The properties of the Na+ current present in thalamocortical neurons of the dorsal lateral geniculate nucleus were investigated in dissociated neonate rat and cat neurons and in neurons from slices of neonate and adult rats using patch and sharp electrode recordings.
The steady-state activation and inactivation of the transient Na+ current (I Na) was well fitted with a Boltzmann curve (voltage of half-maximal activation and inactivation, V 1/2, −29.84 mV and −70.04 mV, respectively). Steady-state activation and inactivation curves showed a small region of overlap, indicating the occurrence of a I Na window current.I Na decay could be fitted with a single exponential function, consistent with the presence of only one channel type.
Voltage ramp and step protocols showed the presence of a noninactivating component of the Na+ current (I NaP) that activated at potentials more negative (V 1/2 = −56.93 mV) than those of I Na. The maximal amplitude ofI NaP was ∼2.5% ofI Na, thus significantly greater than the calculated contribution (0.2%) of theI Na window component. Comparison of results from dissociated neurons and neurons in slices suggested a dendritic as well as a somatic localization of I NaP. Inclusion of papain in the patch electrode removed the fast inactivation of I Na and induced a current with voltage-dependence (V 1/2 = −56.92) and activation parameters similar to those ofI NaP.
Current-clamp recordings with sharp electrodes showed thatI NaP contributed to depolarizations evoked from potentials of approximately −60 mV and unexpectedly to the amplitude and latency of low-threshold Ca2+potentials, suggesting that this noninactivating component of the Na+ channel population plays an important role in the integrative properties of thalamocortical neurons during both tonic and burst-firing patterns.
- action potential
- persistent Na+ current
- burst firing
- dorsal lateral geniculate nucleus
The thalamocortical loop has been the subject of many electrophysiological studies because of its central role in awareness and sleep (Ribary et al., 1991; Steriade et al., 1993; Sillito et al., 1994; Barth and MacDonald, 1996) as well as in a number of neurological disorders (Malafosse et al., 1994; Jeanmonod et al., 1996). In particular, the ionic currents of its constituent neurons have been investigated extensively to determine their contribution to different electrophysiological behaviors (Steriade et at., 1990; Huguenard and Prince, 1991; Llinás et al., 1991;Connors, 1994; McCormick and Bal, 1997). A surprisingly noticeable exception, however, has been the transient Na+current (I Na). Although thoroughly characterized in cortical pyramidal neurons during both developmental and pathological states (Huguenard et al., 1988; Fleidervish et al., 1996), there has been no study aimed at investigating the biophysical properties of this current in, and its precise contribution to the tonic and burst firing of, thalamocortical (TC) neurons. Thus, biophysical modeling of TC neurons (McCormick and Huguenard, 1992;Tóth and Crunelli, 1992; Destexhe et al., 1993; Antal et al., 1996) has relied on the modified description of Na+currents from squid axons and sympathetic, neocortical, and hippocampal neurons (French et al., 1990; Belluzzi and Sacchi, 1991; Traub and Miles, 1991; Traub et al., 1991), a less than optimal compromise in view of the existence of voltage-dependent Na+channel isoforms with different biophysics, pharmacology, and tissue distribution (Noda et al., 1986; Heinemann et al., 1992; Roy and Narahashi, 1992).
Recently, a renewed interest in the noninactivating Na+ current I NaP has developed. This sustained component of the Na+current is seen in many excitable cells, including central neurons (French et al., 1990; Alzheimer et al., 1993; Crill, 1996; Fleidervish and Gutnick, 1996), and it has been implicated in signal amplification (Stuart and Sakmann, 1995; Lipowsky et al., 1996) and intrinsic high frequency oscillations of neocortical neurons (Llinás et al., 1991; Silva et al., 1991). The mechanism or identity ofI NaP is still a matter of some controversy, with multiple theories being advanced in different systems (Alzheimer et al., 1993; Sugimori et al., 1994; Crill, 1996). Indirect evidence for the presence of I NaP in mammalian TC neurons is limited to the effect of TTX in current-clamp recordings (Jahnsen and Llinás, 1984b; Tennigkeit et al., 1996; Pedroarena and Llinás, 1997). A description of the steady-state and kinetic properties of I NaP in these neurons would be significant for a fuller understanding of their signal integration properties in physiological functions and neurological conditions and would also provide the necessary data for ongoing simulation studies of the activity of single and small networks of thalamic neurons.
In this study we have determined the properties ofI Na and I NaP in TC neurons using patch-clamp recordings in the dorsal lateral geniculate nucleus (dLGN) of neonate rats and cats and adult rats, and we have investigated the physiological role of I NaP in tonic and burst firing using current-clamp microelectrode recordings in adult rats.
A preliminary report of some of these results has been published previously (Parri et al., 1996).
MATERIALS AND METHODS
Neonate rat. Slices containing the dLGN were obtained from male Wistar rats as described by Leresche (1992). Briefly, 7- to 11-d-old rats were anesthetized with halothane (2%) and decapitated. The brain was removed, a block of tissue containing the dLGN was separated from the rest of the brain, and 350-μm-thick slices were prepared from this tissue block using a vibratome (Energy Beam Science). Dissection and slicing procedures were performed in ice-cold medium of the following composition (in mm): NaCl 120, KCl 2, MgCl2 4, PIPES 20, CaCl2 1, glucose 25, ascorbic acid 0.3, kynurenic acid 1, pH 7.35 with NaOH. All chemicals were obtained from Sigma (St. Louis, MO) unless stated otherwise. Slices were stored in an oxygenated (100% O2) storage bath until recording commenced. After at least 1 hr, one slice was then anchored in the recording chamber by use of nylon threads fixed across a platinum harp and perfused continuously with the required recording solution (see below).
Dissociated neurons were prepared according to the procedure described by Hernandez-Cruz and Pape (1989), Budde et al. (1992), and Oh et al. (1995). Briefly, slices from neonate rats were prepared as detailed above and treated with 3 mg/ml protease XXIII for 25–40 min in a glass chamber of design similar to that described by Kay and Wong (1986). Neurons were triturated in Ca2+-free Ringer’s solution using fire-polished Pasteur pipettes of decreasing tip diameter. Dissociated neurons were then plated onto coverslips coated with 1 mg/ml poly-d-lysine and allowed to settle for ∼5 min before recording commenced.
Adult rat. Slices were prepared as described previously (Crunelli et al., 1987; Williams et al., 1996). Briefly, male Wistar rats (150–200 gm) were anesthetized (2% halothane) and decapitated. A block of tissue containing the thalamus was dissected, and 400-μm-thick slices containing the dLGN were cut using a vibroslicer (Campden Instruments). All dissection and slicing procedures were performed in ice-cold medium of the following composition (in mm): NaCl 134, NaHCO3 16, KCl 5, KH2PO4 1.25, MgSO4 5, CaCl2 2, and glucose 10. Slices were maintained at room temperature in this Ringer’s solution and bubbled with a 95% O2, 5% CO2 mixture.
Neonate cat. Cats (7–10 d old) were anesthetized (1% halothane, 2% N2O), and the brain was removed as described previously (Pirchio et al., 1997). From a block of tissue containing the dLGN, 400-μm-thick slices were prepared in ice-cold medium (see above) using a Campden vibroslicer, and dissociated neurons were then produced using the same methods as described above for the neonate rat.
Patch-clamp recordings of identified neurons. The slice or the coverslip containing the dissociated neurons was placed in a recording chamber mounted on the stage of a Nikon Axiophot microscope. TC neurons were identified, and were distinguishable from interneurons, by their characteristic size (soma diameter: 20–30 μm) and multipolar morphology (Hernandez-Cruz and Pape, 1989; Leresche, 1992; Williams et al., 1996). Membrane currents were recorded at room temperature (18–22°C) (except those from adult rat slices, see below) using an Axopatch 200A (Axon Instruments, Foster City, CA). Patch electrodes were pulled from borosilicate glass (GC120F, Clark Electromedical Instruments, Pangbourne, UK) using a horizontal electrode puller (Sutter Instruments, Novato, CA) and had resistances of 1–4 MΩ when filled with CsF internal solution. Electrodes were coated with Sylgard (Corning, Corning, NY) to counteract electrode capacitance artifacts. Series resistances (4–10 MΩ) were compensated (60–80%) using the compensatory circuits of the amplifier, and data were not used for analysis if the calculated maximum uncompensated error was ≥5 mV. Voltage protocols, data acquisition, and analysis were controlled with pClamp (Axon Instruments). Currents were sampled at 40 kHz and filtered with a low-pass Bessel filter at 5 kHz. Membrane capacitance was measured using the capacitance compensation circuitry. Currents were corrected on line for linear leakage and capacitative current by scaling the averaged response to four hyperpolarizing steps of 5 mV amplitude obtained at a holding potential of −100 mV.
Blind patch-clamp recordings. In slices of adult rat dLGN, voltage ramp recordings of I NaP were performed at 35 ± 1°C using an Axopatch 1D amplifier (Axon Instruments) and pClamp. Series resistance in these experiments was 17 ± 1 MΩ (n = 4).
Recording solutions. For Na+ current characterization, the internal pipette solution contained (in mm): CsF 120, HEPES 10, EGTA 10, MgCl2 2, CaCl2 1, Na2ATP 4, GTP 0.5, pH 7.3 with TEA-OH (Aldrich, Milwaukee, WI); osmolarity was adjusted to 290 mOsm. The standard extracellular recording solution with physiological levels of Na+ and the K+ and Ca2+ channel blockers contained (in mm): NaCl 120, sodium HEPES 16, KCl 2, glucose 10, TEA-Cl 20, CaCl2 1, 4-aminopyridine 2, MgCl2 4, NiCl2 0.5, CdCl2 0.1, pH 7.4; osmolarity was adjusted to 300 mOsm. Steady-state and kinetic experiments onI Na were performed with a reduced extracellular Na+ concentration (20 mm) to decrease the amplitude of the current and so minimize the impact of possible series resistance-induced errors. All experiments in neonate cats were performed in 30 mm extracellular Na+. The osmolarity of these solutions was maintained at 300 mOsm by increasing the TEA-Cl concentration. In blind patch-clamp experiments in adult slices, NaHCO3 replaced sodium HEPES as the buffer in the bathing solution.
Current-clamp recordings. For sharp electrode current-clamp recordings, adult dLGN slices were placed in an interface-type chamber and continuously perfused with a warmed (35 ± 1°C), oxygenated (95% O2, 5% CO2) medium containing (in mm): NaCl 134, NaHCO3 16, KCl 2, KH2PO4 1.25, MgSO4 1, CaCl2 2, and glucose 10. Intracellular electrodes contained 1 m potassium acetate, recordings were performed using an Axoclamp 2A, and current and voltage records were stored on a Biologic DAT recorder (Intracel Ltd, Royston, UK). Some of these experiments were performed in the presence of 4-(N-ethyl-N-phenylamino)-1,2-dimethyl-6-(methylamino)pyrimidinium chloride (ZD 7288) (kindly donated by Dr. P. Marshall, Zeneca, Macclesfield, UK).
Data analysis. Data were analyzed using the Clampfit program of pClamp (Axon Instruments) and the mathematical transform and curve fitting routines of Sigma Plot (Jandel Scientific, San Rafael, CA). For the construction of steady-state activation and inactivation curves, maximal conductances were estimated using the relationg = I/(V s −V rev), where I is the measured current, V s the voltage step, andV rev the measured current reversal potential. Estimated conductances were normalized and plotted against step potential. Individual traces were fitted in Clampfit using the Chebyshev algorithm to single exponentials of the formy = ae -bt. Data points were fitted with Boltzmann curves of the form y = 1/(1 + e(V1/2-V)/k) (where V 1/2 is the voltage of half-maximal activation, k is the steepness constant), single exponentials of the form y = 1 − ae bt, and double exponentials of the form y = 1 − ae bt +ce dt. All potential values quoted in the text and figures have been corrected for liquid junction potentials (Barry and Lynch, 1991; Neher, 1992): −11 and −8 mV for low (20 mm) and high (130 mm) Na+solution, respectively, calculated using Axoscope (Axon Instruments). All quantitative data in the text and figures are expressed as mean ± SEM unless stated otherwise, and statistical significance was tested using Student’s t test.
The data presented in this paper are based on the following preparations: dissociated neurons (n = 45) from neonate rat, dissociated neurons (n = 7) from neonate cat, neurons (n = 18) in slices of neonate rats, and neurons (n = 14) in slices of adult rats.
Steady-state activation and inactivation
The kinetic and steady-state properties ofI Na were determined using acutely dissociated neurons from neonate rats, which enabled a better space clamp because of the lack of an extensive dendritic arborization. In these experiments the extracellular Na+ concentration was also reduced to 20 mm to decrease the amplitude of the current and further reduce voltage disparities introduced by series resistance errors. To study the activation ofI Na, neurons were clamped at a holding potential of −111 mV, and depolarizing voltage steps of increasing amplitude were delivered every 3 sec. At potentials more positive than −60 mV the voltage steps elicited rapidly activating inward currents that inactivated within 10 msec (Fig.1 A1 ). The elicited currents reached a peak at −23 ± 0.5 mV and reversed at 26.6 ± 4.2 mV (n = 6) (Fig.1 A2 ), a value close to the theoretically determined reversal potential of 23 mV. This result, together with the block by 1 μm tetrodotoxin (TTX) (data not shown), defined this current as a TTX-sensitive neuronalI Na. The steady-state activation ofI Na, constructed from six neurons, could be fitted with a single Boltzmann curve (V 1/2 = −29.84; k = 5.88) (Fig.1 A3 ). To compareI Na activation in different conditions/species (see below) we also fitted the data of each neuron with a Boltzmann curve. From this analysis we obtained a V 1/2 of −29.8 ± 1.74 mV and a k = 5.48 ± 0.18 (n = 6) (Table 1).
Inactivation was investigated using a protocol in whichI Na was elicited with voltage steps to −11 mV every 5 sec from different holding potentials.I Na was reduced when the holding potential was more positive than −90 mV, and no current could be elicited from holding potentials more positive than −40 mV (Fig.1 B1 ). The voltage dependence of steady-state inactivation was plotted by normalizingI Na amplitude at different holding potentials to that elicited from a holding potential of −121 mV. The resulting points (from six neurons) could be fitted with a single Boltzmann curve (V 1/2 = −70; k = 5.88) (Fig.1 B2 ). Again, to compareI Na inactivation in different conditions/species (see below) we also fitted the data of each neuron with a Boltzmann curve and obtained a V 1/2 = −70.04 ± 0.76 mV and a k = 5.8 ± 0.15 (n = 6) (Table 1). Analysis of the steady-state activation and inactivation curves indicated the presence of a small area of overlap in a voltage region centered around −50 mV, thus predicting the existence of a “window component” of I Na (Fig.1 B2 ).
Kinetics of fast activation and inactivation
Because of the fast activation of I Na and the uncertainty inherent in fitting such a rapid rising phase, the time to peak was taken as an indicator of the rate of current activation. Activation was voltage dependent (Fig.2 A,B1 ), with the time to peak decreasing from 2.59 ± 0.79 msec at −41 mV to 0.59 ± 0.01 msec at 9 mV (n = 4) (Fig.2 B1 ). I Na decay could be fitted with a single exponential function (Fig.2 A) and was found to be voltage dependent, with a τ that ranged from 2.19 ± 0.21 msec at −31 mV to 0.64 ± 0.07 msec at 9 mV (n = 4) (Fig.2 B2 ). The relationships of the time to peak and τ of inactivation against membrane potential could not be fitted with simple exponential functions. This finding is consistent with classic models of channel activation (Hodgkin and Huxley, 1952), so that our experimental data in the measured voltage range would be located on the downstroke region of bell-shaped curves.
Onset of, and recovery from, inactivation
The onset of I Na inactivation at a particular potential will affect the action potential firing properties of the neuron, because it defines the rate at which the channel population enters an inactivated state. We investigated this process using a two-voltage step protocol. Neurons were held at a negative potential (−108 mV), and I Na was elicited by 10-msec-long depolarizing steps to −8 mV (Fig.3 A1 ). The onset of inactivation at −68 mV was determined by stepping to this potential for different durations before eliciting I Na. Times at the inactivating voltage were varied between 1 and 300 msec, with currents normalized to the I Na elicited after 1 msec at −68 mV. Analysis of these data showed that the onset of inactivation could be fitted with two exponentials with τ1 = 37 msec and τ2 = 76.9 msec (Fig.3 A2 ).
The rate of recovery from inactivation also has fundamental implications for the frequency and robustness of repetitive firing properties and was investigated using a two-pulse protocol.I Na was elicited with a 10 msec depolarizing step to −11 mV, during which the current was inactivated. A second test pulse was then delivered to elicitI Na, whereas the time between the two pulses (Δt) was varied between 1 and 150 msec to determine the rate of recovery. The experiment was repeated at different holding potentials to investigate the voltage dependence as well as the time course of recovery (Fig. 3 B). We observed that fast recovery from inactivation, investigated by varying Δt in 1 msec steps between 1 and 30 msec, was well fitted by single exponential functions with τ = 5.52 msec and τ = 13.88 msec at holding potentials of −111 mV and −91 mV, respectively. However, recovery at −71 mV was not well fitted with a single exponential (Fig.3 B1 ). Recordings with longer Δtdurations showed that recovery was complete within 30 and 60 msec at −111 and −91 mV, respectively (Fig. 3 B2 ). At −71 mV, however, recovery was not complete in 150 msec and was best fitted with two exponentials (τ1 = 14.74 msec and τ2 = 1.42 sec).
INa in TC neurons of the cat dLGN
Steady-state activation and inactivation ofI Na in dissociated cat neurons was studied using voltage protocols similar to those used in dissociated rat neurons (Fig. 4 A1 ,A2 ). These data could be fitted with a single Boltzmann curve for activation (V 1/2 = −27.2;k = 7.59) and inactivation (V 1/2= −65.7; k = 5.82), respectively. Indeed, the parameters of the Boltzmann curves for I Naactivation in five cat neurons show no difference inV 1/2 (−27.82 ± 2.7 mV) (p = 0.49) but a larger k (6.71 ± 0.25) (p < 0.005) compared with the rat, whereas for the inactivation a slightly more depolarizedV 1/2 (−65.93 ± 1.0 mV) (p < 0.05) but a similar k(k = 5.92 ± 0.25) (p = 0.686) were observed (Fig. 4 B) (Table 1). The kinetic properties of I Na in cat neurons were also similar to those in rat neurons and included, for example, a time to peak of 1.75 ± 0.29 msec and 0.76 ± 0.05 msec at −31 and 9 mV, respectively (Fig. 4 C1 ), and a τ of inactivation of 4.1 ± 1.56 msec and 1.67 ± 0.27 msec at −31 and 9 mV, respectively (n = 5, for all measurements) (Fig. 4 C2 ).
INaP in dLGN TC neurons
A persistent component (I NaP) of the Na+ current evoked by voltage step protocols could be observed by increasing the gain of the amplifier, so thatI Na was saturated. BothI Na and I NaP were blocked by TTX (1 μm) (n = 3) (Fig.5 A1 ). A TTX-sensitive (n = 3), slow component of the Na+ current (i.e.,I NaP) could also be seen when a voltage ramp from −100 to 50 mV was delivered at a rate of 0.2 mV/msec (Fig.5 A2 ). Because the maximal amplitude of this current was relatively small (∼100 pA), this and the following experiments were performed in extracellular solutions containing 130 mm Na+ to determine the amplitude and properties of I NaP under a physiological extracellular Na+ concentration.
The activation range of I NaP was investigated by applying long step depolarizations and measuring the amplitude ofI NaP at 60 msec into the depolarizing step, after I Na relaxation (Fig.5 B1 ). Recordings were performed at a high gain to enable reliable measurements of the sustained component.I NaP activation began at approximately −70 mV, peaked at −39 ± 2 mV, and had an extrapolated reversal potential of 4 ± 6 mV (n = 4) (Fig.5 B2 ). This reversal potential, however, is probably an underestimation of the true reversal potential, because at potentials more than −10 mV an outward current with properties similar to those described by Alzheimer (1994) in pyramidal neurons of the sensorimotor cortex was also activated. The activation ofI NaP could be well fitted with a Boltzmann curve (Fig. 5 B3 ), characterized by a k of 9.09 and a V 1/2 (−56.93 mV) that was more negative than that of I Na (compare Fig.1 A3 , B2 ). To compare I NaP activation in different conditions (see below) we also fitted the data of each neuron with a Boltzmann curve. From this analysis we obtained a V 1/2 of −53.87 ± 3.05 mV and a k = 8.57 ± 1.89 (n = 4) (Table 1).
Properties and occurrence of INaP
Additional properties of I NaP were investigated in adult rat dLGN slices because of the greater amplitude of I NaP in this preparation (Fig.6). The effect of the rate of voltage change on the degree of I NaP activation was investigated by varying the rate of rise of voltage ramps, and the amplitude of the pure I NaP was then measured after TTX and leak subtraction. We also studied the effect of ramp rates in the physiological range from 0.1 to 0.5 mV/msec (Fig.6 A2 ). A rate of 0.1 mV/msec was sufficient to elicit I NaP, and the amplitude of I NaP evoked by a rate of 0.5 mV/msec was double the one that was evoked at the lowest rate. Note that in this preparation, the fastest rates of rise were often sufficient to activate I Na as well (Fig.6 A1 ).
The possible inactivating effect of the holding potential or previous cellular activity was also investigated. For this analysis,I NaP was elicited with a voltage ramp from −100 to 50 mV while a conditioning potential was inserted between successive ramps to determine its effect on the amount ofI NaP evoked during the ramp (Fig.6 B1 ). The magnitude ofI NaP was dependent on the conditioning potential during the inter-ramp interval, with less current being evoked after positive steps were increased. Thus, for example, only 60% of the maximal I NaP could be elicited after a conditioning potential to −50 mV (Fig.6 B2 ).
To investigate possible subcellular, developmental, and species differences, the peak amplitude of I NaP elicited during a 0.2 mV/msec voltage ramp from −100 to 50 mV was compared in different preparations, all perfused with 130 mmextracellular Na+ except for neonate cat (30 mm): neonate rat dissociated neurons, 52.27 ± 10.9 pA (or 4.7 ± 0.5 pA/pF) (n = 11); neonate rat in slices, 86.22 ± 8.6 pA (n = 18); adult rat in slices, 155.75 ± 34.7 pA (n = 4); and neonate cat dissociated neurons, 92.8 ± 28.16 pA (or 4.6 ± 1.3 pA/pF) (n = 5). The difference in I NaPamplitude between dissociated neonate rat neurons and neurons in neonate rat slices was significant (p < 0.05), indicating the presence of this current in TC neuron dendrites.I NaP amplitude in adult rat was significantly greater than I NaP in dissociated neonate rat neurons (p < 0.001) and in neonate slice (p < 0.01). The difference inI NaP between neonate rat and cat dissociated neurons was not statistically significant, even when peak current densities were compared (p = 0.89), but note that in dissociated neonate rat neurons recorded under similar conditions (i.e., in 30 mm extracellular Na+) I NaP was almost immeasurable (i.e., <5 pA).
Mechanism of INaP manifestation
To test the hypothesis that I NaP could be produced by a proportion of the Na+ channel population that was noninactivating, we performed experiments (in dissociated rat neurons and 20 mm[Na+]o) in which 1 mg/ml papain was included in the patch-recording pipette to cause removal of the inactivation gate (Cota and Armstrong, 1992; Brown et al., 1994). Because papain removed inactivation 15–20 min after commencement of whole-cell recording, it was possible to obtain control data onI Na and then compare the current evoked after removal of inactivation. In the low extracellular Na+ solution used (20 mm),I NaP was practically immeasurable, but after 20 min of papain treatment a 0.2 mV/msec ramp protocol from −100 to 50 mV elicited a large inward current (I papain) that peaked between −40 and −30 mV (Fig.7 A), similar to the ramp-evoked I NaP in neurons that were recorded with papain-free electrodes.
Voltage step protocols were then used to obtain a more quantitative analysis of the properties of I papain. We observed that papain treatment transformed transientI Na currents elicited during step depolarizations to noninactivating currents (Fig. 7 B) that activated at more hyperpolarized potentials (around −70 mV) and peaked at −40 mV (Fig. 7 C1, C2 ). In addition, there was no significant difference in theV 1/2 value between I NaP(filled circles in Fig. 7 C2 ) and I papain (−56.92 ± 0.96 mV;n = 3) (open circles in Fig.7 C2 ) (p = 0.86) or theirk value (I papain,k = 6.31 ± 0.56 mV; n = 3) (p = 0.28) (Table 1).
To test the hypothesis that I NaP was a manifestation of the window component ofI Na, we comparedI NaP with the theoreticalI Na window current. The measuredI NaP (52.27 pA, see above) peaked at −39 mV, and its amplitude was 2.5% of the peak I Na(2106 ± 208 pA; n = 9). The theoretical window component was calculated from the product of the Boltzmann fits for the steady-state activation and inactivation curves recorded with papain-free electrodes (compare Fig. 1, A3 andB2 ), and the mean conductance ofI Na from neurons recorded in 130 mmextracellular Na+. This predicted current (dotted line in Fig. 7 C1 ) peaked at −55 mV and had an amplitude of 4.7 pA. The difference in voltage dependence and amplitude, therefore, suggests thatI NaP is not a manifestation of the window current of I Na. We also calculated windowI Na by using the results of a detailed voltage-clamp analysis method (Tóth and Crunelli, 1995) in which the steady-state activation curve is obtained without “contamination” by activation or inactivation kinetics. Even in this case, however, the amplitude of the calculated windowI Na was <1 pA, and thus much smaller than the measured I NaP. Finally, evidence thatI NaP was not a manifestation of windowI Na was obtained by the experiments in which positive holding currents were seen to have an inactivating effect on ramp-elicited I NaP (Fig.6 B1 ), because window currents are not expected to be affected by such potentials (Hirano et al., 1992).
Role of INaP in tonic and burst firing of TC neurons
After the existence and properties of I NaPin TC neurons were established, we investigated the possible physiological role of this current using current-clamp recordings from adult rat dLGN slices (Fig. 8). The membrane properties of these neurons were similar to those described previously for TC neurons in identical recording conditions (resting membrane potential: −62 ± 2 mV; apparent steady-state input resistance: 191 ± 24 MΩ; n = 10). The effect on tonic firing was investigated by recording from TC neurons held at membrane potentials more than or equal to −60 mV, from where low threshold Ca2+ potentials could not be evoked by positive current steps (Fig.8 A1 ). Positive current steps were applied in control conditions and in the presence of 1 μm TTX to investigate the contribution of Na+ currents in this region of the voltage–current relationship. TTX produced a block of the action potentials and also a reduction in the extent of the depolarization elicited by the positive current steps (Fig.8 A2 ). The greatest contribution of the TTX-sensitive component (3.3 ± 0.5 mV; range, 2.5–4.0 mV;n = 3) was at potentials closest to firing threshold (Jahnsen and Llinás, 1984a,b).
To investigate the possibility that I NaP might also have an involvement in the burst firing of TC neurons, neurons were held at −70 mV, and positive current steps were then delivered to activate the low-threshold Ca2+ potential and associated burst-firing response. These experiments were performed in the presence of ZD 7288 (200 μm), a specific blocker of the hyperpolarization-activated inward currentI h (Harris and Constanti, 1995; Williams et al., 1997), to eliminate the effect of changes in this current on the measured parameters (Pape, 1994; Hughes et al., 1996). For the neuron shown in Figure 8 B, a 40 pA current step in control conditions elicited a robust low-threshold Ca2+potential crowned by a burst of action potentials. After 1 μm TTX application, however, the same current step only produced a much smaller and delayed low-threshold Ca2+ potential (Fig.8 B1 , left). With a current step of 50 pA the low-threshold Ca2+ potential recorded in the presence of TTX was also delayed compared with that in control conditions (Fig. 8 B1 ,middle), but with a current step of 80 pA the low-threshold Ca2+ potential profiles in control and TTX conditions were almost indistinguishable (Fig.8 B1 , right). The addition of TTX, therefore, had two effects: (1) a large decrease (67 ± 6%;n = 4) in the amplitude of the low-threshold Ca2+ potential (measured at the lowest input current required to evoke a full-blown Ca2+ potential in control conditions) and (2) a significant increase in the latency of the low-threshold Ca2+ potential, measured from the onset of the current step (control: 83 ± 19 msec; TTX: 115 ± 18 msec; n = 4; p < 0.05; pairedt test). Because these changes in burst-firing properties produced by TTX were observed within a narrow range of membrane potentials (−75 to −65 mV) and with relatively small positive current steps (20–200 pA), extreme care was taken to ensure that cell deterioration and shifts in membrane potential did not occur during the course of these experiments. Results in control and experimental conditions were carefully compared, and any neuron that displayed deterioration in the quality of the recording was excluded from the analysis. In addition, low-threshold Ca2+ potentials were evoked in the presence of TTX not only from the same holding potential as in the control conditions (i.e., −70 mV) (Fig.8 B1 ) but also at potentials slightly more negative and positive (Fig. 8 B2 ,B3 ). Thus, positive current steps delivered from −68 mV (Fig. 88B2 ) elicited low-threshold Ca2+ potentials that were markedly reduced in amplitude and delayed compared with the control potentials at −70 mV, suggesting that the effect observed at −70 mV in the presence of TTX was not caused by a negative shift in membrane potential during the experiment. Similarly, holding the neuron at −72 mV (Fig. 8 B3 ), from where a larger underlying I T Ca2+ current is generated, still elicited low-threshold Ca2+potentials that were delayed and displayed profiles that were not superimposable on the control potentials.
The main conclusions of this study are that (1)I NaP exists in TC neurons; (2) a single Na+ current type appears to underlie bothI Na and I NaP, with the latter being formed by a noninactivating component of INa; and (3) I NaP contributes to both tonic and burst firing of TC neurons.
Steady-state and kinetic properties ofINa
The properties of I Na in TC neurons are in general agreement with measurements of the action potential generatingI Na in other CNS preparations. In particular, for I Na activation our values ofV 1/2 and k fall within the range of those reported by other investigators (from −39 to −28 mV and from 4.2 to 7.1, respectively), and similarly for the inactivation (from −90 to −60 mV and from 4.4 to 10.2, respectively) (Huguenard et al., 1988; Sah et al., 1988; Fan et al., 1994; Magee and Johnston, 1995;Safronov and Vogel, 1995). Single channel recordings have reportedI Na decays to be best fitted with a single exponential in motoneuron soma (Safronov and Vogel, 1995) or two exponentials in hippocampal neuron dendrites (Magee and Johnston, 1995). In TC neurons I Na decays could be well fitted with a single exponential, which suggests therefore a single type of Na+ channel population.
From the steady-state inactivation data it seems that at a typical resting membrane potential of −60 mV only ∼20% of the Na+ channel population is available for activation in TC neurons. A hyperpolarizing episode would therefore be expected to remove inactivation and increase the available Na+channel population, allowing for more robust action potential firing on subsequent depolarization. This feature could explain why the frequency of the burst firing evoked by low-threshold Ca2+potentials in rat TC neurons of the dLGN and other thalamic sensory nuclei reaches a maximum of ∼450 Hz, whereas in tonic firing the maximum frequency does not exceed 200 Hz (Huguenard and Prince, 1994;Jahnsen and Llinás, 1984a; Williams et al., 1996). Indeed, it would be interesting to investigate the onset and recovery from inactivation of I Na in intralaminar thalamic nuclei in which burst firing frequencies of up to 1 kHz have been observed (Steriade et al., 1993).
Properties of inactivation
From the experiments on the onset of inactivation at −68 mV, it is evident that the amplitude of I Na decreases to ∼50% of the initial value within 300 msec. This onset has a rapid and a slow component. In neocortical neurons a slow entry into an inactivated state has been shown to have a marked effect on output firing patterns (Fleidervish et al., 1996), whereas the contribution of this process to TC neuron firing remains to be determined. Recovery from inactivation at holding potentials more negative than −70 mV follow single exponential profiles, whereas at more positive (i.e., physiological) potentials an additional slower recovery from inactivation is also seen. The inability to fit recoveries at these potentials with a single exponential indicates that the relative values of the τ of activation and the τ of inactivation converge. At more negative potentials, single exponential fits show that activation is much faster than inactivation (Tóth and Crunelli, 1996).
The decrease in firing frequency observed in TC neurons during long periods of depolarization more than −40 mV (Williams et al., 1996;Turner et al., 1997) will depend on the interplay between the onset of, and the recovery from, inactivation as well as on the contribution of K+ and high-threshold Ca2+currents. The dynamics of this complex interplay in determining the pattern of TC neuron output under different levels of excitability (Turner et al., 1997) could now be carefully examined using both experimental and simulation studies.
Mechanism of INaP manifestation
The three general hypotheses put forward to explain the occurrence of I NaP have been elegantly summarized in a recent review (Crill, 1996). (1) I NaP is a manifestation of the window component ofI Na; (2) I NaP is a distinct channel type that displays kinetic properties different fromI Na; and (3) I NaPis produced by either a loss, or modulation, of inactivation of the same channels that underlie I Na.
The voltage dependence of I NaP and windowI Na in TC neurons was different, because the maximal amplitude was observed at −40 mV and −55 mV, respectively. Another difference was that the maximal amplitude ofI NaP was ∼2% of the whole-cell current, whereas the calculated peak I Na window current would account for only 0.2% of the whole-cell current. In addition, changes in the inter-ramp holding potential were seen to affect markedly the amplitude of I NaP elicited during a subsequent voltage ramp, whereas window current amplitude should be independent of the holding potential (Hirano et al., 1992). We therefore conclude that I NaP in TC neurons cannot be explained as a manifestation of theI Na window current.
Because of the lack of availability of toxins specific for different subtypes of Na+ channels, it is only possible to distinguish between channel types on the basis of kinetics or single channel conductance. There is a clear difference in the activation range of I NaP and I Na in TC neurons that has been found in almost all other cell types studied (French et al., 1990; Saint et al., 1992). This point, however, is insufficient to prove the involvement of a different channel type and could be attributed to the appearance of another “mode” of channel gating (Alzheimer et al., 1993) or to a subpopulation of Na+ channels having “lost” their ability to inactivate. I NaP could therefore arise from this population of channels, as predicted from the Hodgkin and Huxley model (Hodgkin and Huxley, 1952). The lack of inactivation has been suggested as a likely mechanism in neocortical pyramidal neurons (Brown et al., 1994).
The inactivation hypothesis was tested by including papain in the recording pipette. This caused a removal of inactivation ofI Na, transforming the rapidly inactivating current into a sustained one. The greatly enlarged current seen during ramp protocols with papain-containing electrodes had a voltage dependence similar to the I NaP evoked during similar voltage protocols in control conditions. Current–voltage relations of I papainconstructed with voltage step protocols were also similar toI NaP in untreated cells, displaying comparable voltages of maximal amplitude and similar voltage dependence. Activation curves constructed for I papain also had V 1/2 and k values that were not different from those of I NaP recorded with papain-free electrodes. Cell-attached recordings in other neuronal types have also failed to detect a distinct Na+channel population underlying I NaP (Alzheimer et al., 1993; Magee and Johnston, 1995; Safronov and Vogel, 1995), and it is clear that the only conclusive way of ruling out the involvement of a different Na+ channel subtype would be to perform single channel recordings in TC neurons. In Purkinje neurons, a “resurgent Na+ current” has recently been described that is thought to derive from Na+channels recovering from inactivation via an open state (Raman and Bean, 1997). The properties of such a current, however, cannot explain our findings that the greatest I NaP contribution occurs below firing threshold in current-clamp experiments and thatI NaP is increased by the removal of inactivation by papain. In view of the striking similarity ofI papain and I NaP, however, we suggest that the existence of a different channel type is not required to explain I NaP in TC neurons and that at present the most parsimonious conclusion of our results is that the appearance of I NaP is caused by the absence of inactivation in a subset of the Na+ channel population underlying I Na. The fact that one channel type potentially underlies two very different electrophysiological roles is in itself very interesting, as is the possibility that the mechanism responsible for the loss of inactivation may be under cellular control, a suggestion supported by our experiments on the removal of inactivation by intracellular papain.
Physiological role of INaP
The TTX-sensitive I NaP first activated at membrane potentials around −70 mV, confirming the observation in other systems that this current activates at potentials more negative than those of I Na. Indeed, theV 1/2 of I NaP in TC neurons is ∼20 mV more negative than the V 1/2of the I Na measured in this study and similar to the one found in dorsal root ganglion cells and in hippocampal and neocortical neurons (French et al., 1990; Brown et al., 1994; Baker and Bostock, 1997). This relatively more negative activation ofI NaP suggests a number of putative physiological roles in TC neurons over a wide range of membrane potentials.
A maximum amplitude of ∼150 pA in adult rat TC neurons suggests that for a neuron with an apparent input resistance of 100 MΩ, activation of I NaP could cause a depolarization of 10 mV. The current–voltage relation for I NaP shows that between 60 and 80% of this current will be activated in the membrane potential range between −60 mV and action potential firing threshold, predicting therefore that I NaP would have its greatest influence on membrane potential in this voltage range. Indeed, the contribution of a TTX-sensitive,I NaP-mediated component to the voltage response of TC neurons to positive current steps was confirmed to be maximal in the voltage range immediately below firing threshold, but its amplitude was much less than predicted. This could be attributable to the large effect of K+-dependent rectification in this voltage region or to the reduction of the I NaP elicited during prolonged depolarization at potentials more than or equal to −60 mV (as seen in the conditioning potential experiments). Overall, previous studies in guinea pig and bird TC neurons have shown a larger amplitude (up to 10 mV) of a similar TTX-sensitive component, a result that can be explained by the presence of a higher extracellular K+ concentration and the consequent smaller contribution of K+-dependent outward current in the depolarizing responses observed in these studies (Jahnsen and Llinás, 1984b; Ströhmann et al., 1994; Tennigkeit et al., 1996). In contrast to the prominent role of I NaPin the high-frequency oscillations of cortical neurons, the presence of this current is not essential for the expression of these oscillations in TC neurons (Pedroarena and Llinás, 1997).
The activation of I NaP at around −70 mV would also suggest a possible physiological role at this voltage level. IfI NaP activation was sufficiently rapid, then a synergistic relationship of I NaP andI T in the generation of low-threshold Ca2+ potentials and associated burst firing would be expected. Despite the fact that I NaP is predicted to be small at potentials close to −70 mV, the effect of removing this current on the amplitude and latency of a low-threshold Ca2+ potential was remarkable. Indeed, although the action of I NaP on the low-threshold Ca2+ potential was confined to a small range of membrane potentials, it undoubtedly affected both the efficacy of burst firing and the delay between triggering and firing. To the best of our knowledge this is the first example of I NaPfulfilling an important amplification role in this voltage range, and it suggests a potentially pivotal involvement of this current in the burst-generating mechanism of TC neurons and a possible modulatory target. Moreover, in view of the presence of dendritic T-type Ca2+ channels in TC neurons (Zhou et al., 1997) and their involvement in information processing (Guido and Weyland, 1995), our findings of a contribution of I NaP to burst firing and its somatodendritic location suggest that this current plays a major role in the integration of sensory and cortical inputs over a relatively wide range of membrane potentials.
The work was supported by the Wellcome Trust (Grant 37089). We thank Mr. D. W. Cope, Dr. G. Erdemli, and Dr. N. Leresche for expert advice and assistance during the course of this study, Dr. T. I. Tóth for helpful comments on this manuscript, and Mr. T. Gould for invaluable technical expertise.
Correspondence should be addressed to Vincenzo Crunelli, Physiology Unit, School of Molecular and Medical Biosciences, University of Wales Cardiff, Museum Avenue, Cardiff CF1 3US, UK.