Pore properties and ionic block of the rabbit epithelial calcium channel expressed in HEK 293 cells

1 We have used the whole‐cell patch‐clamp technique to analyse the permeation properties and ionic block of the epithelial Ca2+ channel ECaC heterologously expressed in human embryonic kidney (HEK) 293 cells. 2 Cells dialysed with 10 mM BAPTA and exposed to Ca2+‐containing, monovalent cation‐free solutions displayed large inwardly rectifying currents. Their reversal potential depended on the extracellular Ca2+ concentration, [Ca2+]o. The slope of the relationship between reversal potential and [Ca2+]o on a logarithmic scale was 21 ± 4 mV, compared with 29 mV as predicted by the Nernst equation (n= 3‐5 cells). 3 Currents in mixtures of Ca2+ and Na+ or Ca2+ and Ba2+ showed anomalous mole fraction behaviour. We have described the current‐concentration plot for Ca2+ and Na+ by a kinetic permeation model, i.e. the ‘step’ model. 4 Extracellular Mg2+ blocked both divalent and monovalent currents with an IC50 of 62 ± 9 μM (n= 4) in Ca2+‐free conditions and 328 ± 50 μM (n= 4‐9) in 100 μM Ca2+ solutions. 5 Mono‐ and divalent currents through ECaCs were blocked by gadolinium, lanthanum and cadmium, with a blocking order of Cd2+ >> Gd3+ > La3+. 6 We conclude that the permeation of monovalent and divalent cations through ECaCs shows similarities with L‐type voltage‐gated Ca2+ channels, the main differences being a higher Ca2+ affinity and a significantly higher current density in micromolar Ca2+ concentrations in the case of ECaCs.

as charge carrier and completely abolished by lowering extracellular Ca¥ to 50 nÒ, indicating that a Ca¥-dependent process inhibits ECaC activity. We have further shown that ECaCs become highly permeable to monovalent cations in the absence of extracellular Ca¥ . These findings point to some similarities between ECaCs and voltage-gated Ca¥ channels (VGCCs), which might be reflected in analogous permeation mechanisms. The aim of the present study was, therefore, to further investigate the cationic permeation mechanism of ECaC and its block by divalent or trivalent cations, and to describe the obtained data with a permeation model previously developed for voltage-gated Ca¥ channels.

Vector construction for ECaC-GFP co-expression
The open reading frame of rbECaC was cloned as a PvuII-BamHI fragment in the pCINeoÏIRES-GFP vector (Trouet et al. 1997;Vennekens et al. 2000). This bicistronic expression vector pCINeoÏIRES-GFPÏrbECaC was used to co-express rbECaC and enhanced green fluorescent protein (GFP).

Cell culture and transfection
All experiments were performed using ECaC-expressing HEK 293 cells. The cells were grown in DMEM containing 10% (vÏv) human serum, 2 mÒ ¬-glutamine, 2 U ml¢ penicillin and 2 mg ml¢ streptomycin at 37°C in a humidity controlled incubator with 10% COµ. HEK 293 cells were transiently transfected with the pCINeoÏIRES-GFPÏrbECaC vector using methods described previously (Kamouchi et al. 1999;Vennekens et al. 2000). Approximately 24 h after transfection, cells were used for experiments. Transfected cells were visually identified in the patchclamp set-up. GFP was excited at a wavelength between 450 and 490 nm and the emitted light was passed through a 520 nm longpass filter. The ECaC-expressing cells were identified by their green fluorescence and GFP-negative cells from the same batch were used as controls. Similar results were obtained with cells expressing only GFP and GFP-negative cells.

Electrophysiology
Electrophysiological methods have previously been described in detail . Whole-cell currents were measured with an EPC-9 (HEKA Elektronik, Lambrecht, Germany, sampling rate 1 ms, eight-pole Bessel filter 2·9 kHz) or an LÏM-EPC-7 (List Elektronics, Darmstadt, Germany) using ruptured R. Vennekens and others J. Physiol. 530.2 184 Figure 1. Ca¥ currents through ECaCs in the absence of extracellular Na¤ A, the effect of applying 10 mÒ Ca¥ (horizontal bar) during a 10 s step to −30 mV from a holding potential (Vh) of +20 mV, in the absence of extracellular Mg¥ and with Na¤ replaced by NMDG¤. ECaC-expressing HEK 293 cells were loaded with 10 mÒ BAPTA via the pipette in all experiments. B, current traces recorded in different Ca¥ concentrations (in mÒ) as indicated, in the absence of extracellular Mg¥ and with Na¤ replaced by NMDG¤, in response to 60 ms voltage steps to −140 mV from a Vh of +20 mV. C, currents recorded in different Ca¥ concentrations as indicated in the absence of extracellular Mg¥ and with Na¤ replaced by NMDG¤ in response to 50 ms voltage ramps from −100 to +100 mV with a Vh of +20 mV. D, pooled reversal potentials from 3-5 cells derived from voltage ramps as above. Currents were corrected for the capacitance current. The continuous line represents the fit of the data points with the Nernst equation (slope, 21 ± 4; n = 3-5).
patches. Electrode resistances were between 2 and 5 MÙ, and capacitance and access resistance were monitored continuously. The ramp protocol consisted of linear voltage ramps changing from −100 or −150 to +100 mV within 400 ms applied every 5 s. The step protocol consisted of a series of 60 ms voltage steps between +60 and −140 mV with a decrement of 40 mV. Holding potential was always +20 mV. Reported current densities were calculated from the current at −80 mV during the ramp protocol. Mono-exponential fits of currents were performed using the fitting routine of the WinASCD program (G. Droogmans, KULeuven). Dose-inhibition data were fitted to a logistic dose-response function using Origin 6.0 software (Microcal Software, Northampton, MA, USA).

Solutions
The standard extracellular (Krebs) solution contained (mÒ): 150 NaCl, 6 CsCl, 1 MgClµ, 1·5 CaClµ, 10 Hepes and 10 glucose, adjusted to pH 7·4 with CsOH. Na¤-free conditions were obtained using NMDG-Cl instead of NaCl. The concentration of Ca¥, Ba¥, Sr¥ or Mn¥ was varied between 1 and 100 mÒ as indicated in the text. In 'nominal divalent cation-free solutions', Ca¥ and Mg¥ were omitted from Krebs solution. In these conditions the concentration of free Ca¥ and Mg¥ was estimated to be in the order of 50 nÒ each by means of Fura-2 measurements. In order to remove divalent cations completely, 5 mÒ EDTA or EGTA were added. Various Ca¥-and Mg¥-containing solutions were prepared from this buffer as indicated in the text. The amounts of Ca¥ and Mg¥ to be added were calculated with the CaBuf program (G. Droogmans, KULeuven). The intracellular (pipette) solution in all experiments contained (mÒ): 20 CsCl, 100 caesium aspartate, 1 MgClµ, 10 BAPTA, 4 NaµATP, 10 Hepes, adjusted to pH 7·2 with CsOH.

Statistical analysis
Data are expressed as means ± s.e.m. Overall statistical significance was determined by analysis of variance. In case of significance (P < 0·01), individual groups were compared using Student's t test.

Ca¥ currents through ECaC in the absence of Na¤
We have previously reported that the ECaC, heterologously expressed in HEK 293 cells, is an inwardly rectifying Ca¥ permeable channel that shows a prominent monovalent permeability in divalent cation-free solutions . In order to demonstrate that Ca¥ is indeed the main current carrier in high Ca¥ solutions, we have measured currents through ECaCs in the absence of extracellular Na¤. Figure 1A illustrates the effect of adding 10 mÒ Ca¥ to a divalent cation-free solution containing 150 mÒ NMDG¤ instead of Na¤, during a 10 s step from +20 to −30 mV. The current amplitude shows a sharp increase followed by a rapid and reversible reduction, as was shown before in the presence of Na¤ . This decay process is Ca¥ dependent, but its mechanism is not yet understood. To minimize this decay process in Ca¥-containing solutions, we limited the exposure time to high [Ca¥]ï to 5 s and recorded a single current trace (step or ramp) before switching back to Ca¥-free solution. Figure  Currents recorded in response to a voltage step protocol from +60 to −140 mV in 40 mV decrements (Vh, +20 mV; duration, 60 ms), in the presence of replaced by NMDG¤. It is clear from this that we have to distinguish between (a) the relatively slow Ca¥-dependent current decay process (in Fig. 1A) and (b) the fast Ca¥dependent inactivation process (in Fig. 1B) (see also Vennekens et al. 2000). The amplitude of these currents increased upon increasing [Ca¥]ï with a concomitant shift of their reversal potential to more positive potentials, as illustrated by current traces in response to linear voltage ramps from −100 to +100 mV (duration, 50 ms) in Fig. 1C. Figure 1D shows pooled data of the reversal potentials of 3-5 cells as a function of the extracellular Ca¥ concentration. The fit of these points to the Nernst equation (continuous line) had a slope of 21 ± 4 mV (n = 3-5 cells) per 10-fold change in [Ca¥]ï, which is in good agreement with the theoretical value of 29 mV.
Anomalous mole fraction behaviour Figure 2 shows currents in response to a voltage step protocol applied to ECaC-expressing HEK 293 cells exposed R. Vennekens and others Table 1. Comparison of the step model for ion permeation for ECaCs and L-type Ca¥ channels Comparison of the parameters of the step model (as multiples of RT, where R is the gas constant and T the absolute temperature) applied to ECaCs and those for voltage-gated Ca¥ channels described in Dang & McCleskey (1998). Currents were normalized to the current value for the same cell in buffered divalent-free solution (1082 ± 164 pA pF¢ ranging between 295 and 3000 pA pF¢, n = between 10 and 13). The continuous line represents the current densities as predicted by a model with one high affinity binding site flanked by a low affinity binding site at each side, using the energy profiles for Ca¥ and Na¤ depicted in B. The dashed and dotted lines represent the fractions of the current carried by Ca¥ and Na¤, respectively. B, energy profiles of the ECaC pore along the path of the pore for Ca¥ and Na¤ (as multiples of RT; for details see Results). C, the predicted occupation of the ECaC pore by Ca¥ as a function of the Ca¥ concentration, i.e. the chance to find 1, 2, 3 or no Ca¥ ions bound within the pore (for details see Results section).
to Mg¥-free, 150 mÒ Na¤-containing solutions buffered with EGTA at extracellular Ca¥ concentrations ranging from 10 nÒ to 30 mÒ. It is obvious that the current amplitudes are reduced by increasing extracellular Ca¥ from virtually no Ca¥ to up to 100 ìÒ, but that a further increase of extracellular Ca¥ up to 5 mÒ or higher enhances it again. This finding is reminiscent of the anomalous mole fraction behaviour described for L-type voltage-gated Ca¥ channels in cardiac (Hess et al. 1986) and frog (Almers & McCleskey, 1984) muscle and for the Ca¥ release-activated Ca¥ (CRAC) channel (Hoth, 1995;Lepple-Wienhues & Cahalan, 1996). This anomalous mole fraction behaviour is even more evident if we plot the current amplitude as a function of the extracellular Ca¥ concentration (Fig. 3A, data from 10-13 cells, currents at −80 mV normalized to the corresponding current from the same cell in an EGTAbuffered Ca¥ solution, i.e. 1082 ± 164 pA pF¢; n = 13). It shows that 50% current inhibition occurs at a Ca¥ concentration of about 0·2 ìÒ, a value that is comparable to that of L-type Ca¥ channels (0·7 ìÒ Almers & McCleskey, 1984). In contrast, the current amplitudes at Ca¥ concentrations above 1 ìÒ are much larger than for voltagegated Ca¥ channels, which is consistent with a substantial contribution of Ca¥ to the ECaC current in this concentration range. The continuous line in Fig. 3A represents the current as predicted by the 'step' model, which was previously developed for L-type Ca¥ channels (Dang & McCleskey, 1998), using the energy profiles for Ca¥ and Na¤ depicted in of the central energy well for Ca¥ is −18RT, corresponding to a dissociation constant of 150 nÒ. The heights of the outer energy barriers, i.e. 9 and 10RT for Ca¥ and Na¤, respectively, are in the range expected for diffusional access to and departure of ions from the pore. Other parameter values are summarized in Table 1. The dashed and dotted lines in Fig. 3A represent the fractions of the current carried by Ca¥ and Na¤, respectively. Obviously, the current is mainly carried by Ca¥ at concentrations exceeding 1 ìÒ. The predicted occupation of the channel by Ca¥ as a function of the Ca¥ concentration is shown in Fig. 3C. The anomalous mole fraction behaviour was less pronounced for mixtures of Ca¥ and Ba¥ in nominal Mg¥-free conditions (Fig. 4). The pattern is essentially the same as reported for L-type Ca¥ channels, but a striking difference is the fact that the current carried by Ba¥ through ECaCs is much smaller than that carried by Ca¥ (Almers & McCleskey, 1984;Hess et al. 1986). Because of the limited number of data points, however, it was not possible to fit these data reliably to a kinetic model.

Block of mono-and divalent cation currents by divalent and trivalent cations
We have investigated the concentration dependence of the block by extracellular Mg¥ on currents carried by either monovalent cations in Ca¥-free solutions or Ca¥ in 100 ìÒ Ca¥-containing solutions (Fig. 5). Figure 5A shows the time course of a typical experiment at −80 mV during sequential application of various extracellular Mg¥ concentrations in the absence of extracellular Ca¥. Corresponding I-V curves obtained from voltage ramp protocols at various Mg¥ concentrations are shown in Fig. 5B in the absence of Ca¥ and in Fig. 5C in the presence of 100 ìÒ Ca¥. The current amplitudes were measured at −80 mV to avoid the timedependent component which is apparent in the presence of extracellular Ca¥ at negative potentials . Different Mg¥ concentrations were applied as indicated by horizontal bars (expressed in Ò). Linear voltage ramps as in B were applied every 5 s. B, representative current traces at different extracellular Mg¥ concentrations, as indicated, in the absence of extracellular Ca¥. Cells were loaded with 10 mÒ BAPTA through the patch pipette. Linear voltage ramps from −100 to +100 mV were applied every 5 s (Vh, +20 mV; t, 400 ms). C, representative current traces for the Ca¥ current at different extracellular Mg¥ concentrations, as indicated, in the presence of 100 ìÒ extracellular Ca¥. Cells were loaded with 10 mÒ BAPTA through the patch pipette. Ramp protocol as in B. D, dose-response curve of Mg¥ blocking monovalent and 100 ìÒ Ca¥ currents. Mean current densities at −80 mV were 1085 ± 73 pA pF¢ (n = 4) in buffered divalent cation-free solution (0) and 328 ± 68 pA pF¢ (n = 11) in 100 ìÒ Ca¥, 0 Mg¥ solution (8). ICÛÑ for inhibition of the monovalent current is 62 ± 9 ìÒ (n = 4), in comparison with 328 ± 50 ìÒ (n between 9 and 4) for the Ca¥ current. Currents were normalized to the corresponding value for the same cell in Mg¥-free solution.
normalized to the corresponding current amplitude in a Mg¥-free solution from the same cell (mean current density amounted to 1085 ± 73 pA pF¢ with n = 4 for divalent-free solutions buffered with 5 mÒ EDTA and 328 ± 68 pA pF¢ (n = 11) in the presence of 100 ìÒ Ca¥). The pooled data, as summarized in Fig. 5D, were fitted to a logistic dose-response function. The ICÛÑ values for the Mg¥ block (at −80 mV) were 62 ± 9 ìÒ (n = 4) for the monovalent cation current and 328 ± 50 ìÒ (n = 4-12) for the current mainly carried by Ca¥ ions. Cadmium is another divalent inorganic compound that is frequently used as a blocker of Ca¥ channels. It also blocked divalent currents through ECaCs (in the presence of 100 ìÒ Ca¥) with an ICÛÑ of 2·5 ± 0·6 ìÒ (n = 6). Monovalent currents on the other hand were more sensitive to Cd¥ and were blocked with an ICÛÑ of about 2 nÒ (data not shown). The trivalent cations GdÅ¤ and LaÅ¤ also inhibited ECaC currents. With 30 mÒ Ba¥ as the charge carrier ICÛÑ values were (at −80 mV) 1·1 ± 0·2 ìÒ (n = 13) and 4·6 ± 0·4 ìÒ (n = 11), respectively (Fig. 6). Dose-response curves as shown in Fig. 6C were obtained analogously as described above for Mg¥. At a concentration of 1 ìÒ both trivalent cations completely inhibited the current carried by monovalent cations in the absence of extracellular Ca¥ (not shown).

DISCUSSION
The epithelial Ca¥ channel (ECaC), is a Ca¥-selective channel, with a conductance sequence of Ca¥ > SrÂ¤ Ba¥ > Mn¥. The channel is constitutively opened and is subject to various Ca¥-dependent regulation mechanisms still to be elucidated . The Ca¥ permeability of the ECaC is confirmed in this work by the observation that the dependence of the reversal potential of the currents carried by Ca¥ in the absence of monovalent cations has a slope of 21 mV per 10-fold change in Ca¥ concentration, which is in fairly good agreement with the theoretical value predicted by the Nernst equation (i.e. 29 mV per 10-fold change in Ca¥ concentration). Despite the presence of 10 mÒ BAPTA in the pipette solution a Ca¥-activated Cl¦ current was activated upon addition of extracellular Ca¥ in about 10% of the ECaC-expressing HEK cells. Evidently these cells were omitted from the analysis.

Permeation model for the ECaC
Permeation through ECaCs shows similarities to permeation through other Ca¥-selective channels, such as voltage-gated Ca¥ channels (Almers & McCleskey, 1984;Hess et al. 1986) and CRACs (Lepple-Wienhues & Cahalan, 1996). As shown previously, ECaCs become permeable to monovalent cations upon removal of extracellular Ca¥ and Mg¥ . In the present paper we show that ECaC displays anomalous mole fraction (AMF) behaviour when mixtures of Na¤ and Ca¥ or Ba¥ and Ca¥ are applied. The fact that the AMF effect is not so pronounced in the case of Ca¥ and Ba¥ is, according to us, due to the fact that the difference in permeability through the channel for Ca¥ and Ba¥ is not so great as that for monovalents and Ca¥. Anomalous mole fraction behaviour is generally accepted as evidence for a channel pore containing multiple binding sites occupied by permeant ions moving in single file through the channel (Hille, 1992). Two distinct kinetic models have been developed to account for this anomalous mole fraction behaviour and high ion transfer rate of the Ca¥ channel pore in the case of L-type voltage-gated Ca¥ channels, i.e. the 'repulsion model' proposed by Almers & McCleskey (1984) and Hess et al. (1986) and the 'step model' by Dang & McCleskey (1998). The value of these models is that one can deduce general principles of ion permeation in the absence of an exact knowledge of the underlying molecular structure. In this work we have applied both models to our data but only for the step model were we able to deduce a set of parameters which provided a fairly good description of the data. The step model envisions a channel pore in which two low affinity binding sites flank a central high affinity binding site. In the absence of extracellular Ca¥ the channel pore is available for monovalent permeation through the channel. However, in the presence of [Ca¥]ï the binding sites within the pore will preferentially bind Ca¥, as a result of the higher binding affinity of the binding sites for Ca¥ compared with Na¤. From the Ca¥ occupancy plot it is clear that block of monovalent currents in nanomolar [Ca¥]ï occurs through binding of a single Ca¥ ion in the pore when [Ca¥]ï rises to the micromolar range. The Ca¥ flux at higher [Ca¥]ï is generated in parallel with multiple occupancy of the channel pore, although the chance of finding the ECaC pore in the triple Ca¥ occupied state is very low. Put in a more direct way, Ca¥ flux parallels the occupancy of the internal low affinity binding site. The drive for ion permeation results from the steps in binding affinity provided by the low affinity sites, as if the flanking sites provide stair steps for the ion to mount out of the channel pore (Dang & McCleskey, 1998). The most striking differences between the parameters for L-type Ca¥ channels and ECaCs are the height of the inner barriers for Na¤ and the slightly higher Ca¥ affinity of the central well in the case of ECaCs (Table 1). These features can account for the significantly higher current densities we measured in micromolar Ca¥ concentrations in the case of ECaCs, compared with L-type Ca¥ channels. The reasonably adequate description of our data by this model does not, however, guarantee the uniqueness of the derived parameters. Nevertheless our analysis clearly underscores the similarities in permeation properties of ECaCs and L-type voltagegated Ca¥ channels, properties which can be described by mechanisms which reconcile channel specificity and high Ca¥ fluxes in a multiple occupied, single-file pore through steps in binding energy. As for the structural meaning of the high and low affinity binding sites further molecular characterization of the pore region of ECaC is required.

Block of ECaCs by extracellular di-and trivalent cations
Previously we have shown that Mg¥ blocks monovalent currents through ECaCs in a voltage-dependent manner. It was deduced through a Woodhull analysis that the Mg¥ binding site is located within the channel pore, i.e. 31% within the membrane electrical field . In the current work we found a 5-fold difference between the ICÛÑ of Mg¥ block of monovalent (62 ìÒ) and Ca¥ currents (328 ìÒ). Furthermore in the presence of 2 mÒ [Ca¥]ï the ICÛÑ will shift even further to values up to 9 mÒ (data not shown). Such a difference in blocking capacity depending on the permeating cation seems to be a general property of the inorganic Ca¥ channel blockers we have used in the current work (see also Ellinor et al. 1995;Carbone et al. 1997 [Ca¥] from 0 to 100 ìÒ, rather than 5-fold as observed in our experiments. An alternative though speculative mechanism could be that binding of Ca¥ inside the pore influences the affinity of Mg¥ binding to its binding site. It is clear, however, that molecular identification of the Mg¥ binding site is necessary to further pursue this issue. On the other hand in the case of Cd¥ we indeed found a 10Å-fold difference between the ICÛÑ value in the presence of 100 ìÒ Ca¥ and in the absence of extracellular divalents. This finding therefore suggests that Ca¥ and Cd¥ compete for the same binding site within the ECaC pore. As a comparison, Cd¥ block of Ba¥ currents through voltage-gated Ca¥ channels (ICÛÑ, 300 nÒ in 40 mÒ Ba¥) is 300 times less potent than the block of Li¤ currents (ICÛÑ, 1 nÒ) (Ellinor et al. 1995), which further underlines the differences in pore properties between ECaCs and VGCCs. Currents through ECaCs are also blocked by LaÅ¤ and Gd 3+ , well-known potent blockers of Ca¥ channels (Hille, 1992). The blocking sensitivity for ECaCs is comparable to that described for CRACs (Hoth et al. 1993), whereas L-and Ttype Ca¥ channels are more sensitive (Ca¥ currents through T-type calcium channels: ICÛÑ values of 267 nÒ and 1·02 ìÒ for Gd 3+ and LaÅ¤, respectively, Mlinar & Enyeart, 1993). In contrast, analogues of the transient receptor potential channel, such as hTrp3 (Kamouchi et al. 1999) and dTrpL (Kunze et al. 1997), are less sensitive to LaÅ¤ and Gd