Febrile temperatures unmask biophysical defects in Na_v1.1 epilepsy mutations supportive of seizure initiation

Generalized epilepsy with febrile seizures plus (GEFS+) is an autosomal dominant disorder in which subjects have febrile seizures early in childhood, with increased risk for febrile and afebrile seizures later on in life. Fortunately, most patients have spontaneously remitting epilepsy and a normal development (Deprez et al., 2009). Dravet syndrome (DS) is an intractable epilepsy type with poor prognosis, characterized by fever-induced generalized tonic–clonic seizures during the first year of life. Phenotypic progression for DS patients includes complex seizure phenotypes with impairment of psychomotor development, including ataxia and mental disabilities (Deprez et al., 2009). Even though the outcome of these two different epilepsy syndromes differs dramatically, both are associated with the neuronal Na_v1.1 ion channel gene SCN1A (Escayg et al., 2000; Claes et al., 2003, 2009; Lossin, 2009).

GEFS+ patients harbor predominantly missense mutations, whereas in DS patients, missense, truncation, frameshift, or deletion/duplication mutations have been reported that arise de novo. To date, more than 600 SCN1A mutations are associated with a spectrum of epileptic phenotypes, and a few of these missense mutations have been functionally characterized at room temperature to gain insight into the pathophysiology of GEFS+ and DS (Spampanato et al., 2001, 2003, 2004; Lossin et al., 2002, 2003; Rhodes et al., 2004, 2005; Ohmori et al., 2006). When investigated at room temperature, GEFS+ mutant channels exhibit gain-of-function or loss-of-function gating defects, whereas complete loss of ion channel function is common in DS patients (Yu et al., 2006; Ogiwara et al., 2007; Catterall et al., 2010).

Fever-induced seizures are common in GEFS+ and DS subjects and mouse models (Oakley et al., 2009; Martin et al., 2010) and are typically the first reported symptom. However, no study has investigated the role of elevated temperature in the development of gating defects in these mutant channels. Previous functional studies on several other mutated ion channels associated with temperature-induced disorders have already shown that gating of these ion channels changes at elevated temperatures (Dumaine et al., 1999; Han et al., 2007; Amin et al., 2008; Carle et al., 2009; Samani et al., 2009). We have previously characterized GEFS+ (R859H) and DS (R865G) SCN1A mutant channels at room temperature. The R859H mutant showed mixed biophysical gating defects; the R865G mutant showed overall gain-of-function gating defects that could not explain the disease severity of the DS subject (Volkers et al., 2011). Because GEFS+ and DS share fever-associated seizures, we sought to investigate the temperature-dependent gating of the Na_v1.1 WT, R859H, and R865G sodium ion channels. We hypothesized that the R859H (GEFS+) and R865G (DS) mutations differentially affect the sodium channel function in a temperature-dependent manner. To test this hypothesis, we characterized the biophysical properties of WT and mutant ion channels at 37 and 40°C.

Human-derived tsA201 cells were cultured in Dulbecco’s modified Eagle’s medium supplemented with 10% FBS, 100 U/ml penicillin, 100 µg/ml streptomycin, and 0.05% l-glutamine in a humidified incubator at 37°C with 5% CO₂. All culture media and supplements were obtained from BioWhittaker. Transient transfections were performed with Lipofectamine (Invitrogen) using in total 6 µg of Na_v1.1, β1, and β2 pDNAs in a ratio of 10:1:1, as described previously by Lossin et al. (2002). Cells that expressed β1 and β2 subunits were identified with CD8-positive beads (Invitrogen) and GFP epifluorescence, respectively. Cells that expressed both β subunits were patched 48–72 h after transfection. For electrophysiological measurements, at least two different clones of either WT or mutant constructs were tested. The introduction of R859H (c.G2576A) and R865G (c.G2593G) mutations in the SCN1A plasmid DNA (provided by A.L. George Jr., Vanderbilt University, Nashville, TN) has been described previously (Volkers et al., 2011).

Before patching, cells were incubated for at least 1 h in cell culture medium containing 140 mM NMDG, 4 mM KCl, 1 mM CaCl₂, 1 mM MgCl₂, 14.3 mM Na₂HCO₃, 15.1 mM HEPES, 17.5 mM glucose, 1× amino acids (GIBCO), 1× nonessential amino acids (GIBCO), pH 7.35, supplemented with 10% FCS, 100 U/ml penicillin, 100 µg/ml streptomycin, and 2 mM l-glutamine (all from BioWhittaker) in a humidified atmosphere at 37°C, 5% CO₂. For patch experiments, cells were bathed in modified Tyrode’s solution containing (mM): 140 NaCl, 5.4 KCl, 1.8 CaCl₂, 1 MgCl₂, 6 glucose, and 6 HEPES, pH 7.4. Na_v1.1 sodium currents were measured at 37 and 40°C using the whole-cell patch-clamp configuration with an amplifier (Axopatch 200B; Molecular Devices). In most cells, electrophysiological parameters were measured at both 37 and 40°C. Patch pipettes were pulled from borosilicate glass capillaries with a pipette puller (P2000; Sutter Instrument) and fire-polished. Patch pipettes had a tip resistance between 1.3 and 2.2 MΩ when filled with the following internal solution (mM): 10 NaF, 110 CsF, 20 CsCl, 2 EGTA, and 10 HEPES, pH 7.35. Cells were allowed to stabilize at room temperature (20–22°C) for 10 min after whole-cell configuration was established. Next, temperature was raised to 37 or 40°C using a temperature-controlled perfusion chamber (Cell MicroControls), and cells were allowed to stabilize at each temperature for ∼7 min. Cells expressing sodium current less than −600 pA at room temperature were excluded because of small endogenous sodium currents possibly interfering with the electrophysiological measurements. Cells exhibiting currents greater than −6,000 pA were excluded to avoid voltage-clamp errors. Recordings were ≥90% compensated for pipette series resistance and capacitive transients. Leak currents were subtracted using a P/4 procedure. Sodium currents were filtered at 10 kHz and digitized at 100 kHz using a Digidata1322A (Molecular Devices). For analysis, a 5-kHz low pass–filtering step was used. Voltage-clamp protocols were generated and data acquisition was performed using pCLAMP9.2 (Molecular Devices). Voltage-clamp protocols are depicted in the figures where applicable. Steady-state activation was analyzed from consecutive measurements at 37 and 40°C. Current densities were determined by normalizing peak current amplitudes to cell capacitance. Voltage dependence of activation and slow inactivation were determined by fitting the data with a single Boltzmann equation, I = I_max/(1 + exp((V − V1/2)/k)), and the voltage dependence of fast inactivation with a double Boltzmann equation, I = I₁/(1 + exp((V − V₁1/2)/k₁)) + I₂/(1 + exp((V − V₂1/2)/k₂)), where V is the variable test or conditioning potential, V1/2 are the voltages of half-maximum activation or inactivation, k are slope factors, I₁ and I₂ are the fractional amplitudes of I_max, and I_max is the normalized maximum amplitude of the sodium current. The entry into and recovery from fast and slow inactivation was analyzed by fitting the data with a double exponential, I/I_max = A₁ × (1 − exp(−t/τ₁)) + A₂ × (1 − exp(−t/τ₂)) + C, where τ₁ and τ₂ denote the first and second phase of the inactivation time constant, A₁ and A₂ represent the fractional amplitude of each time constant, and C is the level of noninactivating sodium current. Speed of inactivation was evaluated by fitting the decay phase of the sodium current with a double-exponential function, I/I_max = A₁ × exp(−t/τ₁) + A₂ × exp(−t/τ₂) + C, where the parameters are defined as above. Persistent sodium current was determined by applying a 100-ms depolarizing pulse at voltages −20, −10, and 0 mV from a holding potential of −120 mV. Cells were recorded in the absence and presence of 10 µM tetrodotoxin (TTX; Alomone Labs), and traces were digitally subtracted to obtain persistent current. Next, the average persistent current at the final 10 ms of the depolarizing pulse was normalized to the peak sodium current. Use-dependency was investigated by applying 100 depolarizing test pulses of 2 ms to +15 mV starting from a resting membrane potential of −65 mV at various frequencies (10, 20, 33, 40, 64, 80, 100, 125, and 143 Hz). The peak current of the last pulse was normalized to the peak current generated by the first pulse. The Q₁₀ value of sodium current densities measured at −5 mV was calculated using the formula Q₁₀ = (I_40°C/I_37°C)^{(10/(T2−T1))}, where I_37°C and I_40°C represent the sodium current densities at the indicated temperatures, and T1 and T2 are 37 and 40°C, respectively. Voltages are not corrected for liquid junction potentials of 9.9 mV (37°C) and 10 mV (40°C). Data analysis was performed using Clampfit 9.2 (Molecular Devices), Excel (version 2008; Microsoft), KaleidaGraph (version 4.0; Synergy Software), and OriginPro (version 8.0; GE Healthcare) software. Results are presented as mean ± sem. Statistical analysis was performed using an unpaired Student’s t test. A value of P < 0.05 was considered significant.

Na_v1.1 WT, R859H, and R865G sodium currents were obtained by applying test pulses ranging from −80 to +90 mV from a holding potential of −120 mV. Fig. 1 (A–F) shows sodium current densities for WT, R859H, and R865G channels at room temperature (RT), 37°C, and 40°C. When the temperature was raised from 37 to 40°C, WT ion channels gained approximately −50 pA/pF in peak current density, and a similar increase in sodium current density at elevated temperature was also present in the R859H mutant (Fig. 1, A and B). In contrast, the R865G mutation caused no significant increase at 40°C (Fig. 1 C). Compared with WT, R865G showed a trend toward a reduction in sodium current density at physiological temperature and a significant reduction at voltages −10 to 30 mV at 40°C (Fig. 1, E and F). This reduction was not detected at room temperature (Fig. 1 D). Thus, the R865G mutant shows a loss-of-function defect in current density at febrile temperature. We calculated the Q₁₀ value of the temperature-dependent current density changes between 37 and 40°C at −5 mV. The Q₁₀ value of the R865G mutant (1.17 ± 0.14; t₂₁ = 2.58; P = 0.017; n = 10) was significantly lower than WT (2.38 ± 0.40; n = 13) or R859H (2.13 ± 0.32; n = 12) sodium channels, suggesting that the R865G mutant is less temperature dependent than WT.

To determine whether the speed of activation or inactivation plays a role in the reduced current density of the R865G mutant, we analyzed the activation time by measuring the time from the onset to the peak of sodium currents (time to peak current). Fitting the decay phase of the sodium currents with a double-exponential function to obtain the first and second phase of the fast inactivation time constants determined the inactivation rate. Febrile temperatures increased the rate of activation (decreased time to peak current) of WT channels, whereas the speed of inactivation was unaffected (Fig. 2, A, D, and I). The R859H mutant showed similar activation rates at both temperatures, a significant decrease in time constants for the first phase of fast inactivation at 40°C at membrane potentials between −30 and −20 mV and a trend toward increased speed of inactivation for the second phase of fast inactivation at all voltage ranges (Fig. 2, B, E, and J). The kinetics of the R865G mutant were unaffected by febrile temperature (Fig. 2, C, F, and K) and were comparable to the activation rate and speed of inactivation of WT channels at 40°C (Fig. 2, H and M), indicating that neither time to peak nor speed of inactivation contributes to the reduced current density of R865G at elevated temperatures.

Next, we analyzed the voltage dependence of activation. Febrile temperature caused a depolarizing shift in the voltage dependence of activation of WT ion channels (Fig. 3 A and Table 1). The half-maximal activation (V1/2) at 40°C of R859H was similar to 37°C, whereas the R865G mutant showed a significant depolarizing shift in the voltage dependence of activation at 40°C (Fig. 3, B and C, and Table 1). Compared with WT, however, there was a significant hyperpolarized shift in the voltage dependence of activation of R859H (V1/2 37°C = −28.3 ± 1.0, t₂₃ = 2.45, P = 0.02, and n = 12; V1/2 40°C = −28.1 ± 1.2, t₂₃ = 3.41, P = 0.002, and n = 12) and R865G (V1/2 37°C = −32.9 ± 0.6, t₂₁ = 8.02, P < 0.0001, and n = 10; V1/2 40°C = −31.4 ± 0.6, t₂₃ = 7.09, P < 0.0001, and n = 10) at both temperatures, albeit more prominent for the DS mutation (Fig. 3, D–F). On the other hand, the V1/2 at 37 and 40°C of WT, R859H, and R865G is similar to the values obtained at room temperature (Na_v1.1 WT: −23.1 ± 1.1 mV; R859H: −27.1 ± 0.6 mV; R865G: −31.2 ± 1.1 mV; Volkers et al., 2011), indicating that increased temperatures do not affect the activation gating of the Na_v1.1 WT ion channel and mutants.

Subsequently, we investigated steady-state inactivation by applying a series of depolarizing conditioning pulses in the voltage range of −140 to 0 mV starting from a holding potential of −120 mV, immediately followed by a depolarizing test pulse of −20 mV. Steady-state channel inactivation was best fitted by a double Boltzmann equation. When temperature was raised to 40°C, WT channels showed an increase in channel availability (Fig. 3 A and Table 1). Opposite results were obtained for the R859H mutant, which showed a significant hyperpolarized shift in the voltage dependence of inactivation at elevated temperatures (Fig. 3 B and Table 1). Furthermore, the contribution of the more negative portion of the Boltzmann curve (A₂) for R859H was significantly increased compared with WT channels at 37°C (A₂ R859H = 0.35 ± 0.02; t₁₇ = 2.11; P = 0.049; n = 9) (Table 1). The temperature-dependent negative shift in the steady-state inactivation of sodium ion channels harboring the R859H mutation suggests that fever leads to a decrease in channel availability (Fig. 3, B, E, and G). In contrast, elevated temperatures did not affect the voltage dependence of inactivation of the R865G mutant (Fig. 3 C and Table 1). However, the R865G mutant at 37 and 40°C revealed an increase in the voltage dependence of channel availability compared with WT channels (V₁1/2 37°C = −58.4 ± 1.4, t₁₆ = −5.45, P < 0.0001, and n = 9; V₁1/2 40°C = −57.8 ± 1.3, t₁₆ = −2.14, P < 0.048, and n = 9) (Fig. 3, D, E, and G), a feature that was not observed at room temperature (Volkers et al., 2011).

The steady-state activation and inactivation curves showed an overlap, indicating that a window current was present. Compared with WT, the window current of R865G was enlarged because of the shifts in the voltage dependence of activation and inactivation (Fig. 3, A, C, F, and G). We have therefore analyzed changes in the window current in more detail. The size of the window current produced by WT, R859H, and R865G channels was calculated as the product of the average steady state of activation (m) and inactivation (h). These resultant products are a measure of the probability of incompletely inactivated or deactivated sodium channels to open and were plotted against the membrane potential. The size of the window current of R865G was increased >10-fold at both 37 and 40°C compared with WT (Fig. 4, A–C). This increased window current results in an increase in channel function, especially when the membrane potential is near firing threshold, which could lead to enhanced excitability.

Febrile temperatures accelerated the recovery from inactivation of both WT and mutant channels. However, this effect was less pronounced in both mutants (Fig. 5, A–E, and Table 1). At 37°C, cells expressing R859H had a significantly smaller fraction of channels recovering from the fast inactivation component (A₁) and a larger population recovering from the slower inactivation (A₂) compared with WT (A₁ = 0.48 ± 0.07, t₂₃ = 2.82, P = 0.01, and n = 12; A₂ = 0.68 ± 0.04, t₂₃ = −3.96, P = 0.0006, and n = 12) (Table 1). In addition, there was a significant slowing in recovery from inactivation, as was evident by a larger time constant (τ₂) of recovery (τ₂ = 128 ± 20.8; t₂₃ = −8.63; P < 0.0001; n = 12). At 40°C, the recovery from inactivation of R859H is accelerated but still significantly slower compared with WT channels (τ₁ R859H = 23.8 ± 8.80; t₁₉ = −3.23; P = 0.004; n = 9) (Fig. 5 E). The R865G mutation also caused an overall slowing in recovery from inactivation at 37 and 40°C, as shown by the larger time constants τ₁ (τ₁ t₁₇ = 2.11; P = 0.049) (Table 1 and Fig. 5 D). At 40°C, a smaller R865G channel population was recovering from the fast inactivation component (A₁ t₁₇ = 2.11; P = 0.049) and a significant larger ion channel population was present in the second inactivation status (A₂ t₁₇ = 3.24; P = 0.048) (Table 1). In summary, these results show that both mutants exhibited a slowing in recovery from inactivation at both temperatures, with the largest delay observed for R859H (Fig. 5, A–E).

Slow inactivation is a process distinct from fast inactivation. This independent process is functionally relevant for setting the fractional availability of the sodium channel population at voltages around the resting membrane potential or in response to prolonged bursts of action potentials.

The entry into inactivation was analyzed by fitting the data to a double-exponential function (Fig. 6, A–C). Increased temperatures accelerated the entry into inactivation of WT, R859H, and R865G channels, which is evident by decreased time constants τ₂ (Table 2). Compared with WT ion channels, at 37°C, both mutants have a larger inactivating ion channel population in the first time fraction (A₁) (A₁ R859H = 0.29 ± 0.06, t₁₅ = 2.24, P = 0.04, and n = 9; A₁ R865G = 0.32 ± 0.08, t₁₅ = 2.15, P = 0.049, and n = 9) than WT ion channels (Fig. 6, A–C, and Table 2). At both temperatures, R859H had a larger ion channel population that entered into a slow inactivation state compared with WT, as evident by the smaller value of the residual current (I_residual 37°C = 14 ± 3, t₁₅ = 2.89, P = 0.01, and n = 9; I_residual 40°C = 14 ± 3, t₁₅ = 2.10, P = 0.048, and n = 12) (Fig. 5, A and B, and Table 2). At 40°C, the I_residual of R865G is similar to WT (t₁₈ = 0.57; P = 0.57; n = 10) (Table 2). Compared with WT, the R865G mutant showed an acceleration of entry into steady-state inactivation (τ₂) at both temperatures (τ₂ 37°C: t₁₅ = 2.69, P = 0.02, and n = 9; τ₂ 40°C: t₁₈ = 4.34, P = 0.0004, and n = 10).

Fig. 6 (D–F) shows the recovery from slow inactivation for WT, R859H, and R865G. Compared with WT, both mutations caused a significant slowing in recovery from slow inactivation at both temperatures (Fig. 6, D–F, and Table 2), albeit more prominent for R859H.

At 40°C, both WT and R859H channels showed a hyperpolarized shift in the voltage dependence of steady-state slow inactivation (Fig. 7, A and B). The steady-state slow inactivation of R859H channels at both temperatures was comparable to WT, whereas the R865G mutation caused a hyperpolarized shift in the V1/2 of −10 and −7.7 mV at 37 and 40°C, respectively (V1/2 37°C = −64.3 ± 0.9, t₁₆ = 6.96, P < 0.0001, and n = 7; V1/2 40°C = −65.6 ± 0.8, t₁₅ = 5.26, P < 0.0001, and n = 10) (Fig. 7, A–F, and Table 2).

We tested the use-dependent behavior of WT and mutant ion channels by applying a train of 2-ms depolarizing test pulses at various frequencies to mimic repetitive action potential firing. The use-dependency was analyzed by normalizing the peak sodium currents during the last test pulse to currents generated during the first pulse (Fig. 8). When temperature was increased to 40°C, WT ion channels showed a trend toward decreased use-dependent decay in sodium currents (Fig. 8 A). In contrast, the R859H mutation caused a significant increase in use-dependence in the frequency range of 64–143 Hz (Fig. 8 B), whereas R865G showed a similar use-dependent behavior at both temperatures (Fig. 8 C).

At 37°C, the use-dependency of R859H was significantly less use-dependent than WT and R865G, whereas R865G was similar as WT (Fig. 8 D). At 40°C, R859H channels showed similar use-dependency as WT channels. The R865G mutation caused an increase in use-dependent behavior at frequencies of 20–40 Hz and 80–143 Hz compared with WT sodium channels (Fig. 8 E).

In our previous characterization at room temperature, both mutants exhibited a significant increase in persistent current (Volkers et al., 2011). Because persistent current is rather common in GEFS+ and DS mutant channels at room temperature (Lossin et al., 2002; Rhodes et al., 2004), we examined the persistent current of these mutant channels at physiological and febrile temperatures. In contrast to our previous findings, both mutants had comparable small persistent currents as WT channels at elevated temperatures (Fig. 9, A–C).

Molecular defects in Na_v1.X ion channels have been reported in association with temperature-sensitive disorders such as heat-induced myotonia and cold-induced paramyotonia linked to Na_v1.4 mutations; Brugada syndrome associated with Na_v1.5 mutations; and erythromelalgia, a disabling chronic pain disorder associated with mutations in the Na_v1.7 ion channel. Na_v1.4, Na_v1.5, and Na_v1.7 sodium ion channels are thermosensitive, and mutations in these ion channel proteins can alter the temperature-dependent gating of the channel (Dumaine et al., 1999; Han et al., 2007; Webb and Cannon, 2008; Carle et al., 2009; Samani et al., 2009).

Fever-induced seizures are common in GEFS+ and DS subjects, and GEFS+ and DS mouse models showed increased seizure susceptibility at elevated body temperatures (Oakley et al., 2009; Martin et al., 2010). The physiological mechanisms behind the temperature-dependent reduced seizure threshold are unclear. Reversible temperature-dependent trafficking defects in the GABA_A γ2-subunit receptor—associated with GEFS+—have been reported, which suggest that accelerated endocytosis could be caused by an abnormal folding of the receptor protein at elevated temperatures (Kang et al., 2006). Abnormal folding of the Na_v1.1 ion channel has also been shown for two SCN1A mutations at room temperature (Rusconi et al., 2007, 2009). This suggests that a lowered inhibition by GABAergic interneurons at febrile temperatures is a contributing factor to febrile seizures in individuals harboring a Na_v1.1 mutation. DS Scn1a mouse models already show drastically reduced sodium currents in GABAergic interneurons that cause action potential attenuation during repetitive firing (Yu et al., 2006; Ogiwara et al., 2007). In addition, Scn1a heterozygote animals also show markedly different effects depending on the genetic background, with one strain having a more severe phenotype compared with other strains (Yu et al., 2006). Also, truncation mutations in SCN1A in patients can lead to febrile seizures or GEFS+ instead of DS (Gennaro et al., 2003; Yu et al., 2010). These observations reflect the role of genetic background or modifier genes in the phenotypic outcome of DS patients. Additionally, the role of acute fever in triggering the seizure remains to be clarified.

Previous work on the R859H (GEFS+) and R865G (DS) mutant channels studied at room temperature demonstrated a gain of function in the voltage dependence of activation and a loss of function in recovery of fast inactivation for both mutants (Volkers et al., 2011). In addition, the R859H mutant also showed a loss of function in the voltage dependence and speed of inactivation. The current densities of I_Na carried by both mutated channels were comparable to WT. Although the slow inactivation behavior of these channels was not investigated, the data recorded at room temperature suggest only small gating defects for the R865G mutant that are insufficient to explain the severe phenotype of this patient. Therefore, in this study, we characterized the biophysical behavior of these two mutants, R859H and R865G, at physiological and febrile temperature to mimic the pathophysiological condition at which these epileptic seizures are triggered.

Physiological and febrile temperatures (37 and 40°C) unmasked a depolarized shift in the voltage dependence of inactivation (gain of function) of the R865G channel and a significant decrease in current density compared with WT channels, which was not detected previously at room temperature (Volkers et al., 2011). Opposite results were obtained for R859H channels, where loss of function in voltage-dependent channel availability found at room temperature was virtually abolished at 37°C but unmasked again at 40°C. In addition, the reduced speed of inactivation detected at room temperature was abolished at 37°C and increased at 40°C at a range of −30 to −20 mV. These results emphasize that caution should be used when extrapolating ion channel gating properties obtained at room temperature to 37 or 40°C using a single scale factor. Moreover, our results underline the need of electrophysiological measurements at physiological and febrile temperatures to gather a better insight in the biophysical behavior of Na_v1.1 WT and mutated ion channels.

At room temperature, the R859H and R865G mutations caused an enlarged persistent current (Volkers et al., 2011). Interestingly, this increase in persistent current was abolished completely at 37 and 40°C. This suggests that the stabilization of the inactivation gate of Na_v1.1 channels is highly temperature sensitive and that these mutations hamper complete channel inactivation at lower temperatures. Whether the increased persistent sodium currents measured in other mutants associated with GEFS+ and DS (Lossin et al., 2002; Rhodes et al., 2004) can still be relevant at physiological and febrile temperatures remains to be determined.

Compared with WT and R859H channels, R865G shows a >10-fold increase in window current at both temperatures. The window current occurs at a small range of voltages where sodium ion channels activate and inactivate incompletely. This activity can play an important role in the excitability of a neuron, and an increase in window current is predicted to increase channel function. After correction of liquid junction potentials of the patch pipettes (37°C = 9.9 mV; 40°C = 10 mV), the maximal window current is in range of the resting membrane potential of a neuron (approximately −60 mV). This could produce neuronal hyperexcitability by reducing the threshold for initiating action potential firing in an inhibitory neuron that expresses R865G mutant sodium channels. However, the significant window current observed in the DS mutant may depolarize the membrane and/or increase intracellular Na⁺ to the point of inducing depolarized block of action potential firing or alterations in action potential shape. This would reflect enhanced steady-state inactivation of R865G sodium channels and/or reduce the ionic driving force for Na⁺ compared with WT neurons. Reduced excitability of GABAergic interneurons has been observed in brain slices and dissociated GABAergic interneurons of the Scn1a^+/− DS mouse model (Yu et al., 2006; Ogiwara et al., 2007).

The voltage dependence of fast inactivation at 37°C revealed two distinct gating processes in Na_v1.1 WT and mutant ion channels. The inactivation process present at a more negative voltage range is maybe not very significant because of the hyperpolarized range at which it occurs (Table 1; see V₂1/2). Interestingly, the voltage dependence of fast inactivation of WT Na_v1.4, Na_v1.5, and Na_v1.7 sodium channels at elevated temperatures shows only one inactivation process (Han et al., 2007; Carle et al., 2009; Samani et al., 2009). This could indicate that the temperature-dependent gating characteristics of Na_v1.1 WT ion channels differ between sodium ion channel isoforms.

A DS-like phenotype can be caused by decreased sodium currents in GABAergic inhibitory interneurons, as shown in Scn1a^+/− DS mouse models (Yu et al., 2006; Ogiwara et al., 2007). In our experiments, the DS mutant R865G also showed an acute reduction in current density at a febrile temperature, although the I_Na reduction is not as complete compared with the knockout models. Nevertheless, the reduced neuronal inhibition could contribute to the epileptic severity in DS subjects harboring an R865G mutation. In contrast, at febrile temperatures, the WT sodium currents increase in magnitude and channel availability increases in parallel during repetitive stimulation (gain of function), both of which predict increased GABAergic inhibition and seizure suppression. This divergence in the inhibitory drive of the GABAergic network for WT versus R865G could inform as to the susceptibility of DS patients to febrile convulsions.

In summary, our study indicates that febrile temperatures differentially change the gating of WT and mutated Na_v1.1 channels. For the GEFS+ mutant (R859H), we found a temperature-dependent loss of function in the voltage dependence of inactivation and a use-dependent decrease in channel availability during repetitive firing at 40°C. We hypothesize that the decreased voltage-dependent channel availability in combination with an increased use-dependency contributes to the development of febrile seizures in GEFS+ patients harboring a R859H mutation. In the DS mutant R865G, despite the gain of function in the voltage dependence of activation and inactivation at 37 and 40°C, febrile temperatures reduced peak sodium currents and increased use-dependency compared with WT. These data suggest that fever-induced gating defects in combination with the biophysical ion channel alterations at elevated temperatures could contribute to the phenotypic outcome in DS patients with a R865G mutation. In addition, the faster entry into the slow inactivation state in combination with a loss of function in the voltage dependence of the slow inactivation of R865G at febrile temperatures suggests that prolonged membrane depolarization can ultimately lead to membrane inexcitability in GABAergic inhibitory neurons expressing this mutant, and could explain the severe outcome in this DS subject. The experimental approaches used in this study represent a simplified disease model, and extending these findings to a neuronal or in vivo system is a necessary progression. Indeed, the extensive characterization of the GEFS+ Na_v1.1 mutant channel (R1648H) has used a variety of expression systems as well as a transgenic mouse model to measure different or even opposing gating defects (Spampanato et al., 2001; Lossin et al., 2002; Tang et al., 2009). In contrast, the gating defects of N1417H-mutated Na_v1.1 channels reported using HEK cells were confirmed in GABAergic interneurons of rats carrying this mutation (Mashimo et al., 2010). These reports highlight the important role played by cellular background, including accessory proteins, compensatory mechanisms, and channel trafficking. Future studies characterizing R865H and R865G, as well as other mutant Na_v1.1 channels, in neuronal networks and animal models are necessary to determine how our observations generalize to SCN1A epileptic syndromes.

This work was supported by “Ter Meulen Fund,” Royal Netherlands Academy of Arts and Sciences (to L. Volkers); The Netherlands Organization of Scientific Research and Development (ZonMW; grant 917.66.315 to B.P.C. Koeleman); and the National Epilepsy Fund of The Netherlands (grant NEF 07-21 to M.J.A. van Kempen). The authors declare no competing financial interests.

Lawrence G. Palmer served as editor.