Ca²⁺ influx via the L-type Ca²⁺ channel during tail current and above current reversal potential in ferret ventricular myocytes

Abstract

1. Current through L-type Ca²⁺ channels (I_Ca) was measured electrophysiologically at the same time as Ca²⁺ influx was measured by trapping entering Ca²⁺ with a high concentration of indo-1 (> 1 mm) in ferret ventricular myocytes.

2. Na⁺-free conditions prevented Na⁺–Ca²⁺ exchange and K⁺ currents were blocked by Cs⁺ and TEA. Thapsigargin (5 μm) prevented Ca²⁺ uptake and release by the sarcoplasmic reticulum. I_Ca was pre-activated by brief pulses to +120 mV (the equilibrium potential for Ca²⁺, E_Ca), followed by steps to different membrane potentials (E_m, −80 to +100 mV), in some cases in the presence of the Ca²⁺ channel agonist FPL-64176.

3. Integrated I_Ca (I_Ca) was linearly related to the change in the concentration of Ca²⁺ bound to indo-1, which was assessed by the fluorescence difference signal ΔF_d (F_d = F₅₀₀–F₄₀₀). This created an internal calibration of ΔF_d as a measure of Ca²⁺ influx.

4. The ΔF_d/I_Cadt relationship was virtually unchanged at all measurable inward I_Ca (at E_m from −80 to +50 mV). This indicates that the fractional current carried by Ca²⁺ and channel selectivity are unchanged over this E_m range, and also that the selectivity for Ca²⁺ is very high.

5. Ca²⁺ influx was readily detected by ΔF_d beyond the I_Ca reversal potential (+65 to +100 mV) and was not abolished until E_m was +120 mV (i.e. E_Ca). This is explained by the fact that inward Ca²⁺ flux at the I_Ca reversal potential is exactly balanced by outward Cs⁺ current through the Ca²⁺ channels and can be described by classic Goldman flux analysis with a Ca²⁺/Cs⁺ selectivity of the order of 5000.

6. This result also emphasizes that net Ca²⁺ influx via Ca²⁺ channels occurs over a voltage range where the net channel current is outward.

Ca²⁺ is a second messenger which controls vital events such as neurotransmitter/hormone release (Neher & Augustine, 1992; Zhou & Misler, 1995) and muscle contraction (Bers, 1991). Indeed, Ca²⁺ influx through voltage-dependent Ca²⁺ channels (VDCCs) links electrical activity to output in virtually all excitable cells.

Ca²⁺ influx through Ca²⁺-permeable channels depends not only on the total membrane current, but also on the fraction of the current that is carried by Ca²⁺. While total current can be measured by whole-cell patch-clamp recording, methods to measure fractional Ca²⁺ current through channels (P_f) have not traditionally been practical. Classical measurements of reversal potential using the Goldman-Hodgkin-Katz (GHK) equation with constant field assumptions, can provide only relative Ca²⁺ permeability ratios (e.g. P_Ca/P_Cs; Hille, 1992). The major limitation of GHK permeability ratios is that they do not provide information about actual Ca²⁺ flux, which is functionally important. Zhou & Neher (1993a) devised a method to measure the P_f of nicotinic ACh receptor channels using combined patch-clamp and Ca²⁺-sensitive fluorescence measurements. Since then, P_f values of several other Ca²⁺-permeable cation channels have been determined, including NMDA, AMPA, kainate and ATP receptor channels (Schneggenburger et al. 1993; Burnashev et al. 1995; Rogers & Dani, 1995), Ca²⁺ release-activated Ca²⁺ (CRAC) channels (Hoth, 1995) and cGMP-activated channels (Frings et al. 1995). Because Ca²⁺ influx through these channels has important functional roles (Jonas & Burnashev, 1995; Mollard et al. 1995), P_f is an important basic parameter (Neher, 1995).

In cardiac myocytes and neurons, the reversal potential (E_rev) for I_Ca is typically +50 to +70 mV (Bers, 1991; Hille, 1992), while the equilibrium potential for Ca²⁺ across the cell membrane (E_Ca) is typically +120 mV. Thus there may well be Ca²⁺ entry between a E_m of +50 mV and +120 mV that is associated with either no net current or net outward current. The amount of Ca²⁺ entry over this E_m range cannot be measured by voltage clamp, and consequently there is little direct information about how much Ca²⁺ may enter between E_rev and E_Ca. By using an optical indicator to monitor Ca²⁺ influx we can overcome this electrophysiological limitation.

Ca²⁺ influx via I_Ca occurs in virtually all excitable cells during action potentials. Furthermore, since a large fraction of Ca²⁺ entry during neuronal action potentials (80 %) occurs during repolarization as a tail current (Llinás et al. 1981a,b; McCobb & Beam, 1991; Elhamdani et al. 1998), it is particularly important to evaluate the selectivity of the Ca²⁺ channel at negative E_m. Tail Ca²⁺ current in cardiac myocytes can also trigger sarcoplasmic reticulum (SR) Ca²⁺ release (Cannell et al. 1987). Thus, it is important to measure Ca²⁺ influx and I_Ca to determine P_f at E_m from −80 mV (for tail currents) to +70 mV (close to E_rev). Because the Ca²⁺ channel opens only at positive voltage potentials and closes rapidly upon repolarization, it was not previously possible to directly measure P_f of VDCCs. Here we used the Ca²⁺ channel agonist FPL-64176 (Rampe & Lacerda, 1991) to overcome this limitation, and measured P_f and Ca²⁺ influx by indo-1 fluorescence at all physiological E_m in single ventricular myocytes (including potentials between E_rev and E_Ca).

METHODS

Cells and solutions

All experiments were carried out according to the guidelines laid down by the Loyola University Chicago animal welfare committee. Individual ventricular myocytes were prepared from hearts of adult ferrets, as described previously (Zhou et al. 1998). Briefly, the animals were anaesthetized by i.p. injection of sodium pentobarbital (50–75 mg kg⁻¹) and hearts were removed quickly and perfused with collagenase and pronase (Boehringer-Mannheim). Before experiments cells were placed in an experimental chamber in normal Tyrode solution (NT) containing (mm): 140 NaCl, 6 KCl, 1 MgCl₂, 2 CaCl₂, 10 glucose and 10 Hepes, pH 7.4. Na⁺-free NT (0 Na⁺) was the same except Na⁺ was replaced by TEA and Ca²⁺ was either 2 or 0 mm (with 1 mm EGTA) as noted in Results. To block Ca²⁺-induced SR Ca²⁺ release, cells were incubated in 0 Na⁺, 0 Ca²⁺ NT for 5–10 min with 5 μM thapsigargin before washout with standard 0 Na⁺, 2 mm Ca²⁺ and high TEA Tyrode solution. Figure 1A shows that caffeine-induced contractures after a series of loading pulses were abolished by thapsigargin exposure, confirming complete block of SR Ca²⁺ uptake by thapsigargin (Bassani et al. 1993).

Figure 1. Effects of thapsigargin and FPL.

A, thapsigargin block of SR Ca²⁺ release. In control conditions electrical stimulation of the ventricular myocyte was stopped and 25 mm caffeine was applied, activating SR Ca²⁺ release and a contraction. After the cell had been treated with 5 μM thapsigargin for 5 min the caffeine response was blocked, indicating complete block of SR Ca²⁺ transport by thapsigargin (n = 8). B, ramp potential-induced I_Ca without FPL. C, the same I_Ca ramp after addition of 1 μM FPL (n = 4). The small glitches in the I_Ca trace are due to small depolarizations (5 mV, 65 ms) used to monitor seal resistance.

The standard Na⁺-free, Ca²⁺-free internal pipette solution contained (mm): 80 caesium methanesulphonate, 40 CsCl, 1 MgCl₂, 10 Hepes, 0.3 GTP, 10 TEA and 1 potassium indo-1, pH 7.2. In some experiments, [indo-1] was reduced to 0.1 mm, as noted. Under these experimental conditions, all detectable ionic current is blocked by 1 μM nifedipine or 1 mm Cd²⁺ (Yuan et al. 1996). To increase Ca²⁺ current, particularly the tail current, 1 μM FPL-64176 (RBI, Natick, MA, USA) was added to the bath for parts of certain experiments.

Electrophysiology

Whole-cell patch clamp was used to record membrane currents as previously described (Zhou et al. 1998). In most experiments Ca²⁺ channel currents were the only membrane currents because of the specific experimental conditions: Na⁺ current and Na⁺–Ca²⁺ exchange are blocked by Na⁺-free solutions, K⁺ currents are blocked by internal Cs⁺ and external TEA, and SR Ca²⁺ release is blocked by thapsigargin. The very low intracellular free [Ca²⁺] ([Ca²⁺]_i; due to high [indo-1]) prevented Ca²⁺-activated chloride current (Zygmunt, 1994). Series resistance was always between 2 and 6 MΩ (mean = 4 MΩ) and was not compensated.

Cell shortening (ΔL) was measured by a video edge-detection system (Cresent Electronics, Sandy, UT, USA) as a bioassay of [Ca²⁺]_i (Zhou et al. 1998). Membrane currents, ΔL and indo-1 signals were simultaneously recorded by pCLAMP 6 (Axon Instruments Inc.) and analysed with a Macintosh computer using Igor Pro software (WaveMetrix).

Indo-1 fluorescence microscopy

The indo-1 fluorescence recording system (Zhou et al. 1998) was modified for maximum signal to noise ratio, and photobleaching was less than 10 % for all cell experiments. Fluorescence excitation was at 360 nm. Two different Ca²⁺-sensitive indo-1 emission signals were used for quantitative Ca²⁺ influx measurement. The first was a modified Ca²⁺-sensitive emission signal:

(1)

where F₅₀₀ and F₄₀₀ are emission signals at 500 ± 10 and 400 ± 10 nm, respectively (means ± full width at half-maximum (FWHM)). The second was the ratio R = F₄₀₀/F₅₀₀ used to calculate [Ca²⁺]_i (Grynkiewicz et al. 1985):

(2)

where R_min = 0.68, R_max = 6.31, β = 4.35 and K_d = 0.844 μM (J. W. Bassani et al. 1995).

To reduce the effect of system changes, a standard fluorescence ‘bead unit’ (BU) was used as a fluorescence standard for F_d (Zhou & Neher, 1993a):

(3)

where F_400,b and F_500,b are the F₄₀₀ and F₅₀₀ of a standard fluorescence bead (cat. no. 18340, lot 400103, Fluorescence B/B beads, 4.5 μm, Polyscience, Warrinton, PA, USA). The isocoefficient α was measured to make F_c = F₄₀₀+αF₅₀₀ independent of [Ca²⁺], thus reflecting only [indo-1]_i (Zhou & Neher, 1993b; Zhou et al. 1998). For our set-up, α = 1.0.

Assay of fractional Ca²⁺ current through Ca²⁺ channels

Fractional Ca²⁺ current, P_f (Zhou & Neher, 1993a), combines I_Ca (from patch clamp) and F_d (from fluorescence measurements). Under experimental conditions where Ca²⁺ buffering by indo-1 is 20 or more times larger than the endogenous Ca²⁺ buffer, > 95 % of the Ca²⁺ influx will be bound by indo-1, so that the change in F_d (ΔF_d) is proportional to Ca²⁺ influx (Zhou & Neher, 1993a; Zhou et al. 1998):

(4)

where I_Ca,Ca is the current carried specifically by calcium ions through the L-type Ca²⁺ channel, and f_max is a proportionality constant, or ‘maximum F/Q ratio’ under conditions where 100 % of Ca²⁺ influx is bound by intracellular indo-1. Then P_f is (Zhou & Neher, 1993a):

(5)

where I_Ca is total L-type Ca²⁺ channel current. For convenience we assume that P_f = 100 % for our standard calibration with pulses to +10 mV. To attain the appropriate conditions for eqn (4) and eqn (5), the pipette solution contained 1 mm indo-1 salt (without added Ca²⁺), and based on R measurements the basal [Ca²⁺]_i was ≤ 50 nM for all of the P_f experiments in this paper.

Theoretically, if constant field assumptions apply, the current through the Ca²⁺ channel can be broken down into current carried by Ca²⁺ (I_Ca,Ca) and current carried by Cs⁺ (I_Ca,Cs) such that I_Ca = I_Ca,Ca+I_Ca,Cs, or based on GHK theory (Hille, 1992):

(6)

where k = F/RT = 1/(25.7 mV) (where F is the Faraday constant, R the gas constant and T the absolute temperature), and P_Ca and P_Cs are permeability coefficients. [Ca²⁺]_o is 2 mm here and [Ca²⁺]_i is assumed to be zero (which makes little difference since outward current carried by Ca²⁺ is negligible). In experiments where 6 mm K⁺ was in the bath we used [Cs⁺]_o = 4 mm instead in calculations to simplify analysis and account for a P_K/P_Cs of 1.4 (Hess et al. 1986). The voltage dependence of P_f = I_Ca,Ca/(I_Ca,Ca+I_Ca,Cs) is (Schneggenburger et al. 1993; Burnashev et al. 1995):

(7)

Strategies to quantify Ca²⁺ influx at negative voltages

The L-type Ca²⁺ channel closes quickly at negative E_m. During the transition from open to closed, there is a large tail current reflecting the higher driving force at more negative E_m. Since the duration of the tail current is usually 1 ms or less, the integrated Ca²⁺ influx during the tail current is too small to be readily detected by indo-1 fluorescence. To determine the voltage dependence of P_f we needed to measure Ca²⁺ influx at all physiological E_m, including negative E_m. To increase the tail Ca²⁺ influx and ΔF_d, the Ca²⁺ channel agonist FPL-64176 (FPL) was used. In addition, to allow sufficient time for Ca²⁺ equilibration with indo-1 after Ca²⁺ influx, the voltage pulse protocol had three steps: (1) a pre-activation pulse (to +120 mV for 15 ms) to open Ca²⁺ channels; (2) a test pulse (test voltage and duration) for Ca²⁺ influx; and (3) a Ca²⁺ influx-terminating pulse back to E_Ca (+120 mV for 5–50 s) to stop Ca²⁺ influx and measure ΔF_d induced by the test pulse (see Figs 2–5).

Figure 2. f_max method in thapsigargin-treated myocytes.

A, saturation of the F/Q ratio (ΔF_d/I_Cadt) at f_max in a myocyte, as indo-1 dialyses in from the patch pipette (containing 1 mm indo-1). The [indo-1]_i was inferred from the Ca²⁺-independent F_c value (see Methods). At early times and low [indo-1]_i attention was focused on electrophysiological measurements (so few early f_max data were acquired). Ca²⁺ influx was induced by pulses from −80 to +10 mV during whole-cell voltage clamp. B, after [indo-1]_i > 0.5 mm, ΔF_d is proportional to Ca²⁺ influx and the F/Q slope is constant.

Figure 5. Ca²⁺ and contraction signals evoked by Ca²⁺ influx at E_rev.

A, contraction (ΔL), [Ca²⁺]_i and I_Ca induced by depolarization to +10 mV and E_rev (+70 mV). No FPL was in the bath. B, ΔL and I_Cadt signals upon depolarization from −80 to +90 mV (n = 4). All signals were normalized to that at 10 mV. All signals were evoked by 0.8 s depolarization pulses, except for +80 and +90 mV (1.6 s).

RESULTS

FPL increases tail Ca²⁺ current

Figure 1 shows voltage ramp-induced I_Ca in a cell before (Fig. 1B) and after FPL administration (Fig. 1C). Addition of 1 μM FPL caused three effects: (1) peak I_Ca increased severalfold (note current scales); (2) E_m corresponding to peak I_Ca shifted −18 mV; and (3) tail I_Ca increased in both amplitude and deactivation time, resulting in much more Ca²⁺ influx during the tail current. These observations are consistent with detailed square pulse analysis in rat by Rampe & Lacerda (1991).

The apparent E_rev also shifted in the presence of FPL (from 57.2 ± 1.0 to 70.7 ± 1.9 mV; P = 0.0015, Student's t test). This could reflect either imperfect block of outward Cs⁺ current through other channels (which is unlikely since current in these conditions was fully blocked by nifedipine or Cd²⁺), or increased selectivity with FPL for one or more open channel modes (i.e. P_Ca/P_Cs∼2 times higher). These may merit further exploration, but should not alter the key functional conclusions from the experiments described below.

ΔF_d is proportional to Ca²⁺ influx I_Cadt

Figure 2A shows that as cellular [indo-1] increases with time, the fraction of the Ca²⁺ influx bound by indo-1 reaches the maximum value (as all entering Ca²⁺ binds to indo-1 rather than endogenous Ca²⁺ buffers), as found in chromaffin cells (Neher & Augustine, 1992). Thus the F/Q ratio (ΔF_d/I_Cadt) becomes constant. Figure 2A shows that the F/Q ratio reaches a maximum value (f_max) when [indo-1] > 0.5 mm in a ventricular myocyte. The time constant for the rise in [indo-1] was typically 300 s (Zhou et al. 1998). In this case total Ca²⁺ influx should be directly reported by the ΔF_d (Zhou & Neher, 1993a). This is demonstrated in Fig. 2B, which shows that the ΔF_d signal is proportional to depolarization-induced Ca²⁺ influx charge. In this case I_Cadt was varied by changing pulse duration at +10 mV. Note the F/Q = f_max condition requires that [indo-1] > 0.5 mm and [Ca²⁺]_i≤ 0.25 μM (Neher & Augustine, 1992; Zhou & Neher, 1993b).

Figure 3 shows depolarization-induced Ca²⁺ influx in cells under voltage clamp. Under these conditions (see Methods) the ΔF_d signals should be due purely to Ca²⁺ influx through the Ca²⁺ channels. To assess Ca²⁺ influx at E_m from −80 to +120 mV, a prepulse to +120 mV for 15 ms was applied to activate the Ca²⁺ channel, but at E_Ca no Ca²⁺ flux was expected. Then steps to different test voltages allowed Ca²⁺ influx through open Ca²⁺ channels during tail Ca²⁺ currents at various E_m. Figure 3A shows an experiment without FPL. I_Ca and Ca²⁺ influx (ΔF_d) were measured by patch clamp and indo-1 signals, respectively. Three observations can be noted: (1) for a given pulse duration, the Ca²⁺ influx was maximum in the E_m range from +10 to +40 mV; (2) at the I_Ca reversal potential (+70 mV) where net membrane current is zero, there was still significant Ca²⁺ influx (about 20 % of maximum influx at +10 mV); (3) at −80 mV there was no readily detectable Ca²⁺ influx signal because the deactivation of the channel was too fast.

Figure 3. Voltage dependence of Ca²⁺ inflow through Ca²⁺ channels.

ΔF_d is scaled in both BU and corresponding Ca²⁺ influx units (eqn (4)). A, voltage dependence of I_Ca and ΔF_d without FPL (n = 20). B, similar experiment in a different cell where 1 μM FPL was added to the bath (n = 7). E_test, test pulses.

Figure 3B shows an experiment similar to that in Fig. 3A, except that 1 μM FPL was added to the bath to greatly increase Ca²⁺ influx and allow Ca²⁺ influx at more negative E_m. In this experiment, the pulse duration was increased at E_m values where Ca²⁺ influx was smaller (to increase ΔF_d). With FPL, channel deactivation was greatly slowed, so the Ca²⁺ influx during a tail current was large enough to be detected by ΔF_d even for a E_m of −80 mV. Again there was significant Ca²⁺ influx at the E_rev for I_Ca (+70 mV) and even at +90 mV. This indicates significant Ca²⁺ entry through Ca²⁺ channels at or near the I_Ca reversal potential. Since [Ca²⁺]_i was very low, the outward current was probably carried by monovalent ions passing through the Ca²⁺ channel. At the I_Ca reversal potential this outward I_Ca,Cs exactly balanced the Ca²⁺ influx (resulting in zero net current).

The existence of ion fluxes other than Ca²⁺ through the Ca²⁺ channels raises a question: what is the fractional Ca²⁺ current at a given membrane potential?Figure 4 shows the E_m dependence of Ca²⁺ influx and P_f signals. Figure 4A shows the time course of ΔF_d and I_Ca evoked by test pulses to −70 mV (left) and +10 mV (right). The delay between peak I_Ca and F_d was probably due to the diffusion and equilibration of calcium ions with indo-1 throughout the cell. Figure 4B shows ΔF_d and I_Cadt at different test E_m. With the same pulse duration, Ca²⁺ influx decreased for E_m > 20 mV. I_Cadt became zero at E_rev, while Ca²⁺ influx (ΔF_d) was still detectable. Figure 4C shows that P_f and the F/Q ratio were nearly constant until close to the I_Ca reversal potential. The nearly constant P_f up to +50 mV would be consistent with a GHK prediction (eqn (7)) with P_Ca/P_Cs = 5400. In summary, Fig. 4 demonstrates that I_Ca is carried almost exclusively by Ca²⁺ until fairly near the reversal potential. This is true even for the tail current upon repolarization to −80 mV.

Figure 4. Voltage dependence of fractional Ca²⁺ current (P_f).

A, depolarization-induced Ca²⁺ signals at −70 and +10 mV. B, ΔF_d and I_Cadt at −80 to +90 mV. Pulse duration (120–480 ms) is indicated. C, voltage dependence of P_f with curve based on eqn (7) (n = 7). The reversal potential was +75 mV for this cell.

Figure 5 shows that Ca²⁺ influx at E_rev can trigger significant cell contraction (ΔL) in a cell without FPL in the bath. To reduce exogenous Ca²⁺ buffers, only 0.1 mm indo-1 salt was in the patch pipette, but the cell was still pretreated with thapsigargin and was under Na⁺-free conditions. Figure 5A shows [Ca²⁺]_i and contraction signals at E_m of +10 and +70 mV. At E_rev, both Ca²⁺ and ΔL signals were about 25 % of the signals at +10 mV. According to Figs 3 and 5, the Ca²⁺ influx at E_rev is through Ca²⁺ channels. Figure 5B shows ΔL and Ca²⁺ current signals at E_m from −80 to +90 mV in the same cell. There was a detectable Ca²⁺ influx even at +90 mV, where net I_Ca was outward. In contrast, the tail current at −80 mV did not activate contraction because without FPL I_Ca deactivation was too fast to allow enough Ca²⁺ influx into the cell.

Figure 6A shows current-voltage curves for the Ca²⁺ channel. The total membrane current through the Ca²⁺ channel (I_Ca) was measured by patch clamp. Taking advantage of pure Ca²⁺ influx measured by ΔF_d, the fractional current carried by Ca²⁺ through the channel, I_Ca,Ca, was based on I_Ca,Cadt = ΔF_d/f_max (eqn (4)). Cs⁺ current (I_Ca,Cs) was taken as the difference between the experimentally determined I_Ca and I_Ca,Ca curves. Figure 6B shows theoretical calculations using eqn (6) and illustrates that the general behaviour of the I_Ca was consistent with GHK theory.

Figure 6. Current-voltage curves for L-type Ca²⁺ channels.

A, experimental recording of total Ca²⁺ channel current (I_Ca), Ca²⁺ influx (I_Ca,Ca based on ΔF_d/f_max), and net outward Cs⁺ current (I_Ca,Cs as the difference I_Ca–I_Ca,Ca; n = 5). B, GHK prediction of I_Ca, I_Ca,Ca and I_Ca,Cs for the ionic conditions used here, calculated using eqn (6) (with P_Cs = 3 × 10⁻⁸ cm s⁻¹, P_Ca/P_Cs = 4000, capacitance = 100 pF).

DISCUSSION

We have studied Ca²⁺ influx and Ca²⁺ permeability of L-type Ca²⁺ channels at physiological E_m. The major findings are as follows. (1) The Ca²⁺ permeability or fractional Ca²⁺ current was the same for depolarizing potentials (up to +50 mV) and repolarizing tail current (down to E_m = −80 mV). (2) Throughout this E_m range P_f was constant, consistent with very high Ca²⁺ selectivity (P_Ca/P_Cs > 5000). (3) There was significant Ca²⁺ influx through the Ca²⁺ channels at and above the I_Ca reversal potential.

Ca²⁺ influx during repolarizing tail current

Llinás et al. (1981a,b) predicted that most of the action potential-induced I_Ca in motor neurons is actually a repolarizing ‘tail current’. This was confirmed in motor neurons (McCobb & Beam, 1991) and adrenal chromaffin cells (Solaro et al. 1995; Elhamdani et al. 1998), where the action potential activates the Ca²⁺ channel, but I_Ca lags behind peak E_m, such that it coincides with repolarization, and tail I_Ca accounts for > 80 % of Ca²⁺ influx. Ca²⁺ influx during the tail I_Ca alone triggers transmitter release in neurons (Llinás et al. 1981a; Augustine et al. 1985; Neher, 1995) and SR Ca²⁺ release in myocytes (Cannell et al. 1987), emphasizing the relevance of tail I_Ca.

Total Ca²⁺ influx is determined by both I_Ca and P_f. For E_m at or above 0 mV, the Ca²⁺ permeability of the Ca²⁺ channels is very high (P_Ca/P_Na > 1000; Lee & Tsien, 1984; Hess et al. 1986). However, there was previously no direct measurement at negative E_m (e.g. −30 to −80 mV). P_f is E_m independent in NMDA and AMPA receptor channels (Schneggenburger et al. 1993; Burnashev et al. 1995), but not in nicotinic ACh receptor channels, where P_f is larger at a E_m of −80 mV than at −40 mV (Zhou & Neher, 1993a; Vernino et al. 1995). This was interpreted as either being due to a weak voltage sensor site in the pore which can affect the Ca²⁺ selectivity site, or that the GHK constant field assumptions do not apply. For cardiac I_Ca, P_f was nearly identical from −80 to +50 mV. This may be of minor consequence in cardiac myocytes where most Ca²⁺ influx occurs during the depolarized phase of the action potential. However, in neurons where Ca²⁺ entry occurs mainly during repolarization, altered channel selectivity at negative E_m could be more functionally relevant.

The fact that P_f was the same at −80 and +10 mV also suggests that Ca²⁺ selectivity is independent of the voltage sensor. Thus the site for the voltage sensor must be different from the sites for Ca²⁺ selectivity, as suggested by Yang et al. (1993) and Stühmer et al. (1989).

Ca²⁺ channel selectivity and GHK theory

L-type Ca²⁺ channel permeability ratios have usually been based on E_rev measurements under biionic conditions and GHK constant field assumptions (e.g. Lee & Tsien, 1982; Almers & McCleskey, 1984; Hess et al. 1986; Campbell et al. 1988). In contrast, the present study used fluorescence to measure Ca²⁺ influx and I_Ca simultaneously, so our findings are not dependent on GHK assumptions. Our data provide Ca²⁺ permeation information over a broad range of physiological E_m (from −80 to +100 mV) rather than simply around the reversal potential. One major finding is that I_Ca was carried almost exclusively by calcium ions at E_m from −80 to +50 mV. The current not carried by Ca²⁺ was detectable only at E_m within 10–15 mV of the reversal potential.

P_f is related to actual ionic flux, in contrast to classical GHK permeability coefficient ratios (P_Ca/P_Cs) near zero current. The data here are consistent with a P_Ca/P_Cs of ∼4000 (independent of E_m) for the cardiac L-type Ca²⁺ channel and agree with previous reports based on reversal potential measurements (Hess et al. 1986; Campbell et al. 1988).

Although the I_Ca data presented here qualitatively follow predictions from the GHK current equation and constant field theory (Figs 4 and 6), the behaviour is not precisely ideal. There are localized charges within the Ca²⁺ channel selectivity filter, variations in dielectric and also experimental limitations which could contribute to non-ideal behaviour in this context.

Ca²⁺ influx near apparent reversal potentials

In physiological solutions E_Ca is about +120 mV, while experimental I_Ca typically reverses at a E_rev of +50 to +60 mV with intracellular K⁺ or Cs⁺ (Fenwick et al. 1982). E_rev could be inappropriately taken as the E_m above which no Ca²⁺ influx occurs via the Ca²⁺ channel. However, according to the GHK equation Ca²⁺ entry is expected at E_rev and all the way up to E_Ca (Fig. 6). The present study provides direct quantitative measurement of Ca²⁺ influx in this E_m range (with Ca²⁺ influx at E_rev up to 25 % of that at 10 mV). The physiological implication of this finding is that a considerable amount of Ca²⁺ influx occurs even at the action potential peak in neuronal, endocrine and muscle cells. The point is that net outward current of K⁺ (or in our case Cs⁺) matches inward Ca²⁺ current at E_rev and exceeds it at more positive E_m. Thus to prevent Ca²⁺ entry via Ca²⁺ channels one must depolarize to the theoretical E_Ca of +120 mV or beyond. In cardiac excitation-contraction coupling studies this issue is particularly important because voltage clamp steps to positive E_m can be used to prevent Ca²⁺-induced Ca²⁺ release, but this requires voltage steps to E_m above the thermodynamic E_Ca (≥+120 mV) rather than the current E_rev (∼+60 mV).

Fractional Ca²⁺ flux in other Ca²⁺-permeable channels

Using P_f as a parameter for Ca²⁺ channel selectivity, we can classify Ca²⁺-permeable channels into three groups. High selectivity Ca²⁺ channels with a P_f near 100 % include VDCCs (Lee & Tsien, 1984; Hess et al. 1986; Campbell et al. 1988; present study), CRAC channels (Hoth, 1995) and cGMP-gated channels (Frings et al. 1995). Moderately selective Ca²⁺ channels typically have P_f values between 5 and 20 %. This group includes NMDA receptors (Schneggenburger et al. 1993; Burnashev et al. 1995), α7 nicotinic ACh receptors (Seguela et al. 1993), some subtypes of AMPA receptor channels (Burnashev et al. 1995) and ATP receptor channels in sympathetic neurons (Rogers & Dani, 1995). The low selectivity Ca²⁺ channels have P_f values < 5 %, and include most nicotinic ACh receptor channels (Zhou & Neher, 1993a; Vernino et al. 1995), most AMPA receptor and kainate channels (Schneggenburger et al. 1993; Burnashev et al. 1995), and ATP receptor channels in neuroblastoma cells (Z. Zhou, unpublished data).

Limitations of P_f measurements in ventricular myocytes

Pure Ca²⁺ influx can be measured by the fluorescence signal ΔF_d, since it is proportional to the number of indo-1 molecules that bind Ca²⁺ (Zhou & Neher, 1993a). While P_f measurement does not require GHK constant field assumptions, the assumed value of 100 %P_f here for inward I_Ca should technically be verified in pure Ca²⁺ solutions (not tested here). For these measurements one must also greatly exceed endogenous Ca²⁺ buffering (Berlin et al. 1994) and [Ca²⁺]_i should remain low (Zhou & Neher, 1993b), but with 1 mm[indo-1]_i these constraints should be met. All other ionic currents and Ca²⁺ transporters must ideally be blocked. We blocked all other known ionic currents, the SR Ca²⁺-ATPase and Na⁺–Ca²⁺ exchange. Sarcolemmal Ca²⁺-ATPase and mitochondrial Ca²⁺ uptake could be blocked by carboxyeosin (R. A. Bassani et al. 1995) and Ru360 (Zhou et al. 1998), but both activities are very slow (especially at [Ca²⁺]_i < 100 nM; Bassani et al. 1994). At E_m above +40 mV P_f becomes undefined because I_Ca in the denominator approaches zero, but the ΔF_d signal is still a direct measure of Ca²⁺ flux via the Ca²⁺ channel. Despite these limitations, the P_f and ΔF_d measurements of L-type Ca²⁺ channel fluxes provide valuable fundamental information.