The essential oil of Lippia alba and its components affect Drosophila behavior and synaptic physiology

Plants are estimated to produce more than 200,000 specialized metabolites and many presumably evolved to act as defense mechanisms against predation (Mithofer and Boland, 2012). These metabolites have been found to affect a broad range of organisms, and there is considerable interest in their actions as they represent a source of medicinal and agricultural compounds (Shu, 1998). However, the mechanism of action is known for only a few of these metabolites. The plant Lippia alba, a flowering shrub in the verbena family, is native to southern North America, Central America and South America. Its essential oil (EO) has played a prominent role in Brazilian folk medicine with a wide variety of applications, including its use as an antispasmodic, analgesic, sedative, anxiolytic and antihypertension remedy (Hennebelle et al., 2008). Some of these behavioral effects have been replicated in rodents, where the EO has been reported to act as an analgesic, sedative, anticonvulsant and anxiolytic (Viana et al., 1998, 2000; Vale et al., 1999; Hatano et al., 2012). In aquaculture, the EO has been used to anesthetize fish and reduce their mortality during transport (da Cunha et al., 2010). The L. alba EO has also been tested as an insecticide and found to be toxic to mosquito larvae, and both larval and adult cattle ticks (Vera et al., 2014; Peixoto et al., 2015b). Physiologically, the EO was shown to produce vasorelaxation and decreased nerve membrane excitability in rodents (Maynard et al., 2011; Sousa et al., 2015).

EOs derived from L. alba show great chemical diversity, which is influenced by the plant's origin, its stage of development and the plant part used for extraction (Zoghbi et al., 1998). A number of chemotypes have been characterized based upon their major components; the most prominent components include carvone, limonene and citral (Vale et al., 1999; Hennebelle et al., 2008; Matos et al., 1996). These monoterpenes have been shown to produce sedation or lethality in a broad range of animals: prolonged exposure to citral, carvone or limonene produced lethality in houseflies (Kumar et al., 2014; Rice and Coats, 1994), both citral and limonene were lethal to mosquitoes (Giatropoulos et al., 2012), and citral produced sedation in snails and mice (Price and Berry, 2008; Vale et al., 2002). The physiological actions of citral include altering neuronal firing in the cockroach (Price and Berry, 2006) and reducing nerve membrane excitability and smooth muscle contractions in the rat (Sousa et al., 2015; Sadraei et al., 2003).

Many plant products affect animal behavior by targeting synapses; e.g. nicotine, opiates, tetrahydrocannabinol, cocaine and curare (Pichersky and Lewinsohn, 2011). However, the effect of the EO or monoterpenes on synaptic transmission has not been directly examined. Drosophila melanogaster has proven a valuable model organism that allows combined genetic, behavioral and physiological approaches. It is a particularly good system for synaptic studies due to the accessible and identified neuromuscular synapses found in the larvae. Using larval and adult Drosophila, we examined the behavioral and physiological effects of L. alba obtained from the Amazon region of Brazil. For both the EO and its components, we tested whether they produced sedation and influenced synaptic transmission. Our results showed that the EO produced immobilization of Drosophila and blocked synaptic transmission mainly due to the actions of citral.

Experiments were performed on Canton-S (CS) Drosophila melanogaster, Meigen 1830. For behavioral tests, we used adults less than 7 days old. For physiological studies, experiments were performed in segments 3 or 4 of wandering 3rd-instar larvae. Larvae were pinned out in a physiology chamber and, after an incision through the dorsal body wall, the internal organs were removed to expose the body-wall muscles. Then, the segmental nerves were cut and the brain was removed. All experiments were performed at room temperature.

The aerial part (leaves and fin branches) of Lippia alba (Mill) N.E. Brown was collected from the garden of Local Productive Arrangements of Medicinal Plants and Phytotherapy at Santarém, Pará, Brazil (APL-FITOS), Comunidade de Ponta de Pedras (S=02°26′35.7″ and W=054°54′54.2″). Plant material was identified in the herbarium of Universidade Federal de Juiz de Fora, Minas Gerais state, Brazil, voucher number CESJ 65276. The EO extraction was performed at the Bioprospecting and Experimental Biology Laboratory of Universidade Federal do Oeste do Pará (LabBBEx-UFOPA), Brazil by hydro-distillation, using the Clevenger apparatus (Labor Quimi Vidrolabor, Poa-SP, Brazil) for 2 h. To remove residual water, the EO was centrifuged at 1560 g with anhydrous sulfate, stored in glass bottles and kept under refrigeration at 5°C. The EO used in this study was from a single extraction. The EO chemical analysis was performed with a GCMS-QP2010 Ultra (Shimadzu Corporation, Tokyo, Japan), equipped with AOC-20i auto-injector and using the GCMS-solution software including the Wiley, NIST and FFNSC2 libraries. We used an Rxi-5 ms (30 m×0.25 mm; 0.25 μm film thickness) silica capillary column (Restek Corporation, Bellefonte, PA, USA). The analysis was performed under the following conditions: 250°C injector temperature; oven temperature program was 60–240°C (3°C min⁻¹); carrier gas was helium adjusted to a linear velocity of 36.5 cm s⁻¹ (rate of 1.0 ml min⁻¹); injection was in the split mode of 1 μl sample (5 μl of EO per 500 μl hexane); the split ratio was 1:20; ionization was by electronic impact at 70 eV; and ionization source and transfer line temperatures were 200 and 250°C, respectively. The mass spectra were obtained by automatic scanning every 0.3 s, with mass fragments in the range of 35–400 m z⁻¹. The retention index was calculated for all volatile components using a homologous series of C8-C20 n-alkanes (Sigma-Aldrich, St Louis, MO, USA), according to the linear equation of Vandendool and Kratz (1963). The constituents were identified by comparing their retention indices and mass spectra (molecular mass and fragmentation pattern) with those existing in the GCMS-Solution system libraries, and also with spectra from the literature (Adams, 2007).

We determined whether flies were sedated by the EO or its components: citral, S-(+)-carvone, S-(+)-limonene and β-caryophyllene (Sigma-Aldrich). Ten flies were placed in a 50 ml centrifuge tube and left to acclimate. Then, the EO or its components were applied to filter paper just beneath the lid. Every 30 min for 150 min, the tubes were tapped to knock the flies to the bottom and we determined the number of immobilized flies, i.e. those flies unable to crawl back up the tube wall. Normally, flies show a strong positive geotaxis and rapidly crawl upward after falling to the bottom. After the 150 min exposure, flies were transferred to a separate tube and their recovery was examined at 1, 3 and 6 h post-exposure. Fumigation studies of EOs and their components on insects have typically applied increasing small volumes of the liquid and expressed the dose as liquid volume per container air volume (Phillips and Appel, 2010). In our experiments, we applied 0.5–10 µl volumes to the centrifuge tube, which had a total volume of 55 ml. Greater EO volumes produced a stronger behavioral effect and, since the equilibrium vapor concentration is independent of liquid volume, this was presumably due to a more rapid increase in vapor concentration for large volumes, i.e. the evaporation rate is dependent on both liquid surface area and vapor pressure (Zemansky and Dittman, 1997). In some experiments, we compared the behavioral effects of the EO components after equalizing their vapor pressures. In solution, the partial vapor pressure of a component is equal to its vapor pressure multiplied by its mole fraction (MF), according to Raoult's law (Zemansky and Dittman, 1997). Thus, the vapor pressures for citral (0.09 mmHg), carvone (0.12 mmHg) and limonene (1.98 mmHg) (https://pubchem.ncbi.nlm.nih.gov/) were matched by using the mixtures: carvone/propylene glycol (9.00/1.00) and limonene/propylene glycol (0.95/9.05).

Excitatory postsynaptic potentials (EPSPs) and miniature EPSPs (minEPSPs) were recorded from larval muscle fiber 6 using sharp microelectrodes (15–25 MΩ, filled with 3 mol l⁻¹ KCl) connected to an Axoclamp 2A amplifier (Molecular Devices, Sunnyvale, CA, USA). Data were acquired (sampling rate 10 kHz) using a Digidata 1440A digitizer (Molecular Devices) and pCLAMP 10.5 software (Molecular Devices). To evoke synaptic responses, the cut end of the segmental nerve was stimulated with a suction electrode connected to an S11 stimulator (Grass-Telefactor, West Warwick, RI, USA). Both axons innervating muscle fiber 6 were stimulated to produce a compound EPSP. Spontaneous minEPSPs were recorded in the absence of nerve stimulation. Since we never observed spontaneous nerve activity after cutting the segmental nerve, these spontaneous potentials were minEPSPs and not EPSPs. EPSP amplitude, EPSP decay time constant (decay τ) and the resting membrane potential (RMP) were measured using pCLAMP 10.5 software. To analyze minEPSPs, they were identified using MiniAnalysis (Synaptosoft Inc.) and their amplitude and decay τ were measured from averaged traces as in a previous study (Powers et al., 2016). The decay τs were determined by fitting a single exponential to the decay phase using least-squares regression.

Experiments were performed in HL3 saline (Stewart et al., 1994) and the Ca²⁺ concentration was decreased from 1.5 to 1 mmol l⁻¹ to reduce transmitter release and the size of the EPSP. To examine the effects of the EO and its components on EPSPs and minEPSPs, they were dissolved in ethanol (EtOH) and added to HL3 to give a final EtOH concentration of 0.1% (v/v). The proper amounts of EO or its components were added to EtOH to give the final percentage (v/v) or concentration (molar) stated in the Results. We changed solutions using a gravity-fed perfusion system with a perfusion rate of 1 ml 20 s⁻¹ and a chamber volume of 0.5 ml.

The amount of transmitter released before and after citral application was compared by estimating the quantal content (m) of the EPSPs using the equation m=(EPSP_amp/minEPSP_amp)/[1–(fEPSP_amp/E)], where: EPSP_amp is the average EPSP amplitude; minEPSP_amp is the average minEPSP amplitude; f=synaptic current duration/membrane time constant (τ_m); and E is the RMP– (the synaptic reversal potential) (McLachlan and Martin, 1981). The duration of the synaptic current for MF6 was approximately 10 ms (Powers et al., 2016) and the synaptic reversal potential was –1 mV (Jan and Jan, 1976). Our minEPSP decay τ was used for τ_m; this was supported theoretically and experimentally (Powers et al., 2016).

The input resistance (R_in) was measured with a single electrode using 600 ms current injection pulses delivered at 0.1 Hz before and during citral perfusion. The current was adjusted to give hyperpolarizations of 20–30 mV and the electrode resistance was digitally subtracted from the voltage traces.

Experiments were performed on the Ib terminals innervating muscle fibers 6 and 7, which were loaded with the Ca²⁺ indicator Oregon Green 488 BAPTA-1 (OGB-1 coupled to 10,000 MW dextrans, Invitrogen) by applying the indicator to the cut end of the nerve (Macleod et al., 2002). Measurements were performed in HL3.1 saline with 1.5 mmol l⁻¹ Ca²⁺ and 7 mmol l⁻¹ glutamate, which produces a large presynaptic Ca²⁺ signal without muscle contraction (He and Lnenicka, 2011; Macleod et al., 2004). Terminals were imaged using an upright, fixed-stage BH2 microscope (Olympus Optical, Tokyo, Japan) equipped with epifluorescence, differential interference contrast (DIC), a water-immersion 40× Zeiss (Thornwood, NY, USA) lens (NA 0.75) and a scientific complementary metal-oxide semiconductor (CMOS) camera (PCO.edge, PCO-TECH Inc., Romulus, MI, USA). The technique for image acquisition and OGB-1 excitation during stimulation has been previously described (Lnenicka et al., 2006). Images were streamed at 100 Hz (10 ms exposures) during nerve stimulation at 0.1 Hz. SigmaPlot 12.5 (SPSS Inc.) was used to transform and analyze the fluorescent signals. The percentage change in fluorescence (ΔF) was calculated as 100×(fluorescence–resting fluorescence)/resting fluorescence.

All statistical analyses were performed using SigmaPlot 12.5. All statistical tests were two-tailed. n-values equal the number of tubes (10 flies/tube) for behavioral studies and number of animals (1 experiment/animal) for physiological studies. Values are listed as means±s.e.m.

We first examined whether the EO of L. alba sedated D. melanogaster as previously reported for other animals. Flies were exposed to the EO (0.5–10 µl) in a closed container for 150 min. This produced immobilization of the flies, which increased during the exposure (Fig. 1A). The percentage of flies immobilized was also positively correlated with the EO volume (Fig. 1B). This effect of EO volume was presumably due to a more rapid evaporation rate and increase in vapor concentration for larger volumes (Materials and Methods). The effect of the EO was reversible and almost all flies had recovered by 6 h post-exposure (Fig. 1A). Thus, this period of exposure to the EO produced immobilization but not mortality.

We then tested whether the major components of the EO also produced immobilization and compared their potency. According to gas chromatography–mass spectrometry (Table 1), the three largest components of our EO were citral (58.5%), carvone (7.4%) and limonene (7.3%). We also tested the sesquiterpene β-caryophyllene, which was a minor component (0.6%) of our EO but appears as a major component in one of the chemotypes (Hennebelle et al., 2008). In these experiments, we used a 10 µl volume and again recorded the number of flies immobilized every 30 min for 150 min. Citral and carvone produced similar immobilization (80–90%) at the end of the exposure; however, citral appeared to act faster since, at 60 min exposure, about 70% of flies were immobilized by citral but less than 10% by carvone (Fig. 2A). Nearly all of the flies recovered for both citral and carvone. β-caryophyllene did not produce immobilization and was not further examined. Limonene produced a very strong effect; all of the flies were immobilized after a 90 min exposure and most of these flies remained immobilized 6 h after exposure (Fig. 2A). This exposure was lethal since flies immobilized at 6 h did not recover after 24 h. The strong effect of limonene could have been due to its 20-fold greater vapor pressure compared with citral and carvone. So we added limonene to propylene glycol in the proper MF to give a vapor pressure similar to citral and carvone (see Materials and Methods) and repeated the experiment. Now, limonene had a negligible effect; thus, the effect of citral and carvone was much greater than limonene at comparable vapor concentrations. Next, we examined the relative potency of citral, carvone and limonene when presented at the MF found in the EO (Fig. 2B). Both citral and carvone produced about 80% immobilization at the end of the exposure but, again, citral acted much more rapidly; the effects of limonene were minimal. Finally, we added citral, carvone and limonene together at their MFs found in the EO to produce a ‘reconstituted’ EO (Fig. 2B). This reconstituted EO produced an effect very similar to the EO, indicating that these compounds (predominantly citral and carvone) were sufficient to produce its action. The effect of citral and carvone did not appear to be additive but rather the time course appeared intermediate between the two. In conclusion, it appears that the action of the EO is mainly due to citral and carvone, with citral having the stronger effect.

We examined the effect of the EO and its components on the Drosophila larval neuromuscular junction (NMJ) to test for an effect on synaptic transmission. In these studies, we measured the compound EPSP recorded from muscle fiber 6. Since the EO was dissolved in EtOH, we performed control experiments to test for possible effects of 0.1% EtOH on synaptic transmission. The nerve was stimulated at 0.5 Hz and EPSPs were recorded for 3 min in HL3, 5 min in HL3 with 0.1% EtOH and another 5 min in HL3 (Fig. 3A,B). The average EPSP amplitude (16.2±1.5 mV) measured immediately before EtOH (120–180 s) was not significantly different (P>0.05) from that measured at the end (420–480 s) of EtOH application (15.0±1.4 mV) but was significantly greater (P<0.01) than the amplitude (14.1±1.3 mV) at the end (720–780 s) of the recovery period (one-way repeated measures ANOVA). Thus, EtOH did not appear to have a significant effect on EPSP amplitude, although there appeared to be a small but significant rundown of EPSP amplitude during the experiment.

Initially, pilot experiments were performed with varying EO concentrations to determine whether the EO affected synaptic transmission. We found that 0.012% (v/v) EO produced a major reduction in EPSP amplitude within a few minutes of application (Fig. 3B). The effect was reversible since the EPSP amplitude recovered during the 5 min wash. We then tested the effect of citral and compared it to the EO. We chose 0.007% citral since this should be similar to the amount of citral in the 0.012% EO saline. This was based upon a citral MF of 0.59 and the similarity of the molecular weights and densities of the other major components to citral. We found that perfusion of citral resulted in a reduction in EPSP amplitude that appeared similar to that produced by the EO (Fig. 3A,B). Thus, it appeared that the effect of the EO on EPSP amplitude could be accounted for by the action of citral; however, we also tested the effects of the other EO components. To directly compare their actions to citral, we applied them at the same concentration as citral (0.4 mmol l⁻¹). Carvone reduced the EPSP amplitude but the effect appeared less than seen for citral (Fig. 3B). Limonene did not produce an apparent effect on EPSP amplitude. β-caryophyllene was also tested and it did not produce any effect on EPSP amplitude.

We statistically analyzed the effects of the EO and its components by determining the change in EPSP amplitude produced by the drugs and comparing them to the control (EtOH). In these analyses, we compared averaged values obtained for 1 min before drug application (120–180 s) to those during the last minute of drug application (420–480 s) (Fig. 4A). Both the EO and citral produced approximately a 60% reduction in EPSP amplitude, which was highly significant when compared with the control (Fig. 4B). Carvone produced about a 30% reduction, which was also significant. None of the other components produced a significant change in EPSP amplitude. We examined the EPSP decay τ and the muscle resting membrane potential (RMP). Neither the EO nor its components produced a significant change in the EPSP decay τ; however, both the EO and citral produced a significant reduction (depolarization) in the RMP (Fig. 4B). Since there was no significant effect of EO or its components on the EPSP decay τ, this argues against a major effect on the muscle fiber membrane resistance. To more directly examine the membrane resistance, we measured the input resistance (R_in) during citral application (Fig. 4C). Again, we found no significant effect of citral on the membrane resistance. We also compared measurements made at the end of the recovery period (720–780 s) to those made before drug application. The change in EPSP amplitude, EPSP decay τ and RMP were not significantly different from the control for any of the drugs (P>0.05, one-way ANOVA).

We focused on citral for the remainder of this study since it appeared to be mainly responsible for the synaptic effects of the EO. The decrease in EPSP amplitude could be due to either a decrease in transmitter release or a change in postsynaptic properties. To distinguish between presynaptic and postsynaptic effects, we examined minEPSPs recorded in the absence of nerve stimulation for 3 min before applying citral and 5 min during citral application (Fig. 5A). When we compared the minEPSPs recorded immediately before citral application (120–180 s) to those at the end of citral application, we found a small but significant reduction in minEPSP amplitude and no effect on the minEPSP decay τ. However, the reduction in EPSP amplitude was much greater than the reduction in minEPSP amplitude (Fig. 5B), indicating that the decrease in EPSP amplitude was mainly due to a decrease in transmitter release. To estimate the reduction in transmitter release produced by citral, we calculated the average number of transmitter quanta released (see Materials and Methods). For this calculation, we used the following values before and after citral application, respectively: EPSP amplitude 19.3 mV, 7.9 mV; RMP –69.0 mV, –57.3 mV; minEPSP amplitude 0.58 mV, 0.48 mV and minEPSP decay τ 24.8 ms, 24.5 ms. We found that quantal transmitter release decreased from approximately 38 to 18 (53% reduction) as a result of applying citral to the synapse.

The decrease in transmitter release could be due to a decrease in presynaptic Ca²⁺ influx. To test for this, we imaged single action potential (AP)-evoked Ca²⁺ transients in the presynaptic terminal before and after the addition of citral (0.4 mmol l⁻¹). The nerve was stimulated at 0.1 Hz and the Ca²⁺-transient amplitudes were compared before adding citral and at 3–6 min after the addition of citral (Fig. 6A). We found that the addition of citral resulted in a significant 28% decrease in the amplitude of the Ca²⁺ transient (Fig. 6B). We also performed control experiments where only EtOH (0.1%) was perfused and measured the single AP Ca²⁺-transient amplitudes before and after EtOH application. EtOH produced no significant decrease in the Ca²⁺-transient amplitude (P> 0.10; paired Student's t-test). The effect of citral was further confirmed by comparing the change in the Ca²⁺-transient amplitude produced by citral and EtOH and the difference was significant (Fig. 6B).

We found that 150 min exposure to L. alba EO vapors produced immobilization of adult fruit flies. This fly immobilization is consistent with its sedative effects in rodents and fish (Souza et al., 2017; da Cunha et al., 2010; Vale et al., 1999), and its toxicity to larval and adult insects (Vera et al., 2014; Peixoto et al., 2015a). These latter studies examined long exposure times (usually 2 days) and we assume that the immobilization seen in our studies would have led to lethality during prolonged exposure. We examined the major components of the EO – citral, carvone and limonene – along with the minor component β-caryophyllene. When the components were applied at the same volume, all except β-caryophyllene produced fly immobilization. This is consistent with earlier findings in houseflies: prolonged exposure to citral or limonene was lethal for housefly larvae in both contact and fumigation assays (Kumar et al., 2014), and citral, carvone and limonene were all found to be lethal to adult flies in a fumigation assay (Rice and Coats, 1994; Lee et al., 2003). The toxicity of these components was also demonstrated for other insects. Prolonged exposure to citral or limonene was lethal for mosquito larvae; limonene and carvone were lethal for cockroaches; and citral, carvone and limonene were lethal to various adult insects found in stored grains (Peixoto et al., 2015a; Giatropoulos et al., 2012; Lee et al., 2003). The toxic effects of citral also extends to other invertebrates such as snails (Price and Berry, 2008). We compared the potency of citral, carvone and limonene by equalizing their vapor pressures, since limonene had a vapor pressure 20× greater than the other two components. After reducing its vapor pressure, limonene did not produce significant fly immobilization, demonstrating that it was less potent than citral and carvone.

When the components were applied at MFs similar to those found in the EO, the maximum immobilization produced by citral and carvone was similar but citral acted much more rapidly. The reconstituted EO formed by combining these three major compounds showed an effect similar to the EO. This was presumably mainly due to the actions of citral and carvone, and it was interesting that their combined effects did not appear additive.

The EO produced a more than 50% reduction in EPSP amplitude at the larval NMJ. This effect was replicated by citral applied at the same concentration as found in the EO. In addition, citral was more potent than the other components when applied at the same concentrations. Carvone produced a significant reduction in EPSP amplitude but it was less than for citral, whereas limonene and caryophyllene did not produce a significant effect. It may be that, with longer exposure times, the effect of carvone would become more similar to citral, as seen for the immobilization studies, since, at the end of the exposure, the effect had plateaued for citral but not for carvone. Nonetheless, it is likely that the effect of the EO on synaptic transmission was mainly due to citral since the MF of citral in the EO is about 8× that of carvone. The EO and citral also produced a significant depolarization of the muscle fiber RMP. We currently have no evidence that this resulted from a change in membrane conductance since there was no significant change in the decay τ for EPSPs and minEPSPs or the muscle fiber R_in.

The immobilization produced by the EO, citral and carvone could result from the blocked synapses. Support for this comes from parallels between the effects of the components on immobilization and synapses: citral and carvone produced greater immobilization and a reduction in EPSP amplitude than limonene and β-caryophyllene. In addition, the citral concentration (0.4 mmol l⁻¹) that reduced EPSP amplitude was similar to the concentration that produced toxicity in snails (LC₅₀ 0.3–0.6 mmol l⁻¹) and mosquito larvae (LC₅₀ 0.5 mmol l⁻¹) (Price and Berry, 2008; Giatropoulos et al., 2012). The immobilization could be produced by blocking the NMJ or more likely through a more general block of synaptic transmission. In this case, the inhibitory effects on individual synapses would be multiplicative when considering a series of synapses in a neural circuit.

In the presence of citral, there was a much larger decrease in EPSP amplitude than minEPSP amplitude and we estimated that citral produced about a 50% decrease in transmitter release at the NMJ. A likely candidate for the reduction in transmitter release was a reduction in Ca²⁺ influx at the presynaptic terminal. This appears to be the case since single-AP Ca²⁺ transients imaged before and after citral application showed a significant decrease in amplitude. This decrease in the Ca²⁺-transient amplitude should represent a decrease in Ca²⁺ influx since the Ca²⁺-transient amplitude is dependent on the amount of Ca²⁺ entering and the concentration of fast Ca²⁺ buffers (He and Lnenicka, 2011), and it is highly unlikely that changes in Ca²⁺ buffer concentration occurred during the brief citral application. The postsynaptic effects produced by citral – muscle depolarization and reduced minEPSP amplitude – would contribute to a decrease in EPSP amplitude; however, their effects would be much less than the decrease in transmitter release.

Citral could reduce Ca²⁺ influx and transmitter release by affecting voltage-gated (VG) channels at the presynaptic terminal; in fact, multiple monoterpenes have been found to inhibit the activity of VG channels (Tsuchiya, 2017). For example, a reduction in VG Na⁺ conductance might have reduced the excitability of the synaptic terminals, leading to less depolarization and Ca²⁺ influx through VG Ca²⁺ channels. Citral, carvone and limonene reduced the compound action potential (CAP) amplitude in frog sciatic nerve; citral was the most potent (IC₅₀ 0.5 mmol l⁻¹) followed by carvone (IC₅₀ 1.4–2.0 mmol l⁻¹), and limonene produced a slight reduction in CAP at 10 mmol l⁻¹ (Kawasaki et al., 2013; Ohtsubo et al., 2015). This reduction in CAP amplitude presumably resulted from inhibition of VG Na⁺ channels since a correlation between a reduction in CAP amplitude and VG Na⁺ currents was demonstrated for the monoterpenes linalool and carvacrol (Joca et al., 2012; Leal-Cardoso et al., 2010). Alternatively, if the depolarization seen for the muscle fibers also occurs at the terminals, this could have reduced terminal excitability through increased Na⁺ inactivation. In this case, terminal depolarization would have to occur more rapidly than muscle depolarization since the decrease in EPSP amplitude precedes muscle depolarization (see Fig. 3A).

Citral may have reduced Ca²⁺ influx more directly by blocking the VG Ca²⁺ channel, as indicated by some earlier experiments. Citral blocked smooth muscle contractions in rat and rabbit ileum, and it was proposed that this was due to blocking VG Ca²⁺ channels (Sadraei et al., 2003; Devi et al., 2011). For cockroach neurons, citral eliminated the AP afterhyperpolarization (Price and Berry, 2006); this was likely due to blocking VG Ca²⁺ channels since the afterhyperpolarization resulted from a Ca²⁺-activated K⁺ conductance and it was eliminated by VG Ca²⁺ channel blockers (Lapied et al., 1989). It is possible that citral inhibited both VG Na⁺ and Ca²⁺, since some monoterpenes have been shown to act as nonselective inhibitors of voltage-gated Na⁺, Ca²⁺ and K⁺ channels. Linalool inhibited all VG currents in mammalian olfactory receptor cells, retinal cells and cerebellar Purkinje cells (Narusuye et al., 2005), and limonene inhibited all VG currents in mammalian olfactory receptor neurons and goldfish horizontal cells (Kawai et al., 1997; Kawai and Miyachi, 2000). Alternatively, citral may have acted specifically on the Dmca1A-encoded Ca²⁺ channel, which is mainly responsible for transmitter release at the larval NMJ and has similarities to the vertebrate N-type Ca²⁺ channel (Kawasaki et al., 2000; Rieckhof et al., 2003). Monoterpenes can show specificity in blocking VG Ca²⁺ channels. For example, limonene blocked T- and L-type Ca²⁺ channels with an IC₅₀ of approximately 0.2 mmol l⁻¹ in newt olfactory receptor cells (Kawai, 1999); however, it clearly did not block the Dmca1A channel at this concentration since we found that 0.4 mmol l⁻¹ limonene had no significant effect on EPSP amplitude. If VG Na⁺ channels were inhibited in our experiments, the effects were not large since there was no noticeable increase in the threshold for axon stimulation when citral, carvone or limonene were added to the bath. Nonetheless, the synapse is likely to be particularly sensitive to inhibiting VG Ca²⁺ and/or Na⁺ channels due to the Ca²⁺ cooperativity of transmitter release and nonlinear relationship between Ca²⁺ influx and transmitter release seen at synapses, including the larval NMJ (Stewart et al., 1994; Zhong and Wu, 1991).

The monoterpenes appear to show similarities to local anesthetics since they are both lipophilic and act at relatively high concentrations to block VG ion channels, often nonspecifically. For example, the local anesthetic lidocaine reduced the CAP in the frog sciatic nerve with an IC₅₀ of 0.74 mmol l⁻¹ and blocked Na⁺, Ca²⁺ and K⁺ channels in chick ventricular myocytes (Katsuki et al., 2006; Josephson, 1988). Local anesthetics act by binding to the Na⁺ channel (Catterall and Swanson, 2015) and may also influence ion channel activity by altering the structure of the lipid bilayer (Kopeć et al., 2013); monoterpenes might also act by direct binding to ion channels and altering the lipid bilayer (Oz et al., 2015). For local anesthetics, there was a correlation between potency and lipid solubility as measured by the octanol/water partition coefficient (Strichartz et al., 1990); however, this relationship was not seen for monoterpenes. Instead, their relative potency in reducing the CAP amplitude was monoterpene aldehydes≥alcohols≥ketones >> hydrocarbons (Ohtsubo et al., 2015). The relative potency of our EO components in inhibiting synaptic transmission was consistent with these studies of the CAP amplitude. There was no correlation between lipophilicity and inhibition of synaptic transmission since the relative octanol/water partition coefficient was caryophyllene>limonene>citral>carvone (https://chem.nlm.nih.gov/). In addition, the aldehyde (citral) and ketone (carvone) inhibited synaptic transmission, whereas the hydrocarbons (caryophyllene and limonene) did not. This argues for the importance of the =O group, particularly since carvone is identical to limonene except for the addition of an =O group.

We thank Dr Rabi Musah (University at Albany) for advice in the early stages of this study.