Advances in TRP channel drug discovery: from target validation to clinical studies.

In the neurogenic bladder , the TRPV1-expressing C fibre-driven micturition reflex, which is inactive under physiological conditions, resumes control of the bladder functions 16 . Intravesical administration of a large enough dose of TRPV1 agonist (capsaicin 195 or resiniferatoxin 196 ) to desensitize the C-fibres provides lasting relief in patients with neurogenic bladder by increasing bladder capacity and reducing the number of incontinent episodes 197 . Furthermore, in bladder biopsy samples taken from patients with interstitial cystitis , the density of TRPV1-expressing fibres correlated with clinical symptoms 198 . Yet, in a phase II clinical trial intravesical resiniferatoxin failed to meet the primary end point (symptom improvement according to the Global Response Assessment, a 7-point scale rating overall change in symptoms), faring no better than placebo after 4 weeks 199 . Therefore, it is likely that the increased TRPV1 expression in the bladder biopsy samples was not the cause but a consequence of the disease. Parenthetically, an interesting (and controversial) indication of topical resiniferatoxin desensitization is its application to the penis to prevent premature ejaculation 200 . Animal experiments suggested a therapeutic value for TRPV1 (GRC-6211) 201 and TRPM8 (RQ-00434739 (ref. 202 ) and KRP-2529 (ref. 203 )) antagonists for suppressing hyperactivity in the chronically inflamed bladder, which is yet to be tested in the clinic. In the bladder, TRPV4 is expressed both in urothelium and in the detrusor muscle, where its activation causes sustained muscle contractions 204 . This is consistent with the spotty incontinence phenotype of the Trpv4 –/– mice 205 . By contrast, the TRPV4 agonist GSK1016790A evokes bladder contractions 206 . Of relevance, TRPV4 expression is increased in human overactive bladder mucosa 207 . These and other preclinical findings imply that a TRPV4 agonist may improve the underactive bladder 208 , whereas an antagonist could be valuable in managing the overactive bladder 209 . The TRPC5 antagonist AC-1903 prevented podocyte loss in a rodent model of focal segmental glomerulosclerosis, suggesting a novel interaction for renal allograft protection 210 . TRPC6 is highly expressed in the kidney, but the role of TRPC6 in renal pathology is complex, which has hindered drug development. For example, BI-749327, a selective TRPC6 antagonist, ameliorates renal fibrosis and dysfunction 211 , whereas deletion of Trpc6 worsens glomerular injury in Akita mice 212 . Accordingly, a loss-of-function TRPC6 variant (G757D) is associated with focal segmental glomerulosclerosis 213 . These latter observations warrant caution when using TRPC6 blockers in humans.

Although all TRP channels are evolutionarily highly conserved, their sensitivity to external stimuli show striking species-related differences. For example, TRPV1 is activated by capsaicin in mammals (it was originally cloned as the ‘capsaicin receptor’) 10 , but not in birds 11 . However, changing position 578 in the S4/S5 helix of cTrpv1 from alanine to glutamine renders the chicken receptor capsaicin-sensitive 12 . TRPV1 also shows distinct, species-dependent, heat-activation thresholds, and so TRPV1 is a noxious heat sensor in some mammals (including humans) 13 but not in others (for example, camels that have evolutionarily adapted to desert heat) 14 . Similarly, the sensitivity of TRPA1 to cold differs between rodents and primates 15 . These species-dependent differences in channel sensitivity and function should be considered when selecting experimental animal models and interpreting the results. TRPV1 is a prime example of diversity in TRP channel expression and function. TRPV1 is highly expressed on primary sensory neurons as a major integrator of painful stimuli (afferent function) and a key initiator of neurogenic inflammation (efferent function) 16 (Fig.  2 ). Moreover, TRPV1-expressing sensory neurons have been implicated in warmth sensing 17 , 18 and itching 19 . In the viscera, neuronal TRPV1 triggers reflex pathways like cough, emesis, heart rate, micturition and intestinal peristalsis 16 . Albeit at much lower levels, TRPV1 is also expressed in various brain nuclei 20 and in non-neuronal cells 21 . Fig. 2 TRPV1 in the pain pathway: similarities and differences between desensitization by agonists and blockade by antagonists. TRPV1 is expressed on primary sensory neurons of the unmyelinated (C fibre) type. Activation of TRPV1 by agonists like capsaicin, heat and protons (and putative ‘endovanilloids’) triggers the release of pro-inflammatory neuropeptides like substance P (SP) and calcitonin gene-related peptide (CGRP), thereby initiating the biochemical cascade known as neurogenic inflammation. At the same time, an impulse is generated that is perceived in the brain as pain or itching. Protein kinase C (PKC) and nerve growth factor (NGF) lower the activation threshold of TRPV1 (sensitization). Stimulation of TRPV1 with a therapeutic dose of capsaicin results in a lasting (up to months) but fully reversible state in which the previously excited neurons remain unresponsive (‘silent’) to further challenge; traditionally, this is called desensitization. High-dose capsaicin patches and site-specific injections relieve pain by this mechanism. There is an ill-defined line between reversible desensitization and irreversible neurotoxicity achieved by therapeutic and supratherapeutic doses of capsaicin, respectively. Although both desensitization and neurotoxicity depend on capsaicin-induced Ca 2+ influx through the TRPV1 channel, the downstream molecular mechanisms are still poorly understood. Mitochondrium swelling is an early ultrastructural sign of capsaicin neurotoxicity. Ca 2+ is thought to sequester in mitochondria, where it triggers apoptosis using a molecular pathway that includes caspase activation. Intrathecal resiniferatoxin is used as a ‘molecular scalpel’ to achieve permanent analgesia in cancer patients with chronic, intractable pain by ablating sensory neurons. In principle, a similar strategy can be used to kill cancer cells that overexpress TRPV1 compared with normal counterparts. We note that small molecule TRPV1 antagonists prevent only TRPV1 activation by agonists, leaving other pain sensors intact. TRPV1 is expressed on primary sensory neurons of the unmyelinated (C fibre) type. Activation of TRPV1 by agonists like capsaicin, heat and protons (and putative ‘endovanilloids’) triggers the release of pro-inflammatory neuropeptides like substance P (SP) and calcitonin gene-related peptide (CGRP), thereby initiating the biochemical cascade known as neurogenic inflammation. At the same time, an impulse is generated that is perceived in the brain as pain or itching. Protein kinase C (PKC) and nerve growth factor (NGF) lower the activation threshold of TRPV1 (sensitization). Stimulation of TRPV1 with a therapeutic dose of capsaicin results in a lasting (up to months) but fully reversible state in which the previously excited neurons remain unresponsive (‘silent’) to further challenge; traditionally, this is called desensitization. High-dose capsaicin patches and site-specific injections relieve pain by this mechanism. There is an ill-defined line between reversible desensitization and irreversible neurotoxicity achieved by therapeutic and supratherapeutic doses of capsaicin, respectively. Although both desensitization and neurotoxicity depend on capsaicin-induced Ca 2+ influx through the TRPV1 channel, the downstream molecular mechanisms are still poorly understood. Mitochondrium swelling is an early ultrastructural sign of capsaicin neurotoxicity. Ca 2+ is thought to sequester in mitochondria, where it triggers apoptosis using a molecular pathway that includes caspase activation. Intrathecal resiniferatoxin is used as a ‘molecular scalpel’ to achieve permanent analgesia in cancer patients with chronic, intractable pain by ablating sensory neurons. In principle, a similar strategy can be used to kill cancer cells that overexpress TRPV1 compared with normal counterparts. We note that small molecule TRPV1 antagonists prevent only TRPV1 activation by agonists, leaving other pain sensors intact. TRPV1 is unique among drug targets in that its initial excitation by agonists is followed by a lasting refractory state (traditionally referred to as desensitization) in which TRPV1-expressing neurons are not responsive to both a repeated capsaicin challenge and to various unrelated stimuli 16 (Fig.  2 ). The role of TRPV1 in thermoregulation is well established 22 , 23 . In rodents, activation of TRPV1 with capsaicin initiates heat-loss behaviour at warm ambient temperatures (30–32.5 °C) and mice exhibit ‘red ear’ caused by vasodilation and seek the cool surface of the cage 16 . By contrast, TRPV1-deficient ( Trpv1 –/ – or capsaicin-desensitized) mice display deficiencies in heat-loss mechanisms (such as body licking) and develop hyperthermia when exposed to 35 °C (ref. 24 ). In humans, capsaicin may cause gustatory sweating as a mechanism of heat loss 16 , whereas TRPV1 antagonists can increase or decrease body temperature, or even leave it unchanged (‘thermoneutral antagonists’) 25 – 27 . These thermoregulatory side effects can be exploited for pharmacotherapy: TRPV1 antagonists that cause hyperthermia as a dose-limiting, on-target adverse effect may also help to restore normal body temperature after medical cooling 28 , 29 . In turn, TRPV1 agonists that cause hypothermia may protect the brain after stroke 30 . As reviewed elsewhere 16 , 22 , 23 , there is no consensus in the literature as to the exact anatomic site of the TRPV1-expressing thermoregulatory centre, or the mechanism by which it regulates body temperature. Cryo-electron microscopy and X-ray crystallography have provided novel insights into TRP channel structure and function and highlighted several druggable sites (Fig.  3 ). TRPV1 has fourfold symmetry with different pore profiles for ligand-bound structures 31 , and a vanilloid-binding pocket deep within the membrane bilayer 32 (Fig.  3b , f ). TRPA1 is a sentinel for electrophilic irritants and has a distinct allosteric nexus where a covalent modification of cysteine residues regulates channel activity 33 (Fig.  3c ). Indeed, all known synthetic TRPA1 antagonists, such as A-967079, act as negative allosteric modulators (Fig.  3b ). This is important because they can block TRPA1 (over)activation, and yet leave some level of physiological activity intact, in contrast to traditional orthosteric antagonists. Fig. 3 Structural information on ligand and drug binding by TRP channels, and gain-of-function mutations. Protein colouring is the same as in Fig.  1 . a − d | Schematic descriptions of TRPV ( a ), TRPA1 ( b ), TRPM ( c ) and TRPM1 ( d ) illustrate the general binding site locations identified in structural findings of TRP channels. Similar ligand shapes indicate similar ligand-binding site locations in the respective proteins, and example ligands are labelled. Black stars indicate mutations associated with human diseases that are located at the elbow connection between the S4−S5 linker and the S5 helix, and magenta stars in the S6 helix pore gate indicate mutations that lead to hypersensitive or constitutively active channels and gain-of-function phenotypes. Grey stars mark the TRPM2 D543E hypoactive mutation. e | Sample ligand-binding sites from the antagonist capsazepine bound to rat TRPV1 (PDB ID 5IS0). f | Co-agonists PIP2, Ca 2+ and icilin bound to Ficedula albicollis TRPM8 (PDB ID 6NR3). g | Agonist ML-SA1 bound to human TRPML1 (PDB ID 6E7Z). Protein colouring is the same as in Fig.  1 . a − d | Schematic descriptions of TRPV ( a ), TRPA1 ( b ), TRPM ( c ) and TRPM1 ( d ) illustrate the general binding site locations identified in structural findings of TRP channels. Similar ligand shapes indicate similar ligand-binding site locations in the respective proteins, and example ligands are labelled. Black stars indicate mutations associated with human diseases that are located at the elbow connection between the S4−S5 linker and the S5 helix, and magenta stars in the S6 helix pore gate indicate mutations that lead to hypersensitive or constitutively active channels and gain-of-function phenotypes. Grey stars mark the TRPM2 D543E hypoactive mutation. e | Sample ligand-binding sites from the antagonist capsazepine bound to rat TRPV1 (PDB ID 5IS0). f | Co-agonists PIP2, Ca 2+ and icilin bound to Ficedula albicollis TRPM8 (PDB ID 6NR3). g | Agonist ML-SA1 bound to human TRPML1 (PDB ID 6E7Z). Epigenetic regulation of TRPs is an emerging area of research. In rats, histone H3 acetylation at the Trpv1 promoter region leads to increased TRPV1 protein expression in sensory neurons with concomitant visceral hyperalgesia 34 . In mice, SUMOylation protects TRPV1 from metabolic damage during experimental diabetes and thus delays the development of neuropathic pain 35 . TRPV1 from human donor tissue is also SUMOylated 35 . Likewise, methylation of the human TRPA1 promoter region can result in increased TRPA1 levels and altered pain perception 36 . Indeed, TRPA1 gene methylation is dysregulated in patients with Crohn disease, contributing to visceral pain 37 . Genetic defects in TRPs — or TRP channelopathies — are increasingly recognized causes of hereditary human disease 4 . For example, TRPV4 channelopathy 38 is linked to at least nine different diseases, ranging from autosomal dominant brachyolmia type 3 to Charcot–Marie–Tooth neuropathy type 2 . Furthermore, polymorphisms in TRP genes may regulate disease risk, as exemplified by the reduced migraine incidence in carriers of rs10166942, which correlates to reduced TRPM8 gene expression 39 .

Medical control of chronic pain is frequently unsatisfactory, and the current therapeutic pain market remains dominated by agents that have been around for decades. Narcotics (opioids) are effective painkillers, but, acting in the brain, they are also addictive. A logical strategy to avoid opioid side effects is to target the beginning of the pain pathway: the nociceptor where pain is generated 25 , 40 . Hence there is tremendous interest expressed by pharmaceutical companies in TRP channels that detect noxious stimuli in the periphery (Fig.  4 ). Fig. 4 The pain pathway: wide expression of TRP channels in sensory neurons, skin and brain. Several TRP channels in free nerve endings or sensory corpuscles of sensory neurons transduce painful thermal (TRPV1, TRPA1, TRPM8 and TRPM3), chemical (TRPA1) and mechanical (TRPC1, TRPC3, TRPC6, TRPA1, TRPV2 and TRPV4) stimuli into electrophysiological excitation. TRP channels that do not participate in physiological sensation, but are recruited under pathophysiological conditions (for example, TRPC5 in mechanosensation) are of particular interest. TRP channels in keratinocytes, satellite glia and Schwann cells (TRPV1, TRPA1) also participate in transduction of painful stimuli. TRP channels in the central terminals of sensory neurons amplify and modulate neurotransmitter release in the spinal cord. TRP channels in the brain play critical parts in several physiological processes. It is likely that different TRP channels in the brain have different roles in acute nociceptive versus maladaptive chronic pain. Pain representation in the brain shifts from from nociceptive circuits in acute pain (left brain scan) to emotional circuits during chronic pain in functional magnetic resonance imaging (middle and right brain scans). Therefore, an assessment of central action or peripheral restriction of a TRP channel modulator needs to be carefully performed to maximize efficacy and therapeutic window. Several TRP channels in free nerve endings or sensory corpuscles of sensory neurons transduce painful thermal (TRPV1, TRPA1, TRPM8 and TRPM3), chemical (TRPA1) and mechanical (TRPC1, TRPC3, TRPC6, TRPA1, TRPV2 and TRPV4) stimuli into electrophysiological excitation. TRP channels that do not participate in physiological sensation, but are recruited under pathophysiological conditions (for example, TRPC5 in mechanosensation) are of particular interest. TRP channels in keratinocytes, satellite glia and Schwann cells (TRPV1, TRPA1) also participate in transduction of painful stimuli. TRP channels in the central terminals of sensory neurons amplify and modulate neurotransmitter release in the spinal cord. TRP channels in the brain play critical parts in several physiological processes. It is likely that different TRP channels in the brain have different roles in acute nociceptive versus maladaptive chronic pain. Pain representation in the brain shifts from from nociceptive circuits in acute pain (left brain scan) to emotional circuits during chronic pain in functional magnetic resonance imaging (middle and right brain scans). Therefore, an assessment of central action or peripheral restriction of a TRP channel modulator needs to be carefully performed to maximize efficacy and therapeutic window. Desensitization to capsaicin has a clear therapeutic potential (Fig.  2 ). Indeed, high-dose capsaicin patches 41 , 42 and site-specific injections 43 are clinically proven to provide meaningful pain relief in patients with osteoarthritis, post-herpetic neuralgia and diabetic polyneuropathy. Peripheral expression of TRPV1 on nociceptive fibres is predictive of the patient’s response to capsaicin therapy, which may explain the discrepant outcome in clinical trials with capsaicin in patients with diabetic polyneuropathy in whom nociceptive fibres often degenerate during advanced disease 44 . Capsaicin evokes an intense initial pain reaction that limits the dose that patients can tolerate 16 . To reduce this adverse effect, ‘non-pungent’ TRPV1 agonists like olvanil (NE19550) 45 and MRD-652 (ref. 46 ) have been developed that differ from capsaicin in the activation kinetics of the receptor. These compounds showed promise in animal models of pain 45 , 46 , but their clinical value remains to be demonstrated. The ultrapotent capsaicin analogue, resiniferatoxin, is undergoing clinical trials as a ‘molecular scalpel’ with which to achieve permanent analgesia in severe osteoarthritic pain 47 , 48 , and in cancer patients with chronic intractable pain 49 . This approach has already succeeded in the veterinary clinic, in which intrathecal resiniferatoxin provided lasting pain relief and restored ambulation in dogs with osteosarcoma 50 . Resiniferatoxin has been tested in a small number of female patients with cervical cancer metastatic to pelvic bone, and the results are promising so far 51 . Because TRPV1 agonists that cause desensitization can be perceived as functional antagonists of the receptor (Fig.  2 ), it has been postulated that small molecule TRPV1 antagonists can be also pursued as therapeutic agents 25 , 52 , 53 . Indeed, TRPV1 is a highly druggable target 52 . Various antagonist chemotypes have been discovered, and many have matured into clinical lead molecules 25 , 52 , 53 . The efficacy of TRPV1 antagonists in preclinical pain models varied; some showed efficacy in both inflammatory (for example, complete Freund adjuvant-induced arthritic pain ) and neuropathic pain models (such as the Chung model ), whereas others were active only in the inflammation models 25 , 53 . We note that both the magnitude of the analgesic effect and the dose needed to demonstrate efficacy varied. When using hypothermia by capsaicin as an on-target engagement model, many TRPV1 antagonists actually caused the opposite effect, hyperthermia 25 . However, this febrile reaction could be managed with simple antipyretic agents like acetaminophen and disappeared upon repeated dosing, paving the way to studies in human volunteers 54 . Since Trpv1 –/– mice are less responsive to noxious heat 55 , 56 , and rodents desensitized to capsaicin show dramatically increased noxious heat threshold in the hot plate test 16 , it was unsurprising that TRPV1 antagonists impaired the cutaneous noxious heat sensation in humans, leading to burn injuries as a common adverse effect 53 . Owing to these unacceptable on-target adverse effects, many first-generation TRPV1 antagonists were withdrawn from clinical trials: some, like AMG517, because of febrile reactions 57 and others, like MK2295, because of the burn injuries 58 . Some antagonists that progressed into phase II efficacy trials failed to demonstrate analgesic activity, such as a terminated trial of AZD1386 for osteoarthritic pain 59 . Other clinical trials were terminated without explanation, such as those sponsored by Sanofi/Glenmark (GRC-6211) and PharmEste (PHE575). Since functional TRPV1 expression on nociceptors predicts the patient’s response to agonist (capsaicin) treatment 42 , one may argue that TRPV1 antagonist trial participants should also be selected on the basis of their capsaicin sensitivity. It is also worth mentioning that vitamin D was recently shown to act as a partial agonist of TRPV1 at physiologically relevant free plasma concentrations 60 . A partial agonist can either act as an agonist or antagonist, depending on the presence of other ligands. Such effects may contribute to the variability of TRPV1 antagonists in clinical trials. Although these problems tempered expectations, and many companies abandoned TRPV1 as an analgesic target, there have been a few promising developments. For example, the second-generation molecule, JNJ-39439335 (mavatrep), showed significant improvement versus placebo in stair climbing-induced clinical pain in participants with knee osteoarthritis 61 . Another compound, NEO6860, also showed an analgesic trend, although it did not statistically outperform placebo, without affecting body temperature or heat pain perception 62 . It remains to be seen whether these molecules — or other ‘thermoneutral’ antagonists, like GRTE16523 (ref. 63 ) — can demonstrate meaningful analgesic activity in the clinic. Ultimately, it will probably be easier to dissociate beneficial heat sensing from therapeutic inhibition of TRPV1 activity by targeted antagonist delivery because most, if not all, inhibitors of TRPV1 activation will inhibit heat sensing. The role of TRPV1 phosphorylation by protein kinase-C (PKC) in the development of inflammatory hyperalgesia is well established 64 . Eliminating the PKC phosphorylation site S801 by CRISPR/Cas9 editing in TRPV1 reduced the ongoing pain caused by masseter muscle inflammation without blocking physiological TRPV1 functions 65 . This observation raises the possibility of engineering novel TRPV1 antagonists that selectively interact at the phosphorylated TRPV1. As to future clinical testing, a promising indication for TRPV1 analgesia is gout. In experimental animals, urate crystals activate TRPV1, and the resultant pain is blocked by both genetic inactivation ( Trpv1 –/– mice) and pharmacological blockade (SB-366791) 66 . In a mouse model of gout arthritis, eucalyptol reduced both inflammation and pain by preventing the urate-induced up-regulation of TRPV1 expression in ankle tissues 67 . Furthermore, SB-366791 attenuates dental pain in rats 68 , and, as an added benefit, prevents alveolar bone loss in a rat model of periodontal disease 69 . Other potential indications with increased TRPV1 mRNA levels include endometriosis 70 and chronic lower back pain 71 . However, because much of the earlier preclinical efficacy data that predicted clinical value in inflammatory pain (such as osteoarthritic pain) did not translate in patients, caution is advised for these new potential indications. TRPA1 gene variants have been linked to paradoxical heat (painful cold) sensation 72 , sickle cell pain crisis 73 , carbamazepine-responsive cramp-fasciculation syndrome 74 , and familial episodic pain syndrome (FEPS) 75 . FEPS is a gain-of-function TRPA1 channelopathy that causes debilitating upper body pain due to increased inward currents in response to TRPA1 channel activation at normal resting potentials. From a clinical perspective, this observation is not easy to interpret as patients carry both mutant (human TRPA1 N855S) and wild-type alleles. TRPA1 is a tetramer and it is reasonable to assume that patients express various mixtures of normal and mutant channels 76 . Although one cannot predict the properties of these heteromultimers, the N855S mutation, like the analogous gain-of-function mutations in TRPV4 and TRPM4 , is in the linker that moves during channel gating (Fig.  3b − d ), explaining how it causes channels to open more readily. An antagonist that selectively blocks the mutant FEPS protein would be optimal, but companies are unlikely to invest resources into such a small market. Genetic deletion or pharmacological blockade of TRPA1 vastly attenuate responses to many harmful chemical stimuli, ranging from formaldehyde, through acrolein (present in cigarette smoke) and diesel exhaust, to tear gases 77 , 78 . Moreover, endogenous compounds like methylglyoxal 79 , 80 (a product of aberrant glucose metabolism) and reactive oxygen species (ROS) and nitrogen species all converge on TRPA1 (ref. 81 ): these electrophilic compounds can activate the channel after covalently modifying a hypersensitive cysteine in the cytoplasmic domain of TRPA1 by electrophilic attack 82 (Fig.  3c ). The temperature (cold) sensitivity of TRPA1 is subject to much debate and will not be discussed here, except to mention that Trpa1 –/– mice show attenuated cold allodynia evoked by anticancer drugs like oxaliplatin 83 , and by ischaemia and reperfusion injury of the rodent hindlimb, a murine model of complex regional pain syndrome type I (CRPS-I) 84 . Importantly, TRPA1 antagonists recapitulated the effects of genetic Trpa1 inactivation 84 , indicating a novel therapeutic strategy in CRPS-I patients. Trpa1 –/– mice also displayed reduced visceromotor responses to colorectal distension 85 , indicating a role for TRPA1 in mechanical pain; and Trpa1 –/– rats showed decreased pain behaviours in response to chemical agonists of TRPA1, but normal responses to other pain stimuli, including cold, and itching 86 . Animal studies with TRPA1 antagonists (such as A-967079, HC-030031 and AMG0902) suggested therapeutic potential in patients with neuropathic pain 87 . We note that CHEM-5861528, when given with streptozotocin to rats, ameliorated the pain and reduced the loss of intraepidermal nerve fibres 88 . If these results were to be confirmed in diabetic patients, this observation suggests that TRPA1 antagonists can be disease-modifying by preventing (or at least delaying) the development of diabetic polyneuropathy. As of today, only one TRPA1 antagonist has completed phase II clinical trials, Glenmark’s GRC17536 89 . Although GRC17536 significantly reduced pain scores in the non-denervation group of patients with painful diabetic polyneuropathy without worrisome side effects, the compound had problems with bioavailability/pharmacokinetics and did not progress into phase III. Topical administration may help overcome such problems. Indeed, a topical HC-030031 gel (0.05%) reversed mechanical and cold allodynia in mice after ultraviolet B-induced burn injury 90 . This implies a therapeutic value for TRPA1 antagonist creams for patients with sunburn pain or thermal injury. Interestingly, TRPA1 inhibition produces analgesia against modalities that are not mediated by TRPA1-expressing neurons 91 . This behaviour is not unprecedented, and has been observed in other TRP receptors, such as TRPV1. Although mechanosensitive nerves do not express TRPV1, resiniferatoxin desensitizes capsaicin-sensitive afferents, resulting in ameliorated mechanical hyperalgesia (and, paradoxically, cold hyperalgesia) in a murine model of arthritic pain 92 . The molecular underpinnings of this phenomenon are not clear. One possibility is that TRPV1 expression is plastic and nerve fibres that do not express TRPV1 under normal conditions might do so under pathological conditions, such as pain 93 . Similar considerations may also apply to TRPA1. The journey of TRPA1 antagonists as pain therapeutics to the clinic was given new impetus by the recent purchase by Eli Lilly of the Hydra Biosciences molecules 94 . As of today, TRPA1 small molecule patents have been filed by Ajinomoto Co., Algomedix Inc, Almirall S.A., Boehringer Ingelheim, EA Pharm Co., Eli Lilly, Galderma, Genentech/Roche, Glenmark, Mandom Corp and Orion Corp 95 . According to a new model of sensory coding, there are two distinct neuronal populations involved in warmth perception: one population is excited by warmth, whereas the other is blocked by it 18 . This latter population expresses TRPM8 and displays ongoing cool-driven firing 18 . Accordingly, Trpm8 –/– mice are incapable of distinguishing warm from cold, and show markedly attenuated cold allodynia 96 , 97 . Moreover, in humans, reduced TRPM8 gene expression is associated with a reduced risk for migraine, and reduced sensitivity to cold and cold pain 39 . These observations imply a therapeutic potential for small molecule TRPM8 antagonists in the management of cold-induced pain and migraine 98 . Potent and selective TRPM8 antagonists with acceptable clinical safety profiles have been discovered 98 , many of which bind competitively to the menthol or icilin binding site in the S1−S4 sensor domain (Fig.  3g ). Accordingly, these antagonists demonstrated a clear-cut exposure−efficacy relationship in preclinical models, including icilin-induced wet dog shakes in rats, cold pressor test in rodents, and menthol challenge in guinea pigs; these findings predict target engagement for therapeutic dose in humans 98 . Interestingly, like TRPV1, TRPM8 displayed a basal activation tone 99 . TRPM8 antagonists evoked a mild hypothermic response, but this was not considered a hurdle to clinical use 100 . Some TRPM8 antagonists (such as AMG-333 and PF-05105679) 100 , 101 , have already progressed into clinical studies, in which they reduced pain in the cold pressor test. Although PF-05105679 did not cause perceptible hypothermia, some volunteers reported a ‘hot feeling’ in the perioral area that was deemed to be non-tolerable by two study subjects 100 . AMG-333 also caused a few grade 1 adverse effects related to TRPM8 antagonism 101 . Although most of the attention was focused on antagonists, TRPM8 agonists (such as WS-12 and di-isopropyl-phosphinoyl-alkane, DIPA) also have an analgesic potential. For instance, DIPA was shown to reduce spontaneous painful contractions in the human distal colon 102 . The classical pain targets of TRPV1, TRPA1 and TRPM8 are all expressed in nociceptors 25 , 40 . The expression pattern of novel TRP targets for pain relief is more diverse, ranging from sensory neurons (TRPM3) through brain nuclei involved in pain processing (TRPC4 and TRPC5) to the epithelium (TRPV3 and TRPV4) and immune cells (TRPM2). Gain-of-function mutations in TRPV3 were discovered in patients with Olmstedt disease 103 , 104 , erythromelalgia 104 , and painful plantar keratoderma 105 . As with TRPA1 above, many of these TRPV3 gain-of-function mutations are at the elbow connecting the S4−S5 linker and the S5 helix (Fig.  3b ). Interest in TRPV3 as an analgesic target was first raised by the observation that tissue damage upregulates TRPV3 expression in human sensory ganglia or skin 106 . Several companies developed potent and selective TRPV3 antagonists 107 . In 2014, the Sanofi-Aventis/Glenmark compound GRC15300 failed a phase II trial in patients with peripheral neuropathy, and the programme was terminated. Genetic deletion or knockdown of Trpv4 in mice leads to attenuated pain behaviour in various preclinical models. According to these studies, TRPV4 may play a pivotal part in visceral pain 108 . An interesting approach is to use TRPA1/TRPV4 dual inhibitors for chemotherapy-associated neuropathic pain 109 . The molecular mechanisms that underlie the transition from acute to chronic neuropathic pain are largely unknown, hampering drug development. In this transition, TRPV1, TRPA1 and TRPM2 are emerging as important players 110 , 111 . For instance, acute pancreatitis has a large component of neurogenic inflammation that can be attenuated by TRPV1 and TRPA1 antagonists 112 , 113 . These antagonists also prevented the spouting of pancreatic sensory nerve fibres during the acute pancreatitis attacks, and thereby averted the development of chronic neuropathic pain. As for TRPM2, it is highly expressed in immune cells — including macrophages and microglia — that contribute to inflammatory and neuropathic pain 114 . In a murine model of neuropathic pain, Trpm2 knockdown prevented the development of pain-related behaviour following chronic constriction injury 114 . TRPM3 has a central role in noxious heat sensing 115 , making the ‘triad of TRPV1/TRPA1/TRPM3’ a potential multireceptor pain target 116 . The TRPM3 agonist CIM0216 induced heat hypersensitivity in wild-type mice but not in Trpm3 –/– mice 117 . Conversely, TRPM3 antagonists reduced pain responses in several protocols in preclinical studies 118 – 120 ; again, these results were validated in Trpm3 –/– animals 117 . TRPC4 and TRPC5 are non-selective cation channels that can form homomers and heteromers and are expressed mostly in the amygdala and hippocampus (Fig.  4 ), but also in peripheral sensory neurons both in rodents 121 and humans 122 . Studies on Trpc4 –/– rats showed tolerance to visceral pain responses, whereas somatic pain responses were unaffected 123 . Furthermore, the non-selective TRPC4/TRPC5 antagonist 4-methyl-2-(1-piperidinyl)quinoline (ML-204) inhibited visceral pain responses in wild-type rats 124 , confirming the role of TRPC4 in visceral pain. Local application of ML-204 to amygdala attenuated neuropathic pain behaviour in rats with spared nerve injury 124 . Genetic inactivation ( Trpc5 –/– mice) or pharmacological blockade of TRPC5 by the small molecule antagonist AC1903 prevented the development of mechanical hypersensitivity and persistent spontaneous or tactile pain in a wide range of murine pain models, including that induced by skin incision, chemotherapy and complete Freund adjuvant injection 122 . We note that these pain conditions are associated with elevated lysophosphatidylcholine (LPC) levels, and exogenous LPS triggers TRPC5-dependent pain behaviour in naive mice 122 ; LPS also causes intensive itching during cholestasis both in rodents and nonhuman primates by activating TRPV4 in skin keratinocytes 125 . Because TRPC5 also regulates prolactin release 126 , TRPC5 antagonists may have enhanced analgesic efficacy in females given that prolactin promotes pain only in that sex. These findings suggest that centrally acting TRPC4 and TRPC5 antagonists could relieve visceral and neuropathic pain, or at least make it more tolerable 127 .

Several TRP channels show altered expression in cancers, but it is unclear whether this is cause or consequence of the disease. The use of altered TRP protein expression for cancer diagnosis and prognostication is beyond the scope of this Review. However, such altered TRP channel expression may constitute a therapeutic target. For example, high-grade astrocytoma shows increased TRPV1 expression compared with normal brain 281 . High-dose capsaicin administration can kill neurons owing to the Ca 2+ overload it causes (Fig.  2 ), and thus TRPV1 agonists may help to eradicate this brain tumour. Unfortunately, it is unclear how to deliver capsaicin to the brain in doses sufficiently high to kill tumour cells without causing unacceptable side effects. Oesophageal and head-and-neck squamous cell carcinomas also overexpress TRPV1 and TRPA1 (ref. 282 ). These cancers are more promising targets for TRPV1 and/or TRPA1 agonist therapy inasmuch as they are amenable to topical administration. TRPV1 gene polymorphism has been linked to eating habits in children 283 . There is anecdotal evidence that dietary capsaicin may help maintain a healthy body weight (‘exercise in a pill’) 284 . Even if true, it is unclear whether this is an on-target effect of capsaicin on TRPV1, or whether is due to capsaicin-induced changes in the gut microbiota 285 , as experiments with Trpv1 –/– mice provided conflicting reports 78 . TRPM5 is a calcium-activated cation channel activated downstream of taste receptors and other chemosensory receptors. Unlike wild-type mice, which overeat chocolate and become obese, Trpm5 –/– animals maintain their normal weight when on a carbohydrate-rich diet 286 . Interestingly, these animals also consume less alcohol 287 , implying value for a TRPM5 antagonist in obesity and/or alcohol use disorder. By contrast, Trpm8 –/– mice are obese owing to daytime hyperphagia 288 , whereas the TRPM8 agonist, icilin, increases energy expenditure, and reduces body weight, in mice 289 . These observations identify TRPM8 as another potential target in diet-induced obesity. Type 2 diabetes mellitus (T2DM) has reached pandemic proportions. To date, no disease-modifying treatment is available for T2DM patients. Accumulating evidence suggests that the sensory nervous system is involved in the progression of T2DM by maintaining a low-grade inflammation via TRPV1 (ref. 290 ). Indeed, oral glucose tolerance and glucose-stimulated insulin secretion were improved by both genetic inactivation ( Trpv1 −/− mice) and pharmacological blockade of TRPV1 by the small molecule antagonist BCTC 291 . The TRPV1 antagonist, XEN-D0501, is currently undergoing phase II clinical trials in T2DM patients with good tolerability and safety 292 . A combined TRPV1 and TRPA1 antagonist may also protect sensory nerves and prevent cognitive decline in T2DM patients (Box  2 ).

Although Trpa1 –/– mice lack any obvious neurological phenotype and there is little, if any, detectable mRNA signal in the central nervous system 235 , functional TRPA1 seems to be present, albeit at low levels, in the brain: in glial cells (astrocytes and oligodendrocytes) 236 , in cortical neurons 237 , in cerebral endothelium 238 and in Schwann cells 239 . This discrepancy between positive function-based TRPA1 results and weak or missing Trpa1 mRNA signals is puzzling and yet to be explained. The potential therapeutic use of brain-permeable TRPA1 antagonists for neuroprotection has emerged from an elegant electrophysiological study 240 . White matter is particularly vulnerable to hypoxia associated with stroke. Glutamate release and subsequent activation of N-methyl- d -aspartate (NMDA) receptors in oligodendrocytes is thought to underlie hypoxia-induced white matter injury 240 . Pharmacological or genetic inactivation of TRPA1 prevented hypoxia from damaging and ultimately killing oligodendrocytes after stroke 240 . On a related note, TRPV1-mediated hypothermia reduced stroke volume by 50% and promoted functional recovery after ischaemic stroke in mice 30 . Oligodendrocytes are also affected in multiple sclerosis . Cuprizone-induced demyelination is widely used as a multiple sclerosis model to assess the potential therapeutic efficacy of novel multiple sclerosis treatments. In this model, pharmacological blockade or genetic deletion of TRPA1 reduced oligodendrocyte apoptosis 241 . Furthermore, methylglyoxal, a known TRPA1 activator 79 , accumulated in the white matter lesions of multiple sclerosis 242 . Dimethyl fumarate, used to treat multiple sclerosis and psoriasis, and known to activate the antioxidant response through Nuclear factor erythroid 2-related factor 2 (NRF2), was found to activate TRPA1 in immune cells independent of NRF2 (ref. 243 ). This novel NRF2-independent mechanism may contribute to the peripherally restricted immunosuppressive action of dimethyl fumarate. Trpa1 –/– mice show reduced anxiety-like behaviour, and this beneficial effect can be reproduced with TRPA1 antagonists in healthy, wild-type mice 244 . The neural circuit responsible for the antidepressant action of TRPA1 antagonism is currently unknown. Importantly, ageing Trpa1 –/– mice also exhibited improved memory (fewer reference memory errors), implying a role for TRPA1 in age-related memory decline 245 . Together, these findings identify TRPA1 as a potential target in the pharmacotherapy of anxiety and dementia, a common combination in the elderly (for TRPA1 in capillary endothelium in the brain, and memory decline in diabetic patients, see Box  2 ). However, Trpa1 –/– mice also show defects in white matter myelination, with the myelin basic protein level downregulated and the number of mature oligodendrocytes reduced 246 . According to this observation, TRPA1 may play a critical part in the maturation process of oligodendrocytes, because the myelination process is likely to continue throughout life during new skill learning, and therefore chronic and complete block of TRPA1 in oligodendrocytes may interfere with myelination. However, robust therapeutic efficacy can probably be achieved during intermittent partial pharmacological block of TRPA1 in humans, allowing preservation of the physiological house-keeping function of TRPA1. TRPA1 in astrocytes may constitutively regulate excitatory neurotransmission in healthy brain 237 . If confirmed, central TRPA1 inhibition may interfere with memory formation, contradicting the memory phenotype observed in Trpa1 –/– mice 245 . By contrast, a phase I study of a centrally acting TRPA1 antagonist in healthy volunteers revealed no central nervous system toxicity issues at concentrations that were several-fold above the IC 50 value 247 . One potential explanation for this discrepancy is that constitutive TRPA1 activation in brain slice astrocytes is an experimental artefact: large amounts of reactive lipid hydroperoxides and other putative TRPA1 agonists may be leaking from damaged and dead cells. Hippocampal slices from an Alzheimer disease (AD) mouse model provided compelling evidence that TRPA1 activation by β amyloid contributes to network hyperexcitability, possibly driving silent seizures 248 . This is important because antiepileptic drug use in AD is associated with increased risk for stroke 249 . Central TRPA1 antagonism may provide a safe therapy for seizures in AD. Parenthetically, in the triple-transgenic AD mouse model (3xTg-AD +/+ ), Trpv1 –/– animals had better memory function and lower tau accumulation in the hippocampus compared with Trpv1 wild-type mice 250 . The challenge, of course, is to deliver TRPA1 and/or TRPV1 antagonists to the central nervous system without causing unacceptable systemic side effects. TRPM2 is a well established redox sensor 251 and a potential target in hypoxic-ischaemic brain injury 252 . TRPM2 is implicated in neuronal death by ROS 251 , 252 . Indeed, the TRPM2 antagonist JNJ-28583113 protects cells from oxidative stress-induced death 253 . In a mouse model of cardiac arrest and cardiopulmonary resuscitation, TRPM2 inhibition improved functional recovery following cerebral ischaemia 254 . In the hippocampus of patients with major depressive disorder, increased TRPM2 mRNA levels were detected, suggesting TRPM2 as a target for treating depression 255 . TRPM2 is also a susceptibility gene for bipolar disorder 256 . Trpm2 –/– mice display impaired social cognition, and a subset of bipolar patients possess the hypoactive D543E TRPM2 gene variant 256 (Fig.  3d ). The connection between TRPM2 and bipolar disorder was further strengthened by the finding that TRPM2 regulates the phosphorylation of glycogen synthase kinase-3 (GSK3), the main target of lithium 257 . In the central nervous system, TRPM3 is expressed in both glia and neurons. The discovery that primidone — a drug that has long been approved for the treatment of epilepsy and essential tremor — blocks TRPM3 at concentrations that are achieved in the brain during pharmacotherapy validates TRPM3 as a promising neurological drug target 258 . One may argue that excessive TRPM3 activation can drive white matter injury and thus a centrally acting TRPM3 antagonist may be disease-modifying by protecting white matter. Most recently, gain-of-function TRPM3 variants (such as V837M; Fig.  3d ) have been linked to intellectual disability and epilepsy 259 , although it is not clear how this observation can be exploited for pharmacotherapy. TRPM4 is a Ca 2+ -activated non-selective cation channel in vascular smooth muscle that has a key role in myogenic constriction of cerebral arteries 260 . TRPM4 is closely associated with the sulfonylurea receptor-1 (SUR1) 261 , which explains the sensitivity of TRPM4 to the antidiabetic compound glibenclamide, a SUR1 inhibitor. In rodent models, spinal cord injury upregulated TRPM4 during secondary haemorrhage, and the increase in haemorrhage was prevented by genetic silencing of Trpm4 by antisense in rats or in Trpm4 –/– mice 262 . In rats, siRNA-mediated silencing of Trpm4 reduced infarct volume in a permanent stroke model, indicating that TRPM4 inhibition may improve blood–brain barrier integrity after ischaemic stroke reperfusion 263 . SUR1-TRPM4 forms a heteromer complex with aquaporin-4, which was shown to amplify ion−water osmotic coupling and thereby drive astrocyte swelling in stroke. These observations motivated investigation of the clinically well tolerated glibenclamide as a SUR1-TRPM4 antagonist in stroke patients 264 . In a small-scale study, glibenclamide was shown to improve outcome 234 . A phase III clinical trial with intravenous glibenclamide (BIIB093) is currently recruiting stroke patients ( NCT02864953 ) 265 . TRPM4 was also shown to mediate axonal and neuronal degeneration in experimental autoimmune encephalomyelitis and such effects could be antagonized by glibenclamide, suggesting that TRPM4 inhibition may be a novel target for multiple sclerosis 266 . Behavioural studies with Trpc4 –/– and Trpc5 –/– mice showed positive results in models that predict anxiolytic and antidepressant action 267 , 268 . This effect could be replicated by a small molecule, a TRPC1/TRPC4/TRPC5 pan-inhibitor, HC-070, which implies a therapeutic potential to treat anxiety disorders 269 . Hydra Biosciences and Boehringer Ingelheim have started a phase I trial with their TRPC4/5 inhibitor 270 . The Trpc4 –/– rats also displayed reduced cocaine self-administration without deficits in learning for natural rewards 271 . If this observation is confirmed in humans, a TRPC4 antagonist may prove clinically useful in addiction. TRPC5 has also been implicated in oxidative neuronal death, implying a role in neuroprotection in neurodegenerative diseases like Huntington disease 272 . The recent availability of potent, selective TRPC5 antagonists like NU-6027 and AC-1903 may help test this hypothesis. TRPML1 is a Ca 2+ -permeable non-selective cation channel expressed in lysosomes 273 . Loss-of-function TRPML1 mutations cause type-IV mucolipidosis 274 , a rare genetic disease with predominantly neurological symptoms. Lysosomal Ca 2+ signalling by TRPML1 regulates autophagy and lysosomal biogenesis 275 , and TRPML1 levels in lysosomes are controlled by the transcription factor-EB (TFEB) 276 . The subcellular localization and activity of TFEB is also regulated by phosphorylation mediated by Mammalian Target of Rapamycin (mTOR) 275 . The dephosphorylated form of TFEB translocates into the nucleus, where it induces the transcription of target genes, including TRPML1 (ref. 275 ). TFEB has attracted recent attention for its ability to clear pathogenic molecules in mouse models of Parkinson and Alzheimer disease 277 . In Parkinson disease mouse models, TRPML1 regulates α-synuclein exocytosis in dopaminergic neurons 278 . On the basis of these observations, a small molecule TRPML1 agonist could be a disease-modifying agent in neurodegenerative diseases. A recent structure of TRPML1 with the agonist ML-SA1 bound under the pore helix 279 — analogous to propofol in TRPA1 or cannabidiol (CBD) in TRPV2 — could aid in developing such agonists (Fig.  3e , f ). We note that Merck purchased Calporta, a company developing TRPML1 agonists for US$576 million in November 2019 (ref. 280 ).

The huge interest of pharmaceutical companies in TRPV1 antagonists was based on two premises: first, the animal models in which capsaicin desensitization inhibited pain dramatically, and second, the belief that expression of TRPV1 was restricted to nociceptive fibres. Unfortunately, neither has turned out to be true. Those initial animal models are now recognized as poor predictors of clinical efficacy, and TRPV1 expression is much broader than initially thought. Despite these disappointments, the TRP family remains an exciting and potentially rewarding group of therapeutic targets for a broad range of diseases. Clearly, the one-size-fits-all approach is not feasible and tests need to be developed to identify patient subgroups that may benefit from the therapy. Current TRP channel-modulator drug discovery efforts are guided by human genetics to allow identification of specific indications and patient groups that could benefit from pharmacotherapy (Box  3 ). Furthermore, appropriate target engagement and safety biomarker studies should help define clinically meaningful doses and therapeutic windows. Central unwanted side effects may be minimized by using peripherally restricted antibodies, antibody−drug conjugates, and engineered proteins, whereas systemic adverse effects can be avoided by targeted, site-specific therapy (Box  4 ). We predict that novel treatment modalities, such as engineered proteins, oligonucleotides and gene-based therapies, will be increasingly explored to treat human TRP channel-associated diseases in the future. The complex role of TRP channels in various physiological functions poses the question of whether the risk−benefit ratio of a given TRP channel is attractive enough to start a drug discovery programme that may deliver novel drug candidates after only 5–10 years of hard work. Another issue related to drug target validation is how well efficacy and safety findings in animal models translate to human studies. It is generally accepted that incorporating human-relevant data at early stages of the drug discovery process increases the likelihood of success. A recent exciting development to accelerate drug target validation is to use large-scale, real-world patient cohorts from biobanks and other resources to associate phenotypes with genotypes in an unbiased manner 301 . Indeed, there is evidence that targets with genetic support are approximately twice as likely to lead to an approved drug compared with those without such support. A genetic variant of a given protein may be associated with a particular disease-relevant phenotype without obvious safety implications or, vice versa, a variant may be implicated in a given disease along with several additional unwanted phenotypes. The recent availability of genetic epidemiological tools such as phenome-wide association studies 301 and Mendelian randomization, as well as large-scale real-world clinical datasets, allows for the reliable prediction of in vivo consequences of the modulation of a given target directly in humans. Efficacy and safety signals derived from such ‘virtual’ trials are increasingly likely to drive the benefit−risk discussions for individual TRP channels. Is it possible to achieve further improvement in ‘drug-likeness’ of TRPA1 antagonists: that is, improved solubility and metabolic stability while maintaining high potency? Only a few therapeutic TRP channel selective antibodies have so far been developed. Will more effort be put into developing therapeutic TRP channel antibodies in the future? Will novel technologies such as cryo-EM reveal high-affinity antibody binding sites? Would alternative treatment modalities, such as antibodies, siRNA or gene therapies, offer any advantages over small molecules? What are the roles of TRP channel splice variants under various disease conditions? Can cell compartment and or tissue-specific splice variants be therapeutically targeted? Is it possible to pharmacologically modulate intracellular TRP channels to treat chronic diseases such as lysosomal storage disease, neurodegenerative disease or intracellular pathogens through modulation of lysosomal TRP channels? Most TRP channel modulators are negative allosteric modulators (NAMs). We need to understand how a NAM interacts with the TRP channel activated by natural agonist in a given disease to build a quantitative PK-PD model of drug action in vivo. Do we understand well enough which endogenous agonists drive the disease process? TRPA1 in astrocytes was recently shown to be modulated by non-invasive, low-intensity, low-frequency ultrasound (LILFU) 302 . Will LILFU open up new avenues for non-pharmacological modulation of a disease 303 ? Could LILFU improve focused delivery of big molecules to the brain through the blood–brain barrier in a safe way? Can we use drugs that selectively activate (or block) phosphorylated (or otherwise post-translationally modified) TRP proteins? If so, can such selectivity mitigate adverse effects? Using our understanding of TRP channel structures 304 , can we design drugs that selectively target distinct functional states of TRP channels? Is it possible to achieve further improvement in ‘drug-likeness’ of TRPA1 antagonists: that is, improved solubility and metabolic stability while maintaining high potency? Only a few therapeutic TRP channel selective antibodies have so far been developed. Will more effort be put into developing therapeutic TRP channel antibodies in the future? Will novel technologies such as cryo-EM reveal high-affinity antibody binding sites? Would alternative treatment modalities, such as antibodies, siRNA or gene therapies, offer any advantages over small molecules? What are the roles of TRP channel splice variants under various disease conditions? Can cell compartment and or tissue-specific splice variants be therapeutically targeted? Is it possible to pharmacologically modulate intracellular TRP channels to treat chronic diseases such as lysosomal storage disease, neurodegenerative disease or intracellular pathogens through modulation of lysosomal TRP channels? Most TRP channel modulators are negative allosteric modulators (NAMs). We need to understand how a NAM interacts with the TRP channel activated by natural agonist in a given disease to build a quantitative PK-PD model of drug action in vivo. Do we understand well enough which endogenous agonists drive the disease process? TRPA1 in astrocytes was recently shown to be modulated by non-invasive, low-intensity, low-frequency ultrasound (LILFU) 302 . Will LILFU open up new avenues for non-pharmacological modulation of a disease 303 ? Could LILFU improve focused delivery of big molecules to the brain through the blood–brain barrier in a safe way? Can we use drugs that selectively activate (or block) phosphorylated (or otherwise post-translationally modified) TRP proteins? If so, can such selectivity mitigate adverse effects? Using our understanding of TRP channel structures 304 , can we design drugs that selectively target distinct functional states of TRP channels?

In the skin and the eye, TRPs are attractive pharmacological targets because they are amenable to topical therapy, probably reducing the risk of serious adverse effects. TRP channels are broadly expressed in various skin cell types (including keratinocytes, melanocytes, skin appendage cells, nerve endings and immune cells), with functions ranging from skin barrier function and hair growth through wound healing to cutaneous inflammation and itching 19 , 78 , 214 . TRPA1, TRPV3, TRPV4 and, to a lesser degree, TRPV1 and TRPM8 are all promising targets to relieve itching 19 . TRPA1 expression is elevated in human psoriatic skin biopsies 215 . Accordingly, Trpa1 –/– mice show reduced scratching behaviour in a murine model of chronic itching, and lack the extensive epidermal hyperplasia prevalent in psoriasis and atopic dermatitis 215 . Transgenic mice with skin-targeted, gain-of-function Trpv3 Gly573Ser (Fig.  3b ) exhibit scratching behaviour 216 . In humans, gain-of-function TRPV3 mutations (G568D and Q580P; Fig.  3b ) were described in palmoplantar keratoderma 105 , and increased TRPV3 expression was reported in post-burn pruritus 217 . Similarly, TRPV4 is overexpressed in skin biopsy samples from patients with chronic pruritus 218 , and genetic deletion of Trpv4 ameliorates itching in mouse models of chronic itch 219 . A TRPM8 agonist (menthoxypropanediol) cream was also reported to relieve human itching 220 . The TRPV1 antagonist PAC-14028 (now in phase III clinical trials) improves skin barrier function and relieves pruritus in patients with atopic dermatitis 221 . Although topical TRPV1 antagonists are generally regarded as harmless, a recent study suggests caution. In TRPV1-Ai32 optogenic mice , cutaneous light stimulation activated TRPV1-expressing neurons with resultant local-type 17 immune response 222 , in which a cascade of cytokine-mediated events leads to activation of antimicrobial responses in keratinocytes and recruitment of neutrophils to the skin. If this observation holds true in humans, patients treated with topical TRPV1 antagonists may be susceptible to cutaneous fungal and bacterial infections. TRPA1 is also involved in ultraviolet radiation-induced burn. In mice, topical administration of the TRPA1 antagonist, HC-030031, applied after irradiation blocked the development of mechanical and thermal allodynia 223 . TRPV1 and TRPV3 have been implicated in hair growth. For example, Trpv3 null mice have a thick, wavy coat of hair 224 whereas mice with constitutively active Trpv3 (the DS- Nh strain) are hairless 225 . Indeed, TRPV3 activation has been shown to inhibit human hair growth 226 . Accordingly, topical TRPV3 agonists and antagonists may be beneficial in patients with hirsutism and alopecia, respectively. Nude DS- Nh animals also develop a skin condition similar to human atopic dermatitis 227 . Furthermore, TRPV3 activation blocks lipogenesis in human sebocytes 228 , suggesting a therapeutic potential for a TRPV3 antagonist in dry skin dermatoses. TRPV1 has been implicated in psoriasis 229 and TRPV4 was named as a potential target in rosacea 230 . Recently, gain-of-function TRPM4 mutations (I1033M and I1040T; Fig.  3d ) were linked to erythrokeratodermia 231 . Even if their connection to TRP channels is still weak, all these diseases represent unmet medical needs. Eye drops containing a TRPM8 agonist help to moisturize the cornea in patients with dry eye disease 232 . In animal experiments, the TRPV4 antagonist, HC-067047, protected against the fibrosis (stromal opacification) that develops following alkali burn injury 233 . Last, intraocular TRPV1 antagonist administration has been proposed to relieve allergic conjunctivitis 234 .

TRP channels are attractive targets for the treatment of respiratory diseases 128 given their widespread expression in the lung, both in immune and structural cells, and their central role in evoking respiratory symptoms like bronchospasm and cough. Inhaled capsaicin evokes coughing in both guinea pigs and humans by activating TRPV1 on C-fibre afferents 129 . Indeed, capsaicin-containing sprays and gas canisters are used by individuals for personal protection and by law enforcement for crowd control 16 . Increased cough reflex sensitivity to inhaled capsaicin has been observed in patients with asthma, chronic obstructive pulmonary disease (COPD), idiopathic pulmonary fibrosis (IPF), and chronic idiopathic cough. These findings suggested that TRPV1 is a credible target to treat both chronic idiopathic cough and cough from inflammatory lung disease 130 . The increased TRPV1 activity may be explained by elevated levels of inflammatory mediators (such as prostaglandin E 2 , neurotrophins and bradykinin) in the airways of patients with asthma and with COPD. These endogenous substances activate airway sensory nerves, causing coughing 131 . Moreover, neurons undergo phenotypical changes in respiratory diseases both in experimental animals and human challenge models. For instance, in guinea pigs and rats challenged with ovalbumin, or in guinea pigs treated with neurotrophins, the number of TRPV1-positive neurons (particularly of the nodose-originating Aδ subtype) increases, suggesting that changes in neural pathways influenced by their local environment may result in exaggerated functional responses 132 , 133 . Moreover, TRPV1 single nucleotide polymorphisms (SNPs) have been associated with coughing, suggesting that TRPV1 variants may enhance susceptibility to coughing in smokers and in subjects with a history of workplace exposure 134 . Clinical studies tested the postulated causal role of TRPV1 in patients with chronic idiopathic cough and in chronic cough associated with COPD in patient groups with increased cough sensitivity to capsaicin. In patients with chronic idiopathic cough, the TRPV1 antagonist SB-705498 caused a small but statistically significant inhibition of capsaicin-evoked coughing, but no effect on spontaneous coughing frequency 135 . In subsequent studies, XEN-D0501 (a TRPV1 antagonist with superior efficacy and potency) did not improve spontaneous cough frequency in patients with refractory coughing 136 , nor did it affect spontaneous coughing in patients with COPD 137 , which effectively rules out TRPV1 as a relevant therapeutic target for chronic cough in these patient groups, but not in other respiratory diseases associated with exaggerated coughing. There is widespread non-neuronal TRPV1 expression in the lung, in both structural (such as fibroblast) and immune cells (such as alveolar macrophages) 138 . For example, TRPV1 was detected in human bronchial epithelial cells, with increased expression in patients with refractory asthma 139 . In bronchial epithelium, TRPV1 activation mediates mucin secretion induced by acid and particules 140 , as well as the release of cytokines 141 and ATP 142 . In a preclinical model of cigarette smoke exposure used to mimic the inflammatory response seen in COPD, the TRPV1 inhibitor JNJ17203212 reduced cigarette smoke-induced ATP release from human bronchial epithelial cells 142 . Furthermore, bronchiolar lavage fluid collected from Trpv1 –/– mice following cigarette smoke exposure had diminished ATP release and neutrophilia compared with that of wild-type mice 142 . Conversely, whole-lung homogenates from patients with COPD showed increased TRPV1 mRNA expression compared with samples from smokers without COPD and healthy non-smokers, suggesting an association between TRPV1 expression and disease pathophysiology 142 . The role of TRPV1 in asthma is controversial; some studies show no impact whereas others show protection through TRPV1 inhibition by antagonists or genetic inactivation. For instance, in rodent ovalbumin challenge models, TRPV1 blockade improved standard end points, including eosinophilia, airway hyperresponsiveness and the late asthmatic response 143 . Furthermore, TRPV1 inhibition reduced the interleukin (IL)-13-driven asthma phenotype in mice, and blocked airway hyperresponsiveness to histamine in guinea pigs following ovalbumin exposure 144 . Similarly, data from a murine house-dust mite model suggested that TRPV1, but not TRPA1, inhibition reduced airway cellular inflammation and airway hyperresponsiveness 143 . Although many of the effects seen in this study were not statistically significant, there was a consistent suppression across the functional end points measured, suggesting a role for TRPV1 in CD4 + -dependent allergic asthma models. In summary, the reasons for the observed differences in studies investigating a role for TRPV1 in allergic asthma are unclear, but could be due to differences in species, strains, antigens and interventions used to dissect TRPV1 biology. TRPV1 and TRPA1 are often co-expressed in sensory neurons that innervate the airways, but they are also found separately 128 , 138 , 145 . TRPA1 agonists (like cinnamaldehyde) evoke human vagus nerve depolarization and induce coughing in both guinea pigs and human volunteers 146 . Conversely, TRPA1 antagonists (such as GRC-17536 and HC-030031) are potential anti-tussive agents 147 , 148 . In the lung, TRPA1 is also expressed in bronchial epithelial cells and fibroblasts 149 . TRPA1 is an interesting respiratory target because it is activated by known lung irritants including natural products, such as allyl isothiocyanate, allicin and cannabinol 150 , found in mustard oil, garlic and cannabis, and by environmental irritants, such as acrolein 151 , that are present in air pollution and cigarette smoke 152 . These electrophilic agonists covalently attach to a hypersensitive cysteine in the cytoplasmic domain of TRPA1 (ref. 82 ) (Fig.  3c ). TRPA1 is also activated by reactive and electrophilic by-products of oxidative stress (such as ROS), and electrophiles like hypochlorite and hydrogen peroxide 81 , 153 . Diesel exhaust particles can also activate airway C-fibre afferents via TRPA1 (ref. 154 ). These particles contain polycyclic aromatic hydrocarbons that stimulate mitochondrial ROS production by interacting at the aryl hydrocarbon receptor. This observation links diesel exposure to respiratory symptoms 154 . In addition to environmental irritants, TRPA1 is also activated indirectly by inflammatory mediators 151 (such as bradykinin, PGE 2 and prostaglandin D 2 ), which are elevated in the bronchoalveolar lavage fluid of patients with asthma and COPD 155 . We note that the exhaled breath of patients with inflammatory airway disease is more acidic than that of healthy volunteers 156 . This is interesting because inhalation of low pH solutions, such as citric acid, causes coughing in both guinea pigs and humans. In fact, inhaled citric acid is used in the clinic to evaluate cough reflex sensitivity. Citric acid was thought to evoke cough by virtue of its low pH and subsequent activation of TRPV1 and acid-sensing ion channels (ASIC) 157 , 158 . However, TRPA1 antagonists (such as GRC-17536) also inhibited citric acid-induced coughing in conscious guinea pig models 147 . It is not yet known whether the low pH component of citric acid or the resulting increased osmolarity is responsible for activating TRPA1 (or TRPV1) to evoke coughing. TRPA1 has been implicated in the pathophysiology of allergic asthma 159 , 160 . In allergic individuals, exposure to relevant antigens can lead to an early asthmatic response followed, in certain patients, by a corticosteroid-sensitive, late asthmatic response. Despite its widespread use in the clinical assessment of new therapeutic entities, the mechanisms underlying late asthmatic response remain unclear. Following allergen challenge, activation of TRPA1 stimulates vagal broncho-pulmonary C-fibres and this induces late asthmatic response in the Brown Norway rat asthma model 159 . The mechanism of action is unclear, but the allergen probably stimulates the channel indirectly via the release of endogenous TRPA1 activators, possibly mast cell products like tryptase. TRPA1 may also have a key role in the airway hyperresponsiveness characteristic of asthma. Indeed, the TRPA1 antagonist, HC-030031, reverses airway hyperresponsiveness induced by acetylcholine in an ovalbumin mouse model 159 . Toluene di-isocyanate (TDI), a reactive compound used in the manufacture of polymeric derivatives and known to activate TRPA1 (ref. 161 ), can also evoke respiratory symptoms, including late asthmatic response, in exposed workers 162 . Non-allergic airway hyperresponsiveness can be induced by a single exposure of hypochlorite (a known TRPA1 agonist) in ovalbumin-exposed wild-type mice but not in Trpa1 – /– mice 163 . Ovalbumin-challenged wild-type rats showed signs of airway inflammation 86 , whereas Trpa1 –/– rats did not, which again suggests that TRPA1 inhibition has therapeutic potential for asthma. Lower respiratory tract infections are a leading cause of death in adults and pneumonia is the single largest cause of death in children 164 . Coughing is the main method of spreading bacteria from a human host into the environment. However, the mechanisms of coughing in patients with lower respiratory tract infections are unknown. We note that lipopolysaccharide, which is found in the outer membrane of Gram-negative bacteria, has been identified as a TRPA1 activator that exerts fast excitatory actions via TRPA1 independent of Toll-like receptor-4 (TLR4) activation 165 . This finding implicates TRPA1 as a driver of respiratory symptoms following bacterial infections and warrants further investigation. Although associations have been found between TRPA1 SNPs and susceptibility to airway disease 166 , the evidence linking TRPA1 dysfunction to respiratory disease pathophysiology is still rudimentary. Drug discovery efforts have been hampered by the limited bioavailability of TRPA1 antagonists 167 , 168 . Recently, a potent, selective and orally bioavailable small molecule TRPA1 antagonist, GDC-0334, with good target engagement in human volunteers, has been reported 169 . In preclinical models of respiratory disease, GDC-0334 inhibited cough response, airway smooth muscle hyperreactivity and oedema formation in several species 169 . It is hoped that clinical studies with GDC-0332 (or other compounds with good bioavailability) will answer the question of whether blocking TRPA1 in respiratory disorders has therapeutic promise 168 . TRPV4 is expressed in a wide range of non-neuronal cell types in human airways, including airway and vascular smooth muscle, epithelial cells, fibroblasts and inflammatory cells (such as macrophages and neutrophils) 4 . Multiple endogenous pro-inflammatory and environmental stimuli act in concert to activate TRPV4 (ref. 4 ). Accordingly, TRPV4 activation is implicated in the pathophysiology of chronic lung diseases 170 – 172 . TRPV4 was referred to as ‘the gatekeeper of pulmonary capillary permeability’. Its contribution to respiratory disease is probably best understood in the development of acute lung injury 173 . TRPV4 is involved in disrupting the epithelial and endothelial barrier function 173 , which suggests an important role in lung oedema formation associated with inflammation and tissue injury 174 . In the rodent lung, TRPV4 blockade, or knockdown, inhibited ventilator- and acid-induced lung injury, by significantly reducing the infiltration of inflammatory cells (including neutrophils and macrophages) and lung injury scores 175 , 176 . These data suggest a broad-spectrum inhibition of acute lung injury regardless of causality. Indeed, TRPV4 inhibitors are developed by the National Institute of Health (NIH) as countermeasures against chemical threats 177 . High pulmonary venous pressure is a major cause of heart failure. TRPV4 antagonists inhibit pulmonary oedema associated with heart failure in animal models 178 , paving the way towards clinical trials. On the basis of these observations, calls have been made to evaluate the effect of TRPV4 inhibitors in COVID-19 patients at risk of lung oedema 179 (Box  2 ). TRPV4 is also implicated in chronic respiratory diseases such as asthma, COPD and chronic refractory cough 170 – 172 . Its role in diverse pathologies across respiratory disease has been explained by TRPV4-induced ATP release in a pannexin 1-dependent manner with resultant activation of purinoceptors 180 , 181 . Consistent with this hypothesis, TRPV4-induced ATP release was demonstrated in human airway epithelial cells 142 , 182 and smooth muscle cells 181 . TRPV4 and purinoreceptor P2X3 antagonists were shown to inhibit Aδ sensory afferent nerve fibre activation and coughing induced by TRPV4 agonists or hypo-osmolar solutions (known to activate TRPV4) 183 . This study identified the TRPV4−ATP−P2X3 axis as a driver of airway sensory nerve reflexes such as coughing, and clinical trials with P2X3 receptor antagonists, such as AF-219, would support the role of ATP as a driver of the cough reflex 183 . In the context of asthma, TRPV4 agonists can induce a mast cell-dependent, contractile response in human bronchial and guinea pig tracheal airway smooth muscle in vitro. This effect is attributed to TRPV4-induced ATP release from airway smooth muscle 181 , which activates P2X4 receptors on mast cells to release cysteinyl leukotrienes (Cys LT); this, in turn, activates the Cys LT-1 receptor to evoke contraction 184 . COPD is an inflammatory lung disease associated with cigarette smoking. Smoke exposure was reported to cause a dose-dependent increase in ATP release — which may drive the airway inflammation — from primary human bronchial epithelial cells, and this effect was attenuated by blockers of TRPV1, TRPV4 and pannexin-1 channels 142 . Interestingly, lung tissue from patients with COPD shows an increase in TRPV4 mRNA expression compared with healthy smokers and non-smokers 142 . Genetic variants of TRPV4 have been associated with asthma 185 and COPD 186 . For example, the loss-of-function variant, TRPV4-P19S, was linked to childhood asthma 185 , and a genome-wide association study highlighted seven TRPV4 SNPs that conferred increased susceptibility to COPD 186 . It was postulated that TRPV4 modulators could influence COPD pathophysiology, but this hypothesis has yet to be tested. TRPV4 activity is upregulated in lung fibroblasts derived from patients with IPF 187 . In IPF models, Trpv4 –/– mice are protected from fibrosis 187 . TRPV4 modulates transforming growth factor (TGF)-β1-dependent actions in a SMAD -independent manner with enhanced actomyosin remodelling and increased nuclear translocation of the α-smooth muscle actin-transcription co-activator, myocardin-related transcription factor (MRTF)-A 188 , 189 . These data point to TRPV4 inhibition as a potential therapeutic approach in pulmonary fibrosis 189 . Interestingly, like asthma and COPD, increased amounts of ATP have been detected in the airways of patients with IPF 190 , but currently there is no evidence linking the TRPV4/pannexin/ATP axis to IPF. SARS-CoV-2 may cause potentially fatal acute respiratory syndrome (ARDS) and pulmonary oedema in a subset of patients with COVID-19. As several TRPV4 antagonists that have already been proved safe in clinical trials can protect the alveolo-capillary barrier, these compounds could be protective in patients with COVID-19 at risk of lung oedema 179 . The afferent TRPV1-expressing pulmonary innervation has also been implicated in ARDS; drugs that silence these fibres (for example, resiniferatoxin) can also improve clinical outcome 297 . Deadly viruses can exploit the endo-lysosomal trafficking system of the host cells for penetration and replication. The integrity of this system depends on mucolipins (TRPMLs). Therefore, compounds that interfere with endo-lysosomal maturation and trafficking via TRPMLs may prevent the virus from entering the host cells 298 . In the brain, blood flow must meet the dynamically changing metabolic demands of active neuronal populations. The machinery of neuro-vascular coupling (NVC) senses the change in neuronal activity and redirects the blood flow accordingly. Most recently, TRPA1 expressed in capillary endothelium has emerged as a key player in this functional hyperaemic response 299 . In response to neuronal activity, TRPA1 initiates a rapid Ca 2+ signal, which is dependent on the endothelial pannexin-1 channel and purinergic P2X receptors 299 . This current is subsequently converted into an inward rectifying K + -mediated electric signal that guides blood flow by regulating the tone of the arteriole wall. Disturbed NVC is thought to be a major risk factor for cognitive impairment in diabetic patients 300 . In diabetes, methylglyoxal (and probably also other harmful by-products of glycolysis) bind to and activate TRPA1 (ref. 79 ). Therefore, one can speculate that pathogenic metabolic products may impair NVC in diabetic patients in a capillary endothelial TRPA1-dependent fashion. If so, TRPA1 antagonists could prevent cognitive decline by protecting the NVC in patients with diabetes in addition to protecting nerve fibres 88 and relieving pain during diabetic polyneuropathy 87 . This would be an important added benefit of TRPA1 antagonist therapy. TRPM8 is a cold-responsive receptor 7 , 8 that, in principle, may mediate coughing and bronchoconstriction caused by inhalation of cold air. The role of TRPM8 activation in cough is, however, controversial because menthol, a TRPM8 agonist, paradoxically inhibits citric acid-induced cough in both guinea pigs 191 and humans 192 , and causes a temporary decrease in capsaicin-evoked cough in healthy individuals 193 . Indeed, menthol inhibits sensory nerve activation and is used as an over-the-counter antitussive medication. Menthol was — and in many countries still is — often added to cigarettes to inhibit irritancy in the airways, which may promote nicotine addiction. Therefore, menthol is now banned from cigarettes in the USA. There is still a big question mark over whether the beneficial effects of menthol in the lung are mediated via activation of TRPM8. This is likely to be confounded by the lack of selective tools. For example, in addition to their effects on TRPM8, both menthol and icilin activate and then block TRPA1 (ref. 194 ) (Fig.  3c , d , g ). Several recently developed TRPM8 inhibitors should help to elucidate the role of this channel in respiratory disease biology and pathophysiology.

In 1969, a Drosophila mutant with defective light sensing was identified, which showed only a transient receptor potential (TRP) when exposed to continuous light instead of the expected sustained response 1 . This was later found to be caused by the lack of a functional copy of the gene coding for an ion channel, which was named trp 2 . However, the name ‘TRP’ channel is really a misnomer as the wild-type channel in fact causes a persistent (and not transient) current. This behaviour of TRPs is in contrast to many ion channels, which fully adapt when exposed to constant stimulation. TRPs are multifunctional signalling molecules, expressed in many tissues and cell types 3 , 4 . Most TRPs are polymodal channels, so-called coincidence detectors that are activated by both physical (temperature, voltage, pressure and tension) and chemical stimuli 4 . Beyond that, few generalizations can be made about TRP channels. Some TRPs function as non-selective cation channels in the plasma membrane; others regulate Ca 2+ release in intracellular organelles. The mammalian TRP channel superfamily has 28 members (27 in humans) 3 . On the basis of sequence homology, the superfamily is divided into six subfamilies: canonical (also known as short TRPs, TRPC1−7), vanilloid (also known as TRP channel subfamily V, TRPV1−6), melastatin (also known as TRP channel subfamily M, TRPM1−8), ankyrin (also known as TRP channel subfamily A, TRPA1), mucolipins (TRPML1−3), and polycystins (also known as polycystic kidney disease 2-like 1 protein (PKD2L1, also termed TRPP3) and polycystin-2 (TRPP2)) 3 , 4 . Because these subfamilies were created based on sequence homology and not function, members often have little in common. For example, TRPM2 is a redox sensor in macrophages 5 ; TRPM7 provides a major Mg 2+ uptake pathway in intestinal epithelial cells 6 ; and TRPM8 detects cold and menthol in sensory neurons 7 , 8 , but regulates epithelial growth in response to androgens in the prostate 9 . Despite the structural similarities shared by these proteins (Fig. 1 ; Box  1 ), there are enough differences to develop subtype-selective compounds. Fig. 1 Similarities and differences between the structures of TRP channels. Representative structures for each TRP channel subfamily, coloured to highlight common structural features. The S1−S4 (gold) and the S5 and S6 pore domains (cyan) are the only domains common to all subfamilies. The TRPA, TRPV, TRPM, TRPC and TRPN channels have TRP box helices (dark green). The TRPA, TRPC, TRPN and TRPV channels have amino (N)-terminal cytoplasmic ankyrin repeats (violet; the TRPA1 structure is missing about 11 repeats, as indicated by the question mark and violet shapes). The TRPC, TRPM and TRPN channels have a homologous pre-S1 helical linker (lilac) and C-terminal re-entrant loop and rib helix (brown), which is followed by a coiled coil in the TRPC and TRPM channels (blue; TRPA channels also have a carboxy (C)-terminal coiled coil). The melastatin homology regions (MHRs) of TRPM channels are shown in tan, and the NUDIX domain unique to TRPM2 is shown in red. The TRPML and TRPP channels have a homologous lumenal domain (dark grey). The pre-S1 linkers of the TRPV and TRPA channels and the C-terminal region of the TRPV channels (light grey) share no clear homology outside their respective subfamilies. The structures depicted are human TRPA1 (PDB ID 3J9P), human TRPC3 (PDB ID 6CUD), human TRPM2 (PDB ID 6MIX), Drosophila TRPN (NompC; PDB ID 5VKQ), human TRPP2 (PDB ID 5T4D), human TRPML1 (PDB ID 5JW5) and rabbit TRPV2 (PDB ID 6OO3). Representative structures for each TRP channel subfamily, coloured to highlight common structural features. The S1−S4 (gold) and the S5 and S6 pore domains (cyan) are the only domains common to all subfamilies. The TRPA, TRPV, TRPM, TRPC and TRPN channels have TRP box helices (dark green). The TRPA, TRPC, TRPN and TRPV channels have amino (N)-terminal cytoplasmic ankyrin repeats (violet; the TRPA1 structure is missing about 11 repeats, as indicated by the question mark and violet shapes). The TRPC, TRPM and TRPN channels have a homologous pre-S1 helical linker (lilac) and C-terminal re-entrant loop and rib helix (brown), which is followed by a coiled coil in the TRPC and TRPM channels (blue; TRPA channels also have a carboxy (C)-terminal coiled coil). The melastatin homology regions (MHRs) of TRPM channels are shown in tan, and the NUDIX domain unique to TRPM2 is shown in red. The TRPML and TRPP channels have a homologous lumenal domain (dark grey). The pre-S1 linkers of the TRPV and TRPA channels and the C-terminal region of the TRPV channels (light grey) share no clear homology outside their respective subfamilies. The structures depicted are human TRPA1 (PDB ID 3J9P), human TRPC3 (PDB ID 6CUD), human TRPM2 (PDB ID 6MIX), Drosophila TRPN (NompC; PDB ID 5VKQ), human TRPP2 (PDB ID 5T4D), human TRPML1 (PDB ID 5JW5) and rabbit TRPV2 (PDB ID 6OO3). Most TRPs have restricted expression patterns, but their varied tissue distribution means that the superfamily affects most cells, tissues and organs of the human body. Overall, the diverse physiological functions and regulatory mechanisms of TRPs affect how they are implicated in disease. These include both genetic and acquired channelopathies, as well as many disorders in which targeting one or more TRP channel could alleviate symptoms or provide therapeutic effects 4 . Most TRPs are subjects of intensive drug discovery and development efforts. In this Review, we summarize the crucial advances of the past decade in our understanding of the complex roles that TRPs have in the development and progression of human disease. Whereas our improved understanding of the structures of TRP channels will undoubtedly aid drug discovery (Box  1 ), the increasingly diverse physiological roles of TRPs pose a serious challenge to drug development. Indeed, it has proved difficult to obtain sufficient specificity for clinically useful intervention without unacceptable side effects. Although developing clinically useful TRP modulator drugs is challenging, the potential rewards are enormous given the pathogenic role of TRPs in chronic pain, neurology, oncology, dermatology, pulmonology, cardiology, urology and rare diseases. Effective drug discovery targeting individual TRP channels will need in-depth knowledge of the druggable structural sites and the activation and inactivation mechanisms. This, in turn, requires high-resolution structures in multiple states combined with functional studies. However, drug discovery strategies can also be conceptually enhanced by leveraging the rapidly multiplying structural information to take a bird’s-eye view of the whole family. With at least one structure available for each subfamily, we can now identify several recurring structural features. As predicted from their sequence homology to voltage-gated channels, all TRP channels have the S1−S4 transmembrane segments that form peripheral sensing domains (yellow in Fig.  1 ), whereas S5 and S6 tetramerize to create a central pore (cyan in Fig.  1 ). Other recurring structural features became apparent from sequence analyses, such as the ankyrin repeats in the N-terminal cytosolic domains of the TRPA, TRPC, TRPV and nonmammalian TRPN channels 293 (pink in Fig.  1 ). Moreover, the TRP box, a sequence motif first detected in the TRPC, TRPM and TRPV channels, forms a helix parallel to the membrane (green in Fig.  1 ). This helix, which is a link between the cytosolic and transmembrane domains, is actually found in all TRP subfamilies except TRPML and TRPP. Additional unanticipated homologous structural features are shared between the TRPC, TRPM and nonmammalian TRPN channels. In the Box  1 figure below, these shared features are viewed from above the membrane to better visualize their common structural fold, with the green TRP box helix included for reference. First, the approximately 150 N-terminal residues preceding the transmembrane domain (lilac) form a cytosolic helical platform and a re-entrant loop that penetrates the inner leaflet of the membrane near the S1−S4 domain. Second, the approximately 80–100 C-terminal residues following the TRP box helix (green) also form a re-entrant loop followed by a long helix parallel to the membrane plane (brown). This long helix, named the ‘rib helix’ in TRPM channels 294 , ‘connecting helix’ in TRPC channels 295 and ‘CH2’ in TRPN channels 296 , is then followed by a coiled coil in TRPC and TRPM channels (blue). Shared structural features between TRP channel subfamilies like the ones described here suggest shared regulatory mechanisms. Thus, information gained in individual subfamilies can and should be mined to advance our understanding of channels from other subfamilies with similar structural features.