Naturally Produced Defensive Alkenal Compounds Activate TRPA1

Abstract

(E)-2-alkenals are aldehydes containing an unsaturated bond between the alpha and beta carbons. 2-alkenals are produced by many organisms for defense against predators and secretions containing (E)-2-alkenals cause predators to stop attacking and allow the prey to escape. Chemical ecologists have described many alkenal compounds with 3–20 carbons common, having varied positions of double bonds and substitutions. How do these defensive alkenals act to deter predators? We have tested the effects of (E)-2-alkenals with 6–12 carbons on transient receptor potential channels (TRP) commonly found in sensory neurons. We find that (E)-2-alkenals activate transient receptor potential ankyrin subtype 1 (TRPA1) at low concentrations—EC ₅₀ s 10–100 µM (in 0 added Ca ²⁺ external solutions). Other TRP channels were either weakly activated (TRPV1, TRPV3) or insensitive (TRPV2, TRPV4, TRPM8). (E)-2-alkenals may activate TRPA1 by modifying cysteine side chains. However, target cysteines include others beyond the 3 in the amino-terminus implicated in activation, as a channel with cysteines at 621, 641, 665 mutated to serine responded robustly. Related chemicals, including the aldehydes hexanal and decanal, and (E)-2-hexen-1-ol also activated TRPA1, but with weaker potency. Rat trigeminal nerve recordings and behavioral experiments showed (E)-2-hexenal was aversive. Our results suggest that TRPA1 is likely a major target of these commonly used defensive chemicals.

Introduction

Defensive chemicals, or allomones, are used by many small and slow moving organisms for protection from predators. Allomones are particularly widespread among arthropods, with many lineages having the ability to produce, store, and deploy irritating chemicals ( Blum 1981 ; Eisner et al. 2005 ; Dossey 2010 ). A vast assortment of compounds is utilized for defense, though in many cases we know little about how they affect target predators.

Aldehydes are common constituents of arthropod defensive secretions. Many contain 3–20 carbons, with various substitutions and unsaturated bonds at different positions ( Blum 1981 ; Aldrich 1988 ). (E)-2-alkenals are aldehydes with a double bond between the α and β carbons, and are present in arthropods, plants, and fungi.

Among arthropods, (E)-2-alkenals are found in many groups’ defenses. Some millipedes use (E)-2-alkenals with 7–12 carbons ( Wheeler et al. 1964 ; Bodner and Raspotnig 2012 ), and the Florida “skunk roach” ( Eurycotis floridana ) uses the 6 carbon long (E)-2-hexenal for defense and as an alarm pheromone ( Farine et al. 1997 ). (E)-2-alkenals have been highly elaborated in heteropterans, for example, “stink bugs” ( Aldrich 1988 ; Eliyahu et al. 2012 ; Noge et al. 2012 ). The range of smells emitted when these animals are handled arises from the variety of compounds in their defensive secretions. (E)-2-alkenals are repellent to predators of many taxa, including vertebrates and invertebrates ( O’Connell and Reagle 2002 ; Staples et al. 2002 ; Eliyahu et al. 2012 ). How are (E)-2-alkenals sensed by such a large range of predators and how does that change the predator’s behavior? Why are so many different alkenals used for defense?

Chemosensation in animals begins when chemicals activate receptors in primary olfactory, gustatory, and chemesthetic sensory neurons exposed to the external environment. Aversive chemical signaling involves the gustatory system, via bitter taste receptors, the olfactory system via olfactory receptors, and chemesthesis, via receptors and ion channels on nociceptive neurons ( Viana 2011 ; Green 2012 ; Mickle et al. 2015 ; Roper 2014 ; Li and Liberles 2015 ). Chemesthesis, the detection of noxious chemicals, is present in all animal groups—hence the previous term “common chemical sense.” In vertebrates, chemesthesis occurs over the entire body surface, especially near the animal’s eyes, nose and mouth (via trigeminal nerve innervation). In arthropods, chemesthesis is less well defined, but occurs via sensory neurons in the legs, labellum, and ovipositor ( Kang et al. 2010 ; Zhong et al. 2012 ).

Multiple ion channels in the transient receptor potential (TRP) superfamily contribute to detection of noxious chemicals, depolarizing neurons when activated ( Nilius and Owsianik 2011 ; Viana 2011 ; Mickle et al. 2015 ). TRPV1, for example, responds to heat, protons, and capsaicin, the active ingredient in “hot” chili peppers. It is expressed in nociceptors and plays important roles in pain sensation and inflammatory hyperalgesia (reviewed in Julius 2013 ). TRPV3 is expressed in the taste buds and skin cells, where it responds to carvacrol and thymol, from oregano and thyme, respectively. TRPM8 is also expressed in sensory neurons, and responds to cooling and menthols from mint.

TRPA1 is another prominent chemosensory TRP channel. TRPA1 is expressed in sensory neurons, where its activation contributes to nociception ( Story et al. 2003 ; Jordt et al. 2004 ). TRPA1 is activated by modification of free thiol groups of cysteine side chains, especially those located in its amino terminus ( Hinman et al. 2006 ; Macpherson et al. 2007a ). Pungent chemicals such as allyl isothiocyanate (AITC) from mustard seed, allicin from garlic, and cinnamaldehyde from cinnamon, activate TRPA1 in this way ( Macpherson et al. 2005 ; Bautista et al. 2006 ). Covalent modification leads to sustained activation of TRPA1, with poor reversibility. Chemicals including carvacrol, menthol, and local anesthetics reversibly activate TRPA1 by noncovalent binding ( Xu et al. 2006 ; Karashima et al. 2007 ; Leffler et al. 2011 ).

Compounds with α,β-unsaturated bonds can react with free thiols via Michael addition, and are thus potential TRPA1 activators. Indeed, short alkenals can activate TRPA1, including the shortest, propenal (acrolein), butenal (crotonaldehyde), and pentenal. However, potency appears to weaken as carbon chain length increases ( Bautista et al. 2006 ; Andre et al. 2008 ), and those with 9 or more carbons reportedly fail to activate TRPA1 ( Macpherson et al. 2007b ). In this case, do these deterrents act via other channels or receptors?

We found that (E)-2-alkenals with 6–12 carbons are potent activators of TRPA1, eliciting large, long-lasting currents; activation of other sensory TRP channels was much weaker or absent. We also found that (E)-2-hexenal elicited neural responses from rat trigeminal nerves and behavioral aversion in rats.

Materials and methods

Cells and plasmids

All cells used for calcium imaging and electrophysiology were cultured in Dulbecco’s modified Eagle’s medium/F-12 nutrient mixture (1:1) with 10% fetal calf serum and 100U penicillin/100 µg/mL streptomycin. Media for FlpIn 293 TREx (Invitrogen) stable cell lines was supplemented with 15 µg/mL blasticidin and 150 µg/mL hygromycin B; media for FlpIn 293 TREx parental line used for transient transfections had 15 µg/mL blasticidin alone. Cells were maintained at 37 °C in a humidified 95% oxygen/5% CO ₂ atmosphere. For experiments with transiently expressed channels, ~24–48h prior we transfected nearly confluent 35mm dishes using 5 µL Lipofectamine 2000 (Invitrogen) with 1.5–3 µg of the appropriate channel expression plasmid and 0.1–0.3 µg of either eGFP-C1 or mCherry-N3 (Clontech; we used mCherry for fura-2 imaging to avoid the weak 380nm excitation of eGFP).

TRPA1 experiments

Wild-type (WT) human TRPA1 channels were expressed in a FlpIn 293 TREx cell line stably incorporating the human TRPA1 cDNA. About 18–24h prior to recording, cells were seeded onto 12mm glass coverslips and expression was induced with 0.1 µg/mL doxycycline added to the culture medium. In some later experiments ( Figure 7D ) we used a lower concentration of doxycycline (12.5ng/mL) to reduce maximal current amplitude. In these cases we normalized the current by the amplitude activated by carvacrol (300 µM or 1mM) or AITC (300 µM). To test whether the TRPA1 channel with 3 N-terminal cysteines (at positions 621, 641, 665) mutated to serine responds to alkenals (transient receptor potential ankyrin 1 triple cysteine mutant (C621S, C641S, C665S, TRPA1-3C), we transiently expressed this channel in the FlpIn 293 TREx parental line.

Figure 7.

The aldehydes decanal and hexanal, and the alcohol (E)-2-hexen-1-ol, activate human TRPA1. ( A ) Average fura-2 ratios from WT TRPA1 expressing cells in response to 100 µM decanal ( gray bar ). A later application of 300 µM carvacrol caused a further increase in ratio ( black bar ). There was no response to the 0.1% ethanol, the vehicle for decanal ( open bar ). Average of 54 cells. ( B ) TRPA1-3C channels are also activated by 100 µM decanal (average of 53 cells). ( C ) Average fura-2 ratio of WT TRPA1 expressing cells in response to 100 µM hexanal ( gray bar ). Later application of 300 µM carvacrol again further increased the ratio ( black bar ). There was no response to the ethanol vehicle ( gray bar ). Average of 128 cells. ( D ) Voltage-clamp recordings of WT TRPA1 currents activated by 3mM (E)-2-hexen-1-ol in 0 added Ca ²⁺ Tyrode’s solution. Trace shows average current at +60 mV during ramps. G ray bar shows application of 3mM (E)-2-hexen-1-ol, and black bar shows application of 300 µM carvacrol.

Other TRP channel experiments

TRPV1 and TRPV3 experiments were done using WT human TRPV1 and mouse TRPV3 channels expressed either in an inducible FlpIn 293 TREx line, or after transient transfection. No differences were observed between these approaches. Rat and mouse TRPA1, human TRPV2, mouse TRPV4, and human TRPM8 were transiently expressed in the FlpIn 293 TREx parental line. The TRPV4 channel had yellow fluorescent protein (YFP) fused to its carboxy-terminus. YFP has minimal effects on fura-2 ratios, because excitation by 340 and 380nm light is equal and low.

Calcium imaging

We imaged fura-2 fluorescence to monitor changes in intracellular Ca ²⁺ concentration ([Ca ²⁺ ] _i ) resulting from TRP channel activation. Doxycycline-induced or transiently transfected cells were plated on chambers with glass coverslip bottoms, coated with poly- d -lysine (0.5mg/mL). Cells were loaded with 2 µM fura-2 for 60–120min at room temperature in Leibovitz’s L15 media with 5mM HEPES, pH 7.40 (in some experiments 0.02% pluronic was added to improve fura-2 solubility). The bath solution was Tyrode’s solution containing (in mM): 150 NaCl, 4 KCl, 2 CaCl ₂ , 1 MgCl ₂ , 10 D-glucose, 10 HEPES, pH 7.4.

Cells were imaged using an Olympus IX-70 microscope with a 20×, 0.75 N.A. objective (optimized for UV transmittance) and an Orca-ER camera (Hamamatsu). A Lambda DG-4 xenon lamp (Sutter Instruments) illuminated cells every 3s with 340/10 and 380/10nm filtered light. The dichroic was 400nm long-pass; the 510nm emission filter had 80nm bandwidth (all from Chroma). Excitation durations at 340 and 380nm were equal, ranging from 200 to 275ms depending on loading efficiency. Fluorescence values were averaged over regions containing individual cells after background subtraction, and the ratio of fluorescence excited at 340 divided by 380nm was calculated. Slidebook 4.2.0 (Intelligent Imaging Innovations) controlled the camera and light source. The external solution was changed by pushing 3mL of solution with a syringe connected to the ~150 µL bath via polyethylene tubing. Solution exchange was complete within ~6s; the timing of solution changes has been offset by 3s to account for the delay between the syringe and the bath inlet.

At the end of each experiment we applied a control agonist to determine which cells expressed the channel of interest. Cells that reached a ratio of 2.0 were considered positive. Some cells had high initial [Ca ²⁺ ] _i levels—we did not include those with baseline ratios in control Tyrode’s >2.0, although they still responded to agonist. About 73–100% of inducibly expressing cells, and 12–65% of transiently expressing cells, fit these criteria.

Electrophysiology

HEK293 cells expressing TRP channels were voltage-clamped with an Axopatch 200B controlled by pClamp 9.2 and a Digidata 1320A (Molecular Devices). Pipettes with 2–5 MΩ resistances were pulled from borosilicate glass (WPI) and filled with an internal solution containing (in mM): 122 cesium methanesulfonate, 1.8 MgCl ₂ , 9 ethylene glycol-bis(2-aminoethylether)- N , N , N ′, N ′-tetraacetic acid, 14 creatine phosphate (Na-salt), 4 MgATP, 0.3 NaGTP, 10 HEPES, pH 7.2 (with CsOH). The external solution for recordings was Tyrode’s with 0 added Ca ²⁺ (in mM): 150 NaCl, 4 KCl, 3 MgCl ₂ , 10 D-glucose, 10 HEPES; pH 7.4 (with NaOH; due to trace contamination of salts, final [Ca ²⁺ ] is typically ~10 µM). For the mouse and rat TRPA1 recordings, the external solution was 140mM NaCl and 5mM KCl (all others the same). Human TRPA1 currents recorded in this solution for comparison were not substantially different from those in Tyrode’s. These solutions prevented Ca ²⁺ -dependent potentiation and desensitization of TRPA1 channels ( Wang et al. 2008 ; Ibarra and Blair 2013 ). The Ca ²⁺ -containing Tyrode’s solution was the same except with 2mM CaCl ₂ and 1mM MgCl ₂ .

Cells were held at −40 mV or 0 mV and 200ms ramps from −100 mV to +100 mV were applied every 2s. The voltage was held at −100 and +100 mV for 40ms before and after the ramp; a 40ms repolarization to −60 mV followed the ramp. Currents were filtered at 5kHz and digitized at 20kHz. Series resistances were <10 MΩ and compensated 75–85%. Pipette tips were wrapped with parafilm to reduce capacitance. For analysis, currents were averaged near ± 60 mV during the ramp (± 60.2 to 59.8 mV). Solutions were applied with an array of 200 µm inner diameter quartz tubes after raising the cell off the coverslip. Solution exchange was complete in <1s.

Rat ethmoid nerve recordings

The activity of the ethmoid branch of the trigeminal nerve and breathing rate was measured in response to deterrent chemicals as in Silver et al. (2006) . Briefly, adult male rats (Sprague-Dawley, 250–400g) were anesthetized with 1.0g/kg urethane, an incision was made above the orbit to expose the ethmoid nerve, which was cut and placed on 2 platinum-iridium wire hook electrodes. The incision was filled with halocarbon oil to prevent the nerve from drying out. Multiunit electrical activity was amplified (Grass P-511), integrated, and analyzed using AcqKnowledge 3.7.3 (Biopac). The trachea was cut and 1 tube was inserted rostrally into the nasopharynx to apply solutions to the nasal cavity. A second tube was inserted caudally to allow the rat to breath room air. A thermistor was inserted in this tube to monitor the rat’s breathing. Chemical stimuli were applied in ~1mL of Ringer’s solution into the nasopharynx, with excess solution draining from the nostrils. Rat’s Ringer solution contained (in mM): 135 NaCl, 5.4 KCl, 1.8 CaCl2, 5 HEPES, pH 7.2 (with NaOH). The effects of deterrent chemicals on breathing were assessed by calculating the relative breath duration, defined as the length of the first breath following the stimulus divided by the average breath duration of the preceding 10 breaths in control.

Rat behavioral aversion assay

The aversive effects of (E)-2-hexenal and (E)-2-hexen-1-ol on rats were tested using the following paradigm: Rats were habituated in an opaque-walled plexiglass box (61cm × 61cm × 41cm) with water-filled dishes in each corner for 10min over 3 consecutive days. On the fourth day, (E)-2-hexenal or (E)-2-hexen-1-ol was added to one of the water dishes and the time rats spent in each quadrant during a 10-min trial was monitored with Noldus Ethovision. Control trials had water in all 4 quadrants. All rat experimental procedures were approved by Wake Forest University’s Animal Care and Use Committee.

Reagents

The structures of TRPA1 channel activators we used in this study are shown in Figure 1 . Stock (E)-2-alkenals solutions were made daily in 100% ethanol before dilution into 0 added Ca ²⁺ Tyrode’s. Final ethanol concentration was 0.1%, which had no effect when tested on the TRP channels we studied. Many of the tested chemicals—especially (E)-2-dodecenal and hexanal—degraded over a few hours in Tyrode’s solution. Fresh working solutions for those 2 were made every 60–90min. Between patch-clamp or imaging experiments, we thoroughly cleaned the bath with ethanol, water, and a 0 added Ca ²⁺ Tyrode’s solution containing 300 µM l -cysteine.

Figure 1.

Structure of TRPA1 agonists used in this study. ( a ) (E)-2-hexenal, ( b ) (E)-2-octenal, ( c ) (E)-2-decenal, ( d ) (E)-2-dodecenal, ( e ) hexanal, ( f ) decanal, ( g ) (E)-2-hexen-1-ol, ( h ) carvacrol, ( i ) allyl isothiocyanate (AITC).

Results

Does (E)-2-decenal activate TRP channels known to participate in chemical sensation? We expressed TRPA1, TRPV1-4, and TRPM8 in human embryonic kidney (HEK) cells and imaged fura-2 ratios to detect changes in [Ca ²⁺ ] _i resulting from activation of these Ca ²⁺ -permeable channels ( Figure 2 ). 100 µM (E)-2-decenal elicited large [Ca ²⁺ ] _i increases from only from TRPA1-expressing cells ( Figure 2A ). Addition of (E)-2-decenal caused a rapid and large increase in the fura-2 ratio in every cell expressing TRPA1, remaining high while (E)-2-decenal was present. When (E)-2-decenal was removed, [Ca ²⁺ ] _i declined partially, but remained above the initial baseline. We then added 300 µM AITC to test whether any TRPA1 channels were available to respond, but there was no further change in the fura-2 ratio. The average peak ratio was 10.2±0.2 from 64 cells, peaking 59.8±3.5s after (E)-2-decenal application.

Figure 2.

(E)-2-decenal activates TRPA1 channels. Traces show average fura-2 ratios from HEK293 cells expressing the indicated TRP channel. Gray bars show application of 100 µM (E)-2-decenal, and black bars show application of the control agonist for each channel. About 2min in control Tyrode’s solution separated the 2min (E)-2-decenal application from the control agonist application. Control agonists were: ( A ) TRPA1: 300 µM AITC; ( B ) TRPV1: 10 µM capsaicin; ( C ) TRPV2: 500 µM 2-APB; ( D ) TRPV3: 300 µM carvacrol; ( E ) TRPV4: 100nM GSK-1016790A; ( F ) TRPM8: 500 µM menthol. In some experiments, the 0.1% ethanol vehicle (for (E)-2-decenal) was applied first, as shown by open bars . In separate experiments, 0.1% ethanol had no effect on TRPV2, V4, and M8 expressing cells. Averages are responses from 37 to 129 cells. The external solution was 2 Ca ²⁺ Tyrode’s.

The 5 other TRP channels we tested gave no, or minimal, [Ca ²⁺ ] _i increases when (E)-2-decenal was applied ( Figure 2B -F). In some cases the average response from many cells increased slightly (e.g., TRPV1, TRPV2, and TRPM8). This resulted from asynchronous [Ca ²⁺ ] _i increases in a few cells. The fraction of cells reaching ratios >2.0 was: 8 of 78 TRPV1-expressing cells, 7 of 45 TRPV2-expressing cells, 5 of 127 TRPV3-expressing cells, 0 of 37 TRPV4-expressing cells, and 6 of 108 TRPM8-expressing cells. Furthermore, these responses were unusual, as they were sustained in TRPV1 and TRPV3 cells, but transient in TRPV2 and TRPM8 cells. By comparison, only 2 of 153 uninduced parental HEK FlpIn cells responded to 100 µM (E)-2-decenal (not shown). The (E)-2-decenal response was thus much weaker, and more variable, in these sensory TRP channels.

Although (E)-2-alkenals with 3–20 carbons are used as defensive chemicals by arthropods, 6–12 carbons are most common. The 6-carbon containing (E)-2-hexenal is particularly prevalent, and the 8 and 12 carbon types, (E)-2-octenal and (E)-2-dodecenal, are also widespread ( Aldrich 1988 ; Eliyahu et al. 2012 ). Does the number of carbons affect the activation of TRPA1? Because the 4-carbon crotonaldehyde activated TRPA1-mediated Ca ²⁺ entry with an EC ₅₀ 3-fold higher than the 3-carbon acrolein ( Andre et al. 2008 ), longer carbon chains might be less effective. We tested this by recording TRPA1-mediated currents in patch-clamp to directly measure the activation elicited by (E)-2-hexenal, (E)-2-octenal, (E)-2-decenal, and (E)-2-dodecenal in TRPA1-expressing cells ( Figure 3 ).

Figure 3.

(E)-2-alkenals with 6–12 carbons activate TRPA1 channels with varying potencies. ( A ) TRPA1 currents elicited by 200ms voltage ramps from −100 to +100 mV; holding potential was −40 mV. Sweeps were applied at 0.5 Hz. ( B ) Current-voltage (I–V) relations from 2 cells, extracted from responses during ramps. Lowercase letters indicate timing of sweeps shown in C . ( C ) TRPA1 current averaged at +60 mV (recorded during ramp) in response to application of 30 µM of each (E)-2-alkenal (gray bars show timing). The black bar in the top trace shows the application of 300 µM carvacrol after 5min in 30 µM (E)-2-hexenal. The open bar in the bottom trace shows application of 100 µM HC-030031. ( D ) Concentration–response curves showing average TRPA1 current density versus (E)-2-alkenal concentrations.

About 30 µM (E)-2-hexenal activated a small current during voltage ramps from −100 to +100 mV. There were more available TRPA1 channels expressed in this cell, because addition of 300 µM carvacrol activated very large currents (388 pA at +60 mV in (E)-2-hexenal compared with 5907 pA in carvacrol). These currents were typical of TRPA1 channels: they reversed at ~ −5 mV, and were slightly outwardly rectifying ( Figure 3A , B). Furthermore, 100 µM of the TRPA1 inhibitor HC-030031 blocked 80.3±3.7% ( n = 13) of the current at +60 mV, similar to previous findings ( Figure 3C ) ( McNamara et al. 2007 ). Figure 3C shows the average currents at +60 mV during the voltage ramp (applied at 0.5 Hz), in response to application of 30 µM of each (E)-2-alkenal. On average, 30 µM (E)-2-hexenal elicited 0.056 nA/pF at +60 mV ( n = 12).

(E)-2-alkenals with longer carbon chains were better agonists. At 30 µM, TRPA1 currents activated by (E)-2-octenal, (E)-2-decenal, and (E)-2-dodecenal were very similar in amplitude, averaging 0.71, 0.71, and 0.75 nA/pF, respectively ( Figure 3D , n = 4, 6, 5). Although the current amplitudes were similar, the activation kinetics differed: (E)-2-octenal and (E)-2-decenal required ~5min to reach steady state, whereas (E)-2-dodecenal required just ~1min However, in all cases activation only became apparent after a delay, perhaps reflecting the time needed for alkenals to partition into the cell membrane and enter the cytoplasm. Wan et al. (2013) saw a similar delay before AITC activation of TRPA1 recorded in the absence of external Ca ²⁺ .

The TRPA1 currents we recorded in response to (E)-2-alkenal concentrations from 1 µM to 1mM are plotted as concentration–response curves ( Figure 3D ). These curves show that (E)-2-hexenal was the least potent and efficacious. The other (E)-2-alkenals were similar in efficacy, though we were unable to test higher concentrations, as seals rapidly deteriorated in >300 µM (E)-2-octenal/decenal and >100 µM (E)-2-dodecenal. The relative potency of the alkenals was hexenal < decenal < octenal = dodecenal. This pattern does not quite match the predicted partition ratio (logP) values for the compounds: (E)-2-hexenal (1.79) < (E)-2-octenal (2.81) < (E)-2-decenal (3.83) < (E)-2-dodecenal (4.85). Thus, (E)-2-alkenal potency may be affected by factors other than its ability to partition into the lipophilic cell membrane. TRPA1 activation by these longer alkenals may underlie the Ca ²⁺ increases Nakajima et al. (2013) recently found in an enteric endocrine cell line.

TRPA1 is the sole member of the TRPA subfamily in most vertebrates (teleost fishes have 2 closely related versions, Kang et al. 2010 ). In general, electrophilic agonists of TRPA1 channels activate all of the vertebrate TRPA1 orthologs that have been tested. Invertebrate TRPA channels are also activated, including Drosophila TRPA1 ( Kang et al. 2010 ) and the hymenopteran specific TRPA ( Kohno et al. 2010 ). However, some channel modulators have different effects on different orthologs ( Chen et al. 2008 ; Banzawa et al. 2014 ). Commonly, agonists may result in block of particular orthologs at higher concentrations, leading to a bell shaped dose–response curve. Examples are menthol, eugenol, and thymol, which activate human and rodent TRPA1, but give pronounced block of rodent TRPA1 near millimolar concentrations ( Karashima et al. 2007 ; Xiao et al. 2008 ; Bianchi et al. 2012 ). Are (E)-2-alkenals activators of other TRPA1 orthologs? We tested whether (E)-2-alkenals can activate TRPA1 isoforms from other vertebrate species by recording from cells transiently expressing rat or mouse TRPA1 ( Figure 4 ). We found that a 2–3 minapplication of 100 µM (E)-2-decenal activated 440.9±127.4 pA/pF of current at +60 mV in rat TRPA1 expressing cells ( n = 7). The time course of activation was similar to that recorded in human TRPA1 cells, and the current was reversibly blocked by 100 µM HC-030031 (68.5±2.4% block in 5 cells tested). We then added 300 µM AITC to test whether any further current increase would occur, and found a 33.6±9.8% increase over the (E)-2-decenal activated current ( n = 3).

Figure 4.

Rodent TRPA1 channels are activated by (E)-2-decenal and (E)-2-hexenal. ( A ) Rat TRPA1 is activated by 100 µM (E)-2-decenal. Left shows the average current at +60 mV during a 2-min application of 100 µM (E)-2-decenal ( gray bar ). 100 µM HC-030031 ( open bar ) blocked the current, and 300 µM AITC ( black bar ) further activated TRPA1 slightly. Right : I–V curves taken from voltage ramps applied at the times indicated by lowercase letters. ( B ) Mouse TRPA1 is activated by 1mM (E)-2-hexenal. Symbols as in A, except 1mM carvacrol was applied after (E)-2-hexenal.

We next tested whether the rodent TRPA1 channels would be activated by 1mM (E)-2-hexenal ( Figure 4B ). Rat TRPA1 channels were well activated by 1mM (E)-2-hexenal, with 247.5±102.0 pA/pF at +60 mV elicited during a 2min application ( n = 5). 300 µM AITC added to these cells then increased the current by 40±13.9%.

Mouse TRPA1 mediated currents were small, perhaps simply from weak heterologous expression. Nonetheless, 1mM (E)-2-hexenal activated 17.1±6.5 pA/pF ( n = 5), again with a similar time course to human TRPA1. Currents were potently blocked by 100 µM HC-030031 (98.2±0.8%, n = 5). In 4 of these cells we tested how much more current 1mM carvacrol would elicit: the current increased 50.1±11.5%. This modest amount might result from some direct block at this carvacrol concentration. Structurally similar menthol and thymol both result in net block of rodent TRPA1 channels in the millimolar range, which does not occur in the human isoform ( Karashima et al. 2007 ; Xiao et al. 2008 ). In summary, although the exact potencies of (E)-2-alkenals on rodent TRPA1 channels relative to the human isoform is not clear, they clearly cause activation.

Key cysteine residues of TRPA1 channels that are modified by electrophilic compounds such as AITC and cinnamaldehyde, have been localized to the amino-terminus ( Hinman et al. 2006 ; Macpherson et al. 2007a ). (E)-2-alkenals are likely to modify free thiol groups on cysteines in a similar way, owing to their unsaturated α,β carbon bond. We tested whether the TRPA1-3C channel (cysteines at positions 622, 641, and 665 mutated to serines) is activated by (E)-2-alkenals ( Hinman et al. 2006 ). Figure 5 shows fura-2 imaging of cells expressing human WT TRPA1 and TRPA1-3C channels. About 100 µM (E)-2-hexenal, (E)-2-octenal, (E)-2-decenal, and (E)-2-dodecenal each activate the WT channels and also the TRPA1-3C channels. This suggests there are other targets leading to activation by alkenals. In general, the peak ratios in TRPA1-3C expressing cells were a bit lower than for WT-TRPA1, but expression levels might have also been lower. Activation of TRPA1-3C channels by (E)-2-alkenals did appear to be somewhat slower than WT channels, especially in the case of (E)-2-hexenal ( Figure 5A ).

Figure 5.

(E)-2-alkenals with 6–12 carbons activate both wild-type and A1-3C human TRPA1 channels. ( A ) Response to 100 µM (E)-2-hexenal, ( B ) 100 µM (E)-2-octenal, ( C ) 100 µM (E)-2-decenal, and ( D ) 100 µM (E)-2-dodecenal. Traces are average fura-2 ratios from cells expressing WT TRPA1 channels ( black lines ) or A1-3C channels ( gray lines ). The timing of the 2-min application of 100 µM (E)-2-alkenal is indicated by the gray bars above the traces, and the application of control agonists (300 µM AITC for (E)-2-hexenal, (E)-2-decenal, and (E)-2-dodecenal; 300 µM carvacrol for (E)-2-octenal) is indicated by the black bars . Open bars in (E)-2-hexenal, (E)-2-decenal, and (E)-2-dodecenal recordings show application of 0.1% ethanol, the vehicle used for (E)-2-alkenal dilution. In all panels, lines are the averages of 48–66 cells. The (E)-2-decenal response is the same shown in Figure 2 .

For both WT TRPA1 and TRPA1-3C channels, activation was not complete after 2min in 100 µM (E)-2-hexenal and (E)-2-octenal ( Figure 5A -B). When these (E)-2-alkenals were removed, subsequent addition of 300 µM AITC or carvacrol increased the ratio further. This suggests that there are channels remaining available for activation. In contrast, AITC (and carvacrol, not shown) did not increase ratios in WT and TRPA1-3C channels after removal of the longer chain (E)-2-decenal and (E)-2-dodecenal ( Figure 5C -D). The desensitization of this control agonist response might result from Ca ²⁺ -dependent desensitization of TRPA1 channels following the high Ca ²⁺ elevation reached during (E)-2-decenal and (E)-2-dodecenal application.

There are many agonists that activate both TRPA1 and other sensory TRP channels, with TRPV1 and TRPV3 commonly cotargeted. For example, the TRPA1 agonists AITC and allicin can activate TRPV1, albeit weakly ( Macpherson et al. 2005 ; Gees et al. 2013 ). Eugenol and citral are agonists of TRPA1, TRPV1, and TRPV3 ( Xu et al. 2006 ; Stotz et al. 2008 ), and carvacrol activates TRPA1 and TRPV3 ( Xu et al. 2006 ). TRPM8 is also cotargeted, as icilin and some monoterpenoids such as menthol are agonists of both TRPA1 and TRMP8 channels ( McKemy et al. 2002 ; Story et al. 2003 ).

The fura-2 imaging in Figure 2 suggested that 100 µM (E)-2-decenal elicited variable increases in [Ca ²⁺ ] across the population of TRPV1, TRPV2, TRPV3, and TRPM8 cells. The responses in a few TRPV1 and TRPV3 cells were sustained, suggesting that (E)-2-decenal might activate these channels. To directly test whether (E)-2-decenal can activate TRPV1 and TRPV3, we recorded currents in patch clamp (we did not record from TRPV2 and TRPM8 expressing cells, which gave transient responses).

However, when we recorded TRPV1 currents under voltage clamp, we found that there was little or no activation when adding 100 µM (E)-2-decenal ( Figure 6 ). Addition of 10 µM capsaicin activated large currents, confirming ample TRPV1 expression. From 3 cells, the current at +60 mV after 60s in (E)-2-decenal was 5±2.7 pA/pF, or ~3% of the peak capsaicin-activated current. Voltage-clamp recordings of TRPV3 currents with 100 µM (E)-2-decenal also showed little or no response ( Figure 6B ). No currents were activated by (E)-2-decenal (−0.2 pA/pF relative to baseline, n = 3), compared with robust activation of TRPV3 currents by 150 µM 2-aminoethoxydiphenyl borate (2-APB) (82.9±29.2 pA/pF). The lack of (E)-2-decenal-responsive currents in patch clamp, when some cells did respond in fura-2 imaging ( Figure 2B ), might stem from the tight buffering of Ca ²⁺ in the patch clamp experiments.

Figure 6.

(E)-2-decenal and (E)-2-dodecenal are weak agonists of TRPV1 and TRPV3. ( A ) I–V curves recorded in a TRPV1-expressing cell in control, 100 µM (E)-2-decenal, and 10 µM capsaicin containing solutions. ( B ) Average current at ±60 mV during recording. Gray and black bars show the application of 100 µM (E)-2-decenal and 10 µM capsaicin, respectively. Lowercase letters indicate the timing of the I–V curves. ( C ) Fura-2 ratio traces from TRPV1-expressing cells in response to 100 µM (E)-2-dodecenal. Solid line is the average of 102 capsaicin-responsive cells, dotted lines are the response of 6 individual cells to show the range of responses. ( D – F ) Panels are as above, but TRPV3-expressing cells were used, and 150 µM 2-APB was the control agonist. Due to the strong outward rectification of 2-APB-activated TRPV3 currents, only the average at +60 mV is shown. The fura-2 ratio traces in F are the average of 117 cells carvacrol responsive cells ( solid line ) and 6 individual cells ( dotted lines ). 0 added Ca ²⁺ Tyrode’s was external solution for voltage-clamp recordings; 2 Ca ²⁺ Tyrode’s was bath solution for fura-2 experiments.

We also tested whether (E)-2-dodecenal might be more effective in activating TRPV1 and TRPV3, with fura-2 imaging ( Figure 6A ,B). The average fura-2 ratio in TRPV1 expressing cells in response to 100 µM (E)-2-dodecenal application was 1.4±0.17 ( n = 102 capsaicin-responsive cells). Within those cells, there were 11 with peaks >2.0 (range 2.02–10.35). This response was slow, as cells peaked 109±13s after addition of (E)-2-dodecenal. TRPV3 activation by 100 µM (E)-2-dodecenal was somewhat stronger, as 42 of 117 carvacrol-responsive cells reached ratios >2.0. The average peak ratio following (E)-2-dodecenal application—but before carvacrol application—was 4.7±2.0 ( n = 42). This response was again quite slow, and many cells peaked only after removal of (E)-2-dodecenal (average time to peak was 200±41s in the 42 responding cells). Thus, over most concentrations (E)-2-alkenals activate predominantly TRPA1, though high concentrations may also activate other sensory TRP channels. The weaker activation of TRPV1 and TRPV3 might be enhanced at higher temperatures, allowing for these channels to participate in alkenal detection. A similar effect has been shown for TRPV1 and AITC ( Gees et al. 2013 ).

Although many defensive secretions containing aldehydes are rich in unsaturated forms, others contain aldehydes with saturated carbon chains as major or minor constituents, or alcohols, both saturated and unsaturated. In mixtures with alkenals, the saturated aldehydes might be precursors or biosynthetic derivatives, but aldehydes are deterrent chemicals in their own right ( Blum 1981 ). TRPA1 is activated by short aldehydes, including formaldehyde, acetaldehyde, but not propanal and larger ( Bautista et al. 2006 ; Macpherson et al. 2007b ; McNamara et al. 2007 ). We tested whether decanal and hexanal can activate TRPA1, despite their lack of the α,β unsaturated bond as in (E)-2-alkenals ( Figure 7 ). About 100 µM decanal applied to TRPA1-expressing cells caused a rapid increase in [Ca ²⁺ ] _i . After reaching a peak, the average [Ca ²⁺ ] _i declined, falling 60% over ~2min This decline differs from the sustained plateau response to (E)-2-decenal ( Figure 2A ). After removal of decanal, the [Ca ²⁺ ] _i continued to decline, though it remained elevated above the initial level. Following a 2min recovery in control Tyrode’s solution, 300 µM carvacrol induced a large and rapid increase in [Ca ²⁺ ], suggesting that TRPA1 channels were still available. Presumably 100 µM decanal was not a full agonist or washed away (or both), again distinct from (E)-2-decenal ( Figure 1 ). The TRPA1-3C channel was also robustly activated by decanal ( Figure 7B ). This might be expected given that the saturated aldehyde should not modify free thiol groups. Hexanal also activated TRPA1 channels, possibly with a faster-and more uniform-onset than (E)-2-hexenal. Again, after removal, many TRPA1 channels were ready to respond to application of carvacrol, which caused a very large ratio increase.

Alcohols with the same carbon chain length as other constituents are often found among the complex mixtures of defensive secretions. One example is (E)-2-hexen-1-ol, the alcohol corresponding to (E)-2-hexenal, found in the secretion of plant bug Lopidea robinae ( Staples et al. 2002 ). We tested whether (E)-2-hexen-1-ol could activate human TRPA1 channels recorded in patch-clamp ( Figure 7D ). Application of 3mM (E)-2-hexen-1-ol elicited a slow increase in TRPA1 current, which reached steady-state after ~3min. A subsequent application of 300 µM carvacrol increased the current further. Applying a range of concentrations of (E)-2-hexen-1-ol, we found 300 µM elicited 91.2±24.4 pA/pF ( n = 8), 1mM elicited 70±7.9 pA /pF, ( n = 10) and 3mM elicited 215.9±74 pA/pF current at +60 mV ( n = 5). These current amplitudes were 26.7±2.6% ( n = 10), 36.4±7.1% and 58±3.6% ( n = 5), respectively, of the carvacrol-activated current recorded in the same cell. (E)-2-hexen-1-ol thus appears to activate human TRPA1, yet weakly.

The vertebrate trigeminal nerve innervates the nasal cavity, as well as the eyes and skin covering the head. If alkenals are acting as aversive chemicals versus vertebrates, it is likely that they are first detected by the predator’s trigeminal nerve, in the mouth, the nasal cavity, and/or eyes after being sprayed. We tested whether 2 of the chemicals we found to activate TRPA1—(E)-2-hexenal and (E)-2-hexen-1-ol—could stimulate trigeminal nerve firing ( Figure 8 ). Multiunit recordings from the ethmoid branch of the trigeminal nerve showed that indeed application of both (E)-2-hexenal and (E)-2-hexen-1-ol did cause firing of trigeminal neurons.

Figure 8.

(E)-2-hexenal and (E)-2-hexen-1-ol are trigeminal irritants and are aversive to rats. ( A ) Left : Integrated multiunit activity of ethmoid branch during application of control Ringer’s solution to the nasopharynx of a rat. Right : Breathing pattern during Ringer’s application (marked by arrowhead). ( B ) is same as A, but in response to 10mM (E)-2-hexenal and ( C ) in response to 100mM (E)-2-hexen-1-ol. Time scale is 10s in nerve recordings ( left ) and 20s in breathing responses ( right ). ( D ) Dose–response curve of relative breath duration in response to (E)-2-hexenal and (E)-2-hexen-1-ol ( n = 5 for each; bars show SEM). ( E ) Aversion assay results showing mean time in quadrant with irritant (open bars) or water control (closed bars) during a 10-min trial. Irritant zones contained dishes with either 20% (i.e., 2.04M) (E)-2-hexenal or 10% (~1M) (E)-2-hexen-1-ol ( n = 6, 8). The mean time spent in the (E)-2-hexenal and water zones was significantly different ( P < 0.05, asterisk ).

Upon detection of airborne irritants in the upper respiratory airway, mice respond by reducing the respiratory rate ( Vijayaraghavan et al. 1993 ). We simultaneously measured the respiratory rate during application of (E)-2-hexenal and (E)-2-hexen-ol to the nerve and both chemicals caused a prolonged decrease in respiratory rate. About 100mM (E)-2-hexenal stimulus led to a ~30-fold increase in breath duration and 100mM (E)-2-hexen-1-ol led to a ~10-fold increase.

How do animals respond when encountering sensory irritants such as (E)-2-hexenal? An effective defensive chemical would cause avoidance of the irritation source. We tested whether rats found (E)-2-hexenal and (E)-2-hexen-1-ol aversive by adding each, in separate experiments, to a water dish in 1 quadrant of a behavioral arena ( Figure 8D ). When we measured the amount of time rats spent in the quadrant with the chemical compared with trials with water in all 4 quadrants, rats spent much less time in the quadrant with 20% (E)-2-hexenal (i.e., 2.04M). This time difference indicates the strong aversive effect of these chemicals. The aversion due to 10% (E)-2-hexen-1-ol (i.e., ~1M) was not significant. Given the weak activation of TRPA1 channels by (E)-2-hexen-1-ol (cf. Fig. 5 ), perhaps a higher concentration would be needed to see behavioral aversion. Alternatively, there might be other compounds accompanying (E)-2-hexen-1-ol in defensive mixtures to ensure and enhance a response in the target.

Discussion

Alkenal activation of TRPA1

We find that (E)-2-alkenals, compounds found in the defensive secretions of many animals, are potent agonists of TRPA1. Alkenals with 8–12 carbon chains activated TRPA1 with EC ₅₀ s in the micromolar range. The slow time course and minimal reversibility suggest that activation occurs through modification of cysteine side-chains. The delays before activation may arise from some combination of slow (E)-2-alkenal partitioning into the membrane, their reaction with the channel, and subsequent channel opening. The shorter 6-carbon (E)-2-hexenal was a weaker agonist (EC ₅₀ ~ 100 µM). TRPA1-3C channels were also robustly activated by (E)-2-alkenals, despite lacking 3 amino-terminal cysteines important for activation by other electrophilic agonists, suggesting there are additional sites where modification can occur to activate the channel. This is consistent with recent result from Moparthi et al. (2014) , who showed that TRPA1 with the amino-terminus deleted was still activated by electrophiles. Clearly, other locations in the channel protein are involved in activation, and the recent TRPA1 structure ( Paulsen et al. 2015 ) should aid in defining their location.

(E)-2-alkenal activation of TRPA1 did not appear to cause a Ca ²⁺ -independent desensitization, similar to benzoquinone and N -ethyl-maleimide ( Raisinghani et al. 2011 ; Ibarra and Blair 2013 ). Because the desensitization described there is dependent on cysteine modification as well, it is unclear why (E)-2-alkenals do not also cause it. (E)-2-alkenals are in this way similar to AITC, which causes little or no desensitization. Future experiments on electrophiles with different shapes and locations of the electrophilic moiety may narrow down the parameters controlling desensitization.

Our results add (E)-2-alkenals to the list of natural deterrents that activate TRPA1. Thus, alkenals (from 3-carbon propenal to 12-carbon dodecenal), join pungent chemicals such as alcohols ( Komatsu et al. 2012 ), weak acids ( Wang et al. 2011 ), phenols ( Lee et al. 2008 ), polygodial and isovelleral ( Escalera et al. 2008 ), citral and citronellal ( Stotz et al. 2008 ; Kwon et al. 2010 ) and benzoquinone ( Ibarra and Blair 2013 ). Some of these compounds have α,β-unsaturated bonds and can likely activate by modifying TRPA1 cysteines. Others may reversibly bind elsewhere on the channel, perhaps near the site(s) of carvacrol, menthol and local anesthetic binding. The large chemical diversity of natural TRPA1 agonists, arising independently in multiple lineages, may reflect the value of targeting a predator’s TRPA1 channel.

Some alkenals are produced endogenously in mammalian tissues during inflammation. These alkenals include hydroxyl substituted (e.g., 4-hydroxy-(E)-2-nonenal) and ketone substituted (e.g., 4-oxo-2-hexenal) forms, which can activate TRPA1 ( Macpherson et al. 2007b ; Trevisani et al. 2007 ; Andersson et al. 2008 ). Interestingly, the (E)-2-hexenal and decenal with ketones at the 4 position are used as allomones, and are more behaviorally deterrent than the unsubstituted forms ( Eliyahu et al. 2012 ).

To affect sensory neurons, the deterrent must partition into or through the lipid bilayer to reach channel structures controlling activation. Compounds with longer carbon chains, which have higher logP values, often tend to be better agonists. Ursu et al. (2010) found such a relationship for TRPV1 and a series of 6 agonists, varying potency over 6 orders of magnitude. TRPA1 activation—by alcohols ( Komatsu et al. 2012 ) and weak acids ( Wang et al. 2011 )—and trigeminal neuron Ca ²⁺ signal activation by various organic molecules ( Inoue and Bryant 2005 , 2010 ) both had a similar relationship. This may underlie the lower nasal chemosensory thresholds of longer chain length irritants in humans ( Cometto-Muñiz and Cain 1990 ).

(E)-2-alkenals are deterrents against a broad range of predators

(E)-2-alkenals are effective deterrents to many groups of predators, including ants ( Wheeler et al. 1964 ; Eliyahu et al. 2012 ) and mantids ( Noge et al. 2012 ), fish ( Hepburn et al. 1973 ), lizards ( Krall et al. 1999 ), birds ( Krall et al. 1999 ; Staples et al. 2002 ), and mice ( O’Connell and Reagle 2002 ). The universality of this deterrence likely arises from the actions of (E)-2-alkenals chemesthetic sense ( Conner et al. 2007 ; Green 2012 ). In vertebrates, this may result from activation of TRPA1 and TRPV1 channels, yet these genes are not present in every species. Arthropods lack TRPV1-V4 and TRPM8 orthologs completely. Instead, they may rely upon an expanded TRPA subfamily, which has up to 5 members, including a TRPA1 ortholog and a painless ortholog ( Peng et al. 2015 ). Arthropod TRPA1 responds to electrophiles similarly to the vertebrate isoform via cysteine modification, and some splice variants are additionally activated by increased temperature ( Kang et al. 2010 , 2012 ; Zhong et al. 2012 ). Interestingly, the TRPA1 gene is absent in Hymenoptera. Despite this, hymenopterans retain chemesthesis, perhaps owing to a novel TRPA isoform that is activated by temperature increases and electrophilic chemicals ( Kohno et al. 2010 ). Arthropods may also use the TRPA subfamily member Painless for some chemosensory detection, though exactly how TRPA1 and Painless contribute to signaling is not clear ( Tracey et al. 2003 ; Al-Anzi et al. 2006 ).

Of the sensory-related TRP channels we tested, TRPA1 responded best to (E)-2-alkenals. TRPV4 was not activated, and TRPV1, TRPV2, TRPV3, and TRPM8 gave small and variable responses. The weak activation of these TRP channels might contribute slightly to vertebrate responses, especially if combined with other agonists (e.g., higher temperatures enhance TRPV1 responses to AITC, Gees et al. 2013 ).

Defensive secretions containing both (E)-2-alkenals and benzoquinones

Most animals that use (E)-2-alkenals for defense use primarily mixtures containing alkenals and aldehydes. But there are multiple cases where alkenals have apparently been added to defensive secretions that contain benzoquinones. Examples include the beetles Drusilla canaliculata ( Brand et al. 1973 ) and Eleodes beameri ( Tschinkel 1975 ), and 2 different millipedes ( Wheeler et al. 1964 ; Bodner and Raspotnig 2012 ). Because benzoquinones are potent activators of TRPA1 in their own right (activation thresholds ~3nM, Ibarra and Blair 2013 ), why would adding less potent (E)-2-alkenals be beneficial? Perhaps there are predators resistant to one or the other, and mixtures of allomones might ensure protection from many possible attackers.

Other behavioral effects of (E)-2-alkenals

Although (E)-2-alkenals are typically defensive chemicals, they can also elicit different behavioral outcomes. Certain predators and parasitic species hijack them to localize prey and hosts: for example, parasitic flies targeting heteropterans via (E)-2-octenal and (E)-2-decenal ( Aldrich and Barros 1995 ). How these parasitoids overcome the host (E)-2-alkenals at close range is not clear. Some (E)-2-alkenals producers use their own defensive chemicals as attractants or alarm pheromones, dispersing from a location where a conspecific has deployed their secretion ( Lockwood and Story 1987 ; Farine et al. 1997 ; Fucarino et al. 2004 ). These distinct effects might arise from the interaction of olfactory and chemesthetic senses, though the precise contributions of each are unknown. Alarm pheromones tend to be less specific than other pheromones ( Blum 1996 ), consistent with a single predominant target like TRPA1. More detailed knowledge of the chemical sensing mechanisms used by animals is needed to understand the variable effects of allomones.

Funding

This project was supported by the National Institutes of Health National Institute of Neurological Disorders and Stroke [P01NS072040-02].