Irritating channels: the case of TRPA1

Abstract

Transient receptor potential (TRP) channels have been extensively studied over the past years. Yet, in most cases, the gating mechanisms of these polymodal cation channels still remain a puzzle. Using the nociceptive channel TRPA1 as an example, we discuss the role of dynamic regulation of the pore size (pore dilatation) on channel gating. Additionally, we critically revise current knowledge of the role of intracellular domains, such as ankyrin repeats and EF hand motifs, in channel activation and function. Finally, we assess some problems inherent to activation of TRPA1 by the reaction of electrophilic compounds with the nucleophilic thiol sink of N-terminal reactive cysteines.

Bernd Nilius is Emeritus Professor at the Catholic University of Leuven (KU Leuven). After several positions in Germany, including a Group Leader position at the Max Planck Society, he was appointed as Full Professor of Physiology at the KU Leuven. He worked on many types of ion channels, discovered in 1985 the T-type Ca²⁺ channels in cardiac cells and later focused on ion channels in vascular endothelium, especially Cl^- channels. In Leuven, he initiated very successful research on TRP cation channels, from their molecular characterization, their function in native cells and their impact on human diseases. He is Editor-in-Chief of Pflügers Archiv and serve(d)s as editor for many journals including The Journal of Phyisology.

TRPA1: a puzzling pore

TRPA1 is the only mammalian member of the ‘ankyrin’ type subfamily of TRP channels (Wu et al. 2010). It is widely expressed in many cell types and tissues, including sensory neurons, hair cells, vasculature, heart, brain and pancreas (for details see the TRP channel database, Clapham et al. 2010). Similarly to other TRP channels, it comprises six transmembrane segments and multidomain intracellular N- and C-termini (Fig. 1 and TRPA1 data base).

Figure 1. Structural scheme of TRPA1. Domain structure of TRPA1 monomers.

The domains were elucidated by PROSITE at the ExPASy proteomics server (Swiss Institute of Bioinformatics). Pleckstrin homology (PH) refers to putative phosphoinositol phosphate binding motifs as defined previously (Nilius et al. 2008). CaM, calmodulin; EF hand indicates putative Ca²⁺ interaction sites.

TRPA1 is a relatively high-conductance cation channel. In cell-attached patches, the recording configuration that causes the least interference of the intracellular composition, the single channel conductance measured under physiological ionic conditions (1.5 mm Ca²⁺ and 1 mm Mg²⁺ in the extracellular patch pipette) amounts to ∼65 pS in the inward direction and 110 pS in the outward (Fig. 2). In a divalent-free solution (DVF), i–V curves are linear with a slope conductance of about 110 pS (see e.g. Karashima et al. 2008, 2010). Surprisingly, under these conditions a high open probability is seen at negative potentials and a clear voltage-dependent inactivation at positive potentials. This behaviour varies depending on the experimental conditions and may be related to the state of Ca²⁺-dependent inactivation.

Figure 2. TRPA1 single channels currents from cell-attached patches.

A, cell-attached single channel recording of TRPA1 at different membrane potentials (Note the short burst of opening at positive potentials and the high open probability at negative potentials. The size of the unitary current was measured from amplitude histograms fitted by two Gaussian distributions and plotted in the i–V curve in B. B, from the single channel i–V curves, a single channel conductance under these conditions of 71 pS for negative potentials and 110 pS for positive potentials was calculated. Mean currents were obtained from averaging between 12–36 traces as shown at the left and plotted against voltage (squares). C, dividing the single amplitude of the mean current by the amplitude of the single channel currents at all measured potentials resulted in the P_open–V plot. This curve is fitted by the Boltzmann function which describes the voltage dependence of TRPA1 inactivation. The pipette contained 1.5 mm CaCl₂ and 1 mm MgCl₂. For more technical details see Karashima et al. 2007, 2008, 2010.

Beyond the measurement of the single channel conductance, biophysical properties of the TRPA1 pore have only recently been studied. By using a set of inorganic and organic monovalent cations, we obtained information on the field strength of the pore and also an estimate of the pore's diameter. For inorganic monovalent cations, the TRPA1 cation permeability sequence, Rb⁺≥ K⁺ > Cs⁺ > Na⁺ > Li⁺, corresponds to Eisenman sequence III or IV, implying a weak field strength site. Using a set of organic cations with increasing diameter, we measured a pore diameter of 11.0 Å from the relative permeabilities fitted to the excluded volume equation

were a is the diameter of the ion and d the diameter of the pore (Karashima et al. 2010). A pore diameter of at least 10 Å was also predicted based on experiments using the bivalent dihydrostreptomycin (∼10 Å diameter, 586 Da), which blocked TRPA1 in a voltage-dependent manner but unblocks at very negative potentials. These results were in line with a model in which dihydrostreptomycin binds to a site at 72% of the electric field, and permeates the pore when the electrical field is sufficiently strong (Y. Karashima and B. Nilius, unpublished observations). Importantly, the TRPA1 pore size appears dynamically regulated, similar to what has been described for some ion channel such as P2X7 (North, 2002; Ferrari et al. 2006; Pelegrin & Surprenant, 2006, 2009). Based on the permeation of large organic fluorescent compounds like Yo-Pro-1 (YP, 629 Da), it is very likely that many TRP channels may undergo pore dilatation upon activation (TRPV1 Chung et al. 2008; TRPM8, TRPV1–4 Banke et al. 2010). Thus, the pore size depends on the activating agonist and/or on permeating or non-permeating ions (e.g. Ca²⁺, Banke et al. 2010). In our hands, mustard oil (MO) induced a pore dilatation from 11.0 to 13.8 Å (∼27%), indicating a clear effect on the channel gating (Karashima et al. 2010). Moreover, the dilated TRPA1 channel is permeable to the large fluorescent styrylpyridinium molecule FM1 43 (∼12 Å diameter, 611 Da) (Karashima et al. 2010). Importantly, pore dilatation is accompanied by an increase of the Ca²⁺ selectivity of the pore (P_Ca/P_Na from 5.7 to 7.9) and an increased fractional Ca²⁺ current through TRPA1 (from 17.0 to 23.3%). In addition, we also observed an increase of outward single channel conductance ranging from ∼110 to 130 pS.

Banke et al. (2010) described that the dilatation of the TRPA1 pore is under the strong control of extracellular Ca²⁺. An increase of extracellular Ca²⁺ concentration ([Ca²⁺]_o) leads to a more restricted pore whereas a removal of [Ca²⁺]_o forces the channel to spend more time in the dilated state. This hypothesis is intriguing but can probably be also explained by an interference with Ca²⁺-dependent inactivation. Thus, loading of fluorescent dye FM1 43 might be not only though TRPA1 but could also occur by permeation through hemichannels such as pannexin 1 (Pelegrin & Surprenant, 2006). We cannot confirm yet a striking effect of [Ca²⁺]_o on the pore size. As a matter of fact, the presence of Ca²⁺ in the patch pipette reduces the single channel conductance in the inward direction. This reduction disappears in DVF solution but does not allow the conclusion that Ca²⁺ may cause this effect by favouring pore constriction. In consequence of this TRPA1 pore dilatation mechanism the channel may attain a completely different pharmacology by allowing larger charged or uncharged compounds to enter the dilated pore and to cause pore block. This intriguing mechanism still remains a mystery.

Biophysically, there is no mandatory link between pore size and single channel conductance (γ). Changes in γ are caused by several unexpected interventions. TRPA1 activation (especially in cell-excised patches) may critically depend on the presence of small intracellular compounds such as polyphosphates (PPP_i), which are supposed to bind to the ankyrin repeat domain (ARD) (Kim & Cavanaugh, 2007; Cavanaugh et al. 2008; Karashima et al. 2008). In our hands, application of PPP_i to inside-out patches not only stabilizes channel activity but also increased the single channel conductance (from 115 ± 6 pS to 139 ± 9 pS at +60 mV, 5 mm PPP_i, n = 5, Fig. 3A and B) (B. Nilius and J. Prenen, unpublished data). Thus, our results indicate a structural interaction with both pore and ARD.

Figure 3. TRPA1 single channel currents from an inside-out patch before and after application of PPP_i and from a cell-attached patch expressing the pore mutant D918Q.

A, single channel measurement from an inside-out patch immediately after excision (wild type TRAP1, pipette 0 Ca²⁺, 5 Mg²⁺, +80 mV). B, same conditions as in A. The inner site of the membrane is exposed to 5 mm PPP_i. C, cell-attached measurement of single channel currents through the less Ca²⁺-permeable D918Q mutant (pipette solution as in A, +80 mV). Note that all experiments were performed with a sampling rate of 2.5 kHz and a filtering of 1 kHz. Although it cannot be excluded that very fast openings are not completely resolved, the striking difference to the channel behaviour in A and B is obvious.

There is general agreement that the Ca²⁺ selectivity of the pore is dramatically affected by mutations of a conserved aspartate residue, Asp⁹¹⁸, in the putative selectivity filter (Wang et al. 2008). Our work also supports the dominant role of Asp⁹¹⁸. Different charge-neutralizing mutants, D918A, D918C and D918Q, exhibit a significantly reduced relative Ca²⁺ permeability (Wang et al. 2008), with P_Ca/P_Na values of 0.47 ± 0.03, 1.06 ± 0.06 and 1.04 ± 0.08, respectively (Karashima et al. 2010). We found that mutations at Asp⁹¹⁸ also affect the single channel conductance of TRPA1, which decreases from 111 ± 7 pS for the wild type to 37 ± 5 pS for D918Q (0 Ca²⁺, 5 mm Mg²⁺ in the pipette, +60 mV, B. Nilius and J. Prenen, unpublished observations, Fig. 3C). In contrast to wild type TRPA1, MO stimulation of these pore mutants led to a decrease rather than an increase in relative Ca²⁺ or Mg²⁺ permeability (Karashima et al. 2010). The openings of the mutant channels are only very short (long openings are missing), suggesting that pore mutations destabilize the open pore. Several pertinent questions remain unanswered: Does Ca²⁺ really control pore dilatation? Do non-electrophilic activators induce dilatation? Is there a certain set of new molecules which can block the channels in the dilated but not in the restricted mode?

The ankyrin repeat domain

TRPA1 contains an N-terminal ankyrin repeat domain (ARD) constisting of at least 14 ankyrin repeats (Story et al. 2003). So far, the role of unusually long ARD is not known, although a function in mechano-sensing and channel gating has been proposed (Nagata et al. 2005; Sotomayor et al. 2005; Lee et al. 2006). Very probably, it constitutes a region for interaction with other proteins (Gaudet, 2008) and may form multi-ligand binding sites, as has been found for TRPV channels (Lishko et al. 2007). To assess the role of ARD in regulation of TRPA1 channel function, we constructed mutant channels by truncations of ankyrin repeats in ARD, Δ1–4 and Δ1–12 (deletion of either first 4 or 12 ankyrin repeats, respectively) (Supplemental Fig. S1). Immunocytochemistry experiments show that transiently expressed wild type TRPA1 channels in HEK293 cells co-localize with cadherins, plasma membrane markers involved in cell adhesion (authors’ unpublished observations, Supplementary Fig. S1), and that overexpression of TRPA1 results in an accumulation of cadherins in intracellular membranes together with TRPA1. Possible TRPA1–cadherin interactions had already been hypothesized as one of the potential mechanisms for the putative role of TRPA1 in mechanosensation (Sotomayor et al. 2005). In contrast to wild type TRPA1, ARD truncation mutants did not reach the plasma membrane but still co-localized with cadherins on intracellular membranes. Electrophysiological measurements further showed full loss of activity of both TRPA1 mutant channels, Δ1–4 and Δ1–12 (J. Prenen and B. Nilius, unpublished data, not shown). These results suggest that the intact ARD is necessary for plasma membrane expression of TRPA1, but not for the putative interaction with cadherins. Clearly, more experiments are needed to further our understanding of the role of the ARD and its interaction with other proteins.

Ca²⁺ as a bimodal gating modifier

Ca²⁺ strikingly modulates TRPA1 activity. Micromolar intracellular Ca²⁺ concentrations ([Ca²⁺]_i) activate TRPA1, and also elevation of [Ca²⁺]_o can transiently increase the channel activity. It has been suggested, but is challenged below, that activation may depend on Ca²⁺ binding to an N-terminal EF hand motif (Doerner et al. 2007; Zurborg et al. 2007). In addition to an activating effect, Ca²⁺ also exerts inhibition of TRPA1 (Nagata et al. 2005; Wang et al. 2008). In our view, both effects are not understood so far. Stepwise increases of [Ca²⁺]_o cause a fast transient activation followed by an inactivation. At 30 mm[Ca²⁺]_o, TRPA1 is completely inactivated and cannot be activated by electrophilic compounds. In contrast, at 5 mm[Ca²⁺]_o TRPA1 is transiently activated by MO. In the absence of [Ca²⁺]_o activation by MO is delayed and slow while inactivation is nearly attenuated. This delay of activation and the reduced inactivation are also seen for the less Ca²⁺-permeable TRPA1 mutants (e.g. D918A) in the presence of [Ca²⁺]_o. Thus, these pore mutants behave like TRPA1 in the absence of [Ca²⁺]_o. Wang and colleagues tested this Ca²⁺-dependent phenomenon in more detail (Wang et al. 2008). Activation of TRPA1 by cinnamaldehyde (CA) in the absence of [Ca²⁺]_o induces slow activation. If [Ca²⁺]_o is increased, a short phase of additional activation is followed by a nearly completely inactivation ([Ca²⁺]_o 2 mm). The speed of activation and inactivation depends on [Ca²⁺]_o, e.g. low concentrations delay both activation and inactivation whereas high concentrations accelerate these processes. Intriguingly, if [Ca²⁺]_i is increased by uncaging of DMNP-EDTA, only activation of TRPA1 is observed but no inactivation. Similar to this result, we also observed that dialysing the cell with Ca²⁺-buffered solution (10 μm) increases the current size, but did not cause an acceleration of inactivation. All these data suggest that Ca²⁺-dependent activation may depend on a rise of [Ca²⁺]_i whereas Ca²⁺-dependent inactivation requires Ca²⁺ entry, suggesting that structural domains close to the channel pore might be involved in this mechanism.

The pertinent question is how Ca²⁺ regulates TRPA1. It has been suggested that an EF hand (like) domain in the N-terminus between the 11th and 12th ankyrin repeat (N⁴⁶⁹ISDTRLLNEGDL⁴⁸¹, mouse) is responsible for Ca²⁺-dependent activation (Doerner et al. 2007; Zurborg et al. 2007). In our hands, a deletion of the whole EF hand like domain (Δ460–480) renders the channel non-functional. Most probably, and similar to ARD-truncated variants which do not form functional channels, this deletion prevents plasma membrane insertion of the mutant channel. Interestingly, point mutations D469A and D480A still show functional channels but with significantly decreased current amplitude (431 ± 58 pA pF⁻¹ wild type TRPA1, 42 ± 11 pA pF⁻¹ D480A, 126 ± 26 pA pF⁻¹ D469A, n = 6, at +80 mV). Intriguingly, the D469A mutant can still be activated by Ca²⁺. Step-wise increase of [Ca²⁺]_o from 1.5 to 10 mm Ca²⁺ increases the current at +80 mV from 88.9 ± 29.9 to 201.7 ± 44.2 pA pF⁻¹ (n = 6, J. Prenen and B. Nilius, unpublished data). Thus, the mechanism of Ca²⁺-dependent regulation of TRPA1 still needs to be fully elucidated and understood.

An intriguing alternative hypothesis suggests that modulation of TRPA1 by Ca²⁺ is regulated by TRPV1. In inside-out patches, an increase of [Ca²⁺]_i triggered activation of TRPA1 and this stimulation is attenuated by co-expression with TRPV1 only in the presence of 2 mm[Ca²⁺]_o but not in Ca²⁺-free conditions. Moreover, as shown by TRPV1 mutagenesis, this modulation mechanism depends on the Ca²⁺ permeability of the TRPV1 channel (Patil et al. 2010). Thus, these results question again the necessity of the EF hand domain for regulation of TRPA1 by Ca²⁺.

Ca²⁺-dependent activation and inactivation are especially intriguing for several reasons. First, any screening for TRPA1 agonist or antagonist has to consider a possible interference with Ca²⁺ entry. Second, as long as Ca²⁺-dependent modulation is not understood, it might be difficult to evaluate other mechanisms of gating. Third, Ca²⁺-dependent modulation might be an extremely useful target for pharmacological modulation of the channel.

TRPA1 gating by covalent cystein modification

Chemical activators of TRPA1 include electrophilic compounds like isothiocyanates (the pungent compounds in MO, wasabi and horseradish) (Bandell et al. 2004; Jordt et al. 2004), methyl salicylate (in wintergreen oil) (Bandell et al. 2004), cinnamaldehyde (in cinnamon) (Bandell et al. 2004), allicin and diallyl disulphide (in garlic) (Bautista et al. 2006; Macpherson et al. 2007), acrolein (an irritant in vehicle exhaust fumes and tear gas) (Bautista et al. 2006; Andersson et al. 2008) or endogenous compounds such as H₂O₂, the alkenyl aldehydes 4-hydroxynonena, 4-oxo-nonenal, 4 hydroxyhexenal and the cyclopentenone prostaglandin, 15-deoxy-δ(12,14)-prostaglandin J(2) (15d-PGJ(2) (Trevisani et al. 2007; Andersson et al. 2008), which all activate the channel by covalent modification of cysteines and lysines in the N-terminus (Hinman et al. 2007; Macpherson et al. 2007). The number of activating compounds is steadily increasing. Other compounds like Δ9 tetra-hydrocannabinol (1-HΔ⁹-THC, the psychoactive compound in marijuana (Jordt et al. 2004), menthol and menthol analogues (Karashima et al. 2007), nicotine (Talavera et al. 2009), clotrimazole (Meseguer et al. 2008), dihydropyridines (Fajardo et al. 2008), and general anaesthetics (Matta et al. 2008) activate TRPA1 via another, still unknown, mechanism and mostly show a bimodal action, i.e. inhibition at higher concentrations.

Covalent thiol trapping of N-terminal reactive cysteines by electrophilic compounds seems to be the most powerful mechanism to activate TRPA1. Although this very promiscuous activation of TRPA1 is widely discussed for many physiological and pathophysiological applications, we still know very little about the activation mechanism itself.

Covalent modification of cysteine can occur in several chemical ways such as thiocarbamate forming (e.g. MO), Michael adducts (e.g. enones), disulfides form protein cystein–disulfide products, or alkylation (e.g. iodoacetamides) (Cebi & Koert, 2007). It is still unknown whether different modifications have different effects on channel gating.

Thiol trapping can vary from ‘fast and reversible’ to ‘slow and irreversible’. Thus, Michael acceptors can be sorted into reversible and irreversible thiol sinks. These reaction types can be easily changed by pH (Shi & Greaney, 2005). It is not clear which other possibilities may change the life time of TRPA1 covalent cysteine products. Irreversible electrophiles might be trapped in cell surface layers (like the mucus), whereas reversible electrophiles might escape from trapping. Therefore, the knowledge of the reaction mode of electrophilic compounds would be essential for understanding diverse biological effects. In addition, electrophiles have different lipophilicity. Lipophilic compounds may be rapidly reabsorbed and may easily cross membrane barriers (e.g. passing the blood–brain barrier via a nasal route), whereas electrophiles with a certain water solubility (such as MO) may act more locally. In addition, there is probably a huge biological difference between volatile and non-volatile electrophiles.

Although several cysteines required for channel activation have been identified by mutagenesis (e.g. in AR 11th Cys⁴¹⁵, Cys⁴²², and distal of the ARD Cys⁶²², mouse (Macpherson et al. 2007), and a different set in human (Caterina, 2007; Hinman et al. 2007)), it is not known whether other nucleophilic sinks may contribute to channel activation.

As discussed above, electrophilic modification of N-terminal cysteines in TRPA1 causes dilatation of the permeation pore. It is intriguing that almost all non-electrophilic activators of TRPA1, such as menthol, nicotine, clotrimazol, THC, etc., which have completely different structures, inhibit MO-activated currents, via a mechanism that functionally resembles a pore block. It is completely unknown how these non-electrophilic compounds activate TRPA1, or whether one promiscuous or several more selective binding sites exist. It might be possible that many of these compounds enter the pore (especially in the dilated mode) thereby blocking TRPA1. This raises the question whether similar binding sites as for activation might also be located in or close to the pore.

Conclusions

This short review lists some serious gaps in understanding the permeation and gating behaviour of TRPA1. In the last few years TRPA1 has become a crucial player in many physiological and pathological processes. Moreover, it is also considered as an important target for pharmacological applications. We believe that profound understanding of TRPA1 function is of vital interest not only for fundamental science but also for fine tuning of treatments against TRPA1-related disorders. Curiously, TRPA1 might still belong to the best-studied TRP channels which highlights the ‘state of infancy’ concerning our knowledge on the TRP ‘personalities’, and sketches the lack of an ‘in depth understanding’ of channel permeation, gating, modulation, function in diverse tissues and their role in pathogenesis.

Glossary

Abbreviations

ARD

ankyrin repeat domain

DVF

divalent-free solution

PPP_i

polyphosphates

Author contributions

B.N. drafted the concept of this work, designed and wrote the manuscript. G.O. revised the manuscript, performed and analysed the immuno-histochemistry ARD data. J.P. performed the additional experiments. All authors approved the final version.

Supplementary material

SUPPLEMENT 1

SUPPLEMENT 2

As a service to our authors and readers, this journal provides supporting information supplied by the authors. Such materials are peer-reviewed and may be re-organized for online delivery, but are not copy-edited or typeset. Technical support issues arising from supporting information (other than missing files) should be addressed to the authors