INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The vomeronasal organ (VNO) is a chemosensory structure involved in the detection of pheromones in many mammals (Døving and Trotier 1998). VNO sensory epithelium contains specialized neurons that are thought to transduce the chemical information related to pheromones into action potentials to the brain. Vomeronasal neurons are bipolar cells with an apical dendrite that reaches the epithelium surface and an axon projecting to the accessory olfactory bulb. A wealth of information is now available on the molecular and functional properties of these neurons (reviewed in: Biasi et al. 2001; Keverne 1999; Tirindelli et al. 1998). VNO epithelium also contains supporting, glial cells (Carmanchahi et al. 1999; Garrosa and Coca 1991; Höfer et al. 2000; Naguro and Breipohl 1982; Vaccarezza et al. 1981). Like chemosensory neurons, these cells are bipolar cells with an apical process reaching the epithelial surface, where it branches in several tall microvilli, and a basal process reaching the basal lamina. Unlike neurons, however, data on the functional properties of supporting cells in the VNO are not currently available.

In the olfactory epithelium, supporting cells possess a conspicuous resting K⁺ conductance that likely regulates the extracellular K⁺ (Masukawa et al. 1985; Trotier 1998; Trotier and MacLeod 1986). It has been suggested that this control might be important in setting the olfactory neurons at their maximum sensitivity for odorant detection (Trotier and MacLeod 1986). In other sensory organs, glial cells show quite complex membrane properties. In the retina, for instance, Müller cells are astrocyte-like cells expressing voltage-gated ion channels, neurotransmitter receptors and various uptake carrier systems (reviewed in: Newman and Reichenbach 1996). These properties enable Müller cells to control the activity of retinal neurons by regulating the extracellular concentration of neuroactive substances such as K⁺, GABA and glutamate. In addition, it has been proposed that voltage-gated Na⁺ channels in these cells could be activated by the functioning of adjacent neurons: this way, glial cells could sense the activity of neighboring neurons (Chao et al. 1994). In short, supporting cells seem to be key elements for controlling the transduction and signaling in excitable tissues.

We have addressed the issue of the functional properties of supporting cells in the VNO by studying with the patch-clamp technique their membrane conductances. To this aim, we analyzed the ion currents in voltage-clamp conditions. We found that vomeronasal supporting cells had unique electrophysiological properties, including a rather unusual (for glial cells) ratio for resting ion permeabilities (P_K:P_Na:P_Cl = 1:0.23:1.4). In addition, VNO supporting cells possessed voltage-gated K⁺ and Na⁺ currents with biophysical and pharmacological properties that differed from those of adjacent vomeronasal neurons. Our findings indicate that VNO supporting cells display complex membrane properties and raise the possibility that they might participate in the signal transduction and processing in the VNO.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Dissociation of vomeronasal supporting cells

Adult male C57BL/6J and CD-1 mice were used in this study. Vomeronasal neurons and supporting cells were isolated with a standard enzymatic-mechanical procedure (e.g., Liman and Corey 1996; Maue and Dionne 1987). Briefly, mice were deeply anesthetized by CO₂, followed by dislocation of cervical vertebrae. The VNO was removed within its bony encasing and then carefully dissected free of the bone. The sensory epithelium, which is situated at the medial part of the organ, was separated from the lateral nonsensory epithelium, rinsed in divalent-free Tyrode solution (in mM: 140 NaCl, 5 KCl, 10 HEPES, 10 glucose, 10 Na pyruvate, and 2 EGTA, pH 7.4 with NaOH), and cut into several small pieces. Collagenase (1 mg/ml of type A; Roche, Mannheim, Germany) and trypsin (1 mg/ml; Sigma, St. Louis, MO) were added, and the tissue was incubated at room temperature (22-25°C) for 60-70 min with agitation (for details, see Maue and Dionne 1987). After centrifugation, the surnatant was replaced with regular Tyrode solution (in mM: 140 NaCl, 5 KCl, 2 CaCl₂, 1 MgCl₂, 10 HEPES, 10 glucose, and 10 Na pyruvate, pH 7.4 with NaOH) supplemented with DNase (0.1 mg/ml; type I, Boehringer Mannheim, Germany) and incubated for 15 min with agitation. Finally, small tissue samples were gently triturated with a fire-polished pipette (tip diameter about 100 µm) and immediately plated on the bottom of a chamber that consisted of a standard glass slide onto which a silicon ring 1-2 mm thick and 15 mm ID was pressed. The glass slide was precoated with Cell-Tak (approximately 3 µg/cm²; BD Biosciences, Bedford, MA) to improve adherence of isolated VNO cells to the bottom of the chamber. As a control, we checked that enzymatic treatment with collagenase and trypsin did not affect the membrane properties of VNO supporting cells. To this aim, minced tissue was incubated with agitation in divalent-free Tyrode for 70 min at room temperature and gently triturated with a fire-polished pipette. Although the cell yield was low, this procedure allowed us to established that membrane properties of VNO supporting cells were not affected by enzymatic treatment. VNO neurons were also prepared as for supporting cells and used for comparing membrane properties.

The recording chamber was placed on the stage of an inverted Olympus microscope (model IX70, Olympus, Tokyo, Japan), and isolated VNO cells were viewed with Nomarski optics. During the experiments, VNO cells were continuously perfused with Tyrode by means of a gravity-driven system. Drugs were dissolved in modified Tyrode solution to maintain osmolarity. All chemicals were from Sigma, except tetrodotoxin (Alomone Laboratories, Jerusalem, Israel).

Whole cell recording

Membrane currents of single VNO cells were studied at room temperature by whole cell patch-clamp (Hamill et al. 1981), using an Axopatch 1D amplifier (Axon Instruments, Union City, CA). Signals were recorded and analyzed using a Pentium computer equipped with Digidata 1320 data acquisition system and pClamp8 software (Axon Instruments). pClamp8 was used to generate voltage-clamp commands and to record the resulting data. Signals were prefiltered at 5 kHz and digitized at 50- or 100-µs intervals.

Patch pipettes were made from borosilicate glass capillaries (Garner Glass, Claremont, CA) on a two-stage vertical puller (model PB-7, Narishige, Tokyo, Japan). The standard pipette solution contained (in mM) 120 KCl, 1 CaCl₂, 2 MgCl₂, 10 HEPES, 11 EGTA, 2 ATP, and 0.4 GTP, pH 7.3 with KOH. In some experiments, KCl was replaced by an equal concentration of CsCl, K gluconate, or Cs gluconate. Pipette resistances typically were 5-8 M when filled with intracellular solution. The access resistance of the patch pipette tip was estimated by dividing the amplitude of the voltage steps by the peak of the capacitive transients (from which stray capacitance had been subtracted). Values typically ranged from about 10 to 15 M. Leakage and capacitive currents were not subtracted from currents under voltage clamp, and all voltages have been corrected for liquid junction potentials (Neher 1992).

Input resistance of vomeronasal cells was measured as the slope of the linear current-voltage (I-V) relationship around 80 mV. Cell membrane capacitance was measured by integrating the capacitative current transient during application of a 10-mV voltage step from a holding potential of about 80 mV (Bigiani and Roper 1993).

Analysis of electrophysiological data

Results are presented as means ± SE. Data were analyzed using a Student's t-test. Significance level was taken as P < 0.05.

The Goldman-Hodgkin-Katz equation (GHK) (Hille 2001) was used to evaluate the relative membrane permeabilities to K⁺, Na⁺, and Cl. Values of zero-current potential (V₀) were measured at different concentration of the relevant ion and corrected for the shunt to ground by the seal resistance (Bigiani et al. 1996; Lynch and Barry 1991). This allowed us to estimate the cell's resting potential (V_r) at different concentrations of the relevant ion.

Correction of V₀ for the shunt to ground by seal resistance was obtained by applying the following equation (for details, see Bigiani et al. 1996; Lynch and Barry 1991)

(1)

where G_seal is the seal conductance evaluated from the seal resistance, G_in is the input conductance evaluated from the input resistance, G_m is the membrane conductance evaluated from G_in and G_seal, and E_seal is the potential across the pipette-membrane seal resistance, which is assumed to be nonselective and to represent only the liquid junction potential between the pipette and the bath solution.

Concentration-inhibition curves for the effect of the ion channel blockers, tetraethylammonium (TEA) or tetrodotoxin (TTX), on voltage-gated ion currents were obtained by adding increasing concentrations of the blocker into the bath solution and by measuring the corresponding change in ion current magnitude. The data were fitted to the logistic equation

(2)

where %I is the percent fraction of the ion current blocked by TEA or TTX, C is the blocker concentration, IC₅₀ is the blocker concentration that produces 50% inhibition of the voltage-gated current, and n is the Hill coefficient.

Steady-state inactivation curve for sodium currents were obtained by fitting the data with a Boltzmann equation

(3)

where I/I_max is current elicited during a test pulse and normalized to the maximal current, V is the voltage at which the membrane was held for 300 ms before the test pulse, V_0.5 is the membrane potential at which the current is 50% inactivated, and k the slope.

Immunohistochemical procedures

For immunohistochemistry, glass attached VNO cells were gently perfused with 4% paraformaldehyde, washed in phosphate-buffered saline solution (PBS), and air dried. Cells were blocked in 1% albumin, 0.3% Triton X-100 in PBS for 20 min, and incubated with anti-G8 antibody (1:400) (Tirindelli and Ryba 1996) or olfactory marker protein (OMP) antiserum (1.10000, kindly provided by Dr. F. Margolis, University of Maryland, Baltimore) overnight at 4°C. Specific immunoreactivity was detected by the biotin-avidin-horseradish peroxidase-diaminobenzidine method (ABC kit) as recommended by the supplier (Vector, Burlingame, CA).

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Identification of supporting cells and neurons isolated from the VNO

The VNO sensory epithelium contains three types of cells: sensory neurons, supporting cells, and basal cells (Carmanchahi et al. 1999; Garrosa and Coca 1991; Vaccarezza et al. 1981). Both sensory neurons and supporting cells possess cytoplasmatic processes, whereas basal cells are ovoid elements. Sensory neurons have a round cell body with an apical dendrite reaching the epithelium surface and an axon projecting to the accessory olfactory bulb. Supporting cells have oblong cell bodies with narrow prolongation that lay between the dendrites of the sensory neurons (Carmanchahi et al. 1999; Höfer et al. 2000; Vaccarezza et al. 1981). We used these morphological features to identify VNO neurons and supporting cells after dissociation. Isolated sensory cells were round, ovoid in shape and possessed a clear dendritic process that ended in a knoblike structure from which microvilli protruded (Fig. 1A). Consistently with previous reports (e.g., Liman and Corey 1996; Moss et al. 1998), the axon was usually not present in isolated neurons, likely lost during the dissociation procedure. On the contrary, isolated supporting cells were rectangularly shaped and possessed narrow distal prolongations (Fig. 1B). Supporting cells were further distinguished from sensory neurons by the lack of immunoreactivity for G protein -subunit (G8), a specific marker for vomeronasal neurons (Tirindelli and Ryba 1996) (Fig. 1). In addition, only sensory neurons were immunoreactive for OMP (data not shown), consistently with previous results (Liman and Corey 1996).

View larger version (99K): [in this window] [in a new window]   Fig. 1. Differential interference contrast photomicrographs of a sensory neuron (A) and a supporting cell (B) isolated from the mouse vomeronasal organ (VNO). Cells were immunostained for G8, a specific marker for vomeronasal neurons. Strong staining can be detected in the neuron up to the dendritic knob, whereas the supporting cell was not stained (the darkness in the soma of the supporting cell is shadow of the interference contrast optics). b, basal process; d, dendrite; k, dendritic knob. Scale bar, 10 µm.

In this study, we present data obtained from 159 unambiguously identified supporting cells isolated from the mouse vomeronasal organ. Just for the purpose of comparison, we also report data obtained from 87 sensory neurons. Further information on membrane properties of these sensory neurons isolated from the mouse VNO can be found in a published report (Liman and Corey 1996).

Passive membrane properties

The electrical properties of VNO supporting cells were examined by whole cell patch-clamp recordings. We did not observe any significant differences between the membrane properties of supporting cells dissociated using enzymes, or no enzyme. Therefore data were pooled together.

In a first series of experiments, we determined the passive membrane properties of supporting cells. In the whole cell patch-clamp configuration, a parameter often used to estimate the cell's resting potential is the zero-current potential (V₀). V₀ is the voltage at which no current is required in voltage clamp. With KCl pipette solution, V₀ was 29 ± 1 mV (n = 39), which was significantly more depolarized than V₀ of sensory neurons measured under similar conditions (53 ± 3 mV; n = 19). The input resistance (R_in) of supporting cells was 1.7 ± 0.2 G (n = 14), whereas in sensory neurons, R_in was almost four times larger (6.4 ± 1.3 G; n = 37). Since seal resistance did not change significantly when we patched onto supporting cells or on sensory neurons, findings on R_in suggested that supporting cells were endowed with a substantial "resting" conductance. Finally, in supporting cells the membrane capacitance (C_m: 9.2 ± 0.6 pF; n = 22) was significantly larger than that of sensory neurons (5.1 ± 0.4 pF; n = 39). Normalization of R_in to C_m revealed that the low membrane resistance of supporting cells was due only in part to their larger membrane surface area (R_in/C_m: 0.19 G/pF in supporting cells; 1.26 G/pG in sensory neurons).

Resting membrane permeabilities

Glial cells of several tissues are characterized by a large resting potassium conductance (reviewed in Syková et al. 1998). Substitution of K⁺ with Cs⁺ (a potassium channel blocker; Rudy 1988) in the pipette solution markedly affected V₀ in supporting cells (6 ± 3 mV; n = 8). However, a depolarized V₀ in control conditions (about 29 mV on average) suggested that the membrane of supporting cells was permeable also to other ions. For example, substitution of extracellular Na⁺ ions with N-methyl-D-glucamine (NMDG) caused a pronounced hyperpolarization (V₀ = 45 mV ±1.5; n = 11), suggesting the presence of Na⁺ pathways in the cell membrane. Interestingly, 30-100 µM amiloride, which blocks the passive, epithelial Na⁺ channels in many cell types (Garty and Palmer 1997), had no effect on vomeronasal supporting cells (n = 7).

To estimate the relative membrane permeabilities to the main ions occurring in our experimental solutions, namely K⁺, Na⁺, and Cl, V₀ was measured at different extracellular concentrations of a given ion (K⁺ or Na⁺) and in bi-ionic conditions (i.e., only 2 main permeant ions present in the experimental solutions: either K⁺ and Cl, or Na⁺ and Cl). The osmolarity and the ionic strength of the extracellular solutions were maintained by substituting sodium or potassium with NMDG⁺, a nonpermeable ions. Although V₀ is assumed to be an estimation of the cell's resting potential, it requires correction for the shunt to ground by the seal resistance and also for liquid junction potential changes during solution exchange (see Eq. 1). This correction allowed us to obtain a better estimation of the resting potential (V_r) in supporting cells in different ionic conditions. First we evaluated the relative permeabilities to K⁺ and to Cl in the absence of Na⁺. As revealed by the data shown in Fig. 2A, in resting conditions, the membrane of supporting cells was highly permeable to chloride ions, in addition to potassium ions. Next, we evaluated the relative permeabilities to Cl and to Na⁺, in the absence of K⁺. The data in Fig. 2B clearly show that in resting conditions, glial membranes were six times more permeable to chloride ions than to sodium ions. Moreover, these data show that vomeronasal supporting cells had a significant resting Na⁺ permeability, consistently with the effect of NMDG on V₀.

View larger version (13K): [in this window] [in a new window]   Fig. 2. Effect of variations in extracellular K+ or Na+ concentration ([K+]o and [Na+]o, respectively) on the value of resting membrane potential (Vr) in vomeronasal supporting cells. Resting membrane potential was calculated by correcting V0 for the shunt to ground by seal resistance, and for liquid junction potentials (see METHODS). Points represents means ± SE (bars) from 7 to 11 measurements. A: effect of variation in [K+]o on Vr in the presence of only K+ and Cl as main ions. Intracellular solution: K gluconate; bath solution: Na+-free Tyrode [Na+ replaced by N-methyl-d-glucamine (NMDG)]. Data points were fitted using the Goldman-Hodgkin-Katz equation in the hypothesis that the membrane was permeable only to K+ and to Cl. The best fit of the data points was obtained for a permeability ratio PCl/PK of 1.4. B: effect of variation in [Na+]o on Vr in the presence of only Na+ and Cl as main ions. Intracellular solution: Cs gluconate; bath solution: K+-free Tyrode. Data points were fitted using the Goldman-Hodgkin-Katz equation in the hypothesis that the membrane was permeable only to Na+ and to Cl. The best fit of the data points was obtained for a permeability ratio PCl/PNa of 6.1.

Both ion channels and electrogenic pumps can affect V₀ measurements used to evaluate ion permeabilities. In glial cells of other tissues, the occurrence of electrogenic Na⁺,K⁺ATPase has been documented (e.g., Walz 1989). However, with our ionic conditions it was unlikely that Na⁺,K⁺-ATPase could play any role in setting up V₀ in vomeronasal supporting cells. Intracellular Na⁺ concentration in the millimolar range is required for pump activity (e.g., Glitsch 2001). On the contrary, our standard pipette (intracellular) solution was nominally Na⁺-free (see METHODS). Indeed, application of ouabain (a known Na⁺,K⁺-ATPase blocker at concentrations in the micromolar range) did not significantly affect V₀ (control: 30 ± 6 mV; with 50 µM ouabain: 31 ± 7 mV; n = 4). However, it is possible that Na⁺ influx through opened channels could cause a local increase in the intracellular Na⁺ concentration near the pump (the so-called "fuzzy space": Lederer et al. 1990). This condition could be enhanced when using a high-Na⁺ extracellular solution, while recording with a pipette solution in which K⁺, a competitive inhibitor of Na⁺ at intracellular Na⁺-binding sites of the pump (e.g., Glitsch 2001), is replaced by Cs⁺ (Fig. 2B, rightmost point). However, even in these conditions, application of 50 µM ouabain had no effect on V₀ (control: 15 ± 5 mV; with ouabain: 15 ± 4 mV; n = 4).

In conclusion, the resting membrane of VNO supporting cells was permeable to the main ions occurring in the experimental solutions, namely K⁺, Na⁺, and Cl, with a relative permeability ratio of P_K:P_Na:P_Cl = 1:0.23:1.4. Since the contribution of the Na⁺,K⁺-ATPase to resting potential was negligible, ion permeabilities were likely due to the presence of opened (passive) ion channels.

Voltage-gated K⁺ currents

Glial cells in several tissues display complex membrane properties due to voltage-gated ion channels, including K⁺ channels and Na⁺ channels (reviewed in Sontheimer 1994; Verkhratsky and Steinhäuser 2000). The presence of voltage-gated ion channels in VNO supporting cells was investigated by recording membrane currents elicited by voltage steps from a holding potential of about 80 mV. With normal Tyrode solution in the bath and the pipette solution containing 120 KCl, depolarizing voltage pulses elicited the current time courses shown in Fig. 3A. After an early capacitive transient, current exhibited a pronounced outward trace. These outward currents were observed in all supporting cells we tested under the above ionic conditions, although their magnitude varied from cell to cell. Outward currents were recorded also from sensory neurons. However, in these cells they were preceded by a fast, transient inward current (Fig. 3B) mediated by sodium ions (see also Liman and Corey 1996).

View larger version (30K): [in this window] [in a new window]   Fig. 3. Membrane currents in cells isolated from the vomeronasal sensory epithelium. Cells were held at 84 mV and stepped in 10-mV increments from 74 to +106 mV. Pipette solution: KCl. Bath solution: Tyrode's. A: supporting cells were characterized by outward currents only. Note the large noise that was typically associated with outward currents in these cells. As shown by the current-voltage (I-V) relationship on the right, outward currents typically activated at approximately 30 mV. B: sensory neurons were characterized by both inward currents (shaded dot) carried by Na+, and outward currents (empty dot). As revealed by the corresponding I-V plots (right), inward currents activated at about 50 mV, and outward current at about 30 mV. In all I-V plots, outward currents were measured at the end of 30-ms voltage pulses. Each point represents the means ± SE (bars) from 27 supporting cells, and 20 sensory neurons. Im, membrane current; Vm, membrane potential.

The outward current in both supporting cells and sensory neurons was carried by potassium ions, as indicated by their sensitivity to TEA (Fig. 4, A and B) and by their complete block when CsCl was used instead of KCl in the pipette solution (data not shown). As indicated by the concentration-inhibition curves shown in Fig. 4C, the sensitivity of potassium current (I_K) to TEA was larger in supporting cells than in neurons. Also 4-aminopyridine (4-AP, another K channel blocker, especially for inactivating channels) (Rudy 1988) affected I_K. Interestingly, I_K in supporting cells was less sensitive to 4-AP than to TEA, whereas neuronal I_K was sensitive to both channel blockers (Fig. 4D). Consistent with the weak effect of 4-AP on i_K in supporting cells was the observation that I_K amplitude (at a reference potential of +46 mV) was not significantly affected by the holding potential (V_h): 777 ± 37 pA with V_h = 84 mV; 809 ± 29 pA with V_h = 64 mV; 786 ± 44 pA with V_h = 44 mV; 760 ± 37 pA with V_h = 24 mV (n = 3). Thus I_K in supporting cells did not show any significant inactivation. As reported by Liman and Corey (1996), I_K in sensory neurons displays slow inactivation.

View larger version (39K): [in this window] [in a new window]   Fig. 4. Outward K+ currents in cells isolated from the VNO neuroepithelium. Membrane currents were elicited by a series of depolarizing pulses between 74 and +106 mV, in 10-mV increments, from a holding potential of 84 mV. Pipette solution: KCl. A: in supporting cells, outward currents recorded in regular Tyrode (control) were totally abolished by 10 mM TEA. B: in sensory neurons, on the contrary, 10 mM TEA blocked only partially the outward currents. C: dose-response curves for the effect of TEA on supporting cells () and sensory neurons (). Percentage inhibition of the total K+ current was evaluated for the current elicited by a depolarizing voltage step to +96 mV from a holding potential of 84 mV. Points represent mean ± SE (bars) of 4-10 measurements. Sigmoidal curves are the best fit of the data obtained with the logistic equation (see METHODS) with an IC50 of 0.35 mM and a Hill coefficient of 1 in supporting cells and with an IC50 of 3.39 mM and a Hill coefficient of 0.61 in sensory neurons. D: comparison of the effect of 2 mM TEA and 2 mM 4-AP on the outward potassium current in supporting cells and sensory neurons. Percentage inhibition of the total K+ current was evaluated for the current elicited by a depolarizing voltage step to +96 mV from a holding potential of 84 mV. The histograms represent mean values ± SE (bars) from 6-8 measurements. Note that in supporting cells, K+ currents were weakly sensitive to 4-AP (<10% reduction), whereas they were strongly affected by TEA (90% reduction).

Outward rectification in potassium currents could be due to the activation of voltage-gated channels or to leak channels that rectify because of nonsymmetric potassium concentration gradient (Goldstein et al. 2001). Thus we perfused the cells with a high-potassium Tyrode solution in which Na⁺ was substituted by K⁺ (symmetric K⁺). In these conditions, currents through leak channels (like the KCNK channels of the 2-P-domain potassium channel family; Goldstein et al. 2001) should produce a linear current-voltage (I-V) relationship. However, we found that I-V plots in supporting cells remained nonlinear in symmetric K⁺ (Fig. 5). In addition, the I-V plot exhibited an inward deflection for voltages between 40 and 0 mV (arrow in Fig. 5, right), reflecting the influx of K⁺ ions through voltage-gated channels, which were closed for more negative potential. Thus outward potassium currents were likely mediated by the activation of voltage-gated K⁺ channels in supporting cells.

View larger version (24K): [in this window] [in a new window]   Fig. 5. Outward K+ currents in supporting cells are mediated by voltage-gated channels. Current-voltage (I-V) relationship for membrane currents were evaluated under conditions of high internal and low external K+ (control; pipette solution: 155 mM K+; bath solution: 5 mM K+), and under symmetric K+ (pipette solution: 155 mM K+; bath solution: 155 mM K+). A linear I-V plot would be expected under symmetric K+ in the case of leak channels. Note that under symmetric K+, an inward current (arrow) activates at a membrane voltage of about 40 mV (deflection below the line of leakage current). This is consistent with K+ channels being opened by membrane depolarization and with K+ ions flowing into the cell (EK = 0 mV in symmetric conditions). Each point represents the means ± SE (bars) from 10 supporting cells. Straight lines represent the extrapolated leakage (passive) current obtained by fitting the current values at membrane voltages more negative than 45 mV. Im, membrane current; Vm, membrane voltage.

When external Na⁺ was substituted by K⁺ to obtain symmetric conditions for K⁺, the I-V plot was shifted rightward (Fig. 5, right). In these conditions, K⁺ is the main cation both in the extracellular and intracellular solution. Thus when the voltage-gated channels open up (at about 40 and 30 mV) the potassium current predominates and sets the behavior for the I-V plot. In particular, the membrane current reverses at 0 mV, which is the equilibrium potential of K⁺ (E_K) in symmetric conditions. In addition, the I-V plot was shifted downward for negative membrane potentials (Fig. 5, right), consistently with the presence of resting potassium channels. On the other hand, the increased in the current amplitude for very positive voltages was unusual, since an increase of [K⁺]_o lowers the driving force for outward K⁺ flux and would be expected to decrease rather than increase outward currents. Such a dependence on [K⁺]_o have been reported for outward rectifier K⁺ channels in other tissues (Fink et al. 1996; Pardo et al. 1992; Scamps and Carmeliet 1989), and could be due an action of extracellular K⁺ on the channel protein.

Ca²⁺-dependent K⁺ currents mediated by maxi-K channels (BK channels) have been described in glial cells of other tissues (reviewed in Verkhratsky and Steinhäuser 2000). Application of 20 nM charybdotoxin (CTX), a blocker of BK channels (Knaus et al. 1994), did not affect the outward potassium currents in supporting cells (n = 6; data not shown). Consistent with the absence of BK channels was also the finding that 1 mM Cd²⁺, an inorganic blocker of Ca²⁺ channels (Bean 1992), was unable to reduce outward potassium currents (n = 6; data not shown). As demonstrated by an early study (Liman and Corey 1996), also vomeronasal neurons do not possess Ca²⁺-activated K⁺ currents.

Voltage-gated Na⁺ currents

Voltage-gated, TTX-sensitive sodium currents (I_Na) were elicited regularly in vomeronasal neurons by depolarizing the membrane from a holding potential of about 80 mV (Fig. 3B). In these conditions, on the contrary, supporting cells displayed only outward potassium currents (Fig. 3A). However, when the holding potential was set to more negative values, early, transient inward currents appeared also in the current records from supporting cells (Fig. 6A). Membrane hyperpolarization affected markedly the amplitude of inward currents (Fig. 6B) but not that one of outward potassium currents (Fig. 6C). The transient inward current disappeared when NMDG replaced Na⁺ in the bath (Fig. 7) and was blocked reversibly by 0.5 µM TTX (data not shown): in short, it was a Na⁺ current (I_Na). Activation threshold for I_Na in supporting cells was between 60 and 50 mV, which is similar to the value for neuronal I_Na (Figs. 3B and 6B).

View larger version (33K): [in this window] [in a new window]   Fig. 6. Voltage-gated inward currents in supporting cells isolated from the mouse VNO. A: membrane currents (Im) were elicited by a series of depolarizing voltage pulses in 10-mV increments from different holding potential (Vh). When holding potential was set to value more negative than 84 mV, inward currents appeared in the records (arrows). Note that outward current (upward deflections in the records) were always present. B: current-voltage (I-V) relationships for the inward currents shown in A. These currents activate at about 50 to 60 mV and peaked at about 40 mV. Note that the current amplitude increased by hyperpolarizing the holding potential, suggesting that a large fraction of the channels was removed from inactivation. C: unlike inward currents, outward potassium currents were scarcely affected by hyperpolarized holding potentials.

View larger version (47K): [in this window] [in a new window]   Fig. 7. Effect of Na+-free solution on the inward current in supporting cells isolated from the mouse VNO. Na+ was replaced by NMDG in the Tyrode's solution. Membrane current were elicited by a series of depolarizing pulses between 94 and +106 mV, in 10-mV increments, from a holding potential of 104 mV. Inward current recorded in regular Tyrode (control) were totally abolished by Na+-free solution (middle). The effect was reversible, as indicated by the recovery of the currents during washout (wash). Note that the outward potassium currents was unaffected by the application of Na+-free solution.

To compare the amplitude of I_Na in supporting cells with that one of I_Na in sensory neurons, I_Na was elicited by a two-pulse stimulation as follows: from a holding potential of 84 mV, cell membrane was hyperpolarized to 124 mV for 300 ms and then depolarized to 34 mV to activate I_Na. In these conditions, the maximal amplitude of I_Na was significantly lower in supporting cells (292 ± 114 pA; n = 20) than in sensory neurons ( 868 ± 158 pA; n = 18).

The absence of I_Na in supporting cells held at 80 mV indicated that the inactivation properties of these currents differed from those of the channels expressed by sensory neurons. To study the voltage dependence of the steady-state inactivation, we used a typical two-pulse voltage protocol (prepulse and test pulse) that allowed the evaluation of the noninactivated fraction of the sodium current as a function of a prepulse membrane potential. Steady-state inactivation curves show that in supporting cells inactivation was shifted by about 24 mV in the hyperpolarizing direction (Fig. 8, bottom). Also in astrocytes, V_0.5 is about 25 mV more negative than in neurons (Barres et al. 1989).

View larger version (24K): [in this window] [in a new window]   Fig. 8. Voltage dependence of the steady-state inactivation of voltage-gated Na+ currents in supporting cells and sensory neurons isolated from the mouse VNO. A standard 2-pulse voltage protocol (top) was used for this analysis. Current magnitudes of the test pulse (34 mV) were normalized to its maximal value and plotted against the prepulse potential (bottom). Each point represents the mean ± SE of 15-19 measurements. Data were fitted to a Boltzmann equation. For supporting cells, the half-maximal voltage (V0.5) was 101 mV and the slope (k) was 8.9 mV. For sensory neurons, V0.5 was 77 mV and k was 10.6 mV. Pipette solution: CsCl. Bath solution: standard Tyrode's. Duration of prepulse potentials: 300 ms.

Finally, the sensitivity of I_Na to TTX was less pronounced in VNO supporting cells than in neurons (Fig. 9, bottom). For VNO neuronal Na⁺ channels, a high sensitivity to TTX has been reported also by Liman and Corey (1996).

View larger version (18K): [in this window] [in a new window]   Fig. 9. TTX sensitivity of the voltage-gated Na+ currents recorded in supporting cells and sensory neurons isolated from the mouse VNO. Na+ current was elicited by a test pulse to 34 mV after hyperpolarizing the membrane to 124 mV for 300 ms (top). Percentage inhibition of the maximal Na+ current was evaluated for a series of TTX concentrations and then plotted to obtain the dose-response curves shown (bottom). Points represent means ± SE (bars) of 5-11 measurements. Sigmoidal curves are the best fit of the data obtained with the logistic equation (see METHODS) with an IC50 of 0.48 µM and a Hill coefficient of 0.91 in supporting cells () and with an IC50 of 0.015 µM and a Hill coefficient of 0.54 in sensory neurons ().

As a whole, these data indicated that supporting cells of the VNO expressed voltage-gated Na⁺ channels with specific biophysical and pharmacological properties.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

One of the current challenging issues in olfaction research is the evaluation of the operation of vomeronasal neuroepithelium as detector and transducer of pheromonal signals. The goal of our study was to contribute to the understanding of the functional properties of such a chemosensory structure. We have examined the electrophysiological properties of supporting cells isolated from the mouse VNO. The main finding is that these cells possess quite unique membrane properties that, to our knowledge, have not been described for other types of glial cells.

Resting membrane permeabilities in vomeronasal supporting cells

One of the characteristic features of glial cells in several tissues is their membrane selectivity for K⁺, and a low (in some cases, negligible) permeability to Na⁺ and Cl (Bigiani 2001; Kettenmann et al. 1983; Newman 1985, 1989; Ransom and Sontheimer 1995; Sugihara and Furukawa 1996; Walz and Schule 1982; Walz et al. 1984). Results from the ion substitution experiments described here demonstrate that vomeronasal supporting cells have membranes that are permeable not only to K⁺, but also to Na⁺ and Cl. An estimate based on fitting the data with the GHK equation suggests that the ratio of K⁺ to Na⁺ permeabilities (P_K/P_Na) is about 4, and the ratio of K⁺ to Cl permeabilities (P_K/P_Cl) is about 0.6. These permeability ratios seem quite peculiar to VNO supporting cells (Table 1). Thus there is something fundamentally different about the membrane construction of supporting cells in the VNO.

Role of glial resting ion permeabilities in VNO physiology

The VNO is a chemosensory structure located at the base of the nasal cavity and shaped as a blind-end sac (Døving and Trotier 1998; Halpern 1987). VNO receives stimuli from the nasal cavity through an active vascular pumping mechanism controlled by the autonomic nervous system (Meredith 1994; Meredith et al. 1980). This way, mucus can flow in and out from the VNO (Meredith 1998). Volatile pheromones (Novotny et al. 1985) can reach the VNO after solubilization in the mucous as usual odorants. On the contrary, nonvolatile pheromone substances, such as the proteins of the major urinary protein complex (MUP) (Hurst et al. 2001; Mucignat-Caretta et al. 1995) can reach the vomeronasal neuroepithelium by direct influx of urine into the VNO lumen with licking (Wysocki et al. 1980). Given the relatively high concentration of K⁺, Na⁺, and Cl in rodent urine (150 mM K⁺ and 200 mM Na⁺, with Cl as main counter-anion) (Wüthrich 2000), it is reasonable to conceive that these ions might interfere with the pheromone detection by sensory neurons by altering their membrane excitability at the apical ends. It is therefore tempting to speculate that high resting membrane permeabilities to K⁺, Na⁺, and Cl in supporting cells may play a role in mopping-up the excess of inorganic ions that enter in the VNO lumen with certain pheromone-containing fluids, like urine. Supporting cells are bipolar elements with an apical process reaching the epithelium surface, and a basal process reaching the basal lamina that separates the neuroepithelium from the blood vessels (Carmanchahi et al. 1999; Garrosa and Coca 1991; Höfer et al. 2000; Naguro and Breipohl 1982; Vaccarezza et al. 1981). Apical processes of supporting cells bear tall and wide microvilli that protrude inside the VNO lumen and seem to cover up the short microvilli of sensory neurons (Höfer et al. 2000; Naguro and Breipohl 1982; Vaccarezza et al. 1981). Compared with neuronal microvilli, those of supporting cells reach further into the lumen of the VNO. Thus it is tempting to speculate that at this level excess of ions could be absorbed and then transferred into the blood vessels via glial basal processes coursing through the neuroepithelium. By modifying the ionic strength of the pheromone-containing medium, supporting cells may also influence directly the biophysical properties of pheromonal proteins, such as MUPs. Thereby, VNO supporting cells can affect pheromone detection by neurons. Further studies on the membrane localization of ion permeabilities in glial membranes (apical, basal, lateral) are required to fully elucidate their role in the management of inorganic ions in the VNO.

Voltage-gated K⁺ and Na⁺ conductances in vomeronasal supporting cells

Like glial cells in other tissues (reviewed in Sontheimer 1994; Verkhratsky and Steinhäuser 2000), supporting cells isolated from the mouse VNO possessed both voltage-gated K⁺ and Na⁺ channels. Interestingly, all supporting cells we patched (n = 159) displayed both K⁺ and Na⁺ currents, although their magnitude varied among cells. In addition, we could not reveal any significant variability in the properties of these currents among supporting cells, which in this respect made up a homogeneous cell population. On the contrary, variability in the expression of ion currents has been shown for other types of glial cells. For example, in patch-clamp recordings from hippocampal slices, only 10% of astrocytes expressed Na⁺ currents (Sontheimer and Waxman 1993).

Although voltage-gated K⁺ currents in vomeronasal supporting cells were similar to the delayed-rectifier type (K_DR) described in other glial cells (reviewed in Sontheimer 1994), they presented some unique features. In astrocytes and Schwann cells, K_DR channels are moderately sensitive to TEA, with blocking concentration ranging from 5 to 50 mM, and IC₅₀ ranging from about 6 to 11 mM (Sontheimer 1994). On the contrary, K_DR currents in vomeronasal supporting cells were highly sensitive to TEA, as indicated by an IC₅₀ of 0.35 mM. Moreover, K_DR channels in astrocytes are equally sensitive to both TEA and 4-AP (Sontheimer 1994), whereas in vomeronasal supporting cells they display a lower sensitivity to 4-AP than TEA.

Another unique feature of K_DR currents in vomeronasal supporting cells was their unusual noise (see Fig. 3A). In cone photoreceptors, large fluctuations in the current records have been attributed to the activation of Ca²⁺-dependent K⁺ currents by Ca²⁺ ions entering through voltage-gated channels (Barnes and Hille 1989). However, in vomeronasal supporting cells we found no evidence of such Ca²⁺-dependent currents. Outward currents through two-P-domain K⁺ channels may exhibit conspicuous noise (e.g., Lesage et al. 2000; Patel et al. 1998). However, our experiments with symmetric K⁺ indicated that outward currents in vomeronasal supporting cells were mediated by voltage-gated channels. Molecular studies, such as single-cell RT-PCR (e.g., Rabe et al. 1999) for identifying the potassium channel genes, will be helpful in the identification of the K⁺ channels expressed by vomeronasal supporting cells.

Unlike K⁺ channels, Na⁺ channels in vomeronasal supporting cells were electrophysiologically (steady-state inactivation properties) and pharmacologically (low sensitivity to TTX) similar to the glial type Na⁺ channels expressed in protoplasmic astrocytes (Sontheimer and Waxman 1992). They differed, however, from Schwann cell Na⁺ channels, which are highly sensitive to TTX (IC₅₀ = 2 nM) (Howe and Ritchie 1990).

Role of glial voltage-gated K⁺ and Na⁺ conductances in VNO physiology

As in the case of glial cells in other tissues, the functional importance of voltage-gated K⁺ and Na⁺ channels in vomeronasal supporting cells remains to be established. Spatial buffering, the uptake and redistribution of excess K⁺ from the extracellular space, may involved both resting K⁺ channels and voltage-activated K⁺ channels (reviewed in Sontheimer 1994). Like in astrocytes, Na⁺ currents in vomeronasal supporting cells were usually small in amplitude. As a consequence, current density for I_Na (max amplitude/C_m) was about 32 pA/pF, well below the mean value of 170 pA/pF for I_Na in sensory neurons, which are electrically excitable (Liman and Corey 1996). Furthermore, Na⁺ channels were more than 95% inactivated at a membrane potential of about 75 mV (Fig. 8). V_r of supporting cells could be at best 72 mV in the presence of a Na⁺-free extracellular solution and with a low Cl intracellular solution (see Fig. 2A). Thus as in other glial cells, the mismatch of resting potential and inactivation properties would prevent electrogenesis in vomeronasal supporting cells. It is possible that TTX-blockable Na⁺ channels in glial cells serve roles that do not require voltage-dependent activation. In astrocytes, these channels, even at rest, allow a small percentage of Na⁺ ions to leak into cells (Sontheimer et al. 1994). It has been proposed that glial Na⁺ channels may represent a return pathway for the function of the glial Na⁺/K⁺-ATPase by permitting Na⁺ ions to enter the cell to maintain intracellular Na⁺ at concentrations necessary for the activity of the pump (Sontheimer et al. 1994). Thus it is likely that also in VNO voltage-gated ion channels in supporting cells may play a key role in the control of chemical composition of intercellular fluid, which, in turn, can affect the excitability of sensory neurons.