TRPA1 channels mediate acute neurogenic inflammation and pain produced by bacterial endotoxins

Abstract

Gram-negative bacterial infections are accompanied by inflammation and somatic or visceral pain. These symptoms are generally attributed to sensitization of nociceptors by inflammatory mediators released by immune cells. Nociceptor sensitization during inflammation occurs through activation of the Toll-like receptor 4 (TLR4) signalling pathway by lipopolysaccharide (LPS), a toxic by-product of bacterial lysis. Here we show that LPS exerts fast, membrane delimited, excitatory actions via TRPA1, a transient receptor potential cation channel that is critical for transducing environmental irritant stimuli into nociceptor activity. Moreover, we find that pain and acute vascular reactions, including neurogenic inflammation (CGRP release) caused by LPS are primarily dependent on TRPA1 channel activation in nociceptive sensory neurons, and develop independently of TLR4 activation. The identification of TRPA1 as a molecular determinant of direct LPS effects on nociceptors offers new insights into the pathogenesis of pain and neurovascular responses during bacterial infections and opens novel avenues for their treatment.

Figure 1. TRPA1 mediates stimulation of nociceptor neurons by LPS.

(a–c) Representative examples of the effects of LPS (10 μg ml⁻¹) on [Ca²⁺]_i levels in nodose ganglion neurons isolated from WT (a), Tlr4 KO (b) and Trpa1 KO mice (c). Cinnamaldehyde (CA, 100 μM) and capsaicin (Caps, 100 nM) were applied to identify TRPA1- and TRPV1-expressing neurons, respectively. (d) Percentage of mouse nodose and TG sensory neurons responsive to LPS (blue) or to LPS and CA (black). The labels ‘WT+HC’ and ‘WT after HC’ refer to the responses to LPS (10 μg ml⁻¹) observed in the presence of the TRPA1 inhibitor HC-030031 and after its removal, respectively. (e,f) Percentage of nodose (e) or TG (f) neurons responding to LPS (blue) or to LPS and CA (black) as a function of LPS concentration.

Figure 2. E. coli LPS excites TRPA1-expressing sensory neurons.

(a) Simultaneous recording of [Ca²⁺]_i level (green trace) and instantaneous firing frequency (black dots) in cell-attached mode in a nodose neuron from a Tlr4 KO mouse. The large calcium transients (upper graph) during LPS and MO application correlate with the firing of action potentials (lower graph). (b) Summary results of effects of E. coli LPS (10 μg ml⁻¹) on mean±s.e.m. firing rate in nodose neurons from WT (n=4) and Tlr4 KO mice (n=3), (paired t-test). WT neurons were electrically silent before LPS application. (c) Examples of recordings of membrane potential in the perforated-patch current-clamp mode (I_hold=0 pA) in two nodose neurons. LPS (10 μg ml⁻¹) induces rapid depolarization and firing of action potentials in the TRPA1-expressing neuron (that is, AITC positive; upper traces). In contrast, LPS had no effect on membrane potential in the TRPA1-negative neuron (lower traces). (d) Average±s.e.m. depolarization of the resting membrane potential produced by LPS (n=5) and AITC (AITC) or CA (n=10) in TRPA1-expressing neurons (solid bars). The dashed bars represent the average depolarization produced by the application of LPS, AITC/CA or high extracellular potassium (60 mM) in neurons not expressing TRPA1 (n=10). Statistically significant differences of membrane potential (relative to control) are denoted by *P<0.05 and ***P<0.001, paired t-test).

Figure 3. LPS stimulates recombinant mouse TRPA1 channels.

(a) Ratiometric [Ca²⁺]_i measurement showing the effect of LPS on mTRPA1 cells. Non-transfected cells (red traces) did not respond to LPS. The thick traces represent the means and the dashed traces represent the means±the corresponding s.e.m. (n≥25 cells for each trace). (b) Dose dependency of LPS-evoked [Ca²⁺]_i responses (mean±s.e.m.) (>80 cells per data point). The solid line represents the fit with a Hill equation. (c) Time course of TRPA1 current response to LPS (20 μg ml⁻¹) and AITC (20 μM) monitored at −75 mV. The coloured data points correspond to the current traces shown in the inset. (d) Average±s.e.m. maximal amplitude of the currents (T of recording=25 °C) during extracellular application of LPS (red bar) or AITC (blue bar) in mTRPA1 cells (n=7) and in control (white bar; CHO cells (n=6). Amplitude differences have been compared by an unpaired t-test. *P<0.05, **P<0.01. (e) Whole-cell currents in response to a double-pulse voltage protocol in control and in the presence of E. coli LPS (20 μg ml⁻¹), showing the increase in current amplitude and the slowing of channel closure. (f) Average±s.e.m. voltage dependence of TRPA1 peak tail current at −75 mV in control and in the presence of 20 μg ml⁻¹ LPS (n=6). Data were normalized to the value obtained at +175 mV in control. The solid lines represent the fit of the data with Boltzmann functions.

Figure 4. LPS activates TRPA1 channels in a membrane-delimited manner.

(a) Time course of current amplitudes recorded in outside-out patches of TRPA1-expressing CHO cells at –75 and +75 mV. Currents were elicited by voltage ramps from −100 to +100 mV from a holding potential of 0 mV. The arrow indicates the moment of application of LPS (10 μg ml⁻¹) and the horizontal bar indicates the period of exposure to 50 μM HC-030031. The coloured asterisks mark the time points at which the corresponding current traces shown in b were recorded. Single-channel records in the outside-out configuration at +75 mV, in control solution (c) and in the presence of 10 μg ml⁻¹ LPS (d). Events marked with an asterisk are displayed at an expanded timescale. The histograms on the left show the corresponding distributions of the current amplitudes. All records in the figure were obtained in Ca²⁺-free conditions.

Figure 5. The shape of lipid A determines TRPA1 activation by LPS.

(a) Ratiometric [Ca²⁺]_i imaging experiment showing the effect of extracellular application of lipid A on CHO mTRPA1 cells. Thirty-six percent of the TRPA1-expressing cells (AITC-sensitive) responded to lipid A (red traces, n=50). Non-transfected cells (black traces) did not respond to lipid A (n=20). Solid and broken traces represent mean and ±s.e.m., respectively. (b) Representative responses of mouse nodose neurons to lipid A, CA (100 μM), capsaicin (1 μM) and high K⁺ (30 mM). A fraction of CA-sensitive neurons was also activated by lipid A. (c) Mean [Ca²⁺]_i elevation evoked by LPS extracted from E. coli, S. marcescens and N. meningitidis (20 μg ml⁻¹) in mTRPA1 cells (n>25 for each trace). (d) Average±s.e.m. amplitude of [Ca²⁺]_i evoked by LPSs with different lipid A conformation in mTRPA1 cells (n=26–129). (e) Representative examples of responses in TG neurons to different types of LPS. (f) Percentage of TG neurons responding to different types of LPS. In nociceptive neurons, responses to S. marcescens and S. minnesota LPS were significantly reduced (***P<0.001) compared with E. coli (Fisher’s exact test). In non-nociceptive neurons, responses were not different.

Figure 6. The effects of LPS on TRPA1 channels are independent of TLR4 signalling and require lipid A activity.

(a) Representative example of responses to LPS in nodose neurons following preincubation for 10 min with CLI-095 (1 μM). (b) The percentage of neurons responding to LPS, CA or capsaicin was not affected by preincubation with CLI-095 (n=189) compared with vehicle (n=77), Fisher’s exact test. (c) Mean±s.e.m. amplitude of the response to LPS in CLI-095 (n=24) and vehicle (n=13) was not significantly different (unpaired t-test). (d) Preincubation (30 min) of LPS (3 μg ml⁻¹) with polymyxin B (PMB; 300 ng ml⁻¹) abrogated responses in mTRPA1 cells to LPS. The same incubation had minimal effects on the response to AITC (100 μM). (e) Percentage of CHO-TRPA1 cells responding to PMB, LPS or LPS+PMB (n>50 cells). The inhibition of LPS responses was highly significant (Fischer’s exact test). (f) The amplitude of calcium response to MO was minimally reduced by PMB (mean±s.e.m., n>100 cells).

Figure 7. Neurogenic inflammation and pain by LPS depend on TRPA1 activity.

(a) Trinitrophenol (TNP, 50 μM) and LPS (100 μg ml⁻¹) evoke (mean±s.e.m.) modest but significant (^#P<0.05, Wilcoxon test, n=12/8) tracheal CGRP release; preincubation with TNP potentiates response to LPS and acrolein (30 μM). Trpa1 KO mice do not respond to LPS (*P<0.05, analysis of variance (ANOVA) Fisher’s least significant difference (LSD) test, n=4), TNP or its co-application with LPS or acrolein (**P<0.01). (ANOVA Fisher’s LSD test, n=8, 4, 4). (b) LPS and 4-hydroxy-2-nonenal (HNE) evoke small but significant (^#P<0.05) CGRP responses. Preincubation with low concentration of HNE potentiates tracheal CGRP release evoked by LPS (**P<0.01, ANOVA Fisher’s LSD test, n=4, mean±s.e.m.). (c) Summary of effect of LPS (100 μg ml⁻¹) perfusion on mesenteric artery diameter (mean±s.e.m.) from WT (n=7) and Trpa1 KO (n=8) mice in control condition and in the presence of HC-030031 (10 μM). Arteries precontracted with the alpha1 adrenergic agonist phenylephrine (2 μM). Statistical differences were evaluated with an unpaired t-test. (d) The nocifensive response and the acute local inflammation produced by intraplantar LPS injection (5 μg μl⁻¹ in 10 μl) were nearly abolished in Trpa1 KO animals (n=6) compared with WT littermates (n=7) (unpaired t-test, values are mean±s.e.m.). (e) Mechanical paw withdrawal threshold (mean±s.e.m.) to mechanical stimulation before and after intraplantar injection of LPS (5 μg μl⁻¹) in WT (n=14) and Trpa1 KO (n=14) animals. The asterisks reflect the differences in threshold between WT and KO (unpaired t-test). Comparison between baseline and the different time points gave significant differences (^##P<0.01, ^###P<0.001) in WT animals only (one-way ANOVA with Bonferroni’s correction). (f) Correlation between the hind paw inflammation (mean±s.e.m.) induced by LPSs of different shapes (blue for cylindrical, red for semiconical and black for conical; n=5–9) with the potency for TRPA1 activation (mean±s.e.m.) estimated from the amplitude of Ca²⁺ response evoked in mTRPA1 cells; Fig. 5d).