INTRODUCTION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

The human-ether-a-go-go related gene (h-erg) encodes a potassium channel (HERG) that mediates the cardiac repolarizing current I_Kr (Sanguinetti et al., 1995; Trudeau et al., 1995). The physiological importance of HERG is emphasized by the discovery of certain forms of familial long-QT syndrome linked to mutations in the h-erg gene (Curran et al., 1995; Sanguinetti et al., 1995; Spector et al., 1996a). HERG channels are also molecular targets for widely used pharmacological agents such as class III antiarrhythmics (Spector et al., 1996a), antipsychotic drugs (Suessbrich et al., 1997a), histamine receptor antagonists (Suessbrich et al., 1996; Taglialatela et al., 1998), calcium channel blockers (Chouabe et al., 1998), and sulfonylurea antidiabetic drugs (Rosati et al., 1998). Furthermore, they have been implicated in changes in resting membrane potential associated with the cell cycle, in the control of neuritogenesis and differentiation in neuronal cells, and in the spike-frequency accommodation of neuronal firing (Arcangeli et al., 1993, 1995; Faravelli et al., 1996; Chiesa et al., 1997; Schönherr et al., 1999). Finally, mutations in the erg gene of Drosophila produce neurological defects (Titus et al., 1997; X. Wang et al., 1997), and the rat counterpart of HERG constitutes an important target for the control of electrical activity and hence for the modulation of neurosecretion in adenohypophysial cells (Barros et al., 1994, 1997; Schäfer et al., 1999).

The time and voltage dependencies of HERG currents determine their role in the regulation of cell excitability. This includes a slow activation rate that overlaps with a rapid and voltage-dependent inactivation process, limiting the level of outward current upon depolarization. At negative repolarizing voltages, prominent tail currents are produced after reopening of the channels as they rapidly recover from inactivation, before closing at quite a slow rate (Trudeau et al., 1995; Sanguinetti et al., 1995; Smith et al., 1996; Schönherr and Heinemann, 1996; Spector et al., 1996b; S. Wang et al., 1997). This makes HERG operate as an inward rectifier, although it has the six membrane-spanning domains and the S4 and the regions typical of depolarization-activated channels (Warmke and Ganetzky, 1994).

The molecular basis of HERG kinetic characteristics are not completely understood. Recent work resolving the crystal structure of the highly conserved HERG initial domain (Cabral et al., 1998) identified it as a eukaryotic PAS domain (the eag or PAS domain) involved in the regulation of channel gating (Cabral et al., 1998). Removal of this domain markedly increased HERG deactivation rates (Schönherr and Heinemann, 1996; Spector et al., 1996b; Wang et al., 1998; Cabral et al., 1998) by eliminating its interaction with the gating machinery, probably the S4-S5 linker (Wang et al., 1998; Cabral et al., 1998). One exclusive structural feature of HERG is the presence of a long stretch of amino acids (named here the HERG-specific or "proximal" domain by its proximity to the S1 helix and the channel core) that follows the conserved eag domain and extends from about position 135 to about position 366 (Warmke and Ganetzky, 1994). However, the spatial structure and functional significance of this protein region are unknown.

The best described functional role of K⁺ channel amino-terminal domains is to provide the "ball" and "chain" structures for N-type inactivation (Hoshi et al., 1990). Nevertheless, HERG inactivation takes place through a C-type mechanism not related to the amino terminus (Smith et al., 1996; Schönherr and Heinemann, 1996; Spector et al., 1996b; S. Wang et al., 1997; Wang et al., 1998). The structure of the relatively short protein domains linking the initial ball and the S1 helix in several K⁺ channels has been elucidated recently (Kreusch et al., 1998; Bixby et al., 1999). However, the kind of interaction that takes place between these T1 domains and the channel core is still unknown. Furthermore, whereas the T1 domains are regarded as crucial determinants of channel assembly, HERG channels lacking almost the whole amino terminus are normally expressed in the plasma membrane (Schönherr and Heinemann, 1996; Spector et al., 1996b; Wang et al., 1998; Cabral et al., 1998).

Recent work has implicated the amino terminus of different channels, including HERG, in the regulation of voltage dependence and/or activation kinetics (Schönherr and Heinemann, 1996; Spector et al., 1996b; Terlau et al., 1997; Marten and Hoshi, 1997, 1998; Pascual et al., 1997; Chiara et al., 1999). In this report we tested the relevance of the proximal domain on HERG channel gating. We found that deletion of this domain induces a dramatic acceleration of channel activation associated with a shift in its voltage dependence. However, both deactivation from fully activated states and channel inactivation are poorly affected by the deletion. This and the results obtained with channel variants from which the eag domain has been also deleted lead us to propose a model in which both the long HERG-specific and the initial eag domains play an important role in setting HERG gating characteristics. Hence the HERG-specific proximal domain seems to constitute an important element determining the role of HERG channels in the maintenance of cell resting potential, electrical excitability, and accommodation of neuronal and cardiac firing.

	MATERIALS AND METHODS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Generation of channel mutants

To construct the amino-terminal deletion mutant 2-370, a forward polymerase chain reaction (PCR) primer was synthesized to introduce a HindIII site, 6 bp of 5' untranslated sequence, an ATG, and 24 bp of HERG coding sequence corresponding to amino acids 371-378 (5'-TTG AAG CTT CTC AGG ATG ACT GAG AAG GTC ACC CAG GTC CTG-3'). The reverse primer covered a unique XhoI site in position 2272 (amino acid 697) of the HERG sequence (5'-GAA GTA CTC CTC GAG GCG CTG GCG-3'). The corresponding PCR product was digested with HindIII and XhoI, gel purified, and ligated into HindIII/XhoI-digested wild-type HERG in the pSP64 expression vector. For the fully proximal domain-deleted mutant (138-373) a forward PCR primer containing a HindIII site and covering the start codon (5'-TTG AAG CTT CTC AGG ATG CCG GTG CGG AGG GGC CAC-3') was synthesized and used in PCR reactions with a reverse primer containing 27 bp of coding sequence corresponding to amino acids 129-137, followed by 9 bp of sequence corresponding to amino acids 374-376 and covering a unique BstEII site (5'-GGA AGG TGA CCA TGT CCT TCT CCA TCA CCA CCT CGA A-3'). The partially deleted mutants 170-190, 223-373 and 186-191, 223-373 were generated in a similar way by using the same forward PCR primer (containing a HindIII site and the start codon) and a reverse primer containing 27 bp of coding sequence corresponding to amino acids 214-222, followed by a 9-bp sequence corresponding to amino acids 374-376 and covering the unique BstEII site (5'-GAG GGT GAC CAC GTG GTT GTC CAT GGC TGT CAC TTC-3'). Screening of a number of clones for the 223-373 mutant revealed the systematic introduction of additional internal deletions in positions upstream of amino acid 223. It is probable that the presence of a region particularly enriched in G-C pairs around positions 690-760 of the HERG sequence (corresponding to amino acids 170-193) caused the consistent production of errors during the PCR reactions. Because of this, two partially deleted mutants lacking either 21 or 6 amino acids (corresponding to positions 170-190 and 186-191, respectively), besides the designed 223-373 deletion, were selected for the experiments. The small deletion mutant 355-373 was generated using the same forward PCR primer as above (with a HindIII site and the start codon) and a reverse primer containing 27 bp of coding sequence corresponding to amino acids 346-354, followed by a 9-bp sequence corresponding to amino acids 374-376 also covering the unique BstEII site (5'-CTC GGT GAC CCT GGT GGG CGA AGC CAA GAA GGG GTC-3'). The resulting PCR products were digested with HindIII/BstEII, gel purified, and ligated into HindIII/BstEII-digested wild-type HERG in the pS64 vector. The point mutant S620T was created by site-directed mutagenesis, using a PCR-based overlap extension method as previously described (Ho et al., 1989). The final PCR fragment was digested with BstEII and BglII and reinserted into the wild-type clone at the corresponding sites. In the constructs that combined the S620T mutation with deletions, the BstEII/BglII fragment was reinserted into the deletion mutant clones at the corresponding sites. All constructs were sequenced by the dideoxy chain termination method (Sanger et al., 1977) to confirm the mutations and to ensure the absence of errors.

Plasmids and preparation of cRNA

The plasmid containing the cDNA for the HERG channel was a generous gift of Dr. E. Wanke (University of Milan, Milan, Italy). Plasmids were linearized, and capped cRNA was synthesized in vitro from the linear cDNA templates by standard methods, using SP6 RNA polymerase as described (de la Peña et al., 1992).

Oocyte expression and solutions

Procedures for frog anesthesia and surgery, obtaining oocytes, and microinjection have been detailed elsewhere (Barros et al., 1998). Oocytes were maintained in OR-2 medium (in mM: 82.5 NaCl, 2 KCl, 2 CaCl₂, 2 MgCl₂, 1 Na₂HPO₄, 10 HEPES, at pH 7.5). Cytoplasmic microinjections were performed with 30-50 nl of in vitro synthesized cRNA per oocyte. HERG currents were studied in manually defolliculated oocytes (de la Peña et al., 1992; Barros et al., 1998). Unless otherwise indicated, recordings were obtained in extracellular high-K⁺ OR-2 medium in which 50 mM KCl was substituted for an equivalent amount of NaCl. Functional expression was typically assessed 2-3 days after microinjection.

Data acquisition and analysis

Recordings were made at room temperature with the two-electrode voltage-clamp method as described previously (Barros et al., 1998). Membrane potential was typically clamped at 80 mV, and at 100 or 110 mV for 138-373 and 223-373 deleted channels. Stimulation and data acquisition were controlled with Pulse+PulseFit software (HEKA Elektronic, Lambrecht, Germany) running on Macintosh computers. Data analysis was performed with the programs PulseFit (HEKA Elektronic) and Igor-Pro (WaveMetrics, Lake Oswego, OR). A P/n method was used for leak and capacitive current subtraction. The leak holding potential was set at 100 for basal voltages of 80 and 100 mV, and at 110 mV for cells clamped at 110 mV. Usually, a scaling factor of 0.2 was used. Kinetic parameters of activation, deactivation, inactivation, and inactivation recovery were obtained as previously described (Barros et al., 1998), using the voltage protocols shown on the graphs. The amount of injected cRNA was calibrated to yield inward tail currents in the range of 1-5 µA at repolarization voltages around 100 mV to ensure proper clamp control. Fits to tail currents from inactivating channels began as soon as the clamp settled. A single-exponential function of positive amplitude was applied to the inactivation recovery phase, followed by a negative-amplitude biexponential fit to the decaying phase of the tail. Deactivation parameters for S620T noninactivating channels were obtained from double-exponential fits to the tails, with the first cursor of the fitting window located at the end of the capacitive transient. Data are presented as means ± SE; the number of oocytes is indicated by n and the number of frogs by N.

	RESULTS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Deletion of the HERG-specific proximal domain in the amino terminus selectively modifies activation gating

Effect of proximal domain deletion on HERG activation rates

View larger version (35K): [in this window] [in a new window]   FIGURE 1   Topology of a HERG channel subunit. The transmembrane helices S1-S6 are represented by six cylinders at the top. The positions of the S1 helix, the voltage sensor in S4, and the P region between S5 and S6 are highlighted. The approximate situation of the amino acid residue subjected to point mutation in this study is also indicated by a black dot in the scheme. Schematic representations of the amino terminus in wild-type and deletion mutants are shown at the bottom. The eag and the proximal domains are marked by dotted and striped bars, respectively. Deletions are marked by dashed lines. The size of every domain and the lengths of the deletions are represented on a horizontal scale as proportional to the total length of the amino terminus.

View larger version (20K): [in this window] [in a new window]   FIGURE 2   Effect of HERG proximal domain deletion on channel activation rates. (Top) The time course of voltage-dependent activation was studied by varying the duration of a depolarizing prepulse to 0 mV following the voltage protocols shown at the top of each current trace. Test pulses were applied once every 20 s. Families of currents from wild-type (WT) and proximal domain-deleted (138-370) channels are shown. Note the 80-mV holding potential used for WT channels and the 110 mV used with the 138-373 construct. The zero current level is indicated by the dashed lines. (Bottom) Dependence of activation rates on depolarization membrane potential. The magnitude of the instantaneous tail currents upon repolarization was determined from recordings as shown at the top, by fitting exponentials to the decaying portion of the tail and extrapolating the current to the moment the depolarizing pulse was ended (Barros et al., 1998). Values correspond to 3-12 oocytes from three animals. Plots of normalized current values versus depolarization times were generated at the indicated voltages as shown in the inset for depolarizations to 0 mV, using three oocytes from the same frog. These plots were used to measure times necessary to attain half-maximum current magnitudes.

Effect of proximal domain deletion on HERG steady-state activation properties

View larger version (36K): [in this window] [in a new window]   FIGURE 3   Effect of HERG proximal domain deletion on steady-state voltage dependence of activation. (A) Steady-state voltage dependence of activation was studied following the protocols shown in the two upper insets, with a prepulse of varying magnitude and a duration ranging from 0.4 to 10 s, followed by a pulse test to 100 mV. Holding potentials of 0 mV to hold the channels fully open (left) and 80 (WT) or 110 mV (138-370) to hold the channels fully closed (right) were used as indicated. Families of tail currents during the test pulse for wild-type (WT) and proximal domain-deleted channels (138-370) are shown. The range of potentials covered by the prepulse is stated at the top of each currents. The duration of the prepulse is shown underlined below the current traces. (B) Fractional activation curves for WT and 138-370 HERG channels. The closed and open symbols correspond to data obtained from zero and 80/110 mV holding voltages, respectively. Data obtained at prepulse durations of 0.4 s (circles), 2 s (triangles), and 10 s (squares) are plotted. The continuous lines correspond to Boltzmann curves, h(V) = Imax [1/(1 + exp ((V  Vhalf)/k))], that best fitted the data. The dashed lines represent the deduced position of the activation curves under steady-state conditions obtained as a mean of those corresponding to both holding voltages and prepulses of 10-s duration. Values from at least four oocytes and two frogs are averaged in the graphs.

HERG deactivation time constants are not modified by deletion of the proximal domain

View larger version (32K): [in this window] [in a new window]   FIGURE 4   Lack of an effect of proximal domain deletion on HERG channel deactivation and inactivation rates. (A) Comparison of channel deactivation in wild-type and deleted channels. Families of currents from HERG wild-type (WT) and proximal domain-deleted channels (138-373) are shown on the left. Currents were obtained during steps to potentials ranging from 130 to 0 mV, following depolarizing pulses to 0 mV, as indicated at the top of each trace. The zero current level is indicated by the dashed lines. Deactivation time constants were quantified by fitting a double exponential to the decaying portion of the tails as described in Materials and Methods. Upper right panel: Fast and slow deactivation time constants are plotted for different repolarization voltages. Lower right panel: The relative amplitudes of the fast components of deactivation for the two channel types. Averaged values are shown from nine oocytes expressing WT channels in which biexponential fits to the deactivation time course covering the full range of voltages could be performed for 26 cells (N = 12). Values from 23 oocytes (N = 7) were averaged for 138-373 channels. (B) Effect of proximal domain deletion on inactivation. Onset of fast inactivation was studied using the voltage protocol shown at the top. HERG was activated and inactivated with a 400-ms prepulse to +40 mV. A second short prepulse to 100 was used to recover the channels from inactivation, followed by a test pulse to different voltages to reinactivate the channels. Families of currents for wild-type (WT) and proximal domain-deleted channels (138-373) are shown on the left. Test pulses were applied once every 20 s. Membrane currents are shown starting at the end of the depolarization prepulse. Time constants for the onset of inactivation as shown on the right were obtained from current traces by fitting a single-exponential function to the decaying portion of the currents during the test pulses. Holding potentials of 80 and 100 mV were used with WT and 138-373 channels, respectively. Note the small magnitude of the currents along the test pulses for voltages around the reversal potential of K+ ions ( 20 mV). Statistical significance (0.01 < p < 0.05, Student's t-test) was obtained only for time constant values corresponding to voltages between +10 and 30 mV.

Effects of proximal domain removal are not related to HERG channel inactivation

View larger version (28K): [in this window] [in a new window]   FIGURE 5   Effects of proximal domain removal are independent of HERG inactivation. (A) Families of currents with 400-ms steps to different voltages applied to channels carrying a point mutation to abolish channel inactivation, combined (S620T+ 138-373) or not combined (S620T) with a deletion removing the proximal domain from the amino terminus. Note the faster activation time course of proximal domain-deleted channels at every depolarization voltage. (B) Dependence of activation rates on membrane potential. Times for half-activation were obtained directly from current traces as in A measuring the time necessary to attain half of the maximum current at every depolarization voltage. Averaged values from seven and 10 oocytes (N = 3) are shown for full-length and proximal domain-deleted channels, respectively. (C) Steady-state voltage dependence of activation for S620T and S620T+138-373 channels. The magnitude of the tail currents during test pulses and the Boltzmann fits to the data were obtained as indicated in the legend of Fig. 3.

Effects of proximal domain deletion are not related exclusively to its length, but also to elimination of specific amino acid sequences near the channel core

View larger version (31K): [in this window] [in a new window]   FIGURE 6   Influence of proximal domain length in HERG activation. (A) Dependence of activation rates on membrane potential in channels carrying different deletions of the proximal domain. Times for half-activation were obtained directly from current traces, using noninactivating S620T channels and measuring the time necessary to attain half of the maximum current at every depolarization voltage. Values from wild-type (WT) and whole proximal domain-deleted channels (138-373) like those in Fig. 5 are shown for comparison. (B) Time for half-activation at 0 mV as a function of proximal domain length. The channel variants are arranged in the abscissa in increasing length of the proximal domain, corresponding to 138-373, 186-191 + 223-373 (223-373 on the graph; see Materials and Methods), 355-373, and full-length channels. (C) Steady-state activation voltage dependence of wild-type and amino-terminus-deleted HERG channels. The voltage dependencies of activation under steady-state conditions were derived following the procedures detailed in the legend of Fig. 3. Different HERG channel variants carrying the indicated deletions in the amino terminus were used after being introduced into the background of the S620T noninactivating HERG. The relative positions of the Vhalf values of the Boltzmann curves along the voltage axis are indicated by arrows.

Possible interactions between distal eag and HERG-specific proximal domains in the amino terminus

Modifications in HERG activation induced by removal of the proximal domain are reverted by additional deletion of the eag domain

View larger version (28K): [in this window] [in a new window]   FIGURE 7   Proximal domain deletion-induced accelerations in activation are partially compensated for by the additional removal of the eag domain. (A) Activation rates of HERG 2-370 channels. A family of currents obtained by varying the duration of a depolarizing pulse to 0 mV is shown on the left. Repolarization steps to 60 mV to slow down channel deactivation and allow proper measurements of tail current magnitudes were delivered as shown. The dependence of activation rates on depolarization potential is shown on the right. The magnitude of the instantaneous tail currents upon repolarization was determined from recordings as shown on the left, by fitting an exponential to the decaying portion of the tail and extrapolating the current to the moment the depolarizing pulse was ended. (B) Activation rates of S620T noninactivating HERG channels carrying a 2-370 deletion. A family of currents in response to 400-ms depolarizations to different voltages is shown on the left. Times for half-activation were obtained directly from current traces measuring the time necessary to attain half of the maximum current at every depolarization voltage. Lines showing the variation of activation rates as a function of voltage for wild-type (·····) and 138-373 channels (- - -) are also shown for comparison.

Deletion of the proximal domain accelerates transition of HERG to state(s) showing slow deactivation rates

View larger version (33K): [in this window] [in a new window]   FIGURE 8   Proximal domain removal accelerates transitions to slowly deactivating HERG channel state(s). (A) Scaled deactivating tail currents in response to depolarization steps to different potentials. Noninactivating full-length (S620T) or proximal domain-deleted (S620T+138-373) channels were submitted to 400-ms depolarization pulses to different voltages between 70 and +40 mV (S620T) or 90 and +20 mV (S620T+138-373) in 10-mV steps from a holding potential of 110 mV. Note the slower rates of deactivation induced by increasing depolarization voltage in full-length but not in proximal domain-deleted channels. Biexponential fits are also shown superimposed on the tails. Currents induced by depolarizations and tail currents after repolarization to 110 mV are shown in the insets. (B) Variation of HERG deactivation rates as a function of depolarization voltage. Time constants for deactivation were obtained as indicated in Materials and Methods from tail current recordings as in A. The percentage of the fast decaying tail current component oscillated between 80% at 30 mV and 95% at +40 mV for S620T channels. It also remained fairly constant around 60-70% in the range of 70 to +20 mV for S620T+138-373 channels. Averaged values from six oocytes from the same batch are shown. I/V curves for the tail currents of the insets in A are also shown against an arbitrary vertical scale for better comparison of the voltage dependence of activation and deactivation slowing in S620T (×) and S620T+138-373 (+) channels under the same conditions. (C) Variation of HERG deactivation rate as a function of depolarization voltage in channels carrying different deletions in the amino terminus. Time constants of deactivation were obtained as detailed in the legend of Fig. 8 from tail current recordings of oocytes expressing S620T noninactivating channels, from which nearly half of the proximal domain (223-373), 19 residues (355-373), or both the eag and the proximal domains (2-370) had been deleted. Averaged values from four oocytes are plotted against the amplitude of the depolarization step. Only values corresponding to the major fast component of deactivation are plotted, for simplicity. Repolarization voltage for recording tail currents was 110 mV for 223-373 and 355-373 constructs. Repolarization steps to 60 mV were used with 2-370 channels to slow down channel deactivation and allow proper measurement of deactivation time constants. Nevertheless, although with faster closing kinetics, similar results were obtained with repolarization to 110 mV.

View larger version (29K): [in this window] [in a new window]   FIGURE 9   Effect of depolarization time on closing rates of full-length and proximal domain-deleted HERG channels. (A) Currents induced by depolarizations of different duration in oocytes expressing full-length (S620T) and proximal domain-deleted (S620T+138-373) noninactivating HERG channels. (B) Scaled currents in response to depolarization steps of 40 and 300 ms. Current traces are shown displaced along the horizontal axis for coincidence of the tail current time course. Tail current magnitude has been normalized to peak. Biexponential fits are also shown superimposed on the tail currents. (C) Variation of HERG deactivation rates as a function of depolarization time. Fast and slow deactivation time constants were obtained from tail current recordings of four oocytes (N = 2) as in A. Only values corresponding to the major fast component of deactivation are plotted. (D) Relative amplitude of the fast deactivation component in full-length and proximal domain-deleted channels with depolarizations of increasing duration, as indicated on the abscissa.

	DISCUSSION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

One characteristic structural feature of HERG K⁺ channels as compared to other members of the eag family, is the presence of a long stretch of amino acids (designated here as the HERG-specific or proximal domain) that separates the conserved initial eag or PAS domain from the core of the channel protein (see above). Previous studies with deletion mutants in which either the eag or both the eag and the HERG-specific domains were removed indicated that the amino terminus of HERG does not participate in the fast inactivation process, but contributes to control channel deactivation (Schönherr and Heinemann, 1996; Spector et al., 1996b; Wang et al., 1998; Cabral et al., 1998). By selectively removing the long HERG-specific domain, we demonstrate here a strong acceleration of channel opening. This is associated with a shift of activation voltage dependence in the hyperpolarizing direction. However, elimination of the HERG-specific domain has almost no effect on the closing and inactivation properties of the channel. Thus two different domains regulating HERG gating seem to exist in the amino terminus: a distal eag domain mainly controlling current deactivation, and a HERG-specific domain regulating current activation. Removal of the eag domain from the deleted channels noticeably reverses the effects on activation caused by selective elimination of the proximal domain. This indicates that the functional consequences of proximal domain removal for channel opening are at least partially dependent on the presence of the eag domain and suggests that an interaction of the eag domain with the channel core is involved in modifications of activation caused by elimination of the HERG-specific domain.

These results are compatible with a model in which the presence of the long domain exclusive of HERG limits or slows down the interaction of the eag domain with the channel core. Elimination of the proximal domain would favor the interaction of the eag domain with the gating machinery, speeding up activation. Furthermore, it could be also expected that once activation has been completed at longer times or more positive voltages in full-length channels, dissociation of the eag domain from its interaction site(s) limits deactivation, making this process scarcely dependent on the presence of the proximal domain. In this case, the requirement of channel opening for effective interaction of the eag domain with the channel is supported by the coincidence of the time course and voltage dependence for channel activation and slowing of deactivation, a characteristic dependent upon the presence of an eag domain interacting with the channel core (Wang et al., 1998; Cabral et al., 1998; Sanguinetti and Xu, 1999). Further support for this model comes from the behavior of constructions in which the whole amino terminus has been deleted. In this case, the facilitating contribution of the eag domain to opening will be removed, yielding activation properties more similar to those of the wild-type channel, which has this domain separated from the protein core by the proximal domain. Furthermore, eag domain removal would also remove the limitation imposed by its relatively slow dissociation. This would explain the marked acceleration of closing rates observed with variants lacking either the whole amino terminus or the initial residues of the HERG protein (London et al., 1997; Cabral et al., 1998; Wang et al., 1998). Finally, albeit indirectly, the proposed model is also supported by our recent observation that in full-length channels a temperature increase from 24°C to 35°C causes a nearly fourfold decrease in the time needed for half-activation, but a less than twofold decrease in deactivation time constants (Barros and Pardo, manuscript in preparation). Thus more extensive conformational changes seem to be required for channel opening as compared to deactivation, a process probably rate-limited by unbinding and diffusion of the eag domain.

Opening of HERG channels has been described recently as a multistep sequential process in which the final and voltage-dependent closed-to-open transition is preceded by two steps between closed states: an initial one that is weakly voltage dependent, and a second that is essentially voltage independent (S. Wang et al., 1997; Wang et al., 1998; Kiehn et al., 1999; see also above). Our data indicate that both transitions are clearly accelerated in proximal domain-deleted channels and that the time constant of the slowest voltage-independent transition (S. Wang et al., 1997) is reduced by more than half by the deletion. Although the structural rearrangements associated with individual transitions are not known, this suggests that the need to adequately modify the conformation of the exclusive HERG proximal domain acts as an important constraining factor on efficient progress through the activation pathway.

The exact identity and position of the regulatory residues in the proximal domain are not known, but our results demonstrate that some of them are located between amino acids 223 and 373, close to the core of the channel protein. Our results indicate that a short stretch of 19 amino acids between residues 355 and 373 is important in limiting the interaction of the eag domain with the channel core. However, other residues located between positions 223 and 355 may also help to prevent the effect of this domain favoring activation. Interestingly, removal of residues 355-373 eliminates five positive charges from the HERG sequence. Thus it is tempting to speculate on the possibility that the presence of this region exerts an electrostatic repulsion of the otherwise positively charged distal amino terminus. Systematic mutagenesis studies directed at locating the specific residue(s) involved in proximal domain deletion effects are currently under way.

Previous work with Shaker channels in which the length of the amino-terminal chain domain was systematically changed indicated that ball-dependent channel inactivation was accelerated by chain deletions and slowed by amino acid insertions (Hoshi et al., 1990). As for the interaction of the ball with N-type inactivating channels, the coincidence of voltage dependence for HERG activation and slowing of deactivation suggests that the interaction of the eag domain with HERG is intrinsically voltage independent, and any apparent voltage dependence would arise from direct coupling to activation. Furthermore, the S4-S5 linker seems to act as a receptor site for the very amino terminus in both N-type-inactivating Shaker (Holmgren et al., 1996) and HERG channels (Cabral et al., 1998; Wang et al., 1998; Sanguinetti and Xu, 1999). Thus it could be possible that a mechanism similar to that of N-type inactivation is operative in HERG. However, the concept of "ball" and "chain" domains applied in the framework of a diffusible blocking particle that interacts with the channel core with kinetics dependent on the length of a flexible chain domain (Hoshi et al., 1990) does not seem to be directly applicable to HERG. Thus 1) activation/deactivation gating and not inactivation properties is regulated by the HERG amino terminus; 2) the influence of the proximal domain in HERG gating is exerted through the presence of specific sequence patterns and is not exclusively related to chain length; 3) there is no indication that the amino terminus blocks the HERG pore (Wang et al., 1998); 4) although amino-terminus action is disrupted by elevating external K⁺ in both Shaker (Demo and Yellen, 1991) and HERG (Wang et al., 1998) channels, higher concentrations seem to be needed to destabilize the interaction of the HERG amino terminus with the channel core; 5) only a single ball domain is required to mediate N-type inactivation, but more than one and possibly all four amino termini could be involved in HERG deactivation slowing (Wang et al., 1998); and 6) unlike the case of the ball-inactivating particle, it has been proposed that the distal eag domain is tightly attached to the body of the channel protein (Cabral et al., 1998). It is important to note also that in addition to amino-terminal domains, other cytoplasmic structures could also be involved in the regulatory effects on HERG gating. Thus shifts in activation voltage dependence, such as those induced by removal of the HERG proximal domain, are produced when the HERG carboxy terminus is replaced by the equivalent domain of mouse eag (Herzberg et al., 1998). Similarly, both amino and carboxy termini modulate the voltage sensitivity of KAT1 channels (Marten and Hoshi, 1998).

The regulatory effects of rat eag and KAT1 amino-terminal domains on activation voltage dependence and kinetics seem to be exerted through their molecular interaction with the S4 segment or the S4-S5 linker (Terlau et al., 1997; Marten and Hoshi, 1998). Residues in the S4-S5 linker of Shaker also act as receptor sites for the distal inactivation ball and influence activation gating (Holmgren et al., 1996). Similarly, the interaction site(s) of the very amino terminus with HERG channels seems to be located in the S4-S5 loop (Cabral et al., 1998; Wang et al., 1998; Sanguinetti and Xu, 1999). Whether the modulatory effects of the HERG proximal domain are indirect or are exerted by a direct interaction of this domain with the distal eag domain or with the gating machinery remains to be established.

Regardless of the molecular mechanism(s) involved in controlling gating behavior, the combination of an eag domain with the HERG-specific proximal domain seems to explain two important features of the channel that help it to achieve its physiological role. One is the slow activation rate, essential for use-dependent accumulations implicated in the spike-frequency accommodation of neuronal firing (Schönherr et al. 1999) and, in combination with the very fast inactivation characteristic of HERG, for limiting the outward flow of K⁺ ions and contributing to the maintenance of the plateau potential in cardiac cells (Hancox et al., 1998). This is needed to limit the interaction of the eag domain with the channel core, and hence this would not be achieved without the proximal domain. The other feature is the slow deactivation rate upon repolarization as compared to that obtained without the eag domain. In addition to the small depolarization-induced currents due to superposition of slow activation and fast inactivation, this eag domain-dependent effect would ensure the known operation of HERG as an inward rectifier. This would also determine the important participation of HERG not only in the repolarization phase of the cardiac action potential (and subsequently in controlling the duration of the spike and interspike intervals; Smith et al., 1996, Zhou et al., 1998), but also in the maintenance of the resting potential, the electrical excitability, and the pacemaking activity of adenohypophysial, neuronal, cardiac, and tumor cells (Barros et al., 1997; Bianchi et al., 1998; Schäfer et al., 1999; Schönherr et al., 1999).