Abstract
The ether-à-go-go-related gene (erg) K+ channels are known to be crucial for life in Caenorhabditis elegans (mating), Drosophila melanogaster (seizure), and humans (LQT syndrome). The erg genes known to date (erg1, erg2, and erg3) are highly expressed in various areas of the rat and mouse central nervous system (CNS), and ERG channel blockers alter firing accommodation. To assign physiological roles to each isoform, it is necessary to design pharmacological strategies to distinguish individual currents. To this purpose, we have investigated the blocking properties of specific peptide inhibitors of hERG1 channels on the human and rat isoforms. In particular, we have tested ErgTx1 (from the scorpion Centruroides noxious), BeKm-1 (from the scorpion Buthus eupeus), and APETx1 (from the sea anemone Anthopleura elegantissima). Because these peptides had different species-specific effects on the six different channels, we have also carried out a biophysical characterization of hERG2 and hERG3 channels that turned out to be different from the rat homologs. It emerged that APETx1 is exquisitely selective for ERG1 and does not compete with the other two toxins. BeKm-1 discriminates well among the three rat members. ErgTx1 is unable to block hERG2, but blocks rERG2 and has the lowest KD for hERG3. BeKm-1 and ErgTx1 compete for hERG3 but not for rERG2 blockade. Our findings should be helpful for structure-function studies and for novel CNS ERG-specific drug design.
The ether-à-go-go family of voltage-dependent K+ channels consists of the eag, elk (eag-like), and erg (eag-related gene) subfamilies (Warmke and Ganetzky, 1994). The ERG subfamily comprises ERG1, ERG2, and ERG3 (Shi et al., 1997), which may form either homo- or heterotetramers (Wimmers et al., 2001, 2002).
hERG1 (Warmke and Ganetzky, 1994; Titus et al., 1997) regulates the duration of the cardiac action potential (Sanguinetti et al., 1995); in fact, the slowly decreasing action potential plateau shifts these channels to a less inactivated state that increases the outward K+ current, thus sustaining further repolarization (Sanguinetti and Jurkiewicz, 1990). ERG1 channels may also sustain a process of spike-frequency adaptation (Chiesa et al., 1997), thus contributing to control burst duration in smooth muscle (Akbarali et al., 1999), carotid body (Overholt et al., 2000), lactotrophs (Corrette et al., 1996; Lecchi et al., 2002), human pancreatic β cells (Rosati et al., 2000), and chromaffin cells (Gullo et al., 2003). It has been shown that all the members of the ERG subfamily are expressed in the CNS in both rat (Saganich et al., 2001; Papa et al., 2003) and mouse (Polvani et al., 2003; Guasti et al., 2005). Evidence of the functional significance of this expression has been obtained in mouse cerebellar Purkinje cells and in rat embryonic rhombencephalon neurons (Sacco et al., 2003; Hirdes et al., 2005).
To decipher the functional role of ERG in CNS, it is necessary to isolate the different contributions given by each member of the subfamily. Organic and peptide blockers are known for hERG1 channels. The HERG organic blockers are not sufficiently selective for the different isoforms. However, several hERG-specific peptides have become available: ErgTx1 (CnErg1; Gurrola et al., 1999), ErgTx2 (Lecchi et al., 2002), BeKm-1 (Korolkova et al., 2001), CsEKerg1 (Nastainczyk et al., 2002), and APETx1 (Diochot et al., 2003). The aim of the present study was to examine whether the above ERG1-specific peptides present selective effects on different ERG members. The rationale for this study was the observation that the amino acid sequence of the extracellular “S5-P linker” pore region is different in the three ERG isoforms (Torres et al., 2003). This segment is the putative binding site for the scorpion toxins.
Our results show that 1) the blocking effect of our peptides was reversible (approximately 1 min), with two exceptions (ErgTx1 on hERG3, and rERG2 and BeKm1 on rERG2); 2) APETx1 turned out to be a selective blocker for ERG1; 3) ErgTx1 was unable to block hERG2; and 4) dose-response curves for ErgTx1 and BeKm-1 overlapped. The above differences prompted us to investigate in more detail the biophysical properties of the human and rat ERG2 and ERG3 channels that turned out to be different.
Materials and Methods
Plasmids
rerg1, rerg2, and rerg3 cDNAs were kindly provided by Prof. J. R. Schwarz (University of Hamburg, Hamburg, Germany). Herg2 and herg3 cDNAs were excised from pGH19 vectors (kindly provided by Prof. B. Ganetzky, University of Wisconsin, Madison, WI) and cloned into pCDNA3.1 vector (Invitrogen, Carlsbad, CA). The hERG2 cDNA received from Prof. B. Ganetzky slightly differs from the sequence deposited at NCBI (accession number NP110406) because it lacks a stretch of nucleotides coding for amino acids 718 to 729.
Toxins
ErgTx1, BeKm-1, and APETx1 were purified in the laboratories of Profs. Possani, Grishin, and Lazdunsky, respectively, as described previously (Gurrola et al., 1999; Korolkova et al., 2001; Diochot et al., 2003).
Electrophysiology
Cell Culture. Chinese hamster ovary cells were routinely cultured in Dulbecco's modified Eagle's medium (Invitrogen, Carlsbad, CA) containing 10% fetal calf serum (Invitrogen) and maintained at 37°C in humidified atmosphere with 5% CO2. Approximately 2 × 104 cells were transfected with 1 to 5 μg of the appropriate plasmid, along with 0.2 μg of green fluorescent protein coding plasmid (pEGFP-C1; Clontech, Mountain View, CA). For transfection, a Lipofectamine reagent kit (Invitrogen) was used, following the instructions of the manufacturer. Currents were recorded 24 to 72 h after transfection.
Solutions and Drugs. During the patch-clamp experiments cells were maintained in our standard extracellular solution, containing: 130 mM NaCl, 5 mM KCl, 2 mM CaCl2, 2 mM MgCl2, 10 mM HEPES-NaOH, and 5 mM d-glucose, pH 7.40. During the specific biophysical tests, cells were perfused with a high K+ external solution ([K+]o = 40 mM), in which NaCl was replaced by an equimolar amount of KCl. The pipette solution contained 130 mM K+-aspartate, 10 mM NaCl, 2 mM MgCl2, 10 mM EGTA-KOH, and 10 mM HEPES-KOH, pH 7.30, and nominal [Ca2+]i of ∼50 nM. Appropriate quantities of the toxins from stocks in H2O were dissolved in the extracellular solution immediately before the experiments. When contaminating potassium currents were present, the specific ERG blocker WAY123,398 (Spinelli et al., 1993) was used at 2 μM and the resulting traces were subtracted from control traces to obtain the WAY123,398-sensitive currents. The extracellular solutions were delivered through a nine-hole (0.6-mm) remote-controlled linear positioner with an average response time of 1 to 2 s that was placed near the cell under study.
Patch-Clamp Recordings and Data Analysis. ERG currents were always elicited under conditions of relatively high [K]o (40 mM) to measure currents under optimal signal-to-noise conditions. The currents were recorded at room temperature by means of the MultiClamp 700A (Molecular Devices, Sunnyvale, CA), as described previously (Oliveira et al., 2004). Pipette resistances were approximately 1.5 to 2.2 MΩ. Cell capacitance and series resistance errors were compensated (85-90%) before each voltage clamp protocol run to reduce the voltage errors to less than 5% of the protocol pulse. pClamp 8.2 (Molecular Devices) and Origin 7 (OriginLab Corp., Northampton, MA) software were routinely used during data acquisition and analysis.
The steady-state activation curves were obtained by plotting the peak tail currents at -120 mV versus the preconditioning potential (-80 to +30 mV, for 15 s; Fig. 1A), as reported previously (Schönherr et al., 1999). To generate steady-state inactivation curves, we applied the protocol shown in Fig. 1B. In this way, we obtained the currents at each Vm (Fig. 1B, right inset, a and c, ⋄). The time to peak of these currents was used to choose the time of the subsequent voltage steps at +20 mV to obtain peak tail currents after different degrees of inactivation (Fig. 1B, right inset, b and c,——). To compensate for the channel closing at negative potentials, we extrapolated the decaying phase of deactivation to the onset of the test pulse and applied the same relative correction to the peak tail current, as reported previously (Smith et al., 1996). This correction procedure was applied only at the Vm values indicated by asterisks (Fig. 1B, right inset, c, *). The peak tail currents were plotted versus membrane potentials and the curves were best fitted with a Boltzmann function.
Analysis of the Toxin Blockade Data
According to the Clark occupancy model (whose crucial assumption is that response is generated with sufficiently small delay), we assume that the interaction of a blocking toxin and an ion channel is described, at the steady-state, by the classic Michaelis-Menten-type saturating equation (see Gutfreund, 1995). To generate dose-response curves, we have plotted the fraction of unblocked channels (UCF) versus the toxin concentration ([T]). In Clark's model, this corresponds to the probability of receptor vacancy. It should be remembered in fact that
Because during long electrophysiological experiments the doseresponse curves cannot be obtained under truly steady-state conditions, we preferred to derive the equilibrium dissociation constant KD from the kinetics of our currents, at different toxin concentrations. Under a typical ON kinetics experiment, if a fixed toxin concentration [T] is applied at time 0, the normalized amplitude of the corresponding outward current, originating from the blockade of a K+ ion channel, is an exponential decaying function whose time constant, τon, is given by (k1[T] + k)-1)-1, where k1 and k-1 are the association and dissociation rate constants, respectively. On the contrary, in a washout experiment, the exponential increasing function is given by (1-exp(-t/τoff)), where τoff is 1/k-1, which depends only on the dissociation constant k-1 and not on [T]. In general,
By combining eqs. 1 and 2, we obtain.
The above relation can be used to calculate τoff when it is too long or to estimate τon when it is too short to be measured for technical reasons.
The time constant of an ON kinetic experiment at concentration [T] at known KD can be estimated by
which suggests that the experimental perfusing time should be of the order of 4 τon.
By simple algebraic combination of two ON experiments done at different toxin concentrations, [T1] and [T2], which are governed by the time constants τ1on and τ2on, respectively, it results that KD can be evaluated independently of τoff and is equal to:
which can be easily evaluated by measuring experimentally only the two block time constants and not the washout time constant, as in eq. 2.
Results
The sequences of the S5-S6 linker region of the rat and human ERG channels are shown in Table 1 (the black background indicates the hypothesized α-helix; Zhang et al., 2003). They show that the ERG isoforms present several differences, and there are also differences between human and rat sequences of the same family. According to various cited studies, it is expected that BeKm-1 and ErgTx1 bind to the residues indicated in white. Some of these are targets of both toxins (black background). No similar data are available for APETx1.
Biophysical Differences between Human and Rat Members of ERG2 and ERG3. We have studied the steadystate voltage-dependent activation/inactivation and the time constants of activation in the six isoforms. Figure 1A shows that the activation curves of the human and rat ERG1 and ERG2 are almost superimposable. On the contrary, the hERG3 activation curve was found to be shifted to the left with respect to ERG1 by approximately 7 mV compared with the 17 mV observed in rERG3. The relative positions on the membrane potential axis of the three rat members [see halfactivation potential (V1/2) and slope values in the figure legend] closely correspond to those described previously (Wimmers et al., 2002; Sturm et al., 2005), apart from a few millivolts of negative shift as a result of the use of 40 mM [K]o instead of the normal 5 mM. On the other hand, the voltagedependent inactivation curves of the human ERG1, ERG2, and ERG3 were very similar to each other (Fig. 1B). For rERG3, in the region from -40 to +40 mV, a shift to the right has been reported (Schledermann et al., 2001; Sturm et al., 2005) with respect to rERG1 and rERG2. Our data, obtained with the use of a different method (see inset and Materials and Methods) in a much larger voltage range from -160 to +20 mV, did not evidence this effect.
The time course of activation (Fig. 1C) was investigated with the envelope of the tail currents (see protocol in the inset) obtained at the V1/2 calculated from the experiments shown in Fig. 1A. The time course could be fitted by a doubleexponential function (data not shown). The slower time constant resulted in values larger than 30 s. Because it was not possible to condition the membrane potential for times longer than 40 s (to reach steady state), we decided to consider only the faster time constant, τfast-act (see legend to Fig. 1). These were entirely correlated with the V1/2 values, as shown in Fig. 1D. Only for ERG3 channels were the values of the human and rat τfast-act significantly different (by a factor of ∼3).
Another important property that differentiates the three rat members is the deactivation time course, which has been described by slow and fast components and a variable ratio of the slow to total amplitudes (Wimmers et al., 2002). Because the results we obtained with the hERG2 and hERG3 were different from those described by Wimmers et al. (2001) in the rat, we have tested all the above clones under the conditions applied by Schwarz et al. (2001) (extracellular solutions with 140 [K]o and 75 nM [Ca2+]o). In this case, we found that the biophysical features of the rat clones agreed with those reported by Wimmers et al. (2001) and that the differences between rat and human ERG1 were not significant. On the contrary, the differences between the rat and human ERG2 and ERG3 persisted and thus are to be considered intrinsic properties of the channels (data not shown).
These differences were further investigated (Fig. 2). Normalized traces obtained from cells expressing ERG2 (Fig. 2A) and ERG3 (Fig. 2B) were superimposed, showing that hERG2 deactivates much faster than rERG2 at -120 mV. We ascribe the difference to the presence of a stretch of 36 amino acids in hERG2 but absent in rERG2; in fact, the C terminus has been demonstrated to be important for hERG1 deactivation kinetics (Aydar and Palmer, 2001). Moreover, hERG3 currents, although they decay similarly to rERG3, show a persistent component much larger than that of rERG3.
We measured, at negative membrane potential, the slow and fast time constants (Fig. 2, C and D, left y-axis) and the ratio of the slow to total current amplitudes (Fig. 2, C and D, right y-axis). Figure 2C shows that hERG2 (•) is characterized by τfast and τslow (at -120 mV, 16.7 ± 1.1 and 138 ± 15 ms, respectively) which are approximately half of those observed in rERG2 (○, 31 ± 3.5 and 280.85 ± 32.9 ms, respectively). The ratio of the slow to total current (encircled symbols) has an increasing voltage-dependent behavior that is not very different between rat and human (at -120 mV, 0.14 ± 0.02 and 0.12 ± 0.017, respectively). On the whole, these results account for the differences observed in A. Indeed, it has been reported (Wimmers et al., 2002) that in rERG2, the above ratio is almost 2-fold compared with rERG3, whereas rERG1 is in between.
On the contrary, as shown in Fig. 2D, hERG3 (▴) and rERG3 (▵) have very similar values of τfast (at -120 mV, 16.6 ± 1.2 and 13.99 ± 0.74 ms, respectively) and τslow (at -120 mV, 186.1 ± 12.4 ms and 154.8 ± 17.9 ms, respectively), but the ratio of slow to total component amplitudes is from 3- to 5-fold higher in the human clone (at -120 mV, 0.14 ± 0.01 and 0.04 ± 0.01, respectively). Here again, the analysis substantiates the observations of B.
On the whole, these data suggest species-specific biophysical properties in ERG2 and ERG3 isoforms. For ERG2 channels, the difference consisted of a 2-fold faster deactivation of the human (•) isoform compared with the rat (○). For ERG3, species-related differences were present both in the steadystate activation (Fig. 1) and in the deactivation kinetics (Fig. 2D). In the former, we observed less negative V1/2 values and slower activation in the human clone. In the deactivation kinetics, we found a 3- to 4-fold higher ratio of the slow to total component in hERG3 compared with rERG3. In principle, this diversity could help toxin peptides to discriminate the different channel proteins.
Non-Steady-State Properties of Toxin Blockade and Related Dose-Response Relationships. To test the speed, the amount of block, and the reversibility of action, we continuously perfused and voltage-clamped each cell (Fig. 3) at a fixed toxin concentration, with a protocol consisting of a test pulse at -120 mV delivered every 4 or 10 s and elicited either from an holding potential of -70 or +30 mV (Fig. 3, inset). This experimental protocol was necessary because our peptides have a voltage-dependent action, characterized by a loss of inhibition when the cell is depolarized (Gurrola et al., 1999). The toxin concentrations were chosen according to preliminary experiments to obtain significant inhibition. On the other hand, the dose-response curves were obtained by applying a large range of increasing toxin concentrations [T] (two values for each 10-fold [T] change), for 100 s. The effects were tested at -120 mV, from holding potentials of -70 mV. During these experiments, we could not check whether a steady-state condition was attained at the specific toxin concentration (because of the slowness of the effect). Therefore, some of the dose-response data might be affected by systematic errors (i.e., underestimation of block), especially at the lowest concentrations, in which the “on kinetics” time constant was longer than 100 s. In these cases, the action of single toxin and the putative competition between some of the toxins were studied with different methods (see below).
Properties of ErgTx1 Blockade. As shown in Fig. 3, the blockade rise time, the washout time constant, and the maximal block were different for the six channel types. The data for human and rat ERG1 substantially recapitulate (Gurrola et al., 1999) the fast action, the reversibility, and the voltagedependent efficacy of this toxin (closed and open symbols are related to hyperpolarized or depolarized holding potentials). ErgTx1, up to 210 nM, was unable to consistently block hERG2 channels. On the contrary, full blockade of rERG2 was obtained around 21 nM at a relatively slow rate but with a poor reversibility because of a very slow unbinding kinetics. This effect was also observed on hERG3, but not on rERG3, where a weak blocking and washing-out was also obtained at 210 nM, during a 30-s application. In all clones, we observed a marked loss of potency at depolarized Vm (open symbols). The kinetics of block and recovery are summarized in Table 2.
A comparison of the quasi-stationary dose-response curves of the unblocked channel fraction (UCF; obtained from experiments in which negative holding potentials were used) is shown in the right of Fig. 3. Table 3 gives the parameters of the classic logistic equation with Hill coefficient fixed to 1, fitting the experimental data [we observed that a free Hill coefficient, between 0.98 to 1.26, could improve the fitting (data not shown)]. Table 3 also shows the fraction of current that is insensitive to high toxin concentrations (Ins). ErgTx1 blocked human ERG1 and ERG3 channels with a KD of approximately 4 nM. On the contrary, the rat members were blocked with KD ranging from 2.8 (ERG2) to 38 nM (ERG3).
On the whole, ErgTx1 had a greater affinity for rERG2 than for rERG1. Therefore, ErgTx1 represents the first highaffinity blocker for rERG2 channels, with a selectivity for rat over human members.
Properties of BeKm-1 Blockade. Experiments similar to those described for ErgTx1 were carried out for the other scorpion peptide BeKm-1, as reported in Fig. 4. Data show results obtained by applying 300 nM BeKm-1. In contrast to ErgTx1, BeKm-1 was able to block all the six members, when applied at negative holding potential. Only hERG2 was insensitive to the toxin when the latter was applied at depolarized membrane potential. On washout, the recovery was always relatively fast, except for rERG2, recovery for which was extremely slow. Blocking and washout kinetic properties are summarized in Table 4.
The quasi-stationary dose-response curves (Fig. 4, far right, and Table 5) show an interesting pattern among the rat members, with different KD values in the range from 4.2 to 747 nM. hERG1 and hERG3 were not discriminated by BeKm-1. On the contrary, it should be stressed that BeKm-1 was unique among the three toxins in its ability to block hERG2. Interestingly the rat members were sensitive to BeKm-1 with very different KD values. Therefore, this toxin seems the most promising for selectively recognizing the functional roles of the three different clones.
Properties of APETx1 Blockade. The blocking properties of APETx1 are shown in Fig. 5, A and B. Because the peptide action is highly voltage-dependent (Diochot et al., 2003), experiments were performed at a very low pulse rate to rule out the toxin unbinding caused by the test pulse itself. APETx1 was unable to affect ERG2 and ERG3 currents at concentrations that saturated ERG1 (1 μM) and can therefore be considered a pure ERG1 blocker for both rat and human homologs.
Competition experiments between APETx1 (500 nM) and ErgTx1 (10 nM) or BeKm-1 (10 nM) have been performed, and results are shown in Fig. 5, C and D (human) and E and F (rat). The three columns of each graph show the fraction of unblocked channels under APETx1 alone, the competing toxin alone (ErgTx1 or BeKm-1), or both the competing toxins. These data should be compared with a hypothetical noncompetitive model, shown in Table 6. These results are consistent with the notion that the binding sites on ERG1 for APETx1 are far apart from those of ErgTx1 and BeKm-1 (Diochot et al., 2003; Chagot et al., 2005).
On Kinetics and Competition Experiments with BeKm-1 and ErgTx1. The data shown in Figs. 3 and 4 were obtained with different concentrations of ErgTx1 and BeKm-1. Nonetheless, they suggest important differences between these toxins, if we compare the time constants of washout, which should depend only on the dissociation rates and not on the toxin concentration. Both toxins showed a very fast dissociation from rERG3, whereas the longest time constants for washout were observed for hERG3 and rERG2. The similarities between the two toxins suggest that competition for similar regions of these channels might influence the on kinetics when both toxins are present. Because overlapping binding sites on hERG1 have been suggested for ErgTx1 and BeKm-1 (Zhang et al., 2003), we have also studied their competition for hERG3 and rERG2. These experiments allowed us to calculate the KD of these effects with much more precision than was possible from the data given in Tables 3 and 5.
As for BeKm-1 and its dose-response curves (Fig. 4), we can estimate whether the 100 s used to equilibrate toxin binding, was sufficient to reach the steady-state, based on the expected time constant τon measured during an on-kinetics experiment and theoretically predicted by eq. 4 (see Materials and Methods). The KD values for hERG1 and hERG3 derived from fitting were approximately 9 nM (see Table 5) and the τoff were approximately 50 s (see Table 4). Thus, at 3 nM, the calculated τon is approximately 38 s; i.e., a 115-s delay should occur to approximate the steady-state condition for the dose-response curve.
For rERG2, the KD for BeKm-1, derived from the doseresponse curve, was 4.2 nM and the washout time constant was around 300 s, leading to a τon (at 3 nM) of 175 s and to an experimental time of approximately 9 min. In this case, a specific kinetic experiment is essential to clarify and calculate the correct KD using either eq. 1 or eq. 5.
Competition experiments should be done at concentrations close to the respective KD values to determine whether the results depend on pure summation of the effects or competition. This is particularly difficult, however, from an experimental point of view, because at these concentrations, the on kinetics are very slow.
Competition on hERG3 Channels. Data from Table 5 suggest a putative BeKm-1 KD of 11.5 nM, and we thus studied the kinetics of inhibition at 20 nM. From the average of four experiments, shown in Fig. 6A, we obtained a mean UCF of 0.41 ± 0.03 and τon = 22.8 ± 2.9 s (Fig. 6, D and E, first bars). By using this UCF value in eq. 1, we found KD = 20 × (1/0.41-1)-1 = 13.9 nM, and, by using the τon and τoff values in eq. 2, we found KD = 20 × 22.8/(58-22.8) = 13 nM.
These values are in good agreement with the value of 11.5 nM shown in Table 5. The same procedure was applied at 2 nM ErgTx1 (Fig. 6B, n = 3). In this case, τon was 34.1 ± 4.5 s and UCF was 0.08 ± 0.02 (Fig. 6, D and E, third bar). Because τoff was very poorly defined, we did not use eq. 2. 1 = From eq. 1, as before, we found KD = 2 × (1/0.08-1)- 0.18 nM, which is approximately 20 times smaller than the value derived from the dose-response curves of Fig. 3 (see also Table 3). The dose-response theoretical curve with this KD is shown in short points in Fig. 3, top right.
With the mixture of toxins (BeKm-1was applied first), we tested five cells. One experiment is shown in Fig. 6C, where the block produced by the competing toxins reached 0.11 ± 0.02 and the τon of the current decay was 274 ± 35 s (Fig. 6, D and E, second bar). Both values are in disagreement with a model of independent action because, in this case, we would expect (1) a block down to 0.033, which is approximately 3-fold smaller than the experimental value, and (2) a τon equal to 34.1 s (the line in the plot of Fig. 6C), which is approximately eight times smaller than the experimental result. We were unable to apply the toxins in reverse order, because we should have used a much smaller ErgTx1 concentration, with an impractical increase in the total experimental time, because of current run-down. Overall, these findings suggest a net competition between the two toxins.
Affinity and Competition on rERG2 Channels. Before performing a competition experiment, we wanted a better estimate the KD values of both ErgTx1 and BeKm-1 for comparison with the approximate values of 2.8 and 4.2 nM derived in Figs. 3 and 4.
To this aim, we used eq. 5 at two different pairs of [T1] and [T2] concentrations, either 2 nM and 10 nM or 2 nM and 20 nM. The results were averaged to obtain a more precise value. These τon were measured as shown in Fig. 7, A and B, for the pair of concentrations 2 and 10 nM. At 2 nM, τon resulted 134.3 ± 6.2 (n = 5) for BeKm-1 (□) and 140.2 ± 6.7 (n = 6) for ErgTx1 (○). At 10 nM, τon were 38.2 ± 0.8 (n = 4) for BeKm-1 (▪) and 31.6 ± 0.7 s (n = 7) for ErgTx1 (•). The corresponding values at 20 nM were 20 ± 1.2 s for BeKm-1and 23 ± 1.3 s for ErgTx1 (n = 4). These values are plotted as bars in Fig. 7E. In addition, we measured the UCF at 2 nM, which was 0.18 ± 0.02 for BeKm-1 and 0.13 ± 0.15 for ErgTx1.
By inserting the above data in eq. 5, we calculated the KD values, that turned out to be 1.51 ± 0.12 for ErgTx1 and 1.16 ± 0.09 for BeKm-1. These values were quite different from those shown in Figs. 3 and 4 (see also Tables 4 and 5). Because the latter were obtained at the steady state, their accuracy is expected to be higher. The dose-response theoretical curves with the corresponding KD values are shown in short points in Figs. 3 and 4, far right.
For competition experiments, we decided to use the 2 and 10 nM concentrations as the first and the second applications, respectively. To verify whether competition took place at the moment of the second application, only the time constant of the second application was analyzed. We decided not to include in the analysis the amplitude (small) of the current blocked during the second toxin application, because in each experiment we usually detected a variable amount (3-6%) of unblocked current, which could be a further source of error.
If competition is effective, we would measure a τon significantly longer than the one observed in single toxin experiment. Two exemplary competition experiments (of 11 total) are shown in Fig. 7C (BeKm-1 applied first) and in Fig. 7D (opposite order). Although the observed τon fluctuated considerably among experiments, the relatively fast time course of current decay (white lines through symbols) were little different from the mean values of the corresponding time constants measured during single toxin applications. Averaged data are shown in Fig. 7E, and the experiments to be compared are highlighted by the arrows. In conclusion, it seems that the binding of one toxin is not affected to a greater extent by the presence of the other toxin because the respective time constants were only increased marginally (from 38.2 ± 0.8 s to 51.2 ± 3.5 s for BeKm-1 and from 31.6 ± 0.7 s to 38.2 ± 2.9 s for ErgTx1). Therefore, the two toxins did not compete for the same binding site on rERG2.
Discussion
In general, biophysical properties are the hallmark of voltage-dependent ion channels, but gene subfamilies are often composed of clones with very similar properties. Because subfamilies have subtle differences and are often characterized by unique tissue-specific and sometimes species-specific distributions, the use of highly selective drugs or peptides able to distinguish basically similar biophysical behavior is urgently needed (Wimmers et al., 2001, 2002). In the present study, we showed, for the first time, that some voltage-dependent ERG K+ channels, almost indistinguishable from a biophysical standpoint, could be isolated by using appropriate concentration of pure peptides.
As shown before by Shi et al. (1997), the amino acid identities among hERG1 and rERG2 or rERG3 is around 60 ± 3%. On the contrary, we found that the similarity between the same isoform in two species (human and rat) turned out to be very high: 95% for ERG1, 90% for ERG2, and 94% for ERG3.
Although ERG1, ERG2, and ERG3 are very similar in their primary sequences and biophysical properties, we showed that three ERG1-specific toxins, APETx1, BeKm-1, and ErgTx1, with no sequence homology, can distinguish both the rat and human members, with KD values ranging from 0.18 to 747 nM.
Moreover, we found that the biophysical properties of ERG2 and ERG3, compared across humans and rats, are substantially different. Although there is still no evidence of the regional distribution of hERG2 and hERG3 channels in the human CNS, these biophysical differences suggest that hERG2 and hERG3 exert different functional roles, with possible consequences on the neuronal excitability in specific CNS areas.
Concerning the blocking properties studied here, we observed that the potency rank order for ErgTx1 suggests the following sequence (in parentheses are nanomolar KD values):
hERG3 (0.18) ≫ hERG1 (4.5)
rERG2 (1.16) > rERG1 (6.8) > rERG3 (38)
whereas for BeKm-1, the KD values rank as follows:
hERG1 (7.7) > hERG3 (11.5) ≫ hERG2 (77)
rERG2 (1.51) ≫ rERG1 (19) ≫ rERG3 (747).
As for the species differences, we observed that the sea anemone APETx1 was not able to recognize the less primordial channels ERG2 and ERG3. On the contrary, the ratio of the rat and human KD values for each member (ErgTx1KDr/h (x) and BeKm-1KDr/h (x), x = 1,2,3) suggests that there is a consistency between the two scorpion toxins' preference. Indeed, in the case of ErgTx1, hERG1 and hERG3 are better recognized than the rat members (KDr/h(ERG1) = 1.5, KDr/h(ERG3) = 211), whereas the opposite is true for ERG2, a human clone not recognized by ErgTx1. For BeKm-1, the situation is similar because both KDr/h(ERG1) = 2.7 and KDr/h(ERG3) = 65 are much higher than 1, and KDr/h(ERG2) = 0.02 is much less than 1.
It is known that the S5-P linker of the hERG1 channel is important for high-affinity binding of ErgTx1 and BeKm-1 (Pardo-López et al., 2002; Zhang et al., 2003). Natural amino acid substitutions in this region of ERG2 and ERG3 proteins considerably change toxin affinity. One rERG3 mutation compared with hERG3 (Q591T) results in decrease of BeKm-1 affinity by ∼65-fold, and triple mutation in hERG2 compared with rERG2 makes this channel insensitive to ErgTx1 (P581H, A591V, and Q596R). On the other hand, BeKm-1 and ErgTx1 most likely share a similar mechanism of action, suggesting similarities in their interaction surfaces and overlapping binding sites, at least on ERG1 and ERG3. On the contrary, APETx1 must have a different target region on ERG1. Results shown in the present paper should be helpful for modeling of toxin interaction and to study the structure of outer vestibules of ERG channels.
In addition, our data are relevant for designing experiments aimed at determining the physiological roles of different clones. Moreover, it could be of particular interest in humans, because it cannot be excluded that epilepsy episodes could depend on disorders originating from mutant ERG channels.
The ideal drug should have KD values (with a Hill coefficient around 1) that differ at least 6-fold among different channels. In this case, at a concentration of 6 × KD, only 16% of the channels with the lowest KD will remain open compared with 84% of the channels with the higher KD. To design blocking experiments in rat (and almost certainly in mouse), the best case would be to start with rERG2, because BeKm-1 toxin presents the advantage of having the largest KD difference among the three channels in rat. By using BeKm-1 (at 9 nM) first, it is possible to obtain a 84% rERG2 blockade. This procedure should also produce a small effect on rERG1 (32%). Further application of BeKm-1 (at 114 nM) should almost completely block rERG1; BeKm-1 at very high concentration or ErgTx1 (at 228 nM) should inhibit rERG3. On the contrary, if rERG1 blockage is to be achieved first, APETx1 should be used, followed by either ErgTx1 (at 7 nM) or BeKm-1 (at 9 nM) for rERG2 blockade, and finally ErgTx1 (at 228 nM) for blocking rERG3. Other types of procedures should be used if working on human tissues: first, use APETx1 to block hERG1, then at least 1.8 nM ErgTx1 for an almost complete blockade of hERG3, and finally BeKm-1, at 770 nM, for blocking hERG2.
We have not investigated in the present work the very interesting problems concerning the possibility that heteromultimeric channels could also coassemble in the CNS (Hirdes et al., 2005). It is possible that the KD values for heteromultimers are different from those reported here. This type of study, which complicates the research, is in progress.
Footnotes
-
This study was partially supported by grants from the Italian Ministero dell'Università e della Ricerca Scientifica e Tecnologica (MIUR-PRIN2003-2005-2001055320 and 2003052919, MIUR-FIRB2001-RBNE01XMP4-002, MIUR-FISR2001-0300179), the Università di Milano-Bicocca (to E.W. and to A.A.) (MIUR-PRIN2003-2005-2003054500), and by Telethon Fondazione On-lus (GGP02208 to A.A.). It also was supported by the National Council of Science and Technology (Mexican Government, grant 40251-Q) and from the Direción General de Asuntos de Personal Académico (grant IN206003) of the National Autonomous University of Mexico (to L.D.P.). R.R.-C. is a Ph.D. student of Physiology at the Department of Biotechnologies and Biosciences of the University of Milano-Bicocca. These data were presented in part as a poster at the 2005 Neuroscience Meeting, 152.4.
-
All laboratories contributed equally to the manuscript.
-
ABBREVIATIONS: CNS, central nervous system; ERG, ether-à-go-go-related gene; Erg, ether-à-go-go-related gene; hERG, human ether-à-go-go-related gene channel; rERG, rat ether-à-go-go-related gene channel; WAY123,398, N-methyl-N-(2-(methyl(1-methyl-1H-benzimidazol-2-yl)amino)ethyl)-4-((methylsulfonyl)amino)benzenesulfonamide; UCF, unblocked channel fraction.
- Received October 10, 2005.
- Accepted February 23, 2006.
- The American Society for Pharmacology and Experimental Therapeutics