SEARCH

SEARCH BY CITATION

Summary

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Conclusion
  7. Experimental procedures
  8. Acknowledgements
  9. References

The symbiotic membrane between N2-fixing bacteroids and plant cytoplasm in nodules of soybean contains a sub-picoSiemen cation channel permeable to NH4+. With millimolar concentrations of Ca2+ or Mg2+ on the cytoplasmic side, the channel rectifies current in the direction of cation influx to the cytoplasm. When Ca2+ is present on the bacteroid side of the membrane the current is rectified in the opposite direction. With submicromollar concentrations of divalent on both sides, the channel no longer rectifies. The channel is inhibited by verapamil on the bacteroid side of the membrane with a Kd of 2.6 μM. In the presence of millimolar concentrations of divalents on the cytoplasmic side, the conductance as a function of voltage is fitted by a simple Boltzmann equation with an effective gating charge equal to one. The voltage at which the conductance reaches 50% of maximum is dependent on external NH4+, shifting negative at lower concentrations. The time-course of activation upon hyperpolarisation can be described by the Hodgkin-Huxley equation with Ca2+present on the cytoplasmic side. With Mg2+ the channel activates with single exponential kinetics. The time constant for activation is weakly voltage dependent. Upon depolarisation of the membrane the channel deactivates with double exponential kinetics, the time constants being slightly voltage dependent. We propose a model of the channel in which divalent block is relieved when the blocking ion is dislodged by univalent cation flux into the pore. Mg2+ on the cytoplasmic side may functionin vivoas the gating particle of the channel.


Introduction

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Conclusion
  7. Experimental procedures
  8. Acknowledgements
  9. References

Leguminous plants such as soybean (Glycine max) form symbiotic relationships with nitrogen fixing soil bacteria known collectively as rhizobia. Bacteria infect legume roots via a complex signal interaction process (Mylona et al. 1995) to become established within cortical cells. Endocytosis of the cortical cell plasma membrane contains the symbiotic nitrogen fixing bacteroids within a membrane bound structure known as the symbiosome, which persists throughout the life of the symbiosis. The symbiosome membrane, known as the peribacteroid membrane (PBM), has similarities to both the plasma and vacuolar membranes (for review, see Whitehead & Day 1997).

The major nutrient exchange in the legume/rhizobium symbiosis is fixed nitrogen (NH3/NH4+) from the bacteroids for reduced carbon from the plant. All nutrients are exchanged across the PBM making the PBM an important control point in the symbiosis. Protons are pumped into the symbiosome by a P-type ATPase that creates a trans-PBM pH gradient inside the symbiosomes acidic (Blumwald et al. 1985) by 1–1.6 pH units (Udvardi et al. 1991), and a membrane potential difference inside the plant cytoplasm negative (Udvardi & Day 1989). The mechanism of fixed nitrogen transfer from the bacteroids to the plant involves transport of NH3 and NH4+. NH3 probably diffuses from bacteroids into the space between bacteroids and the PBM known as the peribacteroid space (PBS, Howitt et al. 1986). Some NH3 may then diffuse across the PBM to the plant cytoplasm but the majority will be protonated to NH4+ in the acidic environment of the PBS. Export of NH4+ to the plant cytoplasm would require a transport system. The transport could be passive since assimilation in the cytoplasm by glutamine synthetase ensures a steep concentration gradient and the proton pump generates an electrical gradient.

Transport systems capable of transporting NH4+ have been identified in the soybean and pea PBM (Mouritzen & Rosendahl 1997;Tyerman et al. 1995). The patch-clamp method was used to identify the transport system in soybean PBM whereupon the NH4+ current flow was found to be entirely passive but single-channel currents could not be observed. The noise characteristics of the current suggested that the transport system was a subpico Siemen channel permeable to univalent cations. In this paper we will refer to the transport system as a channel with the acknowledgement that some carrier transporters can show channel-like gating charactersitics (Hilgemann 1996). Under the conditions used by Tyerman et al. (1995), the PBM-cation channel showed strong inward rectification, i.e. it only allowed passage of univalent cations to the cytoplasmic side. The channel activated in a time- and voltage-dependent manner with no inactivation during persistent hyperpolarisation, and was also shown to be inhibited by Ca2+ on the bacteroid side of the membrane (Kd = 22 μM). In these respects, the channel is similar to plant inwardly rectifying K+ channels on the plasma membrane (Schroeder et al. 1994). However, the PBM channel is rather non-selective between monovalent cations at high concentations (150 mm), but is more permeable to NH4+ than to K+ at lower concentrations near to the apparent Km of the channel (20 mm) (Tyerman et al. 1995) and near to the estimated NH4+ concentration in the PBS (Streeter 1989). Most plant K+ channels are permeable to NH4+ (e.g. Hedrich et al. 1995;Schachtman et al. 1992) but to a lesser extent than K+. It seems very likely that the channel identified in the PBM functions as an NH4+ channel in vivo.

The apparent rectification of the NH4+ channel in the PBM may have biological significance even though the conditions in the nodule promote passive NH4+ export. For example, if rectification occurs in vivo it may maintain a balance between H+ and NH4+ transport. Alternatively, rectification may cause the channel to behave as a nutrient valve, allowing only NH4+ efflux under all conditions. The mechanism of rectification in the NH4+ channel is not known and in fact the mechanism of rectification in most plant K+ channels is not well understood. The functional significance of rectification in plant K+ channels is to maintain cytoplasmic K+ concentration and to regulate membrane potential. Studies of animal K+ channels suggest that rectification can be caused by one of two mechanisms, intrinsic voltage-dependent gating or voltage-dependent block by a cytoplasmic ion. Interestingly, most animal outward rectifiers seem to be controlled by the first mechanism, whereas the inward rectifiers appear to be controlled by the second (Nichols & Lopatin 1997). The ion responsible for cytoplasmic block in many cases has been shown to be Mg2+ (Matsuda 1991) but, more recently, polyamines have also been implicated (Ficker et al. 1994;Lopatin et al. 1994). The rectification of plant inward rectifiers in the plasma membrane examined to date has been shown to be independent of Mg2+, and has also suggested to be independent of polyamines (Hoshi 1995;Müller-Röber et al. 1995;Schroeder 1995). The amino terminus and first four membrane spanning domains of plant inward rectifiers probably contains the intrinsic voltage gating residues (Cao et al. 1995). In contrast to these channels, the fast-vacuolar (FV) channel in barley mesophyll tonoplast (technically a K+ outward rectifier) appears to be rectified by Ca2+ (Tikhonova et al. 1997).

In this paper we describe the characteristics of rectification of the NH4+ permeable channel in soybean PBM. The information gained about the channel is used to propose a model of the gating mechanism.

Results

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Conclusion
  7. Experimental procedures
  8. Acknowledgements
  9. References

Effect of Ca2+ and Mg2+ on rectification

Under standard experimental conditions, the NH4+ currents across a patch of the PBM were inwardly rectified (Fig. 1 and Tyerman et al. 1995). In solutions symmetrical for monovalent cations, hyperpolarising voltage pulses resulted in an instantaneous jump followed by a slow time-dependent increase in inward current until a steady state value was reached. Voltage pulses from a holding potential of 0 mV to positive membrane potentials resulted in the time dependent decrease of outward current (also see Fig. 7) to very small outward steady state currents. Instantaneous current-voltage relations were generally linear whereas the time-dependent currents were rectified inward (Fig. 1b). Because no single channel records could be observed, it is difficult to assess the level of the true baseline current (see verapamil block). However, it seems likely that a proportion of the instantaneous current can be attributed to current flow through NH4+ channels at the holding potential (0 mV). A clearer demonstration of this can be seen in later figures showing Ca2+ and verapamil inhibition (Figs 3 and 4).

image

Figure 1. Ammonium currents recorded in the presence of cytoplasmic divalent cations.

Recordings were made of ammonium currents across an inside out patch of symbiosome membrane. Measurements were made with 150 mm NH4+ in the bath.

(a) Currents recorded in standard pipette solution (10 mm Ca2+) in response to voltage pulses of 4 sec in duration at 20 mV intervals from +100 mV to –140 mV (holding potential 0 mV).

(b) The amplitude of instantaneous (○) and steady state (•) currents recorded under the conditions specified in (a) as a function of voltage. Data points represent the mean ± SEM of 10 separate patches. Lines are low order polynomials fitted to the data.

(c) Current recorded with standard bath solution (2 mm Mg2+, 100 nm Ca2+) in the pipette in response to voltage pulses of 3.5 sec in duration at 20 mV intervals from +100 mV to –120 mV (holding potential 0 mV).

(d) The amplitude of instantaneous (○) and steady state (•) currents recorded under conditions specified in (c) as a function of voltage. Data points represent the mean ± SEM of six separate patches. Lines are low order polynomials fitted to the data.

Download figure to PowerPoint

image

Figure 7. Kinetics of time dependent current deactivation.

Time dependent ammonium current deactivation. Measurements were perfomed with the standard pipette solution and a bath solution containing 150 mm NH4+.

(a) Tail currents recorded by pulsing the voltage to more positive values at 20 mV intervals up to +100 mV from a holding potential of –100 mV.

(b) Tail currents recorded in response to four different voltage steps more positive of a holding potential of –100 mV. Lines represent data fitting with a double exponential equation (see text).

(c) Deactivation time constants derived from curve fitting as shown in (b) as a function of voltage. Data points represent the mean ± SEM of five separate patches. Lines represent data fitting with linear regression analysis. Slope of τ2 not significantly different from zero, slope of τ1 data = –1.909 ± 0.254 ms mV–1.

Download figure to PowerPoint

image

Figure 3. The effect of increasing Ca2+ concentration on the bacteroid side of the membrane in the presence of calcium on the cytoplasmic side (a). Steady state ammonium currents as a function of voltage at a range of calcium concentrations on the bacteroid side. The pipette contained the standard pipette solution. The bath contained 20 mm NH4+ and Ca2+ concentrations between 0 and 1 mm.

(b) The dependence of ammonium currents on Ca2+ concentration on the bacteroid side. Currents were measured at –120 mV (circles), – 80 mV (squares), – 60 mV (triangles) and +60 mV (diamonds). The data points for all voltages were fitted simultaneously by a Hill equation, giving a Hill coefficient of 0.76 ± 0.10 and log Kd = 0.92 ± 0.08, i.e. Kd = 8.34 μm.

Download figure to PowerPoint

image

Figure 4. Verapamil inhibition and channel characteristics in the absence of divalent cations

(a) K+ currents as a function of voltage at different external K+ concentrations. The pipette solution comprised 20 mm KCl, 5 mm HEPES/Tris (pH 7.2), 10 mm EGTA, 2.3 mm CaCl2. The bath solution contained K+ at either 150 mm (▪), 60 mm (□), 20 mm (♦) or 2 mm + 50 μm verapamil (▴). Lines represent data fitting with the GHK current equation for 150, 60 and 20 mm K+ and a linear regression for 2 mm K+ + 50 μm verapamil.

(b) The channel is selective for NH4+ over K+. The pipette contained 125 mm KGlu, 10 mm HEPES/KOH (pH 7.2), 10 mm EGTA, 2.3 mm CaCl2. The bath solution contained NH4+ at either 150 mm (▪), 60 mm (□), 20 mm (▴), 6 mm (♦) or 2 mm (•). Lines represent data fitting with the GHK current equation.

(c) The dependence of ammonium currents on verapamil concentration on the bacteroid side of the membrane. The pipette contained standard pipette solution. The bath solution contained 20 mm NH4+ and verapamil concentrations between 0 and 10 μm. Currents were recorded at –120 mV and normalised to the current in 0 μm verapamil. Data points are the mean ± SEM of three separate patches. The data were fitted by the Hill equation, giving a Hill coefficient of 3.00 ± 1.15 and log Kd = 0.42 ± 0.07, i.e. Kd = 2.62 μm.

Download figure to PowerPoint

The pipette solution for the data in Fig. 1(a) contained a high concentration of Ca2+ in order to achieve a tight electrical seal with the membrane. As mentioned earlier, studies of animal K+ channels have shown that multivalent cations in the cell cytoplasm caused inward rectification of currents. We therefore investigated whether Ca2+ or Mg2+ on the cytoplasmic side of the membrane was contributing to the rectification of NH4+ currents.

Membrane patches were formed using a pipette filled with standard pipette solution containing Ca2+ at the tip and a solution containing a high concentration of EGTA in the backfill. Over time, the backfill solution replaced the pipette tip solution by diffusion, gradually reducing the Ca2+ concentration at the membrane surface. Currents were recorded in response to voltage pulses to ± 80 mV (holding potential 0 mV) at 4 min intervals for 30 min (Fig. 2a). Immediately after patching (6 min), the NH4+ currents showed similar characteristics to those seen in Fig. 1. As the experiment progressed, three changes in NH4+ currents could be distinguished. First, the current became less rectified over time, i.e. the magnitude of inward and outward currents became symmetrical. Second, the time dependence of inward current activation was significantly reduced. And third, the magnitude of steady state inward currents increased slightly. The experiment was repeated three times with similar results. The free Ca2+ concentration in the pipette at equilibrium was calculated to be approximately 100 nm.

image

Figure 2. The effect of Ca2+ on ammonium currents

(a) The effect of reducing cytoplasmic Ca2+. Measurements were made with 150 mm NH4+ in the bath. The pipette contained standard pipette solution in the tip and a backfill comprising 125 mm Kglutamate, 10 mm HEPES/KOH (pH 7.2), 10 mm EGTA, 2.3 mm CaCl2. The figure represents the superposition of seven current recordings made 6 min after patching and every 4 min subsequent to this for a total of 30 min. Currents were recorded in response to three voltage pulses at 0 and ±80 (holding potential 0 mV).

(b-d) The effect of increasing Ca2+ on the bacteroid side of the membrane. Measurements were performed in symmetrical 20 mm NH4+. The pipette contained 20 mm NH4+, 10 mm HEPES/Tris (pH 7.2), 10 mm EGTA, 2.3 mm CaCl2. Currents were recorded in response to voltage pulses of 4 sec in duration at 20 mV intervals from +120 mV to –120 mV (holding potential 0 mV).

(b) At the beginning of the experiment, the Ca2+ concentration on the bacteroid side was buffered to 100 nm with 10 mm EGTA.

(c) By bath perfusion the solution was then changed to 10 mm Ca2+.

(d) The amplitude of instantaneous (circles) and steady state (squares) currents as a function of voltage for the two different calcium concentrations shown in b (closed symbols) and c (open symbols).

Download figure to PowerPoint

It is possible that an increased concentration of EGTA in the pipette, rather than a decrease in Ca2+, caused the loss of rectification. However, this seems unlikely in the light of subsequent experiments in which channel rectification is seen in the presence of similar EGTA concentrations and Mg2+ (Fig. 1c and Tyerman et al. 1995). We can exclude the possibility that the increase in currents was due to a break down of the seal upon removal of Ca2+ since the resulting instantaneous currents show the same ion selectivity as the time-dependent currents (Fig. 4). Furthermore, the instantaneous currents are blocked by Ca2+ (Fig. 2c) and verapamil (Fig. 4a).

The steady state Ca2+ concentration in the cytoplasm of many plant cells is approximately 100 nm (Bush 1995). It is therefore unlikely that Ca2+ would cause rectification of the NH4+ current in vivo. Given that Mg2+ has been shown to cause rectification in animal inward rectifiers, the effect of Mg2+ at physiological concentrations (2 mm) was tested. Detached patches were formed using a pipette filled with standard patching solution in the tip and bath solution in the backfill. After an approximately 30 min delay to allow for diffusion of the backfill into the tip, test voltage pulses were applied. The NH4+ currents recorded were similar to those seen in the presence of internal Ca2+ (Fig. 1c,d). The currents were inwardly rectified and showed time-dependent activation and deactivation. These results suggest that Mg2+ on the cytoplasmic side of the membrane can cause channel rectification. The instantaneous currents seen in Fig. 1(d) were larger as a proportion of the steady state current than those in Fig. 1(b) with Ca2+ on the cytoplasmic side. This was probably due to a larger number of NH4+ channels being open at the holding potential of 0 mV since the reversal potential was positive to zero when the K+ concentration in the pipette was less than the NH4+ concentration in the bath.

The effect of applying Ca2+ ions to the bacteroid side of the membrane was also tested. Currents recorded in symmetrical 150 mm K+ concentrations containing submicromollar concentrations of divalent cations were ohmic and showed very little time dependence (Fig. 2b). When 10 mm Ca2+ was added to the bath the current became outwardly rectified and current activation and deactivation developed a time-dependent component (Fig. 2c). The magnitude of the outward steady state currents appeared to be slightly reduced until the membrane potential was taken more positive (Fig. 2d). These results indicate that Ca2+ had a similar effect on the channel when applied to either side of the membrane; it caused rectification by inhibition of the current from the side of the channel exposed to Ca2+.

Adding Ca2+ to the bacteroid side of the membrane when Ca2+ was also present in the pipette inhibited both inward and outward current (Fig. 3a). A dose response curve for the effect of Ca2+ at different membrane potentials (Fig. 3b) revealed that Ca2+ inhibition was not voltage dependent. The Kd in this example was 8.34 μm, Hill slope = 0.76.

The effect of verapamil on the channel

Because no single channel records could be obtained for the NH4+ channel, it was important to distinguish the zero current level from leak current through the seal. We therefore examined some channel blockers and found that verapamil was very potent (e.g. Figure 4a). The addition of verapamil had a similar effect to the addition of Ca2+ to the bacteroid side of the membrane (shown in Fig. 3) when calcium was also present on the cytoplasmic side. Under the same conditions, the addition of verapamil to the bacteroid side of the membrane also caused the inhibition of both inward and outward NH4+ currents (data not shown). To obtain a dose response curve, increasing concentrations of verapamil were added to a bath solution (bacteroid side of the membrane) containing 150 mm NH4+. The pipette contained the standard pipette solution. The dose response curve for verapamil is shown in Fig. 4(c), with an inhibitory constant Kd = 2.62 μm, Hill slope = 3.00.

Behaviour of the channel in low divalent concentrations

When submicromolar concentrations of divalents were present on either side of the membrane, it appeared as if the channel lost voltage dependence so that the open probability was constant with voltage. (Fig. 2b,d and Fig. 4). With K+ on both sides of the membrane at equal concentration (20 mm) the channel showed constant conductance (Fig. 4a). The base-line conductance was very low as indicated by the verapamil blocked currents. As the K+ concentration in the bath was increased to 60 mm and 150 mm the channel showed greater conductance for inward current flow, but approached similar conductance for outward current flow. This intrinsic rectification is entirely predicted by the Goldman-Hodgkin-Katz current equation using a constant K+ permeability and assuming insignificant Cl permeability (fitted solid lines). When NH4+ was the cation used in the bath at various concentrations and the pipette contained 150 mm KCl, the GHK equation could not fit the entire family of curves using a constant NH4+ to K+ permeability ratio (Fig. 4b). Instead, the PNH4+:PK+ ratio had to be increased for lower NH4+ concentrations.

Voltage dependence of current activation in the presence of divalents

Current activation curves for the NH4+ channel were recorded with either Ca2+ or Mg2+ on the cytoplasmic side of the membrane (Fig. 5a,b, respectively). Current activation as a function of voltage was recorded with 150 mm NH4+ on the bacteroid side of the membrane for both divalent cations. Curves were also calculated for decreasing concentrations of NH4+ on the bacteroid side. A Boltzmann function was used to fit the relative conductance (grel) as a function of voltage:

image

Figure 5. Voltage dependence of activation.

Steady state activation curves for ammonium currents obtained from tail currents (as outlined in Hoshi 1995).

(a) Channel conductance as a function of voltage with standard pipette solution. The bath solution contained NH4+ at either 6 mm (♦), 20 mm (▴), 60 mm (•) and 150 mm (▪). Data points represent the mean ± SEM of four separate patches. Lines represent data fitting with a simple Boltzmann function (see text) with slope factors –34.53 ± 2.404, –38.05 ± 1.708, –26.58 ± 1.266 and –26.78 ± 1.437 mV for 6, 20, 60 and 150 mm NH4+, respectively.

(b) Channel conductance as a function of voltage with a pipette solution comprising the standard bath solution. The bath solution contained 150 NH4+. Data points represent the mean ± SEM of three separate patches. Line represents data fitting with a simple Botzmann function with a slope factor of –29.63 ± 2.124 mV.

(c) and (d) are semi-logarithmic plots of the data shown in (a) 150 NH4+ and (b). Lines represent data fitting with a simple Boltzmann function (i.e. n = 1; dashed line) and a Boltzmann function in which n is free running (solid line).

Download figure to PowerPoint

where V50 is the voltage at which the channels have attained half their maximal open probability (when n = 1), S is the slope factor (RT/zδF where R,T and F have their usual values, z represents the valence of the gating charge and δ is the distance moved by the gating charge across the voltage field). The activation curves in 150 mm NH4+ on the bacteroid side were very similar for pipette solutions containing either 10 mm Ca2+ or 2 mm Mg2+. The value of V50 was –62.8 ± 1.65 mV with Ca2+ and –60.2 ± 2.32 mV with Mg2+ and currents activated at about + 20 mV and saturated at about –180 mV. The zδ value calculated from the slope factor in Ca2+ was 0.959 and in Mg2+ was 0.867. Activation curves constructed for currents recorded at different NH4+ concentrations on the bacteroid side showed that V50 became more positive as NH4+ concentration increased (Fig. 5a). However, when the NH4+ concentration on the bacteroid side was increased from 60 to 150 mm NH4+, there was no change in V50. The z-values calculated from the slope factors in 60 mm, 20 mm and 6 mm external NH4+ were 0.966, 0.676 and 0.744, respectively.

Previous authors have found that activation curves were best fit with a Boltzmann equation where n > 1. For example, Hoshi (1995) found that the activation of KAT1 is best fit by a Boltzmann equation with n = 3.3 and suggested this may indicate that the channel is composed of four functionally independent subunits. In order to test the goodness of fit of the data with a simple Boltzmann equation, the data were redrawn on a semi-logarithmic plot and fitted with a Boltzmann equation with a free running value of n (Fig. 7c,d show activation curves for 150 mm external NH4+ concentration). In all cases it was found that the best fit value of n was not significantly different to 1.

Kinetics of channel activation and deactivation

When Ca2+ or Mg2+ was present on one side of the membrane, current activation and deactivation showed time dependence (Figs 6 and 7). We have analysed the kinetics of activation and deactivation of NH4+ currents when divalent cations are present on the cytoplasmic side.

image

Figure 6. Kinetics of time dependent activation of ammonium currents.

Measurements were performed in a bath solution containing 150 mm NH4+.

(a) Currents recorded in response to three voltage steps. The pipette contained the standard pipette solution. Lines represent data fitting with the Hodgkin-Huxley equation (see text) with a fixed value of P = 2.

(b) Currents recorded in response to three voltage steps. The pipette contained the standard bath solution. Lines represent data fitting with a single exponential equation (Y = A*exp(–t/τ) + B).

(c) Activation time constants (τ) derived from curve fitting as shown in (a) and (b) as a function of voltage. Data points represent the mean ± SEM of 13 (data as in a) or seven patches (data as in b). Lines represent data fitting with linear regression analysis. Slope of calcium data = 2.694 ± 0.440 ms mV–1, slope of magnesium data = 6.178 ± 0.786 ms mV–1.

Download figure to PowerPoint

The activation of inward NH4+ current as a function of time with Ca2+ on the cytoplasmic side of the membrane had a sigmoidal time course (Fig. 6a). The inward current normally reached a steady state level within about 1.5 sec of hyperpolarisation. There was no inactivation of this current during prolonged hyperpolarisation over several minutes. Current as a function of time was fitted to the Hodgkin-Huxley equation (Hodgkin & Huxley 1952;Schroeder 1989):

  • image

where I0 is the initial current, Imax is the steady state activated current, τ is the activation time constant and p as defined by Hodgkin and Huxley is the number of independent membrane-bound gating particles which control the opening of the channels. To determine the value of p for the NH4+ channel using the Hodgkin-Huxley analysis, currents elicited by steps from the holding potential (0 mV) to a pulse from 0 to –120 mV were fitted by eqn 2 with free running parameters I0, Imax, τ and p. The mean (± SEM) value of p obtained from 13 patches was 2.25 ± 0.209 and current activation at all voltages was well fit by eqn 2 for P = 2 (e.g. Figure 6a). The value of p was therefore fixed at 2 for all further analysis.

The activation of inward NH4+ currents had different kinetics when 2 mm Mg2+ was present on the cytoplasmic side of the membrane. The kinetics of current activation were analysed as above. The mean (± SEM) value of p obtained from seven patches was 0.860 ± 0.086. Thus, current activation could be best fit by a single exponential equation (Fig. 6b). Further analysis was conducted by fitting activation curves with a single exponential equation (fitting with a double exponential equation did not give better fits).

The activation time constant τ was determined as a function of the membrane potential for NH4+ currents in the presence of Ca2+ or Mg2+ on the cytoplasmic side of the membrane. Figure 6(c) shows that τ is weakly dependent on membrane potential, τ-values decreased with increasing hyperpolarisation. There was large variability in τ between patches. Linear regression analysis showed that the slope was non-zero (P < 0.01) for both data sets. However, there was a significantly stronger voltage dependence of τ in the presence of cytoplasmic Mg2+ than Ca2+ (P < 0.01).

Deactivation of NH4+ currents had an exponential time course with either Ca2+ or Mg2+ on the cytoplasmic side of the membrane (Fig. 7). Tail currents were recorded upon stepping back to a more positive membrane potential (+ 60 mV) from a strongly negative potential (–120 mV). The tail currents recorded at one voltage with either Ca2+ (n = 5) or Mg2+ (n = 3) in the pipette were best fit by a double exponential equation of the form

  • [INSERT EQUATION]

The voltage dependence of deactivation time constants was determined for patches with Ca2+ on the cytoplasmic side of the membrane. Tail currents were elicited by steps from a prepulse potential of –160 mV to tail potentials of –120 to + 60 mV and fitted with a double exponential equation. There was a weak voltage dependence of the slow time constants (Fig. 7e).

Discussion

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Conclusion
  7. Experimental procedures
  8. Acknowledgements
  9. References

Channel gating

It is clear that divalent cations have a primary role in gating of the NH4+ permeable channel in the symbiosome membrane. There appears to be no intrinsic voltage gate since with the removal of divalents from both sides of the membrane the channels behaved as if open probability was constant with voltage. Since we could not observe single channel events we cannot determine the fixed level of open probability. This will require more extensive noise analysis than that carried out previously (Tyerman et al. 1995). Inward rectification of the NH4+ channel can occur with physiological Mg2+ concentrations on the cytoplasmic side of the membrane. To the best of our knowledge, this is the first time that rectification of a plant monovalent cation channel has been attributed to cytoplasmic Mg2+. Other plant channels examined to date appear to be rectified by an intrinsic voltage-dependent mechanism that is unaffected by cytoplasmic ions such as Mg2+ or polyamines (Hoshi 1995;Müller-Röber et al. 1995;Schroeder 1995). Conversely, the inward rectification of many K+ channels from animals has been shown to be at least partially dependent on cytoplasmic Mg2+ (Matsuda 1991). Many of these channels have an underlying intrinsic time- and voltage-dependent rectification in the absence of cytoplasmic cations (Fakler et al. 1995;Matsuda 1988). The addition of divalent cations appears to reduce activation times and in some cases cause rectification of the instantaneous current (Ishihara et al. 1989). While it is possible that rectification of the NH4+ channel involves more than one factor, the work presented here shows that in the absence of multivalent cations the channel does not rectify and that Mg2+ is sufficient to strongly rectify the currents.

The sensitivity of channel gating to NH4+ concentration inside the symbiosome is likely to be important for channel function in vivo (see below). Similar sensitivity to permeant ion concentration has been seen in the gating of the potassium outward rectifier in Vicia guard cells (Blatt & Gradmann 1997). Interestingly, Roberts & Tester (1995) report no such sensitivity in the potassium inward rectifier in maize roots. The significance of channel activation and deactivation kinetics of the NH4+ channel is discussed later. However, both sigmoid and exponential activation time courses have been identified in other plant univalent cation channels (e.g. Fairley-Grenot & Assmann 1993;Van Duijn 1993) although the significance of these kinetics is not generally discussed. The voltage dependence of these time courses seems to be very variable. In most cases, the kinetics of deactivation appear to be best fit by a single exponential rather than a double exponential.

Channel pore characteristics

The current-voltage curves without divalent ions present could be fitted by the simple GHK current equation, albeit with a concentration dependent permeability ratio when different cations were present on either side of the membrane. The fact that the curvature of the current-voltage curves is well described by the GHK current equation indicates that the channel has a relatively symmetrical energy barrier profile and that interaction between ions in the pore is insignificant, at least for K+ and NH4+. However, the saturation of current with increasing cation concentration (Tyerman et al. 1995) and the concentration dependence of PNH4+:PK+ suggests that more complex interactions can occur between the permeating ions and channel. This concentration dependence of PNH4+:PK+ confirms the observation by Tyerman et al. (1995) who demonstrated that the channel becomes more selective for NH4+ at low NH4+ concentrations (20 mm) on the bacteroid side of the membrane. Here we show that at even lower NH4+ concentrations (2 mm) the permeability ratio can increase to nearly 8. Since the current-voltage curves could be well fitted by the GHK current equation, the increase in permeability ratio is reflected by a corresponding increase in NH4+ flux relative to the driving force. The results support the view that the channel functions as a pathway for NH4+ release from the symbiosome in vivo but does not preclude the possibility that other pathways may exist for fixed nitrogen release.

The discovery of verapamil as a blocker of the channel allows a more accurate estimation of the base-line current for the symbiosome membrane patches. This is especially important when voltage- and time-dependence are abolished by removal of divalent cations. The block by verapamil, if specific for the NH4+ channel, will allow analysis of other transporters likely to exist in the symbiosome membrane (Udvardi & Day 1997;Weaver et al. 1994) and derivatives of verapamil may be useful for isolation of the protein (Thuleau et al. 1993). Verapamil is generally used as a Ca2+ channel blocker, and a Ca2+ channel from plant plasma membrane has a reported Kd of 2 μM (Pineros & Tester 1997) which is similar to that observed here for the NH4+ channel. Verapamil inhibited both inward and outward current when presented to the bacteroid side of the membrane. This is similar to the action of verapamil on l-type Ca2+ channels from animals (Lee & Tsien 1983). However, verapamil has also been shown to inhibit outwardly rectifying K+ channels in the micromolar range (Roelfsema & Prins 1997;Terry et al. 1992;Thomine et al. 1994). The sensitivity to verapamil may suggest that the NH4+ channel could be permeable to Ca2+. However, this seems unlikely given that application of Ca2+ to both sides of the channel blocked currents to a similar degree as did verapamil. The reversal potential of the time-dependent current also did not alter with changes in Ca2+ gradient across the membrane (results not shown). It is more likely that verapamil is interacting with the divalent binding sites which gate the channel.

Possible schemes for divalent dependent channel gating

Kuo & Hess (1993) discuss two possible mechanisms of channel gating in relation to Ca2+ effects on univalent ion permeation through l-type Ca2+ channels. The authors describe an allosteric and block model which can be applied to divalent effects on the NH4+ channel described in this paper. The allosteric model involves divalent cations binding to the channel in a manner that causes the channel to become voltage gated. In this model, the divalent binding sites are independent of the univalent cation pathway through the pore. Although an allosteric model can account for many of the characteristics of the NH4+ channel, the explanations tend to be more complex than those of a block model. We have therefore used the block model as the basis for the scheme for channel gating presented here.

A block model was first proposed by Armstrong (1969) for non-triethyammonium ion block of the delayed outward rectifier in the squid axon. Using this model, gating of the NH4+ channel would involve divalent cations blocking univalent cation current when the electrochemical gradient favours cation flow from the side of the membrane containing the divalent cation. Nichols & Lopatin (1997) suggested that the permeant and blocking ions may occupy similar sites within the pore. According to Armstrong’s model univalent cation entry from the other side is sufficient to unblock the channel through interaction with the bound divalent ion. For example, when Ca2+ or Mg2+ are present on the cytoplasmic side of the membrane, deactivation of the outward current observed when the membrane is depolarised results from divalent cations progressively blocking the channels (e.g. Figure 7). NH4+ entry to the channels initiated by hyperpolarising the membrane would unblock the channels and results in a time-dependent increase in inward current across the PBM (e.g. Figure 2). It should be noted that the NH4+ channels have sub-picoSiemen conductances so that the patch-current shows more resemblance to a whole cell or whole vacuole current. Blocking sites would need to be present at both ends of the pore to account for the qualitatively symmetrical effect of divalents. This would also explain our observation that both inward and outward currents are blocked upon the addition of Ca2+ to the PBS side when Ca2+ is already present on the cytoplasmic side. The voltage independence of Ca2+ block suggests that the binding site on the bacteroid side is at the very edge of the voltage profile. A Hill slope of 1 also indicates that there is just one blocking site on the bacteroid side. This is similar to the Ca2+ block observed for the FV channel in barley mesophyll vacuoles (Tikhonova et al. 1997).

The voltage and concentration dependence of current activation (Fig. 5) may also be explained using this model. According to Armstrong (1975), the probability that the univalent cations displace the divalent cations from the pore will depend on the voltage gradient and the concentration of univalent cation. The fact that the V50 was the same in 60 and 150 mm NH4+ could be explained by the saturation of current that is often observed at these NH4+ concentrations (Km = 38 mm, Tyerman et al. 1995). Activation curves could be fit to a simple Boltzmann equation with zδ equal to 1. In a very simplistic interpretation, this may indicate that channel gating involves one univalent cation traversing the full membrane potential gradient to knock off the divalent blocking ion.

The kinetics of channel activation and deactivation under the block model require more explanation. There are several factors to consider: (i) the sigmoidal kinetics (P = 2) observed with Ca2+ and not with Mg2+; (ii) the double exponential kinetics of deactivation; (iii) the slight voltage dependence of the time constants; and (iv) the long time constants for activation and deactivation.

The sigmoidal kinetics observed with Ca2+ but not with Mg2+ may reflect the difference in concentration used in these experiments (10 mm Ca2+, 2 mm Mg2+). The channel may have two binding sites on the cytoplasmic side which are both only occupied at higher divalent concentrations. Channel opening would therefore involve two unblocking steps instead of one. The double exponential kinetics of deactivation (regardless of cytoplasmic divalent) would suggest that the channel can exist in two kinetically distinct open states from which the channel can be blocked by Ca2+ or Mg2+. The time constants for activation and deactivation were generally only weakly voltage dependent. This supports the view expressed above that the blocking sites for divalents are probably positioned at the edge of the voltage profile.

The rates of activation and deactivation may seem rather slow for a simple unblocking and blocking reaction. For example, the Mg2+ blocking rate in the animal K+ inward rectifiers is very rapid (< 50 ms) compared to those for the NH4+ channel (> 500 ms). However, the rate constant for NH4+ channel block by Ca2+ can be calculated to be between 105 and 106 mol–1 s–1 (using the Kd for Ca2+ block on the bacteroid side and taking the unblocking rate constant to be approximately 1/τact (between 2 and 20 sec–1);Hille 1992). This is not an unusual estimate compared to literature values of other blockers (Hille 1992;Lopatin et al. 1995). It should be possible in future experiments to test the effect of varied Ca2+ concentrations which, according to the block model, should affect the time constant of deactivation but not the mean closed time of the channel.

Conclusion

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Conclusion
  7. Experimental procedures
  8. Acknowledgements
  9. References

The results presented here are consistent with a role for the symbiosome channel in transporting NH4+ from N2-fixing bacteroids to the plant in legume nodules. Under in vivo conditions it is likely that Mg2+ in the cytoplasm will function to gate the channel. The proton pump would hyperpolarise the membrane, drive NH4+ through the channel and displace the Mg2+ thereby allowing NH4+ to enter the cytoplasm for assimilation. The high sink strength of ammonium-assimilating enzymes in the plant cytoplasm would provide further directionality for ion flux. Channel activation characteristics also help to understand the significance of rectification under conditions in the nodule. The negative shift in activation potential at low NH4+ concentrations combined with the increased permeability to NH4+ relative to K+ would prevent K+ flux into the symbiosome at times when fixed nitrogen availability was low. Ca2+ on the bacteroid side of the membrane could act to fine tune the flux of NH4+ to the cytoplasm and may be under the control of either the bacteroids which can store Ca2+, or Ca2+–permeable channels that may also occur in the membrane.

Experimental procedures

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Conclusion
  7. Experimental procedures
  8. Acknowledgements
  9. References

Plant material and symbiosome isolation

Soybean, Glycine max (L. cv. Stephens or Forest) were planted in pots filled with a mixture of vermiculite and pearlite and inoculated with Bradyrhizobium japonicum USDA 110. Plants were grown in a naturally illuminated greenhouse as described previously in Day et al. (1989) with supplementary light to ensure a 16 h light/8 h dark cycle. Symbiosomes were isolated from the root nodules (2–3 g) of 5–6-week-old plants by the method of Day et al. (1989). After isolation the symbiosomes were resuspended in buffered medium (350 mm mannitol, 25 mm MES-KOH pH 7.0, 3 mm MgSO4) and kept on ice.

Experimental solutions

The standard bath solution comprised 100 mm Kglutamate, 2 mm MgCl2, 2.3 mm CaCl2, 10 mm EGTA, 5 mm HEPES/Tris (pH 7.0). Osmolalities of this and all other solutions were adjusted to 400 mOsM with mannitol. Patch clamp microelectrodes were filled with 150 mm KCl, 10 mm CaCl2, 5 mm HEPES/KOH (pH 7.2) unless stated otherwise. When alternative pipette solutions were needed, the pipette was backfilled with the new solution before patching. The pipette tip was always initially filled with the standard pipette solution containing a high concentration of calcium in order to achieve good membrane seals. All test bath solutions consisted of 5 mm HEPES/Tris (pH 7.0) with the test cation as a Cl salt. The basic solution for measurements of calcium-dependence comprised 20 mm NH4+ and 2 mm EGTA. Free calcium concentrations in the range 100–107 mm were adjusted by variable additions of CaCl2 to the basic solution. Free calcium concentrations were calculated using the chemical speciation program GEOCHEM (Parker et al. 1987).

Patch-clamp measurements and data analysis

Isolated symbiosomes were placed in the bathing chamber in the standard bath solution. Symbiosomes adhered to the base of the chamber and could be viewed at ×800 magnification. Patch pipettes were made from GC150–10 borosillicate glass capillaries (Clark Electronic Instruments, Reading, UK) using a two-step pulling protocol (electrode puller PP-83, Narishige, Tokyo, Japan). The tip diameters of patch pipettes after fire polishing were 0.5–1 μm. Pipettes filled with the standard solution had resistances of about 40 mΩ. Pipette tips were coated with Sylgard (184 silicone elastomer kit, Dow Corning Co., Midland, MICH).

After the formation of a high resistance seal (10–100 GΩ) of the patch pipette with the symbiosome membrane, the pipette was withdrawn to excise a patch. Bacteroids are released from the symbiosome during patch excision, indicating that an inside out (inside of the symbiosome facing the bath) patch has been formed.

The reference AgCl electrode was connected to the bath via a 1.5% agar bridge filled with 100 mm KCl. Current measurements were made with a List EPC 7 (List Electronics, Darmstadt, Germany), a Dagan 3900 A (Dagan Corporation, Minneapolis, MN, USA) and Axopatch 200 A (Axon Instruments, Foster City, CA, USA) patch amplifiers. Records were filtered at 200 Hz by a low-pass Bessel filter and stored using a Strobes (Strobes Engineering) or P-Clamp6 (Tl–1 DMA Interface digitiser, Axon Instruments, Foster City, CA, USA) analysis acquisition systems. In this work, all bath and electrode solutions contained mainly KCl or NH4Cl, so liquid junction potentials, calculated using the program JPCalc (P.H. Barry, University of New South Wales, Sydney, Australia), were negligible. The sign convention for current and voltage is according to Bertl et al. (1992), i.e. the sign of voltage refers to the cytosolic side, and positive (outward) currents represent efflux of cations into the symbiosome. Data fitting was undertaken with pClamp 6.0 (Axon Instruments, Foster City, CA, USA) and Graphpad Prism (Graphpad Software Inc., San Diego, CA, USA) software.

Analysis of current activation

Relative conductance as a function of voltage was obtained from a double pulse protocol and measurement of tail current amplitude as described by Hoshi (1995). Tail currents were measured at +40 mV after increasingly negative voltage pulses of 4–8 sec in duration (ensuring that currents had reached steady state). Voltage steps were continued until the tail current amplitude at +40 mV had saturated. Tail currents were fitted with a double exponential equation so that the initial current could be extrapolated. These initial tail current amplitudes were normalised relative to the maximum tail current and the relative amplitudes were used as an approximation of the relative conductance (grel) at the preceding negative voltage.

Acknowledgements

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Conclusion
  7. Experimental procedures
  8. Acknowledgements
  9. References

Thanks to Clifford Slayman for alerting us, in 1995, to the possibility of divalent gating. Thanks also to Dianne Trussell and Wendy Sullivan for expert technical assistance. This work was funded by Australian Research Council grants to S.D.T. and D.A.D.

References

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Conclusion
  7. Experimental procedures
  8. Acknowledgements
  9. References