Synergetic Action of Domain II and IV Underlies Persistent Current Generation in Nav1.3 as revealed by a tarantula toxin

The persistent current (INaP) through voltage-gated sodium channels enhances neuronal excitability by causing prolonged depolarization of membranes. Nav1.3 intrinsically generates a small INaP, although the mechanism underlying its generation remains unclear. In this study, the involvement of the four domains of Nav1.3 in INaP generation was investigated using the tarantula toxin α-hexatoxin-MrVII (RTX-VII). RTX-VII activated Nav1.3 and induced a large INaP. A pre-activated state binding model was proposed to explain the kinetics of toxin-channel interaction. Of the four domains of Nav1.3, both domain II and IV might play important roles in the toxin-induced INaP. Domain IV constructed the binding site for RTX-VII, while domain II might not participate in interacting with RTX-VII but could determine the efficacy of RTX-VII. Our results based on the use of RTX-VII as a probe suggest that domain II and IV cooperatively contribute to the generation of INaP in Nav1.3.

V oltage-gated sodium channels (Na v s) are essential for the initiation and propagation of action potentials in excitable tissues such as nerves and muscles [1][2][3] . They consist of a pore forming a-subunit (260 kDa) associated with auxiliary b-subunits (30)(31)(32)(33)(34)(35)(36)(37)(38)(39)(40) 4 . In mammals, nine a-subtypes (Na v 1.1-1.9) have been identified and cloned 5,6 . The a-subunit is composed of four homologous domains (DI-DIV) and each domain contains six transmembrane segments (S1-S6). The S5-S6 segments form a central pore for Na 1 , with the S1-S4 segments from each domain forming the surrounding voltage sensors 2,4 . Each of the four voltage sensors is activated in response to depolarization of membrane potential 7 ; the voltage sensors of the first three domains (DI-III) are responsible for channel activation, while that of the fourth domain (DIV) determined fast inactivation 8,9 . Na v s fail to inactivate in some cases, resulting in the generation of non-inactivating persistent Na 1 currents (I NaP ), which account for up to 5% of the transient peak inward current (I NaT ) in physiological conditions 10,11 . Despite its small amplitude compared with I NaT , I NaP amplifies synaptic potentials and aids in the repetitive firing of action potentials (AP) in neurons because it is activated by sub-threshold voltages 11 . The voltage range in which I NaP is activated is traversed by interspike intervals during an AP train 11,12 . I NaP has been characterized in a variety of neurons 13 . Mutations in some Na v s genes causing enhanced I NaP amplitude are correlated to diseases, such as heart vascular disease and epilepsy 11,14,15 . Therefore, drugs targeting I NaP are expected to have therapeutic benefits [16][17][18] .
The early understanding of the origin of I NaP is based on three hypotheses: (1) the window current hypothesis 19 ; (2) I NaP is generated by unusual subtypes of Na v s which lacking inactivation 20,21 ; (3) I NaP is produced by the same population of channels responsible for I NaT through a distinct gating mechanism 22,23 . Many lines of evidences support the third hypothesis, namely that some channels of a specific Na v subtype occasionally enter the late brief opening and burst of opening state in single channel recordings [24][25][26] . The single channel mechanism for the slowed inactivation of a Na v by site 3 toxins TsIV-5 and anthopleurin B was also attributed to the increment of mean open time as well as prolonged bursting 27,28 ; studies showed that mutations in all four domains of Na v s as well as intracellular loops affect the amplitude of the I NaP 29 ; wild type (wt)-Na v 1.3 naturally develops I NaP in neonatal and axotomized neurons 30,31 , and Na v 1.3 expressed in HEK293T cells displays a clearly detectable I NaP also confirm that I NaP is the intrinsic property of Na v s themselves, although the molecular determinants of I NaP in Na v s are largely unknown.
In the present study, we identified a tarantula toxin named ahexatoxin-MrVII (RTX-VII) that enhances the I NaP of Na v 1.3 and used it as a probe to examine the involvement of each domain of Na v 1.3 in the generation of I NaP . Our results reveal that domains II and IV work in a synergetic manner to determine the toxin-induced I NaP of Na v 1.3.

Results
RTX-VII enhances the I NaP of Na v 1.3. Macrothele raveni (Figure 1a, inset) venom was collected by using an electro-pulse stimulator as described previously 34 . The lyophilized crude venom was fractionated by RP-HPLC (Figure 1a). A comprehensive screening of each eluted fraction against Na v 1.3 transiently expressed in HEK293T cells indicated that the fraction with a retention time of 44.6 min inhibited the fast inactivation of this channel (Figure 1b). This fraction contained a peptide with a molecular weight of 4064.71 Da as determined by MALDI-TOF MS, which was then further purified by analytical RP-HPLC ( Supplementary Fig. S1a). Sequence of the peptide was determined by combining Edman degradation ( Supplementary Fig. S1j) and cDNA sequencing ( Supplementary Fig. S1b), and the toxin was named a-hexatoxin-MrVII (RTX-VII). Blasting the full amino acid sequence of RTX-VII showed that it share 92% identity to a previously known spider toxin Magi-6 ( Supplementary Fig. S1c). However, Magi-6 did not compete with the scorpion toxin LqhaIT in binding site 3 of Na v s, and the symptoms caused by injection of pure Magi-6 to mice could not be directly linked to a particular ion channel receptor 35 . This raise the possibility that the subtle amino acid sequence variation brought by RTX-VII makes it active on mammalian Na v s, as that of APETx3 and APETx1, in which a single amino acid substitution between them confer these two toxins different ion channel selectivity 36 . RTX-VII contains eight cysteine residues forming four disulfide bonds, as the measured molecular weight was 8 Da less than the theoretical one. The conserved arrangement of the cysteine residues in RTX-VII indicated that it contains a cystine knot (ICK) motif ( Supplementary Fig. S1c).
As shown in Figure 1b, RTX-VII had three effects on current of Na v 1.3: (1) it increased the I NaT amplitude at the depolarizing voltage of 210 mV; (2) it inhibited the fast inactivation of the channel as determined by I 5ms /I NaT ratio; (3) and it induced a large I NaP as revealed by the I 45ms /I NaT ratio. At a depolarization of 210 mV, the I NaP generated by Na v 1.3 accounted for little of the I NaT under control conditions, whereas the treatment with 0.1 mM RTX-VII enhanced the I NaP to approximate 25% of the I NaT . The I NaP evoked by the toxin lasted for several seconds and large tail current was where red arrow indicates the fraction containing RTX-VII. (b) Current traces from a representative cell show 0.1 mM RTX-VII enhances the I NaT of Na v 1.3 and induces a large I NaP at the end of a 50-ms depolarization to 210 mV from a holding potential of 2100 mV (n 5 6). Note I NaP was measured at the time point of 45 ms. (c) Time course for the enhancement of I NaP of Na v 1.3 by 0.2 mM RTX-VII and the recovery upon washing with bath solution, t on and t off is 40.9 6 11.3 s and 162.8 6 39.7 s, respectively (n 5 4). (d) RTX-VII dose dependently enhances I NaP of Na v 1.3 with an apparent EC 50 of 0.12 mM (n 5 6). The maximum response (f max ) and the minimum response (f min ) of Na v 1.3 to RTX-VII is 66.90% and 0, respectively.  S1d), distinguishing this toxin from certain a-scorpion toxins. The time course for 0.2 mM RTX-VII activating the I NaP of Na v 1.3 was characterized by a slow onset of action (t on 5 40.9 6 11.3 s) and a slow recovery upon washing (t off 5 162.8 6 39.7 s) ( Figure 1c). The activation of the I NaP of Na v 1.3 by RTV-VII was dose-dependent, with an apparent EC 50 of 120 nM (Figure 1d). The activity and selectivity of RTX-VII were examined against three Na v subtypes (Na v 1.4, 1.5 and 1.7) expressed in HEK293T cells, TTX-R Na v s of rat dorsal root ganglion neurons and Na v s in neonatal rat hippocampal neurons. Among these channels, Na v s of neonatal rat hippocampal neurons were sensitive to this toxin ( Supplementary Fig. S1i), whereas the others were not ( Supplementary Fig. S1e-h).
Kinetics of RTX-VII action on Na v 1.3. The current-voltage (I-V) relationships of the I NaT and I NaP of Na v 1.3 before and after the application of 0.2 mM RTX-VII were explored (Figure 2, Supplementary Fig. S2). Compared with the control, RTX-VII modified the I-V relationship of the I NaT as follows: (1) the activation of the I NaT was potentiated by the toxin at voltages ranging from 245 mV to 10 mV, while no potentiation was observed at voltages . 10 mV ( Figure 2a); (2) the activation voltage of the maximum I NaT was shifted from 5 mV in the control to 210 mV in the presence of the toxin (Figure 2a, Supplementary Fig. S2d). Although RTX-VII did not alter the reversal voltages (approximate 65 mV) of Na v 1.3, toxin application did negatively shift channels' initial activation voltage (Figure 2a and c). These data indicate that the toxin treatment may increase the opening probability of Na v 1.3 channels in cell membrane and facilitate their activation at weak depolarizing voltages. The I-V curves of the I NaP before and after the application of toxin indicate that the enhancement of the I NaP by the toxin occurred across the depolarizing voltages tested (Figure 2a, Supplementary Fig. S2a) and the activation voltage of the maximum I NaP was at about 210 mV. Regarding the I-V curves of the I NaP and I NaT in the presence of toxin, if the amplitudes of the I NaP and I NaT at each depolarizing voltage were normalized to their maximum one, respectively, they overlapped completely (Figure 2b), suggesting that the I NaP and I NaT in the presence of the toxin share rather similar activation voltage. Similar to some a-scorpion toxins 37 , RTX-VII removed the fast inactivation of Na v 1.3 in a voltage-independent way at depolarizing voltages ranging from 220 mV to 130 mV ( Supplementary Fig S2b).
The conductance-voltage (G-V) relationship and the steady-state inactivation of Na v 1.3 before and after the application of RTX-VII were explored (Figure 2d). Compared with the control, RTX-VII increased the conductance of the cell membrane at depolarizing voltages below 10 mV as revealed by an approximate 16 mV negative shift of Na v 1.3 channels' activation curve induced by the toxin (V a 5 212.72 6 4.92 mV for control and V a 5 229.64 6 6.21 mV for the toxin treatment), which is in accordance with the negative shift of the activation voltage for maximum I NaT observed in the I-V curve; the toxin did not significantly alter the slope factor of the activation curve (K a 5 6.83 6 0.98 mV for control and K a 5 5.66 6 1.12 mV for the toxin treatment). The I-V and G-V relationships  of Na v 1.3 before and after RTX-VII application were acquired with stringent controls of the uncompensated series resistance (Rs) caused depolarizing voltage error (the maximum tolerable voltage error was less than 5 mV, the mean maximum Rs-caused voltage error was 2.34 6 1.08 mV, p , 0.001 when compared to the Va shifted amplitude). A steady-state component (approximate 20% of the I NaT ) that was resistant to inactivation was observed in the steady-state inactivation (SSI) curve when conditional voltages were above 220 mV, which should represent the I NaP elicited by conditional pulses. A significant change of V h and K h were observed (V h 5 244.58 6 2.73 mV for control and V h 5 251.12 6 5.00 mV for the toxin treatment, p , 0.05; K h 5 26.98 6 1.44 mV for control and K h 5 212.49 6 0.41 mV for the toxin treatment, p , 0.001). The hyperpolarization shift of the G-V curve and a non-inactivated component in the SSI curve together resulted in an enlarged voltage range for generation of window current, indicating a slower development of closed state inactivation (CSI) in the toxin-treated channels.
The effect of RTX-VII on the repriming kinetics (recovery from fast inactivation) of Na v 1.3 was also investigated. As shown in Figure 2e, the I NaT of Na v 1.3 recovered gradually from fast inactivation with the repolarizing time (recovery time) increasing in the absence (control) and presence (toxin) of RTX-VII. The I NaP induced by the toxin was observed at all recovery time. The I NaP of the toxintreated channels fully recovered at the recovery time of 0 ms, but no I NaT recovery was observed (Figure 2e, toxin). The recovery ratios of Na v 1.3 I NaT before and after the application of toxin were plotted as a function of recovery time (Figure 2f), showing most of channels (.80%) recovered from fast inactivation in 4 ms in both conditions. An apparently faster repriming of the toxin-treated channels than that of control channels within 4 ms was observed, which could be associated with the existence of the I NaP . If the I NaP was subtracted from the I NaT in toxin treated channels, the residual current would exhibit the same repriming kinetics as that of the control ( Supplementary Fig. S2c).
The molecular mechanism of RTX-VII as an excitatory toxin. The enhancement of I NaP of Na v s in hippocampal neurons by RTX-VII may have led to excitatory toxic in mouse. Intracerebroventricular injection of 20 ng RTX-VII dissolved in 20 ml saline caused seizurelike symptoms, as described by circular running in the first several minutes followed by involuntary body twitching, while animals in control group injected with 20 ml saline behaves normal (n 5 5 in each group, Supplementary video). We therefore investigated the mechanism of RTX-VII as excitatory toxin. Na v 1.3 is upregulated in the peripheral nervous system in response to nerve injury, and contributes to the hyperexcitability of nociceptive neurons under neuropathic conditions 38,39 . The fast repriming kinetics and slow development of CSI of Na v 1.3 make it suitable for generating a large response to slowly developing depolarizing inputs (ramp stimuli) 40 . We first tested the effect of RTX-VII on the ramp current (I ramp ) of Na v 1.3 evoked by various ramp stimuli (Figure 3a p1). Consistent with previous studies 33 , Na v 1.3 expressed in HEK293T cells produced a large inward Na 1 current in response to a linearly increasing voltage ramp from 2100 mV to 20 mV at the ramp rate of 1.2 mV/ms; of the two I ramp peaks shown, the first one (I ramp1 ) but not the second one (I ramp2 ) was ramp rate-dependent, with higher rate leading to larger I ramp1 (Figure 3b, black trace). The application of 0.5 mM RTX-VII increased the amplitude of both I ramp1 and I ramp2 generated by Na v 1.3 at all ramp rates tested along with a hyperpolarization shift of the initial activation voltage for I ramp1 (Figure 3b, red trace; the maximum tolerable voltage error was less than 5 mV, the mean maximum Rs-caused depolarizing voltage error was 3.45 6 1.26 mV).The negative shift of I ramp1 of Na v 1.3 was consistent with the channels' negatively shifted activation. The enhanced activation of I ramp1 of Na v 1.3 may have been derived from the larger potential gradient (caused by the negative shift of activation voltage of I ramp1 ) that drives Na 1 to cross the membrane as well as a slowed CSI which makes more channels available for activation.
Na v 1.3 intrinsically produces small I NaP , and the relationship between I NaP and I ramp2 was investigated in a previous study in which a close correlation between them was observed 33 . To clarify the relationship between the RTX-VII evoked I NaP and I ramp2 , the protocol p2 described in Figure 3a was used to elicit two type currents of Na v 1.3 ( Figure 3c). Note only I ramp2 could be evoked at the ramp rate of 0.2 mV/ms (Figure 3b and Figure 3c). RTX-VII dose-dependently enhanced I NaP (I 45ms ) and I ramp2 of Na v 1.3 ( Figure 3c).The apparent EC 50 for RTX-VII activating I ramp2 was 320 nM, as revealed by plotting I ramp2 /I NaT ratios as a function of toxin concentrations ( Figure 3d). This EC 50 value did not differ much from that of RTX-VII activating the I NaP of Na v 1.3 (120 nM). Furthermore, the correlation coefficient between I NaP and I ramp2 was 0.9947 (Figure 3e), indicating a close correlation between them. Thus, data derived from the toxin study further confirmed the conclusion described above.
The effect of RTX-VII on the ramp current of Na v s in neonatal hippocampal neurons was also examined. As shown in Figure 3f, both I ramp1 and I ramp2 of hippocampal Na v s were evoked by a linearly increasing voltage ramp from 2100 mV to 20 mV at the ramp rate of 1.2 mV/ms (Figure 3a, p3); both components of the I ramp of hippocampal Na v s were greatly enhanced by toxin (Figure 3f, upper). As shown in Figure 3f (below), both I ramp1 and I ramp2 of hippocampal Na v s in control conditions displayed voltage-dependent inactivation by reverse ramp (R-ramp) stimulation following forward ramp (Framp) stimulation (Figure 3a, p4); as I ramp1 disappeared, the amplitude of I ramp2 decreased in the R-ramp compared with that in the F-ramp. On the contrary, 1 mM RTX-VII treatment removed the voltage-dependent inactivation of I ramp2 but not I ramp1 , as toxin induced a nearly unchanged I ramp2 in both the forward and reverse ramps, while I ramp1 was absent in the R-ramp. This finding indicates that the amplitude of I ramp2 in the presence of the toxin is only dependent on the transmembrane potential, and this population of Na v s generating I ramp2 should maintain a continuous open state during the entire time course of AP. The toxin negatively shifted I ramp1 and the enhanced activation of I ramp2 might lower the threshold and increase the frequency of AP in hippocampal neurons, respectively, which possibly triggers the spontaneous AP firing in hippocampal neurons at a physiological resting potential. Currentclamp experiments showed that 2 mM RTX-VII triggered spontaneous high frequency AP firing in hippocampal neurons ( Domains II and IV of Na v 1.3 are critical for I NaP generation by RTX-VII. Because Na v 1.5 is resistant to RTX-VII (Figure 4b), a chimera strategy was used to screen the critical modules [voltage sensor domains (VSD) or pore domains (PD)] responsible for the toxin-induced I NaP of Na v 1.3. Each module from the four domains of Na v 1.3 was substituted with the corresponding Na v 1.5 module ( Supplementary Fig. S4). The nomenclature of a specific chimeric channel was defined as follows: for example, Na v 1.3/1.5 DI-VSD chimera is a hybrid channel in which the DI-VSD of Na v 1.3 was replaced with that of Na v 1.5. Eight Na v 1.3 derived chimeric channels were constructed. All chimeric channels except the Na v 1.3/1.5 DII-VSD chimera were functionally expressed in HEK293T cells; therefore, the hybrid channel Na v 1.3/1.5 DII was generated instead of the Na v 1.3/1.5 DII-VSD chimera ( Supplementary Fig. S4). To assess the potency and efficacy of RTX-VII for I NaP generation in each wt-or chimeric channel, a 300-ms depolarization to 10 mV from a holding potential of 2100 mV was applied to evoke the I NaT and I NaP of a specific channel in the absence and presence of various concentrations of   (Figure 3a, p1), the ramp time ranges from 100 ms to 600 ms, 100 ms/step. The Na v 1.3 ramp current (I ramp ) displays two peaks with the first one (I ramp1 ) but not the second one (I ramp2 ) being sensitive to ramp rate in control (black traces). 0.5 mM RTX-VII enhances the amplitudes of both peaks and causes a hyperpolarized shift of the initial activation voltage for I ramp1 (red traces); numbers labeled above the traces indicate the ramp rate (mV/ms) (n 5 10). (c) Representative traces show that RTX-VII dose-dependently enhances the I NaP (I 45ms ) and I ramp2 of Na v 1.3 elicited by the protocol p2 shown in Figure 3a (n 5 5). (d) Dose-response curve for RTX-VII activating the I ramp2 of Na v 1.3, the apparent EC 50 is determined as approximate 0.3 mM; the maximum and the minimum response of Na v 1.3 to RTX-II is 63.03% and 2.15%, respectively (n 5 5). (e) The I 45ms /I NaT ratio was plotted as a function of the I ramp2 /I NaT ratio at each toxin concentration (data from Figure 3c). A linear fit of the dots shows the close correlation between the I NaP and I ramp2 of Na v 1.3 (R 2 5 0.9947) (n 5 5). (f) Compared with control, 1 mM RTX-VII evidently enhances both peaks (I ramp1 and I ramp2 ) of the ramp current of Na v s in rat hippocampal neurons (upper). Protocol p3 shown in Figure 3a was used (n 5 5); Representative traces (below) show that the I ramp of Na v s in hippocampal neurons is elicited by protocol p4 shown in Figure 3a   toxin, and the I NaP was measured at the time point of 295 ms (Figure 4a) because the currents of chimeric channels reached a macroscopic steady state at the time point of 300 ms. To compare data derived from different channels, the relative values of I NaP /I NaT (both from the same current trace) after treatment with different concentrations of toxin were calculated, and the potency of RTX-VII on a specific channel was defined as the EC 50 value, while the efficacy of RTX-VII was determined by steady-state I NaP /I NaT ratio at the saturated concentration of toxin.
The substitution of the VSD/PD of Na v 1.3 with that of Na v 1.5 had different effects on the potency and efficacy of RTX-VII. Compared with the wt-channel, five chimeric channels, namely Na v 1.3/1.5 DI-VSD, DI-PD, DII-PD, DIII-VSD and DIII-PD produced a large I NaP in response to RTX-VII, whereas the other three chimeric channels, Na v 1.3/1.5 DII, DIV-VSD and DIV-PD displayed a smaller I NaP (Figure 4a). The apparent EC 50 values of RTX-VII on these channels were further determined from dose-response curves (Figure 4c and d). The bar diagrams shown in Figure 4e and f indicate the changes in the potency and efficacy of RTX-VII on Na v 1.3 derived chimeric channels, respectively, which could be described as follows: (1) in the chimeras Na v 1.3/1.5 DI-VSD, DII-PD, DIII-VSD and DIII-PD, no significant changes in the potency and efficacy of RTX-VII were observed, as indicated by the negligible reduction in the steady-state I NaP /I NaT ratios and the less than two-fold increments of EC 50 values; (2) the chimera Na v 1.3/1.5 DI-PD generated a large steady-state I NaP , similar to that of the wt-Na v 1.3; however, an approximate 4.8 fold increase of the EC 50 value was observed, indicating that this chimeric channel reduced the binding affinity of RTX-VII but not the efficacy; (3) the chimeras Na v 1.3/1.5 DII and DIV-PD significantly decreased the efficacy of RTX-VII, as revealed by a significantly smaller steadystate I NaP /I NaT ratios (steady-state I NaP only accounts for approximate 24% of I NaT after 2 mM RTX-VII treatment, P , 0.001 when compared to wt-Na v 1.3); however, these substitution resulted in a , 2-fold increase of the EC 50 values; (4) In the chimera Na v 1.3/1.5 DIV-VSD, both the potency and efficacy of RTX-VII was significantly attenuated, because the toxin, even at a concentration of 10 mM, induced a small fraction of steady state I NaP (,10% of the I NaT , P , 0.001, when compared to wt-Na v 1.3) in this chimeric channel and the apparent EC 50 of toxin on this chimeric channel was increased by approximate 25 folds when compared to wt-Na v 1.3. Taken together, these findings suggest that DIV-VSD, DIV-PD, DII and DI-PD of Na v 1.3 play important roles in the RTX-VII-induced I NaP . The different influence of these Na v 1.3 module substitutions on the potency and efficacy of RTX-VII suggest that they play different roles. DIV-VSD and DI-PD might jointly compose the binding receptor for RTX-VII, whereas the loss of efficacy of RTX-VII on the chimeric channel Na v 1.3/1.5 DII was not caused by loss of toxin binding but was rather associated with an intrinsic limitation of this hybrid channel in generating a larger I NaP .
Domain II of Na v 1.3 is not involved in interacting with RTX-VII. Further experiments were performed to clarify the roles of Na v 1.3 DII and DIV. First, we examined whether RTX-VII binds to Na v 1.3 DII. Neurotoxins acting on DII of Na v s often cause a negative or positive shift of the activation kinetics of targeted channels 41 . The substitution of the DII of Na v 1.3 with that of Na v 1.5 should affect the RTX-VII-induced negative shift of activation kinetics of Na v 1.3 if the toxin binds to Na v 1.3 DII, as RTX-VII did not affect the I-V curve of Na v 1.5 (Supplementary Fig. S5a). Therefore, the activation kinetics of the chimeric channel Na v 1.3/1.5 DII was investigated before and after the application of 2 mM RTX-VII. RTX-VII negatively shifted the voltage-dependent activation of the Na v 1.3/ 1.5 DII chimera and increased the I NaT at voltages ranging from 250 mV to 5 mV (Figure 5a). In addition, 2 mM RTX-VIII caused an approximate 14 mV negative shift of the G-V curve of the Na v 1.3/ 1.5 DII chimera without changing the slope factor (V a 5 218.00 6 1.71 mV for control and V a 5 232.09 6 1.99 mV for the toxin treatment; K a 5 7.23 6 1.22 mV for control and K a 5 6.82 6 1.30 mV for the toxin treatment; the maximum tolerable voltage error was less than 5 mV, the mean maximum Rs-caused depolarizing voltage error was 2.92 6 2.10 mV, p , 0.001 when compared to the Va shifted amplitude) (Figure 5b).This raises the possibility that the toxin might not interact with Na v 1.3 DII, which was further confirmed by using a competitive assay. HNTX-III is a tarantula toxin that inhibits the I NaT of Na v 1.3 and Na v 1.7. It was found that this toxin targeted DII S3-S4 linker of Nav1.7 42 . The wt-Nav1.5 channel and the Na v 1.3/1.5 DII chimera were resistant to 1 mM HNTX-III treatment, whereas the Na v 1.3/1.5 DII-PD chimera was inhibited by 1 mM HNTX-III ( Supplementary Fig. S5  b-d). Reconstruction of the DII of Nav1.3 to Nav1.5 (Na v 1.5/1.3 DII chimera) conferred the inhibitory activity of HNTX-III to this channel ( Supplementary Fig. S5 e).These evidences indicate HNTX-III inhibit Nav1.3 by binding to its DII-VSD. If RTX-VII also targeted DII-VSD of Na v 1.3, its binding should prevent the interaction of HNTX-III with Na v 1.3 because of steric hindrance, which would result in an attenuation of the inhibitory potency of HNTX-III on Na v 1.3. As shown in Figure 5c, the inhibitory effects of HNTX-III on Na v 1.3 I NaT did not differ between 0.5 mM RTX-VIIIpretreated and -untreated channels. The dose-response curves were also superimposed well (Figure 5d), providing evidence to rule out the binding of RTX-VII to Na v 1.3 DII. Next, we determined the molecular determinant in DIV of Na v 1.3 for RTX-VII binding. Since Na v 1.5 is resistant to RTX-VII, the residues in S1-S2 and S3-S4 extracellular loops of Na v 1.3 were mutated to the corresponding residues of Na v 1.5, respectively (Figure 5e). A total of seven residues were mutated and six of them were functionally expressed except V1566F. The kinetics for the activation and SSI of all mutants were listed in supplementary Table S1. Compared with wt-Na v 1.3, Four mutant channels (K1503P, M1505K, T1506I, L1507N) carrying mutations in the S1-S2 linker led to a 4-12 folds increase of apparent EC 50 values, whereas the E1562Q and E1562R mutation in the S3-S4 linker resulted in an approximate 5 folds and 20 folds increase of the apparent EC 50 values (Figure 5f and 5g). These data indicate that multiple residues located in Na v 1.3 DIV were involved in interacting with RTX-VII, and that E1562 was the most important residue for the interaction.
Reverse reconstruction of Na v 1.3 DII and DIV into Na v 1.5 fully restores toxin efficacy. Considering the critical role of the DII and DIV of Na v 1.3 in the RTX-VII-induced I NaP , we assumed that reverse reconstruction of Na v 1.3 DII and DIV into Na v 1.5 might restore the efficacy of the toxin. A reversal chimeric strategy was used as follows: four domains of Na v 1.3 were stepwise reconstructed into the scaffold of Na v 1.5 ( Supplementary Fig. S6). The nomenclature of a chimeric channel was defined as follows: for example, Na v 1.5/1.3 DI was a chimeric channel in which the DI of Na v 1.5 was substituted with that of Nav1.3. A total of 11 chimeric channels were constructed and their I NaP generation by the toxin was compared. Again, I NaP was measured at the time point of 295 ms (Figure 6a). The substitution of all four domains of Na v 1.5 with those of Na v 1.3 (Nav1.5/1.3 DI-II-III-IV) almost fully restored the efficacy of RTX-VII, thus eliminating the involvements of the intracellular loops of Na v 1.3 in the toxin-induced I NaP . Of the four single domain replaced chimeric channels, Na v 1.5/1.3 DI, Na v 1.5/1.3 DII and Na v 1.5/1.3 DIII chimeras were resistant to RTX-VII, similar to wt-Na v 1.5, whereas Na v 1.5/1.3 DIV chimera was sensitive to RTX-VII. Furthermore, the toxin slowed the inactivation and induced a small steady-state I NaP in this chimeric channel, indicating that Na v 1.3 DIV is important but not sufficient for RTX-VII inducing large I NaP . Of the two triple domain replaced chimeric channels, Na v 1.5/1.3 DI-III-IV chimera did not fully restore toxin efficacy but Na v 1.5/1.3 DI-II-IV chimera did, which indicates that the DII but not the DI and DIII of Na v 1.3 is required for toxin inducing large I NaP . Of the three double domain replaced chimeric channels, the reconstruction of the DI or DIII of Na v 1.3 into the scaffold of Na v 1.5/1.3 DIV chimera (Na v 1.5/1.3 DIII-IV chimera or Na v 1.5/1.3 DI-IV chimera) had a limited effect on restoring toxin efficacy, whereas the reconstruction of the DII of Na v 1.3 into Na v 1.5/1.3 DIV chimera (Nav1.5/1.3 DII-IV chimera) almost fully rescued toxin efficacy, suggesting the assembly of the DII and DIV of Na v 1.3 should be sufficient for RTX-VII inducing large I NaP . Additionally, the chimeric channel Na v 1.5/1.3 DI-III-IV&DII PD, where only the DII-PD but not the whole DII of Na v 1.3 was present, also attenuated the efficacy of RTX-VII compared with that of Na v 1.5/1.3 DI-II-III-IV chimera, which strongly supports that the DII-VSD of Na v 1.3 plays a vital role in toxin-induced I NaP generation.
The apparent EC 50 values of RTX-VII on the Na v 1.5 derived chimeric channels containing Na v 1.3 DIV were estimated from the dose-response curves (Figures 6b and c), and the changes in the potency and efficacy of RTX-VII on these chimeric channels were showed in Figures 6d and e, respectively. Of the Na v 1.3 DIV-containing chimeric channels, the Na v 1.3 DII-containing ones (Na v 1.5/1.3 DI-II-III-IV, DI-II-IV and DII-IV chimeras), but not those without reconstruction of Na v 1.3 DII or DII-VSD (Na v 1.5/1.3 DIV, DI-III-IV, DI-IV and DI-III-IV&II-PD chimeras), produced a large steady-state I NaP comparable to that of wt-Na v 1.3 in the presence of saturated concentration toxin (Figure 6e). On the other hand, the toxin potency on these Na v 1.3 DIV-containing chimeric channels were only slightly weaker than that of wt-Na v 1.3, although the greatest fold change of EC 50 was observed in Na v 1.5/1.3 DII-DIV chimera (8 folds) (Figure 6d). Moreover, the incorporation of Na v 1.3 DI into Na v 1.5/1.3 DII-DIV chimera (Na v 1.5/1.3 DI-DII-DIV chimera) led to an evident enhancement of toxin potency, which is comparable to that of wt-Na v 1.3 channel. The results are consistent with the interpretation that the DIV of Na v 1.3 was the main toxin binding site, while the DI-PD of Na v 1.3 might construct the low affinity binding site for RTX-VII. Overall, combining data in Figures 4 and 6 confirmed the cooperative involvement of DII and DIV in the toxininduced I NaP of Na v 1.3.

Discussion
Neurotoxins produced by venomous animals, plants, and microorganisms are a valuable pool of molecular probes to investigate the structure-function relationship of Na v s 43 . RTX-VII robustly enhances the I NaP of Na v 1.3 and discriminates Na v subtypes Na v 1.4, Na v 1.5, and Na v 1.7-1.9 from Na v 1.3. The toxin-induced and the intrinsic I NaP share some common features, such as subthreshold activation, a close correlation with I ramp2 , and triggering spontaneous high frequency AP firing. Furthermore, the brief late www.nature.com/scientificreports  opening and burst of openings of Na v s may be the common mechanism underlying the origin of both types of I NaP . However, the intrinsic I NaP of Na v s is small, which hampered the investigation of the mechanism underlying I NaP generation. RTX-VII dramatically enhancing the I NaP of Na v 1.3 enabled detailed investigations of Na v 1.3I NaP generation. In the present study, we clarified the roles of the four domains of Na v 1.3 in I NaP generation by using RTX-VII as a molecular probe.
Along with the enhancement of I NaP , RTX-VII also facilitates Na v 1.3 channel opening at weak depolarizations as revealed by the toxin potentiating I NaT of Na v 1.3 when depolarizing voltages are below 10 mV as well as the toxin negatively shifting channel's steady-state activation. This observation is not without precedent, as some a-scorpion toxins modulate Na v s in a similar way 37 . This phenomenon could be reasonably interpreted as an increase of the maximum opening probability of the toxin-treated channels. However, how the toxin-bound channels open with a greater probability at weak depolarizations remains unclear. Our data indicate that RTX-VII binds to the DIV-VSD instead of the DII-VSD of Na v 1.3, which suggests that the potentiation of Na v 1.3 activation by the toxin might not derive from the toxin facilitating DII activation. Two possible explanations for RTX-VII enhancing the I NaT of Na v 1.3 are proposed: (1) toxin treatment altered single channel conductance 44 of Na v 1.3. This interpretation seems unreasonable because the toxin does not alter the inward and outward Na 1 current of Na v 1.3 evoked by strong depolarizing voltages above 10 mV; (2) RTX-VII tends to stabilize the DIV-VSD of Na v 1.3 in a partially activated state (pre-activated state), which is required for channel activation but not sufficient to trigger channel inactivation (the fully activated DIV-VSD is required for the fast inactivation of Na v s) 45 , thus promoting channel activation in a similar but not identical way with that of b-scorpion toxins 46,47 . The vital difference is that bscorpion toxins trap the DII-VSD but not DIV-VSD of Na v s in the activated state. The latter interpretation seems plausible as emerging evidences support DIV is involved in Na v activation 45,48 . Furthermore, the voltage driving the outward movement of DIV-VSD is the later step in the activation sequence of Na v s 49 . RTX-VII did not alter the I NaT of Na v 1.3 at strong depolarizing voltages (.10 mV), and this phenomenon could be interpreted by the fact that both toxin-free and toxin-bound channels in cell membrane are almost fully activated at 10 mV, which is in consistent with the G-V relationship observed in figure 2d. Taken in all, considering RTX-VII promoting activation and inhibiting inactivation of Na v 1.3 as well as the unique role of DIV-VSD in channel gating, we would like to suggest that RTX-VII might tend to trap and stabilize the DIV-VSD of Na v 1.3 in the pre-activated state during channel activation.
The reconstruction of Na v 1.3 DII but not DI or DIII to Na v 1.5/1.3 DIV chimera fully restored toxin efficacy, but it is interesting that RTX-VII did not bind to Na v 1.3 DII. Therefore, the role of this domain in the toxin-induced I NaP remains unclear. Previous studies Na v 1.5 derived chimeric channels in the absence and presence of various concentration of RTX-VII (n 5 6-9). The chimeric channels were constructed as follows: one or several domains (DI, DII, DIII or DIV) of Na v 1.5 were substituted with the corresponding domain/s of Na v 1.3 (see Supplementary Fig. S6). (b) Dose-response curves for RTX-VII enhancing the I NaP of wt-Na v 1.3 and Na v 1.5 derived chimeric channels that did not or slightly restored toxin efficacy (steady-state I NaP /I NaT ratio at the saturated concentration of toxin) (n 5 6-9). (c) Dose-response curves for RTX-VII enhancing I NaP of wt-Na v 1.3 and Na v 1.5 derived chimeric channels that almost completely restored toxin potency and/or efficacy (n 5 6-9). (d) Bars show the fold changes of the apparent EC 50 of RTX-VII for each Na v 1.5 derived chimeric channels compared with that for wt-Na v 1. showed that the inter-domain interactions of Na v s is necessary for channel gating 50,51 . We proposed in this study that the DII and DIV of Na v 1.3 might cooperate to trigger late brief opening and burst of opening to generate I NaP , and RTX-VII should facilitate/amplify this cooperation to induce large I NaP in Na v 1.3. The subtle amino acid sequence differences of the domain II between Na v 1.3 and Na v 1.5 greatly affect this cooperation, namely the DII of Na v 1.3 can cooperate well with its own DIV, which is not the case for the DII of Na v 1.5 with DIV of Na v 1.3. The roles of Na v 1.3 DI and DIII of in the toxin-induced I NaP generation were unclear. The fact that the replacement of the DI or DIII of Na v 1.3 with that of Na v 1.5 did not affect toxin efficacy could not exclude the possibility that both domains might involve in the I NaP generation, because high sequence similarity of DI and DIII between Na v 1.3 and Na v 1.5 is observed and probably the inter-domain interactions might not be interfered although these two domains were replaced.
RTX-VII induced large I NaP in Na v 1.3 at the end of a 50-ms or a 300-ms depolarization, which differs from some scorpion toxins and sea anemone toxins that slow the inactivation of Na v s but the resultant current decay rapidly in 50 ms (Lqh2 as a representative 52 ). What is the difference derived from? Theoretically, Lqh2 trapping the DIV-VSD of Na v s in the closed state should have induced large I NaP , but the fact is not. How are the toxin-bound channels inactivated? Slow inactivation may not be the underling mechanism. This is because that slow inactivation is rarely observed in a 50 ms depolarization (such a short depolarization is not sufficient to trigger this gating process). The repriming kinetics of the toxin-bound channels is the same as or even faster than that of the toxin-free channels 37 , which is also inconsistent with the fact that the recovery of Na v s from slow inactivation is slow 53 . Based on the unique role of DIV in fast inactivation, a model was proposed to clarify these two problems. Macroscopically, in this model, a depolarization would drive and maintain the first three domains of Na v s in an activated state; Lqh2 and RTX-VII could trap the DIV-VSD of Na v s in the closed 52 and partially activated state, respectively. For Lqh2, such trapping is not very stable, as the depolarization prolongs, the toxin-bound DIV-VSD would be gradually activated, triggering channel inactivation. However, for RTX-VII, the DII of Na v 1.3 might allosterically slow/inhibit this process, which therefore makes RTX-VII stably trap the DIV-VSD of Na v 1.3 in the partially activated state and then the channels would maintain a persistent opening state. The understanding of this process in the single channel level could be as follows: the inactivation ball of a Na v has a ''on state'' (blocking the pore) and an ''off state'' (free in cytosol) which are tightly coupled to the activated and resting state of DIV-VSD, respectively 8 . Normally, DIV-VSD is immobilized in an outward conformation by activation 54,55 . The toxin-bound DIV-VSD could be activated by strong depolarization but not be stably immobilized as toxins tend to ''drag'' the DIV-VSD to its resting state (partially activated state for RTX-VII). Thus, when the toxin-bound DIV-VSD is activated, the inactivation ball is in the ''on state'' and the pore is occluded; when the toxin-bound DIV-VSD is in the resting state (partially activated state for RTX-VII), the channel just opens. The inactivation ball switches between the ''on state'' and the ''off state'' quickly and such inactivation ball movement should trigger the burst opening of the channel in single channel recording. For Lqh2, as the depolarization prolonged, the DIV-VSD of most channels would be stably immobilized and these channels were consequently trapped stably in the inactivated state. On the other hand, for RTX-VII, toxin binding to the DIV-VSD of Na v 1.3 should allosterically affect the conformation of DII-VSD, which would in turn interfere with the time-dependent immobilization of DIV-VSD. We proposed that such gating model underlies the generation of large I NaP in Na v 1.3 by RTX-VII.

Methods
Venom and toxin purification. Spider Macrothele raveni were collected in GuangXi province, China. The spider has a body length of 3-5 cm and the venom was collected by an electric stimulation method as described in another work of our laboratory 34 .
The collected crude venom was lyophilized and preserved at 280uC before use. The crude venom was dissolved in ddH 2 O to a final concentration of 5 mg/ml and subjected to the first round of RP-HPLC purification (acetonitrile gradient: 1%-60%, at an increasing rate of 1% per minute). The fraction containing RTX-VII was then collected, lyophilized and subjected to the second round of RP-HPLC with a slower increasing acetonitrile gradient (acetonitrile at an increasing rate of 0.5% per minute) to obtain the purified toxin.
Toxin sequencing and cDNA of RTX-VII. Partial amino acid sequence of RTX-VII was determined by Edman degradation on an Applied Biosystems/PerkinElmer Life Science Procise 491-A protein sequencer. The cDNA of this toxin was obtained by blasting Edman degradation determined amino acid sequence of RTX-VII against the local cDNA library database of the spider Macrothele raveni (unpublished data).
Constructs and transfection. All Na v clones and beta subunit clones were kindly gift from Dr Theodore R.Cummins (Department of pharmacology and Toxicology, Stark Neurosciences Research Institute, Indiana University School of Medicine, USA). cDNA genes encoding rat Na v 1.3 and rat Na v 1.4 were subcloned into the vectors pcDNA3.1 and pRGB4 40,56 , respectively; the cDNA genes encoding human Na v 1.5 and human Na v 1.7 were subcloned into the vectors pcDNA3.1 and pcDNA3.1-mod 57 , respectively. Auxiliary b1 and b2 subunits both were cloned from human and inserted into an internal ribosome entry site vector 58 . All site mutations of Na v 1.3 were constructed by using the QuikChange II XL Site-directed Mutagenesis kit (Agilent Technologies) according to the manufacture's instruction. The cytosolic boundaries of two adjacent transmembrane segments and two adjacent domains of Na v 1.3 or Na v 1.5 were determined by proteins' topological information deposited in NCBI protein database (for Na v 1.3, the website is http://www.ncbi.nlm.nih.gov/protein/ NP_037251.1, and for Na v 1.5,the website link is http://www.ncbi.nlm.nih.gov/ protein/NP_932173.1). The protein sequence location of each voltage sensor (VSD)/ pore domain (PD) of all four domains of Na v 1.3 and Na v 1.5 are as listed in Supplementary Table S4. A homologous recombination strategy was employed to generate the chimeric channels using the In-FusionHHD Cloning kit (Clontech Laboratories) or CloneEZH PCR Cloning kit (Genscript). For example, for the construction of Na v 1.3/1.5 DI-VSD chimera, the DI-VSD (voltage sensor of domain I) of Na v 1.5 was amplified by PCR using a pair of primers with their 59 end extended by a 15 bp long joint which is homologous or reverse compliment to the upstream or downstream flanking sequence of DI-VSD of Na v 1.3. A pair of oppositely directed primers was used to linearize the whole Na v 1.3 cloned plasmid with the DI-VSD of Na v 1.3 deleted. The PCR amplified segment and the linearized plasmid were subjected to 1% agarose gel electrophoresis, respectively. The corresponding bands were recycled using a DNA gel extraction kit (Sangon biotech) and ligated using the In-FusionHHD Cloning kit (Clontech Laboratories) or CloneEZH PCR Cloning kit (Genscript). Before being transformed to E.coli Top10 competent cell, the ligated product was subjected to FastDigest DpnI (Thermo Scientific) treatment at 37uC for 1 hour to remove the template plasmid. The transformants were verified by colony-PCR using a pair of gene specific primer for each inserted segment and then sequencing (Genscript). The primers used for vector linearization and amplification of Na v domains were listed in Supplementary Table S2 and Table S3. HEK293T cells (ATCC) were grown under the standard cell culture conditions (5% CO 2 and 37uC) in Dulbecco's Modified Eagle Medium (DMEM, Life technologies) supplemented with 10% fetal bovine serum. These Na v constructs were co-transfected with plasmid containing b1 subunit and PEGFP-N1 to HEK293T cells using Lipofectamine 2000 (Life Technology) according to the manufacture's instruction. For wt-Na v 1.3, Na v 1.3 mutants and Na v 1.3 derived chimeric channels, 3 mg Na v plasmid, 1 mg plasmid containing b1 subunit and 0.5 mg PEGFP-N1 plasmid were co-transfected. For wt-Na v 1.5 and Na v 1.5 derived chimeric channels, 1 mg Na v plasmid, 0.3 mg plasmid containing b1 subunit and 0.5 mg PEGFP-N1 plasmid were co-transfected. For ramp test, Na v 1.3 was co-transfected with plasmid containing b1 subunit and plasmid containing b2 subunit 33 . Cells were 80%-90% confluent before transfection, and cells were seeded on a poly-lysine coated Microscope Cover Glass (Fisher scientific) 4-6 hours after transfection. 24 hours after seeding, cells were ready for patch-clamp analysis.
Primary culture of DRG and hippocampal neurons and toxicity test of animals. Animals (Sprague-Dawley rats and Kunming mice) were used according to the guidelines of the National Institutes of Health for care and use of laboratory Animals. The experiments were approved by the Animal Care and Use Committee of the College of Medicine, Hunan Normal University. Acutely dissociated dorsal root ganglion (DRG) cells were prepared from 4 weeks old Sprague-Dawley rats and maintained in short-term primary culture using the method described by Hu, H.Z and Li, Z.W 59 . The dissociated cells were suspended in DMEM supplemented with 10% fetal bovine serum, 50 IU/ml penicillin, and 50 mg/ml streptomycin. Cells were seeded on poly-L-lysine-coated Microscope Cover Glass placed in a cell culture dish (35 3 10 mm, corning) and incubated at 37uC in an atmosphere of 5% CO 2 . Cells cultured for 3-24 h were used in the patch experiments. Experiments were conducted at room temperature (20-25uC). For primary culture of hippocampal neurons, hippocampal tissues of neonatal rats were dissected and treated with 0.25% trypsin in Ca 21 -Mg 21 -free Hank's Buffered Salt solution at 37uC for 15 min, and then were dissociated by trituration with glass Pasteur pipette and seeded on poly-L-lysinecoated Microscope Cover Glass placed in a cell culture dish (35 3 10 mm, corning). Approximate 3.5*10 4 cells in DMEM containing 10% fetal bovine serum were plated in each dish. The culture medium were replaced with serum-free NeurobasalH www.nature.com/scientificreports medium (Life technologies) supplemented by 2% B27 (Life technologies) on the second day after plating, 500 mM glutamine was added to reduce the growth of glial cells. The hippocampal neurons were maintained in a CO 2 incubator at 37uC, onehalf volume of the culture medium was replaced with fresh medium every other day. The neurons were used for patch-clamp analysis after they were maintained in culture for 14-17 days. In order to test the neurotoxicity of RTX-VII, ten mice of either sex with an average weight of 20 g were randomly divided to two groups, animals in the control group were intracerebroventricularlly injected with 20 mL saline, and animals in the experimental group were injected with 20 mL saline containing 20 ng toxin.
Electrophysiology. Cell current recording was made with the whole-cell patch-clamp technique using an EPC 10 USB Patch Clamp Amplifier (HEKA Elektronik). Cells transfected with wt/mutant/chimeric Na v channels and DRG/hippocampal neurons seeding in a glass coverslip were placed in a perfusion chamber in which rapid exchange of solutions around cells could be performed. The recording pipettes were made from glass capillary (thickness 5 0.225 mm) using a PC-10 puller (NARISHIGE).The pipet resistance was controlled at 1.5-2.0 MV by adjusting the pulling temperature. The standard pipet solution contained (in mM): 140 CsCl, 10 NaCl, 1 EGTA, 2 Mg-ATP, and 20 HEPES (pH 7.4). Bath solution contained (in mM): 140 NaCl, 2 CaCl 2 , 1 MgCl 2 , 5 KCl, 20 HEPES (pH 7.4), and 10 glucose. All experiments were conducted at the room temperature (20-25uC). All chemicals were the products of SigmaAldrich and dissolved in water. Data were acquired by PatchMaster software (HEKA Elektronik). Data were analyzed by softwares Igo Pro 6.10A, Excel 2010, Sigmaplot 10.0 and OriginPro 8. Voltage errors were minimized by using 80% series resistance compensation, the speed value of Rs compensation was set to be 10 ms(fast compensation).The capacitance artifact were canceled using the computer-controlled circuitry of the patch clamp amplifier. The pipet capacitance was minimized by filling the pipet with small volume of pipet solution, and the pipet capacitance was controlled to be ,10 pF for effective automatic compensation by EPC-10 amplifier. The pipet capacitance and the cell capacitance was sequentially compensated after the seal and the whole-cell configuration was established, respectively(pipet capacitance and cell capacitance was compensated by automatic fast and slow capacitance compensation, respectively).Stock solution of RTX-VII (1 mM in sterile ddH 2 O) was diluted with fresh bath solution to a concentration of 10 folds of the interested concentration, 30 mL of the concentrated toxin was diluted into the recording chamber (containing 270 mL bath solution) far from the recording pipet (the recording cell) and was mixed by repeatedly pipetting to achieve the specified final concentration 60 . The dose-response curves of toxin on wt/mutant/ chimeric channel were fitted to a Hill equation to estimate the potency of toxin (EC 50 ). The G-V curve before and after toxin treatment and the steady state inactivation (SSI) curve before toxin treatment were fitted using a boltzmann equation: y 5 1/(1 1 exp[(V 1/2 2 V)/K]) in which V 1/2 , V, and K represented midpoint voltage of kinetics, test potential and slope factor, respectively. The SSI curve after toxin treatment was fitted with a modified Boltzmann equation: (Y 2 Y min )/(Y max 2 Y min ) 5 1/(1 1 exp[(V 1/2 2 V)/K)]), Y max and Y min represent the maximum and minimum responses. The time course curve for the I NaP enhancement in response to the toxin application was best fitted by a single exponential rising equation (y 5 y 0 1 a(1 2 e 2x/t )) and the time course for recovery of I NaP upon bath solution washing was best fitted by a single exponential decay equation(y 5 y 0 1 ae 2x/t ), here t represent the time constant for toxin binding to and washing off from channels respectively. In measuring the spontaneous AP firing of neonatal rat hippocampal neuron using current-clamp, the pipette solution contains (in mM): 140 KCl, 5 MgCl 2 , 5 EGTA, 2.5 CaCl 2 , 4 ATP, 0.3 GTP, and 10 Hepes, pH 7.3 (adjusted with KOH). The bath solution contains (in mM): 140 NaCl, 1 MgCl 2 , 5 KCl, 2 CaCl 2 , 10 HEPES, and 10 glucose, pH 7.3 (adjusted with NaOH). During the recording, no current was injected to neurons.
Data analysis. Data were presented as Mean 6 SD. n is presented as the number of the separate experimental cells. Dose response curves were fitted using the following Hill logistic equation: y 5 f max 2 (f max 2 f min )/(11([Tx]/EC 50 ) n ), where f max and f min represent the maximum and minimum response of channel to toxin, [Tx] represent the toxin concentration, n is an empirical Hill coefficient. The Hill coefficient was set to 1 except where indicated otherwise. This is reasonable based on our mutagenesis analysis, which indicated a single high affinity binding site in Na v 1.3 for RTX-VII. Statistical significance was assessed with Microsoft excel 2010 using One-Way ANOVA. Statistical significance was accepted at P values less than 0.05.