutations in Kv7 subunits cause epilepsy in humans
(2) and mice (12). The hippocampus is strongly implicated in
epilepsy (13) and accordingly, previous somatic recordings from
these neurons have indicated that the Kv7 current is involved in
determining several aspects of neuronal excitability, including the
resting membrane potential (RMP), spike frequency adaptation,
and burst suppression (e.g., refs. 14–16). However, the specific
contribution made by Kv7 channels in the AIS to these or other
manifestations of excitability has not been determined. This is
important, because some human epileptogenic mutations impair
axonal Kv7 subunit expression (7).
We have used selective pharmacological and molecular tools
coupled with computer modeling to understand the involvement
of axonal Kv7 channels in shaping neuronal intrinsic activity in
hippocampal CA1 neurons. We show that a high axonal Kv7
N
www.pnas.org兾cgi兾doi兾10.1073兾pnas.0802805105
channel concentration is essential for regulating action potential
threshold and RMP, thereby suppressing inherent spontaneous
activity of individual neurons. Further, we show that, although
there are few dendritic Kv7 channels, the enhanced numbers of
action potentials caused by perisomatic Kv7 channel inhibition
effectively back-propagated into distal dendrites, thereby increasing their activity. These results provide further insights into
the underlying mechanisms responsible for the physiological and
pathophysiological effects after Kv7 channel modulation.
Results
Kv7 Channel Inhibition Induces Spontaneous Action Potential Firing by
Modulating Intrinsic Membrane Properties. To assess the overall
contribution made by Kv7 channels to hippocampal intrinsic
excitability, we first suppressed all Kv7 currents in hippocampal
CA1 neurons by bath-applying the inhibitors, XE991, and linopirdine onto hippocampal slices. Because we wanted to determine how alterations in Kv7 channel activity specifically
affected intrinsic neuronal activity, we added glutamate and
GABA receptor blockers to the external solution (see Methods).
Under these conditions, the soma had a RMP of ⫺67.0 ⫾ 0.4 mV
(n ⫽ 39). Consistent with many previous studies involving
peripheral and central neuronal subtypes (e.g., refs. 14–18), 3
␮M XE991 and 10 ␮M linopirdine substantially reduced somatic
RMP by 8.7 ⫾ 0.9 mV (n ⫽ 9, P ⬍ 0.05, Fig. 1 A and B) and 9.2 ⫾
1.8 mV (n ⫽ 6, P ⬍ 0.05), respectively, and considerably
increased the number of action potentials produced by depolarizing steps [Fig. 1 A; supporting information (SI) Fig. S1]. These
effects occurred over a period of 20 min and did not reverse up
to 45 min after washout.
Surprisingly, most neurons exhibited spontaneous action potentials in the presence of XE991 (13 of 15 neurons; Fig. 1Bii)
or linopirdine (5 of 10 cells), respectively. The stronger effect of
XE991 probably reflects its greater potency as a Kv7 channel
inhibitor (IC50s for XE991 and linopirdine ⫽ 0.2–0.5 and 2.0–7.0
␮M, respectively; refs. 4 and 19). In most cells, spontaneous
firing was fairly regular, but in two neurons, bursting occurred.
The initial frequency ranged from 0.2 to 10 Hz, often increasing
to ⬎30 Hz with time (Fig. 1). Further, the spiking was suppressed
when the RMP was hyperpolarized to the original RMP, indicating that depolarization by Kv7 inhibitors contributed to the
generation of unprompted firing. Importantly, spontaneous spiking persisted in the absence of GABA blockers (n ⫽ 3, data not
shown), showing this was not because of reduced synaptic
inhibition.
To further check that this spontaneous activity was generated
Author contributions: M.M.S. and D.A.B. designed research; M.M.S., M.M., I.V., and E.C.C.
performed research; M.M.S. analyzed data; and M.M.S. and D.A.B. wrote the paper.
The authors declare no conflict of interest.
Freely available online through the PNAS open access option.
‡To
whom correspondence should be addressed. E-mail: mala.shah@pharmacy.ac.uk.
This article contains supporting information online at www.pnas.org/cgi/content/full/
0802805105/DCSupplemental.
© 2008 by The National Academy of Sciences of the USA
PNAS 兩 June 3, 2008 兩 vol. 105 兩 no. 22 兩 7869 –7874
NEUROSCIENCE
Members of the Kv7 family (Kv7.2–Kv7.5) generate a subthreshold
Kⴙ current, the Mⴚ current. This regulates the excitability of many
peripheral and central neurons. Recent evidence shows that Kv7.2
and Kv7.3 subunits are targeted to the axon initial segment of
hippocampal neurons by association with ankyrin G. Further,
spontaneous mutations in these subunits that impair axonal targeting cause human neonatal epilepsy. However, the precise
functional significance of their axonal location is unknown. Using
electrophysiological techniques together with a peptide that selectively disrupts axonal Kv7 targeting (ankyrin G-binding peptide,
or ABP) and other pharmacological tools, we show that axonal Kv7
channels are critically and uniquely required for determining the
inherent spontaneous firing of hippocampal CA1 pyramids, independently of alterations in synaptic activity. This action was
primarily because of modulation of action potential threshold and
resting membrane potential (RMP), amplified by control of intrinsic
axosomatic membrane properties. Computer simulations verified
these data when the axonal Kv7 density was three to five times
that at the soma. The increased firing caused by axosomatic Kv7
channel block backpropagated into distal dendrites affecting their
activity, despite these structures having fewer functional Kv7
channels. These results indicate that axonal Kv7 channels, by
controlling axonal RMP and action potential threshold, are fundamental for regulating the inherent firing properties of CA1
hippocampal neurons.
A
A Effects of ABP on cell excitability
5 min post-patch
min post patch
ii 25
iii ABP + XE991
i (Con)
(ABP)
Evoked Responses
Control
+XE991
20 mV
100 ms
20 mV
250 ms
-65 mV
-70 mV
-58 mV
-60 mV
-64.5 mV
Spontaneous activity
ii
Control
B Effects of ABP on neuronal innate activity
5 min post-patch
min post patch
iii ABP + XE991
i (Con)
ii 25
(ABP)
+XE991
20 mV
10 s
-65 mV
-58 mV
iii
+ XE991 + Cd
2+
iv
C
-58 mV
-58 mV
v
2+
+ XE991 - (Cd + TTX)
RMP (mV)
-70
vi
+ XE991 -ORMP
20 mV
5s
D
* *
*
n=10
-60
Con
ABP +
XE991(n=4)
16
n=10
-65
-55
-60 mV
-64.5 mV
-70 mV
2+
+XE991 + Cd + TTX
n=4
ABP ABP +
XE991
AP No.
B
i
8
* *
*
*
* * *
*
* * *
*
*
*
*
ABP
(DRMP,n=10)
ABP
(ORMP,n=5)
Con(n=10)
0
0 100 200 300
Current Injection (pA)
-65 mV
-58 mV
Fig. 1. Effects of inhibiting Kv7 channel activity on cell excitability All records
are somatic recordings from the same hippocampal CA1 pyramidal neuron. (A
i and ii) Examples of membrane voltage responses to current injections before
and after adding the Kv7 channel inhibitor, XE991 (3 ␮M). Four hundredmillisecond current steps from ⫺150 to ⫹150 pA were applied, as shown in the
schematic under the control trace. (B i and ii) Spontaneous activity at the cell’s
normal RMP recorded before and after application of XE991. (Biii) Addition of
the Ca2⫹ channel inhibitor, Cd2⫹ (200 ␮M), in the continued presence of XE991
did not prevent spontaneous firing but modified its rhythm. (Biv) Subsequent
addition of the Na⫹ channel blocker, tetrodotoxin (TTX) (after application of
XE991 and Cd2⫹) abolished action potentials but did not change resting
potential. (Bv) Spontaneous activity recovered after washout of TTX and Cd2⫹.
The records shown in B were all obtained continuously at a sampling frequency of 10 KHz. [Scale bars (Ai and Bi) apply to all traces within those
sections.]
postsynaptically, we abolished synaptic transmission within the
slice by adding the nonspecific Ca2⫹ channel blocker, Cd2⫹ (200
␮M), after XE991 application (Fig. 1Biii). XE991-induced spiking persisted in the Cd2⫹ solution (n ⫽ 5; Fig. 1Biii), with no
further change in RMP (n ⫽ 5, Fig. 1Biii). The main effect of
Cd2⫹ was to switch regular firing to bursting in all neurons tested
(e.g., Fig. 1B), probably through block of postsynaptic Ca2⫹
channels and Ca2⫹-activated conductances (20, 21). The persistent Na⫹ current (22) did not contribute to the depolarization
after Kv7 channel block, because the Na⫹ channel blocker,
tetrodotoxin (1 ␮M), did not alter RMP after XE991 (n ⫽ 5, Fig.
1Biv). Altogether, the above results show that axosomatic Kv7
channels are vital for preventing spontaneous firing of CA1
pyramidal neurons, and that this property resides in their control
of intrinsic membrane properties.
Ankyrin-Binding Inhibition Results in Spontaneous Action Potential
Firing. We next asked what the specific contribution of AIS Kv7
channels was to the suppression of spontaneous firing. For this,
we designed a peptide (referred to ankyrin G-binding peptide, or
ABP) that had an identical amino acid sequence (YIAEGESDTD) to the Kv7.2/Kv7.3 ankyrin G-binding motif, which plays
a significant role in axonal targeting (5, 7–9). A BLAST (National Institutes of Health Database) search using Swiss-Prot
also revealed that this binding motif is unique to Kv7.2/Kv7.3
7870 兩 www.pnas.org兾cgi兾doi兾10.1073兾pnas.0802805105
Fig. 2. Effects of selective disruption of axonal Kv7 channel function on
intrinsic cell excitability. (A) Effects of including ABP (YIAEGESDTD) in the
internal solution on RMP and action potential numbers measured by applying
400-ms current steps from ⫺150 to ⫹150 pA. The input– output relations at the
normal RMP at the beginning of the recording and after maximal ABP dialysis
are shown in i and ii, respectively. iii illustrates the effect of applying XE991
once the maximal effect of ABP is achieved. (B) Records showing the development of spontaneous activity with inclusion of ABP in the internal pipette
solution and subsequent effects of XE991 after the maximal effect of ABP. The
traces shown in B are obtained from the same cell as in A. [Scale bars (Ai and
Bi) apply to all other records within A and B, respectively.] (C) Mean effects of
ABP and ABP together with XE991 on RMP. (D) Average numbers of action
potentials (AP No.) generated during incremental 400-ms depolarizing current steps in the absence of ABP (control), when the RMP had depolarized
maximally in the presence of ABP only (DRMP), with the RMP restored to that
present at the beginning of the recording (ORMP), and in the presence of ABP
and XE991 at DRMP. *, P ⬍ 0.05.
subunits. Because expression of a construct containing this
peptide sequence has been shown to bind to ankyrin G (7), we
hypothesized that ABP might compete with Kv7.2/Kv7.3 subunits for binding to ankyrin G, thereby perhaps dissociating the
channels from the axonal membrane.
To determine the effects of disrupting axonal Kv7–ankyrin
binding in individual neurons without confounding changes in
network activity, we initially made whole-cell somatic recordings
in the presence of glutamatergic and GABAergic antagonists
with ABP included in the internal pipette solution. [Because the
AIS is ⬍50 ␮m from the cell body (23), changes in AIS
excitability can be monitored at the soma.] ABP delivered in this
manner diffused throughout the cell (Fig. S2). With 10 mM ABP,
the RMP initially remained steady at ⫺66.8 ⫾ 0.8 mV (n ⫽ 10)
for at least 5 min, after which the neuron gradually depolarized
to ⫺61.9 ⫾ 0.6 mV (n ⫽ 10, P ⬍ 0.01) ⬎20 min (Fig. 2 A and
C), reaching steady state thereafter. Following a parallel time
course, action potential firing in response to depolarizing current
steps also increased steadily (Fig. 2 A i and ii and D). This
persisted even when the RMP was restored to that seen when the
cell was first patched onto (Fig. 2D). Spontaneous spiking also
occurred in 7 of 10 neurons tested (Fig. 2C i and ii), albeit at a
much lower frequency (⬇0.01 Hz) than that observed when all
Kv7 channels were blocked with XE991 (Fig. 1Bii). Bath application of XE991 (3 ␮M) subsequent to ABP maximal effect
Shah et al.
20 mV
Capacitative
transients
ABP +
XE991 (red
trace)
Beginning of
recording
20 mV
Following ABP 1 ms
maximal effect
C
AP Threshold (mV)
5 ms
-45
-55
-65
5 min post- 25 min postpatch (ABP)
patch
-52
Time point when
ABP effects were
measured
-56
-60
-64
0
20 30
10
Time (min)
Fig. 3. Effects of reducing axonal Kv7 channel function on action potential
threshold. (A) Superimposed examples of single action potentials 5 (black
trace) and 25 min (blue trace) postpatching with ABP included in the patchpipette solution. Each action potential was generated by a 5-ms depolarizing
current injection from ⫺70 mV, as shown in the schematic. The magnitude of
the current injection was adjusted to ensure that the step produced only a
single action potential. The action potentials are also shown on an enhanced
time scale below. Also superimposed is a trace showing the effects of XE991
if applied after maximal effect of ABP on action potential threshold (red trace)
(B) Average (blue squares) action potential (AP) thresholds at the beginning
and 25 min postpatch with ABP in the internal pipette solution. Open black
squares represent the effects in individual neurons. (C) Graph showing the
time course of the alterations in ST with ABP. *, P ⬍ 0.05.
further depolarized the cells (Fig. 2 A and C) and accelerated the
spiking frequency to ⬎10 Hz (Fig. 2Biii). Concentrations of ABP
⬍10 mM also enhanced depolarization-induced action potential
firing but did not induce spontaneous spiking, presumably
because the amount was too low to disrupt enough binding of
Kv7 channels to ankyrin G. It was difficult to patch cells with ⬎10
mM peptide in the patch pipette.
The peptide targeted Kv7 channels, because ABP had no
effect on cell excitability if neurons were pretreated with 3 ␮M
XE991 (Fig. S3). It also appeared to have no effect on Na⫹
channels, because spike height and width were unchanged (see
below). Further, inclusion of a scrambled version of ABP (sABP,
10 mM; Fig. S3) in the internal pipette solution had no effect on
RMP or action potential firing.
Somatic Input Resistance Is Unaffected by ABP. Although ABP
depolarized the cell, this alone was not the cause of spontaneous
action potential firing, because in control neurons, a comparable
depolarization (⬇5 mV) had no such effect. Hence, another
membrane property must have changed. Because variations in
membrane resistance can also affect cell excitability (e.g., ref.
24), we assessed the effect of ABP on input resistance, RN. No
change in RN occurred with 10 mM ABP included in the pipette
solution, whereas subsequent addition of 3 ␮M XE991 produced
a substantial increase in RN (Fig. S4A), as in control cells with
no intrapipette ABP (Fig. S4B). In support of this, whole-cell
voltage–clamp recordings showed that the somatic Kv7 (M⫺)
current was unaffected after 30-min dialysis with ABP (%
inhibition ⫽ 9.2 ⫾ 5.9%, n ⫽ 3), measured by applying a step
potential from ⫺20 mV to ⫺50 mV (4). These results indicate
that ABP did not affect somatic Kv7 channels.
Axonal Kv7 Channels Critically Regulate the Action Potential Threshold. In central neurons, the AIS is the preferential site for action
potential generation (23), raising the possibility that axonal Kv7
channels may play a significant role in controlling action potential production or kinetics. We tested this by evoking single
spikes (Fig. 3A). With 10 mM ABP in the internal pipette
Shah et al.
solution, the spike threshold (ST) was lowered by 7.8 ⫾ 0.8 mV
(n ⫽ 10, P ⬍ 0.05, Fig. 3B) over a period of 25 min, remaining
steady thereafter (Fig. 3C). Further, a smaller current injection
was required to elicit a single action potential after ABP dialysis
than at the beginning of the recording (data not shown).
Subsequent application of XE991 then had no further effect (ST
difference before and after XE991 ⫽ 0.7 ⫾ 0.7 mV, n ⫽ 4; Fig.
3A). ABP had indeed targeted Kv7 channels, because it had no
effect if neurons were pretreated with 3 ␮M XE991 (ST
change ⫽ ⫺2.0 ⫾ 1.2 mV, n ⫽ 4). (Without ABP, XE991 reduced
ST by 3.4 ⫾ 0.7 mV, n ⫽ 6, P ⬍ 0.05.) The scrambled peptide,
sABP (10 mM), had no effect on ST (change over 45 min ⫽
⫺0.2 ⫾ 1.5 mV, n ⫽ 5). ABP, sABP, and XE991 had no effect
on action potential height or half width or the overall rate of rise,
indicating that Na⫹ and K⫹ channels involved in spike repolarization were unaffected. These results indicate that axonal Kv7
channels play a major role in determining action potential
threshold.
Immunohistochemical Kv7 Subunit Distribution Appears Unaltered by
ABP Treatment. In CA1, intense immunolabeling for ankyrin-G,
Kv7.2, and Kv7.3 subunits is detectable in pyramidal cell AIS (5,
6, 8). We assumed that the ABP disrupts ankyrin-G binding of
Kv7.2/Kv7.3 channels and wondered whether this might cause
subcellular redistribution of Kv7.2 and Kv7.3. We tested this
using a membrane-targeted palmitoylated congener of the peptide (25). This replicated the electrophysiological effects of ABP
delivered intracellularly, whereas a lipidated sABP (N-palmitoyl
sABP) did not (Fig. S5). High densities of Kv7.2 and Kv7.3 were
found in stained CA1 sections that were untreated or treated
with N-palmitoyl sABP (as observed; see refs. 5–6 and 9). The
density of these did not appear to be changed after treatment
with the N-palmitoyl ABP (Fig. S6).
Computer Modeling Verifies That Axonal Kv7 Channels Regulate
Spontaneous Firing. The above observations provide an indication
that the axonal location of Kv7 channels plays a crucial role in
suppressing spontaneous activity. To confirm this, we used
computer modeling (see Methods). Because the biophysical
properties of the Kv7 (or M⫺) current in hippocampal neurons
present in slices have not been clearly defined, we first obtained
an activation curve using voltage–clamp recordings (SI Text and
Fig. S7). These data were then input into the simulation program, NEURON, to obtain a ‘‘near-realistic’’ model of the CA1
pyramidal cell (see Methods). The basic conductance of Kv7
channels calculated from the voltage–clamp recordings was 0.02
S/cm2. Because there seem to be few functional Kv7 channels
present in dendrites (see below and ref. 26), we restricted Kv7
channels to the soma and axon. Further, although immunohistochemistry has indicated that Kv7 protein is three times denser
in the axon than the soma (7, 8), the relative number of
functional axonal Kv7 channels is unknown. We therefore
initially varied Kv7 channel density in axonal and somatic
compartments. Then, to determine their effects on cell excitability, Kv7 channels were selectively removed from either all or
selected compartments by setting their respective Kv conductances to zero. When doing this, we did not modify any other
ionic conductances. However, because inserting axonal Kv7
channels hyperpolarized the model neuron by 1–2 mV (depending on Kv7 channel density), we adjusted the initial RMP to ⫺65
mV to match the experimental data by slightly adjusting the
passive leak current.
With an axon:soma Kv7 conductance ratio of 1:1 or 2:1,
removal of all Kv7 conductance caused a long-lasting depolarization but not spontaneous action potential firing (Fig. 4Ai),
although it did increase the number of action potentials in
response to a step depolarization, as expected (Fig. S8). If,
however, the axonal conductance was set between three and five
PNAS 兩 June 3, 2008 兩 vol. 105 兩 no. 22 兩 7871
NEUROSCIENCE
B
AP Threshold (mV)
A
A
Total cellular Kv7
conductace deletion
from -65 mV
i
Kv7 axon to soma conductance
ratio (a:s) = 2:1
B
Axonal Kv7 conductance elimination from
-65 mV
i
a:s = 2:1
ii
a:s = 3:1
A
Control (-69 mV)
+ XE991 (DRMP,
-65 mV)
250 pA
ii
a:s = 3:1
+XE991 (ORMP,
-69 mV)
Spontaneous
spikes
-300 pA
-60
-50
iv
a:s = 6:1
iv
a:s = 6:1
20 mV
200 ms
point of removal of Kv7 channels
Fig. 4. Relation between the subcellular distribution of Kv7 channels and
spontaneous intrinsic excitability. Computer simulations with the Kv7 channel
conductance in the axon and soma varied. The RMP in all cases was set to be
⫺65 mV. A and B illustrate the changes in unprompted cell firing and excitability if the Kv7 axon to soma (a:s) conductance ratio is set to 2:1, 3:1, 4:1, and
6:1 and Kv7 channel conductance is then either removed from both the axon
and the soma (A) or is deleted from the axon only (B). [Scale bars (Ai) apply to
all traces.]
times that at the soma, then taking away all Kv7 conductance
generated unprompted firing (Fig. 4A ii and iii) in a similar
manner to that caused by applying XE991 (Fig. 1). Interestingly,
if axonal Kv7 conductance was six times or more at the soma,
eliminating all Kv7 conductance produced limited firing (Fig.
4Aiv). This is because Kv7 channel deletion then depolarized the
RMP to values above ⫺50 mV, thereby inactivating Na⫹ channels and abbreviating spiking. Taken together with our experimental data, the model predicts that, for sustained spontaneous
firing to occur after total Kv7 channel blockade (see Fig. 1), the
axonal Kv7 channel density must be between three and five times
greater than at the soma.
To test how axonal Kv7 channels specifically contribute to
spontaneous neuronal activity, we selectively removed the axonal Kv7 channel conductance, leaving the somatic Kv7 channels
intact (thus imitating ABP). The results forecast that a minimum
Kv7 axon:soma conductance ratio of 3:1 is required for spontaneous action potential firing to occur (Fig. 4B i–iii). Again, if the
initial axonal density was six times (or more) than in the soma,
then removal of the axonal channels produced limited action
potential firing for reasons stated above (Fig. 4Biv).
Irrespective of the ratio of axonal Kv7 channels to soma,
selective removal of somatic Kv7 channels did not cause spontaneous firing (data not shown). Thus, the presence of axonal
Kv7 channels is both necessary and sufficient to prevent spontaneous firing.
Computer Modeling Confirms That Axonal Kv7 Channels Control
Action Potential Threshold. Experimental results demonstrated
that changes in action potential threshold (in addition to RMP
depolarization) were required for spontaneous action potential
7872 兩 www.pnas.org兾cgi兾doi兾10.1073兾pnas.0802805105
**
*
*
* * **
A P No .
iii
RMP (mV)
iii
a:s = 4:1
Con
+XE
Control
XE991 (DRMP)
XE991 (ORMP)
0
0 100 200 300
Current Injection (pA)
D
60
(M Ohms)
20
-70
a:s = 4:1
C
n=8
n=8
30
N
p < 0.05,
n=6
R
B
0
Con
+XE
Fig. 5. Effects of XE991 on dendritic activity (A) Recordings from distal CA1
pyramidal dendrites (250 –300 ␮m from the soma) under control conditions
and in the presence of XE991 when the RMP was depolarized (DRMP) or
restored artificially back to that present under control conditions (original
RMP, or ORMP). The traces were generated by applying a series of 400-ms
current steps from ⫺300 to ⫹250 pA, as shown in the schematic. The vertical
and horizontal scale bars correspond to 20 mV and 100 ms, respectively, and
apply to all traces shown. (B–D) Graphs depicting RMP values, action potential
numbers (AP No.) in response to depolarizing current steps, and RN values
before and after application of XE991. RN was calculated as described in
Methods. *, P ⬍ 0.05.
firing to occur after disruption of axonal Kv7 ankyrin binding
(see Figs. 2 and 3). We therefore used computer modeling to
examine how axonal Kv7 channels affected the action potential
threshold, simulating the experimental protocol in Fig. 3A.
Starting with axonal Kv7 conductances two and three times those
in the soma, selective removal of these conductances reduced
action potential threshold by 6.8 and 12.7 mV, respectively.
These results are in reasonable agreement with the change
produced by disrupting axonal Kv7 channel activity with ABP.
Distal Dendrites Have Few Functional Kv7 Channels. The above
evidence points to a critical role for axonal Kv7 channels.
However, Kv7 channels may also be present in dendrites (15, 27,
28). To determine whether this is so, we obtained recordings
from distal dendrites (250–300 ␮m from the soma) and then
bath-applied XE991 (3 ␮M) and linopirdine (10 ␮M). These two
drugs reduced dendritic RMP by 4.7 ⫾ 1.0 mV (n ⫽ 6, Fig. 5 A
and B) and 5.0 ⫾ 0.9 mV (n ⫽ 4), a lesser effect than at the soma
(see above) and, in contrast to the soma (Figs. S1 and S4), did
not affect dendritic RN (Fig. 5D). This implies that Kv7 channels
were not likely to be present in distal dendrites (in keeping with
recent data; see ref. 26).
Nevertheless, the numbers of action potentials recorded in
distal dendrites were significantly enhanced in the presence of
XE991 and linopirdine, like at the soma (Fig. 5 A and C) and
spontaneous spikes were noted in 50% of neurons (see Fig. 5A).
In CA1 pyramidal neurons, most dendritic action potentials are
those that have been initiated at the axon and back-propagated
(29). Thus, the increased numbers of dendritic action potentials
are likely to result from the enhanced axosomatic excitability. In
support of this, the spontaneous activity and enhanced evoked
action potential firing in the presence of XE991 and linopirdine
persisted even when dendritic RMP was artificially restored to
the predrug level (Fig. 5). These results indicate that, despite few
Shah et al.
Discussion
The principal conclusion arising from this study is that the
function of axonal Kv7 channels in hippocampal CA1 pyramidal
neurons is to suppress the inherent spontaneous activity of these
neurons, and that they do this by regulating action potential
threshold and RMP (Figs. 2 and 3). Further, we found that
computer simulations could reproduce our experimental observations if the axonal Kv7 conductance was approximately three
to five times greater than that at the soma (Fig. 4). This accords
well with information from immunohistochemical data (7, 8).
Selectivity and Mechanism of Action of ABP. Our principal experi-
mental evidence for a specific role of axonal Kv7 channels stems
from the use of a peptide (ABP) designed to selectively disrupt
the association between Kv7 subunits and ankyrin G at the AIS.
This peptide affected only Kv7 channels, because its actions were
fully occluded by blocking all Kv7 channels with XE991; and it
affected only axonal, not somatic, channels, because (unlike
XE991) it did not affect somatic input resistance or M current
amplitude. Interestingly, intracellular ABP did not affect Na⫹
channel function (at least within our experimental time frame),
as judged from action potential properties, even though Na⫹
channels have a partly homologous ankyrin-binding motif (5, 9).
This may reflect either the specificity of the peptide or tighter
Na⫹-channel ankyrin binding [perhaps related to the different
locations of the ankyrin-binding motif within the NaV1 and Kv7
channel proteins (9)].
How might ABP be working? Because mutant Kv7 channels
that do not bind ankyrin G are not concentrated at the AIS (5,
8), one possibility is that, by disrupting ankyrin-binding, axonal
Kv7 channels may redistribute after ABP treatment. However,
immunohistochemistry did not reveal any obvious effect of a
lipidated version of the ABP (which had the same functional
effect as the intracellularly injected ABP) on Kv7.2 and Kv7.3
localization (Fig. S6). This may be because diffusion of all
molecules in the AIS membrane is highly restricted, whether
bound to ankyrin or not, because of constraints imposed by the
entire actin-linked cytoskeletal apparatus (30). Alternatively,
ABP might have induced a local reorganization such as redistribution of Kv7 channels from membrane to cytosol. We could
not test this, because all available Kv7 antibodies are directed
toward intracellular domains and so cannot distinguish between
cytoplasmic and membrane protein [the small diameter of the
AIS (⬍1 ␮m) also makes it very difficult to differentiate between
cytoplasmic and membrane fluorescence even with confocal
microscopy]. Nevertheless, the electrophysiological results
clearly indicate that ABP affected axonal Kv7 channels.
Functional Significance of Axonal Kv7 Channels. Kv7 channels are
modulated by a variety of G protein-coupled receptors and
intracellular signaling molecules (1). Our results indicate that
modulating axonal Kv7 channel function can have a significant
impact on neuronal processing under physiological conditions.
For example, Kv7 channel activity is inhibited by acetylcholine
(17), which, in the hippocampus, is probably mediated by the M3
receptor (31). Coincidentally, the M3 receptor is present at the
CA1 neuropil (32). The increased action potential firing and
enhanced back propagation of spikes into dendrites by transient
Kv7 channel inhibition (Fig. 5) will lead to more NMDA
receptor channel openings (because the Mg2⫹ block is removed
by depolarization) and enhanced incoming excitatory synaptic
inputs and, perhaps, synaptic integration (29). This may represent one way by which neurotransmitters such as acetylcholine
contribute to processes such as learning and memory (33).
Shah et al.
Our experiments also provide an explanation for why spontaneous mutations in Kv7 channels that disrupt axonal Kv7
targeting can result in epilepsy (7). In vivo, any spontaneous
firing resulting from axonal channel disruption is likely to be
enhanced by the more global Kv7 inhibition arising from activation of Gq-coupled receptors via afferent glutamatergic and
cholinergic synaptic activity (3), similar to the amplifying effect
of somatic Kv7 block by XE991 (Fig. 2), thereby leading to
convulsions. Thus, a high density of AIS Kv7 channels is
fundamental for regulating hippocampal pyramidal inherent
activity and maintaining neural network homeostasis.
Methods
Electrophysiological Studies. Methods were similar to those described in ref.
24. Briefly, 5- to 6-wk-old rats were anesthetized and perfused intracardially
with ice-cold modified artificial cerebrospinal fluid containing 110 mM choline chloride, 2.5 mM KCl, 1.25 mM NaH2PO4, 25 mM NaHCO3, 0.5 mM CaCl2,
7 mM MgCl2, and 7 mM glucose, and bubbled with 95% O2/5% CO2 (pH 7.2).
Four hundred-micrometer-thick hippocampal– entorhinal slices were then
prepared by using a vibratome (Leica VT1000S) and incubated for 1 h at room
temperature in a holding chamber containing an external recording solution:
125 mM NaCl, 2.5 mM KCl, 1.25 mM NaH2PO4, 25 mM NaHCO3, 2 mM CaCl2, 2
mM MgCl2, and 10 mM glucose and bubbled with 95% O2/5% CO2 (pH 7.2). For
recording purposes, slices were placed in a chamber with external recording
solution supplemented with 0.05 mM 2-amino-5-phosphonovaleric acid
(APV), 0.01 mM 6-cyano-7-nitroquinoxaline-2,3-dione, 0.01 mM bicuculline,
and 0.001 mM CGP 55845 and maintained at 34 –36°C. For some recordings,
the external solution also had 0.01 mM tetrodotoxin and 0.2 mM Cd2⫹. The
internal pipette solution was 120 mM KMeSO4, 20 mM KCl, 10 mM Hepes, 2
mM MgCl2, 0.2 mM EGTA, 4 mM Na2-ATP, 0.3 mM Tris䡠GTP, and 14 mM
Tris-phosphocreatine, pH 7.3 (adjusted with KOH; pipette resistance, 8 –9 M⍀).
Whole-cell current– clamp recordings were obtained by using a bridge-mode
amplifier (Axoclamp 2A), filtered at 10 kHz and sampled at 30 kHz. Series
resistances were in the order of 10 –20 M⍀. Voltage– clamp recordings were
obtained by using an Axopatch 2A amplifier, filtered at 1 kHz and sampled at
5 kHz. All recordings were acquired by using pClamp (Axon Instruments) 8.0
and stored on a computer for further analysis.
Data Analysis. All measurements were made by using Clampfit (v8.0). Because
Kv7 channels are most active at voltages positive to ⫺70 mV (Fig. S7), we
measured RN by applying a 400-ms 100-pA step from ⫺70 mV. To calculate RN,
the steady-state voltage in the last 25 ms of the step was subtracted from ⫺70
mV and divided by the magnitude of current applied. The action potential
threshold was measured by differentiating the action potential voltage with
respect to time (dV/dt). Threshold was defined as the point where dV/dt was
⬎0. Action potential height was measured from dV/dt to the peak whereas
spike width was defined as half the height. Group data are expressed as
mean ⫾ SEM. Statistical significance was determined by using either paired or
unpaired Student’s t tests, as appropriate.
Materials. All chemicals were obtained from Sigma, except for XE991,
linopirdine, CNQX, CGP 55848, bicuculline, and APV, which were purchased
from Tocris. ABP, sABP, and analogues were custom-made either by the
Advanced Biotechnology Center (Imperial College London) or Genscript
and dissolved at the relevant concentration in either the internal or
external solution as appropriate. Stock solutions of bicuculline and CGP
55848 were made in DMSO (final dilution, ⬍0.1%) and stored at ⫺20°C
until use. Aqueous stock solutions of XE991, linopirdine, CNQX, and APV
were also kept at ⫺20°C until use.
Immunohistochemistry. Three hundred fifty- to four hundred-micrometer
slices were treated with 50 ␮M palmitoylated ABP or palmitoylated sABP and
then fixed by using 4% paraformaldehyde for 30 – 45 min. The slices were then
equilibrated in 30% sucrose, embedded (OTC Tissue Tek Compound, Sakura
Finetek), cryostat sectioned at 20 ␮m, and mounted onto Superfrost Plus
(Fisher) slides. Because antigen retrieval was necessary, the sections were
microwave-irradiated in 10 mM Na-Citrate, pH 6.0. Immunoreactions were
then performed by using rabbit Kv7.2-N and guinea pig Kv7.3-N antibodies, as
described (5, 8).
Computational Modeling Methods. All simulations were implemented and run
with NEURON (v5.8) by using the variable time step feature. The 3D reconstruction of the CA1 pyramidal neuron was one of those used previously (neuron
PNAS 兩 June 3, 2008 兩 vol. 105 兩 no. 22 兩 7873
NEUROSCIENCE
functional dendritic Kv7 channels, alterations in firing after
perisomatic Kv7 channels modulation are likely to affect dendritic excitability.
9068802; see ref. 34). Unless otherwise noted, the same standard uniform passive
properties were used for control conditions (␶m ⫽ 28 ms, Rm ⫽ 28k⍀䡠cm2, Ra ⫽
150⍀䡠cm). RMP was set at ⫺65 mV and temperature at 35°C. Active somatic and
dendritic properties included Na⫹, DR-, A-, and Kv7-type potassium conductances, and Ih current. To take into account Ca2⫹ channels opened at rest, a
low-threshold Ca2⫹ conductance, a Ca-dependent K⫹ conductance, and a simple
Ca2⫹ extrusion mechanism were included at uniform density and distribution in
all compartments. Channel kinetics and distribution were based on the available
experimental data for CA1 pyramidal neurons (35). The model and simulation
files are available for public download under the ModelDB section (36) of the
SenseLab database (http://senselab.med.yale.edu). The Kv7 activation was mod-
eled with a Boltzmann curve (v1/2 ⫽ ⫺40, k ⫽ 10). The voltage dependence of the
time constant was obtained from the experimental data (SI Text and Fig. S7) by
using identical activation and deactivation rate constants at V0.
1. Delmas P, Brown DA (2005) Pathways modulating neural KCNQ/M (Kv7) potassium
channels. Nat Rev Neurosci 6:850 – 862.
2. Jentsch TJ (2000) Neuronal KCNQ potassium channels: physiology and role in disease.
Nat Rev Neurosci 1:21–30.
3. Wang HS, et al. (1998) KCNQ2 and KCNQ3 potassium channel subunits: Molecular
correlates of the M-channel. Science 282:1890 –1893.
4. Shah MM, Mistry M, Marsh SJ, Brown DA, Delmas P (2002) Molecular correlates of the
M-current in cultured rat hippocampal neurons. J Physiol 544:29 –37.
5. Pan Z, et al. (2006) A common ankyrin-G-based mechanism retains KCNQ and NaV
channels at electrically active domains of the axon. J Neurosci 26:2599 –2613.
6. Devaux JJ, Kleopa KA, Cooper EC, Scherer SS (2004) KCNQ2 is a nodal K⫹ channel.
J Neurosci 24:1236 –1244.
7. Chung HJ, Jan YN, Jan LY (2006) Polarized axonal surface expression of neuronal KCNQ
channels is mediated by multiple signals in the KCNQ2 and KCNQ3 C-terminal domains.
Proc Natl Acad Sci USA 103:8870 – 8875.
8. Rasmussen HB, et al. (2007) Requirement of subunit co-assembly and ankyrin-G for
M-channel localization at the axon initial segment. J Cell Sci 120:953–963.
9. Lai HC, Jan LY (2006) The distribution and targeting of neuronal voltage-gated ion
channels. Nat Rev Neurosci 7:548 –562.
10. Martire M, et al. (2004) M channels containing KCNQ2 subunits modulate norepinephrine, aspartate, and GABA release from hippocampal nerve terminals. J Neurosci
24:592–597.
11. Roche JP, et al. (2002) Antibodies and a cysteine-modifying reagent show correspondence of M current in neurons to KCNQ2 and KCNQ3 K⫹ channels. Br J Pharmacol
137:1173–1186.
13. Schwartzkroin PA (1994) Role of the hippocampus in epilepsy. Hippocampus 4:239 –
242.
14. Gu N, Vervaeke K, Hu H, Storm JF (2005) Kv7/KCNQ/M and HCN/h, but not KCa2/SK
channels, contribute to the somatic medium after-hyperpolarization and excitability
control in CA1 hippocampal pyramidal cells. J Physiol 566:689 –715.
15. Lawrence JJ, et al. (2006) Somatodendritic Kv7/KCNQ/M channels control interspike
interval in hippocampal interneurons. J Neurosci 26:12325–12338.
16. Yue C, Yaari Y (2004) KCNQ/M channels control spike afterdepolarization and burst
generation in hippocampal neurons. J Neurosci 24:4614 – 4624.
17. Brown DA, Adams PR (1980) Muscarinic suppression of a novel voltage-sensitive K⫹
current in a vertebrate neurone. Nature 283:673– 676.
18. Otto JF, Yang Y, Frankel WN, White HS, Wilcox KS (2006) A spontaneous mutation
involving Kcnq2 (Kv7.2) reduces M-current density and spike frequency adaptation in
mouse CA1 neurons. J Neurosci 26:2053–2059.
19. Zaczek R, et al. (1998) Two new potent neurotransmitter release enhancers, 10,10bis(4-pyridinylmethyl)-9(10H)-anthracenone and 10,10-bis(2-fluoro-4-pyridinylmethyl)-9(10H)-anthracenone: comparison to linopirdine. J Pharmacol Exp Ther
285:724 –730.
20. Haas HL, Jefferys JG (1984) Low-calcium field burst discharges of CA1 pyramidal
neurones in rat hippocampal slices. J Physiol 354:185–201.
21. Hotson JR, Prince DA (1980) A calcium-activated hyperpolarization follows repetitive
firing in hippocampal neurons. J Neurophysiol 43:409 – 419.
22. Yue C, Remy S, Su H, Beck H, Yaari Y (2005) Proximal persistent Na⫹ channels drive spike
afterdepolarizations and associated bursting in adult CA1 pyramidal cells. J Neurosci
25:9704 –9720.
23. Bean BP (2007) The action potential in mammalian central neurons. Nat Rev Neurosci
8:451– 465.
24. Shah MM, Anderson AE, Leung V, Lin X, Johnston D (2004) Seizure-induced plasticity
of h channels in entorhinal cortical layer III pyramidal neurons. Neuron 44:495–508.
25. Robbins J, Marsh SJ, Brown DA (2006) Probing the regulation of M (Kv7) potassium
channels in intact neurons with membrane-targeted peptides. J Neurosci 26:7950 –
7961.
26. Hu H, Vervaeke K, Storm JF (2007) M-channels (Kv7/KCNQ channels) that regulate
synaptic integration, excitability, and spike pattern of CA1 pyramidal cells are located
in the perisomatic region. J Neurosci 27:1853–1867.
27. Chen X, Johnston D (2004) Properties of single voltage-dependent K⫹ channels in
dendrites of CA1 pyramidal neurones of rat hippocampus. J Physiol 559:187–203.
28. Yue C, Yaari Y (2006) Axo-somatic and apical dendritic Kv7/M channels differentially
regulate the intrinsic excitability of adult rat CA1 pyramidal cells. J Neurophysiol
95:3480 –3495.
29. Johnston D, Hoffman DA, Colbert CM, Magee JC (1999) Regulation of backpropagating action potentials in hippocampal neurons. Curr Opin Neurobiol 9:288 –
292.
30. Nakada C, et al. (2003) Accumulation of anchored proteins forms membrane diffusion
barriers during neuronal polarization. Nat Cell Biol 5:626 – 632.
31. Rouse ST, Hamilton SE, Potter LT, Nathanson NM, Conn PJ (2000) Muscarinic-induced
modulation of potassium conductances is unchanged in mouse hippocampal pyramidal cells that lack functional M1 receptors. Neurosci Lett 278:61– 64.
32. Levey AI, Edmunds SM, Heilman CJ, Desmond TJ, Frey KA (1994) Localization of
muscarinic M3 receptor protein and M3 receptor binding in rat brain. Neuroscience
63:207–221.
33. Hasselmo ME (2006) The role of acetylcholine in learning and memory. Curr Opin
Neurobiol 16:710 –715.
34. Migliore M, Ferrante M, Ascoli GA (2005) Signal propagation in oblique dendrites of
CA1 pyramidal cells. J Neurophysiol 94:4145– 4155.
35. Migliore M, Shepherd GM (2002) Emerging rules for the distributions of active dendritic conductances. Nat Rev Neurosci 3:362–370.
36. Migliore M, et al. (2003) ModelDB: Making models publicly accessible to support
computational neuroscience. Neuroinformatics 1:135–139.
7874 兩 www.pnas.org兾cgi兾doi兾10.1073兾pnas.0802805105
ACKNOWLEDGMENTS. We thank Dr. P. Thomas and Prof. M. Duchen for help
in acquiring the ABP-FITC image (Fig. S2) and Mr. T. Kao for aiding with the
immunohistochemistry. This work was supported by a Wellcome International
Prize Travel Research Fellowship [work plan approved in January 2003
(M.M.S.)] and grants from the Epilepsy Research Foundation (M.M.S.), European Union (D.A.B.), and National Institutes of Health (E.C.C.).
Shah et al.