Screening an In-House Isoquinoline Alkaloids Library for New Blockers of Voltage-Gated Na+ Channels Using Voltage Sensor Fluorescent Probes: Hits and Biases

Voltage-gated Na+ (NaV) channels are significant therapeutic targets for the treatment of cardiac and neurological disorders, thus promoting the search for novel NaV channel ligands. With the objective of discovering new blockers of NaV channel ligands, we screened an In-House vegetal alkaloid library using fluorescence cell-based assays. We screened 62 isoquinoline alkaloids (IA) for their ability to decrease the FRET signal of voltage sensor probes (VSP), which were induced by the activation of NaV channels with batrachotoxin (BTX) in GH3b6 cells. This led to the selection of five IA: liriodenine, oxostephanine, thalmiculine, protopine, and bebeerine, inhibiting the BTX-induced VSP signal with micromolar IC50. These five alkaloids were then assayed using the Na+ fluorescent probe ANG-2 and the patch-clamp technique. Only oxostephanine and liriodenine were able to inhibit the BTX-induced ANG-2 signal in HEK293-hNaV1.3 cells. Indeed, liriodenine and oxostephanine decreased the effects of BTX on Na+ currents elicited by the hNaV1.3 channel, suggesting that conformation change induced by BTX binding could induce a bias in fluorescent assays. However, among the five IA selected in the VSP assay, only bebeerine exhibited strong inhibitory effects against Na+ currents elicited by the hNav1.2 and hNav1.6 channels, with IC50 values below 10 µM. So far, bebeerine is the first BBIQ to have been reported to block NaV channels, with promising therapeutical applications.

Isoquinoline alkaloids (IA) represent one of the most diverse and abundant groups of plant compounds. They are biosynthesized from tyrosine, leading to several chemical subgroups such as benzylisoquinoline, aporphine, protoberberine, protopine, morphinan, and many others [7,26]. Until now, only two of the large aporphine alkaloid subgroups (>1000 members) have been described as NaV channel blockers: liriodenine from Fissistigma glaucescens [27] and crebanine from Stephania sp. [28] (Figure 1). Both compounds exhibit anti-arrhythmic activities mediated by the inhibition of multiple cardiac ion channels, such as the NaV, CaV, and KV channels [27][28][29]. However, this IA group has been poorly studied for their effects on NaV channels. Therefore, in this work, we aimed to identify new Nav channel blockers among IA from an in-house library composed of plant alkaloids. Quinoline alkaloids ß-carbolines alkaloids Isoquinoline alkaloids Figure 1. Structural diversity of natural alkaloids active on Na V channels. Tetrodotoxin and saxitoxin are two marine guanidine alkaloids. Veratridine, aconitine, and batrachotoxin are activators of the Na V channels. Lappaconitine, quinidine, and ajmaline are two inhibitors used as clinical drugs. Liriodenine and crebanine, two IA, block the Na V channels. Lappaconitine, which is structurally related to aconitine [15], exhibits potent analgesic properties [16] and has recently been proposed for the treatment of chronic pain [17]. Ajmaline, isolated from Rauwolfia serpentina, inhibits the Na V 1.5 channels [18][19][20][21]. Ajmaline is used in the diagnosis and treatment of cardiac arrhythmia [22]. Brucine, extracted from Nux vomica, exhibits antinociceptive properties mediated by Na V channel inhibition [23]. Harmaline from Peganum harmala [24] and dehydrocrenatidine from Picrasma quassioides are used to treat pain [25].
Isoquinoline alkaloids (IA) represent one of the most diverse and abundant groups of plant compounds. They are biosynthesized from tyrosine, leading to several chemical subgroups such as benzylisoquinoline, aporphine, protoberberine, protopine, morphinan, and many others [7,26]. Until now, only two of the large aporphine alkaloid subgroups (>1000 members) have been described as Na V channel blockers: liriodenine from Fissistigma glaucescens [27] and crebanine from Stephania sp. [28] (Figure 1). Both compounds exhibit anti-arrhythmic activities mediated by the inhibition of multiple cardiac ion channels, such as the Na V , Ca V, and K V channels [27][28][29]. However, this IA group has been poorly studied for their effects on Na V channels. Therefore, in this work, we aimed to identify new Na v channel blockers among IA from an in-house library composed of plant alkaloids.
We focused our studies on the IA subgroups aporphine, 1-benzyl isoquinoline (BIQ), protoberberine, and bisbenzylisoquinoline (BBIQ) in our search for new blockers of Na V channels. The screening assay consists in the measurement of the FRET voltage sensor probe (VSP) signal elicited by the activation of Na V channels that are endogenously expressed in GH3b6 cells by BTX in the absence and in the presence of 62 IA from our in-house natural compounds library. This allowed us to identify five putative Na V channel blockers: liriodenine, oxostephanine, thalmiculine, protopine, and bebeerine. These five IA were assayed for their ability to block the ANG-2 fluorescence emission induced by BTX in HEK293-expressing hNav1.3 and then to inhibit Na + currents elicited by the hNav1.2, hNav1.3, and hNav1.6 channels via automated patch-clamp electrophysiology. Our findings highlighted that the use of the activators of Na V channels, such as BTX, which exhibits a "foot-in-the-door" mechanism and thus freezes the Na V channels in an open state, could compromise the discovery of selective and potent inhibitors.

VSP-FRET Assays Using GH3b6 for Pharmacological Characterization of Na V Channels
The GH3b6 cell line was selected for its wide use as a cellular model for pharmacological studies on Na V channels [30][31][32][33][34][35]. Here, we used RT-qPCR to establish the Na V channel gene expression profile of GH3b6 cells. Our data showed the significant expression of four Nav channel transcripts (scn1a, scn2a, scn3a, and scn8a), which encoded four neuronal and TTX-sensitive Na V channel subtypes (Na V 1.1, 1.2, 1.3, and 1.6, respectively) ( Figure 2). The rank order of the expression levels of these transcripts was scn3a > scn8a > scn2a > scn1a (p < 0.0001). The amplification of scn3a was observed with the lowest Ct-values (Ct = 22.9, for 10 ng of ADNc). Our data indicate that GH3b6 cells mainly expressed the neuronal TTX-sensitive Na V 1.3 channel subtype, which thereby tends to be consistent with earlier observations [35][36][37].
We focused our studies on the IA subgroups aporphine, 1-benzyl isoquinoline (B protoberberine, and bisbenzylisoquinoline (BBIQ) in our search for new blockers of channels. The screening assay consists in the measurement of the FRET voltage se probe (VSP) signal elicited by the activation of NaV channels that are endogenously pressed in GH3b6 cells by BTX in the absence and in the presence of 62 IA from ou house natural compounds library. This allowed us to identify five putative NaV cha blockers: liriodenine, oxostephanine, thalmiculine, protopine, and bebeerine. These IA were assayed for their ability to block the ANG-2 fluorescence emission induce BTX in HEK293-expressing hNav1.3 and then to inhibit Na + currents elicited by hNav1.2, hNav1.3, and hNav1.6 channels via automated patch-clamp electrophysiol Our findings highlighted that the use of the activators of NaV channels, such as BTX, w exhibits a "foot-in-the-door" mechanism and thus freezes the NaV channels in an o state, could compromise the discovery of selective and potent inhibitors.

VSP-FRET Assays Using GH3b6 for Pharmacological Characterization of NaV Channels
The GH3b6 cell line was selected for its wide use as a cellular model for pharm logical studies on NaV channels [30][31][32][33][34][35]. Here, we used RT-qPCR to establish the NaV c nel gene expression profile of GH3b6 cells. Our data showed the significant expressio four Nav channel transcripts (scn1a, scn2a, scn3a, and scn8a), which encoded four neur and TTX-sensitive NaV channel subtypes (NaV1.1, 1.2, 1.3, and 1.6, respectively) (Figur The rank order of the expression levels of these transcripts was scn3a > scn8a > scn scn1a (p < 0.0001). The amplification of scn3a was observed with the lowest Ct-value = 22.9, for 10 ng of ADNc). Our data indicate that GH3b6 cells mainly expressed the ronal TTX-sensitive NaV1.3 channel subtype, which thereby tends to be consistent earlier observations [35][36][37]. GH3b6 cells express TTX-S Nav channels. Nav channel gene expression profile of GH cells. The histogram illustrates the expression repertoire of Nav channel genes in GH3b6 ce established by RT-qPCR. Data are the mean of mRNA relative expression ± SEM (n = 6). ND Detected. Significance tests between groups of data were performed using one-way ANOVA lowed by Bonferroni's multiple comparison test (**** p < 0.0001).
Next, we measured the fluorescent resonance transfer (FRET) between CC2-DMPE DisBAC2(3) in GH3b6 cells to characterize the change in the membrane potential ind Significance tests between groups of data were performed using one-way ANOVA followed by Bonferroni's multiple comparison test (**** p < 0.0001).
Next, we measured the fluorescent resonance transfer (FRET) between CC2-DMPE and DisBAC2 (3) in GH3b6 cells to characterize the change in the membrane potential induced by BTX as it is considered as a full activator of Na V channels [8,13,14]. The fluorescence emission of CC2-DMPE increased, which is in parallel with the decrease of DisBAC2(3) fluorescence in the presence of increasing concentrations of BTX (from 0.01 to 3 µM) (Figure 3a). Subsequently, the VSP fluorescence emission ratio increased in a concentration-dependent manner. The data were best fitted by the Hill-Langmuir equation with a variable slope for a bimolecular reaction with an EC 50 value of 0.27 ± 0.06 µM (Figure 3b). At 100 nM, TTX fully abolished the VSP fluorescence emission ratio induced by 1 µM BTX (Figure 3b). The TTX inhibitory effects were fitted as described above, with an IC 50 value of 8.58 ± 1.67 nM (Figure 3b). Indeed, BTX binds to the open state of Na V channels and consequently prevents their closing in an irreversible manner. Then, the injection of Na + , which depolarized the membrane, led to the activation of the Na V channels and allowed for BTX binding. Since this interaction is irreversible, the cells became leaky to Na + . This led to an Na + influx, which was counteracted by an Na + efflux mediated by Na + /K + ATPase and other Na + transporters (e.g., NCX). This explains the slow kinetic of the fluorescence emission, which reached a plateau. This latter corresponds to an equilibrium between the Na + influx and the Na + efflux. Thus, the VSP signal induced by BTX was inhibited by TTX at a nanomolar concentration, in accordance with the presence of TTX-sensitive Nav channels. Altogether, these results demonstrate that GH3b6 cells are suitable for a FRET-based VSP assay for the screening of Na V channel inhibitors. by BTX as it is considered as a full activator of NaV channels [8,13,14]. The fluorescence emission of CC2-DMPE increased, which is in parallel with the decrease of DisBAC2(3) fluorescence in the presence of increasing concentrations of BTX (from 0.01 to 3 µM) ( Figure 3a). Subsequently, the VSP fluorescence emission ratio increased in a concentration-dependent manner. The data were best fitted by the Hill-Langmuir equation with a variable slope for a bimolecular reaction with an EC50 value of 0.27 ± 0.06 µM (Figure 3b). At 100 nM, TTX fully abolished the VSP fluorescence emission ratio induced by 1 µM BTX ( Figure  3b). The TTX inhibitory effects were fitted as described above, with an IC50 value of 8.58 ± 1.67 nM ( Figure 3b). Indeed, BTX binds to the open state of NaV channels and consequently prevents their closing in an irreversible manner. Then, the injection of Na + , which depolarized the membrane, led to the activation of the NaV channels and allowed for BTX binding. Since this interaction is irreversible, the cells became leaky to Na + . This led to an Na + influx, which was counteracted by an Na + efflux mediated by Na + /K + ATPase and other Na + transporters (e.g., NCX). This explains the slow kinetic of the fluorescence emission, which reached a plateau. This latter corresponds to an equilibrium between the Na + influx and the Na + efflux. Thus, the VSP signal induced by BTX was inhibited by TTX at a nanomolar concentration, in accordance with the presence of TTX-sensitive Nav channels. Altogether, these results demonstrate that GH3b6 cells are suitable for a FRET-based VSP assay for the screening of NaV channel inhibitors.

Screening Vegetal Alkaloids for Blockers of NaV Channels Using VSP in GH3b6 Cells
Sixty-two IA (from IA1 to IA77) from our in-house library were chosen for their structure, none of which, except for liriodenine, had been described as a NaV channel ligand (see supplementary Table S1, for trivial name, chemical class (CAS number), structure,

Screening Vegetal Alkaloids for Blockers of Na V Channels Using VSP in GH3b6 Cells
Sixty-two IA (from IA1 to IA77) from our in-house library were chosen for their structure, none of which, except for liriodenine, had been described as a Na V channel ligand (see supplementary Table S1, for trivial name, chemical class (CAS number), structure, plant origin, and reference). Thus, we first blind-tested these 62 IA for their ability to inhibit a BTX-induced VSP signal in GH3b6 cells ( Figure 4). Figure 4a shows the BTX-induced VSP signal in the absence and in the presence of each IA. Thus, thirty samples exhibited an inhibition rate of 50% and more ( Figure 4b). To select the potent inhibitors of Nav channels, we used a threshold of 75% signal inhibition since the IA had been screened at rather high concentrations (10-50 µg/mL, corresponding to 12-20 µM) (Figure 4b). Then, ten IA were selected for further chemical analysis: (1) HPLC monitoring at two different wavelengths (210 nm and 280 nm) to evaluate their purity in %, (2) NMR structure validation, and (3) if necessary, mass spectrometry (MS) analysis and/or optical rotation to confirm IA identity ( Table 1). It was only at this stage that the different samples were identified and the blind evaluation was stopped. Samples IA50 (tetrandrine), IA31 (tiliacorine), IA36 (sukhodianine), and IA42 (gyrocarpusine) were rejected because of their low purity (<75% at 210 nm and 280 nm). Therefore, we retained the following five IA for further pharmacological characterization: oxostephanine (IA14 and IA24), liriodenine (IA39), thalmiculine (IA49), bebeerine (IA52), and protopine (IA69) ( Table 1).

Characterization of the Effects of Oxostephanine, Liriodenine, Thalmiculine, Bebeerine, and Protopine on Na + Currents Elicited by the hNaV1.2, hNaV1.3, and hNaV1.6 Channels
To validate the inhibitory effects of oxostephanine, liriodenine, thalmiculine, bebeerine, and protopine on NaV channels, automated patch-clamp experiments were performed using three different cell lines stably expressing the NaV channel subtypes detected by RT-qPCR in GH3b6 cells. These five IA were assayed on the Na + currents elicited by the depolarizing step in the three cell lines, which stably expressed the hNaV1.2 (HEK293-hNaV1.2 cells), hNaV1.3 (HEK293-hNaV1.3 cells), and hNaV1.6 channels (HEK293-hNaV1.6 cells). At 10 µM, oxostephanine, liriodenine, thalmiculine, and protopine had no effects on the hNaV1.2, hNaV1.3, and hNaV1.6 channels (Figure 7a and Supplementary Figure S5). At the same concentration, bebeerine was able to inhibit the Na + currents elicited by the hNaV1.2, hNaV1.3, and hNaV1.6 channels (Figure 7a,b). The inhibition rate of bebeerine on these three NaV channels did not depend on the Vm (two-way ANOVA test followed by Holm-Sidak's multiple comparison test, p < 0.001). At a Vm between −20 mV and 30 mV, bebeerine had stronger inhibitory effects on the hNaV1.3 (60-64%, n = 38) and hNaV1.6 channels (70-74%, n = 32) than on the hNaV1.2 channel (42-50%, To validate the inhibitory effects of oxostephanine, liriodenine, thalmiculine, bebeerine, and protopine on Na V channels, automated patch-clamp experiments were performed using three different cell lines stably expressing the Na V channel subtypes detected by RT-qPCR in GH3b6 cells. These five IA were assayed on the Na + currents elicited by the depolarizing step in the three cell lines, which stably expressed the hNa V 1.2 (HEK293-hNa V 1.2 cells), hNa V 1.3 (HEK293-hNa V 1.3 cells), and hNa V 1.6 channels (HEK293-hNa V 1.6 cells). At 10 µM, oxostephanine, liriodenine, thalmiculine, and protopine had no effects on the hNa V 1.2, hNa V 1.3, and hNa V 1.6 channels (Figure 7a and Supplementary Figure S5). At the same concentration, bebeerine was able to inhibit the Na + currents elicited by the hNa V 1.2, hNa V 1.3, and hNa V 1.6 channels (Figure 7a,b). The inhibition rate of bebeerine on these three Na V channels did not depend on the Vm (two-way ANOVA test followed by Holm-Sidak's multiple comparison test, p < 0.001). At a Vm between −20 mV and 30 mV, bebeerine had stronger inhibitory effects on the hNa V 1.3 (60-64%, n = 38) and hNa V 1.6 channels (70-74%, n = 32) than on the hNa V 1.2 channel (42-50%, n = 38, Figure 7c). Notably, bebeerine significantly inhibited the hNa V 1.6 channel to a higher extent (74 ± 3%, n = 32) than the hNa V 1.3 channel (64 ± 4%, n = 48; p < 0.05), at −20 mV and −15 mV, respectively (Figure 7c). Altogether, these data show that bebeerine efficiently blocked the hNa V 1.2, hNa V 1.3, and hNa V 1.6 channels, with the following rank order: hNa V 1.6 > hNa V 1.3 > hNa V 1.2. and hNa V 1.6 channels. The inhibition rate induced by IA52 on the Na + currents was normalized (Vm = −30 to 30 mV). Data were plotted as mean ± SEM and analyzed for statistical significance using a two-way ANOVA test with Holm-Sidak's multiple comparison test (p < 0.0001). Different letters (a, b, and c) above (hNa V 1.2 and hNa V 1.6) and below (hNa V 1. 3) indicate significant differences between groups. p-values are shown in Supplementary Table S1. Figure 8 illustrates the effects of bebeerine (at 10 µM) on the voltage-dependent activation and inactivation of the hNa V 1.2, hNa V 1.3, and hNa V 1.6 channels. All fitting parameters are summarized in Table 3. There was no significant change in both the halfactivation and inactivation potential (V 1/2 ) of the Na + currents elicited by the hNa V 1.2 channels (p = 0.72 and p = 0.69, respectively) nor by the hNa V 1.6 channels (p = 0.71 and p = 0.32, respectively) in the presence of bebeerine (Figure 8a,c). Concerning the hNa V 1.3 channel, bebeerine induced a slight negative shift in V 1/2 activation (−3.6 ± 1.0, p < 0.001), but not in V 1/2 activation (p = 0.08, Figure 8b). However, a significant increase of slope values for both the activation and inactivation curves was observed for each Na V channel subtype in the presence of bebeerine (Table 3). , and hNa V 1.6 (c) channels. Control experiments were performed with extracellular solution containing 0.1% of DMSO corresponding to the vehicle. Data are shown as mean ± SEM. An unpaired t-test was used to analyze the significance between groups. ns: not significant; **, p < 0.01; ***, p < 0.001. Fitted parameters are shown in Table 3. Table 3. Fitting parameters of the voltage-dependent activation and inactivation of Na + currents elicited by the hNa V 1.2, hNa V 1.3, and hNa V 1.6 channels in the absence and in the presence of bebeerine. This table shows the fitting parameters determined from the non-linear regression analysis of the voltagedependent activation and inactivation of Na + currents in the absence or presence of 10 µM bebeerine (IA52, Figure 8). V 1/2 , half potential (mV) for voltage-dependent activation or inactivation. Data are means ± SEM. Statistical significance between the control and test groups was evaluated with an unpaired t-test. 1,2,3,4,5 , p < 0.0001; 6 , p < 0.001.
Oxostephanine and liriodenine did not exhibit significant inhibitory effects at 10 µM on Na + currents elicited by the Na V 1.3 channels, while both IA were able to efficiently inhibit the BTX-induced ANG-2 fluorescence signals in HEK293-hNa V 1.3 cells. Thus, we assayed these two IA on the Na + currents elicited by the hNa V 1.3 channel in the presence of 1 µM BTX. As expected, BTX profoundly altered the Na + currents elicited by the hNa V 1.3 channel by increasing the current densities and inhibition of inactivation (Figure 9a). Interestingly, in the presence of BTX, oxostephanine and liriodenine significantly decreased the current densities of the hNa V 1.3 channel at −60 mV (Figure 9a-c); however, only IA39 positively shifted the V 1/2 activation in the presence of BTX, but to a small extent (p < 0.01, Figure 9b).
We also examined the effects of oxostephanine and liriodenine on the inactivation process of hNa V 1.3 in the presence and in the absence of BTX. We evaluated the AUC of currents traces, which illustrate the inactivation rate of hNa V 1.3 channels. BTX induced the 55-fold increase of the AUC, indicating the decrease of the inactivation kinetic (p < 0.0001, data not shown). Interestingly, the slow-down of inactivation induced by BTX was significantly decreased by both IA14 and IA39 (p < 0.05, Figure 9b,c).   . (b,c) The scatter plots with bars illustrate: the current densities at −60 mV, the V 1/2 of activation, the slope of activation curves, and the area under the curve (AUC in pA.ms) in the absence or in the presence of IA14 (b) and IA39 (c). Data are shown as mean ± SEM. Parametric or non-parametric statistical analyses were performed to determine the significance between groups. *, p < 0.05; ns: not significant.

Discussion
Our study aimed to identify new ligands of the Nav channels by screening vegetal IA from an in-house library. The IA family offers a large molecular diversity that has not been extensively studied for their activities against Na V channels. Using fluorescent VSP, we identified two oxoaporphine alkaloids (oxostephanine and liriodenine), two BBIQ (bebeerine and thalmiculine), and a protopine (protopine) which inhibited BTX-induced depolarization in GH3b6 cells, with IC 50 values in the micromolar range. However, by directly measuring [Na + ] i elevation, using the fluorescent probe ANG-2, only oxostephanine and liriodenine exhibited strong inhibitory effects against the hNa V 1.3 channel, which was stably expressed in HEK293 cells. Moreover, in patch-clamp experiments, only bebeerine was able to efficiently block the Na + currents elicited by the hNav1.2 and hNav1.6 channels. We also showed that by altering the conformation of Na V channels, BTX allowed for the binding of oxostephanine and liriodenine.
The endocrine GH3b6 cell line has been extensively used as a cellular model to study Na V channels in patch-clamp experiments, notably those expressed in the central nervous system [39]. As we have previously demonstrated, these cells mainly express the endogenous Na V 1.3 channel subtype [35]. Thus, these cells represent a good compromise in the search for or in characterizing new ligands of the Na V 1.3 channel, a Na V channel involved in neuropathic pain [40]. To efficiently and rapidly screen our in-house IA library, we used VSP technology. Indeed, this strategy has been proven to be adapted for the detection and selection of ligands of Na V channels from chemical libraries or natural fractions [41][42][43][44][45]. For this strategy, it is necessary to activate Na V channels with selective activators, such as BTX or VTD. In our hands, fluorescent signals induced by BTX were stronger than those induced by VTD (data not shown), likely because BTX acts as a full activator [46]. However, despite an IC 50 value in the nanomolar range for TTX, we were very surprised that BTX-induced VSP signals were inhibited at more than 50%, with almost half of our in-house library in the range of 12-20 µM. Those higher concentrations could lead to a non-specific action on the cell that would modify the VSP signal indirectly. We then decided to select only IA with an inhibition rate of more than 75%, which yielded 10 hits. Among these hits and according to the physico-chemical analysis, five IA were selected: oxostephanine, liriodenine, thalmiculine, bebeerine, and protopine.
The VSP assays indicated that these five IA inhibit the BTX-induced VSP signals in GH3b6 cells with different efficiencies. While the oxoaporphines, oxostephanine, and liriodenine totally suppressed the BTX-induced VSP signals, thalmiculine, bebeerine, and protopine hardly inhibited these signals to 30-65%. This could be explained by the fact that BTX activates the Na V channel by binding to the pore with very high affinity, preventing any inactivation of the channels; this thereby promotes the identification of pore blockers, as it has already been discussed for VTD [47]. Thus, we first hypothesized that oxostephanine and liriodenine would behave as pore blockers, while bebeerine, thalmiculine, and protopine would exhibit a different mechanism of action. However, the use of VSP for the screening assay has been greatly debated, particularly in the search for selective Na V 1.7 inhibitors [47,48]. Notably, this is because this technique requires the use of a neurotoxin activator, which can trap Na V channels in a particular conformation. Thus, this approach needs complementary techniques, such as patch-clamp recording or the use of fluorescent Na + probes, both of which directly measure Na V channel activity. Here, we used ANG-2, which has been proven to be the most efficient in monitoring intracellular Na + concentration [38].
Using this Na + probe, we found that only oxostephanine and liriodenine were able to efficiently inhibit BTX-induced [Na + ] i elevations in hNa V 1.3-HEK293 cells. Indeed, at 10 µM, these two IA almost abolished BTX-induced [Na + ] i elevations, suggesting the micromolar affinities of both IA for the hNa V 1.3 channel in the presence of BTX. However, in our assay, thalmiculine, bebeerine, and protopine exhibited modest effects at 10 µM. Several hypotheses could be made: (1) thalmiculine, bebeerine, and protopine probably have weaker effects on the hNa V 1.3 channel than oxostephanine and liriode-nine, (2) thalmiculine, bebeerine, and protopine could block other Na V channel subtypes, (3) the spatio-temporal resolution of our assay did not allow us to highlight their blocking effects on the hNa V 1.3 channel, and (4) these three IA target other ion channels, such as the Cav and Kv channels, alleviating any change of [Na + ] i by the activation of NCX and other transporters. These data showed the advantages but also the limitations of using fluorescent probes in screening for Na V channel ligands [38,47,49].
Our patch-clamp experiments showed that among the five IA selected by the VSP assays, only bebeerine behaved as a potent blocker of the hNa V 1.2 and hNa V 1.6 channels, with an IC 50 value likely above 10 µM. Neither oxostephanine, liriodenine, thalmiculine, nor protopine had any inhibitory effects on the hNa V 1.2, hNa V 1.3, and hNa V 1.6 channels at 10 µM, revealing that they are likely to be low-potency blockers of these Na V channel subtypes. We could not use higher concentrations of these compounds because of their low solubility and the non-specific effects of the DMSO concentrations. These unexpected results could be explained by the bias induced by BTX as a Na V activator in fluorescentbased assays, as it has been previously discussed for VTD [47]. Just as VTD, BTX is known to bind within the pore, thus trapping Na V channels in an open state [8,13,50], allowing for the binding of low-affinity blockers [47]. This assumption was confirmed by the fact that oxostephanine and liriodenine decreased the BTX effects on both the activation and inactivation properties of the hNa V 1.3 channel, while both IA had no effects on the inactivation properties of this Na V channel in the absence of BTX. Thus, we assumed that both IA bind to Na V channels in the presence of BTX, which traps Na V channels in an open state. Altogether, our data highlighted that fluorescent-based screening assays on BTX-activated Na V channels could compromise the finding of potent and selective blockers of these ion channels, and that the patch-clamp technique remains unavoidable for the accurate characterization of Na V channel ligands.
Oxostephanine and liriodenine are two members of the oxoaporphine subgroup, and they only differ by a methoxy group on the benzyl part of the skeleton of this IA subgroup. Previous studies have demonstrated the ability of liriodenine to suppress Na + currents recorded in isolated cardiomyocytes with an IC 50 value in the micromolar range [27,51], which is close to what we observed. Since cardiomyocytes mainly express Na V 1.5 [2], it is possible that liriodenine preferentially blocks this Na V channel subtype. Several studies have shown that oxostephanine and liriodenine share some biological activities, such as antimicrobial and anti-cancer properties [52,53]. Liriodenine has been proven to target ion channels such as the K V , Ca V, and Na V channels [27], but it also targets receptors such as the GABAA, nicotinic, and muscarinic receptors [54][55][56]. The high structural similarity between oxostephanine and liriodenine certainly supports the fact that they might share the same pharmacological targets, such as the Nav channels.
Finally, for the first time, we determined that bebeerine exhibits inhibitory effects on Na V channels, with a higher potency on the hNa V 1.6 channel than on the hNa V 1.2 and hNa V 1.3 channels. Bebeerine is the enantiomer of curine, which has been extensively studied and previously described as an L-Type Ca V channel blocker [57]. Thus, it is conceivable that bebeerine might also inhibit L-Type Ca V channels. Interestingly, no BBIQ has been reported as a Na V channel blocker so far, and since this IA subgroup is particularly large, further screening by an automated patch-clamp could be of interest. In addition, further studies are required to provide more information on the selectivity and also the blocking mechanism of bebeerine towards all Na V channel subtypes.

In-House Library of Vegetal Alkaloids
More than 300 alkaloids from plants have been collected in a library of the SONAS laboratory (EA 921, University of Angers) in the last four decades. These compounds were extracted and purified from at least 14 plant families (Annonaceae, Berberidaceae, Fumariaceae, Lauraceae, Menispermaceae, etc.), which were collected from all over the world (Bolivia, Colombia, France, Indonesia, Jordan, etc.) (Supplementary Figure S5). They were stored dried at room temperature. In a blind test, 62 IA associated with 10 different structural subtypes (Supplementary Table S2) were screened for their ability to inhibit BTX-evoked depolarization in GH3b6 cells. Each IA or fraction was solubilized in DMSO at 1-5 mg/mL, stored, and sheltered from light at 4 • C. For the fluorescence assays, the cells were plated at a density of 50,000 cells per well in black-walled, clear-bottom, 96-well plates and then cultured at 37 • C and 5% CO 2 for 24 h before the experiments. HEK293 stably expressing hNa V 1.2 and hNa V 1.6 were cultivated in the same conditions mentioned above for the HEK293-hNa V 1.3 cells. For the automated patch-clamp experiments, cells were passaged from 24 to 48 h prior to the experiment and replated in the appropriate quantity to reach 50-70% confluence. For electrophysiological recordings, cells were detached with trypsin-EDTA and diluted in a culture medium. Then, the cells were centrifuged at 800 rpm for 4 min and resuspended (~300,000 cells/mL) in a patch-clamp extracellular solution.

Fluorescent Assays
Voltage sensor probe (VSP) assays were carried out as described previously [59]. VSP were freshly prepared in an Na-free buffer (VSP Buffer) containing (in mM): 160 TEA-Cl, 0.1 CaCl 2 , 1 MgCl 2 , 11 Glucose, and 10 HEPES-K (pH 7.4). Cells were washed with the VSP Buffer and first incubated with CC2-DMPE (5 µM) and Pluronic ® -F127 acid (0.01%) dissolved in the VSP Buffer for 70 min at RT. After washing with the VSP buffer, the cells were loaded with DisBAC 2 (3) (10 µM) and VABSC-1 (250 µM) dissolved and diluted in the VSP Buffer for 45 min at RT. In control experiments, BTX (1-10 µM) with and without TTX (1-10 µM) was added to the VSP buffer containing DisBAC 2 (3) and VABSC-1. The plates were illuminated at λ = 405 nm, and the resulting fluorescence emissions were monitored at λ = 460 and λ = 580 nm. The plate reader measured the fluorescence emission every 5 s up to 145 s. After a 30 s baseline, cell depolarization was induced by the injection of a depolarizing solution (in mM: 160 NaCl, 4.5 KCl, 2 CaCl 2 , 1 MgCl 2 , 11 glucose, 10 HEPES-K, (pH 7.4)), raising the final Na + concentration to 80 mM. The osmolarity of the solution remained the same at around 300 ± 5 mOsmol before and after the addition of the depolarizing solution. For the screening assays, IA were co-incubated at 10-50 µg/mL with 1 µM of BTX in triplicate. In positive control experiments, BTX (1 µM) with TTX (10 µM) and DMSO (1%) were added to the VSP buffer containing DisBAC 2 (3) and VABSC-1. To establish the concentration-response relationships, increasing concentrations of compounds were co-incubated with BTX (1 µM). These experiments were performed in three wells and repeated twice. Intracellular Na + fluorescent assays were performed using the cellpermeant acetoxymethyl esters (AM) form of the Asante NaTRIUM Green-2 (ANG-2) probe, a cytosolic Na + indicator. The HEK293-hNa v 1.3 cells were incubated for 1 h in the dark at 37 • C in the ANG-2 probe (5 µM) prepared with Hank's buffer saline solution (HBSS), which contained (in mM): 2 CaCl 2 , 1 MgCl 2 , 5 KCl, 150 NaCl, 10 D-glucose, 10 HEPES (pH 7.4), and 0.02% pluronic acid. Afterwards, the quencher solution Ponceau 4R (P4R, 1 mM) was added, and the fluorescence signal was monitored at the following wavelengths: excitation 517 nm and emission 540 nm. Thirty seconds after baseline recording, BTX (1 µM) was automatically injected with or without TTX (1 µM) or IA (10 µM), and the fluorescence emission was read for 1000 s. All experiments were performed in triplicate wells and repeated thrice. Fluorescence measurement was carried out using a FlexStation ® 3 Benchtop Multi-Mode Microplate Reader (Molecular Devices, Sunnyvale, CA, USA).

Quantitative Real Time PCR
Total RNA from the GH3b6 cells was extracted using the RNeasy ® micro kit (Qiagen, Courtaboeuf, France). For retrotranscription, 1 µg of total RNA was used, and the reactions were performed using random hexamer primers and the QuantiTect ® Reverse Transcription kit (Qiagen). Real-time PCR assays were carried out on a LightCycler ® 480 Instrument II (Roche, Meylan, France) using the Sybr ® Select Master Mix (Applied Biosystems) with 2.5, 5, and 10 ng of cDNA in duplicate. Gene-specific primers for real-time PCR (Table 4) were designed using the Primer3 Software. Differences in transcript levels were determined using the cycle threshold method, as described by the manufacturer. Specificity of amplification was confirmed by melting curve analysis. Relative quantification of gene expression was normalized to the mean of expression of two validated House-keeping genes: gapdh (glyceraldehyde-3-phosphate dehydrogenase) and Gusb (beta-glucuronidase), according to the formula E = 2 −(Ct(Target)−Ctmean(Reference)) , where C t is the threshold cycle. Amplicon sizes (70-106 bp) and AT content (47-55%) were chosen to allow for a comparison of the relative expression values obtained for each gene [60]. Primer pairs used to amplify Na V channel transcripts by RTq-PCR. GenBank accession number, sequence, and amplicon size are listed. The primer sequences for the housekeeping gene gusb and gapdh are also indicated.

HPLC Analysis
Chromatographic analyses were performed on a 2695 separation module coupled to a 2695 PDA detector (Waters, Saint Quentin en Yvelines, France), assisted by the Empower software (Waters). The IA were dissolved in MeOH at 1 mg/mL and centrifuged for 6 min at 13,000 rpm. The injection volume was 5 µL. Samples were injected into a Lichrospher 100 RP18 column (125 mm × 4 mm, 5 µm, Merck) and were separated by a gradient elution with buffer A (50 mM KH 2 PO 4 , 2 mM heptane sulfonic acid, 0.1% diethylamine, pH = 3) and the MeCN (B) solvent system (T 0 5% B; T 25 100% B and T 30 100% B) [61]. The flow rate was 1 mL/min, with UV detection at 210 and 280 nm, respectively.

Automated Patch-Clamp Experiments
Whole-cell recordings were used to investigate the effects of oxostephanine, liriodenine, thalmiculine, bebeerine, and protopine on HEK293 cells stably expressing the hNav1.2, hNav1.3, and hNav1.6 channels. Automated patch-clamp recordings were performed using the SyncroPatch 384PE from Nanion (München, Germany). Chips with a single hole (n = 384, series resistance 3.93 ± 1.19 MΩ) were used for HEK293 cell recordings. Pulse generation and data collection were performed with the PatchControl384 v1.9.7 software (Nanion) and the Biomek interface (Beckman Coulter). Whole-cell recordings were conducted according to the recommended procedures of Nanion. The cells were stored in a cell hotel reservoir at 10 • C, with a shaking speed of 60 RPM. After initiating the experiment, cell catch, sealing, whole-cell formation, liquid application, recording, and data acquisition were all performed sequentially and automatically. The intracellular solution contained (in mM): 10 CsCl, 110 CsF, 10 NaCl, 10 EGTA, and 10 HEPES (pH 7.2, osmolarity 280 mOsm), and the extracellular solution contained (in mM): 60 NMDG, 80 NaCl, 4 KCl, 2 CaCl 2 , 1 MgCl 2 , 5 Glucose, and 10 HEPES (pH 7.4 (NaOH), osmolarity 280 ± 3 mOsm). Wholecell experiments were performed at a holding potential of −100 mV at room temperature (18-22 • C). Currents were sampled at 20 kHz. For pharmacological assays, alkaloid solutions were prepared at 30 µM in the extracellular solution supplemented with 0.3% bovine serum albumin and distributed in 384-well compound plates according to a pre-established plate plan. This distribution within compound plates could not be randomized since it would then excessively complexify the methods of analyses. The working compound solution was diluted 3 times in the patch-clamp recording well by adding from 30 to 60 µL of the external solution to reach the final reported concentration and the test volume of 90 µL. The effects of IA (10 µM) or of BTX (1 µM) were measured at the end of a 10 min application time. Activation curves were built by depolarizations lasting for 2000 ms, from −130 mV to 40 mV, using 5 mV increment steps. Pulses were applied every 5 s. The use dependency of oxostephanine and the liriodenine effects on hNa V 1.3 with or without the presence of 1 µM BTX were investigated by depolarizing the cells to 10 mV at 1 Hz for 300 s.

Data Bank Search
Structural analogs of oxostephanine and liriodenine were searched for using the search tool from ChemIDplus (https://chem.nlm.nih.gov/chemidplus/ProxyServlet, June 2021). Analogs sharing less than 94% were ignored.

Data Analysis
Fluorescence data analysis was performed using the SoftMax Pro 5.4.1 software (Molecular Devices, Sunnyvale, CA, USA).
With data from VSP assays, calculations made for histograms were performed as follows. For t = 1 s up to 30 s, for each condition, the VSP signal ratio of the GH3b6 cells was measured after 30 min of incubation with 1 µM BTX and an individual alkaloid. The first 30 s showed signal at equilibrium before the addition of the depolarizing solution, resulting in a final Na + concentration of 80 mM. Each curve point is the CC2DMPE signal over the DisBAC(2,3) ratio, as a ratio relative to the response before the addition of the depolarizing solution (calculated by averaging the first 4 time points, T1-4, from each replicate). The baseline of this ratio was shifted to 0 by subtracting the average of the 4 first time points before adding the depolarizing solution. Each value was then expressed as a percentage of the positive control final ratio obtained in each plate/experiment with 1 µM BTX. This allowed us to compare data from different microplates. The equation used was the following: The kinetic traces of the ANG-2 fluorescence were plotted as an emission ratio: λexcitation 517 nm/λexcitation 540 nm. The AUC of the kinetic traces were used to determine the response after BTX injection, alone (1 µM) or with each IA (10 µM).
All graphs and statistical analyses were completed using the GraphPad Prism 7.02 (La Jolla, CA, USA). Data are presented as mean ± SEM, calculated from 3 replicates and representative of 3 independent experiments. Non-linear analysis was used to fit the concentration-response data with the Langmuir-Hill equation with a variable slope. Multiple groups were compared by using a one-way analysis of variance (ANOVA), followed by a Bonferroni post hoc test, when appropriate. Differences between independent groups were assessed by using parametric or non-parametric t-tests (Mann and Whitney) when appropriate. Differences with p < 0.05 were considered significant (* for p < 0.05, ** for p < 0.01, *** for p < 0.001, **** for p < 0.0001).

Conclusions
In conclusion, the VSP screening of the in-house alkaloid library allowed for the identification of five IA-oxostephanine, liriodenine, thalmiculine, bebeerine, and protopine-as inhibitors of BTX-activated Na V channels. However, only bebeerine was validated as a potent blocker of Na V channels by automated patch-clamp experiments, with the following rank order: hNa V 1.6 > hNa V 1.3 > hNa V 1.2. In fact, oxostephanine and liriodenine were able to decrease the BTX-induced alteration of both the activation and inactivation properties of the hNa V 1.3 channel. Our data highlight that the use of the Na V channel activator, which binds within the pore, could lead to the selection of low-potency Na V channel inhibitors, and that electrophysiology is unavoidable for their characterization. Finally, bebeerine is the first BBIQ exhibiting potent inhibitory effects on the hNa V 1.3 and hNa V 1.6 channels.