Intracellular TRPA1 mediates Ca²⁺ release from lysosomes in dorsal root ganglion neurons

The transient receptor potential (TRP) superfamily is composed of a large number of cation channels with diverse functions and expression patterns in mammalian systems (Wu et al., 2010; Montell, 2011). TRPA1 is the sole member of the TRPA subfamily in mammals. It is highly expressed in a subset of primary sensory neurons in the dorsal root and trigeminal ganglia that are known to function in nociception (Story et al., 2003; Jordt et al., 2004). TRPA1 channels are selectively activated by various pungent chemicals, environmental irritants, and endogenous ligands, including mustard oil (allyl isothiocyanate [AITC]), acrolein, and 4-hydroxynonenal (4-HNE; Bandell et al., 2004; Bautista et al., 2006; Trevisani et al., 2007; Andersson et al., 2008; Ruparel et al., 2008; Karashima et al., 2009; Hu et al., 2010). Furthermore, inflammatory factors such as bradykinin indirectly activate TRPA1 channels by activating PLC (Bandell et al., 2004). This uniquely broad agonist profile along with its regulatory properties make it possible for the TRPA1 channel to function in thermosensation, environment-irritant sensing, and nociceptive sensation (Story et al., 2003; Bandell et al., 2004; Bautista et al., 2006; Kwan et al., 2006; Macpherson et al., 2007; McNamara et al., 2007).

The TRPA1 channel has been shown to function on the plasma membrane, and nociceptive signals induce its translocation from a cytosolic pool to the plasma membrane (Bautista et al., 2006; Schmidt et al., 2009). TRPA1 channel activation increases sensory neuron excitability and the intracellular Ca²⁺ concentration ([Ca²⁺]_i; Jordt et al., 2004). It has been shown that ∼17% of the TRPA1 current is mediated by Ca²⁺ influx (Karashima et al., 2010). In contrast to the extensive studies of TRPA1 activators/modulators and the mechanisms of regulation in heterologous expression systems, the regulation and sources of the TRPA1-induced Ca²⁺ signal in dorsal root ganglion (DRG) neurons and other native cells remain largely unknown. In this study, we systematically investigated the Ca²⁺ signal triggered by TRPA1 activation, the mechanisms by which it regulates neuropeptide secretion in mouse DRG neurons, and its relevance to nociceptive sensation.

Although the critical roles of TRPA1 channels in thermosensation, nociception, and environmental-irritant sensing are well known, details about the mechanisms of the TRPA1-mediated [Ca²⁺]_i rise in DRG neurons remained to be elucidated. We therefore monitored [Ca²⁺]_i changes in mouse DRG neurons stimulated with a TRPA1 agonist, AITC. Low concentrations of AITC are specific for TRPA1 (Bautista et al., 2006; Everaerts et al., 2011), whereas high concentrations (>0.1 mM) may activate TRPV1 as well (Everaerts et al., 2011). To ensure that 100 µM AITC, which was used for all of the cellular experiments, specifically activates TRPA1 in DRG neurons, we performed confocal microscopic Ca²⁺ imaging for unbiased screening of AITC-responsive DRG neurons in wild-type (WT) and TRPA1-knockout (KO) mice. Consistent with findings that only a subset of small-diameter DRG neurons express TRPA1 channels (Story et al., 2003), AITC (100 µM) induced a [Ca²⁺]_i rise in 33.5% of Fluo-4–loaded DRG neurons from WT, but not TRPA1-KO, mice (Fig. S1 A). Furthermore, the AITC-induced [Ca²⁺]_i rise remained largely unchanged in TRPV1-KO neurons, indicating that TRPV1 was not involved in the AITC responses in DRG neurons under our experimental conditions (unpublished data). We then quantified the TRPA1-mediated [Ca²⁺]_i rise in Fura-2–loaded neurons by photometric measurements (one cell per measurement). To avoid the desensitization of TRPA1 channels, we first screened for TRPA1-positive DRG neurons by Ca²⁺ imaging using a low concentration (25 µM) of AITC. Those that responded were chosen for subsequent experimental tests. Under these conditions, the [Ca²⁺]_i rise evoked in WT DRG neurons by 100 µM AITC was completely blocked by the specific TRPA1 antagonists HC-030031 and A-967079 (Fig. S2, A and B). The endogenous TRPA1 agonist acrolein (Bautista et al., 2006; Due et al., 2014) also induced a [Ca²⁺]_i increase in WT DRG neurons (Fig. S2 D). In contrast, neither AITC nor acrolein induced any change in [Ca²⁺]_i in TRPA1-KO neurons despite the robust response to capsaicin, a TRPV1 agonist (Fig. S2, C and D). These findings demonstrated that under our experimental conditions, AITC and acrolein induced a [Ca²⁺]_i rise by activating TRPA1 channels in DRG neurons.

TRPA1 is commonly thought to form plasma membrane channels that mediate Ca²⁺ influx. Surprisingly, application of 100 µM AITC in a Ca²⁺-free (0 mM) external solution also led to a [Ca²⁺]_i rise in most of the capsaicin-sensitive WT DRG neurons, whereas little [Ca²⁺]_i change was detected in neurons pretreated with A-967079 (Fig. S3, A and B) or in those from TRPA1-KO mice (Fig. 1, A and B). Similarly, acrolein (Fig. S3, C and D) and another endogenous TRPA1 agonist, 4-HNE (Fig. S1 B; Trevisani et al., 2007; Andersson et al., 2008; Akude et al., 2010; Lupachyk et al., 2011), also triggered a [Ca²⁺]_i rise in WT, but not TRPA1-KO, DRG neurons in the Ca²⁺-free external solution. The [Ca²⁺]_i rise should be completely contributed by intracellular Ca²⁺ sources because 1 mM EGTA was also included in the Ca²⁺-free bath solution, and 100 mM K⁺ failed to trigger any [Ca²⁺]_i rise under this condition (unpublished data). Furthermore, ∼30.4% of DRG neurons were still responsive to AITC in the Ca²⁺-free condition with 2 mM BAPTA (Fig. 1, E and F), but were completely blocked by preloading with a membrane-permeable BAPTA-AM (Fig. 1, G and H). Thus, in addition to mediating Ca²⁺ influx through channels on the plasma membrane, activation of TRPA1 also mobilized Ca²⁺ from intracellular stores in DRG neurons. The AITC-induced Ca²⁺ store release was dependent on AITC concentration and saturable (Fig. 1, C and D). In AITC-positive DRG neurons bathed in the Ca²⁺-free solution, application of AITC at 5, 20, 50, 100, and 300 µM caused increasing degrees of [Ca²⁺]_i elevation (Fig. 1, C and D). However, when DRG neurons were treated twice consecutively with 100 µM AITC in the Ca²⁺-free solution, the second application failed to trigger a discernible [Ca²⁺]_i rise (unpublished data). These results indicated that the Ca²⁺ stores affected by TRPA1 activation were of a limited size and almost completely depleted by the first treatment with 100 µM AITC.

The ER is the major intracellular Ca²⁺ storage pool and responsible for most of the Ca²⁺ mobilization triggered by receptor signaling (Berridge, 2005). To investigate the possible involvement of the ER in mobilizing Ca²⁺ in response to TRPA1 activation, we first emptied the ER store by perfusion with Ca²⁺-free external solution containing 2.5 µM thapsigargin (TG), a blocker of sarcoplasmic/ER Ca²⁺/ATPases, before AITC was applied. However, this treatment did not notably change the AITC-induced release of stored Ca²⁺ (Fig. 2, A and B). Cyclopiazonic acid (CPA), another sarcoplasmic/ER Ca²⁺/ATPase blocker, has been widely used to reversibly deplete the ER Ca²⁺ store (Seidler et al., 1989; Emptage et al., 2001). Similarly, preincubation of cells with Ca²⁺-free solution containing 5 µM CPA also failed to affect the AITC-induced Ca²⁺ release (Fig. 2, C and D). Both TG and CPA triggered a [Ca²⁺]_i rise on their own because of Ca²⁺ leakage from the ER (Fig. 2, A and C). The Ca²⁺ ionophore ionomycin has been shown to preferentially target the ER membrane at low concentrations (up to 1 µM) but also affects other membranes at higher concentrations (e.g., 10 µM; Foyouzi-Youssefi et al., 2000). Therefore, we tested the effects of pre-emptying the internal stores with 1 or 10 µM ionomycin on AITC-induced Ca²⁺ release in DRG neurons. Both concentrations elicited a robust [Ca²⁺]_i rise when applied in the Ca²⁺-free external solution. Consistently, 1 µM ionomycin failed to alter the effect of AITC on Ca²⁺ mobilization (Fig. 2 E), but 10 µM abolished the Ca²⁺ store release induced by TRPA1 activation (Fig. 2 F). These results demonstrated that the stores mobilized by TRPA1 activation did not reside in the ER.

Mitochondria are known to contain Ca²⁺ and participate in intracellular Ca²⁺ signaling (Brocard et al., 2001; Grienberger and Konnerth, 2012), so we next investigated whether mitochondria were involved in the Ca²⁺ mobilization in response to TRPA1 activation. Although the addition of carbonylcyanide-ρ-(trifluoromethoxy)-phenylhydrazone (FCCP; 3 µM), a mitochondrial proton gradient uncoupler that depletes the mitochondrial Ca²⁺ content (Brocard et al., 2001) in the Ca²⁺-free solution caused a sustained [Ca²⁺]_i increase, it failed to affect the subsequent Ca²⁺ release in response to AITC (Fig. S3, E and F), indicating that mitochondria were not involved. A rise in [Ca²⁺]_i has been shown to activate exogenous TRPA1 channels located in the plasma membrane of HEK293 cells (Zurborg et al., 2007). However, Ca²⁺ release from the ER store induced by TG, CPA, and FCCP had no observable effect on the AITC-induced intracellular Ca²⁺ release in DRG neurons (Fig. 2 and Fig. S3, E and F), which might be because of the different recording conditions or the different properties of exogenously expressed TRPA1 channels in HEK293 cells and the endogenous TRPA1 channels in DRG neurons.

Because secretory granules (i.e., secretory vesicles and lysosomes) have also been shown to contain Ca²⁺ and participate in intracellular Ca²⁺ signaling (Haller et al., 1996; Mitchell et al., 2001; Churchill et al., 2002; SantoDomingo et al., 2010), we tested whether acidic organelles play a role in the Ca²⁺ mobilization induced by TRPA1 activation. Bafilomycin A1, a vacuolar-type H⁺-ATPase blocker, blocks the acidification of acidic organelles and thus causes Ca²⁺ depletion from these stores (Dröse and Altendorf, 1997; Sankaranarayanan and Ryan, 2001). Interestingly, after pretreatment with bafilomycin A1 (100 nM for 15 min) in the Ca²⁺-free solution, AITC failed to trigger a [Ca²⁺]_i increase in DRG neurons (Fig. 3, A and B), whereas the subsequent application of 2.5 mM Ca²⁺ induced a robust [Ca²⁺]_i rise by the preactivation of TRPA1 channels in the plasma membrane during AITC stimulation (Fig. 3, A and B). Preincubation with 10 µM ruthenium red (RR) blocked the [Ca²⁺]_i rise through plasma membrane-located TRPA1 (unpublished data). These findings suggest that acidic organelles were essential for the AITC-induced internal Ca²⁺ release. The lysosome is a major acidic organelle. To determine whether lysosomes are critical for the internal Ca²⁺ release evoked by TRPA1 activation, we treated DRG neurons with glycylphenylalanine 2-naphthylamide (GPN; 200 µM for 30 min), a cathepsin C substrate that selectively induces the osmotic lysis of lysosomes (Jadot et al., 1984; Shen et al., 2012). As expected, 200 µM GPN induced a transient [Ca²⁺]_i rise in DRG neurons bathed in the Ca²⁺-free solution, and the fluorescent puncta of preloaded Lysotracker Red DND-99 (1 µM; 30 min) diminished after GPN incubation (Fig. 3, C and D). Strikingly, disrupting lysosomes with GPN abolished the AITC-induced [Ca²⁺]_i rise in DRG neurons in the Ca²⁺-free solution (control, 53 ± 6 nM; GPN, 1.0 ± 0.7 nM; P < 0.001; Fig. 3, E and F), implying that TRPA1 activation induced Ca²⁺ release from lysosome-like organelles. Consistently, GPN pretreatment also substantially reduced the AITC-induced [Ca²⁺]_i rise in external solution containing the normal Ca²⁺ (2.5 mM), and this was completely blocked by pretreatment with 10 µM RR (Fig. 3, G and H).

To establish the presence of TRPA1 channels in lysosomes, we examined their subcellular localization in DRG neurons by immunofluorescent labeling of TRPA1 and lysosome-associated membrane protein 1 (LAMP1) or vesicle-associated membrane protein 2 (VAMP2). The specificity of the TRPA1 antibody was confirmed by comparing the labeling of DRG neurons from WT and TRPA1-KO mice. The TRPA1 antibody labeled 26% of the DRG neurons from WT (32 out of 122), but only 2% (4 out of 161) of those (with weak signals) from TRPA1-KO mice (Fig. S4 A). All neurons from WT and TRPA1-KO mice were positively stained by the LAMP1 antibody (Fig. S4 A). Not surprisingly, we found both peripheral and intracellular punctate staining of TRPA1 in WT DRG neurons (Fig. 4 A). Strikingly, TRPA1 mainly colocalized with the lysosome marker LAMP1, especially near the plasma membrane, but was completely separate from the vesicle marker VAMP2 (Fig. 4, A and B; and Fig. S4 B), suggesting the localization of intracellular TRPA1 channels in lysosome-like organelles. Using immunogold labeling and EM, we also detected TRPA1 on the same type of organelles labeled by the LAMP1 antibody (Fig. 4 C). TRPA1 was also found on the plasma membrane, but not in clear vesicles or clathrin-coated pits (Fig. 4 D). Thus, the intracellular Ca²⁺ store targeted by TRPA1 channel activation is most likely associated with lysosomes.

Recently, several groups including ours have directly recorded ionic currents from enlarged endolysosomal vacuoles isolated from cultured fibroblasts, HEK293 cells, and macrophages after treatment with vacuolin-1, a drug that causes endolysosome fusion (Dong et al., 2008; Cang et al., 2013; Feng et al., 2014). We also attempted to isolate vacuoles from vacuolin-treated DRG neurons for whole-endolysosome recording of TRPA1 currents. Unfortunately, we found that although vacuolin-1 generated enlarged vacuoles normally in HEK293 cells, it failed to do so in DRG neurons (Fig. S5), making it impossible to obtain endolysosomal vacuoles suitable for electrophysiological recording. In addition, overexpression of exogenous TRPA1 in HEK293 cells failed to reproduce AITC-induced internal Ca²⁺ release (unpublished data), probably because of their nonneuronal properties or a different protein-modification system in these cells. Thus, endolysosome patch clamp with HEK293 cells could not be used to confirm the functional expression of TRPA1 in lysosomes.

We therefore chose another method to directly monitor AITC-evoked Ca²⁺ release from lysosomes in DRG neurons. We performed confocal line-scan Ca²⁺ imaging with a protocol similar to that of sarcoplasmic reticulum Ca²⁺ sparks (Cheng and Lederer, 2008). The DRG neurons were transfected with LAMP1-EGFP to label endolysosomes (Fig. 5 A). To capture local Ca²⁺ signals, the neurons were loaded with Rhod2 as the Ca²⁺ indicator and EGTA as an intracellular Ca²⁺ buffer. Strikingly, in the Ca²⁺-free solution, 100 µM AITC induced marked Ca²⁺ transients in EGFP-labeled regions (lysosome Ca²⁺ sparks), representing Ca²⁺ release from single endolysosomes. In contrast, only minimal Ca²⁺ signals were detected in EGFP-negative regions (Fig. 5, B and C). Consistent with the fact that only a subset of DRG neurons express TRPA1, AITC failed to trigger Ca²⁺ release from individual endolysosomes in ∼67% of DRG neurons. These results provided direct evidence that indeed the Ca²⁺ signals arose from lysosome-like organelles upon activation of intracellular TRPA1 channels in DRG neurons.

Previous studies have established that TRPA1 channels mediate Ca²⁺ influx from extracellular space (Karashima et al., 2010). Indeed, a more robust [Ca²⁺]_i increase was detected in DRG neurons when AITC (100 µM) was applied in the normal Ca²⁺-containing solution than in the Ca²⁺-free solution (Fig. 6 A, top traces). To determine whether TRPA1 activation still evoked Ca²⁺ release from the internal store(s) in DRG neurons bathed with normal Ca²⁺-containing solution, we performed the same experiments but with RR included in the bath. RR is a membrane-impermeable inhibitor that blocks many types of cation channels, including TRPA1 (Story et al., 2003; Hu et al., 2010), and 10 µM RR completely blocks TRPA1 localized on the plasma membrane (Bandell et al., 2004; Jordt et al., 2004). In DRG neurons, pretreatment with 10 µM RR (7 min) significantly reduced, but did not abolish, the [Ca²⁺]_i rise induced by TRPA1 activation in normal Ca²⁺-containing solution. The AITC-induced [Ca²⁺]_i elevation was decreased from 408 ± 26 nM in the absence of RR to 56 ± 10 nM in its presence (Fig. 6, A and D). Consistently, the amplitude of the [Ca²⁺]_i rise in the presence of RR was similar to that in the Ca²⁺-free external solution (Fig. 6, A and D).

In addition, pretreatment of DRG neurons with 10 µM RR did not affect the AITC-induced [Ca²⁺]_i rise in the Ca²⁺-free external solution (Fig. 6, A and D), consistent with the view that RR is membrane-impermeable and therefore unable to inhibit the [Ca²⁺]_i rise resulting from internal store release. Collectively, these results showed that activation of TRPA1 channels in normal Ca²⁺-containing external solution induced a [Ca²⁺]_i rise through both extracellular Ca²⁺ influx and Ca²⁺ release from internal stores. It seems that RR selectively blocked the Ca²⁺ influx without affecting the Ca²⁺ store release, effectively separating the two processes in the normal Ca²⁺-containing solution. Given that RR not only blocks Ca²⁺ influx but also inhibits the activity of TRPA1 channels expressed on the plasma membrane, these results further suggested that the Ca²⁺ release induced by TRPA1 activation is independent of not only Ca²⁺ influx from the plasma membrane but also other activities associated with the plasma membrane TRPA1 channels (e.g., Na⁺ influx and membrane depolarization).

Activation of TRPA1 channels also triggers membrane depolarization, which may in turn activate voltage-dependent Ca²⁺ channels, an additional means by which Ca²⁺ influx is mediated. Although RR has been shown to block voltage-dependent Ca²⁺ channels in synaptosomes from rat cortex, the affinity appears to be low, with a 50% inhibitory concentration of 70 µM (Hamilton and Lundy, 1995). Consistently, pretreatment (7 min) with 10 µM RR inhibited only 15% of the depolarization-induced [Ca²⁺]_i rise in DRG neurons (unpublished data), suggesting that the plasma membrane TRPA1 channels were the main target of RR under our experimental conditions.

Because both Ca²⁺ influx and internal Ca²⁺ release contribute to the [Ca²⁺]_i rise elicited by TRPA1 activation, we next investigated the ratio of the internal store contribution to the overall Ca²⁺ signal in response to TRPA1 activation in DRG neurons in normal Ca²⁺-containing extracellular medium (P_TRPA1-store). We used ΔFd analysis to estimate the percentage of Ca²⁺ arising from internal store release in the total Ca²⁺ rise. ΔFd analysis has been widely used to determine fractional Ca²⁺ currents through TRP and other nonselective cation channels (Schneggenburger et al., 1993; Zeilhofer et al., 1997; Yu et al., 2004; Karashima et al., 2010; also see Materials and methods). When most of the increased Ca²⁺ is bound by Fura-2, ΔFd is an accurate measure of the [Ca²⁺]_i rise (Zhou and Neher, 1993).

We measured the F₃₄₀ and F₃₈₀ signals and obtained ΔFd by subtracting F₃₈₀ from F₃₄₀ (Yu et al., 2004) for TRPA1-positive DRG neurons stimulated by 100 µM AITC in the presence and absence of extracellular Ca²⁺ (Fig. 6 A, bottom traces). The AITC-induced F₃₈₀ and ΔFd signals from cells bathed in external solution containing either 2.5 or 0 mM Ca²⁺ are shown in Fig. 6 B. From the Fd values, P_TRPA1-store was determined to be ∼40% (ΔFd: 0.25 ± 0.04 arbitrary units [a.u.] in 2.5 mM Ca²⁺ and 0.10 ± 0.01 a.u. in 0 mM Ca²⁺; Fig. 6, C and E), meaning that internal store release accounts for ∼40% of the overall [Ca²⁺]_i rise resulting from TRPA1 channel activation in mouse DRG neurons. Supporting this estimate, the AITC-induced [Ca²⁺]_i rise in GPN-pretreated DRG neurons in the normal Ca²⁺-containing external solution (contributing extracellular Ca²⁺ influx) was similar to, or slightly larger than, that mediated by store Ca²⁺ release (Fig. 3, G and H).

A rise in [Ca²⁺]_i directly triggers exocytosis in various cell types (Xu et al., 1998; Yu et al., 2004), so we next investigated whether TRPA1 activation directly triggers exocytosis in DRG neurons. Real-time membrane capacitance (C_m) recordings were used to monitor exocytosis with series conductance (G_s) and membrane conductance (G_m) as controls for seal conditions. AITC evoked a robust C_m increase in WT DRG neurons bathed in the Ca²⁺-containing solution (Fig. 7 A), suggesting that the AITC-induced [Ca²⁺]_i rise was sufficient to trigger vesicle exocytosis. In contrast, the AITC-induced C_m increase was drastically decreased in DRG neurons from TRPA1-KO mice (WT, 850 ± 90 fF; TRPA1-KO, 47 ± 1 fF; P < 0.001; Fig. 7, E and F). Importantly, AITC also induced exocytosis in WT DRG neurons bathed in the Ca²⁺-free solution (Fig. 7 B), although the maximal C_m change was 48% smaller than that in normal Ca²⁺-containing solution (2.5 mM Ca²⁺, 850 ± 90 fF; 0 mM Ca²⁺, 450 ± 30 fF; P < 0.05; Fig. 7 F). This is consistent with the finding that the AITC-induced [Ca²⁺]_i increase in the Ca²⁺-free solution was smaller than that in the Ca²⁺-containing solution (Fig. 6). To investigate the lysosome dependence of TRPA1 activation-triggered exocytosis in DRG neurons, we dialyzed these cells with 200 µM GPN to induce lysosome osmolysis and found that the AITC-induced C_m increase was markedly reduced 5 min after the dialysis (Fig. 7, C and F) and further decreased when DRG neurons were bathed in the Ca²⁺-free solution (Fig. 7, D and F). These results demonstrated that TRPA1 activation is coupled to vesicle exocytosis, and TRPA1 channel-mediated Ca²⁺ release from lysosome-like organelles is sufficient to trigger this activity in mouse DRG neurons.

TRP channel activation in peripheral sensory neurons plays essential roles in nociception and pathological pain sensation by causing the release of neuropeptides and other transmitters, such as calcitonin gene-related peptide (CGRP), neuropeptide Y (NPY), and substance P (Ulrich-Lai et al., 2001; Qin et al., 2008; Miyano et al., 2009; Meseguer et al., 2014). To visualize TRPA1 activation-induced neuropeptide release, we transiently expressed NPY-pHluorin in DRG neurons and monitored its release by total internal reflection fluorescence (TIRF) microscopy. Consistent with the idea that TRPA1 activation triggers neuropeptide release, application of AITC induced the exocytosis of large dense-core vesicles labeled with NPY-pHluorin in DRG neurons bathed in both normal and Ca²⁺-free external solutions (unpublished data). The TIRF imaging experiments thus confirmed the capacitance measurements (Fig. 7, A–E), showing that TRPA1-mediated release of stored Ca²⁺ is sufficient to trigger vesicle release from DRG neurons.

To further confirm that neuropeptide release is evoked by TRPA1 activation, we measured CGRP release from DRG neurons using an enzyme immunometric assay (EIA) kit. In control experiments, increasing the extracellular K⁺ concentration from 5 to 100 mM induced a robust CGRP release in 2 mM Ca²⁺-containing solution (Fig. 8, A and B) under our experimental condition. Application of AITC increased CGRP levels both in the presence and absence of extracellular Ca²⁺ (Fig. 8, C and D), indicating that activation of intracellular TRPA1 alone is sufficient to induce the release of neuropeptides from DRG neurons. In addition, CGRP application substantially enhanced the excitability of DRG neurons (Fig. 8, E and F), implying a role of lysosomal TRPA1 channels in the autoregulation of DRG neurons. To investigate whether lysosomal TRPA1-mediated internal Ca²⁺ release contributes to TRPA1-related pain sensation, we compared lysosomal TRPA1-associated acute noxious sensation between WT and TRPA1 KO mice (Bautista et al., 2006; Kwan et al., 2006; McNamara et al., 2007). Strikingly, topical injection of AITC or acrolein to the rodent hind paws induced the spontaneous nocifensive sensation (flinching and licking) and thermal hyperalgesia (withdrawal latency to thermal noxious stimulation), and they were partially relieved by the preinjection of GPN to deplete the lysosomal Ca²⁺ store (Fig. 9). Altogether, these findings indicate that TRPA1 channels are functionally localized to lysosome-like organelles, and TRPA1-mediated lysosomal Ca²⁺ release plays significant roles in DRG neurons at both the cellular and behavioral levels.

Several temperature-sensitive TRP channels (e.g., TRPV1, TRPA1, and TRPM8) are abundantly expressed in DRG neurons and function in temperature sensation and nociception. In this study, we provide the first evidence that, in addition to forming channels on the plasma membrane, TRPA1 is present on lysosomes, and the activation of TRPA1 triggers endolysosomal Ca²⁺ release. Importantly, we demonstrate that the endolysosomal TRPA1 contributes significantly to vesicle exocytosis, neuropeptide/CGRP release, and pain sensation.

The different Ca²⁺ sources and the spatial distribution of Ca²⁺-permeable channels have important implications for the spatial and temporal patterns of the Ca²⁺ signal, often with specific functional consequences. Conventionally, TRPA1, just like other thermosensitive TRP channels, is thought to work at the plasma membrane and mediate Na⁺ and Ca²⁺ influx to elicit nociceptive responses. Although it had been reported that TRPV1 channels may also be present intracellularly and mediate Ca²⁺ release, perhaps from the ER (Eun et al., 2001; Gallego-Sandín et al., 2009), the functional implications were not clear. We focused on endogenous TRPA1 channels in native DRG neurons and identified endolysosomes as the compartments from which these channels mediate significant Ca²⁺ release to account for functions previously thought to be mediated only by Ca²⁺ influx through plasma membrane channels. Using ΔFd analysis, we estimated that the lysosomal store release accounted for ∼40% of the overall [Ca²⁺]_i increase in mouse DRG neurons elicited by 100 µM AITC (Fig. 6). This value is much greater than the contribution of TRPV1-mediated Ca²⁺ release to the overall [Ca²⁺]_i increase elicited by capsaicin in DRG neurons, which we estimated to be 9% by ΔFd analysis (unpublished data). Therefore, not only do intracellular TRPA1 channels reside in different intracellular compartments from TRPV1 channels, but they may also play a more pronounced role in the apparent cytosolic Ca²⁺ signal than TRPV1 channels.

Membrane proteins are synthesized in the ER and then transported to the plasma membrane. It would, therefore, not be surprising if some ion channels functioned in the ER before being sorted out to the plasma membrane. However, TRPA1 does not appear to act in the ER because predepletion of the ER Ca²⁺ store using TG, CPA, or a low concentration of ionomycin (1 µM) did not affect the intracellular Ca²⁺ release induced by a membrane-permeable TRPA1 agonist (Fig. 2). In contrast, all manipulations that disrupt acidic organelles or lysosomes, such as treatment with bafilomycin A1 and GPN, resulted in the loss of TRPA1 activation-induced intracellular Ca²⁺ release (Fig. 3). Consistent with the pharmacological data, immunostaining and immuno-EM revealed the presence of TRPA1 on both the plasma membrane and lysosome-like organelles in DRG neurons (Fig. 4 and Fig. S4). More importantly, by confocal line-scan imaging of DRG neurons, we demonstrated that AITC-induced Ca²⁺ release occurred in close proximity to lysosomes (lysosome Ca²⁺ sparks), whereas the bulk of the cytosol (LAMP1-negative areas) had little response to the TRPA1 agonist under the experimental condition (Fig. 5). Altogether, these data strongly argue for the localization of TRPA1 channels on endolysosomal membranes and their involvement in mediating Ca²⁺ release from these organelles.

Several pieces of evidence suggest that lysosomal TRPA1 channels are critical for TRPA1-mediated physiological functions. First, AITC triggered vesicle exocytosis in Ca²⁺-free solution (Fig. 7, B and E), suggesting that TRPA1-mediated lysosomal Ca²⁺ release is sufficient to induce exocytosis in DRG neurons. The dramatic inhibition of this response by whole-cell dialysis with 200 µM GPN (Fig. 7, C and E) clearly demonstrated the involvement of lysosomes in the AITC-induced exocytosis under Ca²⁺-free conditions. Second, TRPA1 activation evoked neuropeptide release from DRG neurons in the absence of extracellular Ca²⁺. AITC-induced large dense-core vesicle exocytosis was detected by TIRF microscopy imaging (unpublished data) and CGRP release was measured by EIA (Fig. 8, A–D) with samples from DRG neurons bathed in either normal or Ca²⁺-free external solution. Third, TRPA1-mediated lysosomal Ca²⁺ release plays a significant role in nociception and pain hypersensitivity in mice (Fig. 9). These results further support the idea that TRPA1 activation-induced endolysosomal Ca²⁺ release contributes significantly to vesicle exocytosis and neuropeptide release from DRG neurons and may also contribute to nocifensive sensation.

Collectively, we have shown that TRPA1 activation not only triggers Ca²⁺ influx through the plasma membrane but also elicits Ca²⁺ release from endolysosomes in mouse DRG neurons. TRPA1 activation-induced endolysosomal Ca²⁺ release is of considerable size in the overall Ca²⁺ signal. It directly triggers vesicle exocytosis and neuropeptide release and contributes to nocifensive sensation in mice. These results suggest that multiple Ca²⁺ sources combine to generate distinct Ca²⁺ signatures in DRG neurons in response to the activation of TRPA1 channels and reveal new insights into the physiological and pathological functions of the “multifunctional” TRP channels.

The use and care of animals were approved and directed by the Animal Care and Use Committee of Peking University. Animals were housed under a 12-h light/dark cycle at 22 ± 2°C (two to four mice per individually ventilated cage) and had free access to food and water. DRG neurons were isolated from adult WT C57BL/6 mice or TRPA1-KO mice of the same strain (20–25 g, male) as previously described (Zhang and Zhou, 2002). The TRPA1-KO mice were provided by D. Julius (University of California, San Francisco, San Francisco, CA). In brief, DRGs were dissected out and placed in ice-cold L15 medium (Gibco). The tissues were then treated with trypsin (0.3 mg/ml) and collagenase (1 mg/ml) for 30 min at 37°C. Cells were dissociated, collected, and plated on coverslips coated with poly-l-lysine (Sigma-Aldrich). They were maintained at 37°C in a 5% CO₂–95% humidified air incubator and used within 2–8 h after plating. For TIRF imaging, the dissociated cells were transfected by electroporation with 3 µg NPY-pHluorin–expressing plasmid using the Neon Transfection System 100 µl kit (MPK10096; Invitrogen) and cultured for 12–18 h in Neurobasal medium (Gibco) supplemented with 2% B27 (Gibco) and GlutaMAX I (Gibco). Except for the unbiased screening of AITC-responsive neurons by Ca²⁺ imaging with confocal microscopy and EIA assays for CGRP release, only small-diameter neurons (20 ± 5 µm, corresponding to type C) were used (Bautista et al., 2006).

Patch-clamp recordings were performed using an EPC10/2 amplifier with Pulse software (HEKA). The voltage-clamped whole-cell configuration was used to record C_m and membrane currents in mouse DRG neurons (Zhang and Zhou, 2002). Pipette resistance was between 3 and 4 MΩ, and membrane potential was clamped at ∼70 mV. Normal external solution contained 150 mM NaCl, 5 mM KCl, 2.5 mM CaCl₂, 1 mM MgCl₂, 10 mM Hepes, and 10 mM glucose, pH 7.4. In the Ca²⁺-free solution, 2.5 mM CaCl₂ was replaced by 1 mM EGTA. Cells were initially bathed in the normal external solution, and then the Ca²⁺-free solution was locally puffed through a multichannel microperfusion system (MPS-2; INBIO). The standard intracellular pipette solution contained 153 mM CsCl, 1 mM MgCl₂, 10 mM Hepes, and 4 mM Mg-ATP, pH 7.2. C_m was measured using a lock-in module of the Pulse software as described previously (Zhang and Zhou, 2002). In brief, a 1-kHz, 20-mV peak-to-peak sinusoid was applied around a holding potential of −70 mV. Currents were recorded with the Pulse software at a sampling rate of 5 kHz. Drugs were delivered using the MPS-2 system as stated in this section above. Data were analyzed using the Lindau-Neher technique to estimate C_m, G_m, and G_s. Only recordings without significant changes of G_m and G_s were used for analysis (Zhang and Zhou, 2002).

Photometric measurement of [Ca²⁺]_i was performed as previously reported (Zhang and Zhou, 2002; Yu et al., 2004). In brief, DRG neurons were loaded with 5 µM Fura-2 AM (Invitrogen) for 5 min at 37°C. Cells were then washed three times (1 min each) with normal external solution at room temperature. An IX-70 inverted microscope (Olympus) equipped with a monochromator-based system (TILL Photonics) was used to monitor fluorescence changes with alternating excitation wavelengths of 340 and 380 nm. Emitted light at 510 nm was acquired through a photomultiplier tube (TILL Photonics) and recorded by the Pulse software. The fluorescence was sampled at 2 Hz (Zhang and Zhou, 2002). [Ca²⁺]_i was calculated from the ratio (R) of the fluorescence signals excited at 340 nm and 380 nm with the following equation: [Ca²⁺]_i = K_d × (R − R_min)/(R_max − R). K_d, R_min, and R_max are constants obtained from in vitro calibration as previously described (Xu et al., 1998). To avoid the desensitization of TRPA1 channels by drug application, only one cell per coverslip was used for AITC-induced signals. To block TRPA1 channels located on the plasma membrane, DRG neurons were incubated with 10 µM RR for 7 min at room temperature before the ratiometric [Ca²⁺]_i measurement.

For unbiased screening of TRPA1-positive cells, DRG neurons from WT and TRPA1 KO mice were loaded with 5 µM Fluo4-AM and imaged on an inverted confocal microscope (LSM710; ZEISS). Fluorescence images were captured by excitation with a 488-nm laser and emission at 500–560 nm. AITC (100 µM) or KCl (70 mM) was applied locally to the cells by a gravity-fed perfusion system. Fluorescence intensity values from the images were calculated and analyzed using ImageJ (National Institutes of Health). Images were acquired for >20 s before stimulation to establish a stable baseline, and then the fluorescence intensity values during the first 10 s before stimulation were averaged and used for normalization. Neurons that showed fluorescence changes (ΔF) >0.5 in response to stimulation were defined as AITC/4-HNE–responsive cells. This cutoff for selection (5 a.u.) is 1.5-fold that of the full-width half maximum of the Gaussian distribution of fluorescence noise.

For confocal line-scan imaging, mouse DRG neurons were transfected with LAMP1-EGFP and cultured for 12–18 h before experiments. Transfected neurons were incubated with 5 µM EGTA-AM (Invitrogen) for 15 min followed by 5 µM Rhod2-AM (Invitrogen) for 5 min. Cells were washed with standard bath solution immediately after incubation. A z-series of 1-µm optical sections was imaged using a 710 inverted confocal microscope with a 40× oil immersion lens (ZEISS). Line-scan mode was used to measure local Ca²⁺ signals. DRG neurons were bathed in the standard bath solution. The Ca²⁺-free bath solution was locally delivered to the cell by a drug-puffing device (MPS1; Shanghai Yibo Instruments Co., Ltd.) for 60 s before and during the first 20 s of line-scan imaging. The puffing solution was changed to 100 µM AITC in the Ca²⁺-free solution to trigger intracellular Ca²⁺ release during line-scanning. Fluorescence intensities of Rhod2 and GFP (for LAMP1-EGFP) along the line were recorded alternately by excitation at 543 and 488 nm and emission at 560–620 and 500–535 nm at 100 Hz. Changes in Rhod2 fluorescence intensity in EGFP-positive (lysosome) and EGFP-negative (nonlysosome control) regions over time were calculated using the plot profile function of ImageJ and plotted as ΔF/F₀ against time to indicate local [Ca²⁺]_i changes.

DRG neurons were fixed with 4% PFA in PBS for 20 min and permeabilized with 0.3% Triton X-100 (in PBS with 2% BSA) for 5 min. Cells were then blocked with 2% BSA in PBS for 1 h. Primary antibodies were diluted in PBS with 2% BSA and incubated overnight at 4°C. After washing out the primary antibodies with the blocking solution, cells were incubated for 1 h with Alexa Fluor 488– or 594–conjugated secondary antibodies (A11029, A11037, and A11055; Invitrogen). Antibodies against TRPA1 (ANKTM1 [F-20]; sc-32355; Santa Cruz Biotechnology, Inc.), VAMP2 (104211; Synaptic Systems), and LAMP1 (ab25744; Abcam) were used. Images were captured using an N-SIM superresolution microscope system with 78-nm resolution in the x and y axes and 300 nm in the z axis (Nikon). Images were processed using ImageJ.

Tissue slices of lumbar (L4–L6) DRGs were fixed in 2.5% glutaraldehyde and 2% PFA in PBS for 3 h at room temperature. After washing with PBS, they were dehydrated through an ethanol series, embedded in LR white resin, polymerized at 50°C for 24 h, and sectioned at 80 nm. The sections were collected on single-slot, formvar-coated nickel grids. After blocking in 5% BSA in PBS, they were incubated with antibodies against TRPA1 (Santa Cruz Biotechnology, Inc.) and/or LAMP1 (Abcam), washed, and incubated with secondary antibodies (LAMP1: goat anti–rabbit IgG conjugated to 10-nm gold particles, G7277 [Invitrogen]; TRPA1: donkey anti–goat IgG conjugated to 6-nm gold particles, 806.333 [AURION]; or rabbit anti–goat IgG conjugated to 10-nm gold particles, G5402 [Sigma-Aldrich]). Tissue sections were poststained with 2% uranyl acetate for 8–10 min and imaged under a transmission electron microscope (Tecnai G2 20; 200 kV; Thermo Fisher Scientific).

The fractional contribution of the Ca²⁺ store (P_store) was defined as the percentage of Ca²⁺ arising from internal store release in the total [Ca²⁺]_i rise through TRP channels localized both on the plasma membrane and the internal store (Neher and Augustine, 1992; Schneggenburger et al., 1993; Zhou and Neher, 1993; Yu et al., 2004). To control the intracellular Fura-2 concentration ([Fura-2]_i), we kept loading conditions uniform for all DRG neurons and only imaged those that had about the same diameter (20 ± 2 µm). For P_store measurement, Fura-2 AM stock solution was freshly made each time and used at 5 µM. A standard F₃₆₀-[Fura-2]_i plot was generated by loading known concentrations of Fura-2 potassium salt into rat DRG neurons using the whole-cell configuration of the patch clamp method and measuring the corresponding F₃₆₀ signal ([Fura-2] [mM]/F₃₆₀ [a.u.]: 0.125/0.543, 0.25/0.764, 0.5/1.818, and 1/3.038). The F₃₆₀ − [Fura-2]_i relationship was determined to be F₃₆₀ = 2.93 × [Fura-2]_i + 0.17 (R² = 0.99; Zhou and Neher, 1993). The mean F₃₆₀ value of small Fura-2 AM–loaded DRG neurons was 0.52 ± 0.02, thus giving an estimated [Fura-2]_i in the loaded mouse neurons of 120 µM. Because higher [Fura-2]_i reduced the basal [Ca²⁺]_i, which is needed for TRPA1 activation (Bautista et al., 2006; Zurborg et al., 2007), we determined the fractional contribution of Ca²⁺ store release against the total Ca²⁺ signal (release plus influx) under conditions in which intracellular Ca²⁺ buffers were similar at both 0 and 2.5 mM extracellular Ca²⁺. In this way, a similar fraction of Ca²⁺ release and Ca²⁺ influx (termed Pf) would be captured by nonsaturating Fura-2 in the same cell. Assuming the same Pf value for Q_store and Q_total, we have P_store = Q_TRP-store/Q_TRP-total = (ΔFd_0Ca/Pf)/(ΔFd_2.5Ca/Pf) = ΔFd_0Ca/ΔFd_2.5Ca, in which ΔFd_0Ca and ΔFd_2.5Ca are ΔFd values at 0 and 2.5 mM extracellular Ca²⁺.

Basal and stimulated extracellular CGRP concentrations were evaluated in freshly isolated DRG neurons using an EIA kit (Bachem) following the manufacturer’s instructions. The isolated cells were maintained in DMEM/F12 with 10% FBS for 2–3 h in a 24-well plate before EIA experiments. For each well, cells were washed three times with normal external solution and then incubated with normal or Ca²⁺-free external solution for 5 min at room temperature, followed by another 5-min incubation with the same solution containing 100 µM AITC. The incubating solutions were collected for subsequent analysis of basal and stimulation-coupled CGRP levels. A similar protocol was used for 100 mM K⁺-induced CGRP release from DRG neurons in normal external solution. All samples were centrifuged at 13,000 rpm for 5 min and the supernatants processed for CGRP measurement. Samples were analyzed at 450 nm with a microplate reader (Synergy 4; BioTek). CGRP concentrations (in picograms per milliliter) were extrapolated from a best-fit line calculated from serial dilutions of a CGRP standard. All data points were measured in triplicate.

Mice were housed under a 12-h light/dark cycle at 22 ± 2°C, and experiments were performed in the afternoon. Flinching and licking responses to AITC/acrolein were monitored and quantified following previously described methods (Bautista et al., 2006; Kwan et al., 2006). AITC (20 µl, 100 mM) or acrolein (20 µl, 2 mM) was injected into the right hind footpad. Pain was assessed by paw flinching and licking during 5 min after injection. To investigate the effect of GPN on AITC/acrolein-induced spontaneous nocifensive sensation and thermal hyperalgesia, GPN (20 µl, 2 mM) was injected 15 min before the injection of AITC/acrolein plus GPN (20 µl, same concentrations as listed previously). In the thermal nociception assay, mice were allowed to acclimate to a Plexiglas chamber for at least 30 min every day for 1 wk before experiments. The plantar test (3730-002; Ugo Basile) was used to assess thermal hyperalgesia; mice were placed on the surface of a 2-mm–thick glass plate, and a radiant heat stimulus was applied under the glass floor according to literature (Caterina et al., 2000; Bautista et al., 2006). The heat stimulus was terminated with a withdrawal response or a cutoff value of 20 s to prevent injury. The paw withdrawal latency was defined as the time from the beginning of the heat stimulus to hind paw withdrawal. To assess AITC/acrolein-induced thermal hyperalgesia, 20 µl AITC (100 mM) or acrolein (2 mM) was injected into the right hind footpad with or without preinjection of 2 mM GPN 120 (for AITC) or 30 min (for acrolein) before the thermal nociception test. The experimenters were blinded to the genotypes of the mice for all of the behavioral experiments.

All chemicals were from Sigma-Aldrich unless otherwise specified.

All experiments were replicated at least three times. All electrophysiological data were analyzed with Igor software. Statistical comparisons were performed with the two-tailed unpaired Student’s t test or one-way analysis of variance (ANOVA) as indicated using Statistical Package for the Social Sciences 13.0. No statistical methods were used to predetermine sample sizes, but our sample sizes are similar to those generally used in the field. Normality of the data were tested with the Shapiro-Wilk test, and the equality of variance was determined with Lenene’s test. Statistical data are presented as mean ± SEM, and a significant difference was accepted at P < 0.05. Numbers of cells or animals analyzed are indicated in the figures.

Fig. S1 shows that AITC induces a [Ca²⁺]_i rise in DRG neurons bathed in normal Ca²⁺-containing solution and that 4-HNE induces a [Ca²⁺]_i rise in DRG neurons bathed in Ca²⁺-free solution. Fig. S2 shows that TRPA1 activation mediates [Ca²⁺]_i rise in DRG neurons bathed in normal Ca²⁺-containing solution. Fig. S3 shows that TRPA1 activation mediates a [Ca²⁺]_i rise in DRG neurons bathed in Ca²⁺-free solution. Fig. S4 shows the localization of TRPA1 channels in endolysosomes in DRG neurons. Fig. S5 shows that vacuolin-1 fails to enlarge endolysosomes in DRG neurons.

The authors thank Dr. Heping Peace Cheng (Institute of Molecular Medicine, Peking University) for advice on the confocal line-scan imaging of lysosome Ca²⁺ sparks, Dr. Yingchun Hu (Core Facilities, College of Life Sciences, Peking University) for assistance with the transmission EM experiments, and Dr. Iain C. Bruce (Institute of Molecular Medicine, Peking University) for helpful comments on the manuscript.

This work was supported by grants from the National Basic Research Program of China (2012CB518006) and the National Natural Science Foundation of China (31228010, 31171026, 31100597, 31327901, 31221002, 31330024, 31400708, 31670843, 31521062, and SQ2011SF11B01041). M.X. Zhu was supported by the National Institutes of Health (grants R01DK081654 and R01GM081658). C. Wang was supported in part by a Postdoctoral Fellowship from the Peking-Tsinghua Center for Life Sciences.

The authors declare no competing financial interests.