Quaternary Lidocaine Derivative QX-314 Activates and Permeates Human TRPV1 and TRPA1 to Produce Inhibition of Sodium Channels and Cytotoxicity

LOCAL anesthetics are unselective inhibitors of voltage-gated sodium channels and accordingly they induce an unselective block of sensory and motor neurons.¹  Binshtok et al.²  demonstrated that the quaternary lidocaine derivative QX-314 induces a preferential nociceptive block, presumably by permeating through the channel pore of the capsaicin receptor transient receptor potential vanilloid 1 (TRPV1). This effect of QX-314 was reproduced in preclinical studies in rodents,^3–8  and more recent studies indicate that QX-314 can also permeate the pore of transient receptor potential cation channel, subfamily A, member 1 (TRPA1).^3,9,10  Other reports document that QX-314 also induces a strong and long-lasting motor block,¹¹  that is, an inhibition of nerve fibers lacking TRPV1 and TRPA1. Intriguingly, it was recently demonstrated that QX-314 can also block myelinated A-fibers by permeating through the Toll-like receptor 5.¹²  Nevertheless, a possible approach to establish a preferential and long-lasting sensory nerve block is to guide poorly membrane-permeable sodium channel blockers through the pore of TRPV1 and TRPA1 mainly expressed in nociceptive sensory neurons. To our knowledge, permeation of QX-314 has not yet been demonstrated for human (h) TRPV1 or TRPA1. Several studies have demonstrated that the human isoforms of both TRPV1 and TRPA1 display pronounced differences in terms of function and pharmacology compared with the rodent isoforms.^13–19  If the concept of the permeation of QX-314 through TRP channels is to be translated into clinical practice, it seems mandatory to define the effects of QX-314 on hTRPV1 and hTRPA1.

Perisciatic injection of QX-314 in combination with capsaicin resulted in profound neurotoxicity associated with nerve loss.⁵  Moreover, intrathecal injection of QX-314 in mice results in severe irritation or even death.²⁰  Considering that a severe neurotoxic potential of QX-314 would prevent an introduction of this approach into clinical practice, defining the mechanisms mediating a QX-314–induced neurotoxicity seems indispensable. One possibility is that activation of TRPV1 and TRPA1 might trigger a calcium-dependent apoptosis or necrosis leading to cell death.^21,22  Recent studies demonstrated that local anesthetics induce calcium influx in sensory neurons by activating TRPV1 and TRPA1.^19,23–25  However, the jury is still out on whether QX-314 activates TRPV1. Although some reports find that QX-314 fails to activate rodent isoforms of both TRPV1 and TRPA1 due to its poor membrane permeability, a more recent study demonstrated that rat TRPV1 (rTRPV1) can generate membrane currents when challenged by QX-314 at millimolar concentrations.^7,19,25  It is therefore possible that human TRPV1 and TRPA1 are involved in QX-314–induced neurotoxicity either by channel activation or by entry of QX-314 through the channel pore.

In the current study, we aimed to investigate the interaction of QX-314 with hTRPV1 and hTRPA1 with regard to permeation with subsequent inhibition of sodium channels as well as a possible direct channel activation mediating calcium influx and possibly cytotoxicity. We used human embryonic kidney 293 (HEK-293) cells with stable or transient expression of either hTRPV1 or hTRPA1 together with the sodium channel α-subunit Nav1.7. The effects of QX-314 on these ion channels were examined with whole cell patch clamp as well as ratiometric calcium imaging recordings. Cell viability was examined by flow cytometry after double staining with propidium iodide (PI) and annexin V.

QX-314 (lidocaine-N-ethyl chloride) and lidocaine (both from Sigma-Aldrich, Germany) were directly dissolved in water or cell culture medium depending on experiments. 4-(3-Chloro-2-pyridinyl)-N-[4-(1,1-dimethylethyl)phenyl]-1-piperazinecarboxamide (BCTC), a TRPV1 antagonist, HC-030031, a TRPA1 antagonist (both from Tocris, United Kingdom), and the TRPA1 agonists carvacrol and acrolein (both from Sigma-Aldrich) were dissolved in dimethylsulfoxide (DMSO). The TRPV1 agonist capsaicin (Biotrend, Germany) was dissolved in ethanol. All stock solutions were stored at −20°C and diluted into appropriate test concentrations in extracellular buffer solution or cell culture medium before the experiments.

Stable hTRPV1-HEK-293 cells (a kind gift from Dr. Peter Zygmunt, Lund, Sweden), hTRPA1-HEK-293 cells, and HEK-293t cells were cultured in standard Dulbecco’s modified Eagle medium (D-MEM, Gibco, BRL Life Technologies, Germany) with 10% fetal bovine serum (Biochrom, Germany), 100 U/ml penicillin, 100 μg/ml streptomycin (Gibco), and 2 mM Glutamax (Gibco). Five micrograms per milliliter of blasticidin (PAA, Austria) and 0.35% Zeocin (Invitrogen, France) were added for stable expression of hTRPV1, and for induction, tetracycline (Sigma-Aldrich) 0.1 μg/ml was added to the medium 16 to 24 h before the experiments. A stable hTRPA1-HEK-293 cell line was established by using G418 800 μg/ml (both from Sigma-Aldrich) as described previously.¹⁸  Cells were cultivated at 37°C and 5% CO₂ in Petri dishes (Greiner Bio-One, Germany) or 12-well plates (Thermo Fisher Scientific, Germany) 12 to 24 h before the experiments. For transient transfection of HEK-293t cells or the stable cell lines, complementary DNA of rTRPV1 (a kind gift from Dr. David Julius, San Francisco, California), mouse TRPA1 (a kind gift from Dr. Ardem Patapoutian, San Diego, California), rabbit TRPV1 (oTRPV1) and the o/rTRPV1-chimera (kind gifts from Dr. Narender Gavva, Amgen Inc., Thousand Oaks, California), hTRPA1-C621S/C641S/C665S (hTRPA1–3C, kindly provided by Dr. Sven-Eric Jordt, New Haven, Connecticut), or human Nav1.7 (hNav1.7, kindly provided by Dr. Angelika Lampert, Aachen, Germany) was cotransfected with 0.5 μg plasma for enhanced green fluorescent protein (Clontech, USA) by using a nanofectin transfection kit according to the instructions of the manufacturer (PAA). All experiments were performed in accordance with the requirements of the local authorities (Gewerbeaufsicht, Niedersachsen, Germany).

Whole cell patch clamp experiments were carried out with a HEKA Electronics USB 10 amplifier and the Patchmaster software (HEKA Electronics, Germany). Borosilicate pipettes with a resistance of 1 to 2 MΩ were filled with a pipette solution containing 140 mM KCl, 2 mM MgCl₂, 5 mM EGTA, and 10 mM HEPES with pH adjusted to 7.4 by KOH. Standard extracellular solution contained 140 mM NaCl, 5 mM KCl, 2 mM MgCl₂, 1.2 mM CaCl₂, 10 mM HEPES, and 10 mM glucose with pH 7.4 adjusted by tetramethylammonium hydroxide (TMA-OH). Calcium-free extracellular solution contained 140 mM NaCl, 5 mM KCl, 2 mM MgCl₂, 5 mM EGTA, 10 mM HEPES, and 10 mM glucose with pH adjusted to 7.4 by TMA-OH. Transfected cells were visualized by enhanced green fluorescent protein fluorescence before the recording. Only one cell per dish was used for experiments. For sodium currents, data were sampled at 25 kHz and filtered at 5 kHz. The series resistance was compensated by 60 to 80%, and the capacitance artifacts were cancelled using the Patchmaster software. For experiments on tonic block of sodium currents, linear leak subtraction based on resistance estimates from four hyperpolarizing pulses applied after depolarizing test potential was used. For TRPV1- and TRPA1-mediated membrane currents, data were sampled at 10 kHz and filtered at 2 kHz. All substances were applied by a gravity-driven perfusion system allowing a focal application of test solution within a distance of less than 100 μm from the cell. The Fitmaster software (HEKA Electronics) and Origin 8.5.1 software (Origin Lab, USA) were used for data analysis. The availability curves were generated from a double-pulse protocol described in detail in figure 1. Data were fitted by the Boltzmann equation (y = 1/{1 + exp (EPP − 0.5)/k_h}), where EPP is a conditioning pulse potential, 0.5 is the voltage at which y = 0.5, and k_h is the slope factor. To obtain the IC₅₀ value for tonic block by lidocaine, peak current amplitudes at different drug concentrations were normalized to the value obtained in control solution. Data were fitted with the Hill equation y = y_max × {(IV₅₀n / IC₅₀n × Cn)}, where y_max is the maximal amplitude, IC₅₀ is the concentration at which y/y_max = 0.5, and n is the Hill coefficient.

Cells were filled by 4 μM fura-2-acetoxylmethyl ester (fura-2-AM) and 0.02% pluronic for approximately 45 min. After washout to allow fura-2-AM deesterification, coverslips were mounted on an inverse microscope with a 20X objective (Axio Observer D1; Zeiss, Germany). Cells were constantly superfused with extracellular solution containing 145 mM NaCl, 5 mM KCl, 1.25 mM CaCl₂, 1 mM MgCl₂, 10 mM glucose, and 10 mM HEPES by using a software-controlled, seven-channel, gravity-driven, common-outlet superfusion system. Fura-2 was excited using a microscope light source and an LEP filter wheel (Ludl Electronic Products Ltd., USA) to switch between 340- and 380-nm wavelength. Images were exposed for 20 and 40 ms, respectively, and acquired at a rate of 1 Hz with a CCD camera (Cool SNAP EZ; Photometrics, USA). Data were recorded using VisiView 2.1.1 software (Visitron Systems GmbH, Germany). Background fluorescence was subtracted before calculation of ratios. QX-314 was applied in concentrations of 5 and 30 mM for 40 s. To identify TRPV1- or TRPA1-expressing cells on a functional level, 0.3 μM capsaicin (10 s) and 100 μM acrolein (20 s) stimuli were used as positive control, and 5 μM ionomycin was applied as a control at the end of each experiment. Means (± SEM) of area under the curve (AUC) (delta ratio F340/380 nm) for regions of interest are presented.

Cell viability was assessed by double staining with PI and fluorescein isothiocyanate (FITC) annexin V according to the protocol of the manufacturer (FITC Annexin V Apoptosis Detection Kit 1; BD Pharmingen, Germany) and as described before.²⁶  Cells were cultured in 12-well plates (Thermo Fisher Scientific) and exposed to tested substances for 15 min when reaching 80% confluency. Twenty-four hours after incubation, cells were detached using phosphate-buffered saline (PBS) and washed together with the corresponding supernatant with PBS. Cell pellets were resuspended in annexin V binding buffer (1X), and cell count was adjusted to 1 × 10⁶/ml. One hundred microliters of the cell suspension was transferred into 5-ml polystyrene tubes (BD Pharmingen). Staining was performed by adding 5 μl PI and 5 μl annexin V. After incubation for 15 min at room temperature in the dark, 400 μl binding buffer (BD Pharmingen) was added. Specimens were analyzed within 1 h after staining by using a Cytomics FC500 flow cytometer (Beckmann Coulter, Germany). A total of 10,000 cells per sample were measured. Based on previous investigations, experiments were repeated three times. Data were analyzed with FlowJo version 10 (FlowJo LLC, USA).

All data are presented as mean ± SD. Statistical analysis was performed by using GraphPad Prism 5 (GraphPad Software Inc., USA). Comparisons between two groups with different sample sizes were performed by nonparametric, two-tailed testing with the Mann–Whitney U test and between more than two groups with the Kruskal–Wallis test with Dunn post hoc correction. Multiple comparisons for groups with equal, balanced sample sizes were performed by ANOVA with a Tukey post hoc test. Significance was assumed for P value of 0.05 or less. Previous studies were used for estimation of sample sizes, that is, no formal a priori power analysis and sample size projection were performed.

To examine the effect of externally applied QX-314 on sodium current, we used HEK-293t cells expressing hNav1.7 and hTRPV1 channels. In order to examine tonic block, we evoked sodium currents by 40 depolarizing pulses from the holding potential of −70 mV to −10 mV at 0.1 Hz (fig. 1A). At this holding potential, however, we detected a strong rundown of the peak current amplitudes even in control solution (45 ± 28%, n = 23; fig. 1B). When the same protocol was used in cells held at −120 mV, the current amplitude even slightly increased over time (6 ± 10%; n = 25; fig. 1B). In order to find the optimal holding potential for the experiment, we further examined the voltage-dependent fast inactivation. Fast inactivation was induced by 100-ms-long prepulses ranging from −150 to −10 mV applied in steps of 10 mV (fig. 1C). The available current amplitude was then determined by a test pulse to −10 mV, normalized to the amplitude determined at −150 mV, and plotted against the corresponding voltage. Although a considerable fraction of channels are already inactivated at −70 mV, 100% are available at −120 mV (fig. 1C; see dotted lines, midpoint of steady-state inactivation: −79 ± 6 mV, n = 8). We therefore proceeded with protocols by which the sodium currents were monitored in cells held at −120 mV. Previous reports showed a prominent reduction of total sodium currents in rodent dorsal root ganglion cells after application of 5 mM QX-314 or higher.^2,8  As is demonstrated by the cartoon in figure 1D, cells expressing either hTRPV1 or hTRPA1 were treated with the following test solutions for 60 s: (1) 1 μM capsaicin or 200 μM carvacrol alone, (2) capsaicin (fig. 1E) or carvacrol (fig. 1F) in combination with 5 mM QX-314, (3) capsaicin or carvacrol in combination with 30 mM QX-314, or (4) 30 mM QX-314 alone. After application of any combination of substances, peak current amplitudes were reduced in both hTRPV1- and hTRPA1-expressing cells (fig. 1G). When compared with the reduction of the current amplitude by 1 μM capsaicin alone (20 ± 14%, n = 10), the coapplication of neither 5 mM QX-314 (35 ± 24%, n = 16) nor 30 mM QX-314 (23 ± 18%, n = 26, both P > 0.05, Kruskal–Wallis with post hoc Dunn test) resulted in a stronger inhibition in cells expressing hTRPV1. Furthermore, the effect of 30 mM QX-314 alone (34 ± 16%, n = 6) did not significantly differ from the effect induced by 30 mM QX-314 in combination with capsaicin (P > 0.05, Kruskal–Wallis with post hoc Dunn test). In hTRPA1-expressing cells, 200 μM carvacrol (38 ± 25%, n = 10) alone induced a similar reduction of sodium currents compared with the coapplication of carvacrol with QX-314 at both 5 mM (43 ± 21%, n = 20) and 30 mM QX-314 (56 ± 16%, n = 10) (both P > 0.05, Kruskal–Wallis with post hoc Dunn test). Again, the effect of 30 mM QX-314 alone (35 ± 15%, n = 8) did not significantly differ from the effect induced by 30 mM QX-314 in combination with carvacrol (P > 0.05, Kruskal–Wallis with post hoc Dunn test). These data suggest that application of QX-314 does not produce a robust additional blockade of sodium currents compared with application of capsaicin respectively carvacrol alone. It is possible that the amount of cytosolic QX-314 accumulation during 60 s was not enough to induce a robust tonic inhibition. To examine this notion, we determined tonic block of resting Nav1.7 channels (i.e., holding potential of −120 mV) by lidocaine at concentrations from 10 to 30,000 μM (fig. 1, H and I). The calculated (see Hill equation in Materials and Methods) IC₅₀ value for this tonic block was 1,710 ± 242 μM (n = 6 to 11 for each concentration). Assuming that cytosolic QX-314 and lidocaine should be more or less equipotent, this result indeed suggests that very high concentrations of QX-314 need to accumulate in the cytosol in order to induce a substantial inhibition of resting sodium channels. It was previously demonstrated that longer application times of QX-314 result in a stronger inhibition of sodium currents.²  However, when we extended the application time of QX-314 in combination with capsaicin or carvacrol to 3 min or longer, the recordings became progressively instable such that a reliable estimation of sodium current inhibition by QX-314 as opposed to an unspecific rundown seemed inadequate (data not shown). Moreover, as is also demonstrated in figure 1, E and F, inward currents induced by both capsaicin (fig. 1E) and carvacrol (fig. 1F) displayed an almost complete tachyphylaxis within 60 s. When considering the hypothesis that QX-314 needs to permeate through the channel pores of TRPV1 and TRPA1 in order to inhibit sodium channels, it seems difficult to interpret any effects occurring after longer application times during which both TRPV1 and TRPA1 are inactivated.

As an alternative approach, we next examined a QX-314–mediated use-dependent block of sodium currents with or without coapplication of capsaicin or carvacrol (fig. 2). External application of 100 μM (6 ± 4%, n = 5) or 5 mM QX-314 (4 ± 3%, n = 7) did not result in a significant use-dependent block of Nav1.7 currents evoked by 100 depolarizing pulses from −120 to −10 mV at 10 Hz compared with control solution (5 ± 5%, n = 5; P > 0.05, Kruskal–Wallis with post hoc Dunn test; fig. 2C). When 100 μM QX-314 was added to the pipette solution, however, it induced a strong use-dependent block on currents activated at 10 Hz (54 ± 14%, n = 10; P < 0.001, Kruskal–Wallis with post hoc Dunn test; fig. 2, B and C). Thus, comparably low concentrations of intracellular QX-314 induce a pronounced use-dependent block of Nav1.7. We further investigated whether externally applied 5 mM QX-314 induces use-dependent block when hTRPV1 or hTRPA1 is activated by capsaicin or carvacrol. Application of 1 μM capsaicin alone did not result in a prominent use-dependent inhibition of sodium currents (7 ± 5%; n = 6; fig. 2, D and E). However, coapplication of capsaicin and 5 mM QX-314 resulted in a significant use-dependent block (34 ± 18%, n = 13; P < 0.05, Kruskal–Wallis with post hoc Dunn test; fig. 2, D and E). Similarly, 200 μM carvacrol in combination with 5 mM QX-314 produced a significantly stronger use-dependent block of sodium currents in cells expressing hTRPA1 compared with carvacrol alone (32 ± 8%, n = 8; carvacrol alone 6 ± 3%, n = 14; P < 0.01, Kruskal–Wallis with post hoc Dunn test; fig. 2, F and G). These data indicate that QX-314 can indeed permeate activated hTRPV1 and hTRPA1 channels to inhibit Nav1.7. In order to substantiate this finding, we next examined whether this use-dependent block is concentration dependent. Although application of 1 mM QX-314 together with capsaicin did not result in a significant use-dependent block (5 ± 11%, n = 7; P > 0.05, Kruskal–Wallis with post hoc Dunn test; fig. 2E), 30 mM QX-314 induced a strong use-dependent block (51 ± 30%, n = 7; P < 0.01, Kruskal–Wallis with post hoc Dunn test). A similar result was obtained on cells expressing hTRPA1, that is, 1 mM QX-314 (7 ± 17%, n = 6; P > 0.05, Kruskal–Wallis with post hoc Dunn test) failed to induce use-dependent block, whereas 30 mM QX-314 (45 ± 19%, n = 6; P < 0.001, Kruskal–Wallis with post hoc Dunn test; fig. 2G) produced a highly significant inhibition.

We finally explored whether permeation of QX-314 really requires preactivation of hTRPV1 or hTRPA1 channels by potent agonists. In cells expressing hTRPA1, application of only 5 mM QX-314 indeed produced a significant use-dependent block (40 ± 24%, n = 11; P < 0.05, Kruskal–Wallis with post hoc Dunn test; fig. 2G). In cells expressing hTRPV1, however, the weak use-dependent block observed after application of 5 mM QX-314 alone did not reach statistical significance (20 ± 25%, n = 9; P > 0.05, Kruskal–Wallis with post hoc Dunn test; fig. 2E).

Our data suggest that, compared with rodent channels,^19,25  QX-314 might directly permeate or even gate human TRPV1 and TRPA1 channels. To examine this notion, we performed ratiometric calcium imaging and patch clamp recordings in hTRPV1- (fig. 3) and hTRPA1-expressing (fig. 4) cells. Indeed, a 40-s-long application of both 5 (fig. 3A) and 30 mM (fig. 3B) QX-314 induced an increase in intracellular calcium in cells expressing hTRPV1. In HEK-293 cells lacking TRPV1, however, we did not observe any effects by 5 or 30 mM QX-314 (fig. 3A, data shown only for 30 mM; n = 195). A lower percentage of cells expressing hTRPV1 responded to 30 mM QX-314 compared with 5 mM (5 mM: 100 ± 0.7% vs. 30 mM: 58 ± 27%; P ≤ 0.006, Mann–Whitney U test, n = 7 to 13; fig. 3C), and the response to 30 mM QX-314 was significantly lower than to 5 mM QX-314 (AUC, 5 mM: 0.57 ± 0.18 vs. 30 mM: 0.17 ± 0.19; P < 0.001, ANOVA with Tukey post hoc test, n = 886 [5 mM] and 1,030 [30 mM], respectively; fig. 3D).

To confirm that the increase in intracellular calcium is due to activation of TRPV1 channels, we next monitored membrane currents during a 500-ms voltage ramp ranging from −100 to +100 mV (fig. 3E). QX-314 at 5 mM evoked slowly developing (more than 1 min) membrane currents with a strong outward rectification in all cells containing hTRPV1 (n = 11; fig. 3E), that is, inward currents at −100 mV (370 ± 616 pA) were small compared with outward currents at 100 mV (3,246 ± 3,004 pA; n = 11; fig. 3, E and F). Coapplication with the TRPV1 antagonist BCTC (100 nM) resulted in a complete inhibition of QX-314–induced currents (311 ± 223 pA at 100 mV; n = 8; fig. 3, E and F). QX-314 at 5 mM also activated rTRPV1 to a similar extent as hTRPV1 (inward: 132 ± 67 pA; outward: 3,265 ± 2,317 pA; n = 6; fig. 3, G and H). In a previous study, we demonstrated that lidocaine activates TRPV1 partly via interaction with the intracellular vanilloid-binding domain.¹⁶  Therefore, we investigated whether this is also the case for QX-314. To that end, we explored the effects of QX-314 on the capsaicin-insensitive oTRPV1 that exhibits a very low vanilloid sensitivity.¹⁶  QX-314 completely failed to induce membrane currents in cells expressing oTRPV1 (fig. 3, I and J; n = 8), such that both inward currents at −100 mV and outward currents at +100 mV did not increase after application of QX-314 (inward: 62 ± 75 pA; outward: 280 ± 225 pA) compared with control solution (inward: 64 ± 66 pA; outward: 276 ± 222 pA). These data suggest that QX-314 activates TRPV1 at least partly via interaction with the vanilloid-binding domain. We confirmed this notion by using a chimeric o/rTRPV1 construct in which the vanilloid-binding domain of oTRPV1 was replaced by the corresponding domain from rTRPV1. Indeed, 5 mM QX-314 induced prominent currents in this mutant (fig. 3, K and L, n = 7; inward: 1,364 ± 786 pA; outward: 8,293 ± 3,858 pA).

We further examined whether QX-314 also activates TRPA1. Acute application of 5 (fig. 4A) and 30 mM (fig. 4B) QX-314 for 40 s produced a significant increase of intracellular calcium in hTRPA1-expressing cells. Different from hTRPV1, 5 mM QX-314 evoked significantly smaller responses compared with 30 mM (AUC 5 mM: 0.13 ± 0.14; AUC 30 mM: 0.29 ± 0.16, P < 0.001, n = 317 [5 mM] and 399 [30 mM], respectively; fig. 4C). Furthermore, the percentage of responding cells were higher when 30 mM of QX-314 was applied (5 mM: 62 ± 14%; 30 mM: 89 ± 10%, P ≤ 0.006, Mann–Whitney U test, n = 7 to 13 experiments; fig. 4D). QX-314 at 5 mM also induced membrane current in cells expressing hTRPA1 (inward: 78 ± 58 pA; outward: 451 ± 192 pA; n = 10), and this effect was effectively inhibited by 50 μM of the TRPA1 antagonist HC-030031 (inward: 61 ± 70 pA; outward: 192 ± 186 pA; n = 6; fig. 4, E and F). Furthermore, we observed a concentration-dependent activation of mouse TRPA1 by 5 (inward: 69 ± 26 pA; outward: 544 ± 221 pA; n = 10) and 30 mM (inward: 100 ± 39 pA; outward: 910 ± 382 pA) QX-314 (fig. 4, G and H; n = 7). In analogy to a specific binding site of QX-314 in TRPV1, we hypothesized that QX-314 might activate TRPA1 via modification of intracellular N-terminal cysteins whose modulation is crucial for activation of TRPA1 by several reactive agonists including lidocaine.¹⁹  However, the triple mutant hTRPA1-C621S/C641S/C665S generated robust QX-314–induced membrane currents both at 5 and 30 mM, suggesting that N-terminal cysteins are not the site of action of QX-314 in TRPA1 (data not shown).

Our data demonstrated that QX-314 directly activates TRPV1 and TRPA1 channels leading to a subsequent calcium influx (figs. 3A and 4A). It has been demonstrated that TRPV1-mediated calcium influx triggers apoptosis-like cascade leading to cell death.²²  We therefore evaluated the role of both TRP channels for cytotoxicity induced by QX-314 or lidocaine using double staining with annexin V FITC (annexin V) and PI with subsequent flow cytometric analysis (fig. 5). Cells negative for both substances (annexin V and PI) were defined as vital. We incubated the cells for 15 min with different concentrations of QX-314 and lidocaine or 1% PBS (control). After 15 min, substances were removed and cells recovered for 1 h (fig. 5E) or 24 h (fig. 5, A–D) before staining and flowcytometric analysis. Experiments were repeated three times. In hTRPV1-expressing HEK-293 cells, incubation with 5 mM QX-314 for 15 min did not induce a significant cell death, with 91 ± 2% surviving cells compared with control 90 ± 4% (fig. 5A). QX-314 at 30 mM induced a significant reduction of the viable cell fraction to 48 ± 8% (P < 0.001 compared with untreated control, ANOVA with Tukey post hoc test, n = 3). This effect was almost completely inhibited by coapplication of the TRPV1 antagonist BCTC (10 μM; 81 ± 9% viable cells; fig. 5A). QX-314 at 60 mM reduced the viable cell fraction to 31 ± 3%, but this effect was not completely inhibited by BCTC (68 ± 7%, P < 0.01 compared with control, ANOVA with Tukey post hoc test, n = 3). Cell death was not observed with 30 mM QX-314 (83 ± 5% viable cells) in TRPV1-negative HEK-293 cells, but 60 mM QX-314 induced cell death with a viable cell fraction of 70 ± 3% (P < 0.01 compared with control). In hTRPV1-expressing cells, incubation with 10 μM capsaicin induced significant cell death (21 ± 3% viable cells) compared with coincubation with 10 μ M capsaicin and 10 μM BCTC, resulting in significantly less cytotoxicity (77 ± 2% viable cells; P < 0.001, ANOVA with Tukey post hoc, n = 3).

In hTRPA1-expressing HEK-293 cells, neither 5 mM QX-314 nor 50 μM of the potent TRPA1 agonist acrolein induced a significant reduction of the viable cell fraction (fig. 5B). QX-314 at 30 mM reduced the viable cell fraction from 84 ± 1% in control to 69 ± 8%, and 60 mM QX-314 reduced viability to 47 ± 4%. This effect was not inhibited by the TRPA1 antagonist HC-030031 (50 μM). Similar to QX-314, 30 mM lidocaine reduced the amount of viable hTRPV1-expressing cells to 15 ± 11% (fig. 5C). However, this effect was not prevented by coincubation with BCTC (17 ± 16% viable cell fraction; fig. 5C). In hTRPA1-expressing cells, both 5 and 30 mM lidocaine induced significant cell death, 39 ± 10% and 2 ± 1%, respectively, of viable cells (fig. 5D). This effect was not reduced by the TRPA1 antagonist HC-030031 (27 ± 6% and 3 ± 1%, respectively; all P < 0.001 compared with control, ANOVA with Tukey post hoc).

We finally explored whether cell death mediated by hTRPV1 can be detected instantly after induction with QX-314 or lidocaine. Cells were incubated with tested substances for 15 min, and cell death was assessed after 1 h (fig. 5E); 60 mM, but not 30 mM, QX-314 reduced viability to 75 ± 6% (P < 0.05 compared with control). Both 30 and 60 mM lidocaine reduced the viable cell fraction to 11 ± 3% and 2 ± 2%, respectively. These data suggest that QX-314–induced cytotoxicity is in part mediated via TRPV1, but not via TRPA1.

The current study adds new insights on the concept to introduce QX-314 into nociceptors via TRPV1 and TRPA1. QX-314 permeates the human isoforms of TRPV1 and TRPA1 to inhibit sodium channels. However, it also activates TRPV1 and TRPA1 to induce calcium influx, which in case of TRPV1 results in cytotoxicity.

Previous studies suggested that it is possible to use TRPV1 and TRPA1 as shuttles to take QX-314 into the cytosol of nociceptors.²⁷  The permeation of QX-314 through rodent TRPV1 with a subsequent local anesthesia was first demonstrated by Binshtok et al.²  Later studies used various TRPV1 agonists in order to introduce QX-314 through TRPV1.^{2,4,6,8,28,29}  However, other reports demonstrated that QX-314 also induces a TRPV1-independent local anesthesia and even motor block by which TRPV1 agonists were not coadministered.^11,30,31  The ability of QX-314 to induce a motor block suggests that mechanisms other than TRPV1 and TRPA1 are used to induce a nerve conduction block. It is possible not only that extracellular QX-314 blocks voltage-gated ion channels other than sodium channels, but also that membrane proteins other than TRP channels allow entry of QX-314. Indeed, it was recently demonstrated that activation of the Toll-like receptor 5 in A-fibers allows entry of QX-314.¹²  In contrast to TRPV1, comparably little has been published on the interaction of QX-314 with TRPA1. Although Nakagawa and Hiura³²  found that activation of TRPA1 by menthol does not enable permeation by QX-314, more recent studies indicate that activation of TRPA1 by allyl isothiocyanate, cinnamaldehyde, and bupivacaine enable permeation by QX-314.^9,10,33  Furthermore, it has become increasingly clear that several functional and pharmacological properties of TRPA1 and TRPV1 display significant species-specific differences between human and rodent isoforms.^11,12,30,31  As in most cases these differences are encoded by residues within the pore region; it is not clear that rodent and human isoforms share the ability to enable permeation of QX-314.

In contrast to previous studies on cultured sensory neurons, we failed to detect a significant reduction of the total sodium current by the combination of QX-314 and agonists compared with agonists or QX-314 alone. Application of both capsaicin and carvacrol resulted in a strong reduction of sodium currents. Similar effects were demonstrated for both TRPA1 and TRPV1 in cultured sensory neurons,³⁴  and they are probably partly explained by an increased intracellular sodium concentration due to the influx through TRP channels, and a consecutively reduced driving force for sodium through voltage-gated sodium channels.³⁵  However, this possibility does not fully explain why we failed to establish a robust inhibition of sodium currents by introduction of QX-314. Whereas previous studies explored sodium currents in cells at holding potentials close to the physiological resting membrane potential (i.e., −70 to −80 mV),^2,10,29  we performed our experiments at −120 mV. Nav1.7 underwent a considerable rundown when recordings were performed at −70 mV in our experiments, probably due to fast inactivation. Local anesthetics are known to have a low affinity to resting sodium channels,¹  and tonic block of resting Nav1.7 channels revealed an IC₅₀ value of 1.7 mM. Assuming that intracellular lidocaine and QX-314 should be equipotent, this result indicates that a high intracellular concentration of QX-314 would be required to allow detection of tonic inhibition in our experiments. Binshtok et al.⁴  observed a progressive reduction of sodium currents in sensory neurons treated with capsaicin and QX-314 for up to 25 min. Thus, it is possible that the application time (60 s) chosen in our study was too short. However, the quality of our recordings suffered substantially during such prolonged applications of capsaicin and carvacrol. We also observed that both TRPA1 and TRPV1 underwent a strong desensitization during 60 s, an effect that speaks against the notion that longer applications would allow more QX-314 to permeate through the channel pore. Instead, we show here that monitoring of use-dependent block offers a more reliable approach. The addition of 100 μM QX-314 in the pipette solution in cells lacking TRP channels resulted in a stronger use-dependent block compared with the effects induced by externally applied 5 mM QX-314 in cells expressing TRPV1 or TRPA1. Thus, our experimental protocol obviously resulted in an intracellular concentration of less than 100 μM QX-314.

In cells expressing hTRPA1, extracellularly applied QX-314 induced a prominent use-dependent block both with and without coapplication of carvacrol. Although not statistically significant, we also observed a small increase in use-dependent inhibition by application of QX-314 alone in cells expressing hTRPV1. This agonist-independent permeation of QX-314 through hTRPA1 and hTRPV1 was not described in any previous study. However, Puopolo et al.³⁶  observed that QX-314 permeates through TRPV1 without successive dilation of the channel pore. We demonstrate here that both TRPV1 and TRPA1 are activated by externally applied QX-314, that is, QX-314 seems to be able to permeate the channel pores to evoke activation of both TRPV1 and TRPA1. This finding may explain why QX-314 applied alone seems to induce local anesthetic effects as well as irritation in rodents.^11,20,37  However, it is also contradicting our previous studies demonstrating that QX-314 inhibits rather than activates rTRPV1 and hTRPA1.^19,25  Although in these studies QX-314 was only briefly (less than 20 s) applied to the cells, we now find that activation of both TRPV1 and TRPA1 develops slowly and results in rather small inward currents. A more recent study indeed demonstrated that QX-314 can both activate and inhibit rodent TRPV1.^7,38 

Our data indicate that activation of both TRPV1 and TRPA1 by QX-314 does not obey an obvious species specificity between human and rodent isoforms. Similar to what we previously described for lidocaine, experiments on constructs of the capsaicin-insensitive oTRPV1 channel suggest that QX-314 specifically interacts with the intracellularly located vanilloid-binding domain in order to gate TRPV1.²⁵  For TRPA1, however, QX-314 does not seem to interact with the intracellular cysteine residues known to be required for activation of several reactive TRPA1 agonists including lidocaine.¹⁹  Thus, although the mechanism for activation of TRPV1 seems to require a permeation of the channel pore for QX-314 to reach the binding site, we can only assume that this is also the case for TRPA1.

The major finding of our study is that TRPV1 seems to be a relevant mediating mechanism of QX-314–induced cytotoxicity. Activation of TRPV1 by capsaicin and other agonists has been demonstrated to induce cytotoxic effects in neuronal and nonneuronal tissue.^22,39  Although the exact mechanisms of TRPV1-mediated cytotoxicity are unclear, most investigations indicate a calcium-mediated, apoptosis-like cell death.²²  It is also well established that clinically used local anesthetics induce neurotoxic effects.^24,40  Several mechanisms of local anesthetic–induced cytotoxicity have been suggested, including destruction of the outer cell membrane, activation of caspases, and disruption of the mitochondrial membrane.^23,40,41  QX-314 was recently suggested to induce significant neurotoxic effects after coapplication with lidocaine and capsaicin in rats.⁵  In further murine in vivo experiments, very high concentrations of QX-314 (100 mM) induced detectable myotoxicity and soft tissue injury.⁴²  Toxic effects of QX-314 were also found after intrathecal²⁰  and IV injection⁴³  in rat in vivo models. In our study, cytotoxicity of 5 and 30 mM QX-314 was not apparent after a short investigation time of 1 h, which may indicate that activation of TRPV1 induces a regulated, time-dependent cellular death pathway. Although further studies are required to explore whether TRPV1 induces QX-314–induced neurotoxicity in vivo, this finding indicates that the desired permeation of QX-314 through TRPV1 may be directly associated with neurotoxicity. In contrast to QX-314, cytotoxicity of lidocaine was not mediated by TRPV1. Data reporting cytotoxic effects of lidocaine unrelated to cytosolic calcium levels support this notion.⁴⁴  Lidocaine may induce toxicity via TRPV1, but this effect is probably superimposed by other cytotoxic mechanisms described in literature.^40,41  As cytotoxicity of the clinically established local anesthetic lidocaine is considerably stronger and apparently also mediated by TRP channel–independent mechanisms, it may be that the toxic effects of QX-314 observed here are minor clinical relevance. However, unlike the membrane-permeable lidocaine that can readily be cleared from the cytosol via diffusion, it is possible that cytotoxicity induced by QX-314 is enhanced due to the fact that it is likely to be trapped in the cytosol for a longer time than lidocaine. Considering that both TRPV1 and TRPA1 mediate a strong calcium influx, we expected to observe similar cytotoxic mechanisms via TRPA1. Although both QX-314 and lidocaine induced cytotoxic effects in hTRPA1-expressing cells, these effects were not prevented by the TRPA1-blocker HC-030031. Furthermore, activation of hTRPA1 by acrolein induced no cytotoxic effects. These data question the role of intracellular calcium as a key mechanism for cytotoxicity and rather suggest that other mechanisms downstream of TRPV1 are responsible for cytotoxicity. Nevertheless, if TRPV1 proves to mediate a relevant QX-314 neurotoxicity in vivo, our data indicate that an approach by which TRPA1 is selectively targeted may be associated with less or no neurotoxicity.

In summary, our results demonstrate a direct activation of human TRPV1 and TRPA1 by QX-314 with its subsequent permeation into the cytosol. Although this result suggests that QX-314 may be suited for selective sensory nerve blocks in humans, the associated neurotoxicity needs further evaluation.

The authors thank Heike Bürger for excellent technical assistance and Wolfgang Koppert, M.D., for continuous support (both from Department of Anaesthesiology and Intensive Care Medicine, Hannover Medical School, Hannover, Germany).

This study was supported by an Early Career Research Grant by the International Association for the Study of Pain (Washington, D.C.; 2012; to Dr. Leffler) and by the Wilhelm Sander-Stiftung (Munich, Germany; 2014.068.1; to Drs. Kistner and Reeh).

The authors declare no competing interests.