Control of the hyperpolarization‐activated cation current by external anions in rabbit sino‐atrial node cells.

1. Effects of varying concentrations of anions on the hyperpolarization‐activated current (I(f)) were studied in myocytes isolated from the rabbit sino‐atrial node. Substituting Cs+ for the intracellular K+ clearly separated I(f) from the delayed rectifier K+ current. Control properties, including gating kinetics and ion selectivity, similar to previous studies were obtained. 2. Substitution of extracellular Cl‐ with larger anions including isethionate, glutamate, acetate, and aspartate, reduced the amplitude of I(f) without changing the reversal potential. Substitution with small anions such as iodide or nitrate supported an intact I(f). These effects were reproduced in the excised outside‐out patch conformation. 3. The conductance for I(f) was a saturating function of the extracellular Cl‐ concentration ([Cl‐]o) with an equilibrium binding constant (K1/2) of 11 mM and a slope factor of about 1 when substituted with large anions. Total removal of small anions completely abolished I(f). 4. The voltage‐dependent gating of I(f) was not affected by changing ([Cl‐]o), suggesting that Cl‐ modulates conductance properties of I(f). 5. The results indicate that I(f) conductance is unique in that it is dependent on an extracellular anion (Cl‐), yet it is carried exclusively by cations, K+ and Na+. These effects are independent of any measurable voltage‐dependent gating parameters.


INTRODUCTION
In various types of cells having a relatively low resting potential, hyperpolarization of the membrane induces an activation of an inward current (If or Ih). The cell types include cardiac pacemaker cells (Seyama, 1979;Yanagihara & Irisawa, 1980; DiFrancesco, Ferroni, Mazzanti & Tromba, 1986), atrio-ventricular node cells (Noma, Irisawa, Kokobun, Kotake, Nishimura & Watanabe, 1980) and Purkinje cells (DiFrancesco, 1981(DiFrancesco, a, b, 1982, photoreceptors (Bader & Bertrand, 1984), and neurones in spinal sensory ganglia, and central nervous system (Mayer & Westbrook, 1983;McCormick & Pape, 1990). As a common characteristic among different cells, MS 9822 I, shows a very slow time course of activation on hyperpolarization, relative to its quick deactivation on depolarization. The conductance of the If channel is characterized by its multi-ionic nature. The amplitude of If is increased by raising the extracellular K+ concentration ([K+]0) and decreased by reducing [Na+]0. The measurement of the reversal potential suggests that the current is carried by both K+ and Na' (DiFrancesco, 1981 a). However, since the early examination of If (Seyama, 1979), several studies have demonstrated that the current is also sensitive to alteration of [Cl-]0. Substitution of Clwith large anions reduces the amplitude of If (Yanagihara & Irisawa, 1980;Mayer & Westbrook, 1983;McCormick & Pape, 1990).
The finding may suggest that Clalso carries current through the I, channel under an assumption that the intracellular Clis depleted in the absence of external Cl- (Seyama, 1979). Alternatively, the finding may be explained by assuming a nonspecific blocking action of impermeable anions (McCormick & Pape, 1990), or by assuming an activation of the channel by Cl-(present study).
In recent years several groups have refined the isolation technique of single sinoatrial (SA) node cells and are compiling more data on If characteristics as well as its role in the pacemaker mechanism Yatani & Brown, 1990;Denyer & Brown, 1990;van Ginneken & Giles, 1991). We have also developed an improved procedure for isolating SA node cells, in general agreement with these groups. We have undertaken a whole-cell voltage clamp analysis on the activation of If and its dependence on anions by performing systematic ion-replacement experiments on isolated SA node cells.

SA nodal cell isolation
Albino rabbits weighing 1-2 kg were anaesthetized with intravenous injection of sodium pentobarbitone (30-50 mg/kg). The heart was dissected out and mounted on a modified Langendorff perfusion system and perfused with Tyrode solution followed by a Ca2+-free Tyrode solution for 3-5 min. After contractions ceased 300 ml of a Ca2+-free Tyrode solution containing collagenase (Yakult, Tokyo, Japan; 0-13-0-03 mg/ml, depending on lot number) was perfused followed by 150 ml of Ca2"-free Tyrode solution containing protease (Sigma, 0-1 mg/ml). The SA node region was then dissected out, trimmed, and cut into strips. A second Ca2+-free collagenase digestion (06-0-4 mg/ml) was then performed for 10-30 min. The digested SA node tissue was then washed in a 'KB' solution (Isenberg & Kl6ckner, 1982) and dispersed by triturating with a fire-polished pipette. Dispersed cells were stored in 'KB' solution at 4 0C for up to 1 day. Cells used in this study were identified visually as long, spindle-shaped cells approximately 50-80,m in length and 10-15 ,um in width in normal Tyrode solution. These cells have faint striations and a prominent, centrally located nucleus (similar to type A in Denyer & Brown, 1990). Other cell types were generally larger, rod shaped, and striated, or of rounded shape. The present cell-isolation procedure and solutions used provide an If that exhibits very little run-down in normal Na+or K+-Tyrode solution.

Voltage clamp recording and analysis
Whole-cell and isolated patch recordings were performed using the original methods of Hamill, Marty, Neher, Sakmann & Sigworth (1981). Currents were recorded with a patch clamp amplifier (List EPC-7, Darmstadt, Germany). The current, voltage, and trigger signals were stored on video tape using a PCM converter system (NF Electronic Instruments, RP-882, Tokyo, Japan) for subsequent computer analysis (NEC PC98 XL, Tokyo, Japan).
The liquid junction potentials of the aspartate-rich K+ and Cs' internal solutions were -11-4+0-3 mV (n = 10) and -12-3+0-7 mV (n = 7), respectively, when in contact with the control Tyrode solution. The measurement of junction potential drifted by a few millivolts when the electrode pipette was kept in the Tyrode solution, most probably due to diffusion at the pipette tip. We do not know exactly how large the junction potential was between the pipette solution and the intracellular solution, although it should be small. Therefore, voltage recordings measured in the aspartate-rich internal solution were corrected by -10 mV for practical purposes. To minimize junction-potential error during Clvariation a 3 M-KCl leakage electrode was used as a bath reference electrode. This reference system maintains a small, continuous leak of 3 M-KCl solution from a reservoir. The reference electrode was put downstream from the preparation. In other instances a large 3 M-KCl-agar bridge was used. Data are presented as means + standard deviation.

RESULTS
Isolation of the hyperpolarization-activated current using the Cs8-rich internal solution In nearly every spindle-shaped cell examined a large, inward current was activated on hyperpolarization negative to approximately -70 mV. After the initial break-in and voltage clamp, the current often decreased or increased to a steady-state level within minutes, presumably due to equilibration with the pipette solution, but the threshold potential for activation was not obviously changed. Rapid run-down of the current, as described in DiFrancesco et al. (1986), was not evident in normal Tyrode solution or in high-K+ solutions, presumably due to a higher intracellular Ca2+ of pCa 7 (Hagiwara & Irisawa, 1989) and inclusion of cyclic AMP and GTP (Yatani & Brown, 1990).
The activation of If on hyperpolarization is usually overlapped with a timedependent deactivation of the delayed rectifier K+ current. To isolate If, the delayed rectifier K+ current was blocked by adding 2 mm-Ba2+ and 1 mM-Ni2+ in the external solution, which also partially inhibited the time-independent background conductance. For complete suppression of the outward current of the delayed rectifier K+ channel, Cs+-rich internal solution was used. Figure 1A shows a family of If recordings in response to single hyperpolarizing steps. It is evident that the current trace at -60 mV shows no obvious time-dependent change and very slow activation of If is visible at -70 mV. At stronger hyperpolarizations If showed exponential activation at the onset of the pulse. With the Cs+-rich solution the sigmoidal time course of the initial current was not significant. The amplitude of the inward If tail, which was recorded on terminating the hyperpolarizing pulse, increased with increasing conditional hyperpolarization and showed a trend of saturation. Voltage-dependent activation of I was assessed by measuring the tail current. Two different current parameters including integration of the area of the tail current or measurement of the peak amplitude of tail current were used. The values were normalized to the maximum value in each experiment. Data obtained using the K+-rich solution in the pipette are also shown for comparison. We conclude that the activation curve is not affected by replacing internal K+ with Cs+. All data points in Fig. 1B were then fitted with the following Boltzmann relation: where Kact is the activation parameter, Q is valency of gating charge, Vm is the membrane potential, V1 is the potential at half-maximum activation, and F, R and T have their usual meanings in thermodynamics. Regardless of the major intracellular cation or means of measuring the current amplitude, the activation profile in Na+ Tyrode solution displays a threshold near -70 mV, a V of 93-1 + 8-4 mV, Q = 3 03 + 0 55 (n = 6; 4 intracellular Cs+, 2 intracellular K+). These data are shown in Fig. 1B with a representative fitted line.
The 'instantaneous' I-V relations were measured from a double-pulse tail current protocol as indicated in the insets of Fig. 2. Under physiological ionic conditions the tail current reversed from inward to outward with depolarization at -32-7 + 1-4 mV (n = 19), which is slightly less negative than the intersection of the instantaneous I-V curve. This is because the instantaneous I-V relation is also determined by other background conductances. The value of the reversal potential is in good agreement with previous studies if corrected for the junction potential (DiFrancesco, 1981 b;van Ginneken & Giles, 1991). The value of the Na+ conductance relative to the K+ conductance (pNa/pK) is 0-27 according to the Goldman-Hodgkin-Katz equation. With the Cs+-rich solution the tail current was always inward and the instantaneous C1-ACTIVATION OF CARDIAC If I-V relation was a smooth curve (Fig. 2B). Thus, it is concluded that internal Cs' neither carries outward current, nor affects the inward flux through the If channel.
In the following experiments, data obtained with the K+-rich solution were also used when the separation of potassium current (IK) from If was not critical in interpreting the experimental data. ig. 2. Late current (X) and 'instantaneous' current-voltage relationships (0) recorded in potassium aspartate (A) and caesium aspartate pipette solutions (B) with the control Na+ Tyrode solution. The late current was measured near the end of the hyperpolarizing pulse of the single-pulse protocol as shown in Fig. 1A. Current traces and the voltage protocols for the instantaneous I-V relationships are shown in the insets.

Effects of anion replacement
Substitution of the majority of extracellular Clin Na+ Tyrode solution by several organic anions was found to greatly reduce the amplitude of If, whereas substitution with small anions such as Ior NOsupported a full-amplitude If with properties indistinguishable from control. Figure 3A shows representative records comparing the currents between control external Cland NO-(upper panel) or control Cl-and isethionate (lower). The amplitude of If was reduced to about 30 % of the control in the isethionate solution, but the time course of the current was not changed. The effects of different anions on the current amplitude were examined by normalizing the I-V relation referring to the amplitude of If at -120 mV in the control. It was found, regardless of species, that all the organic anions tested reduced If to nearly the same extent, and that the current amplitude was almost identical with Cl-, I-, and NO- (Fig. 3B). The amplitude of the current in the organic anion solutions is about 30 % of the control over the potential range -120 to -90 mV, indicating no obvious voltage dependence of the effect of Clreplacement. These findings are difficult to explain by assuming a blocking action of organic anions (McCormick & Pape, 1990), since the molecular structure of these ions is different. No measurable Clpermeability was evident from experiments of replacing external Cl-. The reversal potential of If was measured by plotting the instantaneous The Clfrom CaCl2, MgCl2, NiCl2 and BaCl2 remains in the extracellular solution (10-6 mM). Currents were normalized to the peak inward current at the maximum hyperpolarized potential. The instantaneous current jumps at the pulse onset were assumed to be due to leak current systems and were subtracted. potential (Fig. 4B). Thus we conclude that Clis not permeating the If channel, which is a cation channel.
Dose-dependent potentiation of If by small anions An activating action by Cl-, rather than If channel block by organic anions, was strongly supported by recording If at very low [Cl-]0. The K+ Tyrode solution was used to obtain the maximum current in these experiments. On complete substitution of Clwith aspartate virtually no If was recorded, even with large voltage steps (Fig.  5A). Addition of only 2, 4, or 14 mM-Clto the external solution induced an If of progressively larger amplitudes during hyperpolarizing pulses (Fig. 5B). In these solutions changes in the concentration of organic anions is less than 10% of the control concentration. It is difficult to assume that decreasing the concentration of the organic anion from 148 to 136 mm progressively released the channel block.
The cell-free patch recording of channel current may be free from complications arising from cell-induced mechanisms, such as changes in cell volume or intracellular ionic concentrations and pH, stimulated by the Clreplacement. Recording of If from a cell-free patch has already been reported, initially by Yatani & Brown (1990). Therefore, the protocol was also performed using isolated outsideout patches. Although the If in this conformation underwent run-down, currents were rapidly reduced in 10 mM-Cland recovered in control Clsolution (Fig. 5C). Also, experiments were conducted with Clreplacing aspartate as the major intracellular anion ([Cl-]i = 140 vs. 24 mm in the normal control) with no change in the effect.
The findings described above strongly suggest that the external presence of small anions is essential for activation of the I channel by hyperpolarization. Dose dependence of this effect was quantified by measuring the steady-state current at  (1), is for the control K+ data. a single, large, hyperpolarizing test potential (-120 mV) over a wide range of [Cl-]o (substituting with glutamate or aspartate) and a dose-response relation was compiled (n = 32). The concentration dependence was fitted with a single-site binding relation with an equilibrium binding constant (Ki) of 11-5 mm and a Hill coefficient of 1-13 (Fig. 6).

Mechanisms of If depression in low-Clsolution
The reduction in the amplitude of If might be caused by a change in the voltageoperated gating kinetics, or by a reduction in the single-channel conductance in the low-Clsolution. We failed to record single-channel events in the cell-attached patch recording or in the excised outside-out patch, most probably due to extremely small single-channel conductance (DiFrancesco, 1986). However, we could test the gating kinetics at low [Cl-]j. Figure  on the voltage-dependent activation parameters over this concentration range. The in low Clwas 94 4 + 3 7 mV and Q was 2-53 + 0-56. 2 The rate of If activation was measured by plotting the current on a logarithmic scale. In Fig. 8, the ordinate indicates the current amplitude measured in reference to an assumed steady-state level. Since the current did not reach a steady level during the pulse, it was necessary to assume a steady-state level. To determine an appropriate value, the reference level was arbitrarily shifted until the semilogarithmic plot became linear near the end of the pulse as shown in the figure. By calculating the regression line over the late linear portion, the slow component was determined first. Then this slow component was subtracted from the original current, and the difference was replotted to measure the fast component. The sum of these two exponential components almost perfectly fitted the original current record as shown in the inset of Fig. 8. It should be noted that at small hyperpolarizations a single exponential often gave a good fit to the currents. On reduction of the current in 14 mm [Cl-]. the current kinetics are nearly unaffected with rate constants (T) of 0-38 and 1-67 as compared with the control values of 0-36 and 1-56 s. Essentially the same results were obtained in three experiments.

DISCUSSION
The results of this paper address the question of how substitution of extracellular Cl-by large anions reduces the size of If. This effect of Cl-substitution has been described, initially by Seyama (1979), followed by Yanagihara & Irisawa (1980) in cardiac SA node. More recently, a similar phenomenon has been described for the very similar If in spinal sensory ganglion neurones (Mayer & Westbrook, 1983) and in thalamic relay neurones (McCormick & Pape, 1990). These latter reports suggested a non-specific blocking activity for the large anions. On further Clremoval or on addition of only a few millimoles of Clin the presence of high concentrations of organic anions (Fig. 3), it is clear that If is directly dependent on small anions in the extracellular solution. Anions of <025 nm seem tobe able to support a normal current while larger, organic anions allow none. The response is a membrane-delimited one in that it is sustained in an excised-patch conformation. Also, intracellular Cldepletion was avoided in experiments containing high Clin the pipette solution.
Dual involvement of anions and cations in a single-conductance system is not a novel result, having been described for anion permeability of gramicidin channels (Neher, 1975) and for cation permeability of an anion channel in hippocampal neurones (Franciolini & Nonner, 1987). However, each of these systems is found to have an experimentally measurable permeability of both anion and cation. We can find no measurable Clpermeability in the If system. No shifts in reversal potential are found after large changes in [Cl-]Q. Furthermore, removing Cl-from the bath solution for the cardiac and neuronal preparations would lead to an increase of inward current rather than a decrease, if anion permeation is a factor.
A model for binding of impermeant anions has been proposed to interpret changes in thallous ion conductance in the gramicidin channel (Eisenman, Sandbloom & Neher, 1978). The model proposes that small anions perform a screening role for cations bound at external sites of a multi-ion pore. The effects on gramicidin channel conductance are quite small in comparison with the present result. The cardiac If conductance was totally eliminated by completely omitting external Cl-. We postulate that such a type of positive charge screening, whether it be a fixed charge at a recessed site on the protein or a bound cation is a necessary step in If channel permeation by cations, such as Na' andK+. Access by the large organic anions to the screening sites may be sterically blocked by a small tunnel structure of the channel protein. The voltage-independent nature of this Cl-effect suggests that these interactions should be out of the membrane potential field in the case of the cardiac If channel. This view is consistent with the finding that the voltage-operated gating properties were not modified by varying the Clconcentration. The voltagedependent gating should depend on a molecular structure within the membrane potential field. The fact that voltage-dependent activation and If kinetics are unaffected by Clvariation suggests that the manner by which Clexerts its effect is by varying channel conductance. While a change in either unit conductance or open probability could produce such a macroscopic result, it seems likely that such a large change in open probability would also be reflected in either gating kinetics or activation of If, as it is in other ion-gated channels (for example the Ca2+-activated K+ channel, Barrett, Magleby, & Pallotta, 1982). Although a novel Cl-binding scheme which produces a functional channel in an all-or-none manner, without subsequent effects on channel properties, could be argued, we reason that a Cl-effect on If channel permeation is the most likely interpretation. Direct single-channel measurement of these parameters is required to test these hypotheses.
The present study succeeded in separating the activation of If from the deactivation of the IK by using Cs+ in the internal solution. It should be noted that the deactivation ofIK causes a time-dependent change of the whole-cell current in the same direction as the activation of If over the potential range positive to the K+ equilibrium potential. In the present study, the outward current through the delayed rectifier K+ channel was suppressed by replacing the intracellular K± with Cs+, but the inward current through If channel was not blocked by the Cs+. The finding is in contrast to the block of the If channel with Cs+ from the external side (DiFrancesco, 1982;Noma, Morad & Irisawa, 1983). It may be speculated that the I, channel does not allow access of Cs+ from the internal mouth of the channel. The asymmetrical effect of Cs+ across the membrane is well established for the inward rectifier K+ channel (see, for example. Matsuda & Noma, 1984). If the ion selectivity of the I, channel in the presence of 5 4 mm-K+ in the external solution is determined by the Goldman-Hodgkin-Katz equation using the reversal potential of about -33 mV in the present study, the pNa/pK ratio is only 027. Thus, the If channel is also relatively selective for K+.