Amongst the nine voltage-gated K+ channel (Kv) subunits expressed in Arabidopsis, AtKC1 does not seem to form functional Kv channels on its own, and is therefore said to be silent. It has been proposed to be a regulatory subunit, and to significantly influence the functional properties of heteromeric channels in which it participates, along with other Kv channel subunits. The mechanisms underlying these properties of AtKC1 remain unknown. Here, the transient (co-)expression of AtKC1, AKT1 and/or KAT1 genes was obtained in tobacco mesophyll protoplasts, which lack endogenous inward Kv channel activity. Our experimental conditions allowed both localization of expressed polypeptides (GFP-tagging) and recording of heterologously expressed Kv channel activity (untagged polypeptides). It is shown that AtKC1 remains in the endoplasmic reticulum unless it is co-expressed with AKT1. In these conditions heteromeric AtKC1-AKT1 channels are obtained, and display functional properties different from those of homomeric AKT1 channels in the same context. In particular, the activation threshold voltage of the former channels is more negative than that of the latter ones. Also, it is proposed that AtKC1-AKT1 heterodimers are preferred to AKT1-AKT1 homodimers during the process of tetramer assembly. Similar results are obtained upon co-expression of AtKC1 with KAT1. The whole set of data provides evidence that AtKC1 is a conditionally-targeted Kv subunit, which probably downregulates the physiological activity of other Kv channel subunits in Arabidopsis.
Voltage-gated K+ (Kv) channels, which mediate sustained K+ influx or efflux across the cell membrane, are currently the best characterized of the K+ transport systems encoded by the Arabidopsis genome (Véry and Sentenac, 2003). These channels result from the assembly of four so-called α subunits related to the product of the Shaker gene identified 20 years ago in Drosophila (Papazian et al., 1987).
In animal cells, the four α subunits making a given Kv channel can be identical, but in many instances they are not. For example, amongst the nine subfamilies of genes that encode subunits forming depolarization-activated Kv channels in human cells, only five (Kv1, Kv2, Kv3, Kv4 and Kv7) encode subunits able to form functional homotetramers (Gutman et al., 2005). The genes from the four other subfamilies encode so-called ‘silent’ or ‘modifier’ subunits that must assemble into heterotetramers to be targeted to the cell membrane in a functional state (Hugnot et al., 1996; Ottschytsch et al., 2002; Post et al., 1996; Salinas et al., 1997). This ability of Kv channel subunits to form either homotetramers or heterotetramers increases the functional diversity of Kv channels produced by a given number of genes. For example, the cardiac Kv2.1 subunit can assemble with a number of other Kv subunits, the expression of which depends on the physiological conditions: each type of resulting heteromeric channel displays its own activation or deactivation kinetics, and/or voltage-dependence (Kerschensteiner et al., 2003; Patel et al., 1997; Vega-Saenz de Miera, 2004).
In Arabidopsis, the Shaker family comprises nine members, which can be segregated into five phylogenetic groups, considering both the encoded polypeptidic sequence and the gene structure (Pilot et al., 2003a). Genes from group I (KAT1 and KAT2, Véry et al., 1995; Pilot et al., 2001), group II (AKT1, Gaymard et al., 1996; SPIK, Mouline et al., 2002; and AKT5, not yet characterized) and group III (AKT2, Lacombe et al., 2000) can produce homotetrameric hyperpolarization-activated K+ channels when expressed alone in heterologous expression systems. In the same conditions, genes from group V (GORK,Hosy et al., 2003; and SKOR, Gaymard et al., 1998) can produce homotetrameric depolarization-activated K+ channels. On the contrary, AtKC1, the single member of group IV in Arabidopsis, initially appeared unable to form functional channels on its own.
Early attempts to determine the electrophysiological activity of AtKC1 channels in Xenopus oocytes (Dreyer et al., 1997) and COS cells (E. Michard, unpublished data) failed, leading to the hypothesis that AtKC1 is a silent Kv subunit, unable to form functional homotetrameric channels (Dreyer et al., 1997). Interestingly, when other plant Kv subunits of known functional characteristics (KAT1 and KST1) were co-expressed with AtKC1 in animal expression systems, their activity appeared to be altered, suggesting that they interacted with AtKC1 and formed heterotetrameric channels endowed with novel properties (Dreyer et al., 1997). More recently, patch-clamp experiments on Arabidopsis root hair protoplasts (Reintanz et al., 2002) provided further support to this hypothesis. Comparison of inward voltage-dependent current recorded in root hairs of wild-type, atkc1- and akt1- knock-out plants suggested that AtKC1 contributed to the inward K+ conductance in the wild-type plants, probably in association with AKT1 within heteromeric channels (Reintanz et al., 2002).
The reason why the AtKC1 Kv subunit is silent when expressed alone as well as the mechanism by which AtKC1 interacts with AKT1 channel activity are still unknown. We investigated these questions by using transiently transformed tobacco mesophyll protoplasts, a plant expression system shown to enable functional characterization of inwardly rectifying plant K+ channels, coupled with subcellular localization studies (Bei and Luan, 1998; Hosy et al., 2005). We show that AtKC1 polypeptides remain in the endoplasmic reticulum (ER) unless they are co-expressed with AKT1. In this case, heteromeric AtKC1-AKT1 channels are formed with functional properties that are different from those of homotetrameric AKT1 channels, and are probably dependent on the stoichiometry of the two subunits. Similar results were obtained upon co-expression of AtKC1 with KAT1, providing direct evidence that AtKC1 is indeed a silent, conditionally targeted and regulatory Kv subunit in Arabidopsis. Finally, the assembly of AtKC1-AKT1 heteromers would be favoured, with respect to that of AKT1 homomers, as suggested by comparing in silico simulations with experimental observations. Such a preferential heteromerization would potentiate the regulatory function of AtKC1 in the plant.
AtKC1 subcellular localization is changed by AKT1
When expressed alone in tobacco mesophyll protoplasts, the AtKC1::GFP fusion was retained in the ER (Figure 1a), whereas in parallel experiments the AKT1::GFP fusion was targeted to the plasma membrane, as indicated by the co-localization of FM4-64 staining (Figure 1b). Interestingly, when the AtKC1::GFP fusion was co-expressed with AKT1, the fluorescence shifted to the plasma membrane (Figure 1c). This suggested that some interaction between the two Shaker channel subunits AKT1 and AtKC1 occurred, and enabled the latter to escape from the ER and to reach the plasma membrane, probably as a component of heteromeric channels, together with AKT1.
Formation of AtKC1-AKT1 heteromers
The patch-clamp technique (whole-cell configuration) was used to study exogenous channel activity resulting from AtKC1 and/or AKT1 coexpression in tobacco mesophyll protoplasts. When expressed alone, AtKC1 did not change the plasma membrane conductance: AtKC1-expressing protoplasts displayed the same current pattern as control protoplasts transformed with the empty vector, i.e. low currents with weak dependence on time and voltage (‘no voltage threshold’, Figure 2a). In the same conditions, as previously reported (Hosy et al., 2005), AKT1 expressed alone formed functional channels enabling large time- and voltage-dependent currents to flow through the protoplast membrane upon hyperpolarization, characterized by a voltage threshold close to −40 mV (‘−40 mV voltage threshold’, Figure 2b, left). A variable feature displayed by the corresponding protoplasts was the current amplitude, probably resulting from differences in AKT1 expression level from one transformed protoplast to another.
All the protoplasts expressing both AtKC1 and AKT1 displayed time- and voltage-dependent currents, but, in most cases, the activation voltage threshold of these currents was more negative than −40 mV. This voltage threshold was obtained as follows. In each AtKC1/AKT1 co-expressing protoplast, and for each imposed voltage, we checked whether the recorded current was significantly larger than the endogenous current recorded in control protoplasts at the same voltage. Stepping from 0 mV (or −20, or −40 mV, in some instances) with a −20 mV decrement, the first voltage step for which this occurred was taken as the threshold for activation of the exogenous current. Activation voltage thresholds were thus estimated to within 20 mV and, with respect to this, AtKC1/AKT1 co-expressing protoplasts could be sorted into three categories, corresponding to threshold values of −40 mV (Figure 2b, right), −100 mV (Figure 2c) and −160 mV (Figure 2d), respectively.
AtKC1 subunits influence the gating of heteromeric channels
Average current–voltage curves with the different observed voltage thresholds are shown in Figure 2(e) (only the time-dependent current was considered in these plots). These different voltage thresholds for current activation corresponded to actual changes in the voltage dependence of the current, as evidenced by G/Gmax analyses (inset in Figure 2e).
If AKT1 and AtKC1 subunits are able to assemble in heteromeric channels, several combinations are possible a priori. Therefore, in protoplasts co-expressing AKT1 and AtKC1, the data (macroscopic currents and derived gating curves) probably reflect the activity of different types of heterotetramers rather than that of a homogenous population of a single type of channel. Thus, in such circumstances, the parameters (Gmax, z and Ea50) shaping the Boltzmann gating curves shown in Figure 2(e) (inset) are purely operational, but cannot be ascribed to a given type of channel. In contrast, the voltage threshold for the whole-cell current activation indicates that channels gating below this threshold (e.g. −100 mV) are numerous enough to yield measurable macroscopic currents, and that channels gating from a higher threshold (e.g. −40 mV) were absent or few in number, i.e. not detectable.
As a consequence of these drastic changes in voltage dependence, no exogenous outward hyperpolarization-activated and time-dependent K+ currents were recorded, in 1 mm external K+, in AtKC1/AKT1 co-transformed protoplasts displaying −100 mV or −160 mV voltage thresholds (Figure 2f). Such outward currents were recorded, however, in the AKT1 transformed protoplasts (Figure 2g).
Depending on which subunits were expressed, the occurrence of the current patterns (relative to the voltage threshold of the inward currents) varied as is summarized in Figure 3. AtKC1- and AKT1-expressing protoplasts displayed a single current pattern, ‘No voltage threshold’ and ‘−40 mV voltage threshold’, respectively. Thus, none of the AtKC1-expressing protoplasts displayed time- and voltage-dependent currents, and all the AKT1-expressing protoplasts displayed such currents below a threshold in the −40 mV range. AtKC1/AKT1 co-expressing protoplasts displayed one of the ‘−40 mV’, ‘−100 mV’ and ‘−160 mV voltage threshold’ current patterns, the first one being the less frequently observed (Figure 3).
We also investigated whether AtKC1 was able to interact with, and to change the channel activity of, a Kv subunit other than AKT1. KAT1 has been found to interact with the C-terminal part of AtKC1 in two-hybrid experiments (I. Chérel, personal communications). Therefore, KAT1 was co-expressed with AtKC1 in tobacco protoplasts, and the resulting heterologous channel activity was compared with the one resulting from the expression of KAT1 alone in the same context. The results (Figure S1) indicate that AtKC1 and KAT1 subunits are able to form heteromeric channels, with negatively shifted activation voltages when compared with homomeric KAT1 channels.
AtKC1 conditional targeting
GFP-tagging experiments such as those illustrated in Figure 1 may yield unexpected results. For example, the KAT1::GFP fusion was hardly seen on the plasma membrane of tobacco protoplasts (Hosy et al., 2005), whereas KAT1 currents could reproducibly be recorded in the same expression system (Figure S1; Li et al., 2002; Hosy et al., 2005). Regarding AtKC1, however, the absence of fluorescence localized to the plasma membrane in AtKC1::GFP-expressing tobacco mesophyll protoplasts, and the cytoplasmic reticular pattern of GFP distribution (Figure 1a), matched the systematic absence of exogenous current in AtKC1-expressing protoplasts (Figures 2a,e and 3). As it has already been reported that the expression of AtKC1 alone failed to yield exogenous current in Xenopus oocytes and COS cells (Dreyer et al., 1997; E. Michard, personal communications, respectively), one may assume that retention of this subunit in the ER makes it unable to form functional channels in the plasma membrane. That co-expression of AKT1 with AtKC1 was able to drive AtKC1::GFP to the membrane (Figure 1c), and that patch-clamped protoplasts co-expressing AtKC1 and AKT1 generally displayed currents that differed from those of AKT1 (Figure 2b–e), demonstrated that the two Kv subunits co-assemble into heteromeric channels with gating properties different from those of homomeric AKT1 channels. A comparable situation may be found in plant species other than Arabidopsis. For example, KDC1, a carrot homologue of AtKC1, has been shown to form heteromeric channels with DKT1, a carrot homologue of AKT1, when both subunits are co-expressed in Xenopus oocytes (Formentin et al., 2004).
The mechanism by which the interaction of AKT1 with AtKC1 may enable the latter polypeptide to be targeted to the plasma membrane is unknown. Tests of protein–protein interactions in the double hybrid system have shown that AtKC1 is not able to interact with itself, whereas it is able to interact with AKT1 (Pilot et al., 2003b). One can propose that AtKC1 is unable to form homodimers, and hence homotetramers, so that when expressed alone it remains in its monomeric form. The AtKC1 monomer could be retained in the ER possibly because it bears some special retention signal. When associated to Kv subunits of another type (AKT1, Figures 1 and 2; and possibly KAT1, Figure S1) within heterotetramers, AtKC1 subunits would have no such grounds for remaining in the ER (e.g. masked retention signal).
Different types of AtKC1-AKT1 heteromers
Shaker subunits first assemble in dimers, and Shaker channels are consequently formed by the association of two dimers (Tu and Deutsch, 1999; Urbach et al., 2000). If AtKC1 is unable to interact with itself, no AtKC1 dimers, but only AKT1-AKT1 and AtKC1-AKT1 ones, will be available for channel assembly in cells expressing both AtKC1 and AKT1. Subsequently, such cells (protoplasts) will have tetrameric channels incorporating from zero to two AtKC1 subunits, as illustrated in the three upper rows of Figure 4(a). As indicated in the fourth row of this figure, homomeric AKT1 channels have a −40 mV activation voltage. As the incorporation of AtKC1 subunits in heteromeric channels produces a negative shift of the activation voltage (Figure 2), it is reasonable to assume that heteromers with two AtKC1 subunits have more shifted activation voltages than those with a single AtKC1 subunit (for example ≤−160 and ≤−100 mV, respectively, as proposed in Figure 4a and discussed below).
Regarding the present macroscopic current data (Figures 2 and 3), a point of debate is that they have to be interpreted in terms of individual channel properties. In protoplasts expressing AKT1 as the sole heterologous gene, one can assume that a homogenous inward K+ channel population was found. This probably underlies the −40 mV voltage threshold displayed by such protoplasts. On the contrary, the macroscopic currents recorded in protoplasts expressing both the AtKC1 and AKT1 genes resulted a priori from the summed contributions of a large number of channels of the three types illustrated in Figure 4(a). The fact that those co-transformed protoplasts displayed different voltage thresholds (Figure 3) suggested, however, that the relative expression of the AKT1 and AtKC1 genes was not controlled in our conditions. A variable expression of the latter, for example, may have resulted in different populations of AtKC1-AKT1 heterotetramers. To evaluate this hypothesis, the respective number (as a percentage of the whole tetramer population) of each channel type obtained in AtKC1/AKT1 co-expressing protoplasts was computed as a function of a variable AtKC1 expression level (expressed in a percentage of the AKT1 expression level in the 0–150% range), assuming that dimers were formed randomly, and that tetramers then assembled randomly from those dimers. This computation was made considering three alternatives regarding the interactions between subunits within a dimer. Firstly, it was assumed that AKT1 has a better affinity for itself than for AtKC1. This was simulated by considering that AKT1-AKT1 dimers are ten times more stable than AtKC1-AKT1 dimers (i.e. preference for homodimerization; Figure 4b). In the second alternative, AKT1-AKT1 dimers were considered to be as stable as AtKC1-AKT1 dimers, [no preference for homo- or heterodimerization, i.e. (purely) random dimerization; Figure 4c]. In the third alternative, AtKC1-AKT1 dimers were considered to be ten times more stable than AKT1-AKT1 dimers (i.e. preference for heterodimerization; Figure 4d).
The results show that the proportion of homomeric AKT1 channels decreases from 100% (if only AKT1 is expressed) to a minimum (AtKC1 expression level reaching 150% of the AKT1 one). This minimum is in the 30% range in the case of a preference for homodimers (Figure 4b), in the 3% range for a purely random channel assembly (Figure 4c), and down to the 0% range in the case of a preference for heterodimers (Figure 4d). The proportions of the two kinds of heteromeric AtKC1-AKT1 channels vary as shown in Figure 4b–d. As discussed above, a −100 mV voltage threshold can only be observed if homomeric AKT1 channels are absent, or so scarce that the membrane conductance remains negligible down to −100 mV. The dashed line in Figures 4b–d marks a 10% threshold, below which the proportion of homomeric AKT1 channels must fall so as not to rank a AtKC1/AKT1 co-expressing protoplast in the ‘−40 mV voltage threshold’ category. This is because the population of homomeric channels (−40 mV voltage threshold) will mask the other populations unless it falls under this limit. The same goes for the −100 mV voltage-threshold population, which will mask the −160 mV voltage-threshold population, unless it declines to below this limit. As expected, the proportion of AKT1 homomers is not predicted to fall below this line (Figure 4b, unless huge AtKC1/AKT1 expression ratios are admitted), so that a −100 mV voltage threshold would never have been observed in the ‘preference for homodimers’ case. In the ‘random assembly’ case it is the proportion of heteromers with a single AtKC1 subunit that never falls below the line (unless an unfeasibly high AtKC1/AKT1 expression ratio is obtained), so that a −160 mV voltage threshold would never have been observed. Finally the observation of a −160 mV voltage threshold (Figures 2 and 3) is predicted in the ‘preference for heterodimers’ case only, when the AtKC1/AKT1 expression level ratio exceeds approximately 85%.
From these computations, it can be concluded that a random assembly scheme including a preference for homodimers (Figure 4b) cannot match the observed diversity in voltage thresholds (Figure 3), even when considering a highly variable expression level of AtKC1. The same holds for a purely random assembly scheme (Figure 4c). Thus, only a random assembly scheme including a preference for heterodimers (Figure 4d) matches the observed data (Figure 3).
Similar findings have recently been reported regarding the carrot Shaker subunit KDC1, an orthologue of AtKC1 (Naso et al., 2006), which preferably forms heterodimers when co-expressed with other Shaker subunits (this has been elegantly demonstrated by comparing the properties of channels resulting from the co-expression of KAT1 and KDC1, and from the expression of KAT1-KDC1 tandems). The activation potential of KDC1-KAT1 heteromers is shifted negatively with respect to the activation potential of KAT1 homomers (Naso et al., 2006). Following the present findings, preference for AtKC1-AKT1 dimers over AKT1-AKT1 dimers is expected to favour the inward-channel regulatory role of AtKC1 expression in plant cells (see below).
AtKC1 heteromerization with other Shaker subunits
The regulatory activity of AtKC1 is unlikely to be restricted to its interaction with AKT1. Indeed, several sets of data indicate that functional interactions can also occur with other Shaker channel subunits. The expression pattern of AtKC1 in the plant is rather widespread and overlaps clearly with that of several Shaker channel subunits, such as KAT1 in etiolated hypocotyls (data not shown; Philippar et al., 2004) and AKT2 in leaf epidermis (Pilot et al., 2003b). Two-hybrid experiments in yeast indicate that, besides AKT1 (Pilot et al., 2003b), several members of the Shaker family can physically interact with AtKC1: at least KAT1 (data not shown) and AKT2 (Pilot et al., 2003b). Co-expression of AtKC1 and KAT1 in tobacco mesophyll protoplasts evidenced a functional interaction between these two Kv α subunits (Figure S1). AtKC1 was found to induce a strong shift of the activation threshold of the inward currents towards a more negative value from ca. −100 mV (KAT1 expressed alone) to −160 mV (Figure S1), as it did in most AtKC1/AKT1 co-transformed protoplasts (Figures 2c and 3). Some AtKC1/KAT1 co-transformed protoplasts displayed no time- or voltage-dependent current in the whole voltage range explored (down to −220 mV): although GFP staining assessed a successful transformation, those protoplasts had the same current pattern as untransformed ones. This situation contrasted with AtKC1/AKT1 co-transformed protoplasts, which always displayed currents that differed from control protoplasts. It can be assumed that these protoplasts had AtKC1-KAT1 heteromeric channels with an activation voltage so negatively shifted that they failed to be activated in our experimental conditions: in those protoplasts, it is likely that when high enough the expression of the AtKC1 subunit completely suppressed the activity of KAT1 (in the physiological voltage range).
Physiological role of AtKC1
The fact that co-expression of AtKC1 with other plant Shaker genes was found to negatively shift the activation voltage of the resulting heteromeric channels, is consistent with previously reported observations in Arabidopsis root hairs (Reintanz et al., 2002). It has been shown that the absence of AtKC1 expression results in a shift towards more positive values of the activation voltage of the major membrane inward K+ conductance (Reintanz et al., 2002), which depends on AKT1 expression (Hirsch et al., 1998).
At the cellular level, the most obvious consequence of AKT1 regulation by AtKC1 is a strong reduction in the inward K+ current. Another consequence was revealed by patch-clamp experiments in bath solutions containing 1 mm K+, a physiological concentration. With 100 mm K+ in the pipette solution, the K+ equilibrium potential (EK) was in the −112 mV range. At membrane potential values between the activation threshold of AKT1 and EK (e.g. at −80 or −100 mV), significant outward currents could be recorded in protoplasts expressing AKT1 alone (Figure 2g; as reported for KAT1 in e.g. Véry et al., 1995), whereas in the same conditions no outward currents were recorded in protoplasts co-expressing AKT1 and AtKC1 (Figure 2f). Thus, the shift in activation threshold towards more negative values upon co-expression of AtKC1 could allow the cell to regulate its membrane inward K+ conductance in response to changes in external K+ concentrations, thereby preventing AKT1 from mediating K+ efflux. In other words, regulation of AKT1 by AtKC1 could decrease K+ influx or prevent K+ efflux, depending on the membrane potential and the external concentration of K+.
In conclusion, targeting to the plasma membrane of AtKC1 depends on its interaction with other α subunits of plant Shaker channels, such as KAT1 and AKT1. The data suggest that heterodimers are favoured over homomers during the process of channel assembly. This yields channels with altered voltage gating, requiring larger hyperpolarizations for opening. It is worth noting that the AKT1 gene shows little transcriptional regulation upon several changes in bulk ionic conditions, such as soil K+ availability and salt stress. In the latter case, although AKT1 expression remained unchanged, AtKC1 expression has been shown to dramatically increase in leaves, and especially in hydathodes (Pilot et al., 2003b), a cell type that also expresses AKT1. Thus, downregulation of the activity of the AKT1 gene product by AtKC1 may, for example, help the plant in facing salt stress.
Subcellular localization of channels fused to the GFP and detection of transformed protoplasts for electrophysiological measurements
Analyses of channel subcellular localization and electrophysiological activity were carried out using transiently transformed tobacco leaf mesophyll protoplasts, as described previously (Hosy et al., 2005). The protoplasts were incubated for 12–38 h at 19°C after transformation. Subcellular localizations of channel-GFP marker constructs, and of FM4-64 labelling, were performed using a Zeiss confocal microscope (LSM510 AX70; Carl Zeiss Inc., http://www.zeiss.com). Electrophysiological analyses were carried out in parallel using protoplasts from the same batch, but transformed using vectors (Hosy et al., 2005) allowing co-expression of one of the two channel subunits under investigation with GFP as a marker of the transformed cells. In this case, the GFP sequence was not fused to the sequence of a channel subunit, but was inserted separately in the transforming vector. To measure currents from only transformed protoplasts, an epifluorescent microscope allowing GFP detection was combined with a patch-clamp set-up. The GFP signal was detected between 489 and 508 nm using an emission filter (piston-GFP; Olympus, http://www.olympus.co.uk) upon excitation at a wavelength of 488 nm emitted by a monochromator (Optoscan C80x; Cairn Research Ltd, http://www.cairnweb.com; Hosy et al., 2005).
GFP-positive protoplasts were analyzed for their exogenous inward K+ conductance. Patch-clamp pipettes were pulled (P97; Sutter Instruments, http://www.sutter.com) from borosilicate capillaries (Kimax-51;Kimble, http://www.kimble.com) to obtain a pipette resistance of about 12 MΩ in the solutions used. Electrophysiological analyses were performed only when seals with resistances higher than 1 GΩ were obtained. Whole-cell recordings were made using an Axopatch 200A amplifier (Axon Instruments, http://www.axon.com). The pCLAMP 8 software (Axon Instruments) was used for voltage-pulse stimulation, online data acquisition and data analysis. The standard protocol consisted of stepping the membrane potential in −20-mV decrements to voltages from 0 to −200 mV, from a holding potential of 0 mV. In some cases, an offset of −20 or −40 mV was set on the amplifier to shift the accessible voltage range to −20/−220 mV or −40/−240 mV (as indicated in the figures). The pipette solution contained 1 mm CaCl2, 5 mm EGTA, 2 mm MgCl2, 100 mm K-glutamate, 2 mm MgATP, 10 mm HEPES–NaOH (pH 7.5), with its osmolarity adjusted to 520 mOsm with d-mannitol. The perfused bath solution contained 10 mm CaCl2, 2 mm MgCl2, 50 mm K-glutamate, 10 mm 2-(N-morpholine)-ethanesulphonic acid (MES)-HCl (pH 5.5), with the osmolarity adjusted to 500 mOsm with d-mannitol. The junction potential was corrected but no leak correction was made. G/Gmax analysis was made from the steady-state value of the time-dependent current (Lacombe and Thibaud, 1998; Véry et al., 1995).
Modelling tetrameric channel assembly from two types of monomers
The currents recorded in protoplasts co-expressing AtKC1 and AKT1 suggested that different kinds of AtKC1-AKT1 heterotetramers were formed (see Discussion). This prompted us to model the assembly of tetramers from a given stock of monomeric AtKC1 and AKT1 α subunits (AtKC1/AKT1 ratio on the x-axis in Figure 4b–d), to predict the subunit composition of the resulting channels, considering a number of assumptions, as described in Discussion. Briefly, once a dimer has been drawn, its fate was as follows: (i) any AtKC1 dimer was disassembled (‘no AtKC1 dimer’ hypothesis), (ii) an AKT1 homodimer (or an AtKC1-AKT1 heterodimer) was either kept for subsequent assembly of tetramers with an x (or a y) probability, or was disassembled with a 1 – x (or 1 – y) probability. Disassembled dimers yielded monomers available for new pairing. In the ‘preference for homodimers’ case (Figure 4b), x was 10 times y. In the ‘random assembly’ case (Figure 4c), x was equal to y. In the ‘preference for heterodimers’ case (Figure 4d), y was 10 times x. The absolute value of x (or y) had no consequence on the resulting dimer population: only the x/y value was of consequence. One thousand drawings for dimers were performed, the outcomes of which were averaged to yield a mean dimer stock. Assembly of tetramers was simulated in silico by drawing pairs of the previously obtained dimers. Any drawn pair of dimers was kept as a tetramer. One thousand drawings for tetramers were made and averaged to yield the results shown in Figure 4b–d.
EH was granted funding by INRA and Région Languedoc-Roussillon. GD was the recipient of a European Marie Curie Fellowship. CA was supported by the ‘ToxNuc-E’ French research programme. AC was the recipient of an ERASMUS fellowship. Drs Benoît Lacombe, Isabel A. Lefevre and Michel Vivaudou made useful comments on the manuscript.