An Arginine Residue in the Pore Region Is a Key Determinant of Chloride Dependence in Cardiac Pacemaker Channels*

The modulation of ion channel activity by extracellular ions plays a central role in the control of heart function. Here, we show that the sinoatrial pacemaker current If is strongly affected by the extracellular Cl– concentration. We investigated the molecular basis of the Cl– dependence in heterologously expressed hyperpolarization-activated cyclic nucleotide-gated (HCN) channels that represent the molecular correlate of If. Currents carried by the two cardiac HCN channel isoforms (HCN2 and HCN4) showed the same strong Cl– dependence as the sinoatrial If and decreased to about 10% in the absence of external Cl–. In contrast, the neuronal HCN1 current was reduced to only 50% under the same conditions. Depletion of Cl– did not affect the voltage dependence of activation or the ion selectivity of the channels, indicating that the reduction of If was caused by a decrease of channel conductance. A series of chimeras between HCN1 and HCN2 was constructed to identify the structural determinants underlying the different Cl– dependence of HCN1 and HCN2. Exchange of the ion-conducting pore region was sufficient to switch the Cl– dependence from HCN1- to HCN2-type and vice versa. Replacement of a single alanine residue in the pore of HCN1 (Ala-352) by an arginine residue present in HCN2 at equivalent position (Arg-405) induced HCN2-type chloride sensitivity in HCN1. Our data indicate that Arg-405 is a key component of a domain that allosterically couples Cl– binding with channel activation.

The slow diastolic depolarization in sinoatrial node (SAN) 1 cells is the motor of cellular automaticity that drives the spontaneous beating of the heart (1)(2)(3). In addition, the slow diastolic depolarization is in the center of the autonomous control of heart rate because its kinetics is speeded up or slowed down by the release of norepinephrine and acetylcholine from sympathetic and parasympathetic nerve terminals, respectively. Several ionic currents have been proposed to control the time course and slope of slow diastolic depolarizations (e.g. T-type and L-type Ca 2ϩ currents and the sustained inward current I st ), some of which are directly regulated by the autonomous nervous system (4). Among these currents, the hyperpolarization-activated ("pacemaker") current I f plays an outstanding role. As the only current in SAN node, I f is dually gated by both hyperpolarized voltages and binding of cAMP, a second messenger, the concentration of which is directly controlled by the activity of ␤-adrenergic and muscarinic receptors. Owing to its dual gating behavior, I f contributes both to basal heart rate and to the autonomous regulation of rhythmicity. I f is carried by a family of four hyperpolarization activated-and cyclic nucleotide-gated channels (HCN1-4) (1,5). Genetic studies in human patients suffering from sick sinus syndrome (6,7) and the analysis of HCN4-deficient mice (8) indicated that the HCN4 channel is required for normal sympathetic stimulation of pacemaker activity. In contrast, HCN2 is not involved in autonomous heart rate control but probably provides a safety mechanism in SAN cells that prevents the diastolic membrane potential from becoming too negative in the presence of excessive hyperpolarizations (e.g. by increased K ϩ -channel activity) (9). In most species, HCN1 and HCN3 are expressed only in minor amounts in the SAN (1). The relevance of these subunits for heart function has not been determined so far. Given their key role for cardiac pacemaking and heart pathology, it is important to understand how the fine-tuning of HCN channels is achieved. Several mechanisms including the formation of heteromers (10 -15), the coassembly with auxiliary subunits (16 -18), the interaction with cellular scaffolding proteins (19,20) and protons (21,22), and the phosphorylation by protein kinases (23)(24)(25)(26)(27)(28)(29), have been implicated to regulate HCN channel activity in the heart. In the clinical setting, modulation of I f by serum electrolytes could be pivotal for cardiac pacemaker function. Indeed, there is experimental evidence that I f activity is sensitive to changes of extracellular Cl Ϫ concentration (30). It has been convincingly demonstrated that I f is a purely cationic current and that Cl Ϫ does not pass HCN channels (5,30). Nevertheless, Frace et al. (31) showed that the amplitude of rabbit sinoatrial I f decreased when extracellular Cl Ϫ was replaced by larger anions such as aspartate (31). This finding could have considerable clinical relevance since in hypochloremia, plasma Cl Ϫ concentration can drop to values as low as 60 mM (32)(33)(34). The inhibition of I f at low Cl Ϫ may well lead to the impairment of pacemaker function and, hence, to the induction of dysrhythmia.
In the last couple of years, genetic mouse models have opened novel perspectives to investigate the (patho)physiological role of ion channels in heart. Since it was not known so far whether or not murine I f is regulated by Cl Ϫ , we set out in this study to characterize in mouse SAN the modulatory effect of this anion. We then determined structural determinants underlying the Cl Ϫ dependence of I f in cloned cardiac HCN channels. To this end, we made use of our finding that HCN2 and HCN4 channels profoundly differ from the HCN1 isoform in terms of their sensitivity to Cl Ϫ. This diversity allowed us to use chimeras constructed between HCN1 and HCN2 to pinpoint individual amino acid residues underlying the Cl Ϫ effect. adjusted with KOH. The extracellular solutions were exchanged by a local solution exchanger. The different solutions reached the cell membrane within less than 100 ms. All recordings were obtained at room temperature. Data were acquired at 10 kHz using an Axopatch 200B amplifier and pClamp 8 (Axon Instruments). Voltage clamp data were stored on the computer hard disk and analyzed off-line by using Clampfit 8 (Axon Instruments). Current amplitudes were measured as the difference between the instantaneous current at the beginning of a 2.5-(HCN1 and HCN2) or 4-s (sinoatrial node cells and HCN4) hyperpolarizing step from a holding potential of Ϫ40 mV to Ϫ140 mV and the steady-state current at the end of the hyperpolarizing step. ] e )) ) in which is the Hill coefficient and K d is the apparent dissociation constant. For determination of half-maximum activation voltage (V 0.5 ), steadystate activation curves were determined by hyperpolarizing voltages (Ϫ140 to Ϫ30 mV) from a holding potential of Ϫ40 mV for 2.25 s followed by a step to Ϫ140 mV. Tail currents immediately after the final step to Ϫ140 mV were normalized to maximal current at Ϫ140 mV (I max ) and plotted as a function of the preceding membrane potential. The data points were fitted with the Boltzmann function: (I Ϫ I min )/(I max Ϫ I min ) ϭ [1 Ϫ exp([V m Ϫ V 0.5 ]/k)], where I min is an offset caused by a non-zero holding current, V m is the test potential, V 0.5 is the membrane potential for half-maximal activation, and k is the slope factor. Time constants of channel activation ( act ) of HCN1 and HCN2 channels were determined by monoexponential function fitting the current evoked during hyperpolarizing voltage pulses to Ϫ140 mV. As has been described earlier, the initial lag in the activation of HCN channel currents was excluded from the fitting procedure (36). For reversal potentials (E rev ), tail currents were recorded immediately after stepping to a family of test voltages ranging from Ϫ70 to 0 mV preceded by a 2.5-s prepulse to either Ϫ140 or Ϫ20 mV. The difference of tail currents resulting from the two prepulse potentials was plotted against the test potentials, and E rev was determined from the intersection of the I/V curve with the voltage axis (35). Changes of liquid junction potentials were either calculated using the JPCalc software (Dr. P. Barry, University of South Wales, Sydney, Australia (37)) or directly measured by a free-flowing KCl electrode. Liquid junction potentials were Ϫ7.4 mV for solution containing 130 mM aspartate, 3.1 mV for solution containing 145 mM Cl Ϫ , and 2.9 mV for solution containing 140 mM I Ϫ . These values were used for a a posteriori correction (38). For all experiments, agar bridges were used as reference electrode to minimize electrode offsets.
Isolation of Murine Sinoatrial Node Cells-Isolated SAN cells were prepared from adult C57BL6/J mice of either sex, aged 8 -12 weeks as described (9,39). The animals were treated in accordance to German legislation on the protection of animals. Single SAN cells were plated onto glass coverslips. For whole cell patch clamp recordings, 5 mM BaCl 2 was added to the recording solution; in experiments with 0 mM Cl Ϫ , 1 mM BaOH was added. Higher BaOH concentrations were not compatible with a pH of 7.4.
Statistics-All values are given as mean Ϯ S.E. n ϭ number of experiments. An unpaired student's t test was performed for the comparison between two groups. Significance was also tested by analysis of variance if multiple comparisons were made. Values of p Ͻ 0.05 were considered significant.

I f of Murine SAN Cells Is Highly Sensitive to Changes in Extracellular Cl
Ϫ Concentration-Spindle-shaped pacemaker cells were isolated from the SAN node region of adult mouse hearts and were analyzed using the whole cell patch clamp method (Fig. 1A). In extracellular solution containing a physiological concentration of Cl Ϫ (105 mM), nearly every cell investigated showed a prominent I f upon hyperpolarizing steps to Ϫ140 mV (Fig. 1B). The current was non-inactivating and reached steady-state values within 3 s. Increasing extracellular Cl Ϫ to 125 mM did not significantly affect the maximal current (I max ), whereas reduction to 85, 65, and 5 mM decreased I max by 9.9 Ϯ 1.9% (n ϭ 23), 21 Ϯ 3.8%., (n ϭ 4), and 54 Ϯ 4.2% (n ϭ 6), respectively (Fig. 1C). On complete substitution of extracellular Cl Ϫ by aspartate, the remaining current (I rem ) was only 12.6 Ϯ 2.1% (n ϭ 14) of the current evoked at 105 mM extracellular Cl Ϫ (Fig. 1, B and C).
HCN2-and HCN4-mediated Currents Are More Affected by Changes in Extracellular Cl Ϫ than HCN1 Currents-To investigate whether or not sensitivity to extracellular Cl Ϫ was also a feature of cloned HCN channels, we expressed the two cardiac HCN channel isoforms, HCN2 and HCN4, as well as the neuronal HCN1, in HEK293 cells. Fig. 2A shows representative current traces of the three channels obtained in 145 mM Cl Ϫ and in the absence of Cl Ϫ . Like the native cardiac I f , HCN2 and HCN4 currents were almost completely abolished (about 90% reduction) when Cl Ϫ was replaced by aspartate. In contrast, the maximal HCN1 current was suppressed by only about 50% in the complete absence of Cl Ϫ . Replacement of intracellular Cl Ϫ by aspartate or gluconate in inside-out patch measurements had no influence on HCN channel currents (data not shown). To exclude that the inhibition of I max was caused by a block induced by aspartate rather than reflecting a true Cl Ϫ depletion effect, HCN1 and HCN2 currents were also measured in the presence of gluconate, an anion that has a comparable size as aspartate but is structurally unrelated to it. The reduction of I max measured under these ionic conditions was virtually identical to that seen in aspartate solution. (HCN1, I rem ϭ 48.6 Ϯ 1.5%, n ϭ 7; HCN2, I rem ϭ 8.3 Ϯ 1.9%, n ϭ 4). There was no difference between the absolute magnitude of HCN1 and HCN2 currents recorded in either gluconate or aspartate solution (not shown). Taken together, these data indicated that I f is indeed controlled by extracellular Cl Ϫ .
The failure of the large anions aspartate and gluconate to compensate for the lack of the smaller Cl Ϫ suggested that the effectiveness of I f activation was determined by the molecular size of an anion. Indeed, the activity of HCN1 and HCN2 was fully maintained if Cl Ϫ was replaced by I Ϫ , an anion that is only slightly bigger than Cl Ϫ . Interestingly, currents in 145 mM I Ϫ were even somewhat bigger than in the presence of equimolar extracellular Cl Ϫ (increase of 8.4 Ϯ 1.2% (n ϭ 18) and 11.1 Ϯ 3.3% (n ϭ 10) for I max of HCN1 and HCN2, respectively (p Ͻ 0.05)). Characterization of the Cl Ϫ Dependence of HCN1 and HCN2-In the following, we set out to determine the structural determinants conferring the differences in the Cl Ϫ dependence between HCN1 and HCN2. We first determined the dose-response relationship of the Cl Ϫ -sensitive fraction of I max versus the extracellular Cl Ϫ concentration (Fig. 2B). Fitting the data to the Hill equation yielded apparent dissociation constants (K d ) of 45.6 and 32.7 mM for HCN1 and HCN2, respectively. The corresponding Hill coefficients were HCN1 ϭ 1.28 and HCN2 ϭ 1.39. We next investigated to which extent the decrease of extracellular Cl Ϫ influenced the biophysical properties of HCN1 and HCN2 currents. Reduction of Cl Ϫ concentration resulted in a slight but significant deceleration of opening kinetics in both HCN1 and HCN2 (Fig. 3, A and B, and Table I). By contrast, the extracellular Cl Ϫ concentration had no influence on the voltage dependence of HCN1 and HCN2 activation (Fig. 3, C and D, and Table I). Similarly, decreasing the Cl Ϫ concentration or replacing Cl Ϫ by I Ϫ did not alter the reversal potential E rev of the I/V curves for fully activated HCN1 and HCN2 channels (Fig. 3, E and F, and Table I). However, reducing the Cl Ϫ concentration decreased the slope conductance of the fully activated I/V curves (Fig. 3, E and F), whereas replacement of Cl Ϫ by I Ϫ increased it (not shown).
Exchange of the Pore Region Is Sufficient to Switch between HCN1-and HCN2-like Cl Ϫ Dependence-The finding that Cl Ϫ and I Ϫ affected the slope of the fully activated I/V curves indicated that small anions control the channel conductance of HCN1 and HCN2. We assumed that the amino acid residues conferring Cl Ϫ dependence were localized in the outer pore region of the channel, where they were involved in the control of ion permeation. To test for this hypothesis, we constructed four chimeric channels in which the region spanning from transmembrane domain S5 to S6 (including the pore) or only the pore region was exchanged between HCN1 and HCN2 (Fig.  4). Chimera (Chi) 1 and Chi2 containing the S5-S6 region or only the pore of HCN2 inserted into the backbone of HCN1, respectively, showed virtually the same sensitivity to depletion of Cl Ϫ as HCN2. The remaining current fraction in the absence of external Cl Ϫ was 7.8 Ϯ 1.5% (n ϭ 11) and 6.4 Ϯ 1.4% (n ϭ 8), which is not significantly different from the value of 6.5% determined for HCN2 (Fig. 2). Similarly, HCN2 acquired a HCN1-like Cl Ϫ sensitivity when its S5-S6 sequence (Chi3, I rem 46.6 Ϯ 1.7 (n ϭ 16)) or pore (Chi4 I rem of 39.2 Ϯ 1.7 (n ϭ 11)) was exchanged by the corresponding portions of HCN1.
Identification of Amino Acid Residues in the Pore Contributing to the Sensitivity to External Cl Ϫ -The sequence that was sufficient in Chi2 and Chi4 to switch the Cl Ϫ dependence from HCN1-type to HCN2-type and vice versa consists of 35 amino acid residues from which all but eight residues are identical between HCN1 and HCN2 (Fig. 5A). To determine the individual contribution of the eight residues for the Cl Ϫ dependence, we constructed a number of chimeras and channel mutants based on Chi4. Our goal was to define the minimal sequence that conferred HCN1-like Cl Ϫ dependence in the context of a HCN2 backbone.
In the chimeras Chi5 and Chi6, the first five and the last three amino acids of the eight different residues were exchanged by the corresponding segments of HCN1 (Fig. 5B). The I rem was 18.7 Ϯ 2.5% (n ϭ 10) for Chi5 and 37.6 Ϯ 1.6% (n ϭ 23) for Chi6. The I rem of Chi6 was almost identical to the I rem of Chi4 (39%) and was very close to that of wild-type HCN1 (47%), indicating that the three HCN1-derived residues present in this chimera were key determinants of the HCN1-type Cl Ϫ dependence. Conversely, the relatively low I rem of Chi5, which was close to that of HCN2 (10%), indicated that the five HCN1derived residues present in this chimera were less influential in inducing HCN1-like behavior. In support of this notion, chimeras in which only subsets of the five residues were exchanged revealed a I rem that was even closer to that of wildtype HCN2 (data not shown).
We sought to define the relative importance of the HCN1derived residues present in Chi6 (F389Y, R405A, and E409V) by analyzing another set of chimeras (Chi7-12; Fig. 5B). First, we replaced Phe-389 in the pore helix of HCN2 by the analogous tyrosine of HCN1 (HCN2 F389Y ; Chi7). The resulting I rem was 29.6 Ϯ 1.8% (n ϭ 29), which is in between the respective values of HCN1 and HCN2. This finding suggested that Phe-389 is influential in determining high Cl Ϫ sensitivity in HCN2.
In support of this notion, a chimera containing all HCN1derived amino acids present in Chi4 with the exception of the F389Y mutation (Chi8) showed a substantial decrease in I rem with respect to Chi4 (I rem(Chi8 ) ϭ 25.3 Ϯ 1.2% (n ϭ 11)). We next explored the role of position 409. A HCN2 channel containing F389Y and R405A mutations (Chi9) did not yield I f currents, indicating that the residue present at position 409 (the only residue in which Chi9 differs from Chi6) is critical for principal channel activity. However, the finding that Chi10 (HCN2 F389Y, E409V ) carrying a HCN1-type valine at position 409 had a somewhat lower I rem than Chi7 (HCN2 F389Y ) carrying a HCN2-type glutamate argues against a major role of position 409 in determining HCN1-and HCN2-type Cl Ϫ dependence. Finally, we replaced Arg-405 by the corresponding alanine present in HCN1 (HCN2 R405A ; Chi 11). Arg-405 is an attractive candidate for controlling Cl Ϫ dependence because it is adjacent to the ion selectivity filter, the narrowest part of the pore, and because its positively charged side chain might well interfere with the flux of cations. Unfortunately, Chi11 did not yield functional channels. Similarly, a chimera in which both Arg-405 and Glu-409 were mutated (HCN2 R405A, E409V ; Chi12) failed to express currents. However, when Ala-405 in Chi6 was back-mutated to arginine (HCN2 F389Y, E409V ; Chi10), I rem decreased from 37.6 Ϯ 1.6% (Chi6) to 17.9 Ϯ 3.5% (n ϭ 7). This result strongly indicated that Arg-405 is an important residue in conferring strong (HCN2-type) Cl Ϫ dependence. Interestingly, Arg-405 is also conserved in the pore of the highly Cl Ϫsensitive HCN4 channel, whereas this channel carries HCN1type tyrosine and valine residues at positions equivalent to residues 389 and 409 of HCN2 (Fig. 5A). To evaluate the role of Arg-405 directly, we introduced this residue into the backbone of HCN1 (HCN1 A352R ). Unlike the complementary HCN2 mutant (Chi11), this channel mutant expressed well and gave robust currents (Fig. 6A). Importantly, the current revealed the same Cl Ϫ sensitivity as wild-type HCN2 (I rem of HCN1 A352R ; 6.3 Ϯ 1.5%, n ϭ 12). In contrast, the mutation had no influence on the voltage dependence of activation ( Fig. 6B; V 0.5, act ϭ Ϫ74.2 Ϯ 5.5 mV (n ϭ 6)), nor did it alter the opening kinetics with respect to wild-type HCN1 ( act ϭ 65.4 Ϯ 13.9 ms (n ϭ 12)). Thus, in the backbone of HCN1, the exchange of Ala-352 by Arg was sufficient to achieve a complete switch of the Cl Ϫ sensitivity from HCN1 to HCN2-type behavior. DISCUSSION In this study, we provide evidence that replacement of extracellular Cl Ϫ by large anions induces a profound reduction of I max and a moderate deceleration of opening kinetics of both native and cloned pacemaker channels. In principal, this effect could result from the blockage of the channels by large anions as has been proposed previously (40,41). However, such a model, which requires that large anions bind to the channel molecule, is challenged by our observation that replacement of Cl Ϫ by either of two structurally unrelated large anions, aspartate or gluconate, inhibited I f equally well. Rather, our data support the model proposed by Frace et al. (31), claiming that Cl Ϫ acts as direct activator of I f channels, whereas large anions do not interact with the channel molecule. In agreement with this mechanism, I Ϫ , the molecular size of which (2.16 Å) is within the range of that of Cl Ϫ (1.81 Å), is also a full activator of I f .
The individual HCN channels could be classified into two groups with respect to the extent of their Cl Ϫ dependence. The native sinoatrial I f as well as the two major cardiac HCN isoforms HCN2 and HCN4 revealed only very low current activity when Cl Ϫ was depleted. By contrast, I max of HCN1 was reduced by only about 50% in the absence of Cl Ϫ . By constructing chimeric and mutant HCN channels, we identified the pore region as the crucial determinant controlling HCN1-or HCN2type Cl Ϫ dependence. Within the identified region, HCN1 and HCN2 differ from each other only in eight amino residues. Among these residues, the residue directly adjacent to the selectivity filter sequence was identified as the key determinant of Cl Ϫ sensitivity. The two channels with high Cl Ϫ sensitivity (HCN2 and HCN4) carry an arginine at this position, whereas HCN1 displays an alanine at equivalent position. Replacement of the alanine in HCN1 by arginine (HCN1 A352R ) was sufficient to endow HCN1 with HCN2-type Cl Ϫ sensitivity. Unfortunately, the reverse HCN2 mutant (HCN2 R405A , Chi11) did not express, preventing a direct assessment of the impact of this residue in the HCN2 backbone. However, analysis of double and triple mutants (Chi10 and Chi6) provided strong evidence that this residue is also functionally important in HCN2. In addition to Arg-405, a phenylalanine in the pore helix (Phe-389) was identified to affect the Cl Ϫ sensitivity in HCN2. However, since HCN4, like HCN1, carries a tyrosine at equivalent position, the impact of this residue may be only relevant in the context of the HCN2 backbone. The same is probably true for the sequence preceding the pore helix, which had a minor influence on Cl Ϫ sensitivity in HCN2 (Chi5). Again, in this region, HCN4 is more similar to HCN1 than to HCN2 (Fig. 5A), arguing against a general role of this domain in determining subtype specificity of Cl Ϫ dependence.
How may Cl Ϫ control the activity of HCN channels? The crystal structure of the HCN channel pore has not been determined so far. However, based on sequence alignments, one can assume that the pore is structurally related to that of the KscA potassium channel (42,43) (Figs. 5A and 7). On the basis of this structure, all eight amino acids that are different between HCN1 and HCN2 are localized within or close to the external portion of the pore and are probably localized outside of the electrical field. This is in line with the finding that Cl Ϫ does not affect the voltage dependence of channel activation. Two models could explain the effect of Cl Ϫ . In the first model, Cl Ϫ promotes channel opening by binding to an activatory domain that is allosterically coupled to the channel gate. Alternatively, Cl Ϫ binding could increase channel activity by relieving the impact of an inhibitory domain that tonically blocks channel conductance in the absence of small anions.
The question arises as to which role the identified residues fulfil in the allosteric regulation of HCN channel activity. The K d values and Hill coefficients deduced from the dose-response relation curves of the Cl Ϫ dependence were virtually identical for both HCN1 and HCN2. Thus, the binding site for Cl Ϫ is most likely conserved between both channels and is probably formed by amino residues of the pore that are identical in both channels. Although unlikely, it cannot be excluded that the binding site of Cl Ϫ is situated even outside of the pore. More- Our data suggest that Arg-405 is part of a regulatory domain that, in addition to the sensors for voltage (S4 segment) and cAMP (the cyclic nucleotide-binding domain), controls HCN channel gating. Since Arg-405 of HCN2 is localized close to the narrow opening of the pore, it may prevent a substantial flux of cations due to its positive charge (Fig. 7). Binding of Cl Ϫ may promote cation flux by inducing a conformational change that alters the relative position of Arg-405. Phe-389, and to minor extent, other residues in the proximal pore region, are probably involved in the coupling between Cl Ϫ binding and pore opening. In support of this model, HCN4, which, like HCN2, is highly Cl Ϫ -dependent, also contains an arginine at the position equivalent to Arg-405. By contrast, HCN1 carries an alanine instead of arginine at a position equivalent to Arg-405. The smaller size and the uncharged nature of the alanine may explain why HCN1 exhibits a substantial ion flux even in the absence of Cl Ϫ .  (42,43). The localization of the S5-and S6-transmembrane helices, the pore helix, and the selectivity filter are indicated on top of the alignment. The amino acid residues different between the three HCN channel types are highlighted. B, schematic representation of Chi4 -12.
Only the eight amino acid residues in the pore that differ between HCN1 and HCN2 are depicted. Residues are highlighted in black if derived from HCN1 or boxed if derived from HCN2. The I rem measured at 0 mM [Cl Ϫ ] e is given beside each chimera. Values are mean Ϯ S.E. with the numbers of experiments (n) shown at the right. n.c., no current upon six independent transfections. A potential role of the Arg-405 in channel gating is also supported by a recent study of Azene et al. (44), showing that this particular residue affects the modulation of HCN channel currents by external K ϩ . Unlike Azene et al. (44), we did not observe a hyperpolarizing shift of V 0.5 and a slow down in activation kinetics in the HCN1 A352R mutant. The reason for this discrepancy is not known but may be due to use of different expression systems (Xenopus oocytes in Azene et al. (44) and HEK293 cells in the present study).
With respect to its strong Cl Ϫ dependence, the sinoatrial I f closely resembles the HCN2-and HCN4-mediated currents. This finding corroborates previous studies, indicating that, at least in mouse and human, HCN2 and HCN4 represent the major molecular correlate of cardiac I f (45). The strong dependence of the sinoatrial I f on the extracellular Cl Ϫ could have significant clinical implications. We observed an about 10 and 20% reduction when Cl Ϫ decreased from its physiological concentration of 105 mM to pathological concentrations of 85 and 65 mM, respectively. In genetic mouse models, it has been shown that a reduction of total I f current amplitude of 20% is sufficient to induce sinus dysrhythmia (9,45). Moreover, such a mechanism could well contribute to arrhythmia occurring during hypochloremia associated with metabolic alkalosis (46,47). Finally, HCN channels are widely expressed in the central nervous system, where they fulfil various functions (2,48). Thus, a decrease in the activity of HCN channels could also be involved in the pathomechanism of important neurological disorders including absence seizures and ataxia (9,49).