Divergent sodium channel defects in familial
Kristopher M. Kahlig*, Thomas H. Rhodes*, Michael Pusch†, Tobias Freilinger‡, Jose ´ M. Pereira-Monteiro§¶,
Michel D. Ferrari?, Arn M. J. M. van den Maagdenberg?**, Martin Dichgans‡, and Alfred L. George, Jr.*††‡‡
Departments of *Medicine and††Pharmacology, Vanderbilt University, Nashville, TN 37240;†Istituto di Biofisica, Consiglio Nazionale delle Ricerche,
16149 Genoa, Italy;‡Department of Neurology, Ludwig-Maximilians-University, 80539 Munich, Germany;§Instituto de Cie ˆncias Biome ´dicas
Abel Salazar, Universidade do Porto, 4099-002 Porto, Portugal;¶Servic ¸o de Neurologia, Hospital Geral de Santo Anto ´nio, 4099-001 Porto,
Portugal; and Departments of?Neurology and **Human Genetics, Leiden University Medical Center, 2300 RC Leiden, The Netherlands
Edited by David E. Clapham, Harvard Medical School, Boston, MA, and approved May 8, 2008 (received for review December 13, 2007)
Familial hemiplegic migraine type 3 (FHM3) is a severe autosomal
dominant migraine disorder caused by mutations in the voltage-
gated sodium channel NaV1.1 encoded by SCN1A. We determined
the functional consequences of three mutations linked to FHM3
(L263V, Q1489K, and L1649Q) in an effort to identify molecular
defects that underlie this inherited migraine disorder. Only L263V
and Q1489K generated quantifiable sodium currents when coex-
pressed in tsA201 cells with the human ?1 and ?2 accessory
subunits. The third mutant, L1649Q, failed to generate measurable
whole-cell current because of markedly reduced cell surface ex-
pression. Compared to WT-NaV1.1, Q1489K exhibited increased
well as delayed recovery from fast and slow inactivation, thus
resulting in a predominantly loss-of-function phenotype further
demonstrated by a greater loss of channel availability during
repetitive stimulation. In contrast, L263V exhibited gain-of-
function features, including delayed entry into, as well as acceler-
ated recovery from, fast inactivation; depolarizing shifts in the
steady-state voltage dependence of fast and slow inactivation;
increased persistent current; and delayed entry into slow inacti-
vation. Notably, the two mutations (Q1489K and L1649Q) that
exhibited partial or complete loss of function are linked to typical
FHM, whereas the gain-of-function mutation L263V occurred in a
family having both FHM and a high incidence of generalized
epilepsy. We infer from these data that a complex spectrum of
NaV1.1 defects can cause FHM3. Our results also emphasize the
complex relationship between migraine and epilepsy and provide
further evidence that both disorders may share common molecular
epilepsy ? SCN1A ? NaV1.1 ? FHM3
estimated 30 million migraine sufferers in the United States
alone represent a significant public health burden. Genetic
factors are important determinants of migraine susceptibility,
and understanding the molecular basis of rare monogenetic
migraine syndromes may help unravel the pathophysiology of
more common forms of migraine (2).
Familial hemiplegic migraine (FHM) is a rare autosomal dom-
inant form of migraine with aura (MA) characterized by the
presence of transient motor weakness and other aura symptoms,
such as visual and speech disturbances (3). FHM probands also
have an increased incidence of typical MA attacks, suggesting that
FHM and MA may exhibit shared pathophysiology (4). Mutations
linked to FHM. FHM1 is caused by mutations in CACNA1A
encoding the pore-forming ?1-subunit of the neuronal calcium
channel, CaV2.1 (5), whereas mutations in ATP1A2 encoding the
catalytic ?2-subunit of the glial Na?/K?-ATPase in adult brain
cause FHM2 (6). Two missense mutations (Q1489K and L1649Q)
in SCN1A encoding the pore-forming ?-subunit of the voltage-
igraine is a common paroxysmal neurovascular disorder
with a lifetime prevalence of 14–18% worldwide (1). An
gated sodium channel NaV1.1 have been linked to FHM3 (7, 8).
Corresponding mutations engineered into the paralogous cardiac
NaV1.5 were observed to cause significant functional disturbances,
further supporting the pathogenicity of these alleles. At the time,
channel were not successful.
SCN1A mutations have been previously associated with various
(generalized epilepsy with febrile seizures plus) to debilitating
(severe myoclonic epilepsy of infancy). Functional studies of epi-
lepsy-associated NaV1.1 mutants have demonstrated either func-
channels (9–15). Among the subset of missense mutations charac-
terized, there are no general correlations between mutation loca-
tion in the sodium channel protein and the functional conse-
quences. Therefore, prediction of the functional effects of FHM3
mutations is not feasible without direct experimental evidence.
Here we determined the molecular basis of FHM3 by examining
the biophysical or biochemical properties of three FHM3 mutants
engineered in a recombinant human NaV1.1 channel. We studied
two previously published mutations (Q1489K and L1649Q) as well
as a third mutation (L263V) recently identified in an FHM family
in which the majority of mutation carriers also have epilepsy (42).
Our data revealed that FHM3 is associated with a wide range of
NaV1.1 defects, suggesting that multiple molecular mechanisms
may underlie FHM3. Furthermore, specific biophysical features
observed for L263V may also explain the higher prevalence of
epilepsy associated with this mutation.
In this study, we determined the functional consequences of three
FHM3 mutations engineered in human NaV1.1 to ascertain the
molecular defects underlying this disorder. Two of the FHM3
mutations we studied were reported previously (Q1489K and
L1649Q) (7, 8). Q1489K introduces a positive charge 11 residues
N-terminal to the Ile–Phe–Met motif within the cytoplasmic linker
between domain III (DIII) and DIV, a structure essential for fast
a conserved arginine (Arg-1648) in the S4 voltage sensor segment
Author contributions: K.M.K., T.H.R., M.P., T.F., J.M.P.-M., M.D.F., A.M.J.M.v.d.M., M.D.,
and A.L.G. designed research; K.M.K., T.H.R., and A.L.G. performed research; M.P., T.F.,
J.M.P-M., M.D.F., A.M.J.M.v.d.M., and M.D. contributed critical information enabling the
study; K.M.K., T.H.R., and A.L.G. analyzed data; and K.M.K., M.D.F., A.M.J.M.v.d.M., M.D.,
and A.L.G. wrote the paper.
The authors declare no conflict of interest.
This article is a PNAS Direct Submission.
Hall, Vanderbilt University, 2215 Garland Avenue, Nashville, TN 37232-0275. E-mail:
This article contains supporting information online at www.pnas.org/cgi/content/full/
© 2008 by The National Academy of Sciences of the USA
July 15, 2008 ?
vol. 105 ?
no. 28 ?
in the D1/S5 segment and was recently identified in a Portuguese
family in which epilepsy co-occurred in three of five mutations
carriers with hemiplegic migraine (42).
Transient coexpression of each FHM3 mutation with the human
sodium channel ?1 and ?2 accessory subunits in tsA201 cells
and Q1489K, but not L1649Q. Fig. 1 illustrates representative
whole-cell current traces recorded from cells expressing WT-
NaV1.1, L263V, Q1489K, or L1649Q. Repeated attempts to record
from cells transfected with L1649Q revealed consistently small
sodium currents (Fig. 1D). However, only 1 of 46 cells expressing
L1649Q exhibited whole-cell peak currents of ?600 pA, and we
could not satisfactorily exclude a contribution of endogenous
channels to these recordings. Thus, we considered L1649Q mutant
channels as nonfunctional.
Impaired Activation and Inactivation for L263V and Q1489K. We
determined the biophysical properties of the two functional mu-
tants L263V and Q1489K. Fig. 2A illustrates that there were no
significant differences in the peak current density for either mutant
at any test potential. Both L263V and Q1489K exhibited trends
toward activating at more negative membrane potentials (Fig. 2B),
but neither shift in activation, V1/2, was statistically significant
fast inactivation was explored by measuring channel availability at
?10 mV after a depolarizing 100-ms prepulse to various test
potentials. Fig. 2C illustrates that L263V exhibits a ?8 mV depo-
larizing shift in the V1/2of steady-state inactivation compared to
WT-NaV1.1, suggesting that L263V resists entry into the fast
inactivated state (Table S1). Moreover, examining the overlap in
‘‘window current’’ for L263V as compared to WT-NaV1.1.
L263V and Q1489K exhibited divergent effects on the time
course of recovery from fast inactivation compared to WT-NaV1.1
(Fig. 2D). L263V exhibited accelerated recovery from inactivation
faster component, ?f (Table S1). In contrast, Q1489K exhibited
slower recovery from inactivation compared to WT-NaV1.1 ex-
plained by a larger time constant, ?s, and increased fractional
amplitude, As, representing the slower component (Table S1).
Abnormal L263V gating is also apparent in the kinetics of the
whole-cell current traces. Fig. 3A illustrates a representative whole-
cell current recorded in response to a voltage step from ?120 to
entry into fast inactivation, as evidenced by a broadening in the
decay phase of the whole-cell current trace. The time constants
corresponding to the fast component of the whole-cell current
decay were significantly larger for L263V compared to WT-NaV1.1
at every test potential (Fig. 3B and Table S2). Time to peak current
was also delayed in cells expressing L263V (Fig. 3C), and this
phenomenon could reflect the slower inactivation kinetics or an
independent effect on activation. There was no significant defect in
activation or inactivation kinetics observed for Q1489K.
Many epilepsy-associated NaV1.1 mutants have been shown to
exhibit a significantly increased persistent current due to impaired
whole-cell currents for WT-NaV1.1, L263V, and Q1489K recorded
L263V and Q1489K exhibit significantly increased persistent cur-
rent levels compared to WT-NaV1.1 (WT-NaV1.1, 0.40 ? 0.06%,
n ? 5; L263V, 1.37 ? 0.22%, n ? 6, P ? 0.01; Q1489K, 1.48 ?
0.19%, n ? 7, P ? 0.01).
Divergent Effects on Slow Inactivation. NaV1.1 channels exhibit
slow inactivation, a process distinct from fast inactivation that
may regulate channel availability during sustained membrane
depolarizations or repetitive neuronal firing. We tested L263V
and Q1489K for defects in slow inactivation by using the
voltage protocols illustrated in Fig. 5 that largely eliminate the
contribution of fast inactivation to channel activity. Compared
to WT-NaV1.1, L263V displayed a delay in the onset (Fig. 5A)
and a ?13-mV depolarizing shift in the V1/2 of steady-state
slow inactivation (Fig. 5B). These findings predict that, similar
to fast inactivation, L263V resists entry into slow inactivated
states. In contrast, Q1489K entered slow inactivation more
rapidly compared to WT-NaV1.1 (Fig. 5A), but there was no
difference in voltage dependence. Both L263V and Q1489K
exhibited small but significant effects on recovery from slow
inactivation compared to WT-NaV1.1 (Fig. 5C and Table S3),
and these effects represent a distinct channel abnormality
from the altered recovery from fast inactivation illustrated in
Fig. 2D. These findings indicate that L263V and Q1489K
exhibit divergent effects on slow inactivation that may influ-
ence steady-state channel availability.
To test channel availability of WT or mutant NaV1.1 channels
during repetitive stimulation (use-dependence), we measured
whole-cell peak current amplitudes during a train of 100 depolar-
izing voltage steps to 0 mV at various frequencies. The peak
currents were recorded from human tsA201 cells coexpressing human ?1and
?2accessory subunits with either WT-NaV1.1 (A), L263V (B), Q1489K (C), or
L1649Q (D). (Inset) Currents were activated by voltage steps to between ?80
and ?20 mV from a holding potential of ?120 mV.
Representative whole-cell sodium currents. Representative sodium
www.pnas.org?cgi?doi?10.1073?pnas.0711717105Kahlig et al.
whole-cell current measured in response to the 100th voltage step
plotted against the stimulation frequency (Fig. 6). L263V exhibited
significantly greater channel availability for pulse train frequencies
100–140 Hz. In contrast, Q1489K displayed significantly lower
channel availability for pulse train frequencies 35–140 Hz. At a
stimulation frequency of 100 Hz, the residual availability of WT-
NaV1.1 channels at the 100th pulse was 0.56 ? 0.07 (n ? 6)
compared to 0.71 ? 0.02 (n ? 10, P ? 0.05) for L263V and 0.33 ?
0.03, (n ? 7, P ? 0.001) for Q1489K. Fig. 6 also illustrates the
use-dependent behavior of an epilepsy-associated mutation
(R1648H) that exhibits a level of residual availability (0.74 ? 0.02,
n ? 10, P ? 0.01 as compared to WT) very similar to L263V, the
FHM3 mutation associated with a high prevalence of seizures.
L1649Q Has Reduced Cell Surface Expression. The FHM3 mutant
L1649Q failed to generate quantifiable whole-cell sodium currents
when transiently expressed in tsA201 cells (Fig. 1). To determine
whether L1649Q non-function was due to impaired trafficking of
channels to the plasma membrane, we used cell-surface biotinyla-
tion coupled with Western blot analysis. Fig. 7A illustrates a
either WT-NaV1.1 or L1649Q exhibit similar levels of total protein
expression (Fig. 7A Left). However, L1649Q channels are not
7A Right). The magnitude of L1649Q cell surface expression was
the protein band in the total fraction. Averaged data from three
independent experiments (Fig. 7B) illustrate that L1649Q (0.14 ?
0.08, n ? 3, P ? 0.01) exhibited a significantly lower level of cell
surface expression compared to WT-NaV1.1 (1.00 ? 0.06, n ? 3).
likely reflects impaired trafficking to the plasma membrane.
We determined the functional consequences of three SCN1A
mutations (L263V, Q1489K, and L1649Q) associated with FHM3
to understand the molecular defects underlying this inherited
migraine syndrome. Our work demonstrated that FHM3 is asso-
ciated with mutant NaV1.1 channels exhibiting a broad range of
abnormalities, including gain of function (L263V), predominant
loss of function (Q1489K), or complete loss of function (L1649Q),
suggesting that this neurological disorder can result from multiple
Two of the mutations (Q1489K and L1649Q) were reported
cardiac sodium channel (NaV1.5) to demonstrate functional aber-
rations caused by the mutations. Specifically, these experiments
identified gain-of-function phenotypes, such as an accelerated
recovery from fast inactivation observed for both mutants or
impaired fast inactivation noted for a NaV1.5 mutation equivalent
to L1649Q. However, these data contrast considerably with our
findings for human NaV1.1, most likely because of intrinsic struc-
tural and functional differences between the two sodium channel
isoforms (11, 20). Therefore, we performed a comprehensive
evaluation of these two FHM3 alleles in NaV1.1. Furthermore, a
third FHM3 mutation (L263V) was recently discovered that we
were also able to study.
Functional Heterogeneity of FHM3 Mutations. The most striking
finding of our work was the diversity of molecular defects observed
for the three FHM3 mutations. On one extreme L1649Q was
nonfunctional, most likely because of its diminished trafficking to
the plasma membrane. Defective trafficking of mutant sodium
channels could be caused by intrinsic misfolding of the large
?-subunit protein or through abnormal interactions with accessory
subunits (10, 21). Q1489K exhibits a mixture of functional defects
including increased persistent current, more rapid onset of slow
inactivation, and greater loss of channel availability during repeti-
causes a predominantly loss-of-function phenotype. In contrast,
L263V exhibited a constellation of biophysical abnormalities that
predict increased sodium conductance and greater channel avail-
ability consistent with a gain-of-function mutation. Specifically,
when compared to WT-NaV1.1, L263V channels exhibited im-
paired fast and slow inactivation, increased persistent current,
depolarized voltage dependence of both forms of inactivation,
and mutant NaV1.1. Biophysical properties of WT-
NaV1.1 (?), L263V (?), and Q1489K (Œ) expressed in
tsA201 cells. (A) Peak current density elicited by test
pulses to various potentials and normalized to cell
capacitance. (B) Voltage dependence of channel acti-
and ?20 mV. (C) Voltage dependence of fast inactiva-
tion assessed in response to inactivating prepulses to
between ?140 and ?10 mV. (D) Time-dependent re-
covery from fast inactivation assessed after a 100-ms
inactivating prepulse to ?10 mV. (A Inset–D Inset)
in Table S1.
Activation and inactivation properties of WT
Kahlig et al.
July 15, 2008 ?
vol. 105 ?
no. 28 ?
accelerated recovery from fast inactivation, and greater channel
availability during repetitive stimulation. Notably, greater channel
availability during repetitive stimulation as seen for L263V also has
been observed for various epilepsy-associated SCN1A mutations
(Fig. 6) (15). Perhaps this biophysical feature correlates with the
unusually high prevalence of epilepsy observed in L263V mutation
carriers. The coincidence of migraine with epilepsy suggests that
these two neurological disorders may share common mechanisms
in L263V mutation carriers.
Pathogenesis of FHM3. Migraine is a complex neurovascular disor-
der. Headache is thought to originate from activation of the
trigeminovascular system (i.e., the trigeminal innervation of cranial
blood vessels and connections to the brainstem), whereas the aura
increase in neuronal activity immediately followed by a refractory
the cortex (22, 23). However, a detailed understanding of the
primary trigger and underlying mechanisms for CSD remains
unclear. Cellular studies with FHM mutants suggest a key role for
increased extracellular glutamate and K?(FHM1 and FHM2) for
lowering the threshold of cortical neurons for CSD. CACNA1A
mutations (FHM1) are thought to enhance presynaptic cortical
glutamate release, whereas ATP1A2 mutations (FHM2) result in
impaired clearance of K?and glutamate form the synaptic cleft
(24). Genetically engineered knock-in mice carrying a pathogenic
FHM1 mutation revealed gain-of-function effects with increased
calcium influx and neurotransmitter release resulting in a reduced
CSD threshold (25), supporting the concept that the migrainous
brain is hyperexcitable (26). Whereas, CaV2.1 channels and ?2-
containing Na?/K?-ATPase pumps are expressed in and around
dritically (27–31) and in the axonal initial segment (32–34), where
they may contribute to integrating incoming signals and controlling
for mutant NaV1.1 channels may cause FHM3 by nonsynaptic
Recent work with two NaV1.1 knock-out mouse models may
offer valuable insight into how our data may integrate into this
pathophysiological model of FHM. Loss of one NaV1.1 allele in
mice preferentially decreased the functioning of inhibitory
GABAergic interneurons without effects on excitatory pyramidal
neurons isolated from the hippocampus (32, 35). In contrast, mice
overexpressing a gain-of-function mutation in NaV1.2 exhibit sei-
zures that correlate with increased persistent current in hippocam-
pal excitatory pyramidal neurons (36). These findings suggest that
ing of inhibitory neurons, whereas gain-of-function mutations may
to peak current for WT-NaV1.1 and L263V. (B) Decay phase of the whole-cell
sodium current was fit with a two-exponential equation, and the inactivation
time constants representing the faster component (?f) for WT-NaV1.1 (?), L263V
(?), and Q1489K (Œ) are plotted against the test potential. Fit parameters are
provided in Table S2. (C) Time to peak sodium current for WT-NaV1.1 (?), L263V
(?), and Q1489K (Œ) plotted against the test potential. L263V exhibited a signif-
icantly delayed peak current compared to WT-NaV1.1. Significant differences
from WT-NaV1.1 are indicated as follows:*, P ? 0.05;**, P ? 0.001.
L263V exhibits delayed whole-cell current kinetics. (A) Representative
to the peak sodium current. (B) Time-expanded view of the sodium currents from A emphasizing the significantly increased persistent current of L263V and
Q1489K during the final 50 ms of the voltage step. (C) Quantification of persistent current as a percent of peak current for L263V and Q1489K. Both mutants
exhibited significantly increased levels of persistent current compared to WT-NaV1.1 indicated as follows:*, P ? 0.01.
L263V and Q1489K exhibit increased persistent current. (A) Representative TTX-subtracted, whole-cell sodium currents recorded for WT-NaV1.1, L263V,
www.pnas.org?cgi?doi?10.1073?pnas.0711717105 Kahlig et al.
predominantly influence excitatory neurons. The basis for these
differential effects may lie in intrinsic differences at the neuron
neurons have higher sodium channel density, exhibit lower action
potential thresholds, and decreased frequency attenuation during
high-intensity stimulation (37). These neurophysiological attributes
are likely critical for inhibitory neurons to exert their influence on
cortical output. We can speculate that the reduced threshold for
CSD in FHM3 is the final common pathway resulting from the
these neuronal subpopulations.
lack of specific information concerning the expression pattern and
functional contributions of this channel to cortical neuronal excit-
ability. Moreover, in contrast to CaV2.1 and ?2-containing Na?/
protocol consisting of a variable length inactivation pulse to ?10 mV followed by a test pulse at ?10 mV. Effects of fast inactivation were minimized by using
a 50-ms interpulse step to ?120 mV. (B) Voltage-dependent entry into slow inactivation was examined by using a two-pulse protocol consisting of a 30-s
conditioning pulse at various potentials followed by a test pulse at ?10 mV. Effects of fast inactivation were minimized by using a 50-ms interpulse step to ?120
fit parameters are provided in Table S3.
Slow inactivation properties of WT and mutant NaV1.1 channels. (A) Time-dependent entry into slow inactivation was examined by using a two-pulse
NaV1.1. Use-dependent channel availability was assessed by stimulating cells
with depolarizing pulse trains (100 pulses, 5 ms, 0 mV) from a holding
potential of ?120 mV at the indicated frequencies. The residual sodium
current (100th pulse) was normalized to the first pulse for each stimulation
train. A recovery period (15 s, ?120 mV) followed each pulse train. Significant
differences from WT-NaV1.1 are indicated as follows:*, P ? 0.05;**, P ? 0.01;
***, P ? 0.001.
Use-dependent effects on channel availability for WT and mutantFig. 7.
nylation was used to measure the cell surface expression of 3XFLAG-tagged
WT-NaV1.1 and L1649Q. (A) Representative Western blot analysis of protein
isolated from tsA201 cells coexpressing ?1 and ?2 alone (Mock) or with either
WT-NaV1.1 or L1649Q. Total protein (Left) and biotinylated cell surface pro-
tein (Right) are illustrated. (B) Densitometric analysis of three independent
experiments illustrates that L1649Q exhibits a significantly lower level of cell
surface expression compared to WT-NaV1.1 indicated as follows:**, P ? 0.01.
L1649Q exhibits reduced cell surface expression. Cell surface bioti-
Kahlig et al.
July 15, 2008 ?
vol. 105 ?
no. 28 ?