In HEK293t cells transfected with hTRPV1 application of 10 mM morphine evoked an inward current, 300 μM capsaicin was applied as a control thereafter (n = 10, sample recording in Fig. 1). In HEK293t cells transfected with mTRPA1 10 mM morphine evoked inward currents that appeared to rapidly inactivate during continued superfusion (n = 10). However, upon offset and washout of morphine regularly large tail currents occurred that revealed the sustained activation of TRPA1 and a slow deactivation. Mustard oil (30 μM) was applied as a control thereafter.

A total of 106 dorsal root ganglion (DRG) neurons were exposed to 10 mM morphine for 40 s, 30 μM acrolein for 30 s, 1 μM capsaicin for 5 s and 60 mM potassium for 30 s with an interval of 240 s between applications. Response rates to morphine were 22%, to acrolein 45% and to capsaicin 58%. Morphine responses in neurons activated by capsaicin but not by acrolein and in neurons activated by acrolein but not capsaicin were larger than in neurons which did not respond to either capsaicin or acrolein (p = 0.024 and p = 0.027, t-test independent samples). Responses to morphine grouped by the sensitivity to capsaicin and acrolein are presented in Fig. 2a. Morphine (10 mM) was applied three times for 40 s. The second application of morphine in the presence of BCTC (10 μM) and HC030031 (50 μM) did not elicit a calcium increase (both p < 0.001 compared to the initial and the third morphine application, n = 62, Wilcoxon, Fig. 2b).

Increasing concentrations of morphine (0.1 – 10 mM) were consecutively applied for 40 s, separated by 240 s washout intervals. From C57BL/6 animals 134 DRG neurons were tested. Morphine at 3 mM and 10 mM evoked a concentration-dependent transient rise in intracellular calcium (ANOVA, F_(4,532) = 16.9, p = 0.005 and p < 0.001, HSD post-hoc tests). The EC₅₀ was 2.6 ± 0.04 mM (Fig. 3). Morphine at 10 mM reached 32% of the 60 mM potassium-induced calcium influx.

This protocol was repeated in neurons from TRPV1^-/- (n = 112), TRPA1^-/- (n = 125) as well as from TRPV1^-/- TRPA1^-/- animals (n = 135). Compared to neurons from C57BL/6 mice, the increases in calcium levels evoked by 10 mM morphine were smaller in animals lacking TRPV1 or TRPA1 and almost absent in double knockouts (ANOVA, F_(9,1485) = 7.0; TRPV1^-/-: 51%, p = 0.008; TRPA1^-/-: 34%, p = 0.005; TRPV1^-/- TRPA1^-/-: 2%, p < 0.001; % of AUC in C57BL/6, HSD post-hoc tests). Compared to baseline, the small morphine-induced increases of intracellular calcium were still significant in neurons from TRPV1^-/- (p < 0.001, ANOVA, F_(4,444) = 13.0) and TRPA1^-/- (p < 0.001, ANOVA, F_(4,468) = 24.6), but not so in neurons from TRPV1^-/- TRPA1^-/- animals (p = 0.10, ANOVA, F_(4,536) = 2.0). Due to the limited solubility of morphine at pH 7.4, no higher concentrations than 10 mM could be tested. Neurons activated by morphine were 20.5 ± 0.9 μm in diameter, smaller than the average of 23.0 ± 0.5 μm (p = 0.041, U-test).

The inflammatory mediator bradykinin is known to activate nociceptive neurons in higher concentrations but in lower concentrations to sensitize TRPV1 and TRPA1. To determine whether calcium increases evoked by 10 mM morphine are increased by bradykinin at a concentration below activation, we applied the second of four consecutive morphine applications in the presence of 5 μM bradykinin (Fig. 4). Bradykinin was applied 40 seconds before the morphine stimulus and did not elicit a calcium response (n = 111). Among the 22.5% of DRG neurons responding to morphine, bradykinin increased the morphine-stimulated calcium influx to 249% (ANOVA, F_(3,330) = 5.1, p= 0.002, HSD post-hoc test), the subsequent two morphine responses were not different compared to the first one (p = 0.33 and p = 0.87, HSD post-hoc tests). This protocol was repeated with morphine replaced by naloxone, which is structurally similar but has no intrinsic activity on opioid receptors. The activation by naloxone was also concentration-dependent (n = 10, data not shown). Bradykinin reversibly increased the naloxone-stimulated calcium increase to 165% of the first response (ANOVA, F_(3,66) = 5.0, p = 0.025, HSD-post hoc test).

In order to reappraise these findings in intact innervated tissue, we investigated the stimulated CGRP release from nerve endings in isolated hind paw skin of C57BL/6 mice. In a first series of experiments (n = 8) we assessed the response to application of 10 mM morphine diluted in SIF (Fig. 5). The basal CGRP concentration in the first incubation period (SIF) was 2 ± 1 pg/ml. When the skin flaps were transferred to the test tube containing 10 mM morphine, the CGRP release increased to 20 ± 5 pg/ml (p = 0.012, Wilcoxon). Final exposure to 60 mM KCl resulted in a CGRP release of 54 ± 11 pg/ml.

A separate series of experiments was performed to test the hypothesis that morphine concentrations without a direct effect on CGRP release may sensitize cutaneous afferents to noxious stimuli activating TRPV1 or TRPA1. CGRP release in the presence of 310 μM morphine was not different from the basal release (Fig. 6). The selective agonist acrolein 100 μM was used to activate TRPA1 at the lower end of the concentration-response relationship; in presence of 310 μM morphine acrolein caused markedly greater increases in CGRP release (p = 0.028, n = 6, Wilcoxon). A similar series of experiments was performed for noxious heat stimulation using SIF at 47°C; heat-induced cutaneous CGRP release depends partly on TRPV1 as a transducer. The exposure to 310 μM morphine caused no rise in CGRP release (Fig. 7), but slightly sensitized the release evoked by heat stimulation (p = 0.018, n = 7, Wilcoxon, compared to 47°C on the control side).

The present study was initiated by the observation of painful sensations caused by topical application of a 2.67 mM morphine gel to inflamed skin following radiation therapy, as reported in the interdisciplinary pain research group (KFO 130). In these patients, the topical morphine formulation (prepared by the pharmacy unit of the university) was preferred because systemic opiates could not be applied systemically due to unwanted side effects. The painful sensations were reported to last only for a few seconds but exceeded seven on a ten-point rating scale. The time course of the pain was similar to that reported after injection of lidocaine, which has recently been shown to activate nociceptive afferents via TRPV1 and TRPA1 receptors before sodium channel blockade leads to local anesthesia 12. Similarly for morphine, an inhibition of sodium channels has been described 131415, masking by conduction block the sustained activation. Sodium channel block also occurs with naloxone 16, shedding doubts on alleged naloxone reversal of peripheral opioid effects.

The radiation dermatitis involves a tailored therapy; topical application of morphine is not the primary choice but rather an uncommon option. This might explain why systematic studies are not available. Common symptomatic therapy of radiation dermatitis is based on topical wound care predominantly involving moisturizers. There is no uniequivocal evidence for a causal therapy, including the topical application of ascorbic acid, corticosteroids or calcineurin inhibitors; opioids have not been investigated so far 17.

There is evidence for peripheral opioid receptor effects 181920. The expression profile of opioid receptors on primary afferent neurons corresponds to the nociceptor population 2122. Peripheral application of opioid receptor agonists (below 100 μM) or antagonists demonstrated their local efficacy but the relative contribution of peripheral opioid receptors after systemic opioid administration is still unclear 23.

In animal experiments involving peripheral opioid injections or topical morphine applications, pain-related behavior has not been reported 242526. However, it seems likely that in these studies morphine concentrations at the nerve endings were below the activation threshold of nociceptors. A behavioural correlate for the activation of TRPV1 and TRPA1 by morphine injection is not easily detected since the immediate pain of skin puncture is closely followed by the nerve conduction block. The only previously reported application of a high concentration of morphine in a well-controlled setup produced nociceptor activation, as observed in the present study. In this preparation, the superior spermatic nerve of dogs in vitro, morphine applied by superfusion at 1 – 310 μM generated action potentials in slowly conducting afferents; the activation was concentration-dependent and the combination of bradykinin and morphine resulted in sensitization 27. In the present study the applied concentrations were higher, and 310 μM morphine caused sensitization but no overt activation. A higher expression of TRPA1 and TRPV1 in visceral afferents 2829 might contribute to the lower concentration of morphine required for activation of testicular nociceptors.

For the reported painful sensations in patients or the activation of primary afferents, a G-protein coupled opioid receptor effect seems unlikely in view of the action of the opioid antagonist naloxone. Since TRPV1 and TRPA1 are the principal detectors for painful chemical compounds, they were the primary targets of the investigation. Considering pharmacological results and activation of neurons from knockout animals, both TRP receptors contributed approximately equal to the activation of DRG neurons caused by morphine. Only the knockout of both receptors eliminated the responsiveness.

The observed morphine-induced increase in calcium levels is only a fraction of that elicited by KCl. This might explain why a lower percentage of neurons was activated by morphine in comparison to acrolein and capsaicin. Inhibition of calcium conductance by morphine is a possible explanation. In fact, morphine and the mu-agonist DAGO (Tyr-D-Ala-Gly-MePhe-Gly-ol) reduce the conductivity of voltage gated calcium channels 3031, which may be activated secondary to neuronal depolarization via TRPV1 and TRPA1. For TRPA1, both morphine and mustard oil seem to inhibit ion flux through TRPA1, a tail current was observed when the application was stopped. The apparent desensitization is likely due to an additional pore blocking effect that morphine at high concentration may exert on TRPA1. Mustard oil, the index agonist of TRPA1, also induced, yet smaller, tail currents followed by slow deactivation. Together with the very slow activation this may suggest that also mustard oil exerts a double action on TRPA1, activation of the receptor-channel and partial pore block.

The neuropeptide calcitonin gene related peptide CGRP is expressed by a large subset of predominantly nociceptive primary afferent neurons and released upon activation of these neurons. The measurement of released CGRP is therefore widely used as an index for the activation of nociceptors. CGRP is expressed in a large proportion of neurons positive for TRPV1 or TRPA1, therefore the release of CGRP is a favorable readout for the activation of these neurons 3233. High concentrations of morphine reversibly stimulated the release of CGRP from the isolated skin. Furthermore, concentrations below the threshold of activation sensitized the CGRP release evoked by noxious heat and acrolein, activators of TRPV1 and TRPA1 receptors, respectively.

In inflamed skin inflammatory mediators and their receptors are frequently overexpressed 3435, among them particularly bradykinin and its receptors 36. Bradykinin sensitizes TRPV1 37 and TRPA1 38. Therefore it was not surprising that morphine responses were sensitized by bradykinin. Sensitization to morphine by bradykinin and a reduced barrier function may explain the particular painfulness of the treated skin areas in radiodermatitis patients.

Animal care and treatment were in accordance with the guidelines of the International Association for the Study of Pain 39. Adult C57Bl/6, TRPV1^-/- and TRPA1^-/- as well as TRPV1^-/-TRPA1^-/- double knockout mice were used. Breeding pairs of TRPV1 and TRPA1 knockout mice were obtained from Dr. John Davis 40 and Dr. David Corey 41 and backcrossed to C57BL/6. Double knockout animals were generated in our animal facility by cross-mating knockouts of both strains. All animals were genotyped by the previously reported primers.

Animals were killed in pure CO₂ atmosphere. Dorsal root ganglia of all lumbar and the first two thoracic segments of the spinal column were excised and transferred in Dulbecco's modified Eagle's medium (DMEM, Invitrogen, Carlsbad, USA) solution containing 50 μg/ml gentamicin (Sigma). The ganglia were treated with 1 mg/ml collagenase and 0.1 mg/ml protease for 30 minutes (both from Sigma) in TNB 100 solution supplemented by TNB 100 lipid-protein-complex, 100 μg/ml streptomycin and penicillin (all from Biochrom, Berlin, Germany) and 200 μg/ml glutamine (Invitrogen, Carlsbad, USA) for 20 minutes at 37°C. After three wash steps, the suspension was triturated in TNB solution using a fire-polished silicone-coated Pasteur pipette. The cells were plated on poly-D-lysine-coated (Sigma) cover slips and cultured in TNB medium overnight at 37°C in a humidified 5% CO₂ atmosphere.

HEK293t cells were transfected with plasmids of hTRPV1 (John Davis) or mTRPA1 (Ardem Patapoutian) along with GFP as reporter plasmid using Polyfect (Qiagen, Hilden, Germany) according to the manufacturer's protocol. After incubation for one day, the cells were replated in 35 mm culture dishes and used for experiments within 1–2 days. GFP-expressing cells were identified by fluorescence.

Recording electrodes were pulled from borosilicate glass tubes (GB150T-8P, Science Products, Hofheim, Germany) to give a resistance of 3.5 – 5.0 MΩ (P97, Sutter, Novato, CA). The external solution contained (in mM) 140 NaCl, 4 KCl, 1.8 CaCl₂, 1 MgCl₂, 10 HEPES, 4 glucose, adjusted to pH 7.4. The internal solution contained (in mM) 140 KCl, 1.6 MgCl₂, 2 EGTA and 10 HEPES and was adjusted to pH 7.4. Recordings were performed at room temperature and cells were held at -60 mV. Membrane currents were acquired with an Axopatch 200B amplifier and pClamp 10 software (Molecular Devices, Sunnyvale, CA). Solutions were applied with a RSC-100 gravity-driven solution changer (Bio-Logic, Claix, France). All cells were first exposed to morphine, then to capsaicin or mustard oil. TRPA1 was sensitized by PLC activator m-3M3FBS (1 μM, Sigma).

Cells were stained by 5 μM fura-2 AM and 0.02% pluronic F-127 (both from Invitrogen) dissolved in the TNB medium for about 30 min in the incubator, followed by a short wash-out period to allow fura-2 AM ester hydrolysis. The cover slips were placed in a custom-made chamber and mounted on a Zeiss Axiovert inverse microscope with a 40× NeoFluar objective. Cells were continuously superfused throughout the experiment with extracellular fluid (in mM: NaCl 145, KCl 5, CaCl₂ 1.25, MgCl₂ 1, Glucose 10, Hepes 10) at a flow rate of approximately 0.3 ml/min at room temperature. Cells were illuminated with a 75W xenon arc lamp and a monochromator alternating between 340 and 380 nm of wavelength (Photon Technology International, New Jersey, USA). Images were acquired at 1 Hz with 200 μs exposure time using a CCD camera, controlled by Image Master software (PTI, Birmingham, USA). Fluorescence ratios were computed for regions of interest adapted to the neurons. All experimental protocols were pre-programmed using a custom-made software that controls the valves of a 7-channel gravity-driven common-outlet superfusion system 42.

C57/Bl6 mice of either sex with an average weight of 20 g (range 12–27 g) were used. The skin of both hind paws distal to the knee was subcutaneously excised. The skin flaps with an average weight of 105 mg (range 57–135 mg, n = 36) were fixed to acrylic rods by surgical threads with the corium exposed. During this procedure the skin flaps were constantly immersed in synthetic interstitial fluid (SIF, in mM: NaCl 107.8, KCl 3.5, NaHCO₃ 26.2, NaH₂PO₄ 1.7, Na-gluconate 9.6, sucrose 7.6, glucose 5.6, CaCl₂ 1.5 and MgSO₄ 0.7 43 equilibrated with carbogen (pH 7.4). The fixed skin flaps were placed in test tubes and mounted in a shaking bath of 32°C. After washing for 30 minutes, each experiment was composed of four consecutive 5 min incubation periods in test tubes filled with 0.8 ml SIF. In the first two periods basal CGRP release was determined, during the third period the preparation was chemically or thermally stimulated, the fourth period was for the post-stimulation control. One separate protocol provided two stimulations in incubation periods two and four. The contralateral hind paw skin of each animal was used as a matched-pairs control without conditioning chemicals.

Immediately after the incubation, 100 μl of the solution were processed using a commercial CGRP enzyme immune assay kit (SPIbio, France). The details of this method were described previously 44. The antibodies used are directed against human α/β-CGRP but are 100% cross-reactive against mouse and rat CGRP. The detection level is about 2 pg/ml. Samples of SIF and all stimulation solutions measured on the same enzyme immune assay 96-well plate served as controls.

The calcium influx elicited by a stimulus was quantified by evaluating the area under the curve (AUC) of the application period. Absolute increases in calcium concentrations were calculated based on R_min and R_max 45. Compared to the foregoing reference period a mean increase of the intracellular calcium concentration of at least 15 nM throughout the application period was considered as activation. At the end of all protocols a 10 s stimulus of KCl 60 mM was applied as control and normalization reference. Cell diameter was calculated as the mean of two orthogonal axes. Repeated measurements and independent groups in calcium imaging experiments were compared by ANOVA and HSD post-hoc test. CGRP release experiments were performed using both hindpaw skin flaps of the same animal and compared using the Wilcoxon matched pairs test. Statistical analysis was performed using Statistica 7 (Tulsa, OK, USA), sigmoidal dose-response curves were fitted by Origin 7.5 (Northhampton, MA, USA). Data are presented as mean ± SEM; p < 0.05 was considered significant.