The Plant Journal (2006) 46, 269–281
doi: 10.1111/j.1365-313X.2006.02690.x
External K+ modulates the activity of the Arabidopsis potassium channel SKOR via an unusual mechanism Ingela Johansson1,†, Klaas Wulfetange2,†,‡, Fabien Pore´e3, Erwan Michard2,3,§, Pawel Gajdanowicz2, Benoıˆt Lacombe4, Herve´ Sentenac4, Jean-Baptiste Thibaud4, Bernd Mueller-Roeber2,3, Michael R. Blatt1 and Ingo Dreyer2,3,* 1 Laboratory of Plant Physiology and Biophysics, IBLS Plant Sciences, Bower Building, University of Glasgow, Glasgow G12 8QQ, UK, 2 Universita¨t Potsdam, Institut fu¨r Biochemie und Biologie, Abteilung Molekularbiologie, Karl-Liebknecht-Strasse 24-25, Haus 20, 14476 Potsdam/Golm, Germany, 3 Max-Planck Institute of Molecular Plant Physiology, Cooperate Research Group, 14476 Potsdam/Golm, Germany, and 4 Biochimie et Physiologie Mole´culaire des Plantes, UMR 5004, Agro-M/CNRS/INRA/UM2, 34060 Montpellier Cedex 1, France Received 18 August 2005; revised 4 December 2005; accepted 21 December 2005. * For correspondence (fax þ49 331 977 2512; e-mail
[email protected]). † These authors contributed equally to this work. ‡ Present address: FU Berlin, Institut fu¨r Biologie, Angewandte Genetik, 14195 Berlin, Germany. § Present address: Instituto Gulbenkian de Cieˆncia, R. Quinta Grande 6, PT-2780-156 Oeiras, Portugal.
Summary Plant outward-rectifying Kþ channels mediate Kþ efflux from guard cells during stomatal closure and from root cells into the xylem for root–shoot allocation of potassium (K). Intriguingly, the gating of these channels depends on the extracellular Kþ concentration, although the ions carrying the current are derived from inside the cell. This Kþ dependence confers a sensitivity to the extracellular Kþ concentration ([Kþ]) that ensures that the channels mediate Kþ efflux only, regardless of the [Kþ] prevailing outside. We investigated the mechanism of Kþ-dependent gating of the Kþ channel SKOR of Arabidopsis by site-directed mutagenesis. Mutations affecting the intrinsic Kþ dependence of gating were found to cluster in the pore and within the sixth transmembrane helix (S6), identifying an ‘S6 gating domain’ deep within the membrane. Mapping the SKOR sequence to the crystal structure of the voltage-dependent Kþ channel KvAP from Aeropyrum pernix suggested interaction between the S6 gating domain and the base of the pore helix, a prediction supported by mutations at this site. These results offer a unique insight into the molecular basis for a physiologically important Kþ-sensory process in plants. Keywords: Arabidopsis, channel protein–cation interaction, channel protein structure, Kþ-dependent gating, Kþ channel outward rectifier.
Introduction As the predominant inorganic ion of plant cells, potassium (K) plays a major role as an osmoticum contributing to cellular hydrostatic (turgor) pressure, growth and responses to the environment. Plants often confront large fluctuations in the ionic environment which affect both Kþ uptake and the electrochemical driving force for Kþ diffusion and its release. Plant cells accommodate changes in the Kþ environment through a number of adaptations in their capacity for Kþ transport and its regulation, processes that rely on a variety of transport proteins (Amtmann et al., 2004; Ve´ry and Sentenac, 2003). An important role is played by voltage-gated Kþ ª 2006 The Authors Journal compilation ª 2006 Blackwell Publishing Ltd
channels of the so-called Shaker family (Pilot et al., 2003). This family is represented in plants by inward-rectifying channels which facilitate Kþ uptake into root cells and guard cells, weakly voltage-dependent channels which are thought to play a role in phloem (un)loading, and outward-rectifying channels which mediate Kþ efflux from guard cells during stomatal closure and from root cells into the xylem for root– shoot allocation of potassium. Like the Kþ channels of animal cells, outward-rectifying Kþ channels in plants open on membrane depolarization. However, intriguingly, they do so only at membrane 269
270 Ingela Johansson et al. voltages positive of the equilibrium potential for potassium, EK. Hence, their gating is sensitive both to membrane voltage and to the prevailing extracellular Kþ concentration (Blatt, 1991). From a physiological standpoint, this ability to sense the prevailing Kþ concentration difference effectively guarantees that the channels open only when the driving force for net Kþ flux is directed outward. In stomatal guard cells, for example, these characteristics help ensure the Kþ efflux needed to drive stomatal closure and control gas exchange, even when the extracellular Kþ varies over concentrations from 10 nM to 100 mM (Blatt, 1988; Blatt and Gradmann, 1997; Roelfsema and Prins, 1997; Schroeder, 1989). In xylem loading these characteristics underlie the tightly controlled Kþ release by stelar cells into the vessels (Roberts and Tester, 1995; Wegner and de Boer, 1997; see also the animated model in the Supplementary Material). Although the physiological significance of the Kþ dependence of plant outward-rectifying Kþ channels is well recognized as one element contributing to a complex web of homeostatic controls within the whole plant (Amtmann et al., 2004), little detail has come to light that could have a bearing on the molecular mechanism of Kþ-dependent gating of these unusual channels. One previous analysis of the outward-rectifying Kþ channels of Vicia faba guard cells indicated that gating was subject to control by Kþ binding to sites with characteristics similar to those of the pore itself (Blatt and Gradmann, 1997). An analogous gating dependence on Kþ is characteristic of the outward-rectifying Kþ channel TOK1/YKC1 of yeast. In this case, mutagenesis studies identified two exterior Kþ-binding domains adjacent to each of the pore loops (Vergani and Blatt, 1999; Vergani et al., 1998) as well as two intracellular domains that contribute to gating (Loukin and Saimi, 1999; Loukin et al., 1997). However, TOK1 is structurally unique among eukaryotes, raising questions about the extent to which these observations might have a bearing on the mechanism of gating in the plant Kþ channels. The Arabidopsis genome includes two genes, GORK and SKOR, that encode Shaker-like, outward-rectifying Kþ channels (Ache et al., 2000; Gaymard et al., 1998). SKOR is expressed in the root stele and has been shown to play a major role in Kþ release to the xylem and therefore in the transport of K from roots towards shoots (Gaymard et al., 1998). GORK is the predominant outward-rectifying Kþ channel in guard cells and plays an important role in stomatal closure (Hosy et al., 2003). GORK and SKOR are structurally very similar, and both channels exhibit the canonical gating dependence on Kþ and membrane voltage of the plant outward rectifiers. GORK senses the Kþ concentration in the guard-cell environment and SKOR the Kþ concentration in the stelar apoplast. So as to analyze the features of this physiologically important sensory process, we took advantage of expression in heterologous expression systems. We carried out site-directed mutagenesis, targeting
domains throughout the SKOR Kþ channel that might contribute to such Kþ sensitivity, and analyzed the mutant currents. Remarkably, we uncovered a protein domain deep within the membrane that affects the Kþ and voltage sensitivity of the channel. These results suggest a sophisticated coupling between Kþ-sensitive gating and permeation and a highly efficient solution for Kþ sensing. The new molecular insights into the Kþ-sensory process of SKOR allowed us to propose a working model which explains all relevant details of Kþ sensing of outward-rectifying plant Kþ channels (animated in the Supplementary Material). Results Alkali cations modulate SKOR activity from outside When expressed in Xenopus oocytes, SKOR yields a slowly activating current with distinctly sigmoid kinetics on depolarizing voltage-clamp steps (Figure 1(a)) that are characteristic of this channel (Gaymard et al., 1998) and similar to those of outward-rectifying Kþ channels described previously in vivo (Blatt, 1988; Roelfsema and Prins, 1997; Schroeder, 1988; Wegner and de Boer, 1999). We found that increasing external [Kþ] from 3 to 30 mM resulted in an increase in the halftime (t1/2) for current activation (Figure 1b); it also depressed the steady-state current (Figure 1c) at any one voltage, and displaced the conductance–voltage curve to more positive voltages in parallel with the equilibrium potential for Kþ, EK (Figure 1d; see also Figure 5). The effect of Kþ on the SKOR current and conductance was independent of the means to balance ionic strength, whether by Naþ, Liþ or N-methyl-D-glucamine (NMDG). However, when substituted for Kþ, Csþ and Rbþ were almost as effective as Kþ in suppressing the current and displacing the conductance–voltage curve to more positive voltages (Figure 1e,f). This Kþ sensitivity of SKOR was effected solely from outside. We recorded the SKOR current in whole-cell mode from COS cells expressing the channel with 10 and 150 mM Kþ outside after dialysis against either 75 or 150 mM Kþ inside the pipette (Figure 1g,h). In each case, increasing [Kþ] outside displaced the current–voltage curve to the right along the voltage axis. Similar effects were observed with Rbþ and Csþ, while substitutions with Naþ and Liþ outside had no effect on voltagedependent SKOR gating. The SKOR channel, unlike the yeast TOK1/YKC1 Kþ channel (Loukin and Saimi, 1999), is not gated by internal [Kþ] in a voltage-dependent manner. However, a study in this issue shows that SKOR is regulated by internal [Kþ] through a voltage-independent mechanism (Liu et al., 2006). The results shown in Figure 1 suggest that, like the outward-rectifying Kþ channels in vivo (Blatt, 1988; Roelfsema and Prins, 1997; Schroeder, 1988; Wegner and de Boer, 1999), the gating of SKOR shows a pronounced sensitivity to voltage and external Kþ. It also shows a close similarity to
ª 2006 The Authors Journal compilation ª 2006 Blackwell Publishing Ltd, The Plant Journal, (2006), 46, 269–281
[Kþ]ext sensing of the SKOR Kþ channel 271 permeation with a selectivity that discriminates among the alkali cations in favor of Kþ, Rbþ and Csþ over Liþ and Naþ. The effect of external [Kþ] on SKOR could be interpreted as an interaction of Kþ ions with a channel-intrinsic cation binding site. As shown in triggered Kþ-diffusion experiments, this cation binding site was accessible from the outside even when the channel was closed (Supplementary Material, Figure S1). We used single-channel analysis to examine the gating kinetic components affected by Kþ. Because single-channel activity was lost rapidly upon patch excision (see also Liu et al., 2005), single-channel experiments were carried out in the cell-attached configuration. Channel lifetime analysis for 10-min recording segments in 3 and 30 mM Kþ at þ30 mV (Figure 2a,b) yielded open lifetimes that were well fitted by a sum of two exponential components but showed little effect of [Kþ] on the time constants. By contrast, closed lifetimes derived from the same recordings yielded two predominant exponential components for 3 mM Kþ and required an additional component with a relaxation time constant near 2 sec, to accommodate the data for 30 mM Kþ (Figure 2b). Similar results were obtained at )10 mV, with an increase principally in mean closed lifetimes of the longest-lived component and a decrease in the greater of the two time constants for the open lifetime (Figure 2c). Thus, in effect, increasing [Kþ] over this concentration range introduces a substantial, long-lived closed state of the channel and has little effect on the short-lived open and closed lifetimes of the channel. Mutations in S6 affect voltage- and Kþ-dependent gating Figure 1. SKOR gating is modulated by extracellular but not by intracellular Kþ. (a) Representative outward Kþ currents obtained from one oocyte injected with SKOR cRNA and recorded in 10 mM Kþ (K10). Currents were elicited by 2.5-sec voltage steps from a holding potential of )100 mV to voltages from )100 to þ50 mV (15-mV increments). (b) Comparison of current relaxations in 3 mM Kþ (K3) and in 30 mM Kþ (K30) at þ35 mV. The dashed line indicates the current trace measured in K30 scaled to the maximum current measured in K3. Activation halftimes were 560 ms in 3 mM Kþ and 1 sec in 30 mM Kþ. (c) Kþ-dependent shift in current activation. Steady-state currents were measured in 3 mM (triangles), 10 mM (circles), 30 mM (squares), and 100 mM Kþ (diamonds). The Nernst potential for potassium, EK, is indicated for all four conditions. (d) Relative conductance determined as described in the Data analysis section. Symbols are as in (c). (e) Rbþ and Csþ substitute for Kþ in modulating gating, but not Naþ and Liþ. Steady-state currents were measured at the end of 2.5-sec voltage steps as in (a) with 3 mM Kþ and 97 mM Naþ (white triangles), 97 mM Liþ (gray triangles), 97 mM Rbþ (gray hexagons), and 97 mM Csþ (gray squares). (f) Relative conductances determined as described in the Data analysis section. Symbols are as in (e). The dashed line is the relative conductance in 100 mM Kþ. (g) Independence of SKOR from intracellular [Kþ]. Current–voltage characteristics were determined in SKOR-expressing COS cells at the end of 1.6-sec steps from a holding voltage of )100 mV. Currents were recorded in 10 mM (circles) and 150 mM (diamonds) external Kþ and normalized to values measured at þ80 mV. The internal (pipette) Kþ concentration was 150 mM (white and black symbols) and 75 mM (gray symbols). Data are for 9 and 17 cells, respectively. (h) Relative conductance– voltage characteristics. Symbols are as in (g).
Extracellular cations affect the activity of several voltagegated ion channels. Among these, components of mammalian Kv channel gating – so-called C-type inactivation – are influenced by extracellular [Kþ] (Baukrowitz and Yellen, 1995; Pardo et al., 1992). Amino acid substitutions at position 449, adjacent to the Shaker Kþ channel pore, strongly affect the kinetics of C-type inactivation (Lopez-Barneo et al., 1993), and residues in the externally exposed fifth transmembrane domain (S5)-pore linker similarly have been implicated in gating of the HERG Kþ channel (Torres et al., 2003) and in Hþ- and Kþ-modulated activation of AKT3 (Geiger et al., 2002). Activation of Kþ channels belonging to the Drosophila melanogaster EAG subfamily is suppressed by elevating the concentration of Mg2þ and other divalent cations outside (Terlau et al., 1996), much as is SKOR by elevated [Kþ]. In this case, mutations in the third transmembrane domain (S3) and externally exposed (S3–S4) linker have been found to reduce or eliminate the Mg2þ sensitivity of the channel (Schonherr et al., 1999; Silverman et al., 2000). Finally, the yeast YKC1/TOK1 Kþ channel is sensitive to both intracellular and extracellular [Kþ] (Ketchum et al., 1995; Loukin and Saimi, 1999; Vergani et al.,
ª 2006 The Authors Journal compilation ª 2006 Blackwell Publishing Ltd, The Plant Journal, (2006), 46, 269–281
272 Ingela Johansson et al. KAT1, that do not show SKOR-like gating to help to pinpoint specific residues unique to SKOR (Figure 3). Mutations at over 30 different sites within these domains – including cysteine-scanning mutagenesis and single residue substitutions with corresponding amino acids from KAT1 – failed to yield evidence of an effect on the Kþ sensitivity of SKOR (Table 1). Besides mutations close to the selectivity filter of the pore, only mutations clustering about residue M313, within the S6 transmembrane domain, either modified or eliminated the Kþ sensitivity of SKOR gating. In each case,
Figure 2. Increasing extracellular Kþ concentration introduces a long-lived closed state of the wild-type SKOR Kþ channel. Cumulative open (left) and closed (right) lifetime analysis of wild-type SKOR was derived from 10-min recordings of single channels (cell-attached) at þ30 mV with 3 mM (a) and 30 mM Kþ (b) at the outer face of the membrane, and with 30 mM Kþ at )10 mV (c). Clamp voltage was corrected for the cell voltage of )40 mV (membrane voltage ¼ cell voltage ) pipette voltage). Representative trace segments are as shown; C: closed state; O: open state. Dwell-time histograms were fitted by non-linear, least-squares analysis to Eqn [2], yielding component time constants (si,
