Effect of phenytoin on sodium conductances in rat hippocampal CA1 pyramidal neurons

Phenytoin is an effective antiepileptic drug. It has commonly been assumed that phenytoin has a specific affinity for the fast-inactivated state of the sodium channel. The current study found no effect on fast inactivation but found that forms of slow inactivation were enhanced by phenytoin. Additionally, a theoretical study is presented showing that the effects of phenytoin can be explained through an interaction with the activated voltage sensor.

Abstract

The antiepileptic drug phenytoin (PHT) is thought to reduce the excitability of neural tissue by stabilizing sodium channels (Na_V) in inactivated states. It has been suggested the fast-inactivated state (I_F) is the main target, although slow inactivation (I_S) has also been implicated. Other studies on local anesthetics with similar effects on sodium channels have implicated the Na_V voltage sensor interactions. In this study, we reexamined the effect of PHT in both equilibrium and dynamic transitions between fast and slower forms of inactivation in rat hippocampal CA1 pyramidal neurons. The effects of PHT were observed on fast and slow inactivation processes, as well as on another identified “intermediate” inactivation process. The effect of enzymatic removal of I_F was also studied, as well as effects on the residual persistent sodium current (I_NaP). A computational model based on a gating charge interaction was derived that reproduced a range of PHT effects on Na_V equilibrium and state transitions. No effect of PHT on I_F was observed; rather, PHT appeared to facilitate the occupancy of other closed states, either through enhancement of slow inactivation or through formation of analogous drug-bound states. The overall significance of these observations is that our data are inconsistent with the commonly held view that the archetypal Na_V channel inhibitor PHT stabilizes fast inactivation states, and we demonstrate that conventional slow activation “I_S” and the more recently identified intermediate-duration inactivation process “I_I” are the primary functional targets of PHT. In addition, we show that the traditional explanatory frameworks based on the “modulated receptor hypothesis” can be substituted by simple, physiologically plausible interactions with voltage sensors. Additionally, I_NaP was not preferentially inhibited compared with peak I_Na at short latencies (50 ms) by PHT.

NEW & NOTEWORTHY

Phenytoin is an effective antiepileptic drug. It has commonly been assumed that phenytoin has a specific affinity for the fast-inactivated state of the sodium channel. The current study found no effect on fast inactivation but found that forms of slow inactivation were enhanced by phenytoin. Additionally, a theoretical study is presented showing that the effects of phenytoin can be explained through an interaction with the activated voltage sensor.

phenytoin (PHT) is an antiepileptic drug (AED) commonly used for a range of epileptic seizures (Browne and Holmes 2001). It is thought to achieve its therapeutic effect mainly by modulating the availability of voltage-gated sodium channels (Na_V; Lipicky et al. 1972; Mclean and Macdonald 1983), which are responsible for the initiation of action potentials in central nervous system neurons. Depolarization of the neuronal membrane results in a rapid phase of “activation” with channels entering open, conductive states. This activation phase is followed by a reduction in sodium current amplitude termed “fast inactivation” (I_F), related to motion of the D3–D4 cytoplasmic linker (Bezanilla 2008). PHT selectively inhibits high-frequency firing (Macdonald and McLean 1986) and also preferentially blocks sustained repetitive firing (Lampl et al. 1998; Ragsdale and Avoli 1998). These effects have been considered to be a result of differential binding affinity of the drug to the different kinetic states of the channel, with highest affinity for inactivated states (the “modulated receptor hypothesis;” Hille 1977). However, although the fast-inactivated state is generally considered the major target (Kuo and Bean 1994), PHT reduces sodium current amplitude in neuroblastoma cells after enzymatic removal of I_F (Quandt 1988). The relatively slow onset and offset of PHT action compared with the kinetic properties of I_F have been ascribed to slow binding of the drug to this state (Kuo and Bean 1994; Rogawski and Löscher 2004) rather than effects on slow inactivation (I_S) as such.

Additionally, several studies have reported a stronger inhibition of the persistent sodium current (I_NaP) compared with the transient component by PHT. This would be expected to reduce repetitive firing (French et al. 1990), augmenting the seizure suppressive action of PHT (Chao and Alzheimer 1995; Colombo et al. 2013).

In the present study we have examined the effects of PHT on equilibrium (steady-state inactivation distributions) and on rates of transitions between closed and closed-inactivated states for I_F and I_S, as well as an inactivation process with time constants intermediate to these processes, termed “I_I.” In addition, the effect of PHT on I_Na was examined after removal of fast inactivation with papaine. PHT did not significantly affect either transition or equilibrium properties of I_F but did affect slower forms of inactivation. I_NaP was not preferentially inhibited by PHT compared with the transient sodium current, I_NaT.

METHODS

All procedures were carried out in accordance with protocols approved by University of Melbourne Animal Ethics Committee.

Solutions and drugs.

Stock aliquots of PHT (50 mM) in DMSO were stored at −20°C and thawed at the time of experiment. Apart from the dose-response experiments, a concentration of 50 μM PHT was used, within the serum therapeutic range (40–80 μM; Kozer et al. 2002). Brain slices were prepared in a high-sucrose bicarbonate-buffered “cutting solution” containing (in mM) 87 NaCl, 2.5 KCl, 25 NaHCO₃, 1.25 NaH₂PO₄, 0.5 CaCl₂, 7 MgCl₂, 25 d-glucose, and 75 sucrose, bubbled with 95% O₂-5% CO₂. Brain slices were maintained in bicarbonate-buffered saline containing (in mM) 125 NaCl, 2.5 KCl, 25 NaHCO₃, 1.25 NaH₂PO₄, 2 CaCl₂, 1 MgCl₂, and 25 d-glucose, bubbled with 95% O₂-5% CO₂ at room temperature. Recordings from dissociated neurons were performed in either HEPES-buffered saline containing (in mM) 125 NaCl, 2.5 KCl, 2 CaCl₂, 1 MgCl₂, 25 d-glucose, and 10 HEPES, with pH adjusted to 7.4 with NaOH, or low-sodium-buffered saline containing (in mM) 35 NaCl, 2.5 KCl, 2 CaCl₂, 1 MgCl₂, 10 d-glucose, 4 4-aminopyridine (4-AP), 55 Tris·HCl, 40 TEA, and 0.1 CdCl₂, with pH adjusted to 7.4 with Tris base at 22°C. Protease type XIV (Sigma-Aldrich) was dissolved in PIPES-buffered saline containing (in mM) 115 NaCl, 5 KCl, 20 PIPES, 1 CaCl₂, 4 MgCl₂, and 25 d-glucose, and pH was adjusted to 7.0 with NaOH. Recording electrodes were filled with intracellular CsF-based patch-pipette solution containing (in mM) 85 CsF, 35 NaCl, 1 CaCl₂, 1 MgCl₂, 10 EGTA, 10 HEPES, and 20 TEA; pH was adjusted to 7.2 with CsOH and osmolality was 280–300 mosM.

Dissociated neuron preparation.

Two- to six-week-old Wistar rats (male and female) were deeply anesthetized with pentobarbital sodium (Lethabarb) and decapitated with rapid brain removal. Coronal brain slices 400 μm thick were sectioned with a vibratome (Microm) at 4°C in cutting solution saturated with 95% O₂-5% CO₂. Hippocampal slices were subsectioned and then equilibrated in bicarbonate-buffered saline at room temperature for 30 min. Small segments (∼1 mm²) of the CA1 region were treated with protease type XIV (1 mg/ml) in PIPES-buffered saline that was constantly perfused with 95% O₂-5% CO₂ for another 30 min at 32°C in a temperature-controlled chamber. Treated slices were washed with HEPES-buffered saline or, where appropriate, with low-sodium-buffered saline before being triturated to isolate individual neurons in the recording chamber.

Electrophysiological recording.

Pyramidal neurons were identified visually. Patch pipettes were made from borosilicate glass tubing (1.5-mm outer diameter, 0.86-mm inner diameter; Harvard Apparatus) with a Sutter P-1000 electrode puller. Neurons were approached with patch pipettes under visual control with positive pressure. On contact the cell membrane, the positive pressure was canceled for seal formation. Access to the cell interior was obtained by applying further suction. The resistance of the recording pipette was between 1 and 2 MΩ with the use of whole cell voltage-clamp configurations. Current and voltage were recorded using a Molecular Devices Axopatch 200B or AM Systems model 2400 patch-clamp amplifier with ∼90% predictive and compensatory series resistance compensation. Capacitive transients were minimized by maintaining the bath solution at a low level and were further reduced by the capacitive compensation circuit of the amplifier. Signals were digitized at between 100 and 200 kHz with Axon 1320 (Molecular Devices, Sunnyvale, CA) or National Instruments devices (NI USB-6259) and low-pass filtered at 10 kHz with a 4-pole Bessel filter. The digitizer was connected to a personal computer for stimulus generation and data acquisition with pClamp 10 (Molecular Devices) or Strathclyde WCP version 4.1.7 software (University of Strathclyde, Glasgow, Scotland). A holding potential of −100 mV was routinely used unless otherwise noted. Potassium channels were blocked by external TEA (20 mM) and Cs ions in the patch electrode, and calcium channels were blocked by Cd²⁺ (0.5 mM) externally. Leakage and capacitive currents were subtracted with a P/4 protocol. Experiments were performed at 22°C with routine verification of chamber temperature. Control and drug containing solutions were applied via a custom-made gravity-flow manifold system, permitting solution changes within ∼2 s, using a 100-μm-diameter outflow tube positioned ∼500 μm from the recorded cell. A liquid junction potential of ∼10 mV was calculated for the pipette and bath solutions (Barry 1994), but potential measurements have not been adjusted for this offset.

Analysis and statistics.

Data analysis was performed with Molecular Devices Clampfit 10, GraphPad Prism, and Microsoft Excel software packages. Statistical errors are presented as standard error of the mean (SE) unless otherwise stated. Rates of onset and offset of inactivation were measured using exponential functions. Nonlinear least-squares parameter estimations were performed with the Marquardt-Levenberg algorithm as implemented in Clampfit and Prism, and with a quadratic minimization algorithm in the Microsoft Excel Solver package. Comparisons between single- and double-exponential data fits were analyzed for significance by using F statistics as implemented in Prism and Clampfit. Student's t-tests were used for comparison of the means with a P value <0.05.

Steady-state inactivation (h_inf) curves were parameterized by a Boltzmann function,

$$h_{\inf }=1/\left[1+e^{\left(V-V^{\prime }\right)/k}\right],$$

where V is the command potential (mV), V′ is the half-inactivation potential (or half-activation potential; mV), and k is the slope factor (mV). A similar equation was used for steady-state activation (m_inf).

Computational modeling.

Modeling of the Na_V kinetics was carried out with JSim (Butterworth et al. 2014) and MATLAB (The MathWorks) by generating Markov state models of the sodium channel that included slow-inactivated states.

RESULTS

PHT does not affect I_Na activation.

The effect of PHT on activation of I_Na was tested with step depolarizations before and after application of 50 μM PHT. Representative current traces with depolarizations from −80 mV to +50 mV with 10-mV increments are shown in Fig. 1A, and the peak normalized currents were fitted with a Boltzmann distribution. Activation occurred at −60 mV, with maximal conductance at 10 mV. The half-activation potential (V_0.5) was −32.1 ± 3.3 mV with a slope factor of 7.6 mV. PHT (50 μM) had no significant effect on V_0.5 or slope. Time to peak, reversal potential, and residual persistent current (I_NaP) were also unaffected by 50 μM PHT (Fig. 1, B and C). The effect on I_NaP is described in greater detail below.

Fig. 1.

Current-potential (I-V) and conductance-potential (G-V) relation before and after 50 μM PHT. A: I-V voltage-clamp protocol is shown at left. Cells were held at −80 mV, with a series of 10-mV increments to 50 mV for 50 ms in duration. Representative current trances shown at right were elicited by the I-V protocol. B: I-V relation plotted as a graph for control (blue circles) and 50 μM PHT (red triangles) (n = 7). CON, control; V_command, command potential. C: G-V relation derived from the I-V relation in B and fitted with Boltzmann's function. Half-activation potentials (V_0.5) were found to be −32.1 ± 3.3 and −32.6 ± 2.7 mV in control and PHT, respectively. Slope factors (k) were 7.6 ± 1.6 and 9.5 ± 1.8 mV in control and PHT, respectively. No statistical significance was found in V_0.5 (P = 0.7) and k (P = 0.1) using paired t-tests.

Application of PHT led to reduction of amplitude of I_Na in a voltage-dependent manner as previously described (Matsuki et al. 1984; Ragsdale et al. 1991; Willow et al. 1985). At a holding potential of −100 mV, I_Na was typically decremented by 20% after application of PHT (50 μM; Fig. 2), and this inhibition was enhanced at more positive holding potentials. The time course of fast inactivation, as measured by the decrement of the macroscopic current, was a biexponential process as described previously (Sah et al. 1988). Neither the time constants nor relative amplitudes of fast inactivation were affected by 50 μM PHT (Fig. 2).

Fig. 2.

Changes in I_NaT amplitude and macroscopic fast inactivation (I_F) with 50 μM PHT. Representative I_NaT were generated by depolarization to −20 mV from holding potentials at −120 (A), −100 (B), and −80 mV (C) before (solid line) and after (dashed line) 50 μM PHT exerted effects. The percent reductions in I_NaT amplitude were 14.1 ± 6.0% (n = 5), 23.8 ± 3.4% (n = 6), and 38.7 ± 9.5% (n = 6) at −120, −100, and −80 mV, respectively. D: a closer look at the macroscopic I_F reveals double-exponential components both before (solid line) and after (dashed line) PHT effect (holding potential −100 mV). There were no significant changes in both time constant components (paired t-test, P = 0.4 for the slow component and P = 0.3 for the fast component) when 50 μM PHT was applied (n = 6).

Dose response of PHT on I_Na.

Concentration dependence of PHT on I_Na inhibition was tested with 50-ms-long depolarizations to −20 mV from a holding potential of −80 mV, activated every 2 s as control and drug-containing solutions were applied to the neuron at concentrations from 1 to 200 μM. Peak I_Na amplitude was normalized to the control solution amplitude and then plotted against the log PHT concentration. An IC₅₀ value of 72.6 ± 22.5 μM was calculated (Fig. 3).

Fig. 3.

Dose-response curve of PHT in CA1 hippocampal pyramidal neurons. Fraction of the maximal response is plotted against the log concentration of PHT (n = 6). A representative I_NaT in control solution as well as in different concentrations of PHT is shown in the inset with the protocol. The holding potential was −80 mV. The IC₅₀ was found to be 72.6 ± 22.5 μM.

Effect of PHT on steady-state inactivation with conditioning prepulses of different durations.

The equilibrium distribution of channels between closed and closed-inactivated states is conventionally measured by the steady-state inactivation (h_inf) relation (Hodgkin and Huxley 1952). A minimum prepulse duration of approximately three to five time constants is needed for the channels to reach equilibrium; however, longer pulses may start to engage other slower processes. Therefore, a prepulse duration of 50 ms was used (Fig. 4C) for the I_F h_inf curve, based on the kinetics measured by Kuo and Bean (1994) as well as our own recordings to isolate the fast inactivation process. Additionally, because a component of inactivation with time constants of seconds had been observed previously (Kuo and Bean 1994) corresponding to previously described “slow inactivation,” or “I_S” (Rudy 1978), h_inf curves were generated with a prepulse duration of 10 s (Fig. 4E). We also used 500-ms-duration prepulses (Fig. 4D), because components of inactivation with time constants of the order of 10² ms have been seen in neuronal preparations (Colbert et al. 1997; Fleidervish et al. 1996), including in this laboratory (French et al. 2016). Interestingly, simply increasing the duration of the prepulses from 50 to 500 ms (Fig. 4D) resulted in an ∼10-mV negative shift of the midpoint of the h_inf curve, suggesting inactivation recruitment over this time period.

Fig. 4.

Steady-state inactivation of sodium current with 50 μM PHT. Data points with SE are fitted to Boltzman distributions. A: representative current traces of the steady-state inactivation protocol are shown. Δt, protocol duration. B: in control solution, increasing conditioning pulse duration (50 ms, 500 ms, and 10 s) produced hyperpolarizing shifts (see text for values). C–E: effects of 50 μM PHT on 50-ms (C), 500-ms (D), and 10-s (E) prepulse durations are shown. No significant change in V_0.5 was found with 50-ms prepulse (P = 0.97, n = 6), but significant changes were found with 500-ms (7 mV; P < 0.001, n = 3∼6) and 10-s (8 mV; P < 0.001, n = 3) conditioning pulse durations. WO, washout.

The I_F h_inf relation had a half-inactivation voltage (V_0.5) of −69.8 ± 2.2 mV and a slope of 8.6 ± 0.4. PHT (50 μM) had no significant effect on these parameters (Fig. 4C). The 500-ms and 10-s V_0.5 were shifted to negative potentials (−80.2 ± 4.1 and −84.8 ± 3.6 mV, respectively) compared with the I_F relation in control solution. Application of 50 μM PHT caused a negative shift of the 500-ms and 10-s relations by 7 and 8 mV, respectively, in marked distinction to the I_F h_inf relation.

Effect of PHT on fast inactivation transition rates.

Rates of transition into the I_F state were unaffected as measured by the rate of decay of I_Na activated by step depolarization to −20 mV (Fig. 2D). Double-pulse experiments were then used to further explore the rates of transition to and from fast-inactivated states before and after PHT application. Entry into fast inactivation was assessed by measuring the decrement of I_Na in response to a series of depolarizing prepulses incremented over 0 to 50 ms to −20 mV before a test depolarization to −20 mV from a holding potential of −100 mV (Fig. 5A). Recovery from I_F was measured at a recovery potential of −100 mV after 20-ms depolarizations to −20 mV at 0- to 50-ms intervals (Fig. 5B). Time constants of onset of inactivation were 1.4 ± 0.ms and 2.0 ± 0.3 ms before and after PHT, respectively (P = 0.1, n = 5). Recovery from fast inactivation at −100 mV was 10.7 ± 1.4 ms in control and 13.0 ± 2.9 ms in PHT (50 μM; P = 0.2, n = 9). Thus, as with the I_F equilibrium relation, there was no effect of PHT on the dynamics of transition into and out of I_F states.

Fig. 5.

Effect of PHT on I_F onset and recovery with double-pulse protocol. Recovery curves were well described by a single-exponential function. A: I_F onset curve generated by the double-pulse onset protocol is shown (n = 5). A depolarizing prepulse to −20 mV for a variable interval (0–50 ms) was followed by a brief 10-ms repolarization to −100 mV, followed by a test pulse to −20 mV as shown in the inset. Time constants of I_NaT decrement of 1.4 ± 0.4 and 2.0 ± 0.3 ms for control and 50 μM PHT, respectively, were not significantly different (P = 0.1). B: recovery from I_F with double-pulse offset protocol is shown (n = 9). An initial 20-ms depolarization from −100 to −20 mV was followed by a 0- to 50-ms recovery period at −100 mV, with a subsequent 20-ms test pulse as illustrated in the inset. Time constants of recovery of I_NaT were 10.7 ± 1.4 and 13.0 ± 2.9 ms in control and PHT, respectively, with no significant difference (P = 0.2).

Effect of PHT on slower inactivated states.

Because there was no significant effect of PHT on onset and offset of I_F and the corresponding h_inf curve (Fig. 4C), effects of PHT on previously identified slow inactivation processes (I_I and I_S) were investigated. This was particularly relevant because the equilibrium states (as determined by the h_inf curves for these processes) were shifted to the left after exposure to 50 μM PHT (Fig. 4, D and E), which would conventionally imply an affinity for those states (see Bean et al. 1983; Hille 1977).

To test the effect of PHT application on the onset of slower forms of inactivation, neurons were depolarized to −20 mV from a holding potential of −100 mV for 10 s. During the depolarization, cells were intermittently repolarized for 15 ms at −100 mV to remove fast inactivation, and a 5-ms test pulse was performed to measure slow inactivation onset. Four initial 250-ms intervals to capture “intermediate” inactivation components and subsequent 1-s intervals were used. The decrement of the I_NaT of the test pulse is shown in Fig. 6A, inset, and was best described by a double-exponential function. The I_S and I_I time constants in control solutions were 4,878.7 ± 230.4 and 111.8 ± 11.4 ms (n = 6), respectively, with the faster component being ∼ 45% of the total decrement. The application of PHT (50 μM) reversibly decreased the I_S time constant to 1,519.6 ± 84.4 ms (P < 0.001, n = 6). In contrast, the I_I time constants were not affected (123.1 ± 8.9 ms; P = 0.1, n = 6). The relative amplitude of I_I to I_S time constants was not altered by PHT (∼45%; P = 0.2), and no additional time constant was seen.

Fig. 6.

Effect of PHT on slow components of inactivation. All current responses were best described by a double-exponential function. A: onset of slow inactivation was observed from 0 to 10 s by using test depolarizations with durations that initially increased in four 250-ms increments and then in nine 1-s increments. PHT decreased the available current from ∼25% in control to ∼15% after 10 s. Two components of inactivation termed “I_I” and “I_S” of 111.8 ± 11.4 (45% of total) and 4,878.7 ± 230.4 ms, respectively, were found. PHT (50 μM) significantly reduced the slower I_S time constant to 1,519.6 ± 84.4 ms (P < 0.001, n = 6) but not the I_I time constant (P = 0.10, n = 6). B: modified double-pulse protocols were used for the recovery of I_S. PHT decreased about 10% channel availability after 9 s. Again, 2 components were constantly seen. Only I_S time constants were affected significantly (P = 0.01, n = 8), not I_I time constants (P = 0.08, n = 8). Paired t-test was used throughout. Both protocols and representative current traces are shown in the insets of the corresponding curves.

The effect of PHT on the offset of slower forms of inactivation was then measured. The protocol was initiated with a 10-s prepulse to −20 mV from a holding potential of −100 mV. A series of 5-ms test pulses were sampled 265 ms apart for the first four pulses (to sample the faster component) and were then sampled at 1-s intervals. The subsequent current amplitudes were normalized to the response at time 0 and plotted against recovery time (Fig. 6B). Again, the recovery was characterized by a sum of two exponentials. Both time constants were significantly increased; the slower component increased from 2,031.5 ± 114.6 to 3,321.8 ± 463.0 ms (P = 0.01, n = 6), and the intermediate component increased from 198.1 ± 22.6 to 538.2 ± 188.0 ms in the presence of 50 μM PHT, which was reversed during washout. The intermediate component increase in amplitude did not reach statistical significance in these experiments (P = 0.08). However, in separate experiments with a larger number of sampling points over a shorter interval (700 ms), the time constant of offset of I_I was increased in the presence of 50 μM PHT. The relative amplitudes of I_I and I_S were not significantly altered (∼65–72%; P = 0.2).

Effect of PHT on closed-state transition at subthreshold potentials.

As noted above, macroscopic I_Na activates at approximately −60 mV. At hyperpolarized potentials, channel opening probability is low, but interchange between closed and closed-inactivated states occurs, as implied by the steady-state inactivation relation and as embodied in the Hodgkin-Huxley equations and Markov-state equivalent models. This can be termed “closed-state transition inactivation” (CSI), as opposed to“ open-state inactivation” (OSI), where channels have passed through the open state before inactivating (Bähring and Covarrubias 2011; Goldman 1995).

Initially, transitions between inactivated and closed states were observed by changing the membrane potential from −80 to −100 mV over an interval of 700 ms. Test pulses (5 ms) to −20 mV assessed channel availability. Figure 7A shows the protocol and a sample current trace (inset). The increase in amplitude of I_Na before and after application of 50 μM PHT was measured, normalized, and plotted against the transition interval. Two components of current recovery were evident in both the control and drug-affected conditions. A fast component corresponding to I_F was present that was not significantly affected by the drug (P = 0.1, n = 4). The slower component corresponding to the previously identified intermediate-duration process was slowed in the presence of drug and changed from 109.7 ± 44.0 to 843.5 ± 229.8 ms (P = 0.04, n = 4). The relative amplitude of the intermediate component increased from 19.7 ± 5.3% to 65.8 ± 10.8% but did not quite reach significance (P = 0.06, n = 4).

Fig. 7.

Effect of PHT on closed-state transition. All cells showed 2 time constants. A: changing the membrane potential from −80 to −100 mV produced a transition from the closed-inactivated to the closed state. The time constants of recovery of I_NaT had components related to I_F and I_I (n = 4). The I_I time constant was significantly slowed by PHT (P = 0.04, n = 4), but not the I_F time constant. B: changing the membrane potential from −100 to −80 mV produced a transition from the closed to the closed-inactivated state. The time constants of decrement of I_NaT were also found to contain I_F and I = (n = 5), but neither was significantly altered by PHT. Both protocols and representative current traces are shown in the insets of the corresponding curves.

The transition from closed to closed-inactivated states was investigated similarly, using voltage steps from a holding potential of −100 mV to −80 mV before a test pulse to −20 mV. The responses and fitted exponential curves in control and 50 μM PHT are shown in Fig. 7B. I_F time constants were again not significantly affected by PHT: 26.5 ± 6.8 and 26.8 ± 6.9 ms in control and PHT, respectively. Notably, I_I time constants of 237.9 ± 100.7 ms in control and 217.3 ± 55.0 ms in PHT were not significantly changed (P = 0.8, n = 5), and there was no change in relative amplitudes. Nonetheless, PHT application decreased channel availability from about 40% to 25% at the end of the 700-ms transition time.

Relationship of use-dependent block by PHT to slow inactivation processes.

PHT and other sodium channel modulators enhance “use dependence,” whereby currents evoked by trains of depolarizations under voltage clamp or action potentials in voltage recording mode are attenuated by drug application. Although commonly associated with drug action, a similar intrinsic pattern of attenuation is seen in control conditions. We therefore sought to discover if intrinsic use dependence could be related to the forms of inactivation identified in this work, and if PHT would affect these processes.

For this experiment, a train of 10-ms depolarizations to −20 mV at 25 Hz over 1 s from a holding potential of −100 mV was employed. A sample protocol is shown in Fig. 8B, inset. I_Na amplitude decreased over time, consistent with the accumulation of inactivated channels with repeated depolarizations. These responses were plotted against time and were best described by a double-exponential time course, with components similar to the previously identified I_F and I_I time constants, plausibly relating these observations to the previously identified inactivation processes.

Fig. 8.

Use-dependent block enhanced by PHT. Repeated stimulation pulses with the interpulse interval of 25 ms are shown before and after 50 μM PHT. A: a representative current trace is shown for the repeated stimulation protocol (inset). B: normalized current peak vs. stimulation duration is plotted. All cells were better fitted with a double-exponential function. Only the amplitude of the I_F component was found to be significantly decreased (P < 0001, n = 4). This suggests that I_I components were significantly increased while the time constants were unaltered.

With the use of a 50-ms interpulse interval protocol, I_F and I_I time constants were 40.2 ± 4.1 and 542.2 ± 146.3 ms before and 52.3 ± 9.0 and 560.1 ± 126.5 ms after 50 μM PHT. The relative amplitude of the fast inactivation component was significantly decreased in the presence of PHT (63.3 ± 5.9% and 44.0 ± 5.4%, before and after PHT; P = 0.01, n = 4). The total inactivated fraction of I_Na was ∼18% in control and ∼30% in PHT after the 1-s-duration protocol. When the same pulse protocol was performed at 40 Hz, the two time constants were again seen, but with smaller amplitudes: 21 and 354 ms, respectively. Similarly, time constants were not affected by PHT, but the proportion of the slower component and total inactivated fraction increased to 55% in the presence of PHT. Although this experiment showed two time constants of inactivation similar to those reported above, because repolarization occurs between pulses, it would not be expected that the time constants correspond exactly with the more specific protocols used in the previous experiments.

Effect of removal of fast inactivation by papaine.

The use of protease or papain to remove macroscopic fast inactivation of I_Na has been used in several previous studies (Armstrong et al. 1973; Oxford 1981; Zilberter and Motin 1991); however, this approach has been less frequently used in central neurons. We found that addition of papaine (0.75 mg/ml) to the patch-pipette solution essentially removed fast inactivation over a duration of about 15 min (Fig. 9A). PHT (50 μM) reversibly decreased I_Na amplitude by a similar proportion to that seen with untreated cells (∼20%, Fig. 9A). When depolarization was prolonged over several seconds, the currents in control solution attenuated with two time constants, similar to the slow and intermediate processes identified above (Fig. 9B). The reversal potential of I_Na remained constant during these depolarizations (data not shown).

Fig. 9.

Enzymatic removal of I_F leaves PHT sensitivity intact. Papain (0.75 mg/ml) was added to the intracellular solution to remove I_F. Representative current traces for a single depolarization to −20 mV from a holding potential of −100 mV are shown with different depolarization durations of 20 ms (A) or 2 s (B). Reversibly, reduction of amplitude of I_Na with 50 μM PHT is evident in the absence of I_F. The longer duration depolarization used in B reveals biexponential attenuation of I_Na similar to the intermediate and slow components of inactivation seen in the normal preparation.

With the use of a 5-s depolarization protocol, the two inactivation time constants (intermediate and slow) were 331.1 ± 85.8 and 1,689.2 ± 404.8 ms in control conditions and 206.5 ± 52.4 and 839.4 ± 212.8 ms in 50 μM PHT, findings consistent with an acceleration of entry into inactivated states (P = 0.02 for intermediate inactivation; P = 0.02 for slow inactivation; n = 6 in each group). No additional time constants of decay corresponding to a possible binding rate of PHT were observed, and these effects were reversible with washout.

We also tested the effect of papaine on h_inf curves (Fig. 10, A and B). Papaine largely removed the transition to inactivation with a 50-ms conditioning pulse, presumably due to the abolition of I_F. However, the conventional sigmoid steady-state inactivation relation was restored with the long-duration (500 ms) conditioning pulse, most likely due to the preservation of I_I with papaine.

Fig. 10.

A: the steady-state inactivation curve shows little decrement with 50-ms conditioning pulses after papaine treatment (red squares) compared with control (black circles; n = 2∼9). B: the steady-state inactivation curve with 500-ms pulses after papaine treatment shows significant decrement with depolarization, consistent with retention of the intermediate inactivation component evident in the long depolarization data (Fig. 9B).

Effect of PHT on I_NaP.

PHT has been reported to inhibit the persistent sodium current I_NaP to a greater extent than the transient component I_NaT (Chao and Alzheimer 1995; Colombo et al. 2013). The effect of PHT (50 μM) on the fractional amplitude change of I_NaP compared with I_NaT was measured at 48 ms, about 24 times the time constant of fast inactivation at −20 mV. I_NaP constituted 2.1 ± 0.4% and 1.8 ± 0.5% of the peak current before and after 50 μM PHT, respectively, a nonsignificant variation (P = 0.6, n = 6).

Computer modeling.

Because the results described above could be consistently explained by effects on slower forms of inactivation, a model for the sodium conductance incorporating a slow inactivation state was developed. The aim of this model was to explore possible mechanisms of action, rather than to generate a detailed reproduction of experimental data. We had observed in separate studies that two of the most commonly described effects of PHT and other sodium channel modulators (SCMs) under voltage-clamp conditions, the negative shift in the h_inf curve and slowing of recovery from inactivation, were reproduced by small changes in the voltage parameter of the forward rate constant of inactivation (α_h) in the Hodgkin-Huxley (HH) equations. Additionally, a recent atomic resolution model of Na_V-PHT interactions revealed a potential binding site in the voltage sensor region in the activated state (Boiteux et al. 2014). A Markov-state model with two closed and one open state, as well as single fast and intermediate-duration inactivation states, was constructed (Fig. 11A) with voltage-sensitive rate constants derived from the HH equations (see Appendix). By simply changing the voltage parameter of the forward rate constant of intermediate inactivation (αh_I) from 100 to 110 mV, the leftward shift in the 500-ms h_inf curve (Fig. 11E) and significant slowing of recovery were reproduced (Fig. 11, C and D). The notable voltage sensitivity of the PHT effect was also reproduced (Fig. 11B). Additionally a proportionally smaller change (25%) in the rate of onset of inactivation was seen, similar to the experimental data for the intermediate form of inactivation. The longer I_S component was not included in this model for simplicity. Of note, no assumptions have been made about relative binding affinities to different states as proposed by the modulated receptor hypothesis and its derivatives.

Fig. 11.

Simulations related to experimental findings. A: state diagram for sodium channel with fast and intermediate inactivation; the slow inactivation state is left out for clarity. The voltage parameter (Vparh_I) for the backward rate constant for αh_I was changed from 100 mV (standard value) to 110 mV to simulate possible PHT interaction with gating charge, resulting in an increased energy barrier to departure from the inactivated state. See Appendix for parameter equations. B: effect of Vparh_I parameter change on I_Na amplitude evoked by depolarizations from a holding potential (V_h) of −100 and −80 mV, respectively. There is an ∼25% reduction in I_Na amplitude at −80 mV compared with control, and a much smaller effect with the same depolarization from −100 mV, i.e., very similar to the classical effect of PHT as shown in Fig. 2, A–C. C and D: recovery from inactivation at −100 mV with Vparh_I = 100 mV (C) and Vparh_I = 110 mV (D) showing slowing of recovery similar to PHT experimental findings (see Fig. 7A). The fast recovery component was unaffected, but the slow component increased from 105 to 227 ms (182%); the rate of onset of the slow component was much less affected, decreasing from 185 to 138 ms (∼25% reduction), again showing a pattern similar to that of experimental results. E: steady-state inactivation curves from the model. Diamonds represent I_F (fast inactivation); squares represent I_I with Vparh_I = 100 mV (standard model); and triangles show shift to left of curve with Vparh_I changed to 110 mV, as seen experimentally. G: modified version of state diagram in A with I_I and I_S, but without I_F, to model PHT interaction in sodium channels in the papaine preparation. Preferential interaction with inactivated states according to the modulated receptor hypothesis is simulated: PHT is assumed to interact with equal affinity for intermediate and slow inactivated states, with rate constants for 48 μM PHT derived from Kuo and Bean (1994) for comparison with experimental data shown in Fig. 9. I_I* and I_S* represent parallel drug-bound slow-inactivated states. F: using the model in G, a third time constant with large-amplitude (2.1 s, relative amplitude 37%) was introduced to the decay phase with the initial intermediate and slow time constants preserved (1.25 s and 230 ms, relative amplitudes 45% and 18%), as opposed to the experimental observation of an acceleration of both components without the addition of a third time constant. A similar pattern was observed when interaction of PHT with I_S alone was modeled. C₁ and C₂, closed states; O, open state.

Although the experimental data and simulation was consistent with modulation of slow inactivation, possibly through effects on voltage sensors, we sought to compare a conventional modulated receptor model with slow binding of PHT with the experiments including papaine-induced removal of I_F. A modified Markov Na_V model with I_I and I_S incorporated, with I_F removed (Fig. 11G), was developed to model the papaine experiments with long depolarizations. A modulated receptor modification of the model was constructed by adding two parallel drug-bound slow-inactivated states (I_I* and I_S*) as illustrated in Fig. 11G with rate constants derived from previous estimates (Kuo and Bean 1994). With long-duration depolarizations, little change in the time constants of intermediate and slow inactivation occurred (230 and 1,250 ms, respectively); however, a large, separate 2,100-ms additional time constant of decay occurred, corresponding to the slow PHT binding. This result is discordant with the experimental finding of an acceleration of the components of slow inactivation, with no additional time constant.

DISCUSSION

Phenytoin (PHT) primarily reduces the amplitude and rate of recovery of sodium currents in neurons (Matsuki et al. 1984; Schwarz and Vogel 1977). Like local anesthetics (LAs) and antiarrhythmic drugs, the PHT inhibition of I_Na is strongly dependent on holding potential.

A cogent explanation of the effect of sodium channel modulators such as LAs and antiarrhythmic drugs was developed by Hille (1977) and Hondeghem and Katzung (1977). It was proposed that these drugs have a higher binding affinity for inactivated states of the channel than other states. This theory explains the leftward shift of the steady-state inactivation relation and slowing of recovery from inactivation observed with LAs (Chen et al. 1975; Courtney et al. 1975). This framework has been extended to PHT, with the relatively slow on and off rates of the drug ascribed to slow binding kinetics (Kuo and Bean 1994). Furthermore, it has been generally interpreted that the fast-inactivated state has the higher affinity for drug binding (for example, Brodie and Sills 2011; Errington et al. 2008, Karoly et al. 2010; Kuo and Bean 1994). Starmer et al. (1983) proposed the “gated receptor hypothesis,” which states that the motion of the activation and inactivation gates results in exposure of binding sites (rather than inactivated states) as the targets of SCMs, explaining the voltage sensitivity of block and other SCM effects. However, Khodorov and Shishkova (1976), Matsuki et al. (1984), and Quandt (1988) have suggested that slower forms of inactivation are involved, which would explain the effects of LAs and PHT. Nevertheless, Karoly et al. (2010), although noting the difficulty in experimentally differentiating effects of SCMs on slow vs. fast inactivation processes, have suggested that fast inactivation is likely to be the main target of these drugs.

Although the present study cannot definitively identify the mechanism of action of PHT, we have attempted to place the drug effects in the context of “normal” channel inactivation behavior throughout these experiments, including “intermediate” timescale processes reported by several groups.

PHT may influence I_F by slow binding kinetics, as suggested by Kuo and Bean (1994), and the current experiments have not completely excluded this possibility. Nonetheless, the qualitative amplitude of the effect of PHT remains despite fast inactivation removal, so it is unlikely that I_F is the only and/or main target. It is also unexpected that the steady-state inactivation relation for fast inactivation remains constant with PHT, as has similarly been reported by Matsuki et al. (1984) and Kuo and Bean (1994).

A second plausible mechanism is that PHT affects slow inactivation processes, possibly including innate “use dependence.” A consistent observation in this study has been the presence and drug modification of two components of slow inactivation: a more “conventional” slow form with time constants on the order of 10³ ms, and a form that has been termed “intermediate” with time constants on the order of 10² ms (French et al. 2016). Both ranges of time constants have been observed previously in CA1 neurons (e.g., faster component: Fleidervish et al. 1996; Migliore 1996; slower component: Kuo and Bean 1994) but have generally been treated as a single process. The shift in the 500-ms and 10-s but not 50-ms h_inf curves with PHT would conventionally suggest a higher affinity for the slower states. Matsuki et al. (1984) and Meves and Vogel (1977) found significant negative shifts (>10 mV) with 60- and 5-s prepulses, respectively, the latter protocol being most similar to the current study. Interestingly, Vandenplas et al. (2013) did not see a significant shift in the steady-state inactivation curve of the slow inactivation process in neuroblastoma cells with PHT, but they used a rather long (500 ms) repolarization before the test pulse, which would likely reduce or abolish such an effect. A notable observation in our study was that the I_S time constant of onset was reduced to 1.6 s by 50 μM PHT, almost exactly the value cited by Kuo and Bean (1994) as the time constant of onset of PHT effect and interpreted as being incompatible with I_S involvement. Additionally, a recent molecular dynamics study of PHT interaction with Na_V channels reported a time constant of interaction with identified potential binding sites of <1 ms (Boiteux et al. 2014), i.e., much faster than that assumed by Kuo and Bean (1994b) to explain the slow onset of PHT. A similarly short binding rate can be calculated from an additional study by Martin and Corry (2014), although these rates are estimates from rather short simulation periods in both studies. Alternatively, a phenylalanine residue, F1764, in the DIVS6 region is commonly considered the binding site for LAs and AEDs (Ragsdale et al. 1996), and mutation of this residue has been found to alter slow inactivation (Bai et al. 2003). It is therefore possible that binding of PHT to this site might affect slow inactivation, to account for at least some of the experimental observations.

Because the steady-state inactivation curve is the equilibrium distribution of closed and inactivated states resulting from rates of entry and exit, it was of interest to observe what effect PHT had on the onset and recovery rates into and out of inactivated states. A simple “increased affinity” for inactivated states would suggest an increased entry rate and slowed exit rate from inactivation. This pattern was seen with conventional slow (approximately on the order of 10³ ms) inactivation (Fig. 5). Intriguingly, a different pattern was seen with the intermediate inactivation component: the rate of entry into this state was not greatly affected, but recovery was slowed considerably, as discussed below.

Effect of removal of fast inactivation with papaine.

The inhibitory effect of PHT was preserved after removal of I_F with cytoplasmic papaine. Prolonged depolarizations for several seconds revealed two components of current decrement, similar to the time constants of onset of I_I and I_S identified with double-pulse protocols. A striking acceleration of both components of inactivation with PHT was evident and reversible. If PHT were slowly binding to Na_V channels without affecting slow inactivation, as suggested by Kuo and Bean (1994b), it might be expected that the original time constants of inactivation would be preserved, with a third component corresponding to the PHT binding rate; however, this was not seen. Rather, the modulated receptor simulation produced an extra time constant corresponding to slow binding (see Fig. 11, F and G), discordant with the experimental observations. These observations were again consistent with PHT enhancing or catalyzing entry of open channels into slow-inactivated states.

Possible role of voltage sensors as PHT binding targets.

Channel voltage sensors have been implicated in LA and antiarrhythmic drug effects. LAs have been found to reduce gating charge (Hanck et al. 1994; Keynes and Rojas 1974) and stabilize the domain III voltage sensor in an activated state (Arcisio-Miranda et al. 2010). A possible interaction of PHT with voltage sensors in Na_V channels was also considered by Kuo and Bean (1994). Additionally, slow inactivation of sodium channels is associated with a slow component of gating current immobilization (Bezanilla et al. 1982), and Silva and Goldstein (2013) found that voltage sensor motion is intrinsically associated with the development of slow inactivation in sodium channels. Although there was no effect found on the steady-state activation curve or rate of fast inactivation in this and previous studies with PHT, we hypothesize that drug interactions with gating charge affect slower inactivation processes that are initiated by outward motion of the charged S4 domains with depolarization. These processes would produce “charge immobilization” and subsequent and possibly related slow inactivation. Entry into slower inactivated states is related to outward voltage sensor movement with depolarization, and the steady-state inactivation relation can be shifted by charge mutations in the S4 region (Muroi and Chanda 2009). Additionally, we could reproduce the salient effects of PHT by adjusting a voltage sensitivity parameter in a simple model of slow inactivation. Specifically, with the use of a Markov model based on the HH equations with an additional slow-inactivated state, a modest 10-mV change in the voltage term of one of the inactivation rate constants (αh_I) produced both a hyperpolarizing shift in the 500-ms steady-state inactivation curve and marked slowing of time constants of recovery (105 to 237 ms at −100 mV), with a much smaller prolongation of entry into slow inactivation (138 to 185 ms at −70 mV). These data are similar to our experimental observations for I_I (see Fig. 11 and Appendix for equations). Another line of evidence supporting a voltage sensor role in PHT action are the molecular dynamical simulations of Boiteux et al. (2014) that have shown a potential binding region for PHT in the voltage sensor region. Notably, the Boiteux model used voltage sensors in the activated state, i.e., in the conformation expected with depolarization. Interactions of PHT with the four voltage sensors could produce the ∼10-kcal energy change needed for this magnitude of voltage sensitivity alteration. It should be noted that this model does not have a strong dependence on the HH model and only uses the equations for voltage sensitivity to provide a simple link between voltage sensor state and the experimentally observed equilibrium between closed and closed slow-inactivated states.

In summary, we propose that PHT exerts at least some of its effects by interactions with S4 gating charges. Depolarization with outward movement of the S4 voltage sensors in Na_V likely exposes binding residues that interact with PHT (Boiteux et al. 2014), enhancing charge immobilization and hence slow inactivation, and slowing return from these states. This is might be considered a form of the “gated receptor hypothesis,” with the addition that the actual binding target is the voltage sensor and that this then enhances entry into slow inactivation. This is reminiscent of a “foot in the door” mechanism, in which exposure of the voltage sensors by outward motion of S4 are “latched” by binding to PHT. In support of this mechanism for the intermediate inactivation component, it was observed that the rate of entry into the intermediate inactivation state with depolarization was minimally accelerated by PHT, compared with repolarization, which was quite slowed. This observation is compatible with “trapping” of voltages sensors once exposed. Both enhancement of transition into and slowing of recovery from I_S was observed, also explainable by this mechanism. The final effect may occur via stabilization of the selectivity filter P-loop structure that is thought to underlie slow inactivation (Ulbricht 2005), also reported by Boiteux et al. (2014). It would therefore be of interest to examine the effects of PHT after mutagenesis of candidate S4 charge residues, as well as to track S4 and P-loop motion in the presence of drug using fluorometric techniques (Chanda and Bezanilla 2002; Loots and Isacoff 1998). A potentially analogous situation has been reported for a potassium channel agonist, NH29, which is thought to produce therapeutic effects by direct stabilization of activated voltage sensors (Peretz et al. 2010).

Lack of PHT effect on I_NaP.

Several studies have reported that the persistent sodium current I_NaP is preferentially affected by PHT (Chao and Alzheimer 1995; Colombo et al. 2013; Lampl et al. 1998; Segal and Douglas 1997). In the current study this was not found, reflecting probably both semantic and methodological issues. If I_NaP is considered as the small residual current, persisting at long durations compared with the time constants of fast inactivation, then it did not appear preferentially affected by PHT in this study using the 50-ms depolarizing protocol. However, we did not measure it during extended depolarizations, where it may be reduced by the slow inactivation processes noted above and enhanced by PHT. Nonetheless, even in this case, the amplitude of I_NaT at the point of measurement should also be evaluated to evaluate whether I_NaP is a more specific target of PHT, as concluded by Chao and Alzheimer (1995). Our conclusions are similar to those of Colombo et al. (2013), who noted that inactivation seems to be the process affected by PHT in the suppression of I_NaP with long-duration depolarizations.

Conclusion.

The mechanism of the inhibitory action of phenytoin on sodium channels has been controversial, and several issues remain to be resolved, including the exact mechanism of interaction with the channel, the nature of the binding site, and whether there is more than one physiologically active site. The present study did not find any clear inhibition of fast inactivation, a commonly assumed functional target for phenytoin, but rather the observed effects could be interpreted as the modulation of slow inactivation mechanisms, including a form that we have termed “intermediate.” This interpretation was extended by showing that both experimental data and simulation studies were compatible with such modulation of slow inactivation. These data are also compatible with the proposal that the Na_V voltage sensors may be targets for sodium channel modulators.

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

Z.Z. and C.R.F. conception and design of research; Z.Z., A.S., and C.R.F. performed experiments; Z.Z., E.L.H.-Y., T.J.O., A.S., and C.R.F. analyzed data; Z.Z., E.L.H.-Y., D.A.W., T.J.O., and C.R.F. interpreted results of experiments; Z.Z., A.S., and C.R.F. prepared figures; Z.Z. drafted manuscript; Z.Z., E.L.H.-Y., D.A.W., T.J.O., and C.R.F. edited and revised manuscript; Z.Z., E.L.H.-Y., D.A.W., T.J.O., A.S., and C.R.F. approved final version of manuscript.

APPENDIX

The following equations were used for the Markov-state model shown in Fig. 11A:

$$\alpha m=0.0978\ast \left[V\left(t\right)+42.13\right]/\left(1-\exp \left\{-\left[V\left(t\right)+42.13\right]/6.49\right\}\right)$$

$$\beta m=3.98\ast \exp \left\{-\left[V\left(t\right)+64.178\right]/18.43\right\}$$

$$\alpha hF=0.08\ast \exp \left\{-\left[V\left(t\right)+92.3\right]/23.6\right\}$$

$$\beta hF=2.54/\left(1+\exp \left\{-\left[V\left(t\right)+7.5\right]/13.4\right\}\right)$$

$$\alpha hI=0.04\ast \exp \left\{-\left[V\left(t\right)+\alpha hIV\right]/11\right\};$$

where αh_i voltage (αh_i V) = 100 or 110 mV, as described in the text.

$$\beta hI=0.0044\ast 1/(1+\exp \left\{-\left[V\left(t\right)+83.7\right]/13.4\right\})$$