Pyramidal cells of rodent presubiculum express a tetrodotoxin-insensitive Na⁺ current

Abstract

Presubicular neurons are activated physiologically by a specific preferred head direction. Here we show that firing in these neurones is characterized by action potentials with a large overshoot and a reduced firing frequency adaptation during repetitive firing. We found that a component of the sodium current of presubicular cells was not abolished by tetrodotoxin (TTX, 10 μm) and was activated at more depolarized voltages than TTX-sensitive currents. This inward current was completely abolished by the removal of external sodium, suggesting that sodium is the charge carrier of this TTX-insensitive (TTX-I) current. The channels responsible for the TTX-I sodium current seemed to be expressed at sites distant from the soma, giving rise to a voltage-dependent delay in current activation. The voltage required for half-maximal activation was −21 mV, and −36 mV for inactivation, which is similar to that reported for Na_V1.8 sodium channels. However, the kinetics were considerably slower, with a time constant of current decay of 1.4 s. The current was not abolished in pyramidal cells from animals lacking either the Na_V1.8 or the Na_V1.9 subunit. This, possibly novel, TTX-I sodium current could contribute to the coding functions of presubicular neurons, specifically the maintained firing associated with signalling of a stable head position.

The presubiculum is a region situated between the subiculum and the entorhinal cortex that contains densely packed pyramidal and stellate cells with small somata. Pyramidal cells are arranged in a columnar, layered structure (O’Mara et al. 2001; Nishikawa et al. 2002); Rolls, 2006) that differs from the loose organisation of somata in the subiculum. Axons of presubicular pyramidal cells ramify in both the entorhinal cortex and subiculum (Funahashi & Stewart, 1997). Reciprocally the presubiculum is innervated by fibres from both the subiculum and the entorhinal cortex, and so may participate in recurrent loops with both neighbouring regions. In vitro studies have shown how re-entrant loops involving the presubiculum may contribute to the spread of synchronous epileptiform activity in both the subiculum and entorhinal cortex (Funahashi et al. 1999; Harris & Stewart, 2001; Menendez de la Prida et al. 2003). Studies on behaving animals suggest that the presubiculum has a specific role in space coding (Rolls, 2006). Some subicular cells are sensitive to head direction (Taube et al. 1990) and others signal a combination of place and direction of movement (Cacucci et al. 2004; Moser et al. 2008). Even so, relatively little is known about the physiology of single presubicular cells and how it contributes to their coding capacities.

Coding and the transmission of information between neurones depends significantly on sodium channels (Bean, 2007). In this study on pyramidal cells in the presubiculum, we observed an inward current that seems to be carried by Na⁺ but is not suppressed by TTX. Most identified Na⁺ channels are antagonised by tetrodotoxin (TTX), but several isoforms have been identified that are resistant to TTX. These TTX-resistant isoforms include the Na_V1.5, Na_V1.8 and Na_V1.9 α subunits. Both Na_V1.5 and Na_V1.8 underlie high-threshold, inactivating currents with fast kinetics while Na_V1.9 gives rise to slowly inactivating persistent currents that activate at subthreshold potentials. Na_V1.5 is expressed by cardiac muscle cells and by some neurones (White et al. 1993). Na_V1.8 and Na_V1.9 are expressed by primary sensory neurones (Souslova et al. 1997) and have been linked with inflammatory processes (Akopian et al. 1996, 1999b) and the perception of pain (Renganathan et al. 2001).

We first examined repetitive firing properties in presubicular neurones in acute slices, and found that they discharged with very little adaptation upon positive current injection. Action potentials were characterized by a large overshoot. Using whole-cell voltage clamp recordings we then showed that besides the fast TTX-sensitive sodium current, presubicular cells possess a sodium current activated at more depolarized voltages, that was insensitive to high doses of tetrodotoxin (TTX-I). We examined the voltage dependence of this TTX-I sodium current, which was similar to that of currents mediated by the Na_V1.8 subunit. However, we found that the current was still present in both Na_V1.8^−/− and Na_V1.9^−/− mice, suggesting that it is not generated by either of these TTX-resistant sodium current isoforms. These results suggest that presubicular neurones express an unidentified, TTX-I sodium current. This current, together with other intrinsic, voltage-dependent conductances could endow presubicular cells with the capacity for persistent firing as long as the angular head direction is maintained.

Methods

Slice preparation

Animal procedures were performed according to the European Committee Council Directive of November 24, 1986 (86/89/EEC) and INSERM guidelines. We took note of the editorial by G.B. Drummond (Drummond, 2009) and confirm that suitable ethical standards have been met. Transverse slices containing the hippocampus, subiculum and entorhinal cortex were prepared in most experiments from Sprague–Dawley rats aged 21–35 days. In some experiments similar transverse slices of the hippocampal formation were prepared from transgenic mice (C57/Bl6) in which either the Na_V1.8 or the Na_V1.9 gene was inactivated (Akopian et al. 1999b; Ostman et al. 2008) and from their non-inactivated littermates. The sodium channel genes SCN10A or SCN11A were inactivated by replacing exon 4 and 5 by a cassette that contains a neomycin-resistance gene (neo) under the control of the phosphoglycerol kinase gene (PGK) promotor, eliminating the S4 voltage sensor of domain I. In Na_V1.8 knock-out mice the PGK-neo cassette terminates with stop codons in all three frames. In Na_V1.9 knock-out a frameshift leads to the introduction of stop codons within exon 6. In either case, animals were anaesthetized deeply with ketamine hydrochloride and xylazine (80 and 12 mg kg⁻¹, respectively). They were perfused via the heart with a cold solution (2–6°C) containing (in mm) 1 KCl, 26 NaHCO₃, 1 CaCl₂, 9 MgCl₂, 248 sucrose, 10 d-glucose and equilibrated with 5% CO₂ in O₂ (cutting solution; the compositions of all solutions are summarized in Table 1). A tissue block containing the hippocampal formation was prepared and 300 μm slices were cut in a Vibratome (Microm HM 650V, Walldorf, Germany) in the same cold, low-Na⁺ ACSF. Slices were stored for at least 1 h in a chamber containing a solution of composition 124 NaCl, 2.5 KCl, 26 NaHCO₃, 1 NaH₂PO₄, 2 CaCl₂, 2 MgCl₂ and 11 d-glucose (in mm, see ACSF, Table 1). They were maintained at 22–25°C and gently bubbled with 5% CO₂ in O₂ (pH 7.3, osmolarity 305–315 mosmol l⁻¹). For recording, one slice was transferred to a chamber, of volume ∼2 ml, mounted on the stage of an Axioskop 2 FS plus microscope (Zeiss, Le Pecq, France).

Table 1.

Solutions

Extracellular	NaCl	KCl	NaH2PO4	NaHCO3	Hepes	CsCl	TEA-Cl	CaCl2	MgCl2	Sucrose	Glucose	NMDG-Cl or TEA-CL or LiCl
Cutting	—	1	—	26	—	—	—	1	9	248	10	—
ACSF	125	2.5	1.25	25	—	—	—	2	2	—	10	—
ExTNM	26	—	—	—	10	10	112	0 or 0.5	4	—	11	—
ExTNM Na+ subst.	—	—	—	—	10	10	112	0.5	4	—	11	26
ExTNM + 2 K+	26	2	—	—	10	10	112	0.5	4	—	11	—
ExTNM Na+
subst. + 2 K+	—	2	—	—	10	10	112	0.5	4	—	11	26
Intracellular	Potassium gluconate			KCl	CsCl	Hepes	EGTA	MgCl2	Tris-GTP	Mg-ATP	Sodium phosphocr.
CsCl	—			—	133	10	10	2	0.4	4	10
Pot gluc	110			20	—	10	10	2	0.4	4	10
CsCl + KCl	—			20	113	10	10	2	0.4	4	10

Concentrations are indicated in mmol l⁻¹. In voltage clamp experiments, the following blockers were routinely added to ExTNM and ExTNM Na⁺ subst : TTX (1 μm), nifedipin (10 μm), mibefradil (10 μm). Flufenamic acid (10 μm) was added for n= 14 recordings as indicated. All intracellular solutions were added with 4 MgATP, 0.4 Tris-GTP and 10 Na₂-phosphocreatine (in mm). The pH of intracellular solutions was adjusted to 7.3 with KOH for the potassium gluconate-based solution and CsOH for the CsCl-based solution.

Recording techniques and solutions

Neurones were visualised with a DAGE-MTI tube camera (Michigan City, IN, USA) using infrared differential interference contrast. The presubiculum was distinguished from the subiculum by the smaller size (10–15 μm) and higher density of neuronal somata. Whole-cell records were made from presubicular cells with glass pipettes. Cells with initial leak conductance >2.8 nS were excluded (ramp leak conductances measured between −100 and −50 mV in ExTNM (see below; see also Table 1)). The pipette solution contained 133 CsCl, 10 Hepes, 10 EGTA, 4 MgATP, 0.4 Tris-GTP and 10 disodium phosphocreatine (in mm, see Table 1) and their resistance was 3–6 MΩ. We used an Axopatch 200A amplifier (Axon CNS, Molecular Devices) in either voltage clamp mode or current clamp fast mode for current clamp recordings. We note that action potential amplitudes and spike afterhyperpolarizations may be exaggerated by this type of amplifier (Magistretti et al. 1998). Reported voltages are not corrected for junction potential. The effective junction potential was estimated as +5.1 mV at 25°C with the pCLAMP (Axon Instruments) calculation function for a Cl⁻-coated silver wire contact with the pipette solution and a 3 m KCl agar bridge contact with the external solution.

The external solution used for current clamp recordings was the same as that used for slice storage. It was perfused at 15 ml min⁻¹ and the recording chamber was heated to 34°C. pH was buffered to ∼7.3 using a bicarbonate–CO₂ system with 26 mm NaHCO₃ and 5% CO₂ in O₂. In voltage clamp experiments, the temperature was reduced to 22–25°C in order to slow current kinetics and aid voltage clamp control. A Hepes pH buffer system was used in the external solution to allow for substitution of Na⁺ by impermeant cations. The external solution used to isolate the TTX-insensitive sodium current (ExTNM, Table 1), contained no added K⁺, NaHCO₃ was absent and the NaCl concentration was reduced to improve space clamp control of Na⁺ currents. ExTNM was composed of 26 NaCl, 10 Hepes, 11 glucose, 10 CsCl, 112 TEA-Cl, 0.5 CaCl₂, 4 MgCl₂ (in mm; pH 7.3, osmolarity 318 mosmol l⁻¹). The Ca²⁺ current antagonists nifedipine (10 μm) and mibefradil (10 μm) were routinely added, as well as TTX (1 μm). Substitution for sodium was achieved by an equimolar, 26 mm, replacement with lithium chloride, N-methyl-d-glucamine (NMDG) or tetra-ethyl ammonium (TEA) (ExTNM Na⁺ subst., Table 1).

A concentration of 1 μm tetrodotoxin entirely blocks TTX-sensitive currents (Roy & Narahashi, 1992; Blair & Bean, 2002). The TTX-resistant isoforms, Na_V1.8 and 1.9, possess values for K_i of 40–60 μm (Sivilotti et al. 1997; Cummins et al. 1999; Renganathan et al. 2002). IC₅₀ values of ∼150 nm are reported for TTX-resistant sodium currents in entorhinal cortex (White et al. 1993), similar to those for TTX-resistant sodium currents in cardiac cells, ranging from 150 nm (Renaud et al. 1983) to 6 μm (Brown et al. 1981). IC₅₀ values of 2–6 μm are found for Na_V1.5-containing subunits in dorsal root ganglion cells (Renganathan et al. 2002).

Flufenamic acid (10 μm) was used to suppress Ca²⁺-activated non-selective cation currents mediated by TRP-like channels (Roy & Narahashi, 1992; Partridge & Valenzuela, 2000). In some experiments we examined the effects of the antagonists ω-conotoxin GIVA (2 μm), ω-agatoxin IVA (200 nm). We also examined effects of the local anaesthetic lidocaine (1 mm) which reportedly blocks TTX-resistant sodium current (Scholz et al. 1998).

Data acquisition and analysis

Pipette current and voltage was controlled and digitized at 10 kHz with a Digidata 1320A analog–digital converter (Axon Instruments) and stored on a PC for analysis with pCLAMP software (Axon Instruments – Molecular Devices, Sunnyvale, CA, USA). Online leak subtraction using a –p/6 protocol was routinely applied to voltage clamp recordings of current activation and inactivation, and voltage ramp protocols. Input resistance and neuronal firing patterns were measured in current clamp mode with current pulses of 200–800 ms duration. Membrane time constant was estimated from the voltage response to an injected pulse of negative current, fitted to a double exponential function using Igor Pro software (WaveMetrics, Lake Oswego, OR, USA). Currents for activation and inactivation curves were measured in voltage clamp experiments as the average value at a delay of 10–100 ms after the start of the pulse (Fig. 5A). Firing rate curves were obtained with incremental current steps of 20 pA up to 140 pA. Action potential properties were analysed using custom software written in Labview (Spikoscope, Ivan Cohen). Voltage threshold was defined as the point where the action potential slope exceeded 10 mV ms⁻¹ (Fricker et al. 1999). Action potential half-width was determined as its duration at half-amplitude (Bean, 2007). The fast afterhyperpolarization (AHP) was measured as the voltage minimum in the 20 ms after the peak of an action potential. Quantitative data are given as means ±s.e.m. and statistical significance was tested using Student's t test.

Figure 5. The TTX-I current may originate at a distance from the soma.

A, depolarizing steps were applied from a holding potential of −80 mV. Reponses to steps to −50, −30 −10, +10 and to +30 mV are shown. For just-threshold steps the current onset was delayed (–30 mV, arrow). B, average delay to peak was voltage dependent. C and D, voltage dependence of deactivation. The TTX-insensitive current was activated by a 100 ms step depolarization to −20 mV, followed by a 1400 ms duration step to various potentials. Holding potential between trials was −60 mV. Sample traces in C show a complex deactivation that occurred in multiple steps. D, voltage dependence of deactivation at two time points, 1 ms after offset of the test potential (blue circles, linear I–V relation) and after 1300 ms (red triangles). The current response to a hyperpolarizing voltage ramp from +70 to −110 mV recorded in the same cell is shown in black. Note the inflection during deactivation. E, current response to triangular-shaped upward then downward voltage command from −140 to +70 mV then back to −140 mV. The holding potential was −60 mV. F, hysteresis in responses to rising and falling ramp. Same current trace as in E, plotted as a function of voltage. The inward current on the downward ramp is smaller than on the ascending ramp due to partial inactivation. Voltage for activation (red dot, ascending ramp) was more depolarized than the voltage for channel closure upon hyperpolarization (red dot, descending ramp).

Anatomy and immunohistochemistry

Neurones were filled passively with biocytin for post hoc morphological identification, using a concentration of 1 mg ml⁻¹ biocytin in the pipette solution. Slices were fixed overnight at 4°C in a solution containing 4% paraformaldehyde in 0.12 m phosphate buffer (PB, pH 7.4), rinsed in PB, cryoprotected in 30% sucrose and frozen on dry ice. Biocytin-filled neurones were revealed in some experiments using the avidin–biocytin peroxidase reaction (Vectastain Elite Rabbit IgG kit, Vector, Burlingame, CA, USA) with 3,3′-diaminobenzidine (DAB, 0.7 mg ml⁻¹) as the chromogen. In other experiments a streptavidin–Alexa Fluor 488 conjugate (1 : 200, Invitrogen, Eugene, OR, USA) was used. Filled cells were visualized with an Olympus IX 81 microscope equipped with a Volocity Grid system (Improvision, Coventry, UK). Anatomical data were stored as stacks of images (50 to 120 images) and analysed. Three-dimensional reconstructions were represented as 2-D images (Fig. 1A).

Figure 1. Cellular properties of presubicular cells.

A, 3-D reconstruction of the soma and principal dendrites of a Alexa Fluor 594-filled presubicular pyramidal cell. B, spontaneous EPSPs initiated action potentials. The membrane potential was −60 mV. C, depolarizing current injections to −51 mV (maintained) and −49 mV (step) initiated cell firing. D, action potential peak, half-width, amplitude and threshold measurements. The amplitude was defined as the spike height relative to the most negative voltage reached during the 20 ms following the peak. The half-width was taken at half-maximal spike amplitude. Threshold was defined as the voltage where dV/dt exceeds 10 V s⁻¹.

Results

Anatomical and electrical properties of presubicular neurones

Neurones recorded in this study were located in the superficial layers of the presubiculum. Their somata visualised under infra-red differential interference contrast (IR-DIC) optics were round or pyramidal shaped with a diameter close to 10 μm. Recorded cells filled with biocytin (n= 18) were revealed as neurones with apical dendrites extending towards the cortical surface (Fig. 1A). Some cells were clearly pyramidal in form (n= 10) while in other cells (n= 8), some proximal apical dendrites extended horizontally before turning towards the cortical surface. There were no consistent differences in the electrical properties of cells with these two distinct dendritic orientations. Neuronal input resistance was 386 ± 40 MΩ, and membrane time constant was τ= 29.5 ± 4.3 ms (n= 8; mean ±s.e.m.). Whole-cell current clamp recordings in our conditions showed that presubicular cells discharged only occasionally, usually when spontaneous EPSPs evoked firing (Fig. 1B). Their resting potential was −61 ± 4 mV. Sustained action potential discharge could be elicited by small step or maintained depolarizing current injections (Fig. 1C). Mean firing threshold was −40.5 ± 1.8 mV (n= 8) and the mean threshold current to initiate firing was 37.5 ± 5.9 pA. As shown in Fig. 1D, action potentials were highly overshooting with a voltage peak reaching 48.1 ± 3.7 mV, and total amplitude of 101.8 ± 4.1 mV, as measured between peak and AHP (see comment on possible distortion in Methods). Action potential duration at mid-height was 1.0 ± 0.07 ms (n= 10 cells).

When long depolarizing current steps were applied, the number of action potentials generated increased sigmoidally with applied current (Fig. 2A and B). Firing frequency showed little adaptation. The mean spike frequency in response to just suprathreshold stimuli was 10.7 ± 1.7 Hz. At higher intensities, intervals between action potentials changed little between the start and the end of the current injection (Fig. 2C). The ratio of the first interspike interval in a train of action potentials to the mean interspike interval was 0.81 ± 0.05 (n= 8). This relative lack of adaptation (Fig. 2D and E) is unusual for hippocampal and cortical pyramidal cells. For instance, the same interspike interval ratio is close to 0.35 for pyramidal cells of rat barrel cortex (Goldberg et al. 2008).

Figure 2. Repetitive firing properties of presubicular cells.

A, repetitive firing in response to step depolarizations (+40, +60 and +100 pA injected current). B, plotting the neuronal firing frequency as a function of the injected current showed a sigmoid input–output relation (n= 8). C, instantaneous firing frequency varies little during long current injections. The different colours show values for a series of five levels of current injection (ranging from +60 to +140 pA) into the same cell. D, frequency increases with current strength in a similar fashion for the 1st, 2nd and 4th action potential and at steady state (mean of the following action potential frequencies), indicating the relative lack of spike-frequency adaptation at different degrees of depolarization. E, cumulative probabilities of instantaneous spike frequencies for a +60 pA current injection gave steep curves for n= 8 cells examined (different colours for different cells), attesting to the relative absence of accommodation in presubicular neurons.

Identification of a TTX-insensitive sodium current in presubicular neurones

In whole-cell voltage clamp records, an inward current persisted in the presence of both the Na⁺ channel antagonist TTX and the Ca²⁺ channel antagonists nifedipine and mibefradil. The current–voltage relations of this TTX-insensitive current were first examined using ramp stimuli (Fig. 3). Holding potential was −60 mV and a ramp was applied at 120 mV s⁻¹ between −140 and +70 mV. Figure 3A illustrates currents evoked by this protocol in control conditions. As the depolarizing ramp was applied, I_h was activated giving rise to an inward current, then sodium action currents escaped the clamp above −50 mV, and at more depolarized potentials voltage-dependent calcium and potassium currents were activated. We then applied the same protocol to cells recorded in conditions designed to isolate TTX-insensitive currents (ExTNM, Table 1, n= 90; Fig. 3B). Current–voltage relations were approximately linear between −100 and −50 mV, with a mean leak conductance of 1.6 ± 0.14 nS (n= 49), corresponding to a membrane resistance of 625 MΩ. At a mean threshold of −17 ± 1 mV, a slow persistent inward current developed producing a negative slope region. It reached a mean peak inward amplitude of −493 ± 36 pA, and changed polarity from inward to outward at 23 ± 1 mV (n= 49). Amplitude measures were taken 1–3 min after the blocking solution ExTNM reached the slice preparation, i.e. less than 5 min after break-in to whole-cell configuration.

Figure 3. A TTX-insensitive Na⁺ current in presubicular cells.

A, current response to a depolarizing voltage ramp (120 mV s⁻¹) applied in the absence of TTX and Ca²⁺ channel blockers (not leak-subtracted). B, ramp in the presence of TTX (1 μm), mibefradil (10 μm) and nifedipine (10 μm) at 0.5 mm Ca²⁺ (not leak-subtracted). C, ramp in the presence of TTX (1 μm), flufenamic acid (10 μm), mibefradil (10 μm) and nifedipine (10 μm) at nominally 0 mm Ca²⁺ (not leak-subtracted). D, the activation threshold for the TTX-insensitive inward current was similar in 0.5 and 0 mm external calcium. E and F, the reversal potential was lower (*P < 0.01, Student's t test) and the inward current amplitude decreased in 0 mm Ca²⁺ (P < 0.1, Student's t test). Error bars represent mean ±s.e.m.G, ramp in the presence of TTX (1 μm), mibefradil (10 μm) and nifedipine (10 μm) at 0.5 mm Ca²⁺ (ExTNM). Substituting TEA for Na⁺ in the external solution abolished the inward current (upper of the two traces, ExTNM Na⁺ subst., see Table 1). H, same protocol as G, now with 20 mm KCl in the internal CsCl based solution (see Table 1) and 2 mm K⁺ externally (ExTNM + 2 K⁺). Substituting TEA for Na⁺ in the external solution abolished the inward current (upper of the two traces, ExTNM Na⁺ subst. + 2 K⁺, see Table 1). The higher noise level in G and H was due to online leak-subtraction. I, 1 mm lidocaine significantly suppresses the inward current over time (lower trace, P < 0.001 at 15 min; n= 4) compared to rundown of the TTX-insensitive current in control (upper trace).

Most cells (63 out of 69 tested) expressed a TTX-insensitive inward current that was larger than 150 pA. Further studies were not performed on cells where the current was small or absent. These cells did not differ in terms of other electrical properties or morphology from the cells that did express the current. The TTX-insensitive current often decreased over time (see Fig. 3I), with a rundown of about 40% over the first 10 min of a recording, which is similar to the kinetics of rundown reported for Na_V1.9 TTX-resistant sodium channels (τ_rundown= 13 min, Coste et al. 2004).

Flufenamic acid was used to suppress Ca²⁺-activated non-selective cation currents mediated by TRP-like channels (Roy & Narahashi, 1992; Partridge & Valenzuela, 2000; Yoshida & Hasselmo, 2009). The potential contribution of a TRP-mediated conductance to the TTX-I current was evaluated by comparing two sets of experiments. Neurones (n= 29) were recorded in the ExTNM solution without flufenamic acid over at least 5 min to assess TTX-I current amplitude and rundown in the presence of the blocking cocktail in the external solution, ExTNM. These data were compared to currents from 14 neurones where flufenamic acid (10 μm) was added on switching from ACSF to ExTNM. The maximal inward current amplitude was −464 ± 31 pA (n= 14) in the presence of flufenamic acid and −525.8 ± 56 pA (n= 29) when it was absent (P= 0.35, not significant), suggesting that I_CAN channels, calcium-activated nonselective cation channels, did not contribute significantly to the TTX-I current.

We tested the possibility that a residual calcium conductance contributed to this inward current. In one set of experiments, the concentration of external calcium in the ExTNM was reduced from 0.5 mm to nominally zero (n= 7, Fig. 3C) for 15–60 min before establishing whole-cell records. Omitting external calcium led to small changes in the activation threshold, which was reduced to −13 ± 2 mV (n= 7, Fig. 3D) and reversal potential hyperpolarized to 18 ± 1 mV (n= 7, Fig. 3E). The maximum inward current amplitude was reduced by about 30% to −351 ± 56 pA (n= 7, Fig. 3F). This remaining current was not affected by the Ca²⁺ channel antagonists agatoxin IVA (200 nm) and conotoxin GVIA (2 μm) (maximal inward current amplitude −479 ± 104 pA, n= 3, P= 0.72, not significant). These data suggest that Ca²⁺ is not a permeant ion, although the TTX-insensitive current may be modulated by Ca²⁺.

If the current is not carried by calcium, it might be carried by sodium ions. Equimolar substitution (26 mm) of sodium by the impermeable cations NMDG (n= 6), lithium (n= 7) or TEA (n= 3) completely abolished the TTX-insensitive current (Fig. 3G). Substituting Na⁺ in the presence of 0.5 mm Ca²⁺ offered further evidence against a Ca²⁺ permeability, since the current was completely eliminated in these conditions (n= 16).

Sodium ions may permeate through some potassium channels when potassium ions are absent from both intra- and extracellular compartments (Callahan & Korn, 1994; Korn & Ikeda, 1995). We therefore asked whether the TTX-I current was abolished by recording in the presence of both external and internal K⁺ (external ExTNM + 2 mm K⁺; internal CsCl-based solution with 20 mm KCl, see Table 1). Under these conditions the inward current was present (maximum inward current −359 ± 36 pA, n= 7; P= 0.016 compared to zero K⁺ solutions ExTNM and CsCl internal). The reversal potential of the current was not significantly changed (24.6 ± 2.4 mV, n= 7; P= 0.59). Furthermore, substitution of sodium ions in the external solution (ExTNM Na⁺ subst. + 2 mm K⁺, see Table 1) by impermeant anions in the presence of external K⁺ (2 mm) completely abolished the inward current. These data indicate that the permeation of sodium ions through potassium channels cannot fully account for the TTX-I (Fig. 3H).

As a further test of the Na⁺ dependence of the current, we examined the effects of the local anaesthetic lidocaine, which is reported to block TTX-resistant sodium currents as well as those sensitive to TTX (Roy & Narahashi, 1992). Since lidocaine acts slowly, we compared TTX-I current amplitudes recorded when it was present or absent at the same time point, 15 min after break-in, to control for the effects of rundown. The current was significantly suppressed by more than 90% after 15 min of exposure to 1 mm lidocaine (n= 4; P < 0.001; Fig. 3I).

Activation by step depolarizations

We next examined the activation and inactivation of the TTX-insensitive current in responses to voltage clamp steps (Fig. 4A). Responses to voltage pulses from a holding potential of −80 mV to test potentials between −60 mV and +40 mV let us construct current–voltage curves. These curves showed a similar voltage dependence to that determined from ramp applications. The current activated close to −30 mV and the peak inward current was elicited by steps to −10 mV. The peak inward current determined from step depolarizations, −457 ± 64 pA (n= 14), was similar to that, −493 ± 36 pA, determined with ramp stimuli (n= 49). Activation curves were constructed by dividing current by the driving force to transform I–V curves to conductance. They were then fitted with Boltzmann equations of the form G_Na/G_Na,max= 1/(1 + exp(−(V−V_1/2)/k))³ where V is the command potential, V_1/2 is the potential with a value of 0.5 for the Boltzmann function, and k is the slope factor (k= 4.5 ± 1.1). This procedure provided a value of −21 mV for the midpoint of activation of the TTX-I sodium current. This value is more depolarized than for TTX-sensitive sodium currents, but similar to that reported for currents carried by channels containing the Na_V1.8 subunit (Renganathan et al. 2001; Blair & Bean, 2002).

Figure 4. Activation and inactivation properties.

A, current–voltage relation of the TTX-insensitive current derived from step depolarizations from −80 to +60 mV (n= 8). The inset shows an example of current traces from a family of depolarizing voltage steps. The mean inward current between 50 to 120 ms after step onset served to construct the I–V curve. B, activation and inactivation curves for the TTX-I Na⁺ current. The activation curve (red) is a conductance transform of the I–V curve in A. The inactivation curve (blue) was obtained from a voltage protocol composed of prepulse of 3.5 s duration at the test holding potentials between −100 to +40 mV, followed by an activating step to 0 mV. The inward currents were normalized to the conductance obtained with a −100 mV prepulse, and plotted as a function of the prepulse holding potential. C, current inactivation is slow. The TTX-insensitive inward current activated by a maintained deploarization to −20 mV decays slowly with time, to 41% of its peak value after 1 s. Curve fitting to a single exponential function yielded τ= 1.41 ± 0.005 s. D, double ramp command from −140 mV to +70 mV and back to −140 mV. The holding potential between trials was −60 mV. The peak inward current on the ascending phase was mirrored by a smaller, sustained peak on the descending phase.

The voltage dependence of inactivation was estimated by measuring currents elicited by an activating pulse to 0 mV applied after an inactivating prepulse of 3.5 s duration to potentials over the range −100 to +40 mV. The protocol was repeated at intervals of 30 s from a potential of −80 mV to prevent cumulative use-dependent effects. Inactivation data were fitted to the equation I/I_max= 1/(1 + exp((V−V_1/2)/k)), where V is the prepulse voltage, V_1/2 is the half-inactivation voltage, and k is the slope factor (k= 15.6 ± 5.6). The resulting value for the midpoint of inactivation was −36 mV (from 8 cells).

Inactivation was slow with a time constant of several hundred milliseconds Figure 4C shows the time course of the current activated by a step depolarization to −20 mV from a holding voltage of −100 mV. Such steps evoked an inward current of peak amplitude −564 ± 68 pA, followed by a slowly decaying, persistent phase (n= 12). After 1 s at −20 mV, the current had decreased to 41% of its initial peak (−229 ± 26 pA, n = 12). We confirmed this slow, time-dependent inactivation by comparing inward currents elicited during the depolarizing and the hyperpolarizing components of triangular-shaped voltage clamp waveforms as in Fig. 4D. The peak inward current during the ascending phase was −494 ± 43 pA, and that recorded during the descending phase was −236 ± 21 pA (n= 32), providing evidence for a reduction of 48% of the current at an interval of about 1.4 s. These data demonstrate a strong persistent component of the TTX-insensitive Na⁺ current.

Evidence TTX-I current is generated at sites distant from the soma

Several observations suggested that channels underlying the TTX-I current were located at a distance from the somatic recording site. The onset of the current could be slow or delayed and with a variable time to peak, particularly for depolarizations to potentials near activation threshold (Fig. 4A, Fig. 5A and B). Such a delay is compatible with a poor point clamp control of inward currents generated at a distant site (Muller & Lux, 1993; Bar-Yehuda & Korngreen, 2008; Williams & Mitchell, 2008). The kinetics of deactivation of TTX-I currents following the offset of depolarizing voltage steps were consistent with this situation. Deactivation was examined at potentials from −80 to +40 mV after a fixed activating pulse to −20 mV for 100 ms (Fig. 5C). At potentials near the midpoint of inactivation (−40 to −60 mV) currents returned to baseline with a delay as long as several hundreds of milliseconds. Deactivation was completed more quickly for steps to more hyperpolarized voltages. In some cells (n= 12; Fig. 5C) deactivation occurred in step-like fashion to an intermediate value before decaying to zero. The duration of steps became shorter in responses to successive trials. Figure 5D plots the voltage dependence of deactivating currents at 1 and 1300 ms after offset of the test potential. At 1 ms the relation is nearly linear, while the voltage dependence of the deactivation process is apparent at 1300 ms. A similar voltage dependence is evident in the current response to a hyperpolarizing voltage ramp from +70 to −110 mV shown as an overlay in Fig. 5D.

Inward currents originating at distal dendritic sites in spinal and hypoglossal motorneurones (Svirskis et al. 2001; Powers & Binder, 2003) underlie a hysteresis in responses to ramp voltage commands. We observed such behaviour in responses of TTX-I currents to triangular ramp commands for most presubicular cells. Figure 5E shows the time course of currents evoked by a sequence of depolarizing and hyperpolarizing ramps and Fig. 5F plots current against voltage during the ramp. During repolarization, a peak current occurring at 11 ± 1 mV mirrored the peak inward current elicited at −15 ± 1 mV during the depolarizing ramp (red dot, ascending ramp; n= 31); although its amplitude was reduced by inactivation (Fig. 4D). A second current component was maintained through more hyperpolarized potentials, before deactivation was completed at −38 ± 2 mV (red dot, descending ramp; n= 31). Together with the variable latency of activation and the current steps seen during deactivation, this hysteresis suggests that the TTX-I current was generated at sites electrically distant from the somatic recording electrode.

Molecular identity of TTX-I sodium current

We pursued the molecular identity of this TTX-insensitive Na⁺ current. Action potentials are typically abolished at concentrations of 0.1 μm TTX, whereas 1 μm TTX is needed to entirely block the classical TTX-sensitive currents (Roy & Narahashi, 1992; Madeja et al. 2001; Blair & Bean, 2002). Na⁺ currents generated by channels containing the Na_V1.5 α subunit are blocked by doses of tetrodotoxin in the range of 10 μm (White et al. 1993). However, 10 μm TTX, a concentration 100 times higher than that which suppressed action potentials in these cells, did not block the current (ramp peak inward currents −453 ± 84 pA, n= 4, Fig. 6A and B). Two other α subunits, Na_V1.8 and Na_V1.9, generate TTX-resistant currents. Those mediated by channels containing the Na_V1.8 subunit activate at similar depolarized potentials, even though these currents tend to inactivate more quickly. The voltage dependence of the current described here (Fig. 4B) differs from that of currents generated by the Na_V1.9 subunit which activate at subthreshold potentials. Both the time-dependent rundown and the kinetics of our TTX-I current are rather similar to Na_V1.9-containing sodium channels. We therefore asked whether a TTX-I current was present in presubicular pyramidal cells of tissue slices prepared from mice in which either the Na_V1.8 subunit or the Na_V1.9 subunit is genetically inactivated (Akopian et al. 1999b; Ostman et al. 2008).

Figure 6. The TTX-I sodium current may be a novel molecular isoform.

A, the TTX-I sodium current is not abolished by high concentrations of TTX (10 μm) as should be the case for Na_v1.5. B, no significant difference in inward current amplitude when applying 1 or 10 μm TTX in the external solution (P > 0.5, Student's t test). C–E, recordings from Na_v1.8^−/− mice showed that the TTX-I sodium current is present in presubicular neurons. Sample traces from step depolarizations (C) and the corresponding I–V curve (D), in recording conditions to isolate the TTX-I sodium current. E, ramp depolarization from −140 to +70 mV with a slope of 120 mV s⁻¹. F, the TTX-I sodium current is present in recordings from Na_v1.9^−/− mice. Same protocol as in A and E. G, the TTX-I ramp peak inward currents are not significantly different between wild-type mice, Na_V1.8^−/− and Na_V1.9^−/− mice, respectively (P > 0.5, Student's t test). Error bars represent mean ±s.e.m.

We first verified that cells of mouse presubiculum exhibit a TTX-insensitive Na⁺ current with similar voltage dependence and kinetics to that of rat presubicular neurones. The average amplitudes of the TTX-I sodium current were smaller in mice, probably due to the smaller cell size. In voltage clamp recordings, we then compared the amplitude of this current in Na_V1.8^−/− and Na_V1.9^−/− mice and their wild-type (WT) littermates (Fig. 6C–F). We included all cells with currents larger than 100 pA, that is 20 out of 21 cells tested. The peak amplitude of the current recorded from WT mice was −249 ± 51 pA (n= 7) whereas the peak amplitude for the current recorded from the genetically inactivated Na_V1.8^−/− animals was −234 ± 53 pA (n= 4), and −254 ± 23 pA (n= 9) for Na_V1.9^−/−. As shown in Fig. 6G, there was no significant difference in the amplitude of TTX-I currents recorded from presubicular cells of WT mice and of Na_V1.8^−/− or Na_V1.9^−/− animals (P > 0.5). Thus, our data do not positively identify the molecular basis of the current.

Discussion

While behavioural work suggests presubicular cells are involved in signalling head direction, there is little information available on their physiology. Here we report that pyramidal cells in superficial layers of the presubiculum express a persistent current that is not blocked by TTX and that seems to be carried by sodium (Fig. 3). It activates and inactivates at more depolarized potentials than the TTX-sensitive Na⁺ current (Fig. 4). Activation and deactivation were delayed suggesting that the current may have been generated at a distance from the soma (Fig. 5). Deactivation occurred at more negative potentials than activation, which may favour bistable membrane behaviour and sustained firing. Our data showed that action potential firing induced in presubicular cells by maintained current was sustained with little accommodation.

A sodium current and not a calcium current?

The current had a mid-point for activation of −21 mV and an inactivation mid-point of −36 mV (Fig. 4B), both considerably more depolarized than the TTX-sensitive Na⁺ current in these cells. Kinetics were very slow with a time constant for inactivation of about 1 s. This time course is comparable to that for inactivation of high-voltage-activated, sustained Ca²⁺ currents (Tsien et al. 1988; Hille, 1992). However the TTX-insensitive current was not affected by nifedipine, a specific antagonist of L-type Ca²⁺ currents, by ω-conotoxin specific for N-type currents, agatoxin specific for P/Q Ca²⁺ currents or by mibefradil which is an antagonist of T-type Ca²⁺ currents. It might still correspond to the R-type Ca²⁺ current which lacks specific antagonists, but contributions of an R-type current should have been abolished in the 0-Ca²⁺ solution used in Fig. 3C. In addition, no residual Ca²⁺ current was detected in experiments where TEA-Cl was substituted for external Na⁺ in an external medium containing Ca²⁺ (Fig. 3D). The lack of antagonism by flufenamic acid suggests the current did not belong to the family of TRP channels. Furthermore, the observation (Fig. 3H) that the current was maintained in the presence of internal or external potassium, suggests that it was not due to the Na⁺ permeability demonstrated for certain potassium channels in the absence of this ion (Callahan & Korn, 1994; Korn & Ikeda, 1995). Instead the observation that the TTX-insensitive current was suppressed by lidocaine suggests it was carried by channels permeable to sodium rather than calcium.

What is the molecular identity of the TTX-I current?

TTX-resistant isoforms of sodium channels include the Na_V1.5, Na_V1.8 and Na_V1.9 α subunits. Na_V1.5 is expressed by cardiac muscle cells (Goldin, 2001; Bennett & Healy, 2008) and perhaps some neurones of the entorhinal cortex (White et al. 1993). Expression of Na_V1.8 and Na_V1.9, which are involved in pain perception (Akopian et al. 1999a; Dib-Hajj et al. 2002; Zimmermann et al. 2007), is thought to be restricted to peripheral sensory neurones (Rush et al. 2007).

The current expressed by presubicular cells seems not to correspond to Na_V1.5 since it inactivates slowly while Na_V1.5 inactivates rapidly (Kerr et al. 2007). Furthermore, while the IC₅₀ for TTX actions on Na_V1.5 is ∼2 μm, the current in presubicular cells was not affected by 10 μm TTX (Fig. 6A). Channels containing the Na_V1.9 α subunit can probably be excluded since they activate close to −60 mV and this current activated close to −20 mV. Furthermore, the TTX-I current was maintained (Fig. 6F) in records from presubicular cells of Na_V1.9^−/− animals (Ostman et al. 2008). The voltage dependence of the current studied here is similar to that of Na_V1.8: its midpoint of activation was −21 mV compared to measured values in dorsal root ganglion cells of −15 mV (Blair & Bean, 2002) and −21 mV (Renganathan et al. 2001) and its midpoint of inactivation was −36 mV compared to −35 mV (Blair & Bean, 2002) and −38 mV (Renganathan et al. 2001). Even so, the kinetics of inactivation were much slower than for Na_V1.8. Furthermore, a TTX-insensitive current of similar amplitude and properties (Fig. 6E) was recorded in presubicular cells from animals genetically deprived of this subunit (Akopian et al. 1999b).

It seems unlikely that this current corresponds to an undescribed molecular species of sodium current. However, currents generated by splice variants of other voltage-gated channels (Tseng-Crank et al. 1994; Bourinet et al. 1999; Chaudhuri et al. 2004; Ouyang et al. 2007; Raingo et al. 2007) including sodium channels (Akopian et al. 1999a; Zimmer et al. 2002; Kerr et al. 2004; Song et al. 2004; Camacho et al. 2006) can exhibit quite different properties. It is interesting that a brain atlas of the distributions of different proteins based on the in situ technique shows signals for both Na_V1.8 and Na_V1.9 in the presubicular region (Scn10a and Scn11a, Allen Institute for Brain Science, http://www.brain-map.org/). However, immunohistochemistry with a poly-clonal antibody raised against the Na_V1.8 subunit from the dorsal root ganglion revealed no specific staining in the presubiculum (C. Dinocourt and D. Fricker, unpublished data; Djouhri et al. 2003).

Unusual kinetic behaviour of the TTX-I current

This current was unusual in that activation could be delayed especially for just supra-threshold depolarizing steps (Fig. 5A). Furthermore, we observed deviations from an exponential decay during deactivation (Fig. 5C). Deactivation tended to occur slowly with one or more steps of faster decay to a smaller current level. While plateaux levels tended to be similar, their duration varied between successive trials.

These phenomena might result from a somatic space clamp that did not perfectly control regenerative electrical events at distant sites (Muller & Lux, 1993). In order to improve space clamp performance, we suppressed most cellular currents and reduced the temperature to 20–25°C. Furthermore, anatomical data (Fig. 1A) show these neurones are rather small with a mean dendritic span of 459 ± 27 μm (n= 18) between the tips of apical and basilar dendrites. Even so, spatially distant sites of current expression may not have been invaded immediately by somatic depolarizing clamp pulses leading to a delayed activation. Inversely, the steps observed during deactivation of the current may reflect a poor voltage control over the termination of plateau potential-like events initiated at distant sites (see Fig. 5F; Svirskis et al. 2001).

A distant site of expression for the current is supported by our observation of a hysteresis, where current–voltage relations during a repolarizing ramp differed from those during a depolarizing ramp (Fig. 5E and F). Not only was the current during the hyperpolarizing ramp smaller due to inactivation but also the return current was prolonged by a plateau-like event. This behaviour might reflect a physiological decoupling between somatic and remote sites permitting the generation of distant and persistent inward currents (Theiss et al. 2007) that could support repetitive firing behaviour as shown in Fig. 2.

Remote sites of expression for the TTX-I current might be dendritic or axonal. A resurgent Na⁺ current in pyramidal cells of the peri-rhinal cortex is expressed at axonal but not somatic sites (Castelli et al. 2007). Persistent Na⁺ currents have been suggested to be expressed by distal dendrites of hypoglossal motoneurones (Powers & Binder, 2003) and in neocortical neurone dendrites (Schwindt & Crill, 1995).

Alternatively, the unusual kinetics might result from an anomalous behaviour of the channel itself. Currents generated by high-threshold Ca_V1 channels in nucleated patches from hypoglossal motoneurones (Moritz et al. 2007) activate with a delay and show a similar step-like relaxation behaviour during deactivation. This may reflect an anomalous voltage-dependent gating where L-type, Ca_V1, channels re-open during hyperpolarizing pulses (Hivert et al. 1999). Since our experiments were done in the presence of nifedipine, the sodium-permeable channel underlying the TTX-I current should also possess anomalous properties. However, in cell-attached patch records made with pipettes containing TTX from presubicular cell somata (D. Fricker, unpublished data), no inward current was detected, arguing against the somatic expression of a current with anomalous gating behaviour.

Role of TTX-I currents in presubicular cells

In the absence of a specific antagonist or a definitive molecular identification it is difficult to pursue the functional significance of the TTX-I current. Sodium currents mediate action potential generation, but the depolarized voltage dependence of the TTX-I currents suggests it does not contribute to the control of the firing threshold. The Na_V1.8 current has been linked to large and strongly overshooting action potentials in dorsal root ganglion cells. The action potential upstroke is controlled by the TTX-sensitive current, whereas the slower TTX-resistant sodium current is the dominant charge carrier at the peak (Renganathan et al. 2001; Blair & Bean, 2002). In presubicular neurones, we noted similarly high amplitude action potentials (Fig. 1), and the TTX-I current could contribute to sodium current at their peak. If the underlying channels are localized on dendrites they would tend to assist action potential retro-propagation (Stuart & Sakmann, 1994; Holthoff et al. 2006).

With particularly slow kinetics, the TTX-I sodium current could act to prolong the decay of large dendritic EPSPs or spine potentials (Lipowsky et al. 1996; Fricker & Miles, 2000; Araya et al. 2007). A current hysteresis such as we observed has been linked with membrane bistability (Hounsgaard et al. 1988; Lee & Heckman, 1998), and sufficiently strong synaptic stimulation may initiate persistent depolarizing potentials (Holthoff et al. 2006; Cai et al. 2007).

Since the TTX-I sodium current is not activated at subthreshold potentials and possesses slow kinetics of inactivation, this current may be especially important during the generation of repetitive firing. Persistent sodium currents may contribute to rhythmicity (Elson & Selverston, 1997; Zhong et al. 2007), repetitive firing (Harvey et al. 2006) and to plateau potentials in spinal and hypoglossal motor neurones (Powers & Binder, 2003; Heckman et al. 2008). We suggest that in presubicular pyramidal neurones the TTX-I sodium current may underlie the limited firing frequency adaptation of these cells (Fig. 2) and could support the role of the presubiculum in maintaining coding of a static head position.

Author contributions

D.F. and R.M. conceived the study and designed the experiments. D.F., C.D. and E.E. performed the experiments. J.W. provided the genetically modified mice. D.F., C.D., E.E., J.W. and R.M. participated in the analysis and the interpretation of the data. D.F. and R.M. wrote the manuscript and C.D., E.E. and J.W. revised and improved it. All the authors have approved the final version of the manuscript.