NH₄⁺ Currents across the Peribacteroid Membrane of Soybean. Macroscopic and Microscopic Properties, Inhibition by Mg²⁺, and Temperature Dependence Indicate a SubpicoSiemens Channel Finely Regulated by Divalent Cations

Abstract

The control of ammonium (NH₄⁺) transport is critical in preventing futile cycles of NH₄⁺/ammonia transport. An unusual nonselective cation channel with subpicoSiemens single-channel conductance permeable to NH₄⁺ had previously been identified in the peribacteroid membrane (PBM) of symbiosomes from soybean (Glycine max) nodules. Here, we investigate the proposed channel mechanism and its control by luminal magnesium. Currents carried by NH₄⁺ were measured in inside-out PBM patches by patch clamp. NH₄⁺ transport corresponding to the physiological direction of net transfer showed time-dependent activation and associated single-channel-like events. These could not be resolved to discrete conductances but had the same selectivity as the total current. The voltage dependence of the steady-state current was affected by temperature consistent with the rate constant of channel opening being reduced with decreased temperature. This resulted in steady-state currents that were more temperature sensitive at voltages where the current was only partially activated. When fully activated, the current reflected more the ion conduction through open channels and had an activation energy of 28.2 kJ mol⁻¹ (Q10 = 1.51, 8°C–24°C). Increased Mg²⁺ on the symbiosome lumen side blocked the current (ID₅₀ = 351 μm, with 60 mm NH₄⁺). Complete inhibition with 2 mm Mg²⁺ was relieved with a small increase in NH₄⁺ on the lumen side of the membrane (shift of 60–70 mm). With Mg²⁺ the selectivity of the transport for divalent cations increased. From these features, we propose a divalent-dependent feedback regulation of the PBM-nonselective cation channel that could maintain a constant NH₄⁺ gradient across the membrane.

The transport of ammonium (NH₄⁺)/ammonia (NH₃) has important roles in nitrogen assimilation in plants, algae, and bacteria, and for pH balance in animals. The control of NH₄⁺ transport is critical to prevent futile cycles that may dissipate pH gradients, waste cell energy, and could account for NH₄⁺ toxicity (Britto et al., 2001a). Recently, the structure of the AmtB NH₄⁺ transporter from Escherichia coli has indicated that the protein functions as a channel transporting NH₃ after the NH₄⁺ is deprotonated in the periplasmic vestibule of the channel at a site that effectively reduces its pKa, and then returning to equilibrium as NH₄⁺ through protonation of NH₃ in the cytoplasmic vestibule (Khademi et al., 2004). This transport would not be electrogenic and also does not require conformational changes during the transport cycle as indicated from the structural study. The human rhesus glycoproteins that share conserved sequences with the Amt family show electroneutral NH₄⁺/H⁺ exchange characteristics (Ludewig, 2004) and NH₃ stimulated NH₄⁺ transport (Bakouh et al., 2004) when expressed in Xenopus laevis oocytes. Aquaporins have also been implicated in NH₃ transport (Niemietz and Tyerman, 2000; Jahn et al., 2004; Loqué et al., 2005) that would function as a low affinity version of the AmtB mechanism. Increased NH₃ permeability facilitated by aquaporins would increase the rate of passive NH₃ flux into an endomembrane-bound acidic compartment provided there is an appropriate gradient (e.g. if cytoplasmic [NH₄⁺] >10 mm; see Britto et al., 2001b), and NH₄⁺ would be trapped in the acid compartment (Loqué et al., 2005). However, NH₄⁺ could move passively back from the acid compartment (Britto et al., 2001b), say through a nonselective cation channel (NSCC), causing a futile cycling of nitrogen and dissipation of the pH gradient. It is thus essential that tight control should occur over the fluxes of both NH₄⁺ and NH₃ across membranes.

In legumes, a membrane that has a high density of NH₄⁺/NH₃ transport that surrounds an acidic compartment is the peribacteroid membrane (PBM; Day et al., 2001). Soil bacteria of the various species of Rhizobium infect roots and induce the formation of nodules. Within certain cells of the nodules, the N₂-fixing bacteria differentiate into bacteroids that are surrounded by the plant-derived PBM forming an organelle-like structure, the symbiosome. The PBM controls the nutrient exchange between the bacteroids and the host cell that is mainly fixed nitrogen in the form of NH₃ or NH₄⁺ from the bacteroids and reduced carbon from the plant (Udvardi and Day, 1997; Day et al., 2001). The transport of NH₃ and NH₄⁺ across the PBM from the symbiosome lumen of legumes has been measured by various techniques. The results to date indicate that there are at least three pathways for reduced nitrogen transport: (1) amino acid transport (permeases and H⁺ symport; Day et al., 2001; Rosendahl et al., 2001); (2) channel-mediated NH₃ pathway sensitive to mercurials (Niemietz and Tyerman, 2000); and (3) a passive electrogenic transport system permeable to univalent cations but with NH₄⁺ having the highest permeability (Tyerman et al., 1995). The latter transport has been proposed to be via a channel and will be referred to as PBM-NSCC.

The PBM-NSCC is novel among NSCC in plants (Demidchik et al., 2002). First, the voltage-dependent properties and inward rectification of the PBM-NSCC require the presence of Mg²⁺ on the cytosolic face of the PBM (Whitehead et al., 1998; Roberts and Tyerman, 2002) and are very sensitive to polyamines (Whitehead et al., 2001), similar to animal inwardly rectifying K⁺ channels (e.g. Doupnik et al., 1995; Soh and Park, 2001). Thus, the channel appears to belong to the class of rectifiers that require the association of diffusible charged gating particles, such as divalent cations or polyamines (for review, see Schroeder et al., 1994; Doupnik et al., 1995; Reimann and Ashcroft, 1999; Oliver et al., 2000; Schachtman, 2000). Most higher plant rectifying cation channels that have been characterized have an intrinsic gate (Schachtman, 2000), although for the nonselective slow vacuolar (SV) and fast vacuolar (FV) channels voltage dependence is strongly modified by Mg²⁺ and polyamines (Tikhonova et al., 1997; Brüggemann et al., 1999; Pottosin et al., 2004). The PBM-NSCC will allow NH₄⁺ transport from symbiosome lumen to the plant cell cytoplasm, where cytosolic Mg²⁺ is in the millimolar or submillimolar range (Brüggemann et al., 1999), but the rectification will prevent back flux to the symbiosome lumen if the NH₄⁺ gradient is reversed. This control may depend on the concentration of divalent cations in the symbiosome lumen. In this respect, previous studies have shown that Ca²⁺ on the symbiosome lumen face of the membrane will inhibit PBM-NSCC (soybean [Glycine max] K_i = 22 μm, Lotus japonicus K_i = 317 μm; Whitehead et al., 1998; Roberts and Tyerman, 2002), but no investigations of the effect of luminal Mg²⁺ have been reported. Luminal Mg²⁺ is known to regulate the SV and FV channels in the tonoplast (Brüggemann et al., 1999; Pottosin et al., 2004).

Another novel feature of the PBM-NSCC is its very low apparent single-channel conductance and very high density in the membrane (Tyerman et al., 1995). So far, no single-channel events associated with the PBM-NSCC have been reported, but noise analysis indicated that the PBM-NSCC could be modeled as many subpicoSiemens ion channels (Tyerman et al., 1995). In contrast to the normal observation of step-like single-channel events observed with the patch-clamp technique, smooth time- and voltage-dependent currents are the normal manifestation of the PBM-NSCC (Tyerman et al., 1995; Whitehead et al., 1998; Whitehead et al., 2001; Roberts and Tyerman, 2002). From protocols normally used for whole-cell voltage clamps, the selectivity of the time-dependent current across single-membrane patches showed that it is passive and relatively nonselective between monovalent cations (Tyerman et al., 1995; Whitehead et al., 1998). In L. japonicus, the PBM-NSCC is also permeable to Ca²⁺ (Roberts and Tyerman, 2002). In both soybean and L. japonicus, the PBM-NSCC conductance near −100 mV saturates with a K_m for NH₄⁺ of 37.5 mm (soybean) and 17 mm (L. japonicus) similar to an estimate of lumen NH₄⁺ concentration (Streeter, 1989).

In summary, the properties of the PBM-NSCC that stand out as being unusual for an ion channel-mediated transport in plant membranes are (1) inward rectification caused by Mg²⁺ on the cytosolic face; (2) the very low open-channel conductance (0.11 pS); (3) saturation of the conductance with increasing NH₄⁺ concentration; and (4) stationary noise spectra (power proportional to inverse of the frequency) that is typical for carriers, or ion channels with high cooperativity or a high number of kinetic states (Tyerman et al., 1995).

In view of the emerging novel mechanisms for NH₄⁺ transport (Khademi et al., 2004) and the importance of NH₄⁺ flux regulation (Britto et al., 2001a, 2001b), we require more information about the nature of the PBM-NSCC that accounts for the NH₄⁺ currents in PBM and its control by divalent cations. Here, we analyzed the PBM-NSCC currents carried by NH₄⁺ in respect of their Mg²⁺ sensitivity on the symbiosome lumen face of the PBM, and their temperature dependence to determine the molecular nature of ion conduction and gating, because previous studies have shown large differences between the temperature dependence of carrier- and channel-mediated transport (Table I). Attempts to identify the gene(s) encoding for PBM-NSCCs will need to take into account its unusual functional characteristics, including its apparent high density on the membrane, when using heterologous expression techniques. The characteristics of the PBM-NSCCs might also confirm the identity of the genes encoding them.

Table I.

Temperature dependence of transport through channels and carriers

Values are selected from the literature and focus on channels and carriers with some unusual characteristics including low conductance (channels), or carriers that can show channel activity.

Channel or Carrier Type	Channel Conductance	EA or Q10 for Transport (Carriers) or Open-Channel Conductance (Channels)	Temperature Dependence of Kinetics	Reference
		EA (kJ mol−1) or Q10	EA (kJ mol−1) or Q10
PBM-NSCC (soybean)	0.16 pS (150 mm NH4+)	EA = 28.2, Q10 = 1.51	Activation EA = 64.2, Q10 = 2.7	This study
			Deactivation EA = 2–6, Q10 = 1.1
ACh receptor channels	Approximately 60 pS at 35°C	EA = 23, Q10 = 1.4	Mean open time EA = 93	Dilger et al. (1991)
Miniature endplate currents, ACh channels			Channel opening EA = 46 Channel closing EA = 64	Stiles et al. (1999)
Shaker K+ channels	10 pS from nonstationary noise analysis (17°C–18°C)	Q10 = 1.44	Gating currents: Early Q10 <2, Late Q10 = 3	Rodríguez et al. (1998)
L-type Ca2+ channels	7 pS	Q10 = 1.2	Activation time const. EA = 78.2	Lux and Brown (1984); Acerbo and Nobile (1994)
	23 pS (110 mm Ba2+)	EA = 23, Q10 =1.4
Voltage-dependent Ca2+ channels, dihydropyridine insensitive	13 pS (110 mm Ba2+)	EA = 18.4, Q10 = 1.3	Activation EA = 146.4, Q10 = 5.9 Inactivation EA = 62.8, Q10 = 2.0	Acerbo and Nobile (1994)
Cystic fibrosis transmembrane conductance regulator channel			Activation by ATP; EA = 100 channel closing; EA = 10	Aleksandrov and Riordan (1998)
Voltage-gated H+ currents (channels)	38 fS at pH 6.5 (current noise, variance analysis)	EA = 52.3–78.2, Q10 = 2.05–3.06	Activation and deactivation, EA = 125–151	Cherny et al. (2003); DeCoursey and Cherny (1998)
Gramicidin A channels	650 pS (1 M HCl) at 19.3°C	EA = 15.3–15.6 (H+)		Chernyshev and Cukierman (2002)
H+ conductance		25 (K+, at T < 27°C)
Ca2+ activated K+ channel, Chara tonoplast	148 pS (150 mm K+)	EA = 13.6 (at −50 mV) 8 (at 50 mV)		Djurišić and Andjus (2000)
Arabidopsis cell suspension	Whole-cell currents	Inward K: EA = 88 at −180 mV		Colombo and Cerana (1993)
		Outward K: EA = 64 at 100 mV
Chara water channels	Hydraulic conductivity	EA =16.8, 21		Hertel and Steudle (1997); Schütz and Tyerman (1997)
Dicarboxuylate transporter on mitochondria inner membrane		EA = 59–125		Liu et al. (1996)
NADPH oxidase, e− transport		EA = 58.9, Q10 = 2.2		Morgan et al. (2003)
Serotonin transporter		Q10 = 3.1		Beckman and Quick (2001)
γ-Aminobutyrate cotransporter (rGAT1)		Transport; Q10 = 2.2 Current; Q10 = 1.8–2.2		Binda et al. (2002)
Inositol/H+ symporter		EA = 64		Seyfang and Landfear (2000)
Glutamate transporter/chloride channel		Coupled d-Asp transport Q10 = 3.2 (at −30 mV); Cl− current Q10 = 1.0 (at 80 mV)		Wadiche and Kavanaugh (1998)

RESULTS

Channel-Like Activity Can Sometimes Be Observed Associated with the NH₄⁺ Currents

After formation of a “giga seal” at the symbiosome membrane in the cell-attached configuration, currents were recorded by clamping the voltage to different positive and negative voltages resulting in a voltage-dependent current with strong rectification properties as reported previously (Tyerman et al., 1995). After withdrawing the patch and establishing the inside-out configuration, a similar time-dependent, negative current at hyperpolarizing voltages was observed under standard conditions (standard bath or NH₄⁺ media and standard pipette medium) with high divalent cation concentrations (10 mm CaCl₂) at the former cytoplasmic side (pipette solution). Such rectification is also observed with physiological levels of Mg²⁺ in the pipette solution (Whitehead et al., 1998), but it is more difficult to achieve giga seals with a Mg²⁺-based solution. We assume in the following experiments that the behavior of the PBM-NSCCs is similar whether Ca²⁺ or Mg²⁺ are used to induce the rectification of the channels.

In approximately 90% of successful inside-out patches (n = 57), only a time-dependent activation of the inward current without any visible single-channel events was observed, although the noise on the currents was variable (Fig. 1A, trace 1). In about 10% of recordings the current had single-channel-like events superimposed (Fig. 1A, trace 2). These events did not have the characteristics of another type of channel superimposed on a rising baseline current; rather, the magnitude of the current deviations increased in proportion to the magnitude of the baseline current. For six different patches displaying this behavior and using data from repetitive pulses to the same voltage, we were able to calculate the maximum fluctuation as a function of time from the start of the current activation. Plotting this as a function of the average current revealed a linear trend. The mean slope was 0.48 (SEM = 0.13), with a mean intercept of −6.5 pA (SEM = 5.27). In other words, the time dependence of the increase in magnitude of the channel fluctuations paralleled the time dependence of the baseline current.

Figure 1.

Single-channel events in recordings of time-dependent symbiosome currents. A, Currents recorded after voltage pulses of −180 mV showing variable noise of the current. Both current traces were obtained immediately after each other in a total number of 40 traces. Arrows at trace two indicate single-channel-like events. Bath solution: 60 NH₄⁺, 0 Mg²⁺. B, Current recorded at −160 mV voltage pulses reflecting various channel-like closing levels. Bath solution: 60 NH₄⁺, 0 Mg²⁺. C, Current recordings obtained at −180 mV voltage pulse followed by a voltage ramp from −180 to 100 mV. Lowest and highest current amplitude are marked as gray traces. All current traces were from a total number of 100 and recorded from the same patch. Bath solution: 150 NH₄⁺, 0 Mg²⁺. D, Voltage dependence of the activated transporter current. The lowest and highest currents were marked as gray lines. The IV diagram was plotted with the data from similar recordings as C. The E_rev for all traces is between −12 and −15 mV.

The occurrence of these deviations in the time-dependent current were unpredictable, and so far did not depend on any pretreatments of the symbiosomes nor on any special experimental conditions. In some instances the two types of behavior (smooth and channel like) were recorded on the same symbiosome patch under the same conditions (Fig. 1, A–C). In some current traces, channel-like closing events could be noticed followed by a slow “opening” showing almost the same time-dependent activation as the macroscopic current (Fig. 1B). It seems to be very likely that channel events and the macroscopic current are closely linked as reported by Figure 1C.

A fully activated, time-dependent current at −180 mV was subject to a depolarizing voltage ramp (−180–100 mV) reflecting the voltage dependence of the activated transporters. Two limiting current amplitudes were observed, and channel-like closing and opening events between these two currents were recorded. No difference in voltage dependence of the current voltage (IV) curve or reversal potential (E_rev) was observed between the current amplitudes in Figure 1D. There appeared to be a continuum between smooth to more noisy, to more channel-like activity. The extremes of these behaviors are illustrated in Figure 1. For the channel-like extreme (Fig. 1D), it was not possible to obtain a consistent single-channel conductance when channel events were clear (Fig. 2). By subtracting the baseline time-dependent current, we then attempted to analyze the transitions using a technique that examines the transition rates between current levels (transition amplitude analysis [TRAMP]; Tyerman et al., 1992). This technique was applied to successive voltage pulses to the same voltage. The responses for a selected set of these are shown in Figure 2A (baseline subtracted). Despite the apparent randomness of the currents, some discrete levels that were common between separate pulses emerged (Fig. 2B, dashed vertical lines); however, it was not possible to distinguish the lowest single-current amplitude. Using patches that did not have obvious channel-like transitions but that had variable degrees of noise, we applied nonstationary noise analysis as used by Tyerman et al. (1995). This analysis gave a single-channel conductance with standard pipette solutions and 150 mm NH₄⁺ in the bath of 0.16 pS (95% confidence; 0.11–0.22 pS; n = 5 patches) confirming the results obtained by Tyerman et al. (1995).

Figure 2.

An example of a set of current responses to the same holding potential of −180 mV in 40 consecutive pulses. A, The time-dependent component of the current was subtracted as a baseline and plotted are the current deviations above this baseline. Two traces out of the sequence are shown against the background of all the current traces to illustrate the variability that is sometimes observed on the channel-like events associated with the time-dependent current. B, Using the TRAMP analysis (Tyerman et al., 1992), the amplitude histograms were extracted for each pulse and are shown superimposed. Some consistent levels were observed between pulses (dotted vertical lines), but it was not possible to extract the lowest common amplitude of the deviations.

Magnesium Ions Block the Current on the Symbiosome Lumen Face

A Mg²⁺ concentration of 1 mm in the bath (corresponding to the symbiosome lumen face of the PBM) blocked about 75% of the time-dependent current (Fig. 3A) when 60 mm NH₄⁺ was in the bath solution. The effective concentration at which 50% of the NH₄⁺ current was inhibited (ID₅₀) was 351 μm (Fig. 3B). The dose response could be well fitted by a dose-response curve where the Hill slope was equal to 1. There was no significant difference between the fitted dose-response curves and hence for the ID₅₀, within the range of −180 to −100 mV for which current activation was clear. Thus, it appears that the voltage dependency of Mg²⁺ block is absent or very weak and not detectable.

Figure 3.

Inhibition of NH₄⁺ currents of soybean symbiosomes by Mg²⁺. Time-dependent currents of excised, inside-out patches of the symbiosome membrane bathed in 60 mm NH₄⁺. The current was recorded during 2-s voltage pulses at 20 mV intervals ranging from 0 to −180 mV. The pipette solution was the same in all recordings (150 mm KCl, 10 mm CaCl₂, 10 mm HEPES/KOH, pH 7.0) and the bath solution always contained 60 mm NH₄⁺ and approximately 100 nM free Ca²⁺ with various Mg²⁺ concentrations. A, The same patch was exposed to increasing Mg²⁺ concentrations in the presence of 60 mm NH₄⁺. B, Mg²⁺ dependence of the relative conductance. The slope conductance (between −180 and −120 mV) was calculated from respective I-V diagrams, and the mean relative conductance of four sets of experiments was plotted against the Mg²⁺ concentration (±SEM), giving an ID₅₀ of 351 μm.

The NH₄⁺ current could be completely inhibited by 2 mm Mg²⁺ when the bath contained 60 mm NH₄⁺; in fact, a small decaying inward current became apparent (Fig. 4B). When the NH₄⁺ concentration was only slightly increased to 70 mm, the time-dependent activation of the NH₄⁺ currents reappeared (Fig. 4C) and were similar to those observed with 1 mm Mg²⁺ at 60 mm NH₄⁺(Fig. 4A). The reduction in the Mg²⁺ inhibition by increasing the NH₄⁺ concentration may indicate a competition between these two ions for the same binding site, or interactions between two separate binding sites. To investigate a possible interaction between NH₄⁺ and Mg²⁺, inward currents were recorded at various NH₄⁺ concentrations in the absence or the presence of 2 mm MgCl₂, and the relative conductance was calculated from the resulting IV curves. The results of Figure 5 showed a saturation of the NH₄⁺ current with bath NH₄⁺ concentration. With no Mg²⁺ present, a K_m of 26.4 mm was determined by nonlinear regression. With 2 mm Mg²⁺, no inward currents are observed until 70 mm NH₄⁺ and further increases in NH₄⁺ concentration indicated that saturation occurred at higher NH₄⁺ concentrations (Fig. 5). With Mg²⁺ present, none of the commonly known interaction types (competitive, uncompetitive, or noncompetitive interactions) were recognized (data not shown).

Figure 4.

Effect of changes in the NH₄⁺ concentration on the Mg²⁺ inhibition of the time-dependent inward current. For A to C, the voltage pulse protocol consisted of 20 mV steps from 0 mV to −180 mV. A, At 60 mm NH₄⁺ and 1 mm Mg²⁺, there is an obvious time-dependent inward current. B, Increasing the Mg²⁺ concentration to 2 mm caused complete block. There was a small decaying inward current that appeared to have similar voltage dependence to the normal activating current (note different current scale and see IV curve in D). C, Increasing the NH₄⁺ concentration to 70 mm caused release of the Mg²⁺ inhibition to give similar time-dependent currents as recorded in A. D, Current versus voltage curves for the time-dependent components of inward current for A, B, and C. The decaying inward current is depicted by the line above the voltage axis. All fitted curves are fourth-order polynomials.

Figure 5.

The relative conductance, G_rel, was calculated from I-V curves such as those shown in Figure 4D, between −180 and −120 mV, and nonlinear regression of the Michaelis-Menton equation gave an apparent K_m of 26.4 mm NH₄⁺ (0 mm Mg²⁺). The pipette always contained the standard pipette medium and bath solutions with increasing NH₄⁺ and various Mg²⁺ concentrations.

Figure 6 shows the effect of Mg²⁺ on the E_rev of the NH₄⁺ current that were obtained by tail current analysis. Without Mg²⁺ present in the bath, the E_rev shifted to more positive potentials when the NH₄⁺ concentration of the bath was increased (Fig. 6A), indicating a flux of cations (NH₄⁺) from the symbiosome lumen (bath) to the cytoplasm (pipette). The shift in E_rev followed closely the predicted Nernst equilibrium potential for monovalent ions with the assumption that the permeabilities of K⁺ and NH₄⁺ were equal (Tyerman et al., 1995), giving a slope of the regression line that was not significantly different to 58 mV per decade NH₄⁺. This contrasts to data obtained under similar solution regimes for L. japonicus PBM where the observed membrane potentials were offset in the negative direction (below the Nernst equilibrium line; Roberts and Tyerman, 2002). Upon addition of Mg²⁺ (2 mm) to the bath solution (lumen face of the PBM), the E_rev of the time-dependent current became more negative, suggesting an additional permeability of a cation whose electrochemical gradient is directed into the symbiosome lumen (bath). Note that an increase in anion permeability under the conditions of the experiment would shift the E_rev more positive because the Cl⁻ concentration in the pipette is higher than in the bath. Previous experiments had also established that the PBM-NSCC showed no appreciable anion selectivity (Tyerman et al., 1995; Whitehead et al., 1998; Roberts and Tyerman, 2002). The high deviation from the predicted Nernstian E_rev is reflected by the regression line with a slope of 95 mV (decade NH₄⁺)⁻¹. When maintaining the NH₄⁺ concentration at 60 mm and increasing the Mg²⁺ concentration, the E_rev shifted to more negative voltages (Fig. 6B).

Figure 6.

Effect of Mg²⁺ on E_rev of the NH₄⁺ current. NH₄⁺ currents were activated by clamping the voltage to −160 mV and tail currents were recorded by stepping the voltages to values between −120 and 100 mV. A, The E_rev were determined, averaged (n = 4–8, ±SEM), and plotted against the NH₄⁺ concentrations of the bath solution (black circles, 0 mm Mg²⁺; white circles, 2 mm Mg²⁺). Lines were fitted to the data giving slopes of 63.82 mV (decade NH₄⁺ concentration)⁻¹ and 94.50 mV (decade NH₄⁺ concentration)⁻¹ for 0 mm and 2 mm Mg²⁺, respectively. The dashed lined indicates the Nernst equilibrium potentials. B, Increasing Mg²⁺ concentrations shift the E_rev of the NH₄⁺ current to more negative voltages indicating a higher conductivity for cations with an electrochemical gradient directed into the bath solution. Ion concentrations in solutions: bath contained (in mm) 60 NH₄Cl, 10 HEPES/Tris, pH 7.0, 1 EGTA, variable MgCl₂; pipette standard.

This shift in E_rev of about 10 to 15 mV relative to the predicted Nernst potential (see Fig. 6A) for univalent cations is not enough to substantially effect the conductances that we measured for the Mg²⁺ data in Figure 5. Also, inspection of Figure 4B shows that there is absolutely no time-dependent activation of inward current (tails also have reversed) at any membrane potential (up to −180 mV) at the threshold point of 60 mm NH₄⁺ with 2 mm Mg²⁺ present. Also note that at 150 mm NH₄⁺ there was no difference between the E_rev with and without 2 mm Mg²⁺ (Fig. 6A), yet there was a reduction in conductance of about 50% (Fig. 5). Thus, the shift in E_rev with Mg²⁺ present in the bath cannot explain the apparent threshold effect illustrated by the data in Figure 5.

Stationary Noise Analysis

We examined the noise associated with the macroscopic time-dependent currents to attempt to clarify the variable behavior of currents (e.g. Fig. 1), and to examine whether or not Mg²⁺ caused additional noise due to open-channel blockade. Some patches were analyzed where differences in noise of the time-dependent current were observed over the duration of the experiment, and where the same voltage pulse protocol was used (40 current traces record; Fig. 7A). The spectral densities of smooth and noisy current traces of the same recording batch were analyzed separately to detect any difference in the Mg²⁺ block on these current types. In total, each of at least 6×40 current traces on three different PBM patches that were recorded under various conditions of NH₄⁺ and Mg²⁺ concentrations (100 mm NH₄⁺, 2 mm Mg²⁺; 150 mm NH₄⁺, 2 mm Mg²⁺; 150 NH₄⁺, 1 mm Mg²⁺; as shown in Fig. 5) were analyzed. A typical example is shown in Figure 7, B and C. No major Mg²⁺ effect on the frequency dependence of the spectral densities was noticed when current traces with small fluctuations were analyzed (Fig. 7B), indicating that the Mg²⁺-induced inhibition of the current does not cause any obvious channel flickering. Nevertheless, a decrease in the averaged spectral density especially at lower frequencies can be observed in these (Fig. 7C) as well as in current traces with initially larger fluctuations present (Fig. 7C, insert). Note that all data were obtained from the same patch and the same recording (40 current traces). In the other two experiments, similar results were observed: Mg²⁺ did not affect the frequency dependence, but only reduced the amplitude of the spectra in both smooth and more noisy time-dependent currents (data not shown).

Figure 7.

Mg²⁺ effects on the noise of NH₄⁺ currents. Forty current traces (=1 batch) were recorded under the indicated conditions (150 mm NH₄⁺ ± 1 mm Mg²⁺) at −160 mm voltage steps of 2 s duration, filtered at 2 kHz and digitized at 5 kHz. A, Examples of smooth and noisy current recordings from the same recording batch in 150 mm NH₄⁺ without Mg²⁺ after removal of the direct current component. Note the channel-like events in the noisy current trace. B, Power spectrum of six traces of smooth current ± 1 mm Mg²⁺, respectively, chosen by chance from two 40 current traces batches of the same patch. C, Mean spectral density calculated from six and 17 smooth current traces in the absence and presence of 1 mm Mg²⁺, respectively. Insert shows the mean spectral density of nine (0 mm Mg²⁺) and eight (1 mm Mg²⁺) noisy current traces from the same recording of A and B. Note that all data were obtained from the same 40 current trace recording on the same symbiosome patch.

Block by Mg²⁺ of the time-dependent NH₄⁺ currents occurred irrespective of the presence or absence of the channel-like activity within the time-dependent current. No effect could be determined on the magnitude of the channel-like steps occasionally observed. Nonstationary noise analysis was performed on the time-dependent currents to determine the effect on the single-channel amplitude. No significant difference was observed ± Mg²⁺ at 100 mm and 150 mm NH₄⁺, but this may be due to the statistical power of the analysis because very high standard errors of the single channel's estimates were obtained.

Temperature Sensitivity of the NH₄⁺ Currents

The temperature sensitivity of the NH₄⁺ currents may differentiate between different molecular mechanisms of transport across the PBM, since channel- and carrier-mediated transport show different temperature dependence (Table I).

To ensure that a good estimate of the temperature dependence of open-channel conductance was made, several precautions were taken in the experiments. We have observed some variability in currents over the duration of experiments on a single patch, apart from the noise described above. Often, the conductance increased as though more transporters were recruited into becoming active. To check that this did not occur during temperature experiments, we compared conductance before and after the temperature variations. On one occasion the conductance increased, and this correlated with a high Q₁₀ and activation energy (E_A) being measured. It was also known from other studies that the activation rates of channels can be extremely temperature sensitive (Acerbo and Nobile, 1994); therefore, we ensured that sufficient time during a voltage clamp pulse was given so that the current became fully activated at a particular voltage and temperature. Figure 8A shows that the full activation of the current with time is delayed at low temperature but that sufficient time is allowed for a steady-state current to be achieved.

Figure 8.

Temperature and voltage dependence of NH₄⁺ currents. Currents were recorded after voltage pulses ranging from −180 to 40 mV at 20 mV intervals at different temperatures: A, 8.9°C; B, 22.7°C. The recording chamber was perfused with 150 NH₄⁺, 0 Mg²⁺ medium preincubated at the indicated temperatures. The temperature was monitored in the bath chamber during the current recordings and calculated from the voltage output of a thermistor. C, IV diagrams of the instantaneous (I_i) and the final current (I_f) from the recordings in A and B.

Another complication arises because we cannot observe single-channel events and therefore cannot get a direct measure of open-channel current. Therefore, time-dependent currents must be activated at voltages where the channel open probability (P_open) is maximal and preferably near 1. To reveal the voltage dependency of current activation, the corresponding IV curves for initial current (I_i) and final current (I_f) were plotted (Fig. 8C). A temperature dependence of the current at all voltage steps was observed where initial current and final steady-state current, as well as the time-dependent component (I_f − I_i), showed no obvious change of their voltage dependence at 8.9°C or 22.7°C.

These current recordings were used to examine the temperature dependence as a function of voltage in more detail. The activation of the current can be described as a Boltzmann function with a V₅₀ of approximately −60 mV and an effective gating charge of near 1 (Whitehead et al., 1998). These data are obtained from the current tails at a voltage positive of the E_rev, where it can be seen that the current tails pile up after very negative voltage pulses in Figure 8, A and B (for details, see Whitehead et al., 1998). At all three temperatures shown in Figure 9A, the current is fully activated at −180 mV. We did observe a negative shift in V₅₀ with decreasing temperature. The average of five experiments gave a shift of 2.32 mV °C⁻¹. The slope factor, which can be used to calculate gating charge, also shifted with a mean coefficient of 0.73 mV °C⁻¹, but this was not significantly different from zero.

Figure 9.

Temperature and voltage dependence of the NH₄⁺ currents. A, The amplitude of the instantaneous tail current was plotted against the prepulse voltage, allowing a direct estimation of the activation characteristics of the current. Currents were recorded at 11.4°C, 12.4°C, and 22.7°C, respectively. The fitted lines are Boltzman curves. Note that the current is fully activated at −180 mV for low as well as for high temperatures. B, Arrhenius diagram of the time-dependent current for different voltages as indicated. Currents were recorded as in Figure 8 at 12 different temperatures ranging from 9.2°C to 21.9°C. Bath solution, 150 mm NH₄⁺, 0 mm Mg²⁺; pipette solution, standard.

The E_A at different voltages was obtained from Arrhenius plots of the time-dependent current amplitudes measured at the same voltage. At voltages near V₅₀, the E_A was much higher (41 kJ mol⁻¹ at −80 mV in Fig. 9B) than at −180 mV (18 kJ mol⁻¹ in Fig. 9B). In total, an average E_A of 28.2 kJ mol⁻¹ ± 2.87 (SEM, n = 7) was measured for the currents when they were maximally activated. The range in E_A for currents that were fully activated was from 18.3 to 39.8 kJ mol⁻¹. The average Q₁₀ was 1.51 ± 0.06 (SEM, n = 7). At a membrane voltage of −100 mV, the average E_A was 48.7 kJ mol⁻¹ ± 8.6 (SEM, n = 4).

We also analyzed the effect of temperature on activation and deactivation kinetics of the time-dependent currents. As reported by Whitehead et al. (1998), the activation could be fit by a Hodgkin-Huxley equation with P = 2 or a single exponential time course. In both cases the time constant of activation was very temperature dependent with an E_A of 64.2 ± 11.1 kJ mol⁻¹ (Q₁₀ = 2.7 ± 0.5). For deactivation, the kinetics were well described by the sum of two exponentials; for both time constants, very low temperature sensitivity was recorded (E_A between 2 and 6 kJ mol⁻¹, not significantly different from 0; Q₁₀ = 1.1, not significantly different to 1).

DISCUSSION

Previous studies have assigned the NH₄⁺ current to a channel mechanism (PBM-NSCC) despite the absence of channel current transitions in patch-clamp recordings so far reported and the equivocal stationary noise characteristics as displayed in this study (Fig. 7). The assignment is based on the passive nature of the currents that are uncoupled from other ion movements and the ability to model the variance during current activation to simple gating (Tyerman et al., 1995). Also, blockade by divalent cations, polyamines (Whitehead et al., 2001), and verapamil (Whitehead et al., 1998) is consistent with channel-mediated transport. In this study, we have demonstrated that single-channel-like events are associated with the NH₄⁺ current. These events appear to lie at one extreme of a variable noise level on the time-dependent currents. The events we observe are closely associated with the time-dependent current, both in terms of their activation and their selectivity, and we propose they represent a form of cooperative gating between variable numbers of the subpicoSiemens channel conductance. We have applied nonstationary noise analysis and have confirmed that the single-channel amplitudes lie in the range of 5 to 50 fA depending on voltage, giving a conductance of between 0.11 and 0.22 pS as reported by Tyerman et al. (1995). From the size of the current steps observed in some patches (e.g. Fig. 1C), this would suggest that up to 1,000 single channels may activate and deactivate simultaneously under some circumstances. Similar behavior has been observed for anion channels in Chara plasma membrane, although with larger unit conductance (McCulloch et al., 1997). In Chara, the anion current could be modeled as unit channels forming a larger cluster that allowed cooperative gating. The specific conditions that cause this behavior are so far unknown.

The temperature dependence of the currents is more consistent with a channel mechanism. An E_A of 28 kJ mol⁻¹ when the current is fully activated lies within the range of values obtained for other ion channels; although slightly on the higher side of the range, it is certainly less than the subpicoSiemens voltage-dependent H⁺ channels (DeCoursey and Cherny, 1998). Transport through carriers show consistently a much higher E_A > 60 kJ mol⁻¹.

The limiting NH₄⁺ conductance in water gives an E_A of 12.3 kJ mol⁻¹ (data from appendix 6.2 in Robinson and Stokes, 1968). Thus, our measurement for NH₄⁺ conductance is over twice this value, probably reflecting either limitation of NH₄⁺ movement within the pore of the PBM-NSCC or some gating still occurring when the channels are fully activated at −180 mV. Although the Boltzmann fits demonstrated that the PBM-NSCC was fully activated at −180 mV, the possibility remains that the P_open of the channel is still less than one. Nonstationary noise analysis had previously demonstrated that current variance as a function of mean current during activation follows a parabolic trajectory, reaching a maximum in variance at about half the full current activation (Tyerman et al., 1995). This can be understood in terms of the channel passing beyond a point of P_open = 0.5 and approaching P_open near 1, but some gating may still be present when the current is fully activated.

The higher E_As recorded at intermediate voltages where the currents are about half activated also indicates that the transitions that determine P_open have some steps with a high E_A. The kinetics of current activation were highly temperature dependent (Q₁₀ = 2.7) in contrast to channel deactivation (Q₁₀ = 1), implying that the rate constant(s) for channel opening is more temperature dependent than channel closing. If this is the case, then the P_open would decline with decrease in temperature, and this should be indicated in our conductance data at voltages where P_open is near 0.5. We observed a negative shift in the V₅₀ of the Boltzman curves fitted to relative conductance as a function of voltage of 2.3 mV per degree reduction in temperature without any change in the shape of the curves. Assuming that the Q₁₀s of the channel opening and closing rate constants are the same as the current activation and deactivation rate constants, we can calculate that at a voltage where P_open was initially 0.5, P_open would decline to 0.3 for a 10°C reduction in temperature. This corresponds very closely to the change in relative conductance at V = −60 mV (0.5–0.3) that we observe from the Boltzman fits for a 10°C reduction in temperature. Note that at a voltage where the current is near maximally activated (−180 mV), the reduction in relative conductance is only from 0.98 to 0.96 (from Boltzman fits). This is explained by the rate constant for channel opening being much larger at this voltage than the rate constant for channel closing (greater than 10-fold), so a reduction in the rate constant of channel opening caused by lower temperature has less of an effect on P_open at more negative voltages.

There are only a limited number of studies on the temperature dependence of plant ion channels (see Table I), with one whole-cell study giving rather high values for well-established classes of plant ion channels (Colombo and Cerana, 1993). This probably is because the currents represent more the gating of the channels rather than the single-channel conductance. However, it is the temperature dependence of the whole-cell/organelle currents representing the ensemble of channels at a particular voltage-dependent P_open that is relevant to the biology of the organism. In our study, we have analyzed the temperature dependence in the context of defining the type of transport reaction (carrier or channel). Such analysis may also be the definitive way to confirm the proposed channel transport mechanism of NH₃ in AmtB.

In terms of the biology of nitrogen fixation, for the energized PBM where hyperpolarization of the symbiosome membrane is proposed to occur via H⁺-ATPase activity (Udvardi and Day, 1989; Udvardi et al., 1991), a membrane voltage near −100 mV would be expected. At this membrane voltage, the NH₄⁺ channels would not be fully activated and the temperature dependence of channel gating would dominate the overall temperature dependence of NH₄⁺ transport. From our results, the mean E_A at −100 mV was 49 kJ/mol (Q₁₀ is approximately 2). This is similar to that observed for NH₃ transport, about 55 kJ mol⁻¹ for soybean PBM (Niemietz and Tyerman, 2000). In that study, addition of aquaporin inhibitors more than doubled the E_A. Thus, both NH₃ and NH₄⁺ transport across the PBM, although channel mediated, would both display high temperature dependence with a Q₁₀ of the order of 2 and similar to that recorded for nitrogenase in intact soybean nodules (Walsh and Layzell, 1986).

Activation of PBM-NSCC is highly temperature dependent (E_A = 64 kJ mol⁻¹), indicating that some conformational changes must occur in the channel protein during opening. This has implications for the proposed model of a simple divalent blocking and unblocking mechanism of PBM-NSCC gating proposed by Whitehead et al. (1998). Such a high E_A is not uncommon for channel activation (see Table I). It is also not uncommon to have a very low E_A for channel deactivation or inactivation (e.g. the cystic fibrosis transmembrane conductance regulator channel). The E_A measured in this study for deactivation of the NH₄⁺ current is not significantly different to a diffusion mechanism. Addressing the Whitehead et al. (1998) model of gating in the PBM-NSCC, it was proposed in that study that Mg²⁺ normally blocked the channel from the cytosolic face of the PBM. During activation of the channel by making the membrane more negative, normally achieved by the H⁺-ATPase on the PBM, the Mg²⁺ is displaced out of the pore by permeating NH₄⁺ from the symbiosome lumen. This “knock-off” effect was justified by the observation that absence of divalent cation on the cytosolic face of the membrane abolished the rectification and voltage-dependent activation of the channels. However, a simple unblocking mechanism for channel activation would not be expected to show such high E_A. Therefore, a modified model of gating requires that the opening of the channel involves a conformational change in the protein; this is dependent on either the presence of Mg²⁺ or Ca²⁺. On the other hand, channel deactivation gives an E_A consistent with simple diffusion kinetics, and this would correspond to blockade of the channel by divalent cations from the cytosolic face of the PBM. To reconcile these two observations of very different temperature sensitivity, we propose that subsequent to the initial blockade (low E_A) during channel closing, the binding of the divalent cation to the closed channel must cause a significant conformational change (stabilization of the closed state?). In this respect, we are proposing that there is more than one closed state. It is this conformational change that must be reversed (high E_A) when the channel is activated by negative PBM voltages. We have previously suggested that there are two open states of the channel based on the double exponential kinetics for deactivation (Whitehead et al., 1998).

It is interesting to note that this mechanism of channel gating is similar to that proposed for the SV channel in Beta vulgaris, where there is a site within the channel that allows low affinity blockade by divalent cations (Ca²⁺ and Mg²⁺) that also permeate the channel, and a Ca²⁺-selective site outside of the voltage field that stabilizes the closed state (Pottosin et al., 2004). However, there are significant differences between the SV channel and the PBM-NSCC. For one, the PBM-NSCC shows voltage-independent channel blockade by Mg²⁺ and Ca²⁺, whereas the SV channel shows voltage-dependent block. Calcium on the symbiosome lumen face blocked the PBM-NSCC with an ID₅₀ of 8.3 μm (Whitehead et al., 1998). Here, we show that Mg²⁺ also blocks PBM-NSCC from the lumenal side with a higher ID₅₀ (351 μm). The blockade by Mg²⁺ was not voltage dependent and showed a Hill slope of one, indicating that only one binding site for Mg²⁺ is involved. This binding site must therefore be very close to the entrance of the pore and outside of most of the voltage gradient across the pore.

Despite the external site for divalent blockade of PBM-NSCC, there were unusual interactions between the permeating ion and Mg²⁺. The interaction cannot be described by a simple competitive inhibition or other known forms of interaction with the permeating ion. The most interesting observation with respect to NH₄⁺ permeation is that at a concentration of Mg²⁺ that totally blocks the channel for a specific NH₄⁺ concentration, it only requires a relatively small increment in NH₄⁺ concentration to relieve this complete blockade. Thus, if the Mg²⁺ concentration was set relatively constant within the symbiosome lumen, there would exist a threshold concentration of NH₄⁺ below which there would be no transport to the cytoplasm. As N₂ fixation by the bacteroids built up the lumen concentration of NH₄⁺, the threshold concentration would be overcome and transport to the cytoplasm would proceed. Complicating this is the substantial increase in NH₄⁺:K⁺ selectivity of the PBM-NSCC when lumen NH₄⁺ declines (Whitehead et al., 1998). However, the set point concentration of NH₄⁺ in the symbiosome lumen for transport to proceed would be mainly determined by the free concentration of Mg²⁺ and Ca²⁺ in the lumen. There are no reported measurements of free divalent concentrations in the symbiosome lumen, which would be difficult to ascertain for in vivo conditions because of the small volume of the space and the sensitivity of N fixation caused by symbiosome isolation. However, symbiosome attached patches revealed similar inward currents (Tyerman et al., 1995; Roberts and Tyerman, 2002), indicating that the free divalent concentration must be low enough and univalent cation concentration high enough under isolation conditions to not substantially block the inward current through PBM-NSCC.

Another interesting observation regarding blockade by Mg²⁺ was the change in selectivity of the channel. With increasing Mg²⁺ concentration, as the channel currents became reduced, they also became more selective for an ion with an electrochemical gradient from the pipette solution to the bath. Under the conditions of our experiment, this is likely to be an increased Ca²⁺ permeability. We believe we can discount the alternative, an increase in K⁺ permeability, because all studies to date have shown the reverse is normally the case, i.e. at lower NH₄⁺ concentrations, the relative permeability of K⁺: NH₄⁺ decreases (Tyerman et al., 1995; Whitehead et al., 1998). Also, the orthologous channel in L. japonicus, in contrast to the soybean channel, shows Ca²⁺ permeability independent of the presence of blocking divalent concentrations (Roberts and Tyerman, 2002). Thus it appears that the PBM-NSCCs in legume PBM have the propensity to conduct divalent cations under certain circumstances. When there is divalent permeation in soybean, there is a substantial reduction in conductance indicating an interaction between univalent and divalent conduction in the pore (Fig. 5).

The permeation of divalents through the PBM-NSCC may determine the concentrations of divalents in the symbiosome lumen, which in turn will determine the set point for NH₄⁺ transport. This may work like a negative feedback mechanism whereby progressive blockade by higher symbiosome lumen concentrations of Ca²⁺ or Mg²⁺ causes the channel to conduct more of these divalents out of the symbiosome lumen to the cell cytoplasm, thus relieving the block. The small volume of the symbiosome lumen would mean that this could be quite a sensitive control system.

The degree of control evident over PBM-NSCC suggests that other NH₄⁺ permeable NSCCs in different membranes could have similar sophisticated control over activity, and this may be very important to prevent futile cycling of NH₄⁺ (Britto et al., 2001a). The FV channel in tonoplast is more permeable to NH₄⁺ than for K⁺, and, thus, if the vacuole is to function as a trap for NH₄⁺, proposed by Loqué et al. (2005) to be contributed by NH₃ permeable aquaporins on the tonoplast, there needs to be a mechanism to prevent NH₄⁺ influx to the cytoplasm. The FV channel is blocked by cytosolic Ca²⁺ and Mg²⁺ and vacuolar Mg²⁺ (Tikhonova et al., 1997; Brüggemann et al., 1999) that may prevent excessive NH₄⁺ influx to the cytoplasm, but it remains to be seen if there is an interaction between NH₄⁺ and Mg²⁺ as displayed by the PBM-NSCC. It is interesting to note that Na⁺ blocks the SV channel from the luminal side, therefore explaining how the SV nonselective cation channel can maintain Na⁺ gradients of 150-fold (Ivashikina and Hedrich, 2005). It is likely that a similar situation may occur with NH₄⁺, but this has not yet been examined in the literature.

The cycling of NH₃ back into the symbiosome lumen from the cytoplasm depends on the cytoplasmic concentration of NH₄⁺ and the NH₃ permeability of the PBM. The PBM has a relatively low NH₃ permeability despite being channel mediated and it is reduced by ATP preincubation (Niemietz and Tyerman, 2000). In the futile cycling across the plasma membrane described by Britto et al. (2001a), energy-dependent net NH₄⁺ efflux occurs in plants (e.g. barley [Hordeum vulgare]) that show maintained negative membrane potentials with high external NH₄⁺ concentrations. From our study, it is likely that the K⁺/NH₄⁺ selectivity and control by divalents of NSCCs are key features that distinguish between plants that differ in NH₄⁺ toxicity.

The molecular identity of the PBM-NSCC has been assigned to GmSAT1, which is a novel protein situated on the PBM that has an N-terminal helix-loop-helix motif similar to transcription factors and a hydrophobic transmembrane domain (Kaiser et al., 1998; Marini et al., 2000). This protein can complement yeast (Saccharomyces cerevisiae) strain 26972c that has certain mutations in two of the three Mep NH₄⁺ transporters (Amt family) that render it dependent on high NH₄⁺ for growth and insensitive to high methylamine concentrations. In 26972c Mep3 is trans-inhibited by the defective Mep1. However, GmSAT1 failed to complement the triple knockout of the Meps (mep1Δ mep2Δ mep3Δ; Marini et al., 2000), casting doubt over the original interpretation that GmSAT1 was the PBM-NSCC. It appeared that GmSAT1 that is targeted to the yeast plasma membrane could be interacting with Mep3 or another cation channel when expressed in 26972c. A proteomic analysis of L. japonicus PBM, which has the PBM-NSCC (Roberts and Tyerman, 2002), failed to detect any class of potassium transport protein (Wienkoop and Saalbach, 2003). Thus, it would appear that the PBM-NSCC may either belong to an as yet uncharacterized family of transporter, or it is manifested via the interaction between membrane proteins that have already been identified in the PBM.

CONCLUSION

The NH₄⁺ current in the PBM of soybean nodules has been positively associated with a channel-mediated transport mechanism from its temperature dependence, despite its unusual characteristics of cooperative gating, stationary noise characteristics, and very low single-channel conductance. Magnesium ions change the channel conformation, and to open the channel a high energy step is required. With partial Mg²⁺ block the transport of divalent cations becomes possible, and this may be involved in setting the threshold for NH₄⁺ transport through the channel by virtue of the novel interaction between divalent concentration and NH₄⁺ concentration in the symbiosome lumen.

MATERIALS AND METHODS

Plant Material

Nodules were collected just before the experiments from 5- to 7-week-old soybean plants (Glycine max L. cv Stephens or Forest) that were grown in a naturally illuminated greenhouse with additional illumination to maintain 16-h-light/8-h-dark cycles. Plants were inoculated with Bradyrhizobium japonicum U.S. Department of Agriculture 110 and grown in pots containing a mixture of perlite and vermiculite. To isolate the symbiosomes, three to five nodules were cut with a razor blade or gently crushed in standard bath medium with 5 mm dithiothreitol added, at 4°C in a petri dish. After 5 to 10 min on ice, a small volume was transferred into the perfusion chamber filled with standard bath medium. Symbiosomes were allowed to settle at the bottom of a glass coverslip for another 5 min. The chamber was then perfused with approximately 5 to 10 mL of fresh standard bath medium, and the symbiosomes that remained adherent to the glass coverslip formed high GΩ seals when they were gently touched by a patch pipette.

Buffers and Solutions

In all experiments, a standard pipette medium was used containing (in mm) 80 mannitol, 150 KCl, 10 CaCl₂, 5 HEPES adjusted to pH 7.0 with KOH. The standard bath solution contained (in mm) 170 mannitol, 100 K⁺Glu, 2 MgCl₂, 10 EGTA, 2.3 CaCl₂ (giving a free Ca²⁺ concentration of 84.8 nm), 5 HEPES adjusted to pH 7.0 with Tris. Other bath media with various NH₄⁺ concentrations were made on the basis of a 10 mm HEPES/Tris buffer, pH 7.0, with 1 mm EGTA. Osmolalities of all solution were adjusted with mannitol to approximately 400 mOsmol. The free Ca²⁺ concentration (approximately 100 nm) was calculated in respect to the Mg²⁺ concentration, pH, and ionic strength of the solutions using the GEOCHEM program (Parker et al., 1987).

Electrophysiological Recordings, Data Acquisition, and Analysis

Patch pipettes were pulled from borosilicate glass capillaries (GC150-10, Clark Electronic Instruments) to an o.d. of approximately 0.5 to 1 μm, and tips were coated with Sylgard (Dow Corning). In general, the pipettes were fire polished and back filled with standard pipette medium. Best sealing rates were obtained with pipettes and showed a resistance of 15 mΩ in standard bath medium. An Ag/AgCl reference electrode was connected by a 100 mm KCl agar bridge to the bath. High resistance (15–30 GΩ) inside-out oriented patches of the symbiosome membrane were obtained as described previously (Whitehead et al., 1998). With this membrane orientation and voltage sign convention (reference at the inside of symbiosome), negative currents reflect a flux of cations from the symbiosome into the cytosol.

The currents were measured using an Axopatch 200A (Axon Instruments) or a List EPC-7 amplifier (List Electronics). The currents were filtered at 1 or 2 kHz and digitized at frequencies of 2 and 5 kHz, respectively, using a Strobes (Strobes Engineering) or DigiData 1200B (Axon Instruments) analog to digital converter running under pClamp8 data acquisition and voltage pulse protocol software. Current data were analyzed by use of Clampfit (versions 8 and 9), MathCad, SigmaPlot, or Graphpad Prism softwares.

A pulse protocol was applied to excised membrane patches starting from a holding potential of 0 mV (0.1 s) and stepping the voltage for 2 s to values between 0 and −180 mV at 20 mV intervals, and finally clamping the voltage at 50 mV for 0.4 s. The time-depended currents were obtained by subtracting the instantaneous current component at the beginning of the negative voltage step from the total current (mean current of last 200 ms).

To obtain the E_rev of the respective currents, the patched membrane was hyperpolarized for 2 s (−160 mV) to activate all channels and then stepped to more positive values ranging between −140 and 100 mV in 20 mV intervals. The resulting tail currents were analyzed according to standard electrophysiological protocols, namely, the potential at which a “zero” tail current was detected was taken as the E_rev.

Current Noise Analysis

Stationary noise analysis examines the fluctuations in current as a function of time while the average current is constant in time. This occurs at a fixed voltage and requires that all time-dependent changes in current have ceased. Nonstationary noise analysis examines the kinetics of current fluctuations during current activation or deactivation, generally after a change in voltage. For the stationary noise analysis carried out in this paper 40 current traces were recorded after at voltages of −120, −140, −160, and −180 mV. The current was filtered at a cut-off frequency of 2 kHz (Bessel-type filter) and digitized at 5 kHz. Noise spectra were generated from every one of these current traces (1 s, 4,250 data points) after manual removal of the direct current component and using a Poisson periodogram (factor α = 6, pClamp9 software).

Temperature Control

To observe the temperature dependence of the currents, a thermistor-based (RS Components, nos. 151–136) thermometer was inserted into the perfusion chamber as close as possible to the patch pipette. After recording the currents at room temperature, prechilled bath medium was perfused through the chamber until a stable temperature was recorded. The bath medium was then allowed to warm up to room temperature again while currents at −180 mV steps as well as the corresponding temperature were recorded. Temperature changes in the chamber were slow and did not change more than 1°C during the recording of one set of currents. The E_A was calculated from the slope of a regression line of an Arrhenius diagram (ln [-I] versus T⁻¹), and Q₁₀ values, the relative changes in a parameter for a 10°C change in temperature, were obtained from a linear regression line of I-T diagrams with T ranging from 8°C to 18°C, or Q₁₀ was calculated for other temperature ranges by:

with I₂ as the current at the higher temperature T₂, and I₁ the current at the lower temperature T₁. Unless otherwise stated, the fully activated time-dependent current at a −180 mV voltage pulse was used for these calculations.