Anion permeability and conduction of adenine nucleotides through a chloride channel in cardiac sarcoplasmic reticulum.

Cardiac sarcoplasmic reticulum (SR) membrane contains several chloride (Cl-) channels. We have characterized a 116-pS Cl- channel (500 mM cis, 50 mM trans Cl-) in cardiac SR that is activated by protein kinase A-dependent phosphorylation. To understand its function further, we examined the permeation of various anions and adenine nucleotides using the planar lipid bilayer-vesicle fusion technique. This Cl- channel showed a high selectivity to anions and its permeability sequence was Br- > Cl- > I- > NO3- > F-. When all anions were replaced with ATP in the cis solution, channel activity persisted. The conductance was 78 pS with 200 mM ATP and 68 pS with 100 mM ATP. The reversal potentials were +63 mV and +41 mV in 200 mM ATP and in 100 mM ATP, respectively. With 100 mM ADP or AMP in the cis solution, channel activities were also observed. The conductances were 87 pS with 100 mM ADP and 115 pS with 100 mM AMP. The apparent adenine selectivity of this channel was ATP > ADP > AMP, assuming that they exist as divalent anions. These results suggest that the SR Cl- channel in cardiac cells may serve as a transporter for the movement of adenine nucleotides between cytosol and SR lumen.

In cardiac as well as skeletal muscles, the SR 1 plays a central role in excitation and contraction coupling (1). Important molecular components for excitation-contraction coupling include the ryanodine receptor (RyR), Ca 2ϩ ATPase, ion channels including K ϩ channels (2), and several kinds of Cl Ϫ channels (3)(4)(5)(6)(7)(8). SR Cl channels have been reported to have conductances of 55 pS (cardiac SR in 260 mM Cl Ϫ ) (4), 200 pS (skeletal SR in 100 mM Cl Ϫ ) and 130 pS (cardiac SR in 250 mM Cl Ϫ ) (8). This latter channel is also voltage-sensitive. We have characterized a protein kinase A-activated Cl Ϫ channel in the cardiac SR with a conductance of 116 pS (500 mM Cl Ϫ in the cis, 50 mM Cl Ϫ in the trans solution) (5,6). Channel opening is regulated by PKA-dependent phosphorylation and by Ca 2ϩ -calmodulin, but insensitive to voltage or the levels of cytosolic Ca 2ϩ . It has been suggested that Cl Ϫ channels may function to maintain charge neutrality across the SR membrane generated by Ca 2ϩ movement during excitation-contraction coupling, but their functional characteristics are not well elucidated.
Recently, it has been suggested that certain types of anion channels in the plasma membrane conduct adenine nucleotides such as ATP (9 -11), but this suggestion has been contested (12)(13)(14). It has also been suggested that the mitochondrial voltage-dependent anion channel (VDAC) (15) may conduct adenine nucleotides. It is not known, however, whether the 116-pS Cl Ϫ channel in cardiac SR conducts ATP or not. In the present study, we examined the anion permeability and whether ATP might permeate through this Cl Ϫ channel or not. We found that this channel could conduct adenine nucleotides, ATP, ADP, and AMP. A preliminary report of this work was presented in abstract form (16).

EXPERIMENTAL PROCEDURES
Preparation-Porcine cardiac heavy SR was isolated by discontinuous sucrose gradient centrifugation as reported previously (5,6).
Solutions and Drugs-The cis bath solution contained (mM) CsCl, 500; EGTA, 1; HEPES, 10; MgATP, 2; and CaCl 2 , 5. The free Ca 2ϩ concentration was 1 M calculated with the method proposed by Fabiato (18). The trans solution contained (mM) CsCl, 50; EGTA, 1; HEPES, 10; and 1 M free Ca 2ϩ . The pH of these solutions was adjusted to 7.3 by adding CsOH. In the experiments to determine anion permeability, CsCl was replaced by 500 mM CsBr, CsF, CsNO 3 , or CsI. The 200 mM ATP solutions were made by replacement of 500 mM CsCl with 198 mM Tris ATP, Na 2 ATP, K 2 ATP, or Cs 2 ATP and 2 mM MgATP. In some experiments, 500 mM CsCl was replaced with 200 mM MgATP. The 100 mM ATP solutions were made by replacing 100 mM CsCl with 98 mM Tris ATP or Na 2 ATP and 2 mM MgATP. 10 M ryanodine (Wako Chemical Co., Osaka, Japan) was added to the cis solution to block the RyR channels. A protein kinase inhibitor (PKI) specific to PKA (Sigma P3294) was purchased from Sigma Chemical Co. Enzymes were dialyzed against the cis solution at 0°C for 1 h before use. DIDS (Sigma) was dissolved in dimethyl sulfoxide (Me 2 SO). Concentrations of Me 2 SO in final solutions were less than 0.1%.
Electrophysiological Methods and Data Analysis-The planar bilayer was composed of brain phosphatidylethanolamine and brain phosphatidylserine (Avanti Polar Lipids, Alabaster, AL) at a ratio of 1:1, dissolved in decane (20 mg/ml). Purified cardiac heavy SR vesicles were added to the cis chamber and fused into the lipid bilayer formed in the hole (0.25 mm in diameter) in a Lexan polycarbonate partition. In the present experiments, the cis chamber was defined as the side to which SR vesicles were added, and the opposite side was referred to as the trans chamber. The cis side was equivalent to the cytoplasmic side of the incorporated channel, and the trans side was equivalent to the lumen of the SR as determined previously (5,6). Currents flowing through the ion channels were measured by using the voltage-clamp technique. Applied voltages were defined with respect to the trans chamber held at ground. Channel activities were recorded at room temperature (22 Ϯ 1°C), amplified by a patch-clamp amplifier (Axopatch 1C, Axon Instruments, Inc, Foster City, CA), and stored on a videocassette tape recorder through a PCM converter system (RP-880, NF Instruments, Yokohama, Japan) digitized at 10 kHz. Data were reproduced and low pass filtered at 2,000 or 1,000 Hz by a filter with Bessel characteristics (octave attenuation, 48 dB) and analyzed off-line on a computer (P5-200, Gateway 2000). For single-channel analysis, the threshold used to judge to open state was set at a half-amplitude of the single-channel currents (19).

Permeation of Various Anions-
We have reported that a 116 pS Cl Ϫ channel in cardiac SR is activated via PKA-dependent phosphorylation or in the presence of MgATP (5, 6). Fig. 1, panel A, a, shows a continuous recording of Cl Ϫ channel openings with 500 mM CsCl in the cis and 50 mM CsCl in the trans chamber solutions. Because SR membrane contained not only Cl Ϫ channels, but also ryanodine receptor Ca 2ϩ release channels (RyRs) and K ϩ channels (2, 7), these channels were blocked by replacement of K ϩ with Cs ϩ and the application of 10 M ryanodine to the cis solution. The current amplitude of the channel remaining after this treatment was Ϫ15 pA at Ϫ40 mV. In the absence of MgATP, channel activity ran down spontaneously within several minutes after incorporation into the lipid bilayer. In the presence of 2 mM MgATP, however, channel openings were maintained until the experiments were interrupted by a break of bilayer ( Fig. 1, panel A, b). The open-time and closed-time histograms could be fitted by the sum of two exponentials (Fig. 5, panel A). The time constants for the open-time histogram were 1.5 Ϯ 0.2 (mean Ϯ S.D.) ms (n ϭ 6) and 43 Ϯ 4 ms (n ϭ 6). The time constants for the closed-time histograms were 0.7 Ϯ 0.1 ms (n ϭ 6) and 5.0 Ϯ 0.5 ms (n ϭ 5). Thus, the kinetic properties and channel conductance were identical to those of the 116 pS Cl Ϫ channel reported previously (5,6). These channel activities could be restored in the presence of 2 mM or higher MgATP or by adding of PKA with 0.05 mM MgATP. Therefore, we performed the following experiments with the cis solution containing 2 mM MgATP.
To test the anion selectivity, current-voltage relationships were obtained with various anions in both the cis and trans solutions. When Cl Ϫ was replaced with equimolar Br Ϫ , the unit amplitude of the current was not changed significantly (Fig. 1,  panel B). The slope conductance with Br Ϫ was about 116 pS, which was identical to that with Cl Ϫ . When Cl Ϫ was replaced with equimolar I Ϫ , NO 3 Ϫ , or F Ϫ , the unit amplitudes were decreased (Fig. 1, panel C). The slope conductances were 40 pS under these conditions. The reversal potential (E rev ) was approximately equal to the calculated equilibrium potential, which was ϩ58 mV (Fig. 1, panel C). From these results, we concluded that this SR-Cl Ϫ channel was highly permeable to these anions.
To determine the anion permeability ratio to Cl Ϫ , E rev was obtained under bi-ionic conditions with 500 mM anion in the cis solution and 50 mM Cl Ϫ in the trans solution (Fig. 2). The permeability ratios were calculated from the Goldman-Hodgkin-Katz equation.
where [X] cis and [X] trans are the cis and trans concentrations of anion, respectively. The values of E rev and the permeability ratios were summarized in Table I. The order of anion permeability (P x ) through the Cl Ϫ channel was determined to be Br   whether adenine nucleotides permeated this Cl Ϫ channel. After incorporation of Cl Ϫ channel into the lipid bilayers, 500 mM Cl Ϫ in the cis solutions was replaced with 200 mM ATP (198 mM Tris ATP and 2 mM MgATP). Fig. 3 shows a continuous recording of channel openings before and after replacement of the cis solution, where the membrane potential was held at Ϫ80 mV. After replacement of 500 mM Cl Ϫ with 200 mM ATP, the current amplitudes decreased (Fig. 3, panel A). We have often observed a subconductance level of the SR-Cl Ϫ channel in the Cl Ϫ solution and similar subconductance levels were recognized in the ATP solution as shown in Fig. 3, panel B.
To confirm the ATP conduction, current-voltage relationships were obtained at two different ATP concentrations in the cis solution and 50 mM Cl Ϫ in the trans solutions (Fig. 4). With 200 mM ATP, the slope conductance of the inward current was 83 pS (n ϭ 15), and E rev was ϩ62.6 mV. With 100 mM ATP, the slope conductance decreased to 68 pS (n ϭ 9) and E rev shifted to ϩ40.5 mV (Fig. 4, panel B). Under these experimental conditions, the only anion present in the cis solution was ATP. Therefore, inward currents could only be carried by ATP at negative potentials. It is difficult to estimate the theoretical E rev for the ATP current, because we do not know the actual free and complexed ATP concentrations under our experimental conditions. If ATP was assumed to exist as a divalent anion, P ATP /P Cl was 0.54 at 200 mM ATP and was 0.5 at 100 mM ATP calculated by the following equation (20), The value of E rev at each condition was almost identical to the theoretical value. From these results, we concluded that the currents were mostly carried by ATP 2Ϫ . The Kinetics of ATP Currents-As reported previously (5,6), the channel open-time and closed-time histograms could be fitted with two exponentials when currents were carried by Cl Ϫ (Fig. 5, panel A). By analyzing the currents carried by ATP, the open-time histograms could be fitted by a single exponential with time constants of 26 Ϯ 11 ms at Ϫ60 mV and 23 Ϯ 10 ms at Ϫ80 mV (n ϭ 4). The fast component of the open-time histogram observed when chloride was charge carrier seemed to disappear when ATP was the charge carrier. The closed-time histogram analyzed with a single-active channel could be fitted by a single exponential with time constants of 0.4 Ϯ 0.4 ms at Ϫ60 mV and Ϫ80 mV (n ϭ 4). The fast component was not different from those in the Cl Ϫ currents. Thus, the slow component of the closed-time histogram with chloride as charge carrier disappeared with ATP as charge carrier (Fig. 5, panel  B). In this analysis we chose the data in which only one channel existed in the bilayer and no subconductance levels were observed.
ADP and AMP Conduction through the Cl Ϫ Channels-We examined whether SR-Cl Ϫ channel might conduct other adenine nucleotides besides ATP. ADP and AMP were tested using the same approach as shown in Fig. 3, panel A. When the cis solution was replaced with 100 mM Na 2 ADP or Na 2 AMP for 500 mM CsCl, the channel activities were maintained (Fig. 6). Thus, the SR-Cl Ϫ channel also conducts these adenine nucleotides. We compared the slope conductances and relative permeabilities in bi-ionic conditions with 100 mM ATP, ADP, or AMP in the cis and 50 mM Cl Ϫ in the trans solution. The slope conductance was 68 pS in 100 mM ATP, 87 pS in ADP, or 115 pS in AMP. The reversal potentials were ϩ41 mV, ϩ20.6 mV, and ϩ3.0 mV, respectively (Fig. 7). The relative permeabilities (P x / P Cl ) are summarized in Table II.
Modulation of the Currents by Protein Kinase A-dependent Phosphorylation-If these currents carried by adenine nucleotides were activated via PKA-mediated phosphorylation as reported previously (5,6), currents should be blocked by removal of MgATP from the cis solution or by the application of PKI. After the ADP currents were activated in the presence of 2 mM MgATP, MgATP in the cis solution was removed. The ADP current was quickly and completely blocked (Fig. 8A), and restored by the reapplication of 2 mM MgATP (Fig. 8B). The application of PKI completely blocked the channel openings (Fig. 8C). Therefore we conclude that the anion channel-conducting adenine nucleotides can be activated via PKA-dependent phosphorylation.

DISCUSSION
Anion Selectivity of Cl Ϫ channel in the cardiac SR-The order of anion selectivity of this Cl Ϫ channel is consistent with the predicted Eisenman's sequence III (21). These results suggest that the electrodiffusion through this Cl Ϫ channel may be controlled by a so-called "weak field strength" selectivity site (21). The selectivity sequence of other Cl Ϫ channels in cardiac or skeletal muscle SR are different from this (3,8,22). The Ca 2ϩ -and voltage-sensitive Cl Ϫ channel in cardiac SR displayed the order of SCN Ϫ , which is consistent with sequence 1 of Eisenman (8). In skeletal muscle SR, the Cl Ϫ channels displayed the sequence of . Therefore, we speculate that SR-Cl Ϫ channel in this study may be a different type of Cl Ϫ channel than others previously described in SR. The cystic fibrosis transmembrane regulator (CFTR) expressed in epithelial cells or cardiac sarcolemma has an anion selectivity sequence similar to the 116 pS SR Cl channel (23,24). In addition, both CFTR and the 116 pS SR Cl Ϫ channel are activated by PKA-dependent phosphorylation ( Fig.  1) (5, 6), and exhibit voltage-independent activation (Figs. 1 and 2; Table I) (25). However, the conductance of the Cl Ϫ channel in cardiac SR is much larger than that of CFTR (Fig.  1). Therefore, it seems unlikely that the SR-Cl Ϫ channel is CFTR.
Permeation Pathway for Cl Ϫ and Adenine Nucleotides-The data in this report first demonstrate ATP currents in cardiac SR by recording single-channel activities (Figs. 3 and 4). Our results suggest that ATP is permeable through the same channel, which mediates Cl Ϫ permeation. First, the continuous channel activity before and after replacement of Cl Ϫ with ATP (Fig. 3, panel A) is consistent with this interpretation. The alternative, that new channels are incorporated into the bilayer as a consequence of ATP addition, seems highly unlikely. Furthermore, we could always detect the channel openings in  the ATP solutions, whenever the activity of Cl Ϫ channels incorporated into the bilayer was recorded (15/15). On the other hand, ATP currents were never observed when Cl Ϫ channels were not incorporated into the bilayer (10/10). Second, the anion channel requires phosphorylation to conduct either Cl Ϫ or ATP (Fig. 8). Third, the pharmacological properties are similar for both currents. We have reported that the SR-Cl Ϫ channel is insensitive to DIDS (5). Likewise, DPC or DIDS did not block ATP or ADP currents (data not shown). Therefore, we conclude that the 116 pS Cl Ϫ channel in cardiac SR can conduct both Cl Ϫ and adenine nucleotides.
Our data are in agreement with other reports indicating the existence of ATP conduction pathway in intracellular membranes (26 -31). It is known that VDAC (mitochondrial porin) is responsible for most of the metabolite flux across the mitochondrial outer membrane and also provides a pathway for nucleotide transport (28 -31). A working model of the VDAC pore proposes a barrel with a diameter of 2.4 -3.0 nM, and it has been reported that VDAC is sufficient to mediate ATP flux through the mitochondrial membrane (32,33). The molecular structures of ATP channels in a variety of intracellular membranes, including SR, have not been identified. Further studies are required to clarify the structural basis for the ATP conduction through the inner membrane channels.
In contrast to the inner membrane channels, there is no consensus regarding the existence of any ATP conduction pathway in sarcolemmal membranes. Although recent studies have suggested ATP conduction through CFTR or the multidrug resistance channel (9 -11), conflicting results have also been presented (12)(13)(14). Furthermore, it is suggested that the size of the ATP anion is much larger than the estimated size of the CFTR pore (34). A different view of these controversial findings have been presented by Pasyk and Foskett (26) and Sugita et al. (34), showing the existence of the CFTR-associated ATP channels in the plasma membrane by patch-clamp technique. Thus, an ATP conducting pathway appears to exist in the plasma membrane, but the identity of the responsible channels, whether they are the same or different from CFTR, are not clear.
Selectivity of Adenine Nucleotides-Based on the theoretical value of E rev at different ATP concentrations, our results were consistent with the assumption that ATP might move predom-inantly as a divalent anion (Fig. 4). Observed P ATP /P Cl ratios did not fit with calculations based on ATP as a monovalent (calculated P ATP /P Cl Ϫ ratio was 0.73 at 200 mM ATP and 1.01 at 100 mM ATP), tetravalent (P ATP /P Cl ϭ 0.527 at 100 mM ATP and 1.168 at 200 mM ATP) or trivalent (P ATP /P Cl is 0.462 at 100 mM ATP and 0.71 at 200 mM ATP) anion. This suggests that ATP permeation as ATP 4Ϫ , ATP 3Ϫ , or ATP Ϫ is unlikely. Sugita et al. (35) showed the ATP currents through CFTR-associated ATP channels with P ATP /P Cl ϭ 0.4. This value is almost identical to our results (0.5). We have also demonstrated the permeation of ADP and AMP through this Cl Ϫ channel (Figs. 6 and 7) and tried to determine the selectivity of adenine nucleotides (Tables I and II). It was quite difficult, however, to estimate precisely free concentrations of these anions in solution or those valences as charge carriers. If ADP and AMP were assumed to be present as a monovalent anion, P ADP /P Cl was 0.71 and P AMP /P Cl was 0.527. The order of apparent permeability became ADP Ͼ ATP Ͼ AMP. If they were assumed to be divalent anions, P ADP /P Cl was 0.25, and P AMP /P Cl was 0.14. Then the order of apparent permeability became ATP Ͼ ADP Ͼ AMP.
Physiological Implication-In the lumen of intracellular organelles, many processes may require adenine nucleotides for the functional source, and intraluminal ATP may be physiologically regulated. ATP is needed for energy-requiring processes (36) and is also the substrate for phosphorylation of intraluminal proteins in ER (37)(38)(39). Luminal ATP is required for protein translocation in the ER (40 -42). Therefore, we speculate that the SR Cl Ϫ channel may mediate the transport of ATP between lumen and cytosol, which may be responsible for important regulatory functions in cardiac excitation-contraction coupling.