An unexpected role for TASK-3 potassium channels
in network oscillations with implications for sleep
mechanisms and anesthetic action
Daniel S. J. Panga,1, Christian J. Robledoa,1, David R. Carra, Thomas C. Genta, Alexei L. Vyssotskib, Alex Caleya,2,
Anna Y. Zechariaa, William Wisdenc, Stephen G. Brickleya, and Nicholas P. Franksa,3
aBiophysics Section, Blackett Laboratory, andcCell Biology and Functional Genomics Section, Division of Cell and Molecular Biology, Imperial College,
South Kensington, London SW7 2AZ, United Kingdom; andbInstitute of Neuroinformatics, University of Zurich/ETH Zurich, Winterthurerstrasse 190,
CH-8057 Zurich, Switzerland
Edited by Richard W. Aldrich, The University of Texas, Austin, TX, and approved August 24, 2009 (received for review June 29, 2009)
TASK channels are acid-sensitive and anesthetic-activated mem-
bers of the family of two-pore-domain potassium channels. We
have made the surprising discovery that the genetic ablation of
TASK-3 channels eliminates a specific type of theta oscillation in
the cortical electroencephalogram (EEG) resembling type II theta
(4–9 Hz), which is thought to be important in processing sensory
stimuli before initiating motor activity. In contrast, ablation of
TASK-1 channels has no effect on theta oscillations. Despite the
absence of type II theta oscillations in the TASK-3 knockout (KO)
in common and is involved in exploratory behavior, is unaffected.
In addition to the absence of type II theta oscillations, the TASK-3
KO animals show marked alterations in both anesthetic sensitivity
and natural sleep behavior. Their sensitivity to halothane, a potent
activator of TASK channels, is greatly reduced, whereas their
sensitivity to cyclopropane, which does not activate TASK-3 chan-
nels, is unchanged. The TASK-3 KO animals exhibit a slower
progression from their waking to sleeping states and, during their
sleeping period, their sleep episodes as well as their REM theta
oscillations are more fragmented. These results imply a previously
unexpected role for TASK-3 channels in the cellular mechanisms
underlying these behaviors and suggest that endogenous modu-
lators of these channels may regulate theta oscillations.
EEG ? knockout ? REM ? theta ? wavelet
able voltage oscillations can be recorded in the electroenceph-
alogram (EEG). These oscillations are observed over a wide
range of frequencies and reflect the synchronous neuronal
activity that occurs during a variety of different behaviors. For
example, as animals explore their environments, as they learn
and lay down memories, as they process sensory input, and as
they sleep, characteristic oscillations occur in the ‘‘theta’’ range
of frequencies (4–12 Hz) (1). These theta oscillations are often
divided into two types (1–5): Type I, which occurs at slightly
higher frequencies (6–12 Hz), and type II (also known as arousal
theta), which occurs at the lower end of the range (4–9 Hz). Type
and rearing, whereas type II theta is associated with immobility
during the processing of sensory stimuli relevant to initiating, or
intending to initiate, motor activity.
The neuronal networks that generate these theta oscillations
involve ascending pathways from the brainstem that project to
the hypothalamus and then to the medial septum/diagonal band
of Broca and the hippocampus (6–9). Where the true pacemaker
is located is unclear, but the basic requirements for a neuron to
oscillate are a depolarizing drive (such as a sodium current)
together with a restoring drive, such as a repolarizing potassium
current. Most computational models (10–12) include several
different ionic currents, some of which are well-characterized
f a sufficient number of neurons participate in network oscil-
lations, then the local field potentials summate, and measur-
and attributed to known ion channels (e.g., HCN channels
underlying Ih), whereas others are only defined operationally
(e.g., slow potassium currents).
We have been studying the role TASK-3 potassium channels
might play in general anesthesia. This channel is a member of a
family of 15 ‘‘background’’ or ‘‘leak’’ potassium channels (13)
that is directly inhibited by acid and activated by certain inha-
lational anesthetics (14, 15). During our initial experiments, we
monitored the cortical EEG as a function of anesthetic concen-
tration and made a striking observation. In wild-type mice, a
highly-tuned anesthetic-induced peak in the theta band of fre-
quencies (4–9 Hz), which appeared at around the concentrations
that induced a loss of righting reflex, was absent in the TASK-3
knockout (KO) animals. TASK-1 KO mice, on the other hand,
appeared identical to wild-type animals.
In this paper we show that the ablation of TASK-3 potassium
channels removes type II theta oscillations, but leaves type I
theta oscillations and exploratory behavior unaffected. More-
over, the TASK-3 KO mice show altered anesthetic sensitivity,
disrupted sleep behavior, and a fragmentation of both sleep
episodes and theta oscillations during REM sleep. These results
suggest that TASK-3 channels play key roles in anesthetic
sensitivity and the regulation of sleep.
Anesthetic-Induced Loss of Righting Reflex. To assess any differ-
ences in anesthetic sensitivity between wild-type, TASK-1 KO,
and TASK-3 KO animals, we used the loss of righting reflex
(LORR) as an assay. In rodents, LORR is observed at the same
concentrations as loss of consciousness in humans (16), with a
comparably steep concentration-response curve reflecting a
sharp transition between the awake and anesthetized states. We
first investigated halothane because of its great efficacy in
activating TASK channels (14, 15, 17–19). We found that the
TASK-3 KO mice were significantly (P ? 0.001) less sensitive to
halothane, with an EC50of 0.94 ? 0.02% atm (n ? 12) compared
with 0.68 ? 0.02% atm (n ? 19) for wild-type animals (Fig. 1A).
In contrast, TASK-1 KO mice showed only a small change in
anesthetic sensitivity with an EC50of 0.78 ? 0.01% atm; n ? 20.
To help assess if the difference between wild-type and TASK-3
Author contributions: S.G.B. and N.P.F. designed research; D.S.J.P., C.J.R., D.R.C., T.C.G.,
A.C., A.Y.Z., and N.P.F. performed research; A.L.V. and W.W. contributed new reagents/
analytic tools; D.S.J.P., C.J.R., D.R.C., T.C.G., A.Y.Z., W.W., S.G.B., and N.P.F. analyzed data;
and W.W. and N.P.F. wrote the paper.
The authors declare no conflict of interest.
This article is a PNAS Direct Submission.
Freely available online through the PNAS open access option.
1D.S.J.P. and C.J.R. contributed equally to this work.
2Present address: School of Pharmacy, University of London, 29/39 Brunswick Square,
London WC1N 1AX, UK.
3To whom correspondence should be addressed. E-mail: firstname.lastname@example.org.
October 13, 2009 ?
vol. 106 ?
KO animals was specifically due to the absence of TASK-3
channels, we next investigated the effects of cyclopropane, an
anesthetic gas that, even at the highest concentrations, does not
significantly activate TASK-3 channels (20). We found that for
cyclopropane, the EC50for LORR was identical (P ? 0.5) for
wild-type and TASK-3 KO animals, being 19.6 ? 0.8% atm (n ?
10) and 19.0 ? 0.8% atm (n ? 12), respectively (Fig. 1B).
An Atropine-Sensitive Theta Oscillation Is Absent in TASK-3 KO Mice.
In parallel with our LORR measurements, we recorded the
cortical EEG as a function of anesthetic concentration. In
wild-type animals, we observed the striking appearance of a
highly tuned peak in the theta range of frequencies (?4–9 Hz)
in the power spectrum when the mice were exposed to halothane
at and above the concentrations that caused a LORR (Fig. 1C).
In contrast, this peak was absent in TASK-3 KO animals (Fig.
1D). We obtained the same results with another inhalational
general anesthetic, isoflurane (n ? 4). In wild-type animals, this
halothane-induced theta oscillation could be greatly (?85%)
inhibited (n ? 3) by systemic atropine (50 mg/kg i.p.), a
nonselective antagonist of muscarinic acetylcholine receptors
(inset to Fig. 1C). The same dose of atropine had no significant
effect (n ? 6) on the power spectra from TASK-3 KO mice in
wild-type (n ? 3) or TASK-3 KO (n ? 7) mice in the absence of
halothane. We also performed the above experiments using
TASK-1 KO mice and found identical results (n ? 4) to those
using wild-type animals (e.g., see Fig. 1E).
Atropine-sensitive theta oscillations (sometimes called type II
theta oscillations) that are resistant to the presence of anesthet-
the medial septum to the hippocampus (21, 22). To confirm that
this septohippocampal pathway was involved, we injected (n ?
3) the local anesthetic lidocaine into the medial septum in
wild-type mice exposed to a concentration of halothane suffi-
cient to induce the sharply tuned theta oscillation. Lidocaine
injection caused an abrupt and almost complete elimination of
the theta peak; the effect reversed after about 20 min. This is
illustrated by the data of Fig. 1F, which shows the Wavelet power
spectrum (see Materials and Methods) of the EEG as a function
of time, just before, and following, lidocaine injection in the
presence of halothane. The involvement of cholinergic pathways
is also suggested by the fact that anticholinesterase drugs such as
physostigmine produce slow type II theta oscillations in many
species, including mice (23–25). We confirmed this in wild-type
mice (n ? 3) and found that an i.p. injection of 0.2 mg/kg
produced a 270 ? 60% increase (P ? 0.05) in peak theta power
centered at 4.5 Hz. In TASK-3 KO animals, in contrast, this dose
of physostigmine had no significant effect (n ? 3; P ? 0.1).
We next examined how the theta oscillation was influenced by
halothane concentration. For all three genotypes (wild-type,
TASK-1 KO, and TASK-3 KO), a peak in the theta range was
Hz), but this gradually shifted to lower frequencies as the
concentration of halothane increased (Fig. 2 A–C). With both
wild-type (n ? 8) and TASK-1 KO (n ? 6) mice, the peak
sharpened and the peak power greatly increased, reaching a
maximum at around 1% halothane. This is shown in Fig. 2 D and
E where the ‘‘Quality’’-factor Q (defined as the peak frequency/
peak width) is plotted against anesthetic concentration. The
Q-factor is a measure of the ‘‘tuning’’ of the oscillation. For
TASK-3 KO mice (n ? 11), this tuned, atropine-sensitive, theta
oscillation was absent (Fig. 2F).
Cyclopropane, which is inactive on TASK-3 channels, caused
no significant increase in theta power for either wild-type (n ?
7) or TASK-3 KO animals (n ? 8) over the range of concen-
trations tested (8–25% atm).
Exploratory Theta Oscillations in Wild-Type and TASK-3 KO Animals
Are Identical. Theta oscillations occur when animals explore their
environments (1, 5, 26), and we next investigated how this
‘‘exploratory’’ or type I theta was affected by the TASK-3
potassium channel ablation and the absence of type II theta.
Mice were placed in an activity monitor so that their walking
power spectra were essentially identical for the wild-type and
TASK-3 KO animals (Fig. 3A). However, because it has been
reported (27–30) that the frequency of the exploratory theta
oscillations increases with the speed of the animal, we investi-
gated whether or not this occurred in our mice. Because of the
limitations of conventional power spectra in terms of time/
frequency resolution, we calculated the Wavelet power spectrum
(31) as a function of time (Fig. 3B) and, using the coordinates
provided by the activity monitor, calculated the frequency at
tivity and eliminates an atropine-sensitive theta oscillation. (A) TASK-3 KO
potently activates TASK channels (14, 17–19). (B) In contrast, the loss of
righting reflex caused by cyclopropane, an anesthetic that has no effect on
TASK-3 channels (20), was unchanged (P ? 0.5). (C) At around loss of righting
spectrum in the theta band of frequencies in wild-type animals that was
sensitive to atropine (Inset). (D) TASK-3 KO mice had a strikingly different
phenotype. Halothane did not induce a tuned theta oscillation at any con-
centration. (E) TASK-1 KO mice displayed an identical behavior to wild-type
animals, with halothane also inducing a tuned theta oscillation sensitive to
in the presence of 1% halothane. (F) Injection of lidocaine into the medial
septum reversibly inhibited the halothane-induced theta oscillation in wild-
type mice. The Wavelet power spectrum shows the appearance of a theta
oscillation during halothane exposure and its abrupt elimination, and then
midline at a depth from the surface of the skull of 5 mm and 0.8 mm from the
The ablation of TASK-3 potassium channels alters anesthetic sensi-
Pang et al. PNAS ?
October 13, 2009 ?
vol. 106 ?
no. 41 ?
maximum theta power as a function of the speed of the mouse.
The data confirmed that the theta frequency does, indeed,
increase with the speed of the animal, but the data for wild-type
(n ? 5) and TASK-3 KO (n ? 5) animals were virtually identical
(Fig. 3C). These data also serve to emphasize that the absence
of TASK-3 channels has selectively ablated one specific type of
Natural Sleep and Theta Oscillations During REM Are Fragmented in
TASK-3 KO Animals. Given that network oscillations feature prom-
the nocturnal period (32), we next investigated if the natural
sleep behavior of the mice had been affected. By using miniature
data-logging devices, we recorded the EEG and EMG of mice in
their ‘‘natural’’ home-cage environments over the full sleep-
wake cycle. There was a clear distinction in sleep behavior
between the genotypes (Fig. 4A). As expected, wild-type animals
(black lines in Fig. 4A, n ? 5) showed a large difference (n ? 5)
in the levels of wakefulness, non-REM and REM sleep during
the ‘‘lights on’’ 12-h period (natural sleep period) compared with
the ‘‘lights off’’ period (natural wake period). Moreover, the
progression between these different behavioral states was rather
abrupt. The TASK-3 KO animals (n ? 5), on the other hand,
showed a much slower progression when moving from the
natural wake period to the sleep period. During the first 2 h after
‘‘lights on,’’ the TASK-3 KO animals spent significantly (P ?
0.05) more time awake, and significantly (P ? 0.05) less time in
REM or non-REM, than wild-type mice, but this difference
progressively reduced over time.
When the distributions of sleep episodes were analyzed, we
found that, during the natural wake period, the number and
average length of sleep episodes was not significantly different
(P ? 0.1 and P ? 0.5, respectively) for wild-type and TASK-3 KO
mice (Fig. 4B, top graph). In contrast, there was an obvious
difference between the genotypes during the natural sleep
period. Here, the number of sleep episodes for the TASK-3 KO
mice was significantly (P ? 0.001) larger than for the wild-type
(36.0 ? 3.5 compared with 19.3 ? 2.5), but their durations were
significantly (P ? 0.005) shorter (655 ? 52 s compared with
1452 ? 180 s). In other words, the sleep episodes were
wild-type, TASK-1, and TASK-3 KO animals. Halothane caused a concentra-
tion-dependent decrease in theta frequency in (A) wild-type mice, (B) TASK-1
KO mice, and (C) TASK-3 KO mice. For both (D) wild-type and (E) TASK-1 KO
animals, however, this was accompanied by a marked increase in peak theta
power and the sharpness of the theta peak at and above the concentrations
that induce a loss of righting reflex. This is shown by the plots of the Q-factor
vs. halothane concentration. The Q-factor of a tuned oscillator is defined as
the frequency divided by the full width at half maximum. (F) With TASK-3 KO
mice, halothane did not increase theta power, and the Q-factor did not
change significantly with halothane concentration.
Characteristics of theta oscillations in the presence of halothane for
ments, ‘‘exploratory theta,’’ were identical in wild-type and TASK-3 KO mice.
(A) The Fast Fourier Transform (FFT) power spectrum of the EEG for mice
identical in wild-type and TASK-3 KO mice. On the left are typical traces
showing the movement of the different genotypes over a 30-min period—as
mice. (B) The top traces show a typical segment of the EEG for a wild-type
mouse with the corresponding speed of the animal averaged over 0.5-s
epochs. The Wavelet power spectrum (see Material and Methods) beneath
how trains of theta oscillations (at ?8 Hz) are often interrupted when the
animal stops moving. (C) A more detailed analysis reveals that the frequency
ANOVA) as a function of the speed of the animal.
The theta oscillations that occur when mice explore their environ-
www.pnas.org?cgi?doi?10.1073?pnas.0907228106Pang et al.
Because the most striking phenotypic difference between the
wild-type and TASK-3 KO animals was the loss of the type II
theta oscillations (Fig. 1 C and D), and because natural sleep had
been disrupted, we next investigated whether the theta oscilla-
tions that occur during REM sleep were also affected. We
calculated the Wavelet power spectrum as a function of time
during periods of sustained REM sleep that occurred during the
natural sleep period (‘‘lights on’’). Increased fragmentation of
the theta oscillations was evident in the Wavelet power spectra
of the TASK-3 KO mice. Fig. 4C shows representative examples
of Wavelet power spectra for the two genotypes, where it can be
seen that in wild-type animals, theta oscillations tend to be less
interrupted. This is shown quantitatively by the autocorrelation
functions shown below (Fig. 4D) calculated on 120-s segments of
data from both wild-type (left graph; n ? 5) and TASK-3 KO
animals (right graph; n ? 5). A test between the peak heights in
the autocorrelation functions show a significant difference (P ?
0.05) between the wild-type and KO animals.
The TASK-3 KO mice were significantly less sensitive to the
general anesthetic halothane. Indeed, they showed the greatest
genetically-engineered decrease in sensitivity to an inhalational
anesthetic yet reported for anesthetic-induced hypnosis (16, 33).
This decrease in halothane sensitivity is comparable with that
observed with TREK-1 KO mice in response to a painful
stimulus (33). The fact that the TASK-3 KO mice displayed an
unchanged sensitivity to cyclopropane, an agent that does not
measurably activate TASK-3 channels (20), supports the idea
that the decrease in halothane sensitivity was a direct conse-
quence of their absence. Thus, involvement of TASK-3 channels
in anesthetic-induced LORR seems likely. A possible role for
anesthetic-activated potassium channels was suggested many
years ago (34). The first such channel that was characterized was
discovered in the pond snail Lymnaea stagnalis (18), and when
recently cloned (17), found to be closely related to mammalian
TASK channels that have been shown to be activated by a variety
of volatile general anesthetics (14, 15). Nonetheless, other
channels, such as GABAAreceptors and other two-pore-domain
potassium channels are almost certain to be involved (16, 35, 36),
because at only 40% higher concentrations, the TASK-3 KO
mice are also anesthetized.
The loss of type II theta oscillations in the TASK-3 KO mice
was completely unexpected. This theta oscillation has been
widely studied, using a variety of different anesthetics (4, 6, 8,
37). The increased tuning of the oscillation under halothane
anesthesia implies that the opening of TASK-3 channels pro-
motes neuronal synchronicity. The oscillation is characterized by
its sensitivity to atropine, and it has been postulated to mediate
the processing of sensory stimuli before initiating motor activity
(2–5). The absence of type II theta was specific to the loss of the
TASK-3 channels, because the removal of TASK-1 channels left
the oscillations unchanged.
Our finding that type I exploratory theta oscillations are
implications for the extent to which the pathways and molecular
mechanisms that mediate the two types of theta oscillations
overlap. Clearly, TASK-3 potassium channels are a necessary
component for type II oscillations but play no evident role in
type I oscillations. Although the circuitry responsible for gen-
erating these oscillations is not certain, it is widely believed that
a pathway from the brainstem, ascending through the hypothal-
This is consistent with our observation that the type II theta
oscillation can be reversibly blocked by lidocaine injection into
the medial septum.
Given that the recombinant TASK-1 and TASK-3 channels
have similar biophysical properties with respect to anesthetic
sensitivity (15), why are the deficits in oscillatory activity and
because of differences in neuronal expression between the two
genes (38); we surmise that the TASK-1 gene is not expressed,
or not expressed highly, in the neurons either driving or sup-
porting the oscillations. In the adult mouse forebrain the
TASK-3 gene has much stronger expression than TASK-1. In
particular, TASK-3 mRNA is abundant in layers 2 to 6 of the
neocortex, CA1 hippocampal pyramidal cells, dentate granule
cells, and the septum (38, 39); the TASK-3 gene is also expressed
in parvalbumin-positive GABAergic interneurons (40) in the
oscillations (41, 42). The TASK-1 gene, by contrast, is poorly
expressed in the mouse hippocampus (38). One scenario is that,
as a consequence of losing TASK-3 channels, a change in the
biophysical behavior of a specific type of hippocampal interneu-
both sleep episodes and theta oscillations. (A) TASK-3 KO mice show a much
natural sleep period (‘‘lights on’’). During the natural wake period (‘‘lights
during the natural wake period. During the natural sleep period (lower
panel), however, the TASK-3 KO animals show a fragmented sleep pattern
with significantly more (P ? 0.001), but significantly shorter (P ? 0.005) sleep
episodes. (C) REM sleep episodes during the natural sleep period showed a
clear difference in the fragmentation of the theta oscillations. The Wavelet
for wild-type and TASK-3 KO mice. The autocorrelations in (D) reflect this
REM theta oscillations (P ? 0.05) than do TASK-3 KO mice (n ? 5). The gray
shading represents the SEM envelope, and the arrows indicate the time at
which a peak in the autocorrelation function is no longer significantly differ-
ent from zero.
Pang et al. PNAS ?
October 13, 2009 ?
vol. 106 ?
no. 41 ?
ron (e.g., a subtype of parv-positive interneuron) might produce
the selective loss of type II theta with minimal impact on type I
theta. It has recently been shown, for example, that the loss of
TASK-3 channels in cerebellar granule neurons affects their
ability to sustain high-frequency firing of action potentials (43).
Alternatively, TASK-3 channels might govern the regulation of
selective cholinergic input to the hippocampus, so their absence
might disrupt theta oscillations. More work is needed to eluci-
date the cellular mechanism.
Previously, establishing the physiological roles of type II theta
when anesthetic is present, or can only be blocked when atropine
is present. Both of these drug treatments will affect neuronal
behavior in ways that do not involve theta oscillations per se.
TASK-3 KO animals should allow us to test hypotheses about
where and when these oscillations might play a role. Whether or
not the theta oscillations that are elicited by various drug
treatments correspond to the theta oscillations that are observed
during specific behavioral states requires further investigation.
Network oscillations and sleep are intimately connected, with
of oscillations in the EEG. Transient spindles and K-complexes
appear before sleep onset, and the relative amplitudes of theta
and delta oscillations (together with the EMG) allow REM and
non-REM sleep states to be distinguished. It is too early to be
certain that the absence of type II theta oscillations in the
TASK-3 KO animals is directly responsible for the changes we
observe in sleep behavior. Nonetheless, the differences we see in
both the time to transition into both REM and non-REM sleep,
as well as the fragmentation in the sleep episodes are a strong
of the mechanism used in natural sleep regulation, perhaps via
muscarinic acetylcholine receptors (44). Interestingly, atropine
treatment, which blocks type II theta oscillations, also shortens
of TASK-3 involvement in sleep mechanisms in general and
theta oscillations in particular.
In summary, we have made the surprising discovery that
TASK-3 channels are required for atropine-sensitive type II
theta oscillations. In contrast, they make no contribution to type
I exploratory theta oscillations. Animals lacking these channels
display a reduced sensitivity to the general anesthetic halothane,
have a significantly slower progression into sleep, and exhibit
fragmented sleep behavior. Because of the many ways in which
might lead to the regulation of behavior via the promotion or
reduction of theta oscillations.
Materials and Methods
Mice. All experiments were in accordance with the United Kingdom Animals
(Scientific Procedures) Act of 1986 and approved by the Ethical Review Com-
mittee of Imperial College London. Animals were housed in a humidity- and
temperature-controlled room, under a 12:12-h light-dark cycle. Water and
with a disruption of the first coding exon, were as described previously.
Anesthetic-Induced Loss of Righting Reflex. An animal was placed in a cylin-
drical glass chamber (900 mL) and, following a 10-min baseline period with
100% oxygen, the anesthetic was introduced, initially at 0.4% for halothane
or 8% for cyclopropane. The anesthetic concentration was then increased
stepwise (steps of 0.1% for halothane and 4% for cyclopropane), and after 10
min equilibration, LORR was assessed by manually rotating the glass cylinder
and scoring a LORR if the animal had all four feet off the ground for 30 s or
more. The observer was blinded to the genotype of the animal. Each animal
was tested once at each anesthetic concentration. Normothermia was main-
tained using a heat lamp placed 45 cm above the glass cylinder. A quantal
concentration-response curve was calculated using the method of Waud (47).
anesthesia. Three gold-plated EEG electrodes (Decolletage AG) were inserted
through the skull onto the dura mater, the first in the frontal bone (?1.5 mm
to Bregma, -1.5 mm from midline), the second in the parietal bone (?1.5 mm
to Bregma, ?1.5 mm from midline), and the third in the interparietal bone
over the cerebellum (?2.0 mm from Lambda, 0.0 mm from midline) for the
reference electrode. Three lengths of Teflon-insulated stainless steel wire
at least 7 days to recover from surgery. The EEG and EMG signals were
chip (http://www.vyssotski.ch/neurologger2). This device was sufficiently
small (about 2 g including batteries) to be attached directly to the animal’s
skull. Four data channels (up to 30 h at 10-bit resolution) could be recorded at
a sampling rate of 400 Hz and were bandpass-filtered (?3 db corner fre-
quency) between 1 and 70 Hz followed by high-pass (0.6 Hz, -3db) offline
in free-moving animals, either in an activity monitor or in their home cages,
and familiar environment.
It should be noted that our EEG measurements were from the cortex, so
that the sleep state of the mice could be determined. Previous work on theta
oscillations has usually been done with rats, often using hippocampal elec-
the cortex in mice, it is almost certain that oscillations generated in the
hippocampus would be detected by our cortical EEG electrodes.
EEG Analysis. EEG data were analyzed using either conventional FFT power
spectra (Fast Fourier transforms of the autocorrelation function) or Morlet
Wavelet analysis. FFT power spectra were calculated using the program Spike
(Spike 2, v5.14; Cambridge Electronic Design) with the area being normalized
to 100. Where a theta peak was observed, a good fit to the data could be
obtained using a Lorentzian function:
p ? p0??
?2? ?f ? f0?2?
where p is the EEG power, p0is a baseline, a0is the height of the Lorenzian, ?
is the half-width at half maximum, f is the frequency and f0 is the peak
frequency. The Q-factor, a measure of the sharpness of the peak, was calcu-
lated as f0/(2?).
EEG data were also analyzed using Wavelet transforms (49), which are
involves convoluting the EEG signal with a series of ‘‘Daughter’’ wavelets,
transform is defined as:
W?s, ?? ?
t ? ?
where s and ? represent the scale and local center of the wavelet ?(s, ?), and
x(t) is the EEG signal as a function of time. We used the most commonly used
Mother wavelet, the Morlet function, which is a complex sinusoid, windowed
by a Gaussian:
?0??? ? ??1/4ei?0?e??2/2
where ? is a dimensionless ‘‘time’’ parameter, and ?0is the dimensionless
wavelet central ‘‘frequency’’ that was set to 6 to satisfy the admissibility
criterion (50). The Wavelet power spectra were calculated using Matlab
(MathWorks) using a script based on that of Torrence and Compo (49).
Sleep Scoring. For the sleep experiments, mice were placed in a temperature-
controlled, sound-proof box illuminated within on a 12:12-h light-dark cycle.
Data were recorded both from animals in their home cages as well as from
Med Associates) and analyzed with Activity Monitor software (Med Associates).
The activity box was thoroughly cleaned with ethanol between experiments.
sleep, non-REM; or wake, W) was scored automatically using an established
www.pnas.org?cgi?doi?10.1073?pnas.0907228106Pang et al.
protocol (51). Briefly, the scoring consisted of filtering the EEG into ‘‘delta’’
(0.5–4 Hz) and ‘‘theta’’ (6–10 Hz) frequency bands and scoring 20-s epochs as
one or other of the three states based on W being periods of high EMG and
intermediate theta/delta ratio, REM being periods with high theta/delta ratio
and low EMG and non-REM being periods with high delta waves, low theta/
delta ratio, and low EMG.
Statistics. Unless otherwise stated, Student’s t-test was used to test for signif-
icance. Where shown, errors bars represent the SEM.
ACKNOWLEDGMENTS.We thank Raquel Yustos for technical assistance and
Geoff Horseman (CED Ltd.) for help with the sleep-scoring algorithm. C.J.R.
held a studentship from the Medical Research Council (UK); D.S.J.P. was the
recipient of a studentship from the Royal College of Anesthetists and the
British Journal of Anesthesia; and A.C., D.R.C., and T.C.G. were recipients of
studentships from the Biotechnology and Biological Sciences Research
Council (UK). This work was supported by grants from the Biotechnology
and Biological Sciences Research Council G021691; Air Products and Chem-
icals, Inc.; and Medical Research Council (UK) Grant G0501584 (to N.P.F.,
S.G.B., and W.W.).
1. Buzsaki G (2002) Theta oscillations in the hippocampus. Neuron 33:325–340.
2. Oddie SD, Kirk IJ, Whishaw IQ, Bland BH (1997) Hippocampal formation is involved in
movement selection: Evidence from medial septal cholinergic modulation and con-
current slow-wave (theta rhythm) recording. Behav Brain Res 88:169–180.
3. Vanderwolf CH (1969) Hippocampal electrical activity and voluntary movement in the
rat. Electroencephalogr Clin Neurophysiol 26:407–418.
4. Kramis R, Vanderwolf CH, Bland BH (1975) Two types of hippocampal rhythmical slow
activity in both the rabbit and the rat: Relations to behavior and effects of atropine,
diethyl ether, urethane, and pentobarbital. Exp Neurol 49:58–85.
5. Oddie SD, Bland BH (1998) Hippocampal formation theta activity and movement
selection. Neurosci Biobehav Rev 22:221–231.
6. Colom LV, et al. (1991) In vivo intrahippocampal microinfusion of carbachol and
bicuculline induces theta-like oscillations in the septally deafferented hippocampus.
7. Kirk IJ (1998) Frequency modulation of hippocampal theta by the supramammillary
nucleus, and other hypothalamo-hippocampal interactions: Mechanisms and func-
tional implications. Neurosci Biobehav Rev 22:291–302.
8. Kirk IJ, McNaughton N (1993) Mapping the differential effects of procaine on fre-
quency and amplitude of reticularly elicited hippocampal rhythmical slow activity.
9. Vertes RP, Kocsis B (1997) Brainstem-diencephalo-septohippocampal systems control-
ling the theta rhythm of the hippocampus. Neuroscience 81:893–926.
10. Acker CD, Kopell N, White JA (2003) Synchronization of strongly coupled excitatory
neurons: Relating network behavior to biophysics. J Comput Neurosci 15:71–90.
11. Rotstein HG, et al. (2005) Slow and fast inhibition and an H-current interact to create
a theta rhythm in a model of CA1 interneuron network. J Neurophysiol 94:1509–1518.
12. Traub RD, et al. (2004) Cellular mechanisms of neuronal population oscillations in the
hippocampus in vitro. Annu Rev Neurosci 27:247–278.
13. Goldstein SA, Bockenhauer D, O’Kelly I, Zilberberg N (2001) Potassium leak channels
and the KCNK family of two-P-domain subunits. Nat Rev Neurosci 2:175–184.
14. Patel AJ, et al. (1999) Inhalational anesthetics activate two-pore-domain background
K?channels. Nat Neurosci 2:422–426.
15. Talley EM, Bayliss DA (2002) Modulation of TASK-1 (Kcnk3) and TASK-3 (Kcnk9)
potassium channels: Volatile anesthetics and neurotransmitters share a molecular site
of action. J Biol Chem 277:17733–17742.
16. Franks NP (2008) General anaesthesia: From molecular targets to neuronal pathways
of sleep and arousal. Nat Rev Neurosci 9:370–386.
17. Andres-Enguix I, et al. (2007) Determinants of the anesthetic sensitivity of two-pore
domain acid-sensitive potassium channels: Molecular cloning of an anesthetic-
activated potassium channel from Lymnaea stagnalis. J Biol Chem 282:20977–20990.
18. Franks NP, Lieb WR (1988) Volatile general anaesthetics activate a novel neuronal K?
current. Nature 333:662–664.
acid pathway inhibitors on K?currents activated by volatile anaesthetics and FMRF-
amide in molluscan neurones. Br J Pharmacol 125:309–318.
20. Gruss M, et al. (2004) Two-pore-domain K?channels are a novel target for the
anesthetic gases xenon, nitrous oxide, and cyclopropane. Mol Pharmacol 65:443–452.
21. Mizumori SJ, Barnes CA, McNaughton BL (1989) Reversible inactivation of the medial
types. Brain Res 500:99–106.
22. Oddie SD, Stefanek W, Kirk IJ, Bland BH (1996) Intraseptal procaine abolishes hypo-
thalamic stimulation-induced wheel-running and hippocampal theta field activity in
rats. J Neurosci 16:1948–1956.
23. Leung LS, Vanderwolf CH (1980) Behavior-dependent evoked potentials in the hip-
Brain Res 198:119–133.
24. Bland BH, Colom LV (1988) Responses of phasic and tonic hippocampal theta-on cells
to cholinergics: Differential effects of muscarinic and nicotinic activation. Brain Res
evoked potentials in the mouse. Neuropsychobiology 40:47–56.
26. O’Keefe J (1993) Hippocampus, theta, and spatial memory. Curr Opin Neurobiol
27. Jeewajee A, Barry C, O’Keefe J, Burgess N (2008) Grid cells and theta as oscillatory
interference: Electrophysiological data from freely moving rats. Hippocampus
28. Rivas J, Gaztelu JM, Garcia-Austt E (1996) Changes in hippocampal cell discharge
patterns and theta rhythm spectral properties as a function of walking velocity in the
guinea pig. Exp Brain Res 108:113–118.
29. Slawinska U, Kasicki S (1998) The frequency of rat’s hippocampal theta rhythm is
related to the speed of locomotion. Brain Res 796:327–331.
30. Gordon JA, Lacefield CO, Kentros CG, Hen R (2005) State-dependent alterations in
hippocampal oscillations in serotonin 1A receptor-deficient mice. J Neurosci 25:6509–
31. Daubechies I (1990) The wavelet transform, time-frequency localization and signal
analysis. IEEE Trans Inf Theory 36:961–1005.
impairments in cognitive functions, and reduced sensitivity to inhalation anesthetics.
J Pharmacol Exp Ther 323:924–934.
33. Heurteaux C, et al. (2004) TREK-1, a K?channel involved in neuroprotection and
general anesthesia. EMBO J 23:2684–2695.
34. Nicoll RA, Madison DV (1982) General anesthetics hyperpolarize neurons in the ver-
tebrate central nervous system. Science 217:1055–1057.
35. Hemmings HC, Jr, et al. (2005) Emerging molecular mechanisms of general anesthetic
action. Trends Pharmacol Sci 26:503–510.
36. Rudolph U, Antkowiak B (2004) Molecular and neuronal substrates for general anaes-
thetics. Nat Rev Neurosci 5:709–720.
37. Bland BH, et al. (2003) Effect of halothane on type 2 immobility-related hippocampal
theta field activity and theta-on/theta-off cell discharges. Hippocampus 13:38–47.
38. Aller MI, Wisden W (2008) Changes in expression of some two-pore domain potassium
channel genes (KCNK) in selected brain regions of developing mice. Neuroscience
39. Aller MI, et al. (2005) Modifying the subunit composition of TASK channels alters the
modulationofaleakconductanceincerebellargranuleneurons. JNeurosci 25:11455–
40. Torborg CL, et al. (2006) TASK-like conductances are present within hippocampal CA1
stratum oriens interneuron subpopulations. J Neurosci 26:7362–7367.
41. Klausberger T, Somogyi P (2008) Neuronal diversity and temporal dynamics: The unity
of hippocampal circuit operations. Science 321:53–57.
42. Wulff P, et al. (2009) Hippocampal theta rhythm and its coupling with gamma oscil-
lations require fast inhibition onto parvalbumin-positive interneurons. Proc Natl Acad
Sci USA 106:3561–3566.
43. Brickley SG, et al. (2007) TASK-3 two-pore domain potassium channels enable sus-
tained high-frequency firing in cerebellar granule neurons. J Neurosci 27:9329–9340.
44. Meuth SG, et al. (2003) Contribution of TWIK-related acid-sensitive K?channel 1
(TASK1) and TASK3 channels to the control of activity modes in thalamocortical
neurons. J Neurosci 23:6460–6469.
Possible implications for REM triggering mechanisms. J Neurosci Res 9:425–435.
46. Mavanji V, Datta S (2002) Sleep-wake effects of yohimbine and atropine in rats with a
clomipramine-based model of depression. Neuroreport 13:1603–1606.
47. Waud DR (1972) On biological assays involving quantal responses. J Pharmacol Exp
48. Vyssotski A, et al. (2009) EEG responses to visual landmarks in flying pigeons. Curr Biol
49. Torrence C, Compo G (1998) A practical guide to wavelet analysis. Bull Amer Meteor
50. Farge M (1992) Wavelet transforms and their applications to turbulence. Annu Rev
Fluid Mech 24:395–457.
51. Costa-Miserachs D, Portell-Cortes I, Torras-Garcia M, Morgado-Bernal I (2003) Auto-
mated sleep staging in rat with a standard spreadsheet. J Neurosci Methods 130:93–
Pang et al.PNAS ?
October 13, 2009 ?
vol. 106 ?
no. 41 ?