Molecular Basis for the Different Activation Kinetics of the Pacemaker Channels HCN2 and HCN4*

The pacemaker channels HCN2 and HCN4 have been identified in cardiac sino-atrial node cells. These channels differ considerably in several kinetic properties including the activation time constant (τact), which is fast for HCN2 (144 ms at –140 mV) and slow for HCN4 (461 ms at –140 mV). Here, by analyzing HCN2/4 chimeras and mutants we identified single amino acid residues in transmembrane segments 1 and 2 and the connecting loop between S1 and S2 that are major determinants of this difference. Replacement of leucine 272 in S1 of HCN4 by the corresponding phenylalanine present in HCN2 decreased τact of HCN4 to 149 ms. Conversely, activation of the fast channel HCN2 was decreased 3-fold upon the corresponding mutation of F221L in the S1 segment. Mutation of N291T and T293A in the linker between S1 and S2 of HCN4 shifted τact to 275 ms. While residues 272, 291, and 293 of HCN4 affected the activation speed at basal conditions they had no obvious influence on the cAMP-dependent acceleration of activation kinetics. In contrast, mutation of I308M in S2 of HCN4 abolished the cAMP-dependent decrease in τact. Surprisingly, this mutation also prevented the acceleration of channel activation observed after deletion of the C-terminal cAMP binding site. Taken together our results indicate that the speed of activation of the HCN4 channel is determined by structural elements present in the S1, S1-S2 linker, and the S2 segment.

The pacemaker channels HCN2 and HCN4 have been identified in cardiac sino-atrial node cells. These channels differ considerably in several kinetic properties including the activation time constant ( act ), which is fast for HCN2 (144 ms at ؊140 mV) and slow for HCN4 (461 ms at ؊140 mV). Here, by analyzing HCN2/4 chimeras and mutants we identified single amino acid residues in transmembrane segments 1 and 2 and the connecting loop between S1 and S2 that are major determinants of this difference. Replacement of leucine 272 in S1 of HCN4 by the corresponding phenylalanine present in HCN2 decreased act of HCN4 to 149 ms. Conversely, activation of the fast channel HCN2 was decreased 3-fold upon the corresponding mutation of F221L in the S1 segment. Mutation of N291T and T293A in the linker between S1 and S2 of HCN4 shifted act to 275 ms. While residues 272, 291, and 293 of HCN4 affected the activation speed at basal conditions they had no obvious influence on the cAMP-dependent acceleration of activation kinetics. In contrast, mutation of I308M in S2 of HCN4 abolished the cAMP-dependent decrease in act . Surprisingly, this mutation also prevented the acceleration of channel activation observed after deletion of the C-terminal cAMP binding site. Taken together our results indicate that the speed of activation of the HCN4 channel is determined by structural elements present in the S1, S1-S2 linker, and the S2 segment.
Hyperpolarization-activated, cyclic nucleotide-gated cation (HCN) 1 channels are thought to underlie the native pacemaker current, termed I f or I h , in the heart and brain where it contributes to the rhythmic activity of cardiac and neuronal pacemaker cells (1)(2)(3)(4). All four members of the mammalian HCN channel gene family that have been cloned recently (5-9) share a highly preserved core region containing 6 transmembrane segments, including a voltage-sensing S4 segment and a pore region between S5 and S6, which is homologous to the S1-S6 region of voltage-gated potassium channels (K v ). The homology of the intracellular N and C termini within the HCN subtypes is less pronounced than for the core region with the exception of the highly conserved 120 amino acid long cyclic nucleotide binding domain (CNBD) starting about 80 amino acids downstream of S6.
All four HCN channels are expressed in cardiac tissue (5,10), but with some variations in the expression intensities among different species. HCN4 is highly expressed in the sino-atrial node. It is generally assumed that the slowly activating HCN4 contributes to the pacemaker activity and the modulation of the heart rate by ␤-adrenergic stimulation, whereas the less expressed, faster activated HCN2 and HCN1 may have additional functions such as maintaining the resting potential of pacemaker and other cells (11)(12)(13)(14).
All four HCN channels are activated upon membrane hyperpolarization. Activation is voltage-dependent, i.e. the more hyperpolarized the membrane becomes, the faster the channels open. The HCN channels differ, however, greatly in their activation kinetics with HCN4 being the slowest, HCN1 the fastest, and HCN2 and HCN3 as intermediate types (8 -10). Activation of HCN2, HCN4, and to a smaller degree that of HCN1 is accelerated by the binding of cAMP to the CNBD (5-10, [15][16][17]. Binding of cAMP to the CNBD releases the inhibition that the C-terminal region exerts on the channel (18).
It has been suggested that crucial components for the activation of HCN channels are within the S1, S1-S2 linker, S2, and S6 C-terminal regions (19,20). Exchanging these regions between HCN1 and HCN4 slows down and speeds up, respectively, the activation kinetics of these channels (19). To observe these effects, several parts of the channels had to be exchanged together, presumably because the difference between HCN1 and HCN4 is rather pronounced as can be seen not only from the basic activation kinetics but also from their different reaction to intracellular cAMP. The different responses to cAMP have been analyzed by HCN1/HCN2 chimeras (18,20). Similarly to the HCN1/HCN4 chimeras, parts of the C terminus and some as yet unknown transmembrane elements accounted for the activation kinetic differences. The channel subtype-specific cAMP modulation of the activation kinetics was always transferred with subtype-specific parts of the C terminus.
Although HCN2 and HCN4 differ considerably in their activation kinetics, they share a relative high sequence homology and are both modulated by cAMP in the same way. In preliminary experiments, we found that the difference in the speed of activation between HCN2 and HCN4 was independent of the binding of cAMP to the C-terminal CNBD. This result suggested that just a short sequence may be responsible for this inherent difference between HCN2 and HCN4.
HCN2 and HCN4 consist of 889 and 1203 residues, respec-tively (8). The difference is primarily due to the significantly longer C terminus of HCN4. In the transmembrane region S1-S6, the homology of the human HCN2 and HCN4 is 90%, leaving only a few amino acids that could account for the differences in activation kinetics not modulated by the influence of the N-and C-terminal regions. We identified 5 amino acid residues in the S1-S2 region that contribute to the activation kinetics of HCN2 and HCN4. One of these amino acids is responsible for the difference in activation kinetics between HCN2 and HCN4.

EXPERIMENTAL PROCEDURES
Molecular Biology-Human HCN2 and HCN4 cDNAs were originally cloned from the atrioventricular node region of a human heart (8). HCN2/4-chimeric channel mutants, and site-directed mutations were constructed in the pcDNA3 mammalian expression vector (Invitrogen) using polymerase chain reaction and restriction sites. Briefly, the Bsu36I-BspLU11I-fragment of HCN4 or HCN2 (a Bsu36I restriction site had been introduced into the HCN2 sequence as a silent mutation at nt 537-543) was replaced by DNA fragments, generated in several overlap PCR steps, which contained the desired sequences and mutations. The correctness of the mutant channels and introduced point mutations was verified by DNA sequencing.
Nomenclature of Mutant Channels-Chimeras are named according to this nomenclature: H4, HCN4-wild type; H2, HCN2-wild type. First wild type in italics is the backbone. Second wild type in italics is the channel from which certain transmembrane segments or amino acids have been introduced into the first wild type. Bold numbers inbetween: transmembrane segments that have been exchanged. Superior or smallcap numbers in between: Single amino acids (numbering from the "backbone"-wild-type HCN) that have been exchanged. Thus, for example H41-2H2 is a chimeric HCN4 channel that contains the transmembrane segments 1 and 2 from HCN2. H4 267/272 H2 is the HCN4 channel in which the amino acids 267 and 272 (HCN4 numbering) have been replaced by the respective amino acids from HCN2, in this example the two leucines have been exchanged for two phenylalanines. In Fig. 7, the classical nomenclature for point mutations is used since only leucine at position 272 in HCN4 has been exchanged for several other amino acids (L272X). Deletion of the C-terminal region of some channels ( . . . ⌬C) was achieved by exchanging the nucleotides encoding amino acid 535 for a STOP codon (E535STOP, HCN4 numbering) leaving a C terminus of only 14 amino acids, which lacks the CNBD. Consult Supplement Table 1, A-D for the exact crossover points and amino acid numbering for all chimeric channels.
Functional Expression and Electrophysiology-HEK293 cells were transiently transfected with expression vectors encoding either wildtype or mutant HCN channels using FuGENE 6 transfection reagent (Roche Applied Science) according to the manufacturer's instructions (transfectant/DNA-ratio: 3/1 v/w). Cells were cultured in MEM supplemented with 10% fetal calf serum and kept at 37°C, 6% CO 2 . Currents were recorded 2-3 days after transfection with the whole cell patch recording technique at a temperature of 23 Ϯ 1°C. The extracellular (bath) solution contained (in mM): 120 NaCl, 20 KCl, 1 MgCl 2 , 1.8 CaCl 2 , 10 HEPES, 10 glucose, pH adjusted to 7.4 with NaOH. The intracellular (pipette) solution contained (in mM): 10 NaCl, 30 KCl, 90 K-Asp, 1 MgSO 4 , 5 EGTA, 10 HEPES, pH adjusted to 7.4 with KOH. For determination of the effect of cAMP on the wild-type and mutant channels ("ϩ cAMP"), 100 M cAMP (Roche Applied Science) was added to the intracellular solution. Patch pipettes were pulled from borosilicate glass and had a resistance of 2-5 M⍀ when filled with intracellular solution.
The membrane potential was held at Ϫ40 mV. To elicit inward currents, step pulses of 5 s duration were applied from Ϫ140 mV to Ϫ30 mV, followed by a step to Ϫ140 mV as described in Fig. 1. For some raw data displays, longer or shorter pulses were applied to more clearly demonstrate the differences in activation kinetics between the (mutant) channels, but these measurements were not included into the voltagedependent steady-state evaluation to obtain V max 1 ⁄2 .
Further analysis was done offline with the Origin 6.0 software (Microcal). Time constants of activation ( act ) were obtained by fitting the current traces of the Ϫ140 to Ϫ100 (Ϫ90) mV steps after the initial lag with the sum of two exponential functions y ϭ A 1 e (Ϫx/t1 ) ϩ A 2 e (Ϫx/t2 ), where 1 and 2 are the fast and slow time constants of activation, respectively; 1 is consequently referred to as act since the slow component (A 2 ) of HCN channel activation generally accounts for Ͻ10% of the current amplitude; thus, the fast component is the major if not the exclusive kinetics-determining component of the HCN channels as has been discussed in Refs. 9 and 19. To obtain voltage dependent steadystate activation curves, tail currents measured immediately after the final step to Ϫ140 mV were normalized by the maximal current (I max ) and plotted as a function of the preceding membrane potential. The curves were fitted with the Boltzmann function in Equation 1, where I min is an offset caused by a nonzero holding current and is not included in the current amplitude, V is the test potential, V max 1 ⁄2 is the membrane potential for half-maximal activation, and k is the slope factor. All values are given as mean Ϯ S.E.; n is the number of experiments. Statistical differences were determined using Student's unpaired t test; p values Ͻ0.05 were considered significant.

HCN2 and HCN4 Differ in Their Activation Kinetics-For
both human HCN2 and HCN4 channels expressed in HEK293 cells, non-inactivating inward currents upon hyperpolarizing steps can be measured (Fig. 1A). In addition, both channels are modulated by intracellular cAMP (Fig. 1B). Application of 100 M cAMP shifted significantly the steady-state activation curves toward more positive potentials. The shift for both channels is about ϩ13 mV resulting in V max 1 ⁄2 of Ϫ86.3 mV and Ϫ84.9 mV for HCN2 and HCN4, respectively, compared with Ϫ98.7 mV and Ϫ99.1 mV without cAMP. The differences in V max 1 ⁄2 between HCN2 and HCN4 are not significant. The slope factor k is generally larger for HCN4 (10.1 versus 7.8 for HCN2, Numbers of experiments (n) are given in the graphics. The C-and N-terminal regions do not change the speed of activation or reaction to cAMP of the core channel, represented by the S1-S6 segments. Deleting the C terminus accelerates the channel activation and abolishes sensitivity to cAMP. B, activation kinetics of the chimera reciprocal to the one displayed in A: The H2NCH4-chimera does likewise not show a difference to the wild-type HCN2 with respect to act or modulation by cAMP.
without cAMP) meaning that HCN4 is activated over a more extended range of potentials around V max 1 ⁄2 .
Determination of the activation time constants ( act ) reveals a major difference between HCN2 and HCN4, as has been described earlier (8): At a potential range from Ϫ140 to Ϫ100 mV, HCN2 is activated 3-fold faster than HCN4 (Fig. 1C). For both channels, act is decreased by cAMP. At Ϫ140 mV, cAMP decreases act for HCN4 from 461 to 165 ms, and act for HCN2 from 144 to 91 ms. However, this difference becomes less dramatic if the relative change in act induced by cAMP is evaluated. At Ϫ140 mV, act decreases 2.8-fold and 1.6-fold for the HCN4 and HCN2 channel, respectively. This relation (Ϯ 0.4 for both) is maintained over the range of activation potentials from Ϫ140 to Ϫ100 mV. But even though act for both channels are much faster in the presence of cAMP, there still remains an about 2-fold difference in act between HCN2 and HCN4 that cannot be explained by a difference in the extend of the cAMPinduced conformational change.
The C-and N-terminal Regions Do Not Account for the Inherent, cAMP-independent Difference in Activation Kinetics of HCN2 and HCN4 -Previously, it was reported that cAMP-dependent differences between HCN1 and HCN2 (17,20) or HCN1 and HCN4 (19) depended on the nature of the C terminus. To test this possibility, we successively exchanged the N and C termini between HCN2 and HCN4 and determined act with and without cAMP. Fig. 2A shows the main results for HCN4: H4NCH2, i.e. HCN4 with both N and C terminus including the CNBD from HCN2, is still activated slowly ( act at Ϫ140 mV ϭ 466 ms, compared with 461 and 144 ms for HCN4 and HCN2, respectively). This chimera is still modulated 2.2to 3.5-fold by cAMP. The same results were obtained when only the N terminus (H4NH2) or the C terminus (H4CH2) were exchanged (data not shown). In addition, deleting the whole C terminus except for 14 amino acids adjacent to the S6 segments (H4⌬C) results in a channel that is activated almost exactly as fast as the cAMP-modulated HCN4 ( act at Ϫ140 mV ϭ 177 ms compared with 165 ms for HCN4 ϩ cAMP). In accordance with the report of Wainger et al. (18) that cAMP releases the inhibition exerted by binding to the C terminus, cAMP has no further stimulating effect on act of the H4⌬C chimera. As for HCN4, the activation kinetics of HCN2 is not influenced by exchange of the N or C terminus (Fig. 2B) indicating that the slow activation kinetics of HCN4 is not transferred with its long C terminus. Deletion of the HCN2 C terminus 14 or 50 amino acids after S6 did not yield functional channels supporting the notion that the C terminus of HCN2 is essential for basic channel activity in addition to mediating cAMP-dependent regulation of channel functions. From these results, we concluded that the inherent difference in act has to lie somewhere in the transmembrane region.
The Difference in Activation Kinetics Lies in the S1-S2 Transmembrane Region-Next, we constructed chimeras in which we exchanged the transmembrane regions between HCN2 and HCN4 to find the regions responsible for the difference. As can be seen in Fig. 3, replacing the S1-S2 region of HCN4 with the HCN2 counterpart (H41-2H2) results in a faster activating channel with act matching act of HCN2. Applying cAMP accelerates this mutant channel to the same extend as the HCN2 wild type. Indeed, all mutant HCN4 channels that contain at least the S1-S2 region of HCN2 (H41-6H2, H41-4H2, H41-2H2) show this fast activation. Interestingly, both H41H2 and H42H2 chimeras, in which only S1 or S2 plus the S1-S2 linker have been exchanged, also show faster activation kinetics. However, act of H42H2 is significantly slower than act of HCN2 wild type (198 versus 144 ms at Ϫ140 mV, without cAMP; p Ͻ 0.05). In fact, act of H42H2 is within the cAMPinduced activation range of HCN4. H41H2 on the other hand is as fast as HCN2 ( act at Ϫ140 mV ϭ 135 versus 144 ms, without cAMP) and is accelerated by cAMP to almost the same extend as HCN2 ( act at Ϫ140 mV ϭ 103 versus 91 ms; p Ͼ 0.05). The S1-S2 reverse chimera H21-2H4 (Fig. 3, last row) in which the S1-S2 region of HCN4 has been introduced into HCN2 confirms the kinetic modulating function of this element: act of H21-2H4 is slow and matches quite exactly act of HCN4, (462 versus 461 ms, and 194 versus 165 ms at Ϫ140 mV, without and with cAMP, respectively). Fig. 4 compares the electrophysiological behavior of the H41-2H2 and H21-2H4 chimeras. Both mutants are yielding inward currents comparable in size to the wild-type channels, but with activation kinetics of H41-2H2 resembling HCN2 and H21-2H4 resembling HCN4. The difference in act between the chimeras and their respective wild-type channels is present over the whole range of activation potentials (Ϫ140 to Ϫ100 (Ϫ90) mV). The changes in act are not the result of a changed steady-state voltage dependence (Fig. 4C). There is no shift in the activation curves. The V max 1 ⁄2 do not differ significantly between any of the mutant and wild-type channels.
A Single Amino Acid in S1 Accounts for the Inherent Kinetic Difference of HCN2 and HCN4 -The sequence comparison of the S1-S2 segment of HCN2 and HCN4 revealed five differing amino acids: two in S1, two in the S1-S2 linker and 1 in S2 (Fig.  5A). These five amino acids were replaced singly or together in the HCN4 channel by the corresponding amino acids from the HCN2 channel (Fig. 5, B and C). Leucine 272 plays a major role in controlling the activation kinetics. Replacing leucine 272 by phenylalanine (H4 272 H2, Fig. 5C, 6 th row) results in an HCN2like, fast activating channel whereas replacing leucine 267 by phenylalanine (H4 267 H2, 8 th row) has no effect on activation kinetics. The double mutant that is identical to H41H2 has the kinetics as chimera H4 272 H2 (Fig. 3). act of H4 272 H2 is 149 ms at Ϫ140 mV compared with 144 ms for HCN2. In addition, the mutant channel is modulated by cAMP like HCN2 with a act of 106 ms at Ϫ140 mV compared with 91 ms for the wild-type channel (p Ͼ 0.05). Deletion of the C terminus of H4 272 H2 results in a functional, fast activating channel with an activation time comparable to that of HCN2 plus cAMP, suggesting that the element for the inherent activation kinetics has been transferred with the mutation L272F. The importance of this amino acid is also shown by the reverse chimera. Replacing the phenylalanine 221 in HCN2 by leucine (amino acid 272 in HCN4 corresponds to amino acid 221 in HCN2) results in a significantly slower activating channel ( act ϭ 353 ms for H2 221 H4 versus 144 ms for HCN2). This value is close to the activation range of HCN4 being 461 ms in the absence of cAMP (Figs. 5C, 9 th row and 6A). In addition, activation of H2 221 H4 is accelerated 2.5-3-fold by 100 M cAMP, a characteristic feature of the wild-type HCN4 channel (see Fig. 1 and above). As already shown for the S1-S2 chimeras (Fig. 4), exchanging the amino acids of S1 did not result in a left or right shift of the steady-state activation curves that could simply account for the changes in activation kinetics at a given potential (data not shown).
The relevance of the three amino acid exchanges in the external S1-S2 linker and the S2 segment was already analyzed by the chimera H42H2. This chimera is identical with the mutations of H4 291/293/308 H2 (see Fig. 3). This exchange results in a faster, but not quite HCN2-like fast activating channel. Replacing both amino acids of the linker together (H4 291/293 H2, Fig. 5C) yields a channel activating still faster than HCN4, but slower than H4 291/293/308 H. Replacing amino acids 291 and 293 individually (H4 291 H2 ϭ N291T and H4 293 H2 ϭ T293A) results in channels with activation kinetics not much different from HCN4. Finally, replacing amino acid 308 (H4 308 H2) results in a channel apparently activating even slower than the wild-type HCN4 channel ( act 507 versus 461 ms at Ϫ140 mV, p ϭ 0.06). Interestingly, all the chimeras in which parts of the S1-S2 linker and/or S2 were mutated show an altered cAMP response. Most notably, the H4 308 H2 chimera can no longer be modulated by cAMP ( act 507 ms versus 520 ms at Ϫ140 mV without and with cAMP, respectively). Furthermore, deleting the C terminus of this mutant (H4 308 H2⌬C, Fig. 5C, second but last row and Fig. 6B) does not yield a fast activating channel as can be observed after deletion of the C terminus of the HCN4 wildtype channel (H4⌬C, Figs. 2A and 5C, last row and 6B). These phenotypes were observed for all chimeras at a voltage range from Ϫ140 mV to Ϫ100 (Ϫ90) mV (Fig. 6) ruling out that the differences in activation time constants were caused by an alteration of the voltage dependence of these channels.
These results suggest, that isolated mutations in the S1-S2 linker and S2 segment interfere with the proper activation of HCN4 channels, independently from the presence or absence of cAMP or a C terminus. However, these effects are not present, if the corresponding mutation(s) is done in the S1 segment. All mutant channels show inward currents comparable in size to the wild-type HCN channels (50 -250 pA/pF). The exchange of amino acid 272 between HCN2 and HCN4 changes HCN4 into an HCN2-like channel and vice versa with respect to the activation kinetics (Fig. 6A). Exchanging amino acid 308 in HCN4 for the respective amino acid of HCN2 channel does not alter significantly the basic activation kinetic of the channel, but renders it unresponsive to cAMP.
Structural Requirements Determining the Effect of Amino Residue 272 of HCN4 -In a last set of experiments, we examined to which extent other amino acids can mimic the effect of phenylalanine at postion 272 in HCN4. We replaced leucine 272 with methionine, tryptophan, and alanine. Methionine is the amino acid found in HCN1 at the respective position, tryptophan is an aromatic, strongly hydrophobic amino acid like phenylalanine and alanine is a neutral amino acid with a short side chain. Methionine at position 272 is not able to speed up the activation of HCN4 (Fig. 7) supporting the previous notion (19) that the activation speed of HCN1 depends on several factors including the S1-S2 transmembrane region and part of the S6 C-terminal region.
A tryptophan at position 272 was able to accelerate the activation of the HCN4 channel significantly ( act at Ϫ140 mV: 186 versus 461 versus 144 ms, H4 272 W versus HCN4 versus HCN2, respectively) (Fig. 7, B and C). However, introduction of tryptophan flattened the activation curve (k ϭ 15.9 versus 10.1 for HCN4) and shifted the steady-state activation curve toward more positive potentials (V max 1 ⁄2 ϭ Ϫ90.8 versus Ϫ99.1 mV for HCN4). These changes could at least partly account for a faster act at a potential of Ϫ140 mV (Fig. 7D)

DISCUSSION
HCN2 and HCN4 are structurally closely related pacemaker channels that are distributed partly in the same, partly in different tissues where they may have different functions (10,14,(21)(22)(23)(24). A characteristic difference of the two channels is their different speed of activation. Part of this difference is carried by their different responsiveness to intracellular cAMP, but apart from this, there is also an inherent difference in activation kinetics rendering HCN4 the slower activating chan- nel. It has been suggested that the molecular basis for this inherent difference resides somewhere in the transmembrane region. In order to find this region, we constructed chimeric HCN4/2 channels. As a basic principle, we tried to turn HCN4 into a mutant channel with the activation kinetics of HCN2, i.e. we tried to speed up the activation of HCN4. We were able to demonstrate that several residues in the S1-S2 transmembrane region determine the activation kinetics of HCN2 and HCN4 channels.
A Single Residue in the S1 Segment Determines the Difference in Activation Speeds-The amino acid at position 272 in the S1 segment of HCN4 is responsible for the difference in the inherent activation kinetics of HCN2 and HCN4. Exchanging this residue between the two channels transfers the activation kinetics to the respective other channel. Especially, HCN4 with a phenylalanine, the respective residue in HCN2, instead of a leucine at position 272 is activated almost exactly as fast as HCN2. The responsiveness to cAMP in this mutant channel is comparable to HCN2. Thus, we conclude that this single residue determines the distinct cAMP-independent activation time constant of HCN2 and HCN4.
Replacement of leucine 272 by tryptophan, an aromatic and hydrophobic amino acid like phenylalanine, had a similar effect on the activation time of HCN4: The speed of activation was significantly increased and the response to cAMP was maintained. However, this mutant channel did not exactly result in an HCN2-like channel since the voltage dependence of activation was shifted toward more positive activation potentials indicating that tryptophan is not sufficient to preserve proper gating in HCN4. Although alanine at position 272 had no effect on the activation kinetic it affected as tryptophan the voltage dependence of activation supporting the notion that gating of HCN4 is rather sensitive to the nature of the amino acid at position 272.
Methionine, the corresponding amino acid of the HCN1 channel, had no effect on the activation kinetics of HCN4. In agreement with Ishii et al. (19), we had to replace the whole S1-S2 segment plus parts of the S6 C-terminal region to change HCN4 into an HCN1-like channel with regard to activation speed (data not shown). The need for this large replacement is not too surprising, if we consider that the effect of cAMP on the activation kinetics of HCN1 is rather modest compared with its effect on HCN4. Ishii et al. (19) already concluded that the functional and structural differences between HCN4 and HCN1 are pronounced and that exchanging single residues in the S1-S2 regions of HCN1 and HCN4 did not markedly affect activation kinetics. The inability of alanine or methionine to change the speed of activation indicates that other parts of the HCN4 channel must contribute to its slow activation time constant. HCN3, an HCN channel, which activates faster than HCN4 but considerably slower than HCN2 (10) has a leucine at the position corresponding to 272. SPIH, an HCN channel from sea urchin testis (25), also has a leucine at the corresponding position but is activated faster than the HCN4 channel. These considerations support the notion that additional elements are needed to affect fast or slow opening gating.
Single Residues in the S2 Segment and in the S1-S2 Linker Influence cAMP Modulation-Replacing the S1-S2 linker and the S2 segment in HCN4 with the respective parts from HCN2 together renders the mutant channel almost as rapidly activated as HCN2 (H4 291/293/308 H2, Fig. 5C). However, activation is only weakly accelerated by cAMP and act of the mutated channel is never faster than act for the cAMP-modulated HCN4 channel. Replacing those three amino acids individually reveals the same principle: The activation time constants of the mutant channels are always within the HCN4 range and the responsiveness to cAMP is limited and even abolished as in the case of the I308M mutation. Most likely, Ile-308 interacts with Leu-272 since the negative effect of the single mutation I308M on cAMP decreased activation time is surmounted if the mutation L272F is combined with I308M. Interaction between these two amino acids are apparently also necessary to increase the activation speed by removal of the C terminus. The presented results show that mutation of the three amino acids  in the S1-S2 linker and S2 segment do not contribute to the inherent act difference between HCN2 and HCN4. Rather, the mutant channels are HCN4-like channels, which are arrested in some rigid state that does not relax if the conformation of the inhibitory C-terminal region is modulated by cAMP.
In conclusion, the basic, cAMP-independent and -dependent activation kinetics of HCN2 and HCN4 channels reside in the S1-S2 transmembrane region. Residue 272 of the S1 segment determines the difference in activation kinetics between HCN2 and HCN4 whereas residues 291 and 293 of the S1-2 linker and 308 of the S2 segment are important for the proper activation gating of these HCN channels. Disrupting the structure in any of the described positions by inserting unsuitable residues results in functional but incorrectly gated channels.
The relative ease with which the basic kinetic properties of HCN2 can be transferred to HCN4, and considering that these basic kinetic properties determines the function of the respective channel in its native environment, suggests that mutations occurring in the described segments of the HCN channels could be associated with diseases based on improper pacemaker activities such as certain cardiac arrhythmias (14), brainstem-based respiratory impairments (21), or epilepsies (14,23,24).