Characterization of a nitrate-permeable channel able to mediate sustained anion efflux in hypocotyl cells from Arabidopsis thaliana

Authors


*For correspondence (fax +33 1 69 82 37 68; e-mail frachisse@isv.cnrs-gif.fr).

Summary

We have characterized a new anionic current in Arabidopsis hypocotyl cells. This current, activated by membrane depolarization, has slow activation and deactivation kinetics in the 10 sec range. It presents many distinct properties from the rapid-type anion current already described on the same membrane. The slow-type channel is highly permeable to nitrate with a PNO3/PCl close to 20, but totally impermeable to sulphate. Activation of the channel requires cytosolic ATP and the slow current is partially inhibited by staurosporin, suggesting that channel regulation involves protein phosphorylation. The slow anion channel displays a unique pharmacological profile different from that of the rapid channel: the slow channel is inhibited by DIDS (4,4′-diisothiocyanatostilbene-2,2′-disulfonic acid) with an IC50 of 26 μM. The slow and rapid anion channels are probably dedicated to specific functions: the first is able to mediate sustained anion efflux, while the second is a good candidate to be involved in fast electrical signalling.

Introduction

Plant anion channels are involved in important cellular functions, including nutrient transport, signal transduction and cell turgor regulation ( Barbier-Brygoo et al. 1999 ; Barbier-Brygoo et al. 2000 ; Tyerman 1992; Ward et al. 1995 ). The physiological role of anion channels has been particularly well studied in stomata where their activation is one of the limiting steps in the loss of guard cell turgor leading to stomatal closure ( Ward et al. 1995 ), and in algal cells where they contribute to membrane excitability through the generation of action potentials ( Barbier-Brygoo et al. 2000 ; Hope & Findlay 1964; Kourie 1994). With the exception of these two examples, the detailed characterization of anion channels is still far from complete in differentiated plant cells. In guard cells, at least two different types of anion channels are present at the plasma membrane ( Keller et al. 1989 ; Schroeder & Hagiwara 1989; Schroeder & Keller 1992): the slow anion channel (S-type) and the rapid anion channel (R-type). A combination of pharmacological and genetic approaches has provided strong evidence that S-type anion channels catalyse the sustained anion efflux leading to stomatal closure in response to abscisic acid (ABA) ( Pei et al. 1997 ; Schroeder et al. 1993 ; Schwartz et al. 1995 ).

Recently, we have characterized a voltage-dependent anion channel ( Thomine et al. 1995 ) at the plasma membrane of the hypocotyl epidermal cells of Arabidopsis thaliana. This channel, tightly controlled by the transmembrane voltage, is deactivated at resting membrane potentials and activated by depolarization and shows rapid kinetics in the millisecond range, similar to those of the R-type guard cell anion channel. The voltage regulation of this channel is controlled by cytoplasmic nucleotides and does not require nucleotide hydrolysis ( Thomine et al. 1997a ). This rapid channel is selective for nitrate and sulphate relative to chloride, and sulphate has been shown to exert a regulatory effect on the channel by preventing the run-down of the anion current ( Frachisse et al. 1999 ).

To explore a possible role of anion channels in the regulation of hypocotyl development, Thomine et al. (1997b) tested a variety of anion channel blockers on hypocotyl elongation. Some of the blockers were able to counteract the inhibitory effects of auxins on hypocotyl elongation, providing some evidence for the involvement of anion channels in auxin-regulated elongation processes. The pharmacological profiles of the rapid-type voltage-dependent anion channel from hypocotyl cells ( Thomine et al. 1997a ) and of the auxin-regulated hypocotyl elongation do not match, making participation of this channel in this response unlikely ( Thomine et al. 1997b ). A blue-light-stimulated anion channel has also been described in Arabidopsis hypocotyl cells ( Cho & Spalding 1996), but its high sensitivity to the anion channel inhibitor nitrophenyl-propylamino-benzoic-acid (NPPB) together with the fact that NPPB was unable to counteract auxin inhibition of hypocotyl growth argue against its involvement in hypocotyl elongation. Therefore, these data suggest that other types of anion channels are involved in the auxin-regulated hypocotyl elongation, and call for a more extensive description of the diversity of anion channels in hypocotyl cells.

Here we describe a new type of anion channel co-residing with the rapid voltage-dependent anion channel formerly described at the plasma membrane of hypocotyl epidermal cells ( Frachisse et al. 1999 ). In terms of kinetics, this new channel resembles the guard cell slow anion channel. Channel properties such as voltage dependence, selectivity, nucleotide regulation and pharmacology were investigated in order to compare this new channel with the rapid-type channel and to obtain indications as to the possible functions of this slow-type channel in hypocotyl cells.

Results

A channel with slow kinetics co-resides with the rapid anion channel at the plasma membrane of Arabidopsis hypocotyl cells

While we were investigating the ionic selectivity of the fast anion channel in Arabidopsis hypocotyl cells ( Frachisse et al. 1999 ), we noticed that a significant anionic current component remained after run-down of the R-type current when nitrate was present in the pipette. To characterize this new anionic current, we performed additionnal experiments using 50 m m NO3 as the cytosolic anion. After entering the whole-cell configuration, we imposed a transmembrane potential consisting of a fast ramp and a combination of two long pulses to monitor both the rapid current and the new current which appeared to follow slow kinetics. The ionic current ( Fig. 1) across the plasma membrane 10 min after the establishment of the whole-cell configuration is composed of a rapidly activating anion current (‘Rapid’) with a strong voltage dependence observed during the ramp, followed by a steady outward current during holding at 71 mV, and then a slow deactivating inward current (‘Slow’) is recorded when stepping the voltage to −189 mV. Super-imposition of the currents recorded successively at 10, 13 and 18 min shows the run-down of the rapid anion current while the slow current remains unchanged ( Fig. 1).

Figure 1.

Evolution over time of rapid and slow anion currents of the plasma membrane of Arabidopsis thaliana hypocotyl cells, after entering the whole-cell configuration.

A protocol composed of a 5 sec ramp from −189 to 71 mV followed by two step pulses of 36 sec at 71 mV and 40 sec at −189 mV is applied every 90 sec. The ionic current across the plasma membrane 10 min (trace 1) after the establishment of the whole-cell configuration is composed of a rapidly activating anion current (‘Rapid’) with a strong voltage-dependence observed during the ramp, followed by a stationary outward current during holding at 71 mV, and then a slow deactivating inward current (‘Slow’) is registered when stepping the voltage to −189 mV. Super-imposition of the currents recorded successively at 10 min (trace 1), 13 min (trace 2) and 18 min (trace 3) shows the run-down of the rapid anion current while the slow current stayed unchanged. The bottom part shows, on an expanded time scale, the rapid current recorded during the ramp. The pipette solution contains 50 m m KNO3, 2 m m MgCl2, 4.2 m m CaCl2, 5 m m EGTA, 5 m m MgATP, 5 m m LiGTP, 10 m m HEPES/Tris pH 7.2, while the bath includes 50 m m CaCl2, 5 m m MgCl2, 0.5 m m LaCl3, 10 m m MES/Tris, pH 5.6.

The voltage-dependent activation of the slow deactivating current was investigated in detail. To exclude any contamination of the slow current by the rapid current, we imposed the pulse protocol shown in Fig. 2 after complete run-down of the fast anion current. The membrane potential was held for 20 sec at a potential ranging from −109 to +71 mV and then stepped back for 30 sec at −189 mV. Depolarizing steps ( Fig. 2a) elicit an inward anion current with higher intensities for negative potentials and nearly zero current observed at +71 mV. The hyperpolarizing step induces a slow deactivating inward anion current, the instantaneous amplitude of which increases with the magnitude of the depolarizing pre-pulse. The instantaneous amplitude at the beginning of the hyperpolarizing step (tail current amplitude) reflects the activation status of the slow current at the end of the preceding depolarizing step. The plot of tail current amplitudes against the pre-pulse potentials ( Fig. 2b, open circles) shows the voltage-dependent activation of the slow channel by depolarization. At the end of the −189 mV voltage pulse ( Fig. 2b, filled circles), the slow current is completely deactivated.

Figure 2.

Slow activation of the current in response to depolarizing voltage steps.

The pulse protocol is composed of a 20 sec depolarizing voltage step ranging from −109 to +71 mV, followed by a 30 sec hyperpolarization to −189 mV.

(a) Outward currents of increasing intensities are elicited by increasingly depolarized voltage steps. Repolarization of the membrane to −189 mV results in large slow relaxation currents whose instantaneous amplitude increases with the amplitude of the depolarizing pre-pulse.

(b) Plot of the tail current (open circles) and of the current at the end of the −189 mV voltage pulse (closed circles) against the pre-pulse potential. Note that more positive activation pulses result in larger instantaneous negative current, while at the end of relaxation the current amplitude is independent of the voltage pulse. Pipette and bath compositions are the same as in Fig. 1.

The activation kinetics of the slow current were investigated by performing the pulse protocol shown in Fig. 3. This protocol allowed monitoring of the time dependence of the activation during the depolarizing step by progressively increasing (by increments of 5.6 sec) the duration of this step before stepping back to hyperpolarized potentials. Here again, the amplitude of the tail current magnifies the activation status of the slow current at the end of the preceding depolarizing step. Figure 3 shows, on the same protoplast, the slow deactivating current induced in these conditions for three different membrane potentials. The more depolarized potentials led to a higher activation of the slow channel, illustrated by a higher amplitude of the tail current at +71 mV than at −89 mV. The magnitude of the tail current, as a function of the duration of the activation pre-pulse, reflects the time-dependent activation of the slow anion current upon depolarization. This activation kinetics could be fitted by a mono-exponential equation giving time constants for activation (in sec) of 12 ± 1 (n = 3), 23 ± 11 (n = 4) and 18 ± 6 (n = 3) for the activation potentials at +71, −9 and −89 mV, respectively. These activation time constants in the 10 sec range are close to those described by Schroeder & Keller (1992) for guard cells of Vicia faba.

Figure 3.

Activation kinetics of the slow current.

The membrane potential is stepped from a holding potential of −189 mV to a depolarized membrane voltage of +71 mV for seven progressively increased durations (2.8–36.5 sec); for a second and a third set of measurements the depolarizing voltage is −9 and −89 mV, successively. The instantaneous tail current is dependent on the number of activated slow anion channels, therefore the activation is higher for the highest depolarizing potential and the activation kinetics fit a mono-exponential function. Pipette and bath compositions are the same as in Fig. 1.

To study the current–voltage relationship of the slow current, we performed a pulse protocol in which the membrane potential was clamped at +71 mV during 36 sec for channel activation and then stepped back to potentials varying from −189 to +91 mV by increments of 70 mV. The deactivation kinetics of the slow current at different membrane potentials ( Fig. 4a) were fitted by a mono-exponential equation, giving (in sec) values of 5.9 ± 1.5 (n = 34), 7.2 ± 2.1 (n = 33), 7.3 ± 3.2 (n = 34) and 5.2 ± 1.3 (n = 4) for membrane potentials of −189, −119, −49 and +21 mV, respectively, showing no significant voltage dependence of these time constants. The instantaneous and steady-state currents were plotted as a function of the imposed membrane potential ( Fig. 4b). The average maximal amplitude of the steady-state slow current was 12.1 ± 1.2 pA/pF (n = 70).

Figure 4.

Steady-state and instantaneous current–voltage relationships of the slow channel current.

The membrane potential was polarized from +71 mV to potentials ranging from −189 to +91 mV. (a) During the holding potential of +71 mV the channels activate, and the subsequent polarization step results in large tail currents for the more hyperpolarized potentials followed by a slow relaxation traducing channel deactivation. (b) Current–voltage relationship of the slow channel current. The plot of the current amplitude at the end of the voltage pulse (♦) against the transmembrane potential shows the voltage dependence of the slow anion channel in steady-state conditions, while plotting the tail current amplitudes (▪) against the transmembrane voltage shows the current driven by the electrochemical gradient allowed to flow through the open channel. Pipette and bath compositions are the same as in Fig. 1.

The slow current is carried by anions

To determine the ion selectivity of the slow current, we studied the current–voltage (I–V) relationship of the steady-state slow current for different nitrate concentrations in the bath, after the complete run-down of the rapid anion channel. In all experiments, the anion loaded into the cytosol of hypocotyl protoplasts was NO3 at a concentration of 50 m m. While decreasing the external NO3 concentration from 111.5 m m to 12.5 m m and to 2.6 m m while keeping Ca2+ and Mg2+ constant ( Fig. 5a), the mean reversal potential shifted from −14.2 ± 4.6 to 41.0 ± 7.8 and 76.3 ± 18.0 mV, respectively (n = 3). The measured reversal potential corresponded to the theorical Nernst potential ENO3 ( Fig. 5a, inset) indicating that the slow anion current is carried by nitrate. To exclude any contribution of a calcium component to the inward current due to the 50 m m CaCl2 in the bath, we switched the bath solution from 50 m m Ca2+ to 95 m m TEA+ plus 5 m m La3+, a large-spectrum calcium-channel blocker. Figure 5(b) shows that the I–V steady-state curve obtained in the standard CaCl2 bath medium does not significantly differ from the I–V curve recorded 5 min after perfusion with the TEACl bath. Reversal potentials measured on the same protoplasts were 77.8 ± 4.4 (n = 3) and 83.7 ± 7.6 mV (n = 3) for the calcium chloride and the TEA chloride baths, respectively. Figure 5(c) shows the shift of the reversal potential from 80.0 ± 6.6 (n = 3) to −14.9 ± 1.5 mV (n = 3) when external chloride was replaced by nitrate. The shift of the reversal potential by external NO3 and its value close to the calculated ENO3 of −18 mV, correlate with a high permeability for nitrate. We calculated the relative selectivity of the anion channel from the reversal potential measured with 50 m m NO3 in the cytosol and 110 m m Cl in the bath using an equation derived from the Goldman–Hodgkin–Katz current equation (see Eqn 1 in Experimental procedures). The permeability ratio obtained, PNO3 /PCl = 23.2 ± 1.3 (n = 69), indicates a large selectivity for nitrate relative to chloride. We recently showed that sulphate permeates through the rapid anion channel ( Frachisse et al. 1999 ). To test whether sulphate also permeates through the slow channel, the protoplasts were loaded with 25 m m potassium sulphate. In these conditions, for which rapid anion channel does not undergo run-down ( Frachisse et al. 1999 ), we performed the same ramp–pulse protocol as in Fig. 1, designed to monitor both rapid and slow currents. Figure 5(d) shows, 32 and 40 min after establishing the whole-cell configuration, that the ramp elicits a rapid voltage-dependent current, indicating that sulphate permates the rapid channel, while stepping the membrane potential from 71 to −189 mV does not elicit the slow relaxation typical for the slow current (compare the 10 min trace in Fig. 1 with the 32 min trace in Fig. 5d). This result indicates that, in contrast to the fast anion channel, the slow anion channel is not permeable to sulphate.

Figure 5.

Nitrate and sulphate permeability of the slow anion channel.

(a) Dependence of the slow anion current on the external nitrate concentration. The pipette solution with 50 m m KNO3 is same as in Fig. 1. Bath solutions containing 50 m m Ca2+, 5 m m Mg2+ and 111.5 or 12.5 or 2.6 m m NO3 and kept at pH 5.6 with MES are applied successively to the same protoplast. The inset shows the mean reversal potential (± SE) of the current as a function of external NO3 concentration (n = 3), and the dotted line represents the theorical Nernst potential ENO3.

(b) Calcium does not permeate through the slow anion channel, as when exchanging the standard bath (♦: 50 m m CaCl2, 5 m m MgCl2) for a free calcium bath (⋄: 95 m m TEACl, 5 m m LaCl3), reversal potential and current amplitude were unchanged. Pipette solution is the same as in Fig. 1.

(c) The current–voltage relationship with 50 m m KNO3 in the cytosol and 50 m m CaCl2 and 5 m m MgCl2 in the bath (♦) is compared with the current–voltage relationship obtained on the same protoplast after exchanging the bath with 50 m m Ca(NO3)2 and 5 m m MgCl2 (⋄). The shift in reversal potential from 80 ± 7 to − 15 ± 2 mV (n = 3) demonstrates a high nitrate selectivity relative to chloride; the large outward nitrate current obtained with the nitrate bath shows the large nitrate permeability of the channel. Pipette solution is the same as in Fig. 1.

(d) Sulphate does not permeate through the slow anion channel. Using a similar voltage protocol as in Fig. 1, only the rapid anion current is elicited by the voltage ramp with 25 m m K2SO4 instead of 50 m m KNO3 in the pipette (other ions same composition as Fig. 1), while the slow deactivating inward current is not observed when stepping the voltage from +71 to −189 mV. The two current traces super-imposed in the figure were been recorded 32 (trace 1) and 40 min (trace 2) after establishing the whole-cell configuration.

Nucleotide dependence of the slow anion channel

Different modes of regulation of anion channels by nucleo- tides have been proposed: a putative nucleotide binding site for the rapid-type anion channels of Arabidopsis hypocotyl epidermal cells ( Thomine et al. 1997a ) and Vicia faba guard cells ( Schulz-Lessdorf et al. 1996 ), or protein phosphorylation/dephosphorylation for the slow-type channel of Vicia faba and Arabidopsis guard cells ( Pei et al. 1997 ; Schmidt et al. 1995 ). To gain further insight into the regulation of the slow anion channel from hypocotyl cells, different nucleotides have been loaded in the cytosol. To determine whether nucleotides are required to observe the slow anion current, ATP and GTP were omitted from the pipette solution. In such conditions, the instantaneous and steady-state slow currents are almost completely abolished within 10–20 min after establishing the whole-cell configuration ( Fig. 6a). This shows that nucleotides are necessary to maintain the slow anion current.

Figure 6.

Nucleotide regulation of the slow anion current involves protein phosphorylation.

(a) Current–voltage curves obtained after 20 min of perfusion of the cytosol with standard medium containing 5 m m MgATP + 5 m m LiGTP (closed symbols) or no nucleotide (open symbols) are shown for the instantaneous (squares) and steady-state currents (diamonds).

(b) Mean whole-cell current densities (± SE) measured at the beginning of the pulse at −189 mV for the instantaneous current (dashed bar) or at the maximum of the steady-state current (grey bar) as a function of the intracellular nucleotides: (from left to right) 5 m m MgATP + 5 m m LiGTP, 0 m m nucleotide, 5 m m LiAMP-PNP. Numbers in brackets refer to the number of protoplasts.

(c) Effect of the membrane-permeable kinase inhibitor staurosporin on the slow anion current with pipette containing 5 m m MgATP. Time course of the instantaneous slow inward current at −189 mV measured every 90 sec. DMSO 0.5% was perfused in the bath from 20 to 73 min, and the kinase inhibitor staurosporin 10 μm was applied in the bath from 35 to 47 min. DMSO and staurosporin induced reversible inhibition of the current. The inset represents the residual instantaneous current (as a percentage of the control) measured at different points of the experiment. Pipette and bath compositions are the same as in Fig. 1; numbers in brackets represent the number of protoplasts tested.

Figure 6(b) shows the whole-cell current density measured either at the tail at −189 mV (dashed bars) or at the steady-state maximum current (filled bars). Instantaneous currents at −189 mV or steady-state maximum currents follow the same dependence with regard to intracellular nucleotides. In the absence of intracellular nucleotides, the slow anion current is abolished. When the non-hydrolysable 5-adenylylimidodiphosphate (AMP-PNP) (5 m m) was loaded in the pipette, a significant fraction of the slow anion current was present. This may be indicative of a complex regulation of this current by nucleotides, involving some mechanisms that require hydrolysis and colleagues that do not.

To determine whether the requirement for hydrolysable ATP for activation of the slow anion channel is linked to protein phosphorylation, the effect of the membrane-permeable kinase inhibitor staurosporin on the slow anion current was tested. Hypocotyl protoplasts were perfused with a pipette solution containing 5 m m MgATP. As shown in Fig. 6(c), the staurosporin solvent, 0.5% dimethyl sulphoxide (DMSO), itself induces a partial decrease of the current to 76.3 ± 9.3% (n = 7) of the control. After getting a steady-state current in the presence of DMSO, 10 μm staurosporin added to the bath induces a further decrease of the current to 46.8 ± 10.8% (n = 7), corresponding to 39% of inhibition of the DMSO control. Removing staurosporin from the bath restores the current after a 15–30 min lag. The DMSO effect was also reversible since withdrawing this solvent restored the initial current. To confirm this result, the protein kinase inhibitor K252a was also tested at a concentration of 10 μm. An inhibitory effect similar to that induced by staurosporin was observed (44.6% of inhibition of the DMSO control, n = 4, data not shown).

Inhibition of the slow anion current by anion channel blockers

To be able to link the function of the slow anion channel identified at the plasma membrane of Arabidopsis hypocotyl cells with more integrated physiological responses, we analysed the pharmacological properties of this current. The effects of six anion channel inhibitors – NPPB (nitrophenyl-propylamino-benzoic-acid), 9-AC (anthracene-9-carboxylic acid), DIDS (4,4′-diisothiocyanatostilbene-2,2′-disulphonic acid), SITS (4-acetamido-4-isothiocyanostilbene-2,2′-disulphonic acid), IAA-94 (indanyl-oxyacetic acid (6,7-dichloro-2-cyclopentyl-2,3-dihydro-2-methyl-1-oxo-1H-indan-5-yl oxy)) and niflumic acid were tested in the whole-cell configuration. After complete run-down of the rapid current, a pulse protocol consisting of 25 sec at +71 mV followed by 30 sec at −189 mV was applied each 60 sec. When getting a stable slow deactivating anion current, a pulse protocol similar to that described in Fig. 4(a) was applied to record the I–V relationship in control conditions. Then the inhibitor to be tested was added to the perfusion bath at a concentration of 100 μm while performing the +71/−189 mV pulse protocol in order to follow the evolution of the current. A second I–V curve was recorded when a stable inhibition of the current was observed after 3–5 min of presence of the inhibitor in the bath solution. Figure 7(a) shows I–V curves for instantaneous and steady-state currents in control conditions and in the presence of DIDS. The extent of inhibition was similar for the tail current and the steady-state maximum current, but varied from one inhibitor to the another ( Fig. 7b). DIDS was the most potent inhibitor; SITS, niflumic acid and NPPB showed moderate inhibitiory effects; and IAA-94 and 9AC were totally inactive. Washing out the different inhibitors resulted in recovery of the slow current, except with DIDS for which the inhibition was irreversible even after a long washing time of up to 30 min. Figure 7(c) shows the concentration-dependent inhibition of the slow channel by DIDS. From this curve, a DIDS concentration for half inhibition (IC50) of 26 μm was deduced.

Figure 7.

Inhibition of whole-cell slow anion currents by anion channel blockers.

(a) Current–voltage curves before (closed symbols) and after (open symbols) addition of the most potent inhibitor DIDS (100 μm) in the bath. I–V curves for instantaneous (squares) and steady-state (diamonds) currents are derived from the same pulse protocol as for Fig. 4.

(b) Mean percentage (± SE) of residual current measured at the steady-state maximum current (grey bar) and at the tail (instantaneous current at −189 mV, white bar) after inhibition by 100 μm of IAA-94, NPPB, niflumic acid, 9-AC, DIDS and SITS. Numbers in brackets refer to the number of protoplasts tested.

(c) Concentration-dependent inhibition of the slow anion current by DIDS. Steady-state; each point results from 3–5 independent measurements. Pipette and bath compositions are the same as in Fig. 1.

Discussion

Characteristics of the slow anion channel current of hypocotyl cells

In addition to the previously identified rapid voltage-dependent anion channel at the hypocotyl epidermal cells ( Frachisse et al. 1999 ; Thomine et al. 1995 ; Thomine et al. 1997a ), we have characterized in the present work a second type of channel with slow activation and deactivation kinetics and time constants in the 10 sec range. This channel is activated by voltages positive to −100 mV and deactivated at more negative potentials, and it displays little or no inactivation. In terms of activation–deactivation kinetics and voltage-dependence, the slow anion channel of hypocotyl cells is very close to the S-type anion conductance from Vicia faba guard cells ( Schroeder & Keller 1992) and shows similarities with the Gs anion conductance described by Dieudonnéet al. (1997) in Coeffea arabica protoplasts. The hypocotyl slow anion channel also resembles the slow anion channel in Arabidopsis guard cells ( Pei et al. 1997 ), although the absence of detailed analysis of kinetic parameters for this channel does not allow a close comparison.

We found a high selectivity of the channel for nitrate relative to chloride (PNO3 /PCl = 23.2). High-intensity outward anion current was recorded when external chloride was substituted by nitrate, indicating a good permeation of this anion through the channel. In terms of nitrate selectivity, the hypocotyl slow channel is very similar to the S-type channel of Vicia faba guard cells which shows a PNO3 /PCl of 20.9 ( Schmidt & Schroeder 1994). Unlike nitrate, sulphate does not permeates through the hypocotyl slow channel since no slow deactivating current can be observed when the cytosol is loaded with sulphate. In a recent study ( Frachisse et al. 1999 ), we have shown by capillary ion electrophoresis analysis that aerial parts of Arabidopsis plantlets, in our culture conditions, exhibit a high endogenous level of nitrate of about 70 m m and contain 20 m m sulphate. Taking into account the negative membrane potential of hypocotyl cells of −150 mV (A. Kurkdjian, personnal communication), the channel permeability and endogenous anion concentrations, the opening of the slow channel should lead to a sustained nitrate efflux.

The activity of the slow anion channel is nucleotide-dependent since removing ATP and GTP from the cytosol completely abolished the anion current, while the non-hydrolysable ATP analogue AMP-PNP maintains only a small proportion of the current. Staurosporin, a membrane-permeable non-specific protein kinase inhibitor which is known to block protein phosphorylation efficiently ( Ruegg & Burgess 1989) was able to inhibit about 40% of the slow current. These results indicate that ATP acts at least partly through phosphorylation processes. Hence, the nucleotide regulation shares similarities with S-type anion channel of Vicia faba guard cells: phosphorylation activates the current. However, the nucleotide regulation also involves mechanisms that do not require nucleotide hydrolysis. These additional regulatory elements could be ATP allosteric interactions as reported for the CFTR (cystic fibrosis transmembrane regulator) Cl channels ( Foskett 1998; Hwang et al. 1994 ), in agreement with the fact that ATP binding cassette modulators affect a guard cell slow anion channel ( Leonhardt et al. 1999 ). Further investigations will be necessary in order to obtain a detailed description of the regulation pathways involving nucleotides and controlling the activity of the hypocotyl slow anion channel.

It is of interest that in Arabidopsis hypocotyl cells the slow anion channel is regulated in a different manner than its homologue in guard cells. The presence of intracellular MgATP which activates the channel of hypocotyl cells does not produce slow anion channel current in guard cells ( Pei et al. 1997 ), as channel activation requires addition of the stress hormone ABA. This difference in nucleotide sensitivity probably indicates a differential regulation of related channels between two different organs of the same Arabidopsis plant. This idea is reinforced by the observation that in our patch–clamp conditions on hypocotyl cells, i.e. in the presence of intracellular nucleotides, we did not find any effect of ABA on the slow channel activity. In particular, unlike Vicia faba guard cells ( Schwartz & Schroeder 1998), ABA treatments did not maintain the activity of the slow channel in ATP-depleted Arabidopsis hypocotyl cells (Colcombet et al. unpublished results).

The pharmacological profile of the Arabidopsis hypocotyl slow channel differs from that of the S-type channel of broad bean guard cells. In Vicia faba, strong inhibition by NPPB, niflumic acid and IAA-94 (IC50 = 5–10 μm), moderate inhibition by 9-AC (IC50 = 55 μm), and almost no inhibition by DIDS were observed ( Schroeder et al. 1993 ; Schwartz et al. 1995 ), while in Arabidopsis hypocotyl cells the most efficient blocker is DIDS (80% inhibition at 100 μm), with SITS, niflumic acid and NPPB being less active (60% inhibition for 100 μm), and 9-AC totally inefficient. This shows that two channels with similar kinetics, voltage-dependence and selectivity can have very different blocker sensitivities, probably reflecting differences in the structure/affinities of the blocker binding sites within the channel pore. Among the blockers tested, stilbene derivatives seem to be the more efficient on the hypocotyl channel. The irreversibility of the inhibitory effect of DIDS is not surprising, since this type of molecule has been synthesized to covalently bind animal Cl channel proteins. Indeed, the irreversibility of the inhibition by DIDS has already been described on the chloride channel of torpedo ( White & Miller 1979), further identified as CLC-0. It is noteworthy that 100 μm DIDS or even a higher concentration (300 μm, not shown) blocks the whole voltage-dependent component of the slow current without shifting the reversal potential, revealing a novel voltage-independent, DIDS-insensitive anion current. A recent study of slow anion currents in intact guard cells of different species ( Forestier et al. 1998 ) has shown that in Arabidopsis only 9-AC is able to block the channel. Such a result illustrates that the pharmacological properties of anion channels are both species- and organ-dependent.

Two types of anion channels are active simultaneously in hypocotyl cells

In most studies, one channel-type is associated with one experimental condition. S-type or R-type anion channels have been observed independently but the occurrence of both types in the same cell has been shown only in Vicia faba guard cells ( Schroeder & Keller 1992). The present work demonstrates that both slow and rapid anion channels can be recorded under the same ionic conditions in the same cell, and function at the same time as revealed by the ramp–pulse protocol. The question of whether slow and rapid channels in a given cell type represent two functioning modes of the same molecular entity or two different channels has been raised by different authors ( Dietrich & Hedrich 1994; Pei et al. 1997 ) but is still unanswered. Comparison of the kinetics, nucleotide-dependence and pharmacological properties of the hypocotyl slow and fast anion currents described in this paper supports the hypothesis that they are actually carried by two separate channels.

As already mentioned above, the activation time constant of the slow current is in the 10 sec range while the rapid one is in the millisecond range ( Thomine et al. 1995 ). Activation time constant measurements both for the slow and the rapid anion current give, respectively, 18 ± 6 sec (n = 3) and 4.4 ± 0.6 msec (n = 5) at −89 mV, showing a difference of three orders of magnitude between the two. The same holds true for the deactivation constants, with values of 5.9 ± 1.5 sec (n = 34) and 1.9 ± 0.6 msec (n = 2) at −189 mV for the slow and rapid channels, respectively. An important point to consider is that the slow mode of the rapid anion current described in the absence of ATP ( Thomine et al. 1995 ) represents a particular functioning mode of this channel and should not be confused with the slow anion current described here. The rapid anion channel in its slow mode is observed in the absence of intracellular nucleotides while in this same situation the slow current is not activated. Nucleotide regulation is a second element which strongly discriminates slow and rapid anion channels. The slow channel is inactive in the absence of nucleotide but is activated by intracellular ATP. In contrast, we have shown earlier that the effect of nucleotides on the rapid channel was to shift the voltage gate towards depolarized potentials which leads to a reduction of the current ( Thomine et al. 1997a ). Slow and rapid anion channel currents also display marked differences in their selectivity. The selectivity for nitrate relative to chloride is close to 20 for the slow channel. In contrast, it was previously shown to be only 2.6 for the rapid channel ( Frachisse et al. 1999 ). Furthermore, sulphate does not permeate the slow anion channel but is a good substrate for the rapid channel. Another feature to be mentioned is the distinct pharmacological profiles of the two channels: DIDS is the best inhibitor of the slow channel whereas it is inactive on the rapid anion channel ( Thomine et al. 1997a ). Slow-type and rapid-type guard cell anion channels of Vicia faba also show a marked difference in their pharmacological signatures ( Marten et al. 1992 ; Marten et al. 1993 ; Schroeder et al. 1993 ; Schroeder 1995; Schwartz et al. 1995 ). All these biophysical and pharmacological differences between the slow-type and rapid-type anion currents from the hypocotyl cells suggests strongly that they are carried by distinct channels. However, the only way to answer this question unambiguously will be through cloning of the genes encoding the channel proteins.

Possible physiological functions of the slow anion channel in hypocotyl cells

The role of plant plasma membrane anion channels in a well-defined physiological function has been clearly illustrated only in few cases. The best documented example concerns the requirement of a slow anion channel for the osmo-regulation of stomatal guard cells. Assorted pharmacological evidence obtained in Vicia faba either with anion channel blockers ( Schroeder et al. 1993 ; Schwartz et al. 1995 ) or with kinase and phosphatase inhibitors ( Schmidt et al. 1995 ) shows a good correlation between the ability of these agents to reduce the anion current and to counteract the stomatal closure induced by ABA. Genetic evidence has also been provided in Arabidopsis thaliana by the study of mutants affected in both stomatal responses to ABA and regulation of the slow anion channel ( Pei et al. 1997 ; Pei et al. 1998 ). The similarity between the slow anion current described in guard cells and the current that we have characterized in this work suggests that they may be involved in similar functions. Thus, the slow anion current could be involved in long-term anion efflux which, when it is associated with potassium efflux, can regulate the osmotic pressure of hypocotyl cells. Its activation by phosphorylation, probably in response to a signal, would then lead to a decrease in osmotic pressure, while maintaining a high osmotic pressure is thought to be a prerequisite for cell expansion. Slow anion channels might thus be involved in the down-regulation of hypocotyl cell elongation.

On the basis of their regulation properties and/or their pharmacology, anion channels have been proposed to participate in the control of cell expansion by light in Arabidopsis hypocotyls ( Cho & Spalding 1996) or pea leaves ( Elzenga & Van Volkenburgh 1997), and by auxin in oat coleptiles ( Keller & Van Volkenburgh 1996). To investigate a possible role of anion channels in the regulation of hypocotyl cell elongation in Arabidopsis thaliana, Thomine et al. (1997b) also used a pharmacological approach. A variety of anion channel blockers were studied for their effect on hypocotyl elongation and its regulation by auxin. Several anion channel blockers such as DIDS, 9-AC and SITS were able to counteract the inhibition of elongation induced by auxin, whereas IAA-94 showed a moderate efficiency, and NPPB and niflumic acid only inhibited hypocotyl elongation ( Thomine et al. 1997b ). Furthermore, a small stimulating effect on the hypocotyl elongation was observed for 9-AC, DIDS and SITS when applied alone. This pharmacological signature described at the whole-plant level is distinct from the profiles of sensitivity to blockers of the rapid-type anion channel (insensitive to DIDS, Thomine et al. 1997a ) and of the blue-light-stimulated anion channel (highly sensitive to NPPB ( Cho & Spalding 1996) formerly described at the plasma membrane of hypocotyl cells. While stilbene derivatives DIDS and SITS are both active on the newly identified slow anion channel and on the hypocotyl elongation, 9-AC is only able to counteract auxin inhibition of hypocotyl elongation but not to block the slow channel. Two interpretations may account for this result: (i) the slow channel is not involved in the hypocotyl developmental response, or (ii) the slow anion channel is involved in the control by auxin of hypocotyl elongation and the discrepancy between the effect of 9-AC at the plant level and at the channel level might reflect an action of this molecule on another target, such as an anion transporter, other channel types or other anion channels located on the vacuolar membrane. In favour of this hypothesis, several examples of non-specific inhibition of plant ion channels by molecules developed to block animal ion channels have already been described. Garrill et al. (1996) have shown that anion channel blockers interact with K+-outward rectifiers, and Greger (1990) has emphasized that some Cl channel inhibitors are also able to interact with anion transporters. This illustrates the difficulty of linking the transport activity of a channel protein monitored in patch–clamp experiments to an integrated phenomenon such as cell elongation. In addition to its specific pharmacology, a remarkable feature of the slow anion channel is its specificity for NO3 over other anions, which may provide another clue to its physiological function. Under our culture conditions, NO3 is the major anion stored by the plantlets ( Frachisse et al. 1999 ). This suggests that, besides a general role in osmotic regulation, this channel could play a specific role in maintaining the homeostasis of the cytosolic nitrate concentration.

In conclusion, we have identified for the first time on the hypocotyl plasma membrane a slow anion channel sharing similar kinetic properties with the slow guard cell anion channel. This channel, which is highly permeant for nitrate, co-resides with the rapid anion channel formerly characterized. By mediating sustained anion efflux due to its slow deactivation kinetics, the slow anion channel is proposed to be involved in cell volume regulation. By contrast, with its fast kinetics of activation and deactivation, the rapid channel co-residing on the same cell would be presumed to participate in fast electrical signalling. Characterization of the in vivo biological functions of the slow and rapid anion channels in Arabidopsis hypocotyls, as well as identification of the genes encoding channel proteins, represent the major directions for further investigations.

Experimental procedures

Plant material and protoplast isolation

Seedlings (Arabidopsis thaliana ecotype Columbia) were grown on a medium containing 5 m m KNO3, 2.5 m m K2HPO4/KH2PO4, pH 6, 2 m m MgSO4, 1 m m Ca(NO3)2, 1 m m MES, 50 μm Fe-EDTA, Murashige and Skoog micro-elements ( Murashige & Skoog 1962), 10 g l−1 sucrose and 7 g l−1 agar. Culture conditions were 21°C, with a 16 h day length at lighting levels of 120 μE m−2 sec−1 with neon tubes (a combination of Mazdafluor Blanc Industrie and Mazdafluor Prestiflux). Seedlings aged 7–8 days were used for electrophysiological investigations. Hypocotyls were excised from 30–40 seedlings and protoplasts were isolated according to Elzenga (1991) as described in Thomine et al. (1995) .

Electrophysiological investigations

Patch–clamp experiments in the whole-cell configuration were performed as described by Hamill et al. (1981) using an EPC 7 patch–clamp amplifier (List Electronic, Darmstadt, Germany) with a low-pass filter (eight-pole Bessel filter) or an Axon 200A amplifier (Axon Instruments, Foster City, CA, USA). During measurements, freshly isolated epidermal protoplasts from Arabidopsis hypocotyls were maintained in a bathing medium containing 50 m m CaCl2, 5 m m MgCl2, 0.5 m m LaCl3, 10 m m MES-Tris, pH 5.6. The pipettes were filled with 50 m m CsNO3, 2 m m MgCl2, 5 m m EGTA, 4.2 m m CaCl2, 10 m m Tris/HEPES, pH 7.2, supplemented with 5 m m MgATP and 5 m m Li GTP, or other nucleotides as indicated in the figure legends. The osmolalities of bath and pipette solutions were adjusted to 450 mOsmol with mannitol using a Wescor 5500 vapour pressure osmometer (Wescor, Logan, UT, USA). Giga-ohm resistance seals between pipettes (pipette resistance 1–5 MΩ) coated with Sylgard (General Electric, New York, USA) pulled from Kimax-51 capillaries (Kimble Glass Inc., Owens, IL, USA) and protoplast membranes were obtained, with gentle suction leading to whole-cell configuration. The liquid junction potential was measured according to Neher (1992) and all potentials given have been corrected for the junction potential. Unless otherwise indicated, the figures show values for one representative protoplast, and statistics are given as mean ± SE (n indicates the number of protoplasts tested).

Permeability ratios for nitrate relative to chloride were calculated from reversal potential measurements using Eqn 1, derived from the Goldmann–Hodgkin–Katz equation (see Hille 1992).

PNO3/PCl = (Clo e(EF/RT) – (Cli)/(NO3i)      (1)

where E is the reversal potential, R is the gas constant, T is the absolute temperature and F is the Faraday's constant, have their usual meanings and Clo, Cli and NO3i represent the extracellular and intracellular activities of chloride and the intracellular activity of nitrate, respectively. Activity coefficients were estimated according to Dean (1985).

Acknowledgements

The authors thank Dr Sebastien Thomine for helpful discussion and critical reading of the manuscript. This work was supported by the CNRS grant to UPR 0040.

Ancillary