1The ability of three structurally homologous scorpion toxins to block voltage-dependent K+ currents in rat dorsal root ganglion neurones was examined using the patch-clamp technique.
2Neurones with a diameter > 35 μm had two identifiable components of macroscopic K+ current. The outward current during depolarizations had both inactivating and non-inactivating components, and the tail currents had both a fast component (IKf) with a time constant of about 2.5 ms and a slow component (IKs) with a time constant of about 10 ms.
3The functional properties of IKf and IKs differed in several ways: (i)IKf activated over a more negative voltage range than IKf; (ii)IKf partially inactivated during a depolarization to +70 mV, whereas IKs did not inactivate during a 1 s depolarization to +70 mV; (iii)IKf activated more rapidly than IKs;and (iv)α-dendrotoxin selectively blocked IKf.
4Tityustoxin-Kα (TsTX-Kα) selectively blocked IKf, with little or no effect on IKs. The block was concentration dependent, with 50% of the current inhibited at a toxin concentration of about 38nM.
5TsTX-Kα block of IKf was completely reversible, but the washout rate was slow. The time constant of recovery from TsTX-Kα block was about 11 min.
6Charybdotoxin (CTX) also selectively blocked IKf in a reversible manner, but was about 10 times less potent than TsTX-Kα. The CTX washout rate was over 10 times faster than that of TsTX-Kα; the time constant of recovery was 0.8 min.
7Pandinotoxin-Kα (PiTX-Kα) also selectively blocked IKf; the IC50 for block of IKf was about 8.1nM. In contrast to the other two toxins, however, PiTX-Kα was poorly reversible.
8The block of IKf produced by CTX was voltage dependent. In the voltage range from -10 to +70 mV, the fraction of blocked IKf fell from 91 to 37%. In contrast, both TsTX-Kα and PiTX-Kα blocked IKf in a voltage-independent manner.
9The backbone structure and many of the amino acid side-chains on the presumed docking surfaces of the toxins are identical or conservatively replaced in all three toxins. Thus, some small differences in a few side-chains that influence electrostatic, hydrophobic/hydrophilic and/or steric interactions probably account for the marked differences in affinities and dissociation rates.
Many of the characteristics of toxin-K+ channel interactions have come from studies of cloned K+ channels (Goldstein et al. 1995; Aietar et al. 1995; Gross & MacKinnon, 1996). It now seems appropriate to examine, in parallel, the effects of some of these toxins on various native K+ channels. This is necessary in order to identify toxins that can selectively block specific subtypes of native K+ channels, and to characterize differences as well as similarities between the toxin–channel interactions in native versus cloned K+ channels.
In the study described here, we examined the ability of three scorpion toxins to inhibit voltage-dependent K+ currents in rat dorsal root ganglion (DRG) neurones. The three toxins are: tityustoxin-Kα (TsTX-Kα) from Tityus serrulatus, pandinotoxin-Kα (PiTX-Kα) from Pandinus imperator, and charybdotoxin (CTX) from Leiurus quinquestriatus hebraeus (Lqh).
Dissociation and culture of DBG neurones
Adult male rats (200–250 g) were killed by exposure to a rising concentration of CO2, followed by exsanguination. The sensory ganglia were dissected free, transferred to cold Locke solution (containing (mM): 156 NaCl, 5.6 KC1, 4 NaHCO3, 1.2 Na2H2PO4, 2.2 CaCl2, 1.2 MgCl2, 11.5 dextrose and 10 Na-Hepes; pH 7.4) and dissociated using a modification of a published procedure (Christian, Togo, Naper, Koschorke, Taylor & Weinreich, 1993). Briefly, the ganglia were sectioned and then incubated for 10 min at 37°C in Ca2+- and Mg2+-free Hanks’ balanced salt solution (HBSS; Sigma) containing 0.2 mg ml−1 L-cysteine and 0.15 mg ml−1 papain (pH 7.2). The ganglia were then washed in HBSS, followed by incubation for 10–15 min at 37°C in HBSS containing 1.6 mg ml−1 dispase (Type II; Boehringer Mannheim) and 0.8 mg ml−1 collagenase (pH7.2; Type IA; Sigma). Occasional vortexing and gentle trituration through a fire-polished Pasteur pipette was used to aid dispersion. The cells were centrifuged, washed once in culture medium (Dulbecco's modified Eagle's medium (Gibco) containing 10% (v/v) fetal bovine serum), and plated onto slivers of uncoated coverslips sized to fit our experimental patch-clamp chamber. The cells were maintained at 37 °C in a humidified atmosphere containing 8% CO2. Immediately after dissociation, and for 1–2 days in culture, these neurones are spherical, 25–50μm diameter cells with a total membrane capacitance of 53.4 ± 16.2pF (n= 13), and are thus ideally suited for whole-cell voltage clamp experiments.
Whole-cell patch-clamp recording
The whole-cell configuration of the patch-clamp technique was used for these experiments (Hiriart & Matteson, 1988; Parsey & Matteson, 1993). The solution in the recording chamber (volume ∼0.2 ml) was exchanged at a rate of about 2 ml min−1, so that the superfusion fluid was completely replaced in a few seconds. Pulse generation, and data acquisition, recording and analysis were performed with 386 or 486 IBM-type computers (Sala & Matteson, 1991; Parsey & Matteson, 1993). A homemade patch-clamp amplifier was designed along the lines described by Hamill, Marty, Neher, Sakmann & Sigworth (1981). Experiments were carried out at room temperature (20–22°C) at a holding potential of -80 mV.
A custom-designed hardware interface was used to communicate between the analog electronics and the computer. For stimulus-evoked data, the computer delivers the voltage-clamp command pulse to the patch-clamp amplifier, and can acquire data at rates up to 100 kHz. When performing macroscopic current recordings of voltage-dependent channel activity, linear currents were eliminated with a modified P/x procedure (Armstrong & Bezanilla, 1974) in which four to eight control pulses of one-third the amplitude of the test pulse were given, and the responses averaged.
Several approaches were taken to minimize series resistance errors (Armstrong & Gilly, 1992). Firstly, low resistance (0.5–1.5MΩ) patch electrodes were used. Secondly, when recording the relatively large K+ currents (IK) from DRG neurones, the internal K+ concentration was lowered to reduce the magnitude of IK, and thereby minimize voltage errors. Finally, electronic series resistance compensation was employed. To improve the stability of this compensation when large amounts of correction were required, the series resistance correction signal was delayed by passing it through a one-pole filter before summing the signal with the command voltage. Under the conditions of the experiments described here, the mean settling time constant of the whole-cell clamp was 140 ± 56μs (n= l3), without series resistance compensation. A further improvement in time resolution was obtained by applying this compensation.
Whole-cell recordings of voltage-dependent IK were made after eliminating current flow through other channels in the following ways. Na+ currents were eliminated by removing Na+ from both sides of the membrane, Ca2+ currents were blocked using extracellular Co2+, Ca2+-activated K+ currents were inhibited by using intracellular EGTA, and ATP-blockable K+ channels were blocked with intracellular ATP. The external solution had the following composition (mM): 170 choline chloride, 5 KCl, 1 MgCl2, 2.5 CaCl2, 2 CoCl2, 5 glucose and 10 Hepes (pH7.4). The pipette filling solution contained (mM): 120 N-methyglutamine glutamate, 30 KCl, 2 MgCl2, 10 EGTA, 4 MgATP and 10 Hepes (pH 7.2). For toxin experiments, a concentrated toxin stock solution (300μM in distilled water) was prepared and stored at 5°C. Immediately prior to the experiment, the toxin was diluted with the external solution to the appropriate concentration.
Preparation and application of toxin solutions
Recombinant TsTX-Kα and PiTX-Kα (Rogowski et al. 1996), and native CTX purified from Lqh venom, were used in these experiments. Recombinant TsTX-Kα (available from Research Biochemicals International), has the same amino acid sequence, chromatographic characteristics and physiological activity as does the toxin purified from T. serrulatus venom (Blaustein et al. 1991; Werkman et al. 1993; Rogowski, Krueger, Collins & Blaustein, 1994). The native and recombinant toxins selectively block only the non-inactivating K+ channel in synaptosomes. In contrast, a substance which was synthesized chemically (Peptides International, Louisville, KY, USA), and which is also sold commercially as Tityustoxin-Kα’, blocks both inactivating and, at higher concentration, non-inactivating K+ channels in rat brain synaptosomes (M. P. Blaustein & R. S. Rogowski, unpublished observations). Figure 1 shows the amino acid sequences and toxin structures (determined by NMR methods (Bontems, Roumestand, Gilquin, Menez & Toma, 1991) or calculated on the basis of homologies (Rogowski et al. 1996)) for the three scorpion toxins employed in this study. The toxins were applied to the cell by bath exchange of the external solution. The solution in the chamber (volume ∼0.2 ml) was exchanged at a rate of about 2 ml min−1, so that the superfusion fluid was completely replaced in a few seconds. The α-dendrotoxin (α-DaTX) was purified from green mamba venom in our laboratory (Benishin, Sorensen, Brown, Krueger & Blaustein, 1988).
Estimation of fast and slow tail currents
Voltage-dependent K+ currents in large diameter rat DRG neurones contain two predominant kinetic components in their tail currents (Fig. 2; see Results). These two currents are identified as IKf and IKs, for fast and slow tail current kinetics, respectively. In order to quantify the ability of various scorpion toxins to selectively block either IKf or IKs, the relative contribution of each component to the total K+ current was determined by fitting the sum of two exponentials to the K+ tail currents (Fig. 2E and F). The extrapolated amplitude of the fast or slow component of the tail current provides an estimate of the relative number of IKf or IKs channels, respectively, that are opened during the preceding depolarization. Potassium currents generated by steps to -60 mV contain mainly IKf. Therefore, the validity of using tail currents to quantify toxin block was examined in the following manner. The effect of each toxin on the IKf current generated at -60 mV was quantified in two ways: (i) by measuring the change in magnitude of the outward current during the pulse; and (ii) by measuring the change in magnitude of the extrapolated fast tail current. The two methods gave similar results.
Voltage-dependent K+ currents
Figure 2 illustrates the two most common patterns of whole-cell outward currents observed in these experiments. These currents were voltage dependent, and began to activate between -40 and -30 mV. Significant variability in the time course of outward currents generated by rat DRG neurones was observed (Akins & McCleskey, 1993). For example, in the cell illustrated in Fig. 2A, the outward currents activated relatively quickly and reached a peak in 4–7 ms at +70 mV. As shown in Fig. 2B, the current then partially inactivated over the next several tens to hundreds of milliseconds. This current pattern was observed in forty-six of fifty-three cells with diameters > 35μm, and was rarely observed in smaller diameter cells. The family of currents illustrated in Fig. 2C and D, was common in smaller diameter cells. This pattern consisted of a very fast, apparently transient, outward current followed by a relatively slowly activating current which reached a maximum in 20–30 ms at +70 mV. The currents in this type of cell did not have a rapidly inactivating current component at positive voltages (Fig. 2D) and did not respond to the polypeptide toxins discussed in this report (not shown). The remainder of this report will focus only on the large diameter neurones containing IKf and IKs.
It appears that several distinct subtypes of voltage-dependent K+ channels (e.g. A-type and delayed rectifier type) contribute to these macroscopic currents, and that the relative proportion of the different subtypes varies from one cell to another in the DBG. The analysis of K+ tail currents (next section) enabled us to identify two kinetically distinct types of voltage-dependent K+ currents which accounted for most of the outward current in neurones like the one in Fig. 2A.
Potassium tail currents
Under the ionic conditions of our experiments, the tail currents were inward in direction when the cells were repolarized to -80 mV (Fig. 2E and F). These tail currents did not decay along a single exponential time course, but contained at least two or three clearly identifiable exponential components: (i) a very fast component with a time constant (τ) of about 250μs; (ii) a component with a τ of ∼1.8–2.5 ms; and (iii) a relatively slow component with a τ of about 10ms. The tail current illustrated in Fig. 2E (noisy, dotted trace) is not well fitted by a single exponential (dashed line), whereas the sum of two exponentials (continuous line overlying the data) describes the current accurately, indicating that two kinetically distinct components are present in the tail. The time constants of the two components were 1.8 and 9.9 ms. A third exponential component was apparent in the tail current shown in Fig. 2F from another cell, where the sum of three exponentials (τ1= 0.334 ms, τ2= 1.87 ms and τ3= 10.4 ms) were required to fit the data. In the experiments described below, the amplitudes of the tail currents were obtained from the extrapolated amplitudes of the fitted exponentials.
Reversal potential measurements indicate that only two of these tail current components are K+ currents. The very fast tail current (τ∼0.25ms at -8OmV) had a reversal potential near 0 mV, and was unaffected by changes in external K+ concentration (data not shown). The other two tail current components (τ∼2 and 10 ms at -80 mV) had reversal potentials that shifted by 15 mV when external K+ was reduced from 5 to 2 m (data not shown). This is close to the expected shift of 23 mV, suggesting that K+ was the major current carrier. We will refer to these two tail current components as IKf and IKs, for fast and slow K+ currents, respectively. As illustrated in the following sections, these two components may reflect the activity of two channel populations that have different kinetics.
The two K+ tail current components activate over very different voltage ranges, as shown in Fig. 3A. In this experiment, tail currents were measured following 50 ms pulses to various potentials. The normalized amplitude of each tail current component is plotted as a function of pulse amplitude in Fig. 3A. Clearly, the fast tail current (τ= 2 ms; •) activated over a more negative voltage range than the slow component (τ= 10 ms; ▪). Fits of a Boltzmann distribution to the data (smooth curves in Fig. 3A) show that, in this cell, the voltage for half-maximal activation (V1/2) was -17.5 mV for IKf, and +26.5 mV for IKs. In four cells, V1/2 averaged -9.7 ± 3.5 mV (mean ±s.e.m.) for IKf and +16-5 ± 4.1 mV for IKs.
Inactivating and non-inactivating K+ channels
The inactivation gating of voltage-dependent K+ channels can be assessed by examining the effect of pulse duration on the amplitude of the tail currents. Figure 3B shows a plot of the tail current amplitude as a function of the duration of a preceding depolarization to +70 mV. The amplitude of IKf(•) reached a maximum within a few milliseconds, but then declined as the pulse duration increased (note that the tail currents are inward and therefore negative). The kinetics of this decline are similar to the kinetics of inactivation of the outward current at +70 mV (see inset in Fig. 3B), suggesting that the fast tail current reflects the activity of the inactivating K+ channels. In contrast, the slow component of the tail (IKs, ▪ in Fig. 3B) activated more slowly and did not decline in amplitude as pulse duration increased, indicating that the slow tail reflects K+ channels which do not inactivate. Similar results were obtained in three other cells: at the end of a 500 ms step to +70 mV, IKs did not inactivate and IKf inactivated by 75 ± 8.0%.
In addition to different deactivation and inactivation gating behaviour, IKf and IKs also have different activation kinetics. Activation kinetics were characterized with tail currents using the pulse protocol shown in Fig. 3C. A variable duration pulse to +70 mV was used to activate the channels, and the number of activated channels was evaluated from the amplitude of the tail current following repolarization. Figure 3C shows that at +70 mV, the channels giving rise to the fast tail current activate rapidly, reaching a maximum open probability within about 5–7 ms. The channels giving rise to IKs take much longer to activate: the slow tail current reached a maximum only after 20–50 ms at +70 mV. Similar results were obtained in three other cells.
In summary, these data support the hypothesis that IKf reflects the activity of a population of K+ channels that deactivates with a time constant of about 2 ms at -80 mV, activates and inactivates rapidly upon depolarization, and activates over a relatively negative voltage range (V1/2∼−10 mV). The slow component of the tail current (IKs) is associated with a population of K+ channels that deactivates slowly (τ∼10 ms) at -80 mV, activates slowly upon depolarization, does not inactivate, and has a relatively high threshold for activation (V1/2∼+16 mV). As described below, this identification of two voltage-dependent K+ current components is supported by the finding that several K+ channel toxins isolated from scorpion and snake venoms selectively block only one of the K+ current components.
Inhibition of K+ currents by α-DaTX
The snake toxin, α-DaTX, selectively blocks inactivating voltage-dependent K+ channels in rat brain synaptosomes (Benishin et al. 1988), and was expected to block IKf selectively in rat DRG neurones, as shown in Fig. 4. Whole-cell currents were recorded before (control trace) and after applying 100 nM α-DaTX externally; the difference trace is the toxin-sensitive component of the current. At either -30 mV (Fig. 4A) or -10 mV (Fig. 4B), α-DaTX blocked an early, transient component of the current.
Most of the current activated at -30 mV is of the fast-tail type, which we have identified as inactivating, A-type current (Fig. 3 A). Figure 4A shows that most of the current activated at -30 mV was inhibited by the K+ channel blocker α-DaTX. Accordingly, α-DaTX should selectively block the fast component of the tail. In Fig. 4C, the whole-cell current at +10 mV is shown before (Control) and after adding 100nM α-DaTX. The tail currents consisted of both fast (τ= 2.10 ms) and slow (τ= 10.9 ms) components; only the fast component was blocked by α-DaTX. The toxin-sensitive current (Fig. 4D) was a rapidly activating and partially inactivating current that deactivated with fast kinetics (τ= 2.12 ms) upon repolarization. Block of IKf by α-DaTX could be reversed by washing with toxin-free solution (data not shown).
Inhibition of voltage-dependent K+ currents by TsTX-Kα
The three scorpion toxins that we tested all selectively blocked only one of the two principle components of the voltage-dependent K+ current. Figure 5 shows the effects of TsTX-Kα on the macroscopic K+ current. The records in Fig. 5A illustrate the current generated by 50 ms voltage-clamp steps to -10mV, before and after adding 20 or 100nM TsTX-Kα. The tail current in the control record in Fig. 5A contains primarily a fast component, because most of the current generated by such a small depolarization is IKf. During the 50ms step to -10 mV, 20nM TsTX-Kα reduced the magnitude of the outward current by about 30%; TsTX-Kα also decreased the amplitude of the fast tail current by about 48% upon repolarization to -80 mV. At 100nM, TsTX-Kα reduced the outward current by 74%, and the fast tail current by 87%. In three cells, 100nM TsTX-Kα blocked the IKf tail current by 6O±6.3%. Figure 5B shows the toxin-sensitive K+ currents, which were obtained by subtracting the current recorded in the presence of TsTX-Kα from the control current. Clearly, the toxin blocks a current component that activates relatively quickly, inactivates partially and deactivates rapidly; these are the characteristics of IKf.
A similar effect can be seen in the records obtained at +30 mV (Fig. 5C and D). At this voltage, both IKf and IKs channels are activated, as indicated by the fact that the control tail current has both a fast and a slow component. TsTX-Kα decreased the magnitude of the outward current during the pulse, and reduced only the fast component of the tail current. Consistent with the results at -10 mV, the toxin-sensitive current at +30 mV reflects IKf: it activates rapidly, inactivates partially and deactivates rapidly (Fig. 5D).
At a concentration of 200nM, TsTX-Kα blocked IKf nearly completely (Fig. 6). The currents shown in Fig. 6A were generated by 50 ms steps to −10mV before and after adding 200nM TsTX-Kα. In the presence of TsTX-Kα, the remaining outward current was a small, slowly developing current that was followed by a slow tail current; it therefore represents current flow through IKs channels.
Block by TsTX-Kα is not voltage dependent
To examine the effect of voltage on the block of IKf by TsTX-Kα, we measured K+ currents generated by 50 ms steps to various voltages in the range from -40 to +70 mV. The tail currents recorded upon repolarization were separated into fast and slow components as described in Methods. Figure 7A shows the amplitudes of the fast tail current (IKf) as a function of voltage before and after adding 100nM TsTX-Kα: the fast tail current was nearly completely blocked by TsTX-Kα at all voltages. Figure 7B shows that TsTX-Kα had relatively little effect on the slow tail current at all voltages. Thus, at concentrations up to 100nM, TsTX-Kα selectively blocked IKf by a voltage-independent mechanism.
Dose–response relationship of TsTX-Kα block
The sensitivity of IKf to block by TsTX-Kα is illustrated by the dose–response curve in Fig. 8 (▪), which summarizes results from eleven cells. The IC50 for block of IKf by TsTX-Kα, obtained by fitting a conventional sigmoidal relationship to the data in Fig. 8, was 38nM. The large error bars (± 1 s.e.m.) in this figure indicate that the neurones varied somewhat in their sensitivity to TsTX-Kα. One possible reason is that IKf may not represent a single population of K+ channels. If there are multiple subtypes of voltage-dependent K+ channels with fast deactivation kinetics and varying sensitivity to TsTX-Kα, and if the relative proportion of these channel subtypes varies from cell to cell, the variability shown in Fig. 8 would be expected.
Dissociation of TsTX-Kα is relatively slow
Block of IKf by TsTX-Kα was reversible, but the dissociation of the toxin from its binding site was relatively slow (Fig. 9). After exposure to a blocking concentration of TsTX-Kα, the bathing solution was switched back to the control solution without toxin; this corresponds to time ‘zero’ in Fig. 9. During the next several minutes the K+ current gradually recovered, with an exponential time course, to the control level. Slow washout of TsTX-Kα was observed in four other cells: in the two cells in which this was studied systematically, the time constant of recovery averaged l1.2min. This is much slower than the rate of chamber washout (< 1 min; see Methods).
CTX block of IKf
The effects of CTX have been extensively studied. In addition to blocking large conductance Ca2+-activated K+ channels (MacKinnon & Miller, 1988; Stampe et al. 1995), CTX blocks voltage-dependent, Shaker-type K+ channels (MacKinnon, Reinhart & White, 1988; Goldstein & Miller, 1993). CTX also blocks a rapidly inactivating voltage-dependent K+ channel as well as a Ca2+-activated K+ channel in rat brain synaptosomes (Schneider, Rogowski, Krueger & Blaustein, 1989; Blaustein et al. 1991). The blocking action of CTX on voltage-dependent K+ channels in rat DRG cells was investigated in order to compare the effects of this toxin with those of TsTX-Kα, which blocks a different (non-inactivating) K+ channel in rat brain synaptosomes (Blaustein et al. 1991). Qualitatively, CTX block (Fig. 10) looks similar to the block by TsTX-Kα. Much of the outward current at -10 mV (which reflects mainly IKf) was blocked by 200nM CTX (Fig. 10A). This CTX-sensitive current is the rapidly activating, rapidly deactivating current (Fig. 10B). At +40 mV, where both IKf and IKs are activated, 200nM CTX only blocked IKf (Fig. 10C and D). Thus, both CTX and TsTX-Kα selectively block IKf in rat DRG cells. Nevertheless, CTX block of IKf differs from that of TsTX-Kα in two important ways, as described below.
Voltage-dependent block of IKf by CTX
CTX block of Ca2+-activated K+ channels, as well as Shaker K+ channels, is voltage dependent, and can be reduced with depolarization (Park & Miller, 1992b; Goldstein & Miller, 1993). A similar effect was seen in rat DRG neurones: CTX block of IKf could be partially relieved by depolarizations to voltages positive to 0 mV, as illustrated in Fig. 11 (using the procedure described for Fig. 7). At voltages from -40 to -10 mV, 200nM CTX blocked most of IKf (Fig. 11A). However, as the voltage increased from 0 to +70 mV the block of IKf decreased. In the two cells where this was studied, the fraction of IKf blocked at -10mV averaged 91%, while at +70 mV it declined to 37%. CTX, like TsTX-Kα, had little or no effect on IKs (Fig. 11B).
Recovery from CTX block is rapid
In contrast to the moderately slow recovery from TsTX-Kα block, recovery from CTX block was relatively fast. In Fig. 12, time ‘zero’ corresponds to the time that the external solution was switched from one containing 200nM CTX back to the control solution (without toxin). As CTX dissociated from its binding site, the K+ current rapidly increased toward the control level along an exponential time course. Rapid recovery of IKf from CTX block was also observed in three other cells, and the recovery time constant, quantified in two cells, averaged 0.81 min; this is more than an order of magnitude faster than recovery from block by TsTX-Kα.
Pandinus toxin block of IKf is poorly reversible
The Pandinus toxin, PiTX-Kα, also selectively blocked IKf in rat DRG cells, as shown in Fig. 13. At either -10 mV (Fig. 13A) or +30 mV (Fig. 13C) 10nM PiTX-Kα blocked only the fast component of the tail current. As was the case with the other scorpion toxins we tested, the PiTX-Kα-sensitive current (Fig. 13B and D) was the rapidly activating, partially inactivating and rapidly deactivating current, indicative of IKf. The degree of block produced by PiTX-Kα was not a function of voltage: in four cells IKf was reduced by 69 ± 3.0% at -10 mV, and by 71 ± 4.7% at +60 mV.
In contrast to the block induced by TsTX-Kα and by CTX, block of IKf by PiTX-Kα was poorly reversible. Recovery was studied in four cells exposed to PiTX-Kα: washing with toxin-free solution for up to 1 h resulted in recovery of 0, 32, 34 or 42% of the K+ current. This observation appears to be consistent with earlier reports that crude Pandinus venom irreversibly blocks K+ current in frog myelinated nerve (Pappone & Cahalan, 1987) and in GH3 cells (Pappone & Lucero, 1988). Our toxin washout experiments in rat DRG cells indicate that the toxin dissociation rates follow the sequence: CTX > TsTX-Kα > PiTX-Kα.
Relative sensitivity of IKf to block by TsTX-Kα, CTX and PiTX-Kα
The three scorpion toxins tested in this study blocked IKf in rat DRG neurones at -10 mV with significantly different potencies (Fig. 8). PiTX-Kα was the most potent blocker of IKf, with an IC50 of 8.1nM. TsTX-Kα was approximately 4-fold less potent than PiTX-Kα, with an IC50 of 38nM. CTX was the weakest of the three blockers of IKf: 200nM CTX blocked 60 ± 7.4% (n= 3) of the current. Although we did not systematically study the dose–response relationship for CTX block, we would predict an IC50 of ∼125nM, based on 60% block at 200nM.
Two predominant voltage-dependent K+ channels in rat DRG
In this study, we have shown that there are two primary components of the macroscopic K+ current in rat DRG neurones. Using the whole-cell patch-clamp technique, K+ tail current analysis was used to identify two kinetically distinct tail current components with time constants that differ by a factor of 5, and are thus easily separable. The fast component of the tail current, IKf, was associated with a population of voltage-dependent channels that activated over a relatively negative voltage range (V1/2∼−10 niV), had relatively fast activation and deactivation kinetics, and at least partially inactivated with rapid kinetics. Thus, these channels have many properties similar to ‘A-type’, or Shaker-type, K+ channels (Hille, 1992). The slow component of the tail current, IKs, was associated with a population of voltage-dependent K+ channels that activated at more positive voltages (V1/2∼+16 mV), had relatively slow activation and deactivation kinetics and did not inactivate during a 1 s depolarization.
An alternative explanation for the complex K+ tail current might be that it was generated by a single population of voltage-dependent K+ channels which have a complex gating scheme. For example, a channel with two open states could perhaps generate the kinetic properties of the whole-cell current observed in DRG neurones. This explanation seems unlikely in view of two observations. First, the relative proportion of IKf and IKs in the total K+ current varied significantly in different cells. This seems to be consistent with multiple populations of channels which vary in their relative expression among cells in the DRG. Second, α-DaTX and the three scorpion toxins all selectively blocked only one component of the current, IKf. The simplest explanation for this result is that these toxins selectively blocked one subtype of K+ channel.
Scorpion toxin selectivity for K+ channel subtypes
We have investigated the ability of three scorpion peptide toxins (TsTX-Kα, PiTX-Kα and CTX) to inhibit the natively expressed voltage-dependent K+ channels found in rat DBG neurones. The three K+ channel toxins were selected for this study, in part, because they are known to have different selectivities in rat brain synaptosomes: PiTX-Kα and CTX both block rapidly inactivating, voltage-dependent K+ channels, whereas TsTX-Kα selectively blocks non-inactivating, voltage-dependent K+ channels, and only CTX blocks Ca2+-activated K+ channels (Blaustein et al. 1991; Rogowski et al. 1996). Thus, we were expecting to observe similar selectivity in rat peripheral nerve ganglia.
Many DRG neurones express primarily two types of voltage-dependent K+ channels: IKf, a low voltage activated, rapidly deactivating, partially inactivating current, and IKs, a high threshold, slowly deactivating, non-inactivating current. Surprisingly, all three scorpion toxins selectively blocked only IKf, with little or no effect on IKs. However, the three toxins did vary significantly in both potency and washoff rate. CTX was the weakest of the three blockers, and it dissociated from its binding site with the fastest kinetics. PiTX-Kα, the most potent blocker of IKf, was, at best, only partially reversible. TsTX-Kα was intermediate: it reversibly blocked IKf with a KD about an order of magnitude lower than that of CTX, and its washoff rate was about 10 times faster.
The selectivities of TsTX-Kα, PiTX-Kα and CTX for voltage-dependent K+ channels reported in this paper are different from previous reports in other preparations. For example, in rat brain synaptosomes TsTX-Kα blocks only a non-inactivating component of the 86Rb efflux (Blaustein et al. 1991), whereas PiTX-Kα blocks only an inactivating component of the current (Rogowski et al. 1996), as does CTX (Schneider et al. 1989). In addition, in rat hippocampal neurones TsTX-Kα also blocks a non-inactivating component of the K+ current (Eccles, Rogowski, Gu, Alger & Blaustein, 1994). The present results are somewhat different, in that TsTX-Kα and PiTX-Kα both selectively blocked the same component of the voltage-dependent K+ current: a partially inactivating, Shaker-like K+ current. One possible explanation for the difference is simply that different subtypes of voltage-dependent K+ channels are present in different neurones; in this case, there appears to be a clear difference between the K+ channels in brain and those in peripheral ganglion neurones from the same animal. K+ channels are very diverse: numerous subtypes of voltage-dependent K+ channels have been identified (Hille, 1992). Thus, the channels that generate IKf in rat DBG neurones, which are blocked by both TsTX-Kα and PiTX-Kα, probably represent a different subtype than either the inactivating, voltage-dependent K+ channels in rat brain synaptosomes, or the ‘A-type’ K+ channels in hippocampal neurones, as these two types are insensitive to TsTX-Kα. It is interesting that, as in DRG neurones, TsTX-Kα and PiTX-Kα both block a cloned Shaker-type K+ channel, the Kv1.2 channel, but the affinity of the Kv1.2 channel for the toxins is much higher than the channels studied in our experiments: the Kv1.2 IC50= 32 pM for PiTX-Kα (Rogowski et al. 1996) and 210pM for TsTX-Kα (Werkman et al. 1993).
Correlation between affinity and washout rate
An inverse correlation was observed between the apparent affinity of the toxin for its binding site and the apparent toxin-dissociation rate: the higher the affinity, the slower the dissociation. This appears to be a common feature of the interaction of many drugs with their binding sites on ion channels, as similar results have been obtained for block of Ca2+-activated K+ channels by mutants of CTX. Park & Miller (1992a) tested the ability of CTX mutants with altered electrostatic properties to block Ca2+-activated K+ channels. They identified several toxin mutants that exhibited altered affinity for the channel; the change in affinity was due mainly to alterations in the toxin-dissociation rates.
One interesting aspect of the relatively weak interaction between CTX and IKf channels is the fact that block by CTX can be partially relieved by making the membrane potential more positive. Neutralizing the positive charge at residue 27 in CTX (Fig. 1A) by mutating lysine to glutamine eliminated the voltage dependence of block of Ca2+-activated K+ channels (Park & Miller, I992α). This type of effect could not account for the lack of voltage dependence in the block of IKf produced by PiTX-Kα and TsTX-Kα, because both of these toxins also contain a lysine at the position corresponding to residue 27 in CTX (Fig. 1A).
Toxin structure-activity relationships
The observations reported here raise critical questions about the structure–activity relationships of the three scorpion K+ channel toxins examined in this study. Which specific toxin amino acid side-chains account for the markedly different affinities and different dissociation rates of the toxins? Also, how can the unique voltage sensitivity of CTX block be explained? Definitive answers to these questions surely require additional experimentation. Nevertheless, some speculation is warranted to narrow the focus of such inquiries in view of the substantial body of information that is already available about structure- activity relationships of CTX and mutant analogues (Goldstein et al. 1995).
Virtually all of the scorpion K+ channel toxins that have been studied to date (Miller, 1995; Rogowski et al. 1996; Tenenholz, Rogowski, Collins, Blaustein & Weber, 1997) appear to be structurally similar molecules. They are thirty-five to thirty-nine amino acids long, and they all consist of a β-sheet at the C-terminus that contains three cysteines. The β-sheet is linked to the central α-helix by two disulphide bridges (C13–C33 and C17–C35, based on the CTX numbering; see Fig. 1A and C), and to an N-terminal fragment by a third disulphide bridge (C7–C28). The backbone is thus held rigid, and is virtually identical in all of these scorpion K+ channel toxins.
The K+ channel toxins all block by binding to the external surface of the channel (Werkman et al. 1993; Goldstein et al. 1995; Rogowski et al. 1996). The specific amino acids that interact with the external vestibule of the channel, and thereby account for the specific binding properties of each toxin, are those that surround the invariant Lys-27 that apparently inserts into the channel pore (Park & Miller, 1992b). Figure 1A shows the sequences of the three toxins we studied, and Fig. 1B shows space-filling models based on NMR determination of CTX structure (Bontems et al. 1991). The views depicted in Fig. 1B show the surfaces of the toxins that probably face the channel vestibule (Goldstein et al. 1995), with Lys-27 in the centre. The models of PiTX-Kα and TsTX-Kα were based on sequence homologies that provide evidence for similarities to CTX in tertiary structure (for details see Rogowski et al. 1996); this has largely been confirmed for PiTX-Kα on the basis of NMR data (Tenenholz et al. 1997).
Amino acids identified as ‘critical’ or ‘influential’ for CTX binding to a mutant, cloned Shaker K+ channel (F425G; Goldstein et al. 1995) are shown in Table 1; many of these amino acids are labelled in Fig. 1B. The other amino acids were termed ‘indifferent’, as they apparently had negligible influence on CTX binding to the Shaker K+ channel, but this does not preclude the possibility that they may influence binding to the DRG ‘fast’ K+ channel. Table 1 also shows the identically positioned amino acids in TsTX-Kα and PiTX-Kα (and see Fig. 1B). Note that all five ‘critical’ amino acids in the CTX sequence are either identical or conservatively replaced: positively charged Arg-34 is replaced by the slightly larger, positively charged lysine in both TsTX-Kα and PiTX-Kα; and Tyr-36 is replaced by phenylalanine, which also has an aromatic ring, in PiTX-Kα. It seems probable that amino acid side-chain differences at other, less conserved positions on the docking surface of the toxins (Fig. 1B) must account for the 10- to 20-fold difference in affinities, and 10- or > 100-fold difference in dissociation rates between CTX, on the one hand, and TsTX-Kα and PiTX-Kα, on the other. A significant difference is that the positively charged Arg-25 is replaced by an uncharged alanine in TsTX-Kα, and by a small, uncharged but dipolar residue, asparagine, in PiTX-Kα.
Table 1. Critical and influential amino acids in CTX and the homologous amino acids in TsTX-Kα and PiTX-Kα
Toxin/K+ channel target
*These data refer to a mutant Shaker K+ channel (F425G), with a bulky phenylalanine in position 425 replaced by a smaller glycine to enhance CTX binding by about three orders of magnitude. From Goldstein et al. 1995; DR, delayed rectifier.
The possible impact of these differences on the binding surfaces of the toxins may be inferred from the models of the three toxins in Fig. 1B. The backbones of the three toxins are nearly identical (Fig. 1C), as are the orientations of the Lys-27 side-chains (facing in Fig. 1B and pointing downward in Fig. 1C). The nearby side-chains at position 25 are, however, very different for the three toxins: Arg-25 in CTX protrudes further than the side-chains on the other two toxins (Ala-25 in TsTX-Kα is unlabelled, and lies between Lys-18 and Pro-37; see Fig. 1B). We speculate that the nature of this side-chain, and possibly others in the vicinity, may markedly influence the characteristics of binding of the various toxins to the DRG IKf channel.
The superimposed stick models of the toxin backbones (with a few side-chains labelled), also illustrate some differences (Fig. 1C). The Ser-10 in CTX and TsTX-Kα, and the Pro-10 in PiTX-Kα appear to be closely aligned. But note the differences in the positions of the small, hydrophobic Leu-14 in TsTX-Kα, and the aromatic side-chains in Trp-14 in CTX and Tyr-14 in PiTX-Kα, as well as the differences between the C-terminal Ser-37 in CTX and Pro-37 in TsTX-Kα, and the positively charged Arg-38 in PiTX-Kα.
The significance of the absent three TV-terminal amino acids in PiTX-Kα is also unknown. It might, for example, minimize possible steric hindrance that could arise if the other toxins are so large that they may tend to pop out of the channel vestibule more readily than PiTX-Kα. This could contribute to the much slower dissociation of PiTX-Kα from its binding site on the DBG IKf channels, as compared with the other two toxins.
Analysis of the specific interactions between scorpion toxins and cloned homotetrameric K+ channel α-subunits has been very instructive (e.g. Goldstein et al. 1995; Stampe et al. 1995; Hidalgo & MacKinnon, 1995; Aiyar et al. 1995). Nevertheless, native K+ channels are usually hetero-multimeric proteins that contain four β- and four α-subunits (Dolly et al. 1994; Sewing, Roeper & Pongs, 1996); indeed, it seems possible that the β-subunit may modify toxin binding and toxin structure-activity relationships. Thus, in the absence of detailed structural information about the native K+ channels, it also seems to be instructive to employ comparative studies such as ours to identify the critical features of toxin structure that may influence toxin selectivity, affinity and dissociation rate.
We thank R. S. Rogowski for preparation of the toxins used in this study, T. C. Tenenholz and D. J. Weber for helpful discussions concerning toxin structure–activity relationships and T. C. Tenenholz for modelling the toxin structures used in Fig. 1B and C. This work was supported by National Institutes of Health grants NS16106 and NS34627, and by funds from the University of Maryland School of Medicine and the University of Maryland-Baltimore Graduate School.