Journal of Neurophysiology

ω-Conotoxin GVIA Alters Gating Charge Movement of N-Type (CaV2.2) Calcium Channels

Viktor Yarotskyy, Keith S. Elmslie


ω-conotoxin GVIA (ωCTX) is a specific blocker of N-type calcium (CaV2.2) channels that inhibits neuropathic pain. While the toxin appears to be an open channel blocker, we show that N-channel gating charge movement is modulated. Gating currents were recorded from N-channels expressed along with ß2a and α2δ subunits in HEK293 cells in external solutions containing either lanthanum and magnesium (La-Mg) or 5 mM Ca2+ plus ωCTX (ωCTX-Ca). A comparison showed that ωCTX induced a 10-mV right shift in the gating charge versus voltage (Q-V) relationship, smaller off-gating current time constant (τ QOff), a lower τ QOff voltage dependence, and smaller on-gating current (QOn) τ. We also examined gating current in La-Mg plus ωCTX and found no significant difference from that in ωCTX-Ca; this demonstrates that the modulation was induced by the toxin. A model with strongly reduced open-state occupancy reproduced the ωCTX effect on gating current and showed that the gating modulation alone would inhibit N-current by 50%. This mechanism of N-channel inhibition could be exploited to develop novel analgesics that induce only a partial block of N-current, which may limit some of the side effects associated with the toxin analgesic currently approved for human use (i.e., Prialt).


ω-conotoxin GVIA (ωCTX) is a highly specific blocker of N-type calcium channels (Olivera et al. 1994) that is thought to act by occluding the pore (Ellinor et al. 1994; Liang and Elmslie 2002; McDonough et al. 2002). The supporting evidence includes 1) the putative toxin binding site is located near the pore (Ellinor et al. 1994; Feng et al. 2001), 2) the speed of toxin binding is negatively affected by increased divalent cation concentration (Boland et al. 1994; Elmslie et al. 1994; Liang and Elmslie 2002), and 3) ωCTX block is prevented by ω-AgaIIIa binding, which partially occludes the pore to reduce channel conductance (McDonough et al. 2002). The toxin is a highly basic peptide that carries a net +5 charge, and several charged amino acids have been found to be important for toxin binding (Kim et al. 1995; Lew et al. 1997). The binding of this highly charged peptide to external surface of the N-channel could affect gating charge movement and, thus, channel opening, but pore block by this toxin prevents direct observation of ionic currents to test this idea. However, Jones et al. (1999) found that ωCTX affected inactivation-induced immobilization of N-channel gating currents at depolarized voltages; this supports toxin modification of N-channel gating and demonstrates a method of examining that modification. Several toxins (e.g., ω-AgaIVa, grammotoxin) are known as gating modifiers of voltage-gated calcium channels. The general blocking mechanism is to stabilize close state occupancy of the channels, which results in right-shifted activation voltage dependence and slowed activation kinetics (Catterall et al. 2007; McDonough 2007). We test the hypothesis that ωCTX is also a gating modifier of N-type calcium channels.

We compared gating currents in three external solutions: lanthanum-magnesium (La-Mg) (Jones et al. 1997–1999), ωCTX in Ca2+ (ωCTX-Ca) and ωCTX in La-Mg (ωCTX-LaMg), and found that the toxin alters the voltage dependence and kinetics of gating charge movement independent of the ionic composition of the external solution (Ca2+ or La-Mg). We conclude the ωCTX-modulation of N-channel gating would inhibit action potential induced Ca2+ influx. Thus the toxin-binding site provides a target for the development of drugs that could specifically modulate N-channel gating to reduce, instead of completely block, function.


HEK cell transfection

HEK293 cells were maintained in standard DMEM/Glutamax medium containing 10% fetal bovine serum (FBS) and 1% antibiotic-antimycotic (regular medium), at 37°C in 5% CO2 incubator. For transfection, the medium was changed to DMEM/F-12 medium containing 10% FBS and 1% antibiotic-antimycotic. HEK293 cells were transfected using calcium phosphate precipitation method (Yarotskyy and Elmslie 2007) by adding to the media 1 ml of precipitated transfecting solution containing: HEPES-buffered saline (HeBS), 50 mM CaCl2 and cDNA plasmids as follow: 11.5 μg α1B-CFP, 8.5 μg α2δ, 5.5 μg β2a subunits, 2.15 μg TAG (to increase expression efficiency), and incubating for 8 h after which the medium was replaced by the standard DMEM medium. N-channel expressing cells were visualized by CFP, which was attached to the N-terminus of α1B1B-CFP). Our preliminary data have shown no effects of this attachment on either ionic or gating currents. All clones used in this study were obtained from Dr. Blaise Peterson (Penn State College of Medicine). Transfected cells were split into 35-mm dishes that also served as the recording chamber.

Measurement of currents

Cells were voltage-clamped using the whole cell configuration of the patch-clamp technique. Pipettes were pulled from Schott 8250 glass (Garner Glass, Claremont, CA) on a Sutter P-97 puller (Sutter Instruments, Novato, CA). Currents were recorded using an Axopatch 200A amplifier (Molecular Devices, Sunnyvale, CA) and digitized with ITC-18 data acquisition interface (Instrutech, Port Washington, NY). Experiments were controlled by a Power Macintosh G3 computer (Apple Computer, Cupertino, CA) running S5 data-acquisition software written by Dr. Stephen Ikeda (National Institutes of Health, NIAAA, Bethesda, MD). Leak current was subtracted on-line using a −P/4 protocol for ionic currents and −P/8 for gating currents. All recordings were carried out at room temperature, and the holding potential was −120 mV. Both ionic and gating currents were digitized at 50 kHz after analog filtering at 5–10 kHz for ionic current and at 10 kHz for gating currents. Gating currents were recorded using 40–60% series resistance compensation. The maximum voltage-clamp error was <5 mV in 13 cells, but in one cell, the error ranged from 5.8 to 7.5 mV. There was no correlation between gating current parameters and maximum voltage error and/or series resistance.


For the transfection, HeBS contained (in mM) 140 NaCl, 25 HEPES, and 1.4 Na2HPO4 and the pH = 7.10 was adjusted using 5 N NaOH. The internal pipette solution contained (in mM) 104 N-methyl-d-glucamine-Cl (NMG-Cl), 14 creatine-PO4, 6 MgCl2, 10 NMG-HEPES, 5 Tris-ATP, 0.3 Tris-GTP, and 10 NMG-EGTA with osmolarity = 280 mosM and pH = 7.4. The external solution for recording ionic currents contained (in mM) 5 CaCl2, 145 NMG-Cl, and 10 NMG-HEPES, osmolarity = 325 mosM and pH = 7.4. We utilized three external solutions for recording gating currents. The La-Mg solution contained (in mM): 0.2 LaCl3, 5 MgCl2, 0.1 NMG-EGTA, 145 NMG-Cl, and 10 NMG-HEPES with osmolarity = 325 mosM and pH = 7.4. The ωCTX-LaMg solution was made by adding 5 μM ωCTX to the La-Mg solution, whereas the ωCTX-Ca solution was made by adding 5 μM ωCTX to the ionic current external solution. This ωCTX concentration was chosen to obtain a reasonable toxin blocking speed so that recordings in La-Mg ± ωCTX could be done within the same cell. The ωCTX on-rate is highly sensitive to divalent cation concentration (Boland et al. 1994; Liang and Elmslie 2002) so that complete block can require >10 min at low micromolar concentrations (Liang and Elmslie 2001). At 5 μM, complete block of ionic currents was obtained after ∼5 min so that the effect of ωCTX on gating currents could be regularly assessed within a single cell. In all La-Mg-containing solutions, the free [La3+] was 0.1 mM because La3+ was chelated by 0.1 mM EGTA (Jones et al. 1997–1999). Test solutions were applied from a gravity-fed perfusion system with an exchange time of 1–2 s.

Data analysis

Data were analyzed using IgorPro (WaveMetrics, Lake Oswego, OR) running on a Macintosh computer. Voltage dependence of activation (activation I-V) was measured from ionic tail currents by averaging 0.3 ms of current beginning 0.3 ms following repolarization to −60 mV. The tail current amplitude was plotted against step voltage and fitted by a Boltzmann function from which we determined half activation voltage (V0.5) and slope factor. Deactivation voltage dependence was determined from a protocol that consisted of 10-ms steps to +60 mV followed by 22-ms tail voltage steps ranging from +20 to −160 mV. Ionic tail currents were fitted by a single-exponential function to determine τDeact. The voltage dependence (Ve) of τDeact was determined by fitting the τDeact-voltage relationship with a single-exponential function. Gating currents were digitally filtered using the IgorPro binomial smooth function (a Gaussian filter) with a 2.3-kHz cutoff. The gating current waveform was not altered, but filtering greatly improved the signal-to-noise ratio. Small steady-state step/tail currents introduced by the −P/8 leak subtraction protocol was measured at the end of the step/tail and subtracted from gating currents. Total charge moved by the depolarization or repolarization (QOn or QOff, respectively) was calculated by integrating gating currents over the entire voltage step. The Q versus voltage curve was fitted by a Boltzmann function to yield V0.5, slope, and maximum Q (Qmax). The Off-gating current τ (τ QOff) was calculated as for ionic current τDeact (described in the preceding text). The magnitude of Off-gating currents was calculated by averaging 0.3 ms at the peak. The on-gating current τ (τ QOn) was determined by fitting a single-exponential equation to the decay phase of QOn.

Group data were calculated as means ± SD throughout the paper. Paired t-test was used for within-cell comparisons. One-way ANOVA with Tukey HSD post hoc test was used to test for differences among three or more independent groups.

Computer simulations

Simulated currents were generated using Axovacs 3 (written by Stephen W. Jones, Case Western Reserve University) on a Macintosh G4 computer running Virtual PC 6 (Microsoft, Seattle, WA). Voltage dependent rate constants (kx) in the model were calculated from: kx = Ax exp(VzxF/RT), where Ax is the rate constant at 0 mV, V is the voltage, and zx is the charge moved, and R, T, F are the gas constant, absolute temperature and Faraday's constant, respectively. Simulated currents were analyzed using IgorPro.


ω-conotoxin GVIA was purchased from Bachem Americas (Torrance, CA). DMEM/F12, DMEM, FBS, 100× antibiotic/antimycotic were from Invitrogen (Carlsbad, CA). Other chemicals were obtained from Sigma (St. Louis, MO).


N-channel gating current in La-Mg

We substituted Ca2+ with equimolar Mg2+ and 100 μM free La3+ (La-Mg) to isolate gating currents from ionic current (Jones et al. 1997, 1999). The ability of our −P/8 leak subtraction method to remove linear capacitative currents was checked by stepping from −120 to −100 mV and from −120 to −140 mV, which revealed no signs of nonlinearity. HEK293 cells with no detectable ionic currents were used to further show that our −P/8 method removed linear capacitative and leak current at all test voltages (not shown, n = 6).

Well-resolvable gating currents were observed for cells with peak ionic currents (5 mM Ca2+) ≥2 nA (Fig. 1A). The voltage dependence of charge movement (Q-V) was obtained from 20-ms steps to voltages ranging from −120 to +80 mV (QOn) followed by a 25-ms repolarization to −60 mV (QOff). These durations allowed gating currents to completely develop at all test voltages. The integrated gating current (QOn and QOff) showed a monotonic increase with step voltage that saturated at voltages ≥ +20 mV (Fig. 1C). The normalized QOn-V and QOff-V were nicely superimposed and were ∼20 mV left-shifted compared with the I-V (Fig. 1D), which is similar to previous observations (Jones et al. 1999). This shift was quantified by fitting the Boltzmann equation to the I-V and Q-V relationships to determine V0.5 and slope. On average the V0.5 for the I-V in 5 mM Ca2+ was 13.1 ± 3.2 mV (n = 15), whereas in La-Mg, the QOn-QOff V0.5 = −7.6 ± 5.7 mV (n = 11). Thus the ΔV0.5 = −20.7 mV between the I-V in Ca2+ and Q-V in La-Mg. However, in making these comparisons, we noticed that the QOff V0.5 (V0.5 = −5.0 ± 5.4 mV) was consistently more positive than that for QOn (V0.5 = −10.1 ± 6.2 mV, n = 11, P < 0.05). In addition, the maximum QOff (0.46 ± 0.17 pC) was consistently larger than the QOn max (0.39 ± 0.13 pC) so that the ΔQmax = 0.08 ± 0.04 pC, which is statistically significant (P < 0.05). This ∼20% increase in QOff max likely results from a small contamination by ionic current (relatively large driving force at −60 mV), which could also explain the ∼5-mV right-shift in QOff V0.5 (relative to QOn). Indeed, Boltzmann fitting of a theoretical Q-V relationship that was comprised of 80% QOn (V0.5 = −10 mV) and 20% tail current (V0.5 = 13 mV) yielded V0.5 = −5.9 mV, which is very close to the V0.5 = −5.0 that we measured from the QOff-V relationship and supports the idea that QOff had a minor (20%) contamination by ionic current. One caveat to this conclusion is that the theoretically blended Q-V relationship showed a larger slope factor (11.3) than that for either the Q-V (10.0) or I-V (11.0) relationships. We conclude that ionic current contamination was minimal and that the Q-V relationship was ≥20 mV left-shifted relative to the I-V.

FIG. 1.

N-channel gating currents in lanthanum-magnesium (La-Mg). A: the representative traces show ionic currents recorded in the 5 mM Ca2+-containing extracellular solutions during voltage steps to −10, 10, and 40 mV (top). The voltage protocol is shown below the traces. B: the representative traces show gating currents recorded in La-Mg for voltage steps to −50 mV (top), 0 mV (middle), and 80 mV (bottom). C: the Q-V relationship is shown for both QOn (•) and QOff (○) in La-Mg. The smooth curves are single Boltzmann equation fits to the Q-V relationships with V0.5 = −9.6 and –6.3 mV, slope factor = 10.0 and 8.6, and Qmax = 0.28 and 0.32 for QOn and QOff, respectively. D: the steady-state activation I-V for ionic currents recorded in 5 mM Ca2+ (ICa, Embedded Image) is right shifted compared with Q-V recorded in La-Mg from the same cell as A and B. ▪ (on), QOn; □ (off), QOff. Q-V and I-V relationships were fitted by single Boltzmann functions (—) with V0.5 = 9.9 mV, slope factor = 9.7 for the I-V and the same V0.5 and slope factor as in C for QOn-V and for QOff-V. All data are from the same cell.

ωCTX modulates gating current

We began investigating potential effects of ωCTX by measuring gating current in 5 mM Ca2+ + ωCTX (Fig. 2A). If ωCTX was without effect, we expected to observe a Q-V relationship similar to that in La-Mg. However, the actual Q-V curve was right-shifted from that in La-Mg so that it was much closer to the I-V relationship (Fig. 2). Boltzmann fitting of the Q-V relationship revealed that the V0.5 in ωCTX-Ca was significantly more depolarized than that in La-Mg (Fig. 2E) with ΔV0.5 = 9.3 mV. In addition, the slope factor of the Q-V relationship was significantly increased by ωCTX (Fig. 2F). We thought of two possibilities to explain the difference in V0.5 between La-Mg and ωCTX-Ca. The first is that surface charge screening in La-Mg is different from that in Ca2+ so that the true voltage dependence of charge movement is shown by ωCTX-Ca. However, previous results have demonstrated that the QOn-V relationship in La-Mg is identical to that in Ca2+ when equimolar substitutions are made between Ca2+ and Mg2+, which suggests that surface charge screening is similar in these two solutions (Jones et al. 1999). The second possibility is that ωCTX alters N-channel gating charge movement. These possibilities were distinguished by examining the effect of ωCTX on gating current in La-Mg (ωCTX-LaMg), which will have an identical surface charge screening effect as La-Mg. The results clearly demonstrate that the Q-V relationships in ωCTX-Ca and ωCTX-LaMg are superimposed and that both are right-shifted relative to that in La-Mg measured in the same cell (Fig. 2D). Furthermore, averaged results show that V0.5 in ωCTX-Ca and ωCTX-LaMg are indistinguishable from each other, but both are significantly depolarized from that in La-Mg (Fig. 2E). The Boltzmann slope factor is increased by ωCTX relative to that in La-Mg, and this effect is also independent of the ionic composition of the external solution to which the toxin was added (Fig. 2F). Thus ωCTX modulates N-channel gating current by shifting activation to more depolarized voltages. The depolarized shift by ωCTX is reminiscent of a surface charge screening effect, which could be induced by the net positive charge of the toxin. However, such an effect cannot explain the larger slope factor of the Q-V relationship in ωCTX.

FIG. 2.

The Q-V relationship is right-shifted by ωCTX. A: representative gating currents in ωCTX-Ca are shown for voltage steps to −50 mV (top), 0 mV (middle), and 80 mV (bottom). B: the Q-V relationship in ωCTX-Ca is shown for both QOn (▪) and QOff (□) from the same cell as in A. The Boltzmann equation fits (—) generated V0.5 = 5.3 and 7.5, slope factor = 13.2 and 12.3, and Qmax = 0.32 and 0.34 for QOn and QOff, respectively. C: the activation I-V (ICa, Embedded Image) for ionic currents is shown along with the QOn-V (on, ▪) and QOff-V (off, □) from the same cell. —, Boltzmann equation fits (as described for Fig. 1) with V0.5 = 11.3 mV, slope = 10.5 for I-V and the same V0.5 and slope factor as in B for QOn-V and for QOff-V. —, left-shifted: the Boltzmann fit for QOn-V in La-Mg from the same cell and is reproduced here to highlight the ωCTX-induced shift. D: comparison of the QOn-V relationships in La-Mg (Embedded Image), ωCTX-Ca (○), and ωCTX-LaMg (•) from the same cell. The Boltzmann fits (—) yield V0.5 = −12.7, −4.7, −1.1 mV, slope factor = 8.3, 13.7, and 12.7, and Qmax = 0.57, 0.56, and 0.56 for La-Mg, ωCTX-Ca, and ωCTX-LaMg, respectively. E and F: the average (±SD) V0.5 (E) and slope factor (F) is shown for ionic (ICa) and gating currents (La-Mg, ωCTX-Ca, and ωCTX-LaMg). The significant differences for all comparisons were determined using ANOVA with Tukey HSD post hoc test (P < 0.05). The lower case letters above each column indicate the data sets that differ significantly. The number of cells averaged for E and F were 14 for ICa, 11 for La-Mg, and 4 for both ωCTX-Ca and ωCTX-LaMg.

Like La-Mg, we observed that QOff max (0.44 ± 0.17 pC) was consistently larger than QOn max (0.38 ± 0.16 pC) in ωCTX-Ca (n = 4). Interestingly, the 16% increase in Qmax was associated with a right shift in V0.5 for QOff (4.0 ± 1.1) versus QOn (−0.5 ± 4.1). As with La-Mg, it seems that QOff in ωCTX-Ca was slightly contaminated with ionic current. This is supported by measurements in ωCTX-LaMg, which also showed a slightly larger (12%) QOff max (0.35 pC) versus QOn max (0.31 pC) and no significant shift in V0.5 (QOn V0.5 = 2.0 ± 7.1 mV vs. QOff V0.5 = 0.3 ± 1.2 mV, n = 4). The minor ionic current contamination of QOff in ωCTX-Ca likely results from the very slow time course of N-current block by ωCTX in the presence of mM divalent cation concentrations (Boland et al. 1994; Elmslie et al. 1994; Liang and Elmslie 2001).

ωCTX speeds gating current kinetics

The steady-state Q-V relationship is right-shifted 10 mV by ωCTX, which could result from the positively charged toxin screening negative charges on the surface of the channel. This idea was examined by measuring the ωCTX affect on gating current kinetics. The surface charge hypothesis will be supported if kinetic parameters are also shifted 10 mV depolarized by the toxin. We initially investigated the QOff time constant (τ QOff), which was assessed by fitting a single-exponential equation to the off-gating currents. When compared with the deactivation time constant for ionic current (τDeact), τ QOff in La-Mg was larger at all voltages examined and appeared to reach an asymptotic value of ∼0.3 ms at voltages less than −100 mV, which suggests the presence of a rate limiting voltage-independent transition to charge movement that has not been observed with N-current (Fig. 3) (Buraei et al. 2005). In addition, the e-fold change of τ QOff with voltage (Ve) was not as steep as that for τDeact (Fig. 3B). Thus channel closing is faster and more voltage dependent than QOff movement. The addition of ωCTX induced a significant reduction in τ QOff at voltages greater than −100 mV (Figs. 3B and 4C), but τ QOff appeared to be approaching the same limiting value (∼0.3 ms) at strongly hyperpolarized voltages. To test the surface charge hypothesis of the toxin effect, we shifted the τ QOff-V relationship 10 mV hyperpolarized, but found that this shift was not sufficient to normalize the data in La-Mg ± ωCTX (Fig. 4). Indeed, τ QOff in La-Mg was still significantly larger than that in ωCTX at voltages greater than −80 mV (Fig. 4C). Shifting the τ QOff-V relationship by 20 mV yielded good agreement between the two data sets (± toxin), whereas a 30-mV shift was clearly too much (Fig. 4C). As with the Q-V relationship, τ QOff in ωCTX was identical when 5 mM Ca2+ was substituted for La-Mg, so that the effect is clearly isolated to the toxin. It appears that the ωCTX effect on gating current is not consistent with a simple charge screening effect.

FIG. 3.

ωCTX speeds off-gating current. A: typical gating currents recorded in La-Mg at tail voltages of −20, −40, and −60 mV following a 10-ms 60-mV step. B: ionic current τDeact recorded in 5 mM Ca2+ (○, n = 12), τ QOff obtained in La-Mg (•, n = 11) and τ QOff recorded in ωCTX-LaMg (•, n = 4) were determined by fitting currents with a single-exponential function and are plotted against tail voltage. The τ vs. voltage relationships between −10 and −80 mV were fit (—) by an exponential function to yield Ve for each condition, with Ve = e-fold for −24.8 mV, e-fold for −40.6 mV, and e-fold for −54.4 mV for ionic current τD, QOff τ in La-Mg and QOff τ in ωCTX-LaMg, respectively. C: mean Ve (±SD) is shown for ionic currents (ICa) and gating currents (La-Mg, ωCTX-Ca, and ωCTX-LaMg). The significant differences were determined using ANOVA with Tukey HSD post hoc test (P < 0.05). The lower case letters above each column indicate the data sets that differ significantly.

FIG. 4.

ωCTX induces a 20-mV shift in τ QOff. All data were obtained in La-Mg ± ωCTX. A: typical gating currents in La-Mg with (black) and without (gray) ωCTX. B: τ QOff from single-exponential fitting of off-gating current is plotted vs. tail voltage for data with (open squares) and without ωCTX (open circles). The gray filled squares (+ωCTX ΔV) represents the +ωCTX τ QOff left-shifted 20 mV to illustrate the close correspondence with the –ωCTX results. C: the difference between τ QOff ± ωCTX (Δτ QOff) is plotted vs. voltage for the data shifted 0 (open circles), −10 (open squares), −20 (filled circles), and −30 mV (open triangles). The results were averaged from 5 cells.

This conclusion was further supported by examining the effect of ωCTX on τ QOn (Fig. 5), which was measured by fitting a single-exponential equation to the decay phase of the on-gating current. The τ QOn versus V relationship was bell-shaped with a peak near 0 mV in La-Mg and +10 mV in both ωCTX-LaMg and ωCTX-Ca (Fig. 5B). However, the most striking difference induced by ωCTX was the strong reduction of τ QOn at voltage < +20 mV. Surprisingly, ωCTX did not affect τ QOn at more depolarized voltages (Fig. 5), which is inconsistent with the simple surface charge screening hypothesis. Once again the substitution of 5 mM Ca2+ for La-Mg had no impact on the ωCTX effect of τ QOn (Fig. 5).

FIG. 5.

ωCTX speeds on-gating charge movement. A and B: comparison of gating currents recorded in La-Mg (thick black trace), wCTX-LaMg (thin gray trace), and ωCTX-Ca (thin black trace) elicited during voltage steps to 0 (A) and +25 (B) mV (QOn) and –60 mV (QOff). C: the mean ± SD τ QOn vs. step voltage plot for gating currents recorded in La-Mg (black squares), wCTX-LaMg (gray squares), and ωCTX-Ca (gray circles). The values ranging from −15 to 0 mV were significantly different (P < 0.05, paired t-test, n = 4) between La-Mg and ωCTX-Ca. The τ QOn values at −15, −10, and 0 mV were significantly different between La-Mg and ωCTX-LaMg, and there were no significant differences between ωCTX-Ca and ωCTX-LaMg at any test voltages.

N-channel model

Having examined the effect ωCTX on gating current, we were interested in determining how the toxin was affecting the channel to generate these effects. Our approach was to utilize our five-state model that was previously developed to explain the effect of the N-channel agonist roscovitine on ionic current (Fig. 6A) (Buraei et al. 2005). This model was developed for N-current (3 mM Ba2+) recorded in bullfrog sympathetic neurons, and the only change was to adjust the rate constants to account for a 20-mV depolarization in gating parameters that was likely induced by recording expressed N-channels in 5 mM Ca2+ (Table 1). This model that was originally developed to reproduce ionic currents was able to nicely reproduce our gating current results (Fig. 6). The Q-V relationship was left-shifted relative to the I-V (Fig. 6C) but only 10 mV instead of the 20 mV observed in our recordings. The QOff τ was larger than τDeact, but the Ve was not decreased as for our recorded currents (Fig. 6D). The simulated QOff τ also did not reach an apparent asymptote of 0.3 ms at voltages less than −80 mV. It is likely that the model requires a voltage-independent step that limits QOff kinetics but not channel closing. A voltage-independent closed state (mean closed time = 0.3 ms) was observed in single N-channel recordings, but it was not clear if that state was along the pathway to channel opening/closing (Lee and Elmslie 1999). Together these results suggest that this voltage-independent transition is along the pathway to and from opening/closing. We have yet to incorporate this transition into our model, which could explain the similar Ve between QOff τ and τDeact in our model. The QOn-V relationship was bell-shaped as were the data, but the model values were smaller. Thus the model nicely captured the voltage-dependent properties of N-channel charge movement.

FIG. 6.

A N-channel gating model can reproduce the ωCTX effect on gating current. A: the model scheme is reproduced from Buraei et al. (2005), and the parameters for each transition rate are given in Table 1 for both Cntl and toxin models. B: simulated gating currents at +60 (QOn) and −60 (QOff) mV using the Cntl and toxin models were filtered using an 2.3-kHz digital RC filter (R = 8.5 MΩ * 0.6 and C = 15 pF) to mimic the filtering induced by our average uncompensated series resistance (40% compensation) and average cell capacitance. C: a comparison of the I-V and Q-V relationships for the Cntl and toxin models. The activation I-V (□) was measured from tail currents at −60 mV following 10-ms steps to voltages ranging from −90 to +60 mV using the Cntl model. The on-gating charge (Q) was determined using the same voltage protocol by integrating the on-gating current over the 10-ms step for both the Cntl (▪) and toxin (Embedded Image) models (compare with Fig. 2). —, Boltzmann fits to the simulated data to yield V0.5 = 9, −2, 7 mV and slope factor = 10, 10, 14 for Cntl I-V, Cntl Q-V, and toxin Q-V, respectively. D: a comparison of simulated ionic current τDeact (○) from the Cntl model with gating current τ QOff vs. voltage relationships from both the Cntl (▪) and toxin (Embedded Image) models. The ionic tail current and off-gating currents were generated at voltages ranging from 0 to −140 mV following a 10-ms step to +60 mV and were fit using a single-exponential equation as for the recorded currents. The smooth lines are single-exponential fits with Ve = −25 mV (Cntl τDeact), −25 mV (Cntl τ QOff), and −30 mV (toxin τ QOff, •; compare Fig. 3). The gray diamonds represent the τ QOff toxin data shifted –20 mV to show the correspondence with the Cntl τ QOff values (compare Figs. 3 and 4). E: a comparison of τ QOn for the Cntl and toxin models. Note that the values are similar at the extreme negative and positive voltages as with the data shown in Fig. 5. F: action-potential-generated simulated currents are compared between the Cntl and toxin models. The action potential was generated using the Hogkin-Huxley (1952) model with a sodium conductance = 40 pS and a potassium conductance = 36 pS. The calcium conductance for each simulation was 1 pS, which permitted the observation of the action potential induced Ca2+ influx without affecting the action potential waveform (Buraei et al. 2005).

View this table:

Rate parameters for control and toxin models

We altered the model rate constants to determine the minimal changes needed to simulate the effect of ωCTX on gating current. The toxin induced changes we wanted to reproduce were 1) a 10-mV right shift in the Q-V relationship, 2) a lower slope to that relationship, 3) a 20-mV right shift in the QOff τ versus V relationship, and 4) a smaller τ QOn at voltages less than +20 mV. As a first step, we adjusted all the rate constants so that they reflected a 10-mV depolarization in gating (e.g., as expected for a surface charge screening effect). This global change yielded a shift in the Q-V relationship, but the slope factor was not altered. We next specifically increased the rate constants between the high and low Po open states (excluding C3 ←→ O4), which achieved changes in QOff τ but induced only minor shifts in the Q-V or QOff τ-V relationships. Based on this preliminary work, it appeared that changes in the gating transitions around the open states produced the best results, but we need a larger voltage shift. This was achieved (toxin model) by including changes to C3 ←→ O4 transitions along with those between the open states (O4 ←→ O5), which resulted in a 10-mV right shift in the Q-V relationship with lower slope and a 20-mV right shift in the QOff τ-V relationship (Fig. 6). The model also predicted the toxin-induced decrease in QOn τ at voltages near 0 mV with little or no change in τ at more depolarized and hyperpolarized voltages (Fig. 6E). Our simulations revealed that ωCTX affects N-type channels by destabilizing the open states.

With this model that reproduces the toxin-induced changes in gating charge movement, we wanted to investigate the impact of these gating changes on ionic current. As expected from destabilization of the open state, the model generated smaller ionic currents in the presence of the toxin. However, we were most interested in the effect on action-potential-induced Ca2+ influx because this is most relevant physiologically. As we had done previously (Buraei et al. 2005), we determined the Ca2+ influx during an action potential generated by the Hodgkin-Huxley model (Hodgkin and Huxley 1952). The Ca2+ influx was calculated by integrating the current over the entire waveform, which showed that the toxin reduced Ca2+ influx by 52%. The implication is that the ωCTX would maximally block 1/2 of the N-type calcium current even if it did not block the pore.


Our fundamental finding is that ωCTX modulates N-channel gating charge movement. The toxin induced a 10-mV right shift in the Q-V curve, faster off-gating charge movement at V > −80 mV, and faster on-gating charge movement at V < +20 mV. Our work also supports the previous conclusion that the voltage dependence of charge movement is not affected by substitution of external Ca2+ with equimolar Mg2+ plus 100 μM free La3+ (Jones et al. 1997, 1999).

The Yue lab previously demonstrated that Mg2+ + a free La3+ concentration of 100 μM (200 μM La3+ and 100 μM EGTA) generated good gating current isolation. In agreement with them, we found that La-Mg solution blocks ∼99.7% of ionic currents during voltage steps (Jones et al. 1997, 1999), which yielded a maximum contamination of ∼20% ionic current in our gating current recordings in La-Mg. Some ionic current contamination may have also been present in ωCTX-Ca (∼16% maximum). However, this cannot explain the toxin effect on gating currents because the contamination is smaller than that estimated for La-Mg and results were identical in ωCTX-LaMg in which no contaminating ionic current could be detected. Thus it is clear that N-channel gating charge movement is modulated by ωCTX, and our modeling reveals that this modulation alone would inhibit ionic current by 50%.

ωCTX as a gating modulator

Previous reports describing ωCTX block of N-type channels concluded that the toxin was a pore blocker (Ellinor et al. 1994; Liang and Elmslie 2002; McDonough et al. 2002). The supporting evidence included 1) the putative toxin binding site was found to lie near the pore within domain III (Ellinor et al. 1994; Feng et al. 2001), 2) the speed of toxin binding was negatively affected by increased divalent cation concentration (Boland et al. 1994; Liang and Elmslie 2002), and 3) ωCTX block was prevented by ω-AgaIIIa binding, which partially occludes that pore to reduce channel conductance (McDonough et al. 2002). However, Jones et al. (1999) found that ωCTX affected inactivation-induced immobilization of N-channel gating currents at depolarized voltages, which suggested that the toxin could also modify N-channel gating. We have extended this observation to show that the bound toxin affected several properties of N-channel charge movement. The Q-V relationship in either ωCTX-Ca or ωCTX-LaMg was 10-mV right-shifted relative to that in La-Mg. While this effect was surprising, it was small relative to the other effects of ωCTX that we observed. The τ QOff versus V relationship in ωCTX needed to be shifted −20 mV to match that in La-Mg. τ QOff is likely to be more sensitive to changes in open-open and/or open-closed transitions than the steady Q-V relationship, which implies that these transitions are more strongly affected by ωCTX. The QOn τ was strongly decreased by the toxin but only at intermediate voltages. The effect of ωCTX was minimal at the voltage extremes, where N-channel Po is either near zero (≤ −20 mV) or maximal (≥ +20 mV), which explains the significantly lower slope to the Q-V relationship in ωCTX. Within our model, the toxin effect was reproduced by increasing the rate constants out of and decreasing the rate constants into the open states. Together, these changes decreased open state occupancy, which resulted in the toxin-induced 52% reduction in Ca2+ influx during the action potential. It is interesting that changes in open state transitions best reproduced the toxin effects on gating current, which suggests that closed state transitions are relatively toxin insensitive.

A previous study on N-channels expressed in Xenopus oocytes concluded that ωCTX bound with higher affinity to inactivated channels (Stocker et al. 1997). This effect predicts that the toxin would immobilize N-channel gating charge movement (Jones et al. 1999), but maximum QOn and QOff were not altered by ωCTX in our study. One potential reason is that we expressed N-channels with β2a subunits, which abrogates voltage-dependent inactivation (Dolphin 2003), and maintained holding potential at −120 mV to minimize inactivation. Both of these techniques helped to maximize gating current and enhanced our ability to detect the impact of ωCTX on the kinetics of charge movement. However, these techniques also caused us to miss a potential effect of ωCTX to immobilize charge movement by enhancing N-channel inactivation (Stocker et al. 1997).

The mechanism for the ωCTX effect on N-channel gating charge kinetics is not known, but there are several possibilities. One is that toxin binding allosterically destabilizes S4. The putative toxin binding site is located within domain III between S5 and the P-loop (Ellinor et al. 1994; Feng et al. 2001). The site is close to the pore, which likely explains the primary blocking mechanism, but the presence of the toxin could allosterically affect S4 movement (perhaps domain III S4). A second possibility is that toxin binding electrostatically affects S4 movement as a result of the positively charged amino acids in ωCTX (2 lysine and 2 arginine residues). ωCTX binding may screen negative surface charges so that stronger depolarizations are required to induce voltage sensor (S4) movement, and the positively charged toxin residues may be positioned to electrostatically destabilize S4 when it is in the outward (activated) confirmation. Regardless of the specific mechanism, the major implication is that toxin binding destabilizes one or more activated S4 segments to inhibit N-channel gating.

Implications for the treatment of neuropathic pain

ωCTX has been shown to be an efficacious pain blocker in several animal models (Vanegas and Schaible 2000), and its closely related cousin, ω-conotoxin MVIIA, (also called SNX-111, ziconotide, and Prialt) was recently approved by the Federal Drug Administration [FDA application (NDA) 021060, original approval December, 28 2004] for the treatment of chronic pain in humans (Elmslie 2004; Snutch 2005; Wermeling and Berger 2006). However, serious side effects are associated with the use of this drug in humans, including dizziness, confusion, and death (Elmslie 2004; McGivern 2006; Penn and Paice 2000; Snutch 2005; Wermeling and Berger 2006). One reason for these side effects is that the toxin block of N-channels is all or none. Toxin binding to the channel completely blocks Ca2+ influx and the subsequent physiological effects. Perhaps the prevalence of these side effects could be reduced if drugs could be developed that partially blocked N-current. Our findings demonstrate that the ωCTX binding site may provide such an effect if drugs could be developed that bound to the site without plugging the pore. Based on our simulations, such a drug would maximally block only 50% of the Ca2+ influx during an action potential, which could reduce, instead of prevent, neurotransmitter release. We cannot currently predict what form such a drug would take. If the toxin electrostatically affects gating, the relevant drug could be a smaller version of ωCTX that carries a strong positive charge and binds to the site, but is sufficiently small to not interfere with the conductance pathway. On the other hand, if the toxin allosterically affects gating, it may be possible to generate small organic compounds that bind to the toxin site to mimic the gating effects without blocking N-channel conductance. There are hundreds of such compounds that have been screened for the ability to displace ωCTX binding (Zhang et al. 2006). The most promising of these compounds should be examined to determine if they modulate N-channel gating as predicted by our results.


This work was supported by a grant from the Pennsylvania Department of Health using Tobacco Settlement Funds (KSE). The PA Department of Health specifically disclaims responsibility for analyses, interpretations, and conclusions presented here.


We thank Dr. Stephen W. Jones for helpful discussion on early version of this manuscript.


  • The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.


  1. Boland et al. 1994.
  2. Buraei et al. 2005.
  3. Catterall et al. 2007.
  4. Dolphin 2004.
  5. Ellinor et al. 1994.
  6. Elmslie 2004.
  7. Elmslie et al. 1994.
  8. Feng et al. 2001.
  9. Hodgkin and Huxley 1952.
  10. Jones et al. 1999.
  11. Jones et al. 1997.
  12. Jones et al. 1998.
  13. Kim et al. 1995.
  14. Lee and Elmslie 1999.
  15. Lew et al. 1997.
  16. Liang and Elmslie 2001.
  17. Liang and Elmslie 2002.
  18. McDonough 2007.
  19. McDonough et al. 2002.
  20. McGivern 2006.
  21. Olivera et al. 1994.
  22. Penn and Paice 2000.
  23. Snutch 2005.
  24. Stocker et al. 1997.
  25. Vanegas and Schaible 2000.
  26. Wermeling and Berger 2006.
  27. Yarotskyy and Elmslie 2007.
  28. Zhang et al. 2006.
View Abstract