REVIEWS

Breaking barriers to novel analgesic drug development

Ajay S. Yekkirala1–3, David P. Roberson1–3, Bruce P. Bean1 and Clifford J. Woolf1,2 Abstract | Acute and chronic pain complaints, although common, are generally poorly served by existing therapies. This unmet clinical need reflects a failure to develop novel classes of analgesics with superior efficacy, diminished adverse effects and a lower abuse liability than those currently available. Reasons for this include the heterogeneity of clinical pain conditions, the complexity and diversity of underlying pathophysiological mechanisms, and the unreliability of some preclinical pain models. However, recent advances in our understanding of the neurobiology of pain are beginning to offer opportunities for developing novel therapeutic strategies and revisiting existing targets, including modulating ion channels, enzymes and G-protein-coupled receptors.

Pain is the primary reason why people seek medical blockbuster model of one treatment for all pain condi- Analgesics 12 Pharmacological agents or care: more than 40% of the US population is affected tions is not tenable . The poor predictability of preclin- 1 ligands that produce analgesia. by chronic pain . In the United States alone in 2013, the ical ‘pain’ models can result in candidates being selected overall cost of treating certain chronic pain conditions that do not have activity in the conditions suffered by Tolerance amounted to more than US$130 billion2. Currently avail- patients13 (BOX 1). Conversely, difficulty in ensuring A state in which the drug no analgesics longer produces the same able — nonsteroidal anti-inflammatory drugs target engagement, lack of sensitivity of clinical trials effect and a higher dose is (NSAIDs), amine reuptake inhibitors, anti­epileptic and distortions induced by placebos increase the risk therefore needed. drugs and opioids — have varying but typically low that potentially effective compounds or targets may levels of analgesic efficacy, which is generally coupled be prematurely abandoned14,15. These issues have led Dependence with deleterious effects2. Indeed, opioids, which are the to most drug developmental efforts being devoted to An adaptive state that develops when a pharmacological agent most commonly used (with ~240 million prescriptions reformulations of existing validated analgesic classes — 3 is used repeatedly and leads to in 2014 in the United States) and often the most effec- opioids, NSAIDs, antiepileptic agents and amine reuptake withdrawal on cessation of the tive class of analgesics, produce tolerance, dependence inhibitors — despite their well-known limitations16. drug regimen. and constipation, and are associated with major abuse Given the prevalence of opioid diversion and over- 1Department of liabilities. Furthermore, the respiratory depression asso- dose, there is an urgent need to develop effective and Neurobiology, Harvard ciated with high doses has led to a catastrophic increase safe analgesics without a liability for abuse. Work from a Medical School, Boston in the number of drug overdose deaths in the United large number of laboratories has shed light on the mech- Children’s Hospital, 300 States3 (see the US Centers for Disease Control and anisms and pathways that mediate specific aspects of the Longwood Avenue, Boston, 8,10,11,17–19 Massachusetts 02115, USA. Prevention website). nociceptive nexus (FIG. 1), revealing the poten- 2FM Kirby Neurobiology Diverse pathological situations at different anato­ tial for more fine-tuned targeting of acute and chronic Center Boston Children’s mical sites can lead to pain. Causes of pain include pain in different clinical conditions. Completely new Hospital, 300 Longwood cancer, inflammation or tissue injury, as well as injury therapeutic strategies targeting ion channels, enzymes Avenue, Boston, 4–8 Massachusetts 02115, USA. or lesions of the nervous system . Diverse chronic and G-protein-coupled receptors (GPCRs), often with 3Blue Therapeutics, Harvard widespread pain syndromes may also occur owing to human genetic validation, are emerging. In this Review, Innovation Launch Lab, 114 abnormal amplification states in the central nervous sys- we present these emerging mechanisms and assess tar- Western Avenue, Allston, tem (CNS)9–11. All of these can lead, through abnormal gets and agents in preclinical and early clinical stages of Massachusetts 02134, USA. Correspondence to A.S.Y. activity in nociceptive systems, to pain in the absence development that show promise for the development and C.J.W. of a stimulus (spontaneous pain), exaggerated responses of innovative analgesics (TABLES 1,2). [email protected]; to noxious stimuli (hyperalgesia) and pain evoked by Clifford.Woolf@childrens. normally innocuous stimuli (allodynia). Neurobiology of pain harvard.edu The heterogeneity of clinical pain conditions and The mechanisms that sustain and drive chronic pain doi:10.1038/nrd.2017.87 the complexity and multiplicity of underlying patho- (FIG. 1) have been exhaustively reviewed elsewhere7,10,17,20–23. Published online 9 Jun 2017 Corrected online 23 Jun 2017 physiological mechanisms has made it difficult to iden- In physiological situations, pain is a necessary protective and 6 Oct 2017 tify tractable targets with broad involvement — the response that is initiated by high-threshold peripheral

NATURE REVIEWS | DRUG DISCOVERY VOLUME 16 | AUGUST 2017 | 545 ©2017 Mac millan Publishers Li mited, part of Spri nger Nature. All ri ghts reserved.

REVIEWS

Box 1 | The challenges of preclinical models of pain Another driver for spontaneous pain in the absence of a peripheral stimulus is the ectopic activity of primary 255,256 Preclinical rodent efficacy models are essential for analgesic development , but sensory neurons (FIG. 1). Several voltage-gated sodium their predictive validity has been questioned owing to several high-profile programmes channels (Nav1.3 (REFS 27,28), Nav1.7 and Nav1.8 in which rodent behavioural readouts predicted analgesic effects that were absent in (REFS 29–34)), potassium channels (KCNQ), calcium humans. For example, fatty acid amide hydrolase (FAAH) inhibitors were found to be antinociceptive in a range of animal models, but compounds such as PF‑04457845 channels (Cav2.2) and hyperpolarization-activated produced no analgesic effect in people with osteoarthritis despite lowering FAAH cyclic nucleotide-modulated channel (HCN) have been activity by >96%244. Similarly, NK1 (substance P) antagonists were shown to robustly implicated in the modulation of ectopic activity and reverse rodent nociceptive responses in the context of inflammation and nerve injury membrane excitability in sensory neurons, especially but failed to produce analgesia in subsequent clinical trials257. Nonetheless, many after peripheral nerve injury35–38. However, a direct rela- clinically used analgesics, such as nonsteroidal anti-inflammatory drugs (NSAIDs) and tionship between such ectopic activity of primary sen- opioids, produce antinociceptive effects in rodents256, albeit typically at much higher sory neurons with spontaneous pain is yet to be proven, doses than those used in patients. mainly due to inconsistent results and the paucity of reli- Using animal models of pain to detect putative analgesics faces several challenges. able preclinical models, although live-cell imaging with First, how do you measure pain, a conscious subjective report of an unpleasant sensory genetically encoded activity reporters should change this. experience, when you have no access to how an animal feels? Second, are the models true surrogates for the conditions or diseases that commonly produce pain in patients? One major consequence of peripheral inflamma- Third, how do you eliminate the confounders of bias, observer-induced changes and tion is the hyperfunction of nociceptive transduction lack of reproducibility? And last, it is possible that drugs that target human proteins ion channels on the peripheral terminals of sensory may not be active on their rodent homologues. neurons. These channels are responsible for converting Effectively measuring pain is perhaps the most difficult challenge, as we can only noxious stimuli into inward currents. The peripheral measure outcomes that may correlate with some aspect of pain, such as withdrawal sensitization of these transducers and ion channels from a stimulus or learned avoidance from a situation that may be painful. For involved in membrane excitability leads to a reduction reflexive measures of pain, typically a brief stimulus lasting for seconds is applied to in the threshold for activation and hyperexcitability of 5,6 a part of an animal’s body and a response measured . This clearly bears little nociceptor sensory neurons17. However, peripheral sen- correspondence to the clinical situation of ongoing spontaneous pain. Attempts have sitization alone does not fully account for the prolonged been made to develop outcome measures that may reflect the presence of discomfort, but these require more effort and validation to make them robust and presence of pain, dynamic tactile allodynia, temporal useful255. Some classes of analgesics, such as opiates, can reduce nociceptive reflexes summation of pain and the spread of pain hypersensi- 26 at high doses; however, this does not mean that such reflexes have predictive validity tivity to non-­injured tissue (secondary hyper­algesia) . for all classes of analgesics. These features of pain hypersensitivity are the conse- Some disease surrogates, such as pain in response to an acute noxious stimulus, quence of augmented sensory signalling in the CNS incision wound or traumatic injury to a nerve, are easy to model5,6. However, for chronic due to a prolonged increase in excitability and synap- widespread pain, such as fibromyalgia, osteoarthritis and low back pain, there are no tic efficacy of central nociceptive neurons and loss of rodent models that phenocopy the human disease precisely. inhibitory activity26. This phenomenon, called central Finally, a major difficulty with preclinical models is that the human observer may sensitization, is common to all types of clinical pain, introduce changes in the animal’s behaviour that may distort the outcome measure some of which are driven by activity in nociceptors as through induction of fear and anxiety253. Recently, even the gender of the investigator has been shown to have a significant impact on pain-related behaviour in mice258. Best a form of activity-dependent plasticity, and some by laboratory practices can eliminate bias through thorough blinding, and independent autonomous changes in the CNS. replication should be standard21,22. Several mechanisms contribute to central sensiti- In conclusion, although preclinical models are invaluable for the validation of targets zation; increased synaptic strength due to an upregula- and the measurement of drug action, they currently lack predictive power, and findings tion of presynaptic ion channels, increased presynaptic must be interpreted with caution. neuro­transmitter release or enhanced postsynaptic neuro­ transmitter receptor activity17,26. Disinhibition — in which inhibitory noradrenergic, opioid and GABAergic trans- sensory neurons to warn about the risk of harm by tis- mission is reduced through various synaptic changes — Hyperalgesia Enhanced nociceptive sue injury or infection. Indeed, the presence of pain and the loss of spinal interneurons also have a major role, 17,26 response to a noxious stimulus, acts as a homing beacon to alert the body to the pres- especially in neuropathic pain . In addition, injury to leading to greater discomfort ence of a noxious stimulus so that necessary correc- the nervous system may produce structural changes in than before. tive responses can be mounted. For tissue injury and nociceptive circuits that lead to increased pain sensitivity. infection, pain hypersensitivity accompanies the asso- Depending on the underlying cause of pain, many of the Allodynia Nociceptive response elicited ciated inflammation and persists for the duration of the above processes may also be acting in concert, further 17,24 7 even to previously non-noxious inflammatory response ; by encouraging the patient complicating treatment options . stimuli. to avoid use of the affected body part, pain helps heal- In addition to the above mechanisms, there are ing to occur. However, pain can also be maladaptive multiple local and distributed pain modulatory influ- Phenocopy When an organism shows and pathological when it arises as a result of dysfunc- ences. Endocannabinoids, for example, target cannab- phenotypic characteristics that tion of the nociceptive system, genetic factors (for inoid receptors in both GABAergic and glutamatergic reflect a different genotype example, paroxysmal extreme pain disorder caused by synapses, where they serve as retrograde transmitters from its own. gain‑of‑function mutations in Nav1.7 (REF. 25)), injury and block the transmission of pain signals39. In addi- to the nervous system that leads to neuropathic pain17,26 tion, endogenous opioid peptides such as endorphins, Neuropathic pain A condition leading to pain due and abnormal central amplification syndromes. In all enkephalins, dynorphin and endomorphins target the to damage or disease of of these cases, pain is no longer the symptom but the different opioid receptor subtypes (μ, κ and δ) to provide nervous system tissues. disease itself. analgesia in various contexts40–42.

546 | AUGUST 2017 | VOLUME 16 www.nature.com/nrd ©2017 Mac millan Publishers Li mited, part of Spri nger Nature. All ri ghts reserved. ©2017 Mac millan Publishers Li mited, part of Spri nger Nature. All ri ghts reserved.

REVIEWS

Peripheral mechanisms Central (dorsal horn) targets

Primary afferent Microglial cell Primary afferent neuron (central neuron (peripheral terminal) terminal) CGRP CB cannabinoids, 1 mGluRs, NK1 TRKA GPCR–ion • Pain GABA channel • Itch receptors interaction • Nociceptor sensitization AMPA or NMDA Glutamate receptors Opioid Opioid Selective • Cav Biased GPCR receptors receptors Nav and • TRP GPCR dimers Kv ligands • P2X signalling antagonists Dorsal horn neuron Analgesia with reduced side effects? Astrocyte

Figure 1 | Mechanisms relevant for nociception and analgesia. There are several targets in peripheral neurons and in the nociceptive synapse between glia, peripheral nervous system neurons and central nervousNature system Reviews neurons | Drug in Discovery the spinal dorsal horn that provide interventional strategies for the development of analgesics. The central terminals of nociceptor sensory neurons synapse on neurons in the dorsal horn of the spinal cord, which, after processing by local circuits, relay the information to the brain. The presynaptic nociceptor terminals deliver input generated by noxious stimuli in the periphery, inflammation and peripheral nerve injury to the postsynaptic neurons. Several of the transmission and modulation mechanisms involved represent potential analgesic targets; excitatory transmissions could be reduced or inhibitory transmissions could be augmented. Some of the key players — opioids, cannabinoids, NK1 and other G-protein-coupled receptors (GPCRs) — are discussed throughout the main text, and others include ion channels such as NMDA (N-methyl-d-aspartate), AMPA (α-amino‑3‑hydroxy‑5‑methyl‑4‑isoxazole propionic acid) and GABA (γ-aminobutyric acid). Another key player is tyrosine kinase receptor A (TRKA; also known as NTRK1), which is targeted by nerve growth factor (NGF); NGF–TRKA-signalling blockade has been shown to produce potent analgesia. However, in many cases, a lack of specificity for nociceptive pathways makes on‑target undesirable effects problematic. Cav,

voltage-dependent calcium channel; CB1, cannabinoid receptor 1; CGRP, calcitonin gene-related peptide; Kv, potassium channel; mGluR, metabotropic glutamate receptor; Nav, sodium channel; P2X, purinoceptor; TRP, transient receptor potential cation channel.

An often overlooked and underappreciated aspect Abuse-deterrent opioids of pain is that chronic pain can enhance negative emo- Several companies have focused on modifying exist- tional states along with comorbidities such as anxiety and ing opioids to develop abuse-deterrent formulations depression43,44. Analogously, studies have also shown that (ADFs)46,47. For example, extended release oxycodone analgesia can be emotionally rewarding. The brain meso- (OxyContin; Purdue Pharma) and oxymorphone hydro- corticolimbic region encodes reward and motivation, chloride (Opana ER; Endo Pharmaceuticals) are com- and the coupling of nociception to this region mediates monly prescribed on the basis of the assumption that pain aversion learning and memory44. The motivation to abuse only occurs with drugs that have rapid kinetics, obtain pain relief can be very strong, and recent studies which increases euphoria. However, this may be a miscon- on the impact of chronic pain on reward and motiva- ception. The clinical data are scant and mostly inconclu- tion circuits suggest that neurochemical changes in these sive, but some data show that the incidence of abuse may Central sensitization brain regions may serve as objective markers for pain44. actually increase with ADF opioids46,47. Indeed, although A condition of the nervous formulated differently, the active agents are still euphoric system in which neurons in the Targeting the opioid system opioids, and the drugs retain the potential side effects of central nervous system are in a state of prolonged increase in Exogenous opioid ligands isolated from opium poppy the class, including the potential to overdose on high dose 47 excitability and synaptic seeds have been used as potent analgesics for more than exposure . In addition, although ADF OxyContin has efficacy, coupled with the loss 3,000 years41, and opioids such as hydrocodone and reduced the number of prescriptions for and abuse of of inhibitory activity. morphine still remain the analgesics of choice for many OxyContin48,49, many abusers have either learnt to tam- Analgesia physicians in the clinic. However, although targeting opi- per with the abuse deterrence or have moved to other 50,51 A lack of, or insensibility to, oid receptors can lead to efficacious inhibition of pain, prescription opioids, such as fentanyl or heroin . pain. especially in an acute injury or palliative setting, opioid receptors also induce deleterious effects, such as toler- Peripherally restricted receptor ligands Nociception ance, physical dependence and an addictive potential that All major opioid analgesics currently on the market are Sensory neuronal responses to noxious or damaging stimuli has led to a crisis of prescription and recreational opioid μ-opioid receptor agonists. Early efforts were made to 45 that attribute the sensation abuse . Numerous approaches are therefore being tested develop κ- or δ-selective ligands because some of the of pain. to mitigate the deleterious effects of opioid analgesics. initial selective compounds showed a lesser propensity

NATURE REVIEWS | DRUG DISCOVERY VOLUME 16 | AUGUST 2017 | 547 ©2017 Mac millan Publishers Li mited, part of Spri nger Nature. All ri ghts reserved. ©2017 Mac millan Publishers Li mited, part of Spri nger Nature. All ri ghts reserved.

REVIEWS

Table 1 | GPCR ligands at various stages of preclinical and clinical development Target/category Name Mechanism Preclinical data Clinical status Refs Abuse-deterrent MS‑Contin, Extended release opioids Similar to opioid pharmacophores Clinically used. 46,47,50,51 formulations of OxyContin, etc. Controversial data suggest opioids (Purdue Pharma) increased misuse of opioids. More studies are necessary Peripheral GPCR CR845 (Cara Peripherally restricted Similar to other κ-agonists in Currently under clinical See CR845 ligand Therapeutics) κ-opioid agonist rodent pain models (http://files. development: phase IIa, phase IIa (i.v. and oral) shareholder.com/downloads/ oral; phase III, i.v. results press AMDA-2C4IM7/4486893744x0x release and 707472/8DFBB204-4BC0-4A66- CR845 phase ACF7-1AC8E9C31FA5/707472. III patient pdf). Poor CNS penetration (see recruitment Cara Therapeutics — Kappa press release Opioid Receptor Agonists) Heteromer GPCR NNTA (Blue Selective μ–κ-opioid Analgesic, no dependence, no Preclinical IND 66 ligands Therapeutics) heteromer agonist (initial CPP development indication of greater potency and receptor expression in the spinal cord) MDAN‑21 Bivalent μ-agonist–δ- Analgesic, no tolerance or CPP. No current information on 59,60,62,64 antagonist ligand Extremely potent analgesic in clinical development several rodent models MMG‑22 Bivalent μ-agonist– Potent analgesia in inflammatory ND 64 mGluR5 antagonist ligand and neuropathic pain models MCC‑22 Bivalent μ-agonist–CCR5 Potent analgesia in inflammatory ND 65 antagonist ligand and neuropathic pain models Biased GPCR RB‑64 Biased KOR ligand Analgesic, CPA ND 74 ligands TRV130, TRV250, Biased μ-opioid agonists Varying levels of analgesic Phase I or II trials 86,265 TRV734 (all efficacy, no respiratory depression completed, phase III Trevena), PZM21 and reduced abuse potential and initiated for TRV130 other side effects Truncated GPCRs IBNtxA 6TM μ-opioid truncated Analgesic, no CPP ND 70–72,266 receptor (site of expression and extent of expression still to be determined)

AT2R Mycolactone AT2R blockers (mechanism Analgesic, neuralgia indication Under development for 90,92,95 (Spinifex Pharma), still to be determined) post-herpetic neuralgia. EMA401 Phase II data showed increased efficacy when compared with placebo C5AR DF2593A (Dompé C5AR allosteric antagonist Allosteric C5AR inhibitor; effective ND 267 Pharma) in rodent models of inflammatory and neuropathic pain

Cannabinoid APD371(Arena CB1 and CB2 agonists Antinociception, attenuate APD371 in phase II; 87,88, receptors Pharmaceuticals), (spinal and supraspinal morphine tolerance KHK6188 and LY2828360 268,269 LY2828360 (Eli mechanisms) discontinued in Phase II; Lilly), S-777469 S-777469 diverted to atopic (Shionogi), dermatitis studies in the KHK6188 (Kyowa clinic Hakko Kirin) GPR55 ML‑193 GPR55 antagonist Analgesia; agonists are ND 270 pronociceptive α2a adrenergic Clonidine, α2a adrenergic agonists Some efficacy in neuropathic pain Meta-analysis of 28 clinical 97–100, receptor tizanidine, models in rodents. trials suggests some clinical 102,103 dexmedetomidine efficacy Chemokine Morphine + Opioid + CCR antagonist Reduced OIH and opioid ND 108,119 receptors AMD3100 tolerance in rodents, reduced (Genzyme) neuropathic pain AZD2423 CCR2 antagonist Potent analgesia in neuropathic Failed clinical trial 120 (AstraZeneca) pain and inflammatory pain models RAP‑103 (Rapid Mixed CCR2–CCR5 Attenuates neuropathic pain ND 121 Pharmaceuticals) antagonist

6TM, six transmembrane; AT2R, angiotensin type 2 receptor; C5AR, C5a anaphylatoxin chemotactic receptor 1; CB1, cannabinoid receptor 1; CB2, cannabinoid receptor 2; CCR, C-C chemokine receptor; CNS, central nervous system; CPA, conditioned place aversion; CPP, conditioned place preference; GPCR, G-protein-coupled receptor; GPR55, G-protein-coupled receptor 55; IND, investigational new drug; i.v., intravenous; KOR, κ-opioid receptor; mGluR5, metabotropic glutamate receptor 5; ND, no data on clinical development; OIH, opioid-induced hyperalgesia.

548 | AUGUST 2017 | VOLUME 16 www.nature.com/nrd ©2017 Mac millan Publishers Li mited, part of Spri nger Nature. All ri ghts reserved. ©2017 Mac millan Publishers Li mited, part of Spri nger Nature. All ri ghts reserved.

REVIEWS

Table 2 | Developmental status of ligands targeting ion channels involved in nociception Target/ Name Mechanism Preclinical data Clinical status Refs category TRPV1 Currently in trials: Agonists as patch Potent analgesia in Clinical trials failed owing 131,271–274 JNJ‑38893777 or injectable; inflammatory and to hyperthermia and other (Janssen), AZD1386 antagonists for pain neuropathic pain. Especially side effects. New ligands (AstraZeneca), relief beneficial for thermal currently under clinical NEO6860 (NeoMed). nociceptive component development do not have Failed: AMG517 the above side effects (Amgen), AZD1386 (AstraZeneca), ABT‑102 (Abbott), MK‑2295 (Merck & Co.) Cavs Ziconotide, Blockers for various Efficacy for currently Z944 and CNV2197944 in 174; see Z944 phase Ib gabapentin, pain indications approved ligands in models phase II results press release and pregabalin. of neuropathic pain, but CNV2197944 phase II Currently in trials: only moderate clinical trial announcement Z944 (Zalicus), efficacy (BOX 2). New ligands press release CNV2197944 have good potency in (Convergence), inflammatory, neuropathic TROX‑1 and visceral pain models (Grünenthal). Failed: Z160 (Zalicus) Nav1.7 TV‑45070 (Xenon Blockers for pain. Knockouts are insensitive to TV‑45070 failed phase II; 140,152,154; see and Teva); Human genetic data nociceptive stimuli; loss of PF‑05089771, GCD‑0276 BioTuesdays — Xenon GCD‑0276, show that mutations inflammatory pain and GCD‑0310 in phase I still committed GCD‑0310 (Xenon confer congenital to TV‑45070 for and Genentech); pain insensitivity. neuropathic pain and PF‑05089771 (Pfizer) Controversially, some Xenon and Genentech argue that increased collaboration press endogenous opioids release in knockouts are producing the insensitivity Nav1.8 PF‑04531083 (Pfizer) Target is widely PF‑04531083 was effective PF‑04531083 failed in a 154 expressed in in producing analgesia clinical trial of dental pain nociceptors, and in neuropathic and blockers are proposed inflammatory pain models to be analgesic Cavs, voltage-gated calcium channels; Nav, voltage-gated sodium channel; TRPV1, transient receptor potential cation channel subfamily V member 1.

for physical dependence and abuse40–42. Indeed, many has been developed that, in contrast to the conven- κ-agonists produce mild or severe aversive behaviour. tional opioid fentanyl, binds specifically to periph- However, such efforts have failed owing to ineffective eral μ-opioid receptors in acidified peripheral tissues analgesia (δ) or severe psychotomimetic and dysphoric to provide antinociception without the side effects effects (κ). For instance, κ-ligands such as enadoline and of respiratory depression, sedation, constipation or butorphanol were developed to have less abuse potential, addiction potential55. but they are still rewarding, and also induce fatigue, seda- These early preclinical and clinical data are prom- tion, confusion, anxiety, visual and auditory distortions ising, but questions still remain as to the efficacy of and depersonalization52,53. peripherally restricted ligands in reducing most forms As many of the side effects of κ-agonists are generated of chronic pain, especially those sustained by central in the CNS, Cara Therapeutics is currently developing a sensitization, as described above. peripherally restricted κ-agonist, CR845, to be adminis- tered either intravenously or orally. Intravenous CR845 Heteromers, bivalent ligands and isoforms has been shown to successfully inhibit postoperative Another approach to eliminate μ-opioid-associated side Depersonalization abdominal pain in phase II trials54, and an oral formula- effects is based on burgeoning evidence that GPCRs, A state in which an individual’s thoughts, feelings and tion is currently being tested in people with chronic knee including opioid receptors, form both oligomers and 56–58 emotions seem to not belong pain in a phase II trial. higher-order complexes (FIG. 2). Homomers (com- to them. A recent study has reported an approach for tar- plexes of similar receptors) and heteromers (complexes geting peripheral μ-opioid receptors without caus- of different receptors) provide exciting potential targets Antinociception ing narcotic side effects. Painful conditions are often for pharmacological intervention. Inhibition of sensory neuronal response to noxious stimuli associated with localized tissue acidification. A fluor- Several bivalent ligands have been generated that that leads to reduction of pain inated fentanyl analogue, NFEPP ((±)-N-(3‑fluoro‑1‑ contain distinct pharmacophores for two recep- sensation. phenethylpiperidin‑4‑yl)-N-phenylpropionamide), tors tethered together with a carbon atom linker of

NATURE REVIEWS | DRUG DISCOVERY VOLUME 16 | AUGUST 2017 | 549 ©2017 Mac millan Publishers Li mited, part of Spri nger Nature. All ri ghts reserved. ©2017 Mac millan Publishers Li mited, part of Spri nger Nature. All ri ghts reserved.

REVIEWS

varying lengths. These ligands demonstrate some prom- ising pharmaco­logical profiles (TABLE 1). For example, Biased GPCR GPCR GPCR–ion 59–62 signalling dimerization channel MDAN‑21 (μ-agonist–δ-antagonist) , a series of interaction • TRV130 • NNTA μ-agonist–cannabinoid receptor 1 (CB1) antagonist • TRV250 • MDAN-21 • Bradykinin ligands63, MMG22 (μ-agonist–metabotropic gluta- • TRV734 • MMG22 • Prostanoids mate receptor 5 (mGluR5) antagonist)64 and MCC22 • RB-64 • MCC22 • Histamine • Chemokines (μ-agonist–C-C chemokine receptor type 5 (CCR5) antagonist)65 have demonstrated potent antinocice- ptive activity in various preclinical rodent models of pain and are also devoid of tolerance or place prefer- G-protein β-arrestin ence, a measure of abuse liability. However, these mol- signalling signalling Novel Nociceptor pharmacology sensitization ecules are rather large, raising questions about their bioavailability. Analgesia with There is therefore interest in the development of reduced side effects small-molecule ligands for heteromeric opioid recep- tors. N-Naphthoyl-β-naltrexamine (NNTA), a selec- tive agonist for μ–κ-opioid receptor heteromers, is an example66. μ–κ heteromeric opioid receptors have been found to form heteromers in the rat spinal cord, and Dorsal root there seems to be more of these receptors in female rats ganglion than in male rats67. NNTA targets μ–κ heteromers at sub-­ picomolar concentrations and is a potent antinociceptive Spinal cord Primary afferent agent that does not seem to result in tolerance, physical cross-section neuron dependence or drug-seeking behaviour in rodents66. (peripheral terminals) NNTA is currently being investigated for safety by Blue Therapeutics through investigational new drug (IND)- enabling studies, and the company has plans for human clinical trials on successful filing of the IND package. The newly described buprenorphine small-molecule analogue BU08028 — a bifunctional ligand for μ and nociceptin opioid receptors — is another potential drug candidate that has displayed potent antinociception without concomitant respiratory depression or physical dependence in monkeys68. A different, but complementary, approach has focused on targeting different isoforms of the μ-opioid receptor (TABLE 1). The opioid receptor mu 1 (Oprm1) gene has two independent promoters and can thus form several splice variants, including some unex- pected six-transmembrane (6TM) domain receptors • TV-45070 • JNJ-38893777 • Retigabine that are structurally quite different from 7TM domain • GCD-0276 • NEO6860 • Flupirtine GPCRs69,70. For instance, the splice variant MOR1C • Lidocaine • AZD1386 • 4-Aminopyridine • Z944 results in a functionally active 6TM receptor, and the nal- Selective Nav • CNV2197944 Potassium trexone derivative, iodobenzoylnaltrexamide (IBNtxA), blockers channel that targets this receptor produces antinociception in TRP and Cav openers antagonists mice without respiratory depression or drug-seeking behaviour in conditioned place preference studies71,72. This suggests that the carboxy‑terminal transmem- Nature Reviews | Drug Discovery brane domain of the μ-opioid receptor may not be nec- Biased ligands essary for the docking of certain ligands. Similarly, the GPCRs were initially considered to signal only through modified C terminus of the functional splice variant the specific G proteins that they couple with, but the MOR1D mediates its dimerization with gastrin-releas- recent discovery of ligand bias has enhanced our under- ing peptide receptor (GRPR) to generate opioid-induced standing of GPCR signal transduction and provided new itch in mice, providing further evidence that the MOR GPCR pharmacology74. GPCR coupling with β‑arrestins C terminus may have a key role in producing adverse can be promoted or avoided by ligands that restrict the Ligand bias effects of opioids73. Several questions remain, however, receptor to particular conformations, and can thus lead Occurs when a ligand shows as to how these truncated receptors signal, the extent to several distinct pharmacological outcomes from the selectivity or preference to a and level of tissue expression, and why targeting them same receptor75–82 (FIG. 2). particular signal transduction mechanism for a target alters μ-opioid side effects. In addition, further research Several biased agonists have been developed target- receptor. Also called is warranted to determine the similarity between human ing opioid receptors that produce reduced opioid side ‘functional selectivity’. and murine opioid receptor isoforms. effects with similar or better antinociception, at least

550 | AUGUST 2017 | VOLUME 16 www.nature.com/nrd ©2017 Mac millan Publishers Li mited, part of Spri nger Nature. All ri ghts reserved. ©2017 Mac millan Publishers Li mited, part of Spri nger Nature. All ri ghts reserved.

REVIEWS

▶ Figure 2 | Peripheral analgesic drug discovery targets. The peripheral terminals of Pharmaceuticals has published positive phase Ib clinical primary afferent nociceptor neurons express G-protein-coupled receptors (GPCRs) and trial results on its selective CB2 agonist, APD371, and various ion channels that mediate the transduction of different sensory modalities, some is planning phase II trials (see Arena Pharmaceuticals of which have specific coupling and signalling features that offer opportunities for novel press release on APD371). Several other candidates87 drug classes. GPCRs can utilize both G proteins and ‑arrestins as secondary messengers, β (TABLE 1) and biased ligands can engage particular signalling pathways to obtain increased are also being evaluated in either phase I or II efficacy and reduced adverse effects. GPCRs can also form complexes with other trials (ClinicalTrials.gov identifiers: NCT01319929, receptors, leading to the formation of homomers or heteromers. Furthermore, they NCT00697710, NCT01544296 and NCT01789476), the functionally interact with ion channels such as transient receptor potential cation results of which are awaited. channel subfamily V member 1 (TRPV1), TRPA1 and G-protein-activated inward rectifier potassium channels (GIRKs). Anti-nerve growth factor (NGF) antibodies have also been Angiotensin type 2 receptor developed with the intent of alleviating certain pain modalities by blocking the Another intriguing GPCR, angiotensin type 2 recep- activation of tyrosine kinase receptor A (TRKA). In addition, inhibitors for sodium tor (AT2R), belongs to the renin–angiotensin system, channels (Nav) and calcium channels (TRP and voltage-gated calcium channels (VGCCs)), which is now recognized as contributing to neuropathic and potassium channel openers have been developed with intriguing pharmacology. pain in preclinical rodent models90,91. Whereas AT R Ligands for some of these targets that are currently in analgesic clinical trials are shown. 1 has a major role in hypertension, AT R is expressed in NNTA, N-naphthoyl-β-naltrexamine 2 human sensory neurons and, when activated, sensitizes transient receptor potential cation channel subfamily V in preclinical animal models74,83 (TABLE 1). A prominent member 1 (TRPV1) ion channels that mediate ther- 92,93 example is oliceridine (TRV130; Trevena), a μ-opioid mal pain . Although the mechanism by which AT2R receptor biased agonist that activates G-protein signal- inhibitors reduce neuropathic pain remains unknown, ling, but not β‑arrestin. Clinical trials with TRV130 have the inhibition of p38 and extracellular-signal-regulated been largely positive, showing that it is well tolerated and kinase–mitogen-activated protein kinase (ERK–MAPK) has comparable analgesic efficacy to morphine but with pathways has been implicated90,94. a faster onset84,85. TRV130 also produces significantly Spinifex Pharmaceuticals has developed a highly

reduced respiratory depression, nausea and vomiting selective AT2R inhibitor, EMA401, that inhibits in patients; however, this seems to be dose-dependent84. capsaicin-­induced calcium fluxes in human and rat dor- The fact that respiratory depression is not completely sal root ganglia (DRGs) and produces dose-­dependent eliminated with this approach is concerning, as the abuse anti­nociception in mouse neuropathic pain models, 90 (TABLE 1) liability of this compound has not yet been determined effects that are abolished in AT2R-null mice . in humans. Phase I and II clinical trials have shown that EMA401

The recent exciting discovery of PZM21 — a Gi-biased is well tolerated and produces significant analgesia μ-opioid agonist — clearly illustrates that differential with daily dosing in people suffering from post-herpetic pharmacology can be achieved by selectively activating neuralgia92. However, by contrast, a mycobacterial

specific signalling messengers through the same recep- polyketide mycolactone has been shown to activate AT2R tor86. The compound is the product of extensive in silico-­ to induce analgesia in mice with Buruli ulcer95. The fact assisted screening of >3 million ligands and is devoid of that both activators and antagonists of the same recep- the common opioid side effects of constipation, respira- tor produce analgesia complicates the delineation of the

tory depression and drug-seeking behaviours. However, mechanism of action of AT2R in pain. some questions remain as to the analgesic efficacy of the molecule, as it was found to be four times less potent α2 adrenergic receptors than morphine in rodent studies73. Studies of the role of the α2 adrenergic system in pain are conflicting. Pharmacological studies and studies in Non-opioid GPCRs null mice show that α2 receptors seem to play a part Outside the opioid receptor system, a few other receptors in diverse painful syndromes by inhibiting the presyn- have been found to be directly involved in pain path- aptic release of neuropeptides in the spinal cord and act- ways, including cannabinoid receptors, angiotensin ing on microglia96–100. However, pontine α2 adrenergic type 2 receptor and α2 adrenergic receptors. activation increases pain by attenuating descending inhi- bition101. A meta-analysis of 28 clinical trials found that Cannabinoid receptors although intraoperative administration of the α2 adren-

The cannabinoid receptors, CB1 and CB2, have been ergic receptor agonist dexmedetomidine (DEX) reduced intensively investigated as targets for potential pain post­operative pain and opioid consumption, and lowered 87 Psychotomimetic effects therapeutics . Whereas CB1 receptor agonists pro- the risk of opioid-related adverse effects, some patients A state of psychosis of the duce severe psychotomimetic effects and have a large suffered from intraoperative bradycardia, and the post- mind leading to delusions, hallucinations, and so on abuse potential, selective CB2 agonists produce analge- operative administration of DEX did not have signifi- 102,103 that are precipitated by sia in preclinical models without any major CNS side cant effects on pain management . Recro Pharma 87,88 a pharmacological agent effects . CB2 receptors were initially thought to be is developing an intranasal DEX formulation (DEX-IN) or ligand. peripherally restricted targets, but they are expressed that is administered at one-eighth to one-tenth of the in the CNS (FIGS 1,2), especially by spinal microglia and intravenous recommended dose, which could reduce Post-herpetic neuralgia Pain caused by nerve damage interneurons. Many pharmaceutical companies have adverse effects. In phase II trials, DEX‑IN produced due to infection with varicella developed selective CB2 agonists, but initial clinical tri- positive efficacy results for treatment of postoperative zoster virus. als failed owing to lack of efficacy87,89. However, Arena (bunionectomy) pain, but likewise generated adverse

NATURE REVIEWS | DRUG DISCOVERY VOLUME 16 | AUGUST 2017 | 551 ©2017 Mac millan Publishers Li mited, part of Spri nger Nature. All ri ghts reserved. ©2017 Mac millan Publishers Li mited, part of Spri nger Nature. All ri ghts reserved.

REVIEWS

effects that are characteristic of α2 adrenergic agonists, in other neuronal types, whereas Nav1.7 sodium chan- including reduced blood pressure and bradycardia. Plans nels are expressed in a wider variety of peripheral sen- for phase III trials of DEX‑IN for postoperative pain are sory and sympathetic neurons but have little expression pending US Food and Drug Administration (FDA) input in the CNS123,124. (see DEX‑IN phase II efficacy press release and DEX‑IN Although current strategies have focused on channels phase II side effects press release). in primary sensory neurons, there is potential to target ion channels in higher-order neurons. In this case, it would be Chemokine receptors difficult or even impossible to confine drug effects only Pro-inflammatory chemokines contribute to chronic to nociceptive pathways. However, this is not necessarily inflammatory pain104. Chemokines modulate nocicep- an insurmountable problem. Many existing ion channel-­ tors through the activation and sensitization of sodium based therapies for other conditions are also targeted to and TRPV1 channels and also communicate between channels with wide expression in the nervous system, immune, neuronal and glial cells105–107. The injection of including anti-epileptic drugs targeted broadly to sodium certain chemokines leads to pain106,107, whereas chemo­ channels and anxiolytic drugs targeted to γ-aminobutyric 108 16 kine antagonists produce antinociception . However, acid type A (GABAA) channels . In general, our knowl- given the numbers of these signalling molecules and edge of the organization of neurons and neuronal circuits their importance in immune function109, chemokines is too primitive to understand how such drugs targeted to are challenging therapeutic targets. widely expressed ion channels can achieve clinical efficacy There seems to be some crosstalk between opioid without debilitating side effects. As we gain more detailed and chemokine pathways — especially in the con- knowledge about the patterns of expression, regulation text of opioid tolerance, opioid withdrawal-induced and function of various channels in higher-order pain sig- hyperalgesia and opioid-induced hyperalgesia. This nalling neurons, it may be possible to target the activity of crosstalk may reflect hetero-desensitization and hetero-­ these neurons in a rational manner. oligomerization between opioid and chemokine recep- tors110–113. Co‑administration of morphine with C-X-C TRPV1 channels motif chemokine 12 (CXCL12; also known as SDF1), The molecular cloning of TRPV1 channels and their C-X3‑C motif chemokine 1 (CX3CL1; also known as identification in 1997 as a receptor in primary sensory fractalkine) or C-C motif chemokine 5 (CCL5) leads to neurons mediating thermal pain125 (FIG. 1) led to an ava- reduced opioid analgesia in rodents114–118. Conversely, lanche of studies implicating TRPV1 activation in various administration of antibodies to CCL2 or a CXCR4 inhib- preclinical pain models. Drug development programmes itor (plerixafor (AMD3100; Mozobil; Genzyme)) leads targeting TRPV1 channels followed quickly, with patents to reduced opioid-induced hyperalgesia and tolerance filed for more than 1,000 compounds by approximately 50 in rodents115,119. Bivalent ligands for μ‑CCR5 hetero-­ companies126. So far, the results have been disappointing. oligomers show potent antinociception without any In general, the efficacy of TRPV1 antagonists in clinically tolerance, dependence or abuse potential in mice65. relevant pain conditions in humans has been modest, However, clinical trials with chemokine antagonists and there are two problematic adverse on‑target effects have not been successful. AZD2423 (AstraZeneca), a seen with many of the compounds: reduced sensitivity to selective CCR2 antagonist, effectively reduced neuro­ noxious heat, potentially leading to accidental burns, and pathic pain in rodent models but did not produce an increase in body temperature, probably reflecting the analgesia in people with post-herpetic neuralgia120. participation of TRPV1 channels in normal temperature Species differences or poor CNS activity may have led regulation126–128. These issues, along with modest efficacy, to this failure, and dual-targeting antagonists such as have resulted in the discontinuation of most programmes. RAP‑103 (Rapid Pharmaceuticals; a CCR2–CCR5 dual However, recent work suggests that it may be possi- antagonist) are under development with the aim of ble to develop compounds that retain anti-nociceptive achieving a better efficacy profile121. activity without adverse effects. This would require agents that can inhibit the activation of TRPV1 chan- Ion channels as drug targets nels by endo­genous ligands that mediate pain but that Ion channels underlie all electrical activity by neurons. do not inhibit the activation of TRPV1 channels in In principle, targeting ion channels to disrupt pain sig- response to increases in temperature or low pH (which nalling could be done at all stages, from inhibiting the has been suggested to mediate the control of body tem- firing of primary nociceptors to inhibiting second- or perature129). Such ‘modality-specific antagonists’ are higher-order neurons in the spinal cord and brain. The based on evidence that vanilloid ligands such as capsa- most obvious way of selectively targeting pain signalling icin (and perhaps endogenous agents mediating pain) is to target ion channels that have preferential expression activate TRPV1 in a different way from temperature or in primary nociceptors. This is the strategy embodied in pH, and that these modalities of activation can be differ- drugs targeted to TRPV1 channels, which are strongly entially inhibited128–130. NEO6860 (Neomed) is one such expressed in primary nociceptors and have little or no modality-specific agent131 and is currently in phase II tri- Withdrawal expression in most other neuronal types122. Several other als for osteoarthritis. New technology that can provide Symptoms such as anxiety and shaking that develop on potential drug targets, including Nav1.8 and Nav1.9 high-resolution structures of ligand-bound TRPV1 chan- 132,133 cessation of a drug that has voltage-dependent sodium channels, are also highly nels using cryo-electron microscopy should greatly been used repeatedly. expressed in primary nociceptors with little expression facilitate further development of this approach128.

552 | AUGUST 2017 | VOLUME 16 www.nature.com/nrd ©2017 Mac millan Publishers Li mited, part of Spri nger Nature. All ri ghts reserved. ©2017 Mac millan Publishers Li mited, part of Spri nger Nature. All ri ghts reserved.

REVIEWS

Sodium channels In addition, peptide inhibitors of Nav1.7 channels Voltage-dependent sodium channels underlie electri- have been developed by cleverly modifying naturally cal signalling in all neurons. There are nine distinct occurring peptides from spider venom. This approach types of sodium channels encoded by different genes, exploits amino-acid differences among different sodium of which three (Nav1.1, Nav1.2 and Nav1.6) are widely channel types in extracellular loops of the channel pro- expressed in both the CNS and peripheral nervous sys- teins and has yielded peptide inhibitors with very high tem (FIG. 2) and three more (Nav1.7, Nav1.8 and Nav1.9) affinity for blocking Nav1.7 channels and a high degree of are expressed mainly in subsets of peripheral neurons. subtype selectivity150. However, the delivery of these large Currently used sodium channel blockers, including lido- peptide compounds may present a substantial challenge caine, amitriptyline and mexiletine, display relatively for clinical translation. A potential problem with the clin- little selectivity among different sodium channels, and ical use of Nav1.7 blockers is that although the channels their therapeutic window is limited by unwanted side are minimally expressed in the CNS, they are expressed effects, mainly from actions on the CNS. This has stimu- in primary olfactory neurons151, meaning that a potent lated efforts to develop compounds that are selective for Nav1.7 blocker may well produce anosmia, a side effect the sodium channels with little expression in the CNS. that has been found intolerable by a substantial fraction of patients in clinical trials of drugs with other targets. Nav1.7. Nav1.7 channels are expressed in nociceptors Another potential issue relates to the recent finding and sympathetic neurons, with only low expression in that sensitivity to pain can be restored by naloxone in the CNS123,134. They emerged as prime candidates for both Nav1.7‑null mice and in a human lacking Nav1.7 novel analgesics following the identification of rare (REF. 152). This raises the possibility that the loss of pain human mutations in which loss of function in Nav1.7 sensation in null mutants is more complicated than sim- channels resulted in congenital insensitivity to pain135,136. ple functional removal of Nav1.7 channels and may result In addition, a number of gain‑of‑function mutations in from upregulation of opioid pathways inhibiting pain. Nav1.7 have been identified, in which patients experi- However, it is possible that such upregulation reflects ence pain, sometimes severe, either spontaneously or complex compensatory events during development that triggered by normally innocuous stimuli124. Similarly, would not be mimicked by acute inhibition. mouse knockout models of Nav1.7 show nearly complete insensitivity to nociceptive stimuli and loss of inflamma- Nav1.8. Nav1.8 channels are tetrodotoxin (TTX)- tory pain, whereas the mice respond normally to various resistant, voltage-gated sodium channels that are non-painful stimuli137–139. expressed in small-diameter primary nociceptive neu- Nav1.7 is currently being actively pursued by the rons, with little or no expression in other neuronal pharmaceutical industry. Potent small-molecule inhib- types. In principle, their selective expression and domi- itors of Nav1.7 channels have been found140–142, most nant contribution to the overall sodium current in small-­ notably a class of sulfonamide compounds that can be diameter DRG neurons (at least in the soma) make made highly selective for Nav1.7 channels over other Nav1.8 channels attractive as drug targets. Moderately sodium channels, and even selective for human Nav1.7 selective Nav1.8 inhibitors have been developed and channels over those from other species143. Clinical trials have exhibited anti-nociceptive properties in rodent of several of these compounds are underway (TABLE 2), models30,153. There have been efforts to develop Nav1.8 including the Icagen/Pfizer compound PF‑05089771, inhibitors for clinical use in humans140. However, the which completed phase II trials for wisdom tooth only reported clinical test of a Nav1.8‑selective inhibitor, removal and primary erythromelalgia144, and two com- the Pfizer compound PF‑04531083 (REF. 154), resulted pounds (GDC‑276 and GDC‑0310) developed by Xenon in a terminated trial on dental pain, and there is no Pharmaceuticals, which are being tested in phase I trials evidence of other efforts120. Although Nav1.8 channels in collaboration with Genentech145. are prominent in the cell bodies of nociceptive DRG The binding site for sulfonamide inhibitors has neurons, axonal propagation in these neurons generally recently been determined and is completely different to depends on TTX-sensitive channels32,155. In a rat model that of classic small-molecule sodium channel blockers of neuropathic pain, expression of Nav1.8 in axons is such as lidocaine, carbamazepine and other local anaes- sufficiently upregulated to allow TTX-resistant propaga- thetics and anti-epileptic agents, which interact with tion of action potentials in some C fibres32, but it remains the pore region of the sodium channels and generally unclear under what conditions Nav1.8 inhibition can have little selectivity among different sodium channel inhibit the generation or propagation of action potentials subtypes146,147. The high specificity of sulfonamide com- in human pain conditions. pounds and the unusually detailed knowledge of their binding site should allow the design of even more potent Nav1.9. In a similar way to Nav1.8 channels, Nav1.9 Trigeminal neuralgia and selective small-molecule Nav1.7 inhibitors. channels are expressed in small-diameter primary noci- A painful condition caused by Several non-sulfonamide-class compounds that ceptive neurons. Nav1.9 channels activate and inactivate disease affecting, dysfunction are targeted primarily to Nav1.7 channels, but are gen- very slowly, and under normal physiological condi- of or damage to the erally less selective, are also being tested. Raxatrigine tions they seem to be mostly inactivated. Their normal trigeminal nerve. (Convergence/Biogen) has been tested in phase II trials in physio­logical function may be to provide a small cur- 148 Anosmia trigeminal neuralgia , and funamide (TV‑45070, Xenon/ rent to help depolarize the resting potential and amplify Loss of the sense of smell. Teva), is in phase IIb trials for post-herpetic neuralgia149. slow subthreshold depolarizations between the resting

NATURE REVIEWS | DRUG DISCOVERY VOLUME 16 | AUGUST 2017 | 553 ©2017 Mac millan Publishers Li mited, part of Spri nger Nature. All ri ghts reserved. ©2017 Mac millan Publishers Li mited, part of Spri nger Nature. All ri ghts reserved.

REVIEWS

potential and spike threshold156. Nav1.9‑knockout mice release is triggered by various GPCRs172. Activation of exhibit normal sensitivity to acute pain but reduced these receptors is sufficient to facilitate QX‑314 inhibi- hyperalgesia in response to inflammation, suggesting tion of experimentally induced itch172. During inflam- that these channels may represent therapeutic targets157. mation, TRPV1 and TRPA1 channels are activated by Screening for blockers of Nav1.9 has been hampered by endogenous agents, and in this circumstance it is likely the inability to express Nav1.9 in heterologous systems, also unnecessary to co‑apply an exogenous TRP channel although an expressible chimeric Nav1.9–Nav1.4 channel activator. Moreover, because the activation of C fibres has recently been reported158. and release of neuropeptides is often part of the inflam- matory process, inhibiting C‑fibre electrical activity can Targeted delivery of charged sodium channel blockers. actually reduce inflammation, not just block the pain or An alternative strategy for targeting Nav1.7 or Nav1.8 itch resulting from the inflammation. C‑Fibre silencing channels is to use large-pore channels that are selectively by entry of charged sodium channel blockers can reduce expressed by primary pain-initiating nociceptor neurons airway inflammation in animal models of asthma146. as portals for delivering drugs (FIG. 2). Many primary A limitation of this strategy is that the cationic nociceptors express either TRPV1 channels or transient sodium channel blockers must be applied in close prox- receptor potential cation channel subfamily A member 1 imity to the nerve fibres being silenced, because as cat- (TRPA1) channels, or both. These channels are unusual ionic molecules they have limited tissue penetration. in that they form a pore that allows the entry of very large Suitable application methods include direct instillation cations159, including N‑methyl-d‑glucamine (195 Da)160 into wounds, local injections and delivery by aerosol to and the cationic dye FM1‑43 (MW 452 Da)161. This affect lung inflammation. Application to broken skin in property can be exploited to deliver cationic drug mole- dermatitis or similar conditions could also be effective. cules into the neurons. The lidocaine derivative QX‑314 Notably, this limited tissue penetration means that there (N‑ethyl-lidocaine) is a cationic molecule (263 Da) that is is little or no systemic exposure of the cationic agents, ineffective in inhibiting neuronal sodium channels when greatly minimizing the chance of unwanted effects. presented on the extracellular side of the channel because it is too large to enter the sodium channel pore from Calcium channels the outside; however, it can enter and inhibit channels Voltage-dependent calcium channels (Cavs) are when presented on the intracellular side162–164. Such entry calcium-­selective channels that are closed at normal occurs only when channels are open and the QX‑314 resting potentials and opened by depolarization. They molecule can be trapped inside the channels when the underlie many neuronal functions, including helping to channel closes again, so the blocking effect accumulates regulate cellular excitability and mediating the release in a use-dependent manner. The pores of both TRPV1 of synaptic vesicles. Of ten voltage-dependent calcium and TRPA1 channels are large enough for QX‑314 to channels encoded by the mammalian genome, eight permeate into the cell162–164, with entry of the cationic have wide expression in neurons. molecule driven not only by the concentration gradient The most widely used current therapeutic agents but also by the negative intracellular membrane voltage, targeting voltage-dependent calcium channels are gab- promoting higher concentrations inside the cell. Thus, apentin (Neurontin; Pfizer) and pregabalin (Lyrica; externally applied QX‑314 can enter and inhibit sodium Pfizer), which are thought to act by affecting the traf- channels in neurons when TRPV1 or TRPA1 channels ficking and recycling of calcium channel proteins147,173. are activated162,164. Some P2X family channels, including For treating hyperalgesia and neuropathic pain, gab- P2X purinoceptor 2 (P2X2), P2X4 and P2X7, also have apentin and pregabalin are effective in only a minority large pores and can probably admit QX‑314 and similar of patients, and the reasons for this are still unknown. molecules165–167. In initial experiments studying peri- It has been proposed that improved therapeutic selec- neural application to sciatic nerves in normal animals, tivity might be attained by agents that are selective for QX‑314 co‑applied with capsaicin produced long-lasting α2δ1 subunits, as these subunits seem to mediate the inhibition of pain signalling with little effect on motor antihyperalgesic effects of gabapentin and pregabalin. function mediated by the same nerve trunk162,168,169. In However, in addition to α2δ1 subunits, gabapentin and later experiments, QX‑314 was co‑applied with lidocaine, pregabalin bind α2δ2 subunits, which may mediate side which acts as an agonist at both TRPV1 and TRPA1 effects173. Furthermore, recent advances in our under- channels at millimolar concentrations170,171. Lidocaine standing of a couple of calcium channels have opened blocks all nerve function but only transiently (~1 hour), up some additional avenues for analgesic drug design. whereas co‑application with QX‑314 produced noci­­cep­ tive block lasting up to 9 hours or more168,169, probably Cav2.2 splice variants. Synaptic transmission is triggered reflecting long residency of QX‑314 after it has entered by calcium entry through presynaptic calcium channels. TRPV1- and TRPA1‑containing fibres. Most synaptic transmission in the mammalian brain and In many clinical situations, TRPV1 and TRPA1 spinal cord is mediated by Cav2.1 (P/Q‑type) and Cav2.2 channels may be sufficiently activated endogenously (N‑type) calcium channels173,174. Interestingly, transmis- that there is no need to apply an exogenous TRP chan- sion at the first synapses in the pain pathway, glutamatergic nel activator along with the cationic sodium channel synapses from primary sensory neurons to second-order blockers. For example, during itch, TRPV1 or TRPA1 neurons in the dorsal horn, seems to rely nearly exclu- channels are activated by second messengers whose sively on presynaptic Cav2.2 channels. Cav 2.2‑knockout

554 | AUGUST 2017 | VOLUME 16 www.nature.com/nrd ©2017 Mac millan Publishers Li mited, part of Spri nger Nature. All ri ghts reserved. ©2017 Mac millan Publishers Li mited, part of Spri nger Nature. All ri ghts reserved.

REVIEWS

mice have reduced sensitivity to some but not all types such as Cav2.1) have been described and reported to of acute, inflammatory and neuropathic pain (reviewed be effective in rat models of neuropathic pain, inflam- in REF. 37). Cav2.2 channels can be blocked selectively matory pain and mechanical allodynia185–187. However, by several peptides from the venom of cone snails; one whether related compounds are being pursued for of these, ω‑conotoxin MVIIA, also known as ziconitide possible clinical use is unclear. (Prialt; Elan Pharmaceuticals), is currently used to treat pain, with application by intrathecal pump. However, its Potassium channels utility is limited by a small therapeutic window, probably Mammalian neurons express many different types of reflecting the participation of Cav2.2 channels in other potassium channels, with significantly different expres- functions of the spinal cord and perhaps spillover of sion patterns among various types of neurons. A num- peptide into higher regions of the nervous system. ber of different potassium channels are downregulated The efficacy of Prialt has stimulated efforts to find in primary sensory neurons in animal models of hyper- small-molecule inhibitors of Cav2.2 channels that could algesia and neuropathic pain38,188–193, probably contrib- be effective against pain. Several such inhibitors have uting to consequent hyperexcitability of pain-signalling been reported, including TROX‑1 (REFS 175,176). Like neurons. In principle, enhancing current through potas- most small-molecule sodium channel inhibitors, these sium channels could be an effective way of inhibiting are state-dependent inhibitors, which bind more tightly neuronal activity, and the differential expression pat- to activated states of the channel than resting, closed terns of various types of potassium channel offers the states. Therefore, state-dependent inhibition results in possibility of inhibiting the activity of a specific neu- greater inhibition of channels that are repetitively cycling ronal population, such as nociceptors, without excessive quickly through activated states, potentially conferring disruption of other neuronal functions194. The search for at least some selectivity for neurons that are hyperactive. enhancers of potassium currents to inhibit pain signal- Although several such state-dependent calcium chan- ling is relatively recent, but already it is evident that the nel inhibitors have demonstrated effective pain relief general strategy may be promising195,196. So far, the most in several rodent models175–177, efficacy in human trials attention for this approach has been on Kv7 (KCNQ) remains to be demonstrated. family channels35,197–199, and several members of the As Cav2.2 channels are widely expressed in the brain, ‘two-pore’ (K2P) family of channels188,200. An alternative a systemically available channel blocker that crosses approach to small-molecule enhancers of channel func- the blood–brain barrier would probably exhibit seri- tion is to attempt to correct the changes in potassium ous unwanted effects. However, the Cav2.2 channels in channel gene expression that are associated with pain191. primary sensory neurons have different splice variants from the predominant channels in the brain (reviewed Kv7 channels. Kv7 channels are voltage-dependent in REF. 178), raising the possibility of developing small channels that typically activate and deactivate relatively molecules with sufficient selectivity to reduce spinal slowly and often can be activated by small subthreshold cord transmission without deleterious central effects. depolarizations. The original description of a neuronal Kv7 current was ‘M‑current’, named after its ability to Cav3.2 channels. Cav3.2 channels are low-threshold be inhibited by muscarinic stimulation in sympathetic T‑type channels that in various central neurons can neurons and later shown to be mediated by Kv7.2–Kv7.3 mediate burst firing after hyperpolarization (which heteromeric channels201. By activating slowly with small removes resting inactivation of the channels). Their nor- depolarizations, Kv7 current produces strong adaptation mal physiological function in sensory neurons is enig- of firing evoked by steady depolarization. Kv7.2–Kv7.3 matic, but knockout of Cav3.2 channels or knockdown channels seem to be widely expressed in the nervous sys- by antisense RNA has an impressive analgesic effect on tem, often with strong expression at the axon initial seg- mechanical and thermal pain in rats179,180, raising the ment, where they are well positioned to modulate action possibility that they could be targeted therapeutically. potential firing202,203. A number of compounds have been The AbbVie compound ABT‑639, an orally available found that enhance the activation of Kv7 family channels Cav3.2 inhibitor with reasonable selectivity, produced by shifting the voltage-dependence of activation to more dose-dependent antinociception in a rat model of knee- negative voltages. The best known of these is retigabine joint pain and reduced tactile allodynia in several rodent (Trobalt; GlaxoSmithKline), which was introduced for models of neuropathic pain, including spinal nerve liga- use as an anti-epileptic drug. Retigabine inhibits noci- tion and chronic constriction injury (CCI)181. However, ceptive behaviours in rodent models of neuropathic subsequent clinical trials of ABT‑639 with human vol- pain204,205 and hyperalgesia36. Whether the effects are unteers showed no effect in an acute intradermal cap- mediated by central or peripheral neurons is unclear. saicin pain model in which pregabalin was effective182. Disappointingly, however, a clinical study on human Similarly, ABT‑639 did not reduce pain in patients with neuropathic pain failed to meet its primary efficacy end diabetic neuropathy183. Other Cav3.2 inhibitors have point (NCT00612105), adding retigabine to the long list been described to be effective in pain models, including of compounds with efficacy in rodent models of pain a rat model of pain from irritable bowel syndrome184. that seem to have much less dramatic effects in humans. Several molecules that have state-dependent block- Another compound that enhances Kv7 channel-­ ing activity at both Cav2.2 and Cav3.2 channels (and mediated current by shifting activation to more negative variable potency for blocking other calcium channels voltages is flupirtine (Katadolon; Asta Medica), which is

NATURE REVIEWS | DRUG DISCOVERY VOLUME 16 | AUGUST 2017 | 555 ©2017 Mac millan Publishers Li mited, part of Spri nger Nature. All ri ghts reserved. ©2017 Mac millan Publishers Li mited, part of Spri nger Nature. All ri ghts reserved.

REVIEWS

in the same chemical class as retigabine (differing only Kv2–Kv9 channels. Kv2 channels are voltage-activated by substitution of a pyridine ring for a phenyl ring). delayed rectifier potassium currents that are widely Interestingly, flupirtine was introduced to clinical prac- expressed in both peripheral and central neurons. Kv2.1 tice as a centrally acting non-opioid analgesic in Europe subunits have especially wide and high expression in in 1984. The effects of flupirtine may be mediated cen- many neuronal types. A particularly interesting prop- trally and/or peripherally. Small-diameter DRG neu- erty in regards to possible drug development is that Kv2 rons probably corresponding to C fibres express Kv7.2 subunits can make heteromeric channels with subunits subunits, and a component of their potassium current is from other gene families, including Kv5, Kv6, Kv8 and inhibited by the Kv7 blocker XE991 (REF. 206). Knockout Kv9 subunits — these so‑called ‘silent’ subunits do not of Kv7.2 in all somatic sensory neurons produces a mod- seem to make functional channels on their own214,215. In est enhancement of withdrawal from painful mechani- the context of pain, attention has been focused particu- cal and thermal stimuli, mediated primarily by loss of larly on Kv9.1 subunits, for which a common human Kv7.2 from Aδ fibres207, perhaps reflecting the promi- allelic variant is strongly associated with increased nent expression of Kv2 channels in nodes of Ranvier of susceptibility to neuropathic pain6. Subsequent animal many myelinated nerves. Of particular interest, the pri- studies showed that Kv9.1 subunits are strongly down- mary Kv7 subunit expressed by small-diameter C‑fibre regulated in primary sensory neurons after nerve injury, nociceptors is Kv7.5 (REF. 208), raising the possibility that in parallel with the development of pain behaviours216. screening for selective enhancers of this subunit might The reduction of Kv9.1 expression is hypothesized to identify novel peripherally acting analgesics. However, produce spontaneous firing of Aδ fibres210,216,217. Thus, Kv7.5 is also prominently expressed in vascular smooth a plausible pharmacological strategy would be to find muscle, so such agents may also decrease blood pres- small-molecule enhancers of Kv2.1‑containing chan- sure. But if the channels in C‑fibre nociceptors are Kv7.5 nels, either targeted to remaining Kv2.1–Kv9.1 channels homomers and those in vascular muscle are heteromers or to whatever heteromeric combinations are favoured with other Kv7 subunits, there may be an opportunity after downregulation of Kv9.1. Kv2.1 channels are also for selective targeting. strongly expressed in small-diameter C‑fibre DRG neu- rons, where they are potentially associated with various K2P channels. Two-pore-domain (K2P) family channels other silent subunits, including Kv6.1, Kv8.1, Kv9.2 and constitute a diverse group of 15 gene products. Each pro- Kv9.3 (REF. 218). The potential for Kv2‑based channels tein contains four transmembrane segments and two pore made by highly diverse combinations of subunits in domains, with channels formed by dimers of subunits209. different neuronal types raises the possibility of finding K2P channels are widely expressed in both neurons and small molecules with differential selectivity for particular non-neuronal cells. Many K2P channels are constitutively subunit combinations that could preferentially reduce the open, and K2P channels underlie the main resting potas- excitability of pain-signalling neurons. The possibility of sium conductance that is responsible for the negative pharmacological selectivity targeted to different subunit resting potential of cells. Because of their regulation of combinations is validated by differential sensitivity to the both resting potential and input conductance of neurons, broad-spectrum inhibitor 4‑aminopyridine219, but this K2P channels can exert strong control over neuronal has not yet been systematically explored. excitability. For that reason, compounds that enhance the opening of K2P channels (or perhaps upregulate their Ion channels in higher-order neurons expression) are an attractive possibility for drug devel- In neuropathic pain, there is likely to be altered excitability opment. Compared with many other channels, there is and circuitry of higher-order nociceptive neurons in the relatively little known about the functional contributions spinal cord and brain. Little is known about the ion chan- of specific K2P channels in various neuronal types, largely nels involved in controlling changes in the excitability of because there are currently few specific pharmaco­logical higher-order neurons, but as our understanding increases, tools. Nevertheless, recent findings suggest that several it may be possible to selectively target such activity. Agents K2P channels could be attractive clinical targets for pain such as amitryptiline that reduce sodium channel activity relief. A number of K2P channels expressed in DRG neu- broadly, without subtype selectivity, could act in pain relief rons are downregulated after nerve injury and in various by modifying the activity of higher-order neurons as well other animal models of neuropathic pain associated with as primary sensory neurons (FIG. 1). neuronal hypersensitivity210. Of particular interest is Discussion of the mechanism of action of gabapen- TREK1 (also known as KCNK2) on the basis of the strik- tinoids has generally focused on synaptic transmission ing finding that it can be activated by morphine through at first-order synapses in the spinal cord, as their effects μ-opioid receptors; the analgesic action of morphine in on the trafficking of calcium channels have been studied TREK1‑knockout animals is reduced by about half 211. in DRG neurons147, but such effects could also occur at Currently, too little is known about relative expression higher-order synapses. Positive allosteric modulators of

of various K2P channels in different neuronal (and non-­ GABAA chloride channels could have potential as pain Allosteric modulators neuronal) types to predict which might be the best targets relievers220,221, but generally they also produce strong Ligands that alter the activity for pain relief while minimizing the disturbance of other sedation. It was recently found that flupirtine enhances of an agonist, antagonist or inverse agonist of a target by functions. The development of specific pharmacological GABA-mediated currents at concentrations compara- 212,213 222 binding to a site distinct from tools, which is underway in several laboratories , ble to those enhancing Kv7 activity . GABAA channels the active site. should be a key step. are highly diverse in terms of subunit structure, being

556 | AUGUST 2017 | VOLUME 16 www.nature.com/nrd ©2017 Mac millan Publishers Li mited, part of Spri nger Nature. All ri ghts reserved. ©2017 Mac millan Publishers Li mited, part of Spri nger Nature. All ri ghts reserved.

REVIEWS

pentameric combinations of 19 different potential sub- nervous and cardiovascular systems owing to decreased

units, and they are found in different combinations in synthesis of biogenic amines and nitric oxide if BH4 falls different neurons and subcellular regions — the reasons excessively below basal levels. An alternative strategy to for which are incompletely understood. Thus, it may minimize the risk of side effects is targeting sepiapterin

be possible to find agents with selectivity for specific reductase (SPR), the terminal enzyme in the BH4 syn-

subunit combinations that enhance pain-relief prop- thetic pathway, blockade of which allows enough BH4 223 erties relative to sedation . A better understanding production through the BH4 salvage pathway to prevent

of the expression patterns of specific GABAA receptor adverse effects while nevertheless reducing the increased subunit combinations in the higher-order neuronal levels that contribute to neuropathic and inflammatory pain-­mediating circuitry would be useful for a rational pain225. SPR inhibition produces analgesia in multiple approach in this direction. acute and chronic neuropathic and inflammatory pain models with no tolerance and no effect on acute noci- Enzymes as analgesic drug targets ception225. One attractive feature of SPR inhibition for NSAIDs, the most widely used analgesics, target the drug development is that target engagement can be enzyme cyclooxygenase (COX), and there is grow- determined by measuring sepiapterin levels in plasma; ing interest in targeting other enzymes for analgesic sepiapterin can be used as a biomarker of efficacy for (TABLE 3) 225 development . the reduction in BH4 levels . Interestingly, a yeast three hybrid screen using drugs with clinical activity but no

BH4 synthesis: sepiapterin reductase known target led to the discovery that sulfasalazine, an Axonal injury increases the production of tetrahydro- FDA-approved drug that is used to reduce inflamma-

biopterin (BH4) from GTP in injured neurons owing to tion and its associated pain in rheumatoid arthritis and increased expression of GTP cyclohydrolase 1 (GCH1), inflammatory bowel disease, inhibits SPR, although 226 the rate-limiting enzyme in the BH4 pathway, in the with very low potency . Quartet Medicine, through 224 injured cells . BH4 is an essential cofactor for aromatic a strategic partnership with Merck, is developing novel amine hydroxylases, nitric oxide synthases and alkyl­ SPR inhibitors for the treatment of neuropathic and glycerol monooxygenase. Nerve injury is also accom- inflammatory pain.

panied by increased BH4 synthesis in the macrophages that infiltrate proximal to the injury225. Human genetic Microsomal prostaglandin E2 synthase 1

studies reveal an association between GCH1 poly­ Microsomal prostaglandin E2 synthase 1 (mPGES1)

morphisms that decrease BH4 levels and reduced pain is responsible for prostaglandin E2 (PGE2) produc-

in patients. Furthermore, decreasing BH4 levels by tion during inflammation as the terminal enzyme inhibiting or knocking out GCH1 in sensory neurons in the prostanoid pathway, acting downstream of produces analgesia in rodents224,225. Inhibition of GCH1 COX2 (REF. 227). PGE2 is a major inflammatory sen- does, however, create a risk of adverse effects in the sitizer of nociceptors, and mPGES1 inhibition using

Table 3 | Ligands targeting various enzymes relevant to pain Target/category Name Mechanism Preclinical Clinical status Refs SPR • Sulfasalazine, SPRi3 SPR inhibitors, Analgesia: CFA, CCI, SNI ND 225

• QM729 (Quartet BH4 pathway Medicine) FAAH and MGL PF‑04457845 (Pfizer), Endocannabinoid Analgesia in several Several clinical trials failed 244,275–278 BIA 10‑2474 (Bial) pathway models

mPGES1 PF‑4693627 (Pfizer), PGE2 pathway Inflammatory pain LY3023703 in phase II study 228,279–281 LY3023703 (Eli Lilly) (NCT01872910) of post-dental surgical pain iNOS Cindunistat (Pfizer) iNOS inhibitor Significant reduction on No clinical efficacy in people 282 osteoarthritis lesions in with osteoarthritis rodent and canine models Glutamate E2072, 2‑MPPA Inhibitors Neuropathy, neuropathic ND 283,284 carboxypeptidase II pain d‑Amino acid oxidase SUN Inhibitor Neuropathic pain, bone ND 285 cancer pain Leukocyte elastase Sivelestat (Ono Inhibitor Neuropathic pain Approved for use during acute 237 Pharma) respiratory distress, no clinical trials yet for neuropathic pain p38 Losmapimod Inhibitor Inhibition of neuropathic No differentiation from 240 (GlaxoSmithKline) pain placebo

BH4, tetrahydrobiopterin; CCI, chronic constriction injury; CFA, complete Freund’s adjuvant; FAAH, fatty acid amide hydrolase; iNOS, inducible nitric oxide synthase; MGL, monoacylglycerol lipase; mPGES1, microsomal prostaglandin E2 synthase 1; ND, no data on clinical development; PGE, prostaglandin E; SNI, spared nerve injury; SPR, sepiapterin reductase.

NATURE REVIEWS | DRUG DISCOVERY VOLUME 16 | AUGUST 2017 | 557 ©2017 Mac millan Publishers Li mited, part of Spri nger Nature. All ri ghts reserved. ©2017 Mac millan Publishers Li mited, part of Spri nger Nature. All ri ghts reserved.

REVIEWS

MF‑63 (2-(6‑chloro‑1H‑phenanthro-[9,10‑d]imidazol‑­ Mice lacking Serpin A3N develop more mechanical 2‑yl)isophthalonitrile) and two novel 2‑acylaminoim- allodynia after nerve injury than wild-type mice, and idazole inhibitors has been shown to reduce inducible exogenous delivery of Serpin A3N attenuates mechan- PGE synthesis without suppressing prostacyclin genera- ical hypersensitivity­ 237. Serpin A3N inhibits leukocyte tion. The two novel mPGES1 inhibitors produced anal- elastase (also known as neutrophil elastase) which is gesia equivalent to diclofenac in a guinea pig model of released by the T lymphocytes that infiltrate the DRG knee-joint pain (mono iodoacetate model of arthritis). after nerve injury. Genetic loss of leukocyte elastase or mPGES1 inhibition therefore represents a novel approach administration of the leukocyte elastase inhibitor sive- for the management of inflammatory pain, probably lestat (Elaspol; Ono Pharmaceuticals) has been shown to without the adverse effects of COX2 inhibitors: that is, attenuate mechanical allodynia in neuropathic models237. myocardial infarction and gastrointestinal bleeding228. These findings reveal complex crosstalk between injured sensory neurons and immune cells, and identify a novel Soluble epoxide hydrolase immune cell target for further evaluation as an analgesic Epoxyeicosatrienoic acids (EETs) are cytochrome for neuropathic pain. P450‑epoxygenase-derived metabolites of arachidonic acid that act as endogenous signalling molecules in mul- Challenges and considerations tiple biological systems. Soluble epoxide hydrolase (sEH) The assessment of failure in proof‑of‑principle efficacy is the enzyme that degrades EETs, the actions of which clinical trials for novel analgesics is commonly beset by involve analgesia. An inhibitor of sEH, trans‑4-[4-(3‑ insufficient reporting by industry. Consequently, it is often trifluoromethoxyphenyl‑1‑ureido)-cyclohexyloxy]- uncertain whether the wrong target, drug, outcome meas- benzoic acid (t‑TUCB) stabilizes bioactive mediators — ure or patient was selected. It is also unclear whether there epoxy fatty acids (EpFAs) that have low abundance but was a failure of target engagement or too strong a placebo are highly potent bioactive lipids that maintain homeo­ response. If nothing is learned from these phase II trials, it stasis and have analgesic action in a mouse model of is cumulatively an enormous waste of resources, and some diabetic neuropathy229. EpFAs act as upstream modu- drugs may be prematurely abandoned. lators of endoplasmic reticulum stress pathways, which Particularly worrying is the possibility that preclinical are engaged in diabetic neuropathy and whose reduction models are not true surrogates of human disease and leads to analgesia230. However, because some EETs sen- that reflexive-stimulus-evoked ‘pain’ measurement may sitize nociceptors231, sEH inhibition is likely to increase not be a sensitive predictor of efficacy in patients (BOX 1). inflammatory pain232. This would probably complicate In many cases, adverse effects mean that it is simply not the development of these inhibitors as analgesics, as possible to achieve in humans the drug exposure levels inflammatory and neuropathic pain may coexist. that are required in animals to show efficacy. In addi- tion, a large placebo effect and emotional control of pain, Proteases among other reasons, further complicate analgesic clin- Several proteases — serine proteases (trypsin), matrix ical trial design (BOX 2). Some of the key challenges in metalloproteinases (MMPs), cysteine proteases (cathep- assessing novel analgesic action are illustrated by recent sin S), caspases and proteinase-activated receptor 2 pain trials exploring novel ‘pain’ targets. (PAR2) — have been implicated in pain modula- tion233–236. Increased levels of serine proteases and PAR2 Predictive failure of preclinical models have been observed in mice with cancer pain, which is Rodents, especially mice, remain the preclinical model attenuated by inhibition of protease levels or completely organism of choice for the validation of targets and eliminated in mice that are deficient in PAR2 (REF. 233). the testing of their ligands, typically using pain-related Transient increases in MMP9 levels and delayed but behavioural measures. Because many neurobiological persistent increases in MMP2 levels have been shown and disease mechanisms and pathways are evolutionarily to induce and maintain neuropathic pain in rodents234. conserved, much useful information can be gleaned by The same study also showed that MMP inhibitors can using wild-type or genetically modified animals. However, produce antinociception, suggesting that MMPs may be major problems remain. First, there is considerable debate a target for developing therapeutics. It should be noted, as to the relevance of reflexively evoked responses to nox- however, that MMPs are also involved in several other ious stimuli to clinical pain conditions, especially sponta- pathways, and broad-spectrum inhibitors can cause toxic neous pain and changes in cognitive function and mood. side effects. Indeed, clinical trials with MMP inhibitors Second, many of the pain models are not good surrogates for cancer have failed as a result of toxicity and lack of of human disease (BOXES 1,2). Inflaming or damaging a efficacy. Similarly, both caspase 6 and cathepsin S have joint, for example, does not make a good model of osteo­ been shown to regulate neuronal–glial interactions arthritis. Below are two examples in which preclinical during neuropathic pain and potentially lead to central efficacy could not be replicated in human clinical trials. sensitization235,236. Serine protease inhibitor A3N (Serpin A3N), an Propentofylline. The glial-modulating agent propentofyl- endogenous serine protease inhibitor, is upregulated line failed in a study conducted by Solace Pharmaceuticals in injured sensory neurons after axonal injury and to decrease pain in people with post-herpetic­ neuralgia, seems to protect against neuropathic pain in those despite some activity in rodent peripheral neuropathic animals where the degree of upregulation is high237. models. The likely explanation for this failure could be a

558 | AUGUST 2017 | VOLUME 16 www.nature.com/nrd ©2017 Mac millan Publishers Li mited, part of Spri nger Nature. All ri ghts reserved. ©2017 Mac millan Publishers Li mited, part of Spri nger Nature. All ri ghts reserved.

REVIEWS

major species difference between rodents and humans, show efficacy in neuropathic pain for some time. At the both in microglial activation and in propentofylline time that EMA401 was tested, its analgesic mechanism action238. The involvement of microglia and astrocytes of action was uncertain94,241. TRPV1 antagonists, how- in pain in humans remains to be tested, and the failure of ever, have had to be abandoned because of on‑target propentofylline indicates the importance both of human hyperthermia242. target validation, most commonly by genetics, and of drug action on human cells. Achieving CNS penetration Many of the targets relevant to pain are expressed in the p38 pathway inhibitors. The p38 MAPK inhibitor losma- CNS. However, developing analgesics that are capable pimod exhibited no efficacy in people with neuropathic of penetrating the CNS is a significant challenge. The pain from lumbosacral radiculopathy, despite preclin- blood–brain barrier prevents ligands of certain chemi- ical studies showing the involvement of p38 signalling cal structures from effectively penetrating the CNS and pathways in the development of persistent pain after reaching their targets. In addition, after compounds peripheral nerve injury239. It is uncertain whether the trial enter the brain, they may be actively pumped out by failed because of a lack of adequate target engagement various efflux pumps243.

or differences between chronic (conditions lasting for Cannabinoid agonists (CB1) have major actions in years) lumbosacral radiculopathy and much more acute humans, including claims of analgesia87,88. However, the

(conditions lasting for days or weeks) animal models of clinical use of CB1 agonists or cannabis is complicated by neuropathic pain240. Because the preclinical data were not the drug abuse liability from psychoactive effects; there- aligned with the patient cohort tested, the significance of fore, increasing the level of endogenous endocannabi- the failed trial for other pain types is uncertain, which noids represents an attractive alternative strategy. The is true for most neuropathic pain trials. However, p38 endocannabinoids anandamide and 2‑arachidonoylglyc- activation is downstream of nerve growth factor (NGF), erol, acting at CB receptors, are metabolized by hydro- a validated target for novel analgesics (see below) and lytic enzymes (fatty acid amide hydrolase (FAAH) and for angiotensin, which seems to act in the periphery to monoacylglycerol lipase (MGL)) and by oxygenating 90 block TRPV1 sensitization . Notably, an AT2R antago- enzymes (COX2). However, an irreversible fatty acid nist, EMA401, has efficacy in post-herpetic neuralgia92, amide hydrolase 1 inhibitor, PF‑04457845, failed to representing the first drug acting on a novel target to reduce pain from osteoarthritis of the knee, despite decreasing FAAH activity by >96% and substantially increasing fatty acid amide endogenous substrates; fur- Box 2 | Pain clinical trial design thermore, many preclinical studies showed the efficacy 244 Phase II proof-of-concept clinical trials for novel analgesics are beset by a number of of the drug in models of inflammatory pain . Whether major problems: how to measure the pain, how to differentiate placebo from drug the lack of effect of PF‑04457845 on pain reduction rep- effect and how to select patients. Pain is a subjective experience that is influenced resents an inability to access FAAH in the CNS remains not only by heritable traits but also by psychosocial processes, culture, experience, unknown. Despite this failure, other FAAH inhibitors social and biological context, and mood259. Translating this dynamic complex internal are entering clinical trials245,246, and there is preliminary experience into verbal descriptors and numerical or analogue scales of intensity may human genetic support for FAAH involvement in cold not capture comparable levels of sensation between or within subjects, even with pain sensitivity247. training260, and certainly does not reflect the degree or nature of the precipitating initiator or maintainer of the pain. In addition, the high level of placebo and nocebo Adverse effects responders in chronic pain trials is a major confounder261. Attempts are being made to deal with bias and unreliability, most notably by Although chronic pain can be debilitating and can reduce ACTTION (Analgesic, Anesthetic, and Addiction Clinical Trial Translations, quality of life, in several clinical conditions ongoing pain Innovations, Opportunities, and Networks)16, a public–private partnership with the protects damaged tissue from further insult. For exam- US Food and Drug Administration. Trial design can be improved by better diagnostics, ple, anti-NGF antibodies (tanezumab, fulranumab and blinding and choice of primary outcome measure, adaptive designs, crossovers and fasinumab) have, surprisingly, proven to be more effica- random withdrawal, use of adequate power and responder analyses262. However, cious in people with osteoarthritis than would be expected there is a lack of predictive biomarkers, either of pain or of target engagement, which from the preclinical data248, achieving greater efficacy than makes it difficult to understand the underlying pathophysiology that leads to various that of the standard of care (NSAIDs and opioids) for clinical presentations of pain and whether a drug is having any action, although new chronic low back pain, osteoarthritis and diabetic neuro­ imaging technology may change this263. pathy249–251. However, tanezumab resulted in the rapid Although preclinical pain research has evolved into classifying different types of pain on the basis of mechanism, little of this has translated into pain diagnosis in the progression of osteoarthritis, particularly when admin- 252 clinic12. Recent studies have proposed several ways to approach this issue: istered at high doses and with NSAIDs . Although the a successive classification methodology can be used to identify the pain state, pain cause of this is uncertain, one possibility is that the over- mechanism and molecular target in sequence, emphasizing the development of use of damaged joints because of high analgesia cover- individualized therapy or precision medicine12; clinically relevant patient clusters can age may lead to accelerated joint destruction. Additional be identified on the basis of pathophysiological mechanisms254; and, similarly, concern comes from the role of NGF as a growth fac- patients can be matched on the basis of phenotypic characteristics that are tor in development. In pregnant cynomolgus monkeys, 264 predictive of individual variation . These proposals are a first effort to classify intra­venous tanezumab increased stillbirth and infant clinical pain in a fashion that lends itself to improved clinical trial design and to mortality, and decreased infant growth with changes in meaningful pharmacological intervention, where efficacy can be readily identified the peripheral sympathetic and sensory nervous system and quantified. including developmental neurotoxicity149,144.

NATURE REVIEWS | DRUG DISCOVERY VOLUME 16 | AUGUST 2017 | 559 ©2017 Mac millan Publishers Li mited, part of Spri nger Nature. All ri ghts reserved. ©2017 Mac millan Publishers Li mited, part of Spri nger Nature. All ri ghts reserved.

REVIEWS

Phenotypic screens Conclusions: the road ahead Finally, focusing on targets such as Nav1.7 that already Unbiased screening strategies There are several promising emerging targets for the have human genetic data supporting their involvement in which the functional output treatment of acute and chronic pain. However, the key in pain may provide less risky opportunities to develop is a pre-determined alteration question is still whether preclinical efficacy data will new effective analgesics than a reliance on rodent-based of phenotype. translate to humans, and if not, why not. Although models of pain. However, studies should not be restricted high-throughput screening in heterologous cell lines only to targets with human genetic data, as many targets generates selective ligands, the lack of normal path- of the most widely clinically used drugs would not have ways that affect the target of interest in its native cellular been selected on the basis of genetic variants producing environment may lead to the selection of suboptimal clear disease-related phenotypes. In addition, large, rare ligands. In this context, phenotypic screens performed phenotypes, such as insensitivity to pain, may not be use- directly in human neurons differentiated from patient ful for identifying targets engaged in clinical conditions, induced pluripotent stem cell lines may help in greatly such as after damage to the nervous system or in response enhancing the quality of patient-relevant data133,145. A to chronic inflammation. similar argument exists for many of the commonly Despite this, combinations of phenotypic screens in used reflexive-avoidance pain behavioural models in patient-derived neurons, human genetics, genome editing rodents (BOX 1). Given that rodents are prey animals, in mice and human cell lines, as well as improvements in their behaviour can be highly influenced by environ- drug design and kinetics, the identification of biomarkers ment and investigator253. Together, these issues suggest and clinical trial design (BOX 2), offer an increasing prob- that we need to make a considerable effort to develop ability of success. As new therapeutic opportunities arise assays that can measure spontaneous pain automati- from such efforts, they may enable a precision-medicine cally and model human disease conditions as closely approach in which treatment can be targeted at the par- as possible (BOX 1). ticular mechanisms driving pain in individual patients254.

1. Dubois, M. Y., Gallagher, R. M. & Lippe, P. M. 10. Basbaum, A. I., Bautista, D. M., Scherrer, G. & 20. Woolf, C. J. & Salter, M. W. Neuronal plasticity: Pain medicine position paper. Pain Med. 10, 972–1000 Julius, D. Cellular and molecular mechanisms of pain. increasing the gain in pain. Science 288, (2009). Cell 139, 267–284 (2009). 1765–1769 (2000). 2. Johannes, C. B., Le, T. K., Zhou, X., Johnston, J. A. & In this piece, the authors review the biological In this review, the authors conceptualize how Dworkin, R. H. The prevalence of chronic pain in United underpinnings of somatosensation at the circuit, plasticity in ascending sensory pathways may elicit States adults: results of an Internet-based survey. cellular and subcellular levels, with particular pain hypersensitivity by increasing signal gain. J. Pain 11, 1230–1239 (2010). emphasis on pain-related receptors and 21. Chiu, I. M., von Hehn, C. A. & Woolf, C. J. Neurogenic 3. Volkow, N. D. & McLellan, A. T. Opioid abuse in chronic mechanisms. inflammation and the peripheral nervous system in pain — misconceptions and mitigation strategies. 11. Ossipov, M. H., Dussor, G. O. & Porreca, F. host defense and immunopathology. Nat. Neurosci. N. Engl. J. Med. 374, 1253–1263 (2016). Central modulation of pain. J. Clin. Invest. 120, 15, 1063–1067 (2012). This review highlights common misconceptions 3779–3787 (2010). 22. Woolf, C. J. What is this thing called pain? J. Clin. about abuse-related liabilities of prescription This review explores evidence that central Invest. 120, 3742–3744 (2010). opioids and proposes strategies that could help modulatory circuits can dramatically change 23. Julius, D. & Basbaum, A. I. Molecular mechanisms to mitigate these risks. the subjective experience of painful stimuli. of nociception. Nature 413, 203–210 (2001). 4. Decosterd, I. & Woolf, C. J. Spared nerve injury: 12. Vardeh, D., Mannion, R. J. & Woolf, C. J. Toward a Here, the authors describe molecular mechanisms an animal model of persistent peripheral neuropathic mechanism-based approach to pain diagnosis. J. Pain of primary afferent neurons, their modality pain. Pain 87, 149–158 (2000). 17, T50–T69 (2016). sensitivities and various transducers, peptides, In this study, the authors describe a technique Here, the authors propose that identifying specific lipids and growth factors that signal pain and for modelling peripheral neuropathic pain in mechanisms that underlie chronic pain could mediate pain-related signals. laboratory rodents. provide the basis for a personalized-medicine 24. Michaud, K., Bombardier, C. & Emery, P. Quality 5. Honore, P. et al. Murine models of inflammatory, approach to analgesia. of life in patients with rheumatoid arthritis: does neuropathic and cancer pain each generates a unique 13. Woolf, C. J. Overcoming obstacles to developing new abatacept make a difference? Clin. Exp. Rheumatol. set of neurochemical changes in the spinal cord and analgesics. Nat. Med. 16, 1241–1247 (2010). 25, S35–S45 (2007). sensory neurons. Neuroscience 98, 585–598 (2000). This article discusses the many complexities that 25. Drenth, J. P. & Waxman, S. G. Mutations in 6. Costigan, M. et al. Multiple chronic pain states are have made the development of new analgesics so sodium‑channel gene SCN9A cause a spectrum associated with a common amino acid-changing allele challenging. of human genetic pain disorders. J. Clin. Invest. in KCNS1. Brain 133, 2519–2527 (2010). 14. Andrews, N. A. et al. Ensuring transparency and 117, 3603–3609 (2007). This research article describes a putative human minimization of methodologic bias in preclinical 26. Woolf, C. J. Central sensitization: implications for the pain gene that could inform the selection of novel pain research: PPRECISE considerations. Pain 157, diagnosis and treatment of pain. Pain 152, S2–S15 drug targets for patients with neuropathic pain. 901–909 (2016). (2011). It may help to explain why some, but not all, people Here, members of the Preclinical Pain Research 27. Hains, B. C. et al. Upregulation of sodium channel with nerve injury progress to chronic pain. Consortium for Investigating Safety and Efficacy Nav1.3 and functional involvement in neuronal 7. von Hehn, C. A., Baron, R. & Woolf, C. J. Deconstructing (PPRECISE) working group propose new voluntary hyperexcitability associated with central the neuropathic pain phenotype to reveal neural standards of scientific rigour and transparent neuropathic pain after spinal cord injury. J. Neurosci. mechanisms. Neuron 73, 638–652 (2012). reporting to promote more-efficient advancement 23, 8881–8892 (2003). This work describes how the variable expression of of the search for new pain treatments. 28. Nassar, M. A. et al. Nerve injury induces robust sensory nerve injury symptoms can provide insights 15. Singla, N. et al. Assay sensitivity of pain intensity allodynia and ectopic discharges in Nav1.3 null mutant into the underlying pathophysiological mechanisms versus pain relief in acute pain clinical trials: ACTTION mice. Mol. Pain 2, 33 (2006). and guidance for the development of personalized systematic review and meta-analysis. J. Pain 16, 29. Dong, X. W. et al. Small interfering RNA-mediated

pain therapies. 683–691 (2015). selective knockdown of NaV1.8 tetrodotoxin-resistant 8. Ji, R. R., Xu, Z. Z. & Gao, Y. J. Emerging targets in This meta-analysis found that for preclinical acute sodium channel reverses mechanical allodynia in neuroinflammation-driven chronic pain. Nat. Rev. Drug pain trials, readouts of total pain relief may be neuropathic rats. Neuroscience 146, 812–821 Discov. 13, 533–548 (2014). more sensitive to treatment than summed pain (2007). Here, the authors discuss emerging intensity differences. 30. Jarvis, M. F. et al. A-803467, a potent and selective neuroinflammatory pain targets and describe 16. Finnerup, N. B. et al. Pharmacotherapy for Nav1.8 sodium channel blocker, attenuates potential therapeutic opportunities to target neuropathic pain in adults: a systematic review and neuropathic and inflammatory pain in the rat. excessive neuroinflammation. meta-analysis. Lancet Neurol. 14, 162–173 (2015). Proc. Natl Acad. Sci. USA 104, 8520–8525 (2007). 9. Latremoliere, A. & Woolf, C. J. Central sensitization: 17. Costigan, M., Scholz, J. & Woolf, C. J. Neuropathic 31. Ekberg, J. et al. muO-conotoxin MrVIB selectively a generator of pain hypersensitivity by central neural pain: a maladaptive response of the nervous system blocks Nav1.8 sensory neuron specific sodium plasticity. J. Pain 10, 895–926 (2009). to damage. Annu. Rev. Neurosci. 32, 1–32 (2009). channels and chronic pain behavior without This work describes the mechanisms and triggers 18. Patapoutian, A., Tate, S. & Woolf, C. J. Transient motor deficits. Proc. Natl Acad. Sci. USA 103, that underlie the initiation and maintenance of receptor potential channels: targeting pain at the 17030–17035 (2006).

central sensitization, and how they are altered by source. Nat. Rev. Drug Discov. 8, 55–68 (2009). 32. Gold, M. S. et al. Redistribution of NaV1.8 in uninjured changes in the properties and expression patterns 19. Woolf, C. J. Pain: morphine, metabolites, mambas, and axons enables neuropathic pain. J. Neurosci. 23, of glutamate receptors. mutations. Lancet Neurol. 12, 18–20 (2013). 158–166 (2003).

560 | AUGUST 2017 | VOLUME 16 www.nature.com/nrd ©2017 Mac millan Publishers Li mited, part of Spri nger Nature. All ri ghts reserved. ©2017 Mac millan Publishers Li mited, part of Spri nger Nature. All ri ghts reserved.

REVIEWS

33. Joshi, S. K. et al. Involvement of the TTX-resistant 54. Webster, L., Menzaghi, F. & Spencer, R. CR845, 76. Drake, M. T. et al. β-Arrestin-biased agonism at sodium channel Nav 1.8 in inflammatory and a novel peripherally-acting kappa opioid receptor the β2‑adrenergic receptor. J. Biol. Chem. 283, neuropathic, but not post-operative, pain states. agonist, has low abuse potential compared with 5669–5676 (2008). Pain 123, 75–82 (2006). pentazocine. J. Pain 16, S81 (2015). 77. Wisler, J. W., Xiao, K., Thomsen, A. R. & 34. Roza, C., Laird, J. M., Souslova, V., Wood, J. N. & 55. Spahn, V. et al. A nontoxic pain killer designed by Lefkowitz, R. J. Recent developments in biased Cervero, F. The tetrodotoxin-resistant Na+ channel modeling of pathological receptor conformations. agonism. Curr. Opin. Cell Biol. 27, 18–24 (2014). Nav1.8 is essential for the expression of spontaneous Science 355, 966–969 (2017). In this review, the authors describe how different activity in damaged sensory axons of mice. J. Physiol. 56. Milligan, G. The prevalence, maintenance, and relevance ligands can induce distinct receptor conformations 550, 921–926 (2003). of G protein-coupled receptor oligomerization. at the same GPCR to elicit unique downstream 35. Fritch, P. C. et al. Novel KCNQ2/Q3 agonists as Mol. Pharmacol. 84, 158–169 (2013). signalling profiles. They provide support for how potential therapeutics for epilepsy and neuropathic 57. Rozenfeld, R. & Devi, L. A. Receptor these properties could be exploited to develop pain. J. Med. Chem. 53, 887–896 (2010). heteromerization and drug discovery. Trends analgesics with different or reduced side-effect 36. Dost, R., Rostock, A. & Rundfeldt, C. The Pharmacol. Sci. 31, 124–130 (2010). profiles. anti‑hyperalgesic activity of retigabine is mediated 58. Yekkirala, A. S. Two to tango: GPCR oligomers 78. Shukla, A. K. et al. Distinct conformational changes by KCNQ potassium channel activation. Naunyn and GPCR–TRP channel interactions in nociception. in beta-arrestin report biased agonism at seven- Schmiedebergs Arch. Pharmacol. 369, 382–390 Life Sci. 92, 438–445 (2013). transmembrane receptors. Proc. Natl Acad. Sci. USA (2004). 59. Yekkirala, A. S., Kalyuzhny, A. E. & Portoghese, P. S. 105, 9988–9993 (2008). 37. Lee, S. Pharmacological inhibition of voltage-gated An immunocytochemical-derived correlate for 79. Whalen, E. J., Rajagopal, S. & Lefkowitz, R. J. Ca2+ channels for chronic pain relief. Curr. evaluating the bridging of heteromeric mu‑delta Therapeutic potential of beta-arrestin- and Neuropharmacol. 11, 606–620 (2013). opioid protomers by bivalent ligands. ACS Chem. Biol. G protein‑biased agonists. Trends Mol. Med. 17, 38. Stemkowski, P. L., Noh, M. C., Chen, Y. & Smith, P. A. 8, 1412–1416 (2013). 126–139 (2011). Increased excitability of medium-sized dorsal root 60. Lenard, N. R., Daniels, D. J., Portoghese, P. S. & 80. Zidar, D. A., Violin, J. D., Whalen, E. J. & ganglion neurons by prolonged interleukin‑1β Roerig, S. C. Absence of conditioned place preference Lefkowitz, R. J. Selective engagement of G protein exposure is K+ channel dependent and reversible. or reinstatement with bivalent ligands containing coupled receptor kinases (GRKs) encodes distinct J. Physiol. 593, 3739–3755 (2015). mu‑opioid receptor agonist and delta-opioid receptor functions of biased ligands. Proc. Natl Acad. Sci. USA 39. Zogopoulos, P., Vasileiou, I., Patsouris, E. & antagonist pharmacophores. Eur. J. Pharmacol. 566, 106, 9649–9654 (2009). Theocharis, S. E. The role of endocannabinoids in pain 75–82 (2007). 81. Gesty-Palmer, D. et al. A beta-arrestin-biased agonist modulation. Fundam. Clin. Pharmacol. 27, 64–80 61. Aceto, M. D. et al. MDAN‑21: a bivalent opioid ligand of the parathyroid hormone receptor (PTH1R) (2013). containing mu‑agonist and delta-antagonist promotes bone formation independent of G protein 40. Gutenstein, H. & Akil, H. in Goodman and Gilman’s pharmacophores and its effects in rhesus monkeys. activation. Sci. Transl Med. 1, 1ra1 (2009). Pharmacological Basis of Therapeutics Ch. 21 Int. J. Med. Chem. 2012, 327257 (2012). 82. Strachan, R. T. et al. Divergent transducer-specific (eds Brunton, L., Lazo, J. & Parker, K.) 547–590 62. Daniels, D. J. et al. Opioid-induced tolerance and molecular efficacies generate biased agonism at a (The McGraw Hill companies, 2006). dependence in mice is modulated by the distance G protein-coupled receptor (GPCR). J. Biol. Chem. Chapter 21 of this authoritative pharmacology text between pharmacophores in a bivalent ligand series. 289, 14211–14224 (2014). provides a thorough overview of opioid analgesic Proc. Natl Acad. Sci. USA 102, 19208–19213 83. Violin, J. D., Crombie, A. L., Soergel, D. G. & pharmacology. (2005). Lark, M. W. Biased ligands at G‑protein-coupled 41. Fries, D. S. in Principles of Medicinal Chemistry 63. Le Naour, M. et al. Bivalent ligands that target mu receptors: promise and progress. Trends Pharmacol. Ch. 14 (eds Foye, W. O., Lemke, T. L. & Williams, D. A.) opioid (MOP) and cannabinoid1 (CB1) receptors are Sci. 35, 308–316 (2014). 247–269 (William & Wilkins, 1995). potent analgesics devoid of tolerance. J. Med. Chem. 84. Soergel, D. G. et al. Biased agonism of the mu‑opioid Chapter 14 of this essential medicinal chemistry 56, 5505–5513 (2013). receptor by TRV130 increases analgesia and reduces source describes the medicinal chemistry of 64. Akgun, E. et al. Ligands that interact with putative on‑target adverse effects versus morphine: common opioid and anti-inflammatory MOR‑mGluR5 heteromer in mice with inflammatory a randomized, double-blind, placebo-controlled, analgesics. pain produce potent antinociception. Proc. Natl Acad. crossover study in healthy volunteers. Pain 155, 42. Lesniak, A. & Lipkowski, A. W. Opioid peptides in Sci. USA 110, 11595–11599 1829–1835 (2014). peripheral pain control. Acta Neurobiol. Exp. (Wars.) (2013). 85. Soergel, D. G. et al. First clinical experience with 71, 129–138 (2011). 65. Akgun, E. et al. Inhibition of inflammatory and TRV130: pharmacokinetics and pharmacodynamics in 43. Navratilova, E. et al. Positive emotions and brain neuropathic pain by targeting a mu opioid receptor/ healthy volunteers. J. Clin. Pharmacol. 54, reward circuits in chronic pain. J. Comp. Neurol. 524, chemokine receptor5 heteromer (MOR‑CCR5). 351–357 (2014). 1646–1652 (2016). J. Med. Chem. 58, 8647–8657 (2015). 86. Manglik, A. et al. Structure-based discovery of opioid 44. Navratilova, E. & Porreca, F. Reward and motivation in 66. Yekkirala, A. S. et al. N‑Naphthoyl-beta-naltrexamine analgesics with reduced side effects. Nature 537, pain and pain relief. Nat. Neurosci. 17, 1304–1312 (NNTA), a highly selective and potent activator of mu/ 185–190 (2016). (2014). kappa-opioid heteromers. Proc. Natl Acad. Sci. USA 87. Tabrizi, M. A., Baraldi, P. G., Borea, P. A. & 45. Kolodny, A. et al. The prescription opioid and heroin 108, 5098–5103 (2011). Varani, K. Medicinal chemistry, pharmacology, and crisis: a public health approach to an epidemic of 67. Chakrabarti, S., Liu, N. J. & Gintzler, A. R. potential therapeutic benefits of cannabinoid CB2 addiction. Annu. Rev. Public Health 36, 559–574 Formation of mu-/kappa-opioid receptor heterodimer receptor agonists. Chem. Rev. 116, 519–560 (2016). (2015). is sex-dependent and mediates female-specific 88. Han, S., Thatte, J., Buzard, D. J. & Jones, R. M. 46. Cassidy, T. A., DasMahapatra, P., Black, R. A., opioid analgesia. Proc. Natl Acad. Sci. USA 107, Therapeutic utility of cannabinoid receptor

Wieman, M. S. & Butler, S. F. Changes in prevalence of 20115–20119 (2010). type 2 (CB2) selective agonists. J. Med. Chem. 56, prescription opioid abuse after introduction of an 68. Ding, H. et al. A novel orvinol analog, BU08028, 8224–8256 (2013). abuse-deterrent opioid formulation. Pain Med. 15, as a safe opioid analgesic without abuse liability 89. Nevalainen, T. Recent development of CB2 440–451 (2014). in primates. Proc. Natl Acad. Sci. USA 113, selective and peripheral CB1/CB2 cannabinoid 47. Kunins, H. V. Abuse-deterrent opioid formulations: part E5511–E5518 (2016). receptor ligands. Curr. Med. Chem. 21, 187–203 of a public health strategy to reverse the 69. Pasternak, G. W. & Pan, Y. X. Mu opioids and their (2014). opioid epidemic. JAMA Intern. Med. 175, 987–988 receptors: evolution of a concept. Pharmacol. Rev. 65, 90. Anand, U. et al. Mechanisms underlying clinical (2015). 1257–1317 (2013). efficacy of angiotensin II type 2 receptor (AT2R) 48. Chilcoat, H. D., Coplan, P. M., Harikrishnan, V. & 70. Marrone, G. F. et al. Truncated mu opioid GPCR antagonist EMA401 in neuropathic pain: clinical Alexander, L. Decreased diversion by doctor-shopping variant involvement in opioid-dependent and opioid- tissue and in vitro studies. Mol. Pain 11, 38 (2015). for a reformulated extended release oxycodone independent pain modulatory systems within the 91. Danser, A. H. & Anand, P. The angiotensin II type 2 product (OxyContin). Drug Alcohol Depend. 165, CNS. Proc. Natl Acad. Sci. USA 113, 3663–3668 receptor for pain control. Cell 157, 1504–1506 221–228 (2016). (2016). (2014). 49. Coplan, P. M. et al. The effect of an abuse-deterrent 71. Majumdar, S. et al. Truncated G protein-coupled mu 92. Rice, A. S. et al. EMA401, an orally administered opioid formulation on opioid abuse-related outcomes opioid receptor MOR‑1 splice variants are targets for highly selective angiotensin II type 2 receptor in the post-marketing setting. Clin. Pharmacol. Ther. highly potent opioid analgesics lacking side effects. antagonist, as a novel treatment for postherpetic 100, 275–286 (2016). Proc. Natl Acad. Sci. USA 108, 19778–19783 neuralgia: a randomised, double-blind, 50. Cicero, T. J., Ellis, M. S. & Surratt, H. L. Effect of (2011). placebo‑controlled phase 2 clinical trial. abuse-deterrent formulation of OxyContin. N. Engl. 72. Wieskopf, J. S. et al. Broad-spectrum analgesic efficacy Lancet 383, 1637–1647 (2014). J. Med. 367, 187–189 (2012). of IBNtxA is mediated by exon 11‑associated splice 93. Anand, U. et al. Angiotensin II type 2 receptor (AT2 R) 51. Cicero, T. J., Ellis, M. S. & Kasper, Z. A. A tale of 2 variants of the mu‑opioid receptor gene. Pain 155, localization and antagonist-mediated inhibition of ADFs: differences in the effectiveness of abuse- 2063–2070 (2014). capsaicin responses and neurite outgrowth in human deterrent formulations of oxymorphone and 73. Liu, X. Y. et al. Unidirectional cross-activation of GRPR and rat sensory neurons. Eur. J. Pain 17, 1012–1026 oxycodone extended-release drugs. Pain 157, by MOR1D uncouples itch and analgesia induced by (2013). 1232–1238 (2016). opioids. Cell 147, 447–458 (2011). 94. Smith, M. T., Woodruff, T. M., Wyse, B. D., 52. Walsh, S. L., Strain, E. C., Abreu, M. E. & 74. White, K. L. et al. The G protein-biased kappa-opioid Muralidharan, A. & Walther, T. A small molecule

Bigelow, G. E. Enadoline, a selective kappa opioid receptor agonist RB‑64 is analgesic with a unique angiotensin II type 2 receptor (AT2R) antagonist agonist: comparison with butorphanol and spectrum of activities in vivo. J. Pharmacol. Exp. Ther. produces analgesia in a rat model of neuropathic pain hydromorphone in humans. Psychopharmacology 352, 98–109 (2015). by inhibition of p38 mitogen-activated protein kinase (Berl.) 157, 151–162 (2001). 75. Tang, W., Strachan, R. T., Lefkowitz, R. J. & (MAPK) and p44/p42 MAPK activation in the dorsal 53. Pallasch, T. J. & Gill, C. J. Butorphanol and Rockman, H. A. Allosteric modulation of root ganglia. Pain Med. 14, 1557–1568 (2013). nalbuphine: a pharmacologic comparison. beta‑arrestin‑biased angiotensin II type 1 receptor 95. Marion, E. et al. Mycobacterial toxin induces Oral Surg. Oral Med. Oral Pathol. 59, 15–20 signaling by membrane stretch. J. Biol. Chem. 289, analgesia in buruli ulcer by targeting the angiotensin (1985). 28271–28283 (2014). pathways. Cell 157, 1565–1576 (2014).

NATURE REVIEWS | DRUG DISCOVERY VOLUME 16 | AUGUST 2017 | 561 ©2017 Mac millan Publishers Li mited, part of Spri nger Nature. All ri ghts reserved. ©2017 Mac millan Publishers Li mited, part of Spri nger Nature. All ri ghts reserved.

REVIEWS

96. Lemmens, S., Brone, B., Dooley, D., Hendrix, S. & 118. Rittner, H. L. et al. Pain control by CXCR2 ligands 142. Focken, T. et al. Discovery of aryl sulfonamides Geurts, N. Alpha-adrenoceptor modulation in central through Ca2+-regulated release of opioid as isoform-selective inhibitors of NaV1.7 with efficacy in nervous system trauma: pain, spasms, and paralysis peptides from polymorphonuclear cells. FASEB J. 20, rodent pain models. ACS Med. Chem. Lett. 7, 277–282 — an unlucky triad. Med. Res. Rev. 35, 653–677 2627–2629 (2006). (2016). (2015). 119. Wilson, N. M., Jung, H., Ripsch, M. S., Miller, R. J. & 143. McCormack, K. et al. Voltage sensor interaction site for 97. Giovannitti, J. A. Jr, Thoms, S. M. & Crawford, J. J. White, F. A. CXCR4 signaling mediates morphine- selective small molecule inhibitors of voltage-gated Alpha‑2 adrenergic receptor agonists: a review of induced tactile hyperalgesia. Brain Behav. Immun. 25, sodium channels. Proc. Natl Acad. Sci. USA 110, current clinical applications. Anesth. Prog. 62, 31–39 565–573 (2011). E2724–E2732 (2013). (2015). 120. Kalliomaki, J. et al. A randomized, double-blind, 144. Butt, M. et al. Morphologic, stereologic, and 98. Mori, K. et al. Effects of norepinephrine on rat placebo-controlled trial of a chemokine receptor 2 morphometric evaluation of the nervous system in cultured microglial cells that express alpha1, alpha2, (CCR2) antagonist in posttraumatic neuralgia. Pain young cynomolgus monkeys (Macaca fascicularis) beta1 and beta2 adrenergic receptors. 154, 761–767 (2013). following maternal administration of tanezumab, a Neuropharmacology 43, 1026–1034 (2002). 121. Padi, S. S. et al. Attenuation of rodent neuropathic pain monoclonal antibody to nerve growth factor. Toxicol. 99. Lavand’homme, P. M. & Eisenach, J. C. Perioperative by an orally active peptide, RAP‑103, which potently Sci. 142, 463–476 (2014). administration of the alpha2‑adrenoceptor agonist blocks CCR2- and CCR5‑mediated monocyte chemotaxis 145. Wainger, B. J. et al. Modeling pain in vitro using clonidine at the site of nerve injury reduces the and inflammation. Pain 153, 95–106 (2012). nociceptor neurons reprogrammed from fibroblasts. development of mechanical hypersensitivity and 122. Szallasi, A., Cortright, D. N., Blum, C. A. & Eid, S. R. Nat. Neurosci. 18, 17–24 (2015). modulates local cytokine expression. Pain 105, The vanilloid receptor TRPV1: 10 years from channel 146. Talbot, S. et al. Silencing nociceptor neurons reduces 247–254 (2003). cloning to antagonist proof‑of‑concept. Nat. Rev. Drug allergic airway inflammation. Neuron 87, 341–354 100. Feng, X. et al. Intrathecal administration of clonidine Discov. 6, 357–372 (2007). (2015). attenuates spinal neuroimmune activation in a rat 123. Habib, A. M., Wood, J. N. & Cox, J. J. Sodium 147. Bauer, C. S. et al. The increased trafficking of the model of neuropathic pain with existing hyperalgesia. channels and pain. Handb. Exp. Pharmacol. 227, calcium channel subunit alpha2delta‑1 to presynaptic Eur. J. Pharmacol. 614, 38–43 (2009). 39–56 (2015). terminals in neuropathic pain is inhibited by the 101. Wei, H. & Pertovaara, A. Spinal and pontine 124. Waxman, S. G. et al. Sodium channel genes in pain- alpha2delta ligand pregabalin. J. Neurosci. 29, alpha2‑adrenoceptors have opposite effects on related disorders: phenotype-genotype associations 4076–4088 (2009). pain‑related behavior in the neuropathic rat. and recommendations for clinical use. Lancet Neurol. 148. Lawrence, J. Nav1.7: a new channel for pain treatment. Eur. J. Pharmacol. 551, 41–49 (2006). 13, 1152–1160 (2014). Pharm. J. http://dx.doi.org/10.1211/ 102. Schnabel, A., Meyer-Friessem, C. H., Reichl, S. U., 125. Caterina, M. J. et al. The capsaicin receptor: a heat- PJ.2016.20200841 (2016). Zahn, P. K. & Pogatzki-Zahn, E. M. Is intraoperative activated ion channel in the pain pathway. Nature 149. Bowman, C. J. et al. Developmental toxicity assessment dexmedetomidine a new option for postoperative pain 389, 816–824 (1997). of tanezumab, an anti-nerve growth factor monoclonal treatment? A meta-analysis of randomized controlled 126. Nilius, B. & Szallasi, A. Transient receptor potential antibody, in cynomolgus monkeys (Macaca trials. Pain 154, 1140–1149 (2013). channels as drug targets: from the science of basic fascicularis). Reprod. Toxicol. 53, 105–118 (2015). 103. Schnabel, A. et al. Efficacy and safety of research to the art of medicine. Pharmacol. Rev. 66, 150. Murray, J. K. et al. Single residue substitutions that intraoperative dexmedetomidine for acute 676–814 (2014). confer voltage-gated sodium ion channel subtype postoperative pain in children: a meta-analysis of 127. Moran, M. M., McAlexander, M. A., Biro, T. & selectivity in the NaV1.7 inhibitory peptide GpTx‑1. randomized controlled trials. Paediatr. Anaesth. 23, Szallasi, A. Transient receptor potential channels J. Med. Chem. 59, 2704–2717 (2016). 170–179 (2013). as therapeutic targets. Nat. Rev. Drug Discov. 10, 151. Weiss, J. et al. Loss‑of‑function mutations in 104. Melik Parsadaniantz, S., Rivat, C., Rostene, W. & 601–620 (2011). sodium channel Nav1.7 cause anosmia. Nature 472, Reaux‑Le Goazigo, A. Opioid and chemokine receptor 128. Carnevale, V. & Rohacs, T. TRPV1: a target for rational 186–190 (2011). crosstalk: a promising target for pain therapy? drug design. Pharmaceuticals (Basel) 9, E52 (2016). 152. Minett, M. S. et al. Endogenous opioids contribute Nat. Rev. Neurosci. 16, 69–78 (2015). 129. Lehto, S. G. et al. Antihyperalgesic effects of (R,E)-N- to insensitivity to pain in humans and mice lacking 105. Abbadie, C. et al. Chemokines and pain mechanisms. (2‑hydroxy‑2,3‑dihydro‑1H‑inden‑4‑yl)-3-(2- sodium channel Nav1.7. Nat. Commun. 6, 8967 (2015).

Brain Res. Rev. 60, 125–134 (2009). (piperidin‑1‑yl)-4-(trifluorom ethyl)phenyl)-acrylamide 153. Kort, M. E. et al. Subtype-selective NaV1.8 sodium 106. Reaux‑Le Goazigo, A., Van Steenwinckel, J., (AMG8562), a novel transient receptor potential channel blockers: identification of potent, orally active Rostene, W. & Melik Parsadaniantz, S. Current status vanilloid type 1 modulator that does not cause nicotinamide derivatives. Bioorg. Med. Chem. Lett. 20, of chemokines in the adult CNS. Prog. Neurobiol. hyperthermia in rats. J. Pharmacol. Exp. Ther. 326, 6812–6815 (2010). 104, 67–92 (2013). 218–229 (2008). 154. Bagal, S. K. et al. Recent progress in sodium channel 107. Van Steenwinckel, J. et al. CCL2 released from 130. Watabiki, T. et al. Amelioration of neuropathic pain modulators for pain. Bioorg. Med. Chem. Lett. 24, neuronal synaptic vesicles in the spinal cord is a by novel transient receptor potential vanilloid 1 3690–3699 (2014). major mediator of local inflammation and pain after antagonist AS1928370 in rats without hyperthermic 155. Wilson, M. J. et al. μ‑Conotoxins that differentially peripheral nerve injury. J. Neurosci. 31, 5865–5875 effect. J. Pharmacol. Exp. Ther. 336, 743–750 (2011). block sodium channels NaV1.1 through 1.8 identify (2011). 131. Chiche, D., Brown, W. & Walker, P. NEO6860, those responsible for action potentials in sciatic nerve. 108. Xie, F., Wang, Y., Li, X., Chao, Y. C. & Yue, Y. a novel modality selective TRPV1 antagonist: results Proc. Natl Acad. Sci. USA 108, 10302–10307 (2011). Early repeated administration of CXCR4 antagonist from a phase I, double-blind, placebo-controlled 156. Rush, A. M., Cummins, T. R. & Waxman, S. G. Multiple AMD3100 dose-dependently improves neuropathic study in healthy subjects. J. Pain 17, S79 sodium channels and their roles in electrogenesis within pain in rats after L5 spinal nerve ligation. Neurochem. (2016). dorsal root ganglion neurons. J. Physiol. 579, 1–14 Res. 41, 2289–2299 (2016). 132. Gao, Y., Cao, E., Julius, D. & Cheng, Y. TRPV1 (2007). 109. Talbot, S., Foster, S. L. & Woolf, C. J. Neuroimmunity: structures in nanodiscs reveal mechanisms 157. Dib-Hajj, S. D., Black, J. A. & Waxman, S. G. NaV1.9: physiology and pathology. Annu. Rev. Immunol. 34, of ligand and lipid action. Nature 534, 347–351 a sodium channel linked to human pain. Nat. Rev. 421–447 (2016). (2016). Neurosci. 16, 511–519 (2015). 110. Szabo, I. et al. Heterologous desensitization of opioid 133. Cao, L. et al. Pharmacological reversal of a pain 158. Goral, R. O., Leipold, E., Nematian-Ardestani, E. & receptors by chemokines inhibits chemotaxis and phenotype in iPSC-derived sensory neurons and Heinemann, S. H. Heterologous expression of NaV1.9 enhances the perception of pain. Proc. Natl Acad. Sci. patients with inherited erythromelalgia. Sci. Transl chimeras in various cell systems. Pflugers Arch. 467, USA 99, 10276–10281 (2002). Med. 8, 335ra56 (2016). 2423–2435 (2015). 111. Szabo, I. et al. Selective inactivation of CCR5 and 134. Dib-Hajj, S. D., Yang, Y., Black, J. A. & Waxman, S. G. 159. Owsianik, G., Talavera, K., Voets, T. & Nilius, B.

decreased infectivity of R5 HIV‑1 strains mediated by The NaV1.7 sodium channel: from molecule to man. Permeation and selectivity of TRP channels. Annu. opioid-induced heterologous desensitization. Nat. Rev. Neurosci. 14, 49–62 (2013). Rev. Physiol. 68, 685–717 (2006). J. Leukoc. Biol. 74, 1074–1082 (2003). 135. Cox, J. J. et al. An SCN9A channelopathy causes 160. Hellwig, N. et al. TRPV1 acts as proton channel to 112. Grimm, M. C. et al. Opiates transdeactivate congenital inability to experience pain. Nature 444, induce acidification in nociceptive neurons. J. Biol. chemokine receptors: delta and mu opiate 894–898 (2006). Chem. 279, 34553–34561 (2004). receptor‑mediated heterologous desensitization. 136. Cox, J. J. et al. Congenital insensitivity to pain: novel 161. Meyers, J. R. et al. Lighting up the senses: FM1‑43 J. Exp. Med. 188, 317–325 (1998). SCN9A missense and in‑frame deletion mutations. loading of sensory cells through nonselective ion 113. Pello, O. M. et al. Ligand stabilization of CXCR4/ Hum. Mut. 31, E1670–E1686 (2010). channels. J. Neurosci. 23, 4054–4065 (2003). delta‑opioid receptor heterodimers reveals a 137. Nassar, M. A. et al. Nociceptor-specific gene deletion 162. Binshtok, A. M., Bean, B. P. & Woolf, C. J. Inhibition of mechanism for immune response regulation. reveals a major role for Nav1.7 (PN1) in acute and nociceptors by TRPV1‑mediated entry of impermeant Eur. J. Immunol. 38, 537–549 (2008). inflammatory pain. Proc. Natl Acad. Sci. USA 101, sodium channel blockers. Nature 449, 607–610 114. Rivat, C. et al. Src family kinases involved in 12706–12711 (2004). (2007). CXCL12‑induced loss of acute morphine analgesia. 138. Minett, M. S. et al. Distinct Nav1.7‑dependent pain 163. Puopolo, M. et al. Permeation and block of TRPV1 Brain Behav. Immun. 38, 38–52 (2014). sensations require different sets of sensory and channels by the cationic lidocaine derivative QX‑314. 115. Zhao, C. M. et al. Spinal MCP‑1 contributes sympathetic neurons. Nat. Commun. 3, 791 (2012). J. Neurophysiol. 109, 1704–1712 (2013). to the development of morphine antinociceptive 139. Gingras, J. et al. Global Nav1.7 knockout mice 164. Brenneis, C. et al. Bupivacaine-induced cellular entry tolerance in rats. Am. J. Med. Sci. 344, 473–479 recapitulate the phenotype of human congenital of QX‑314 and its contribution to differential nerve (2012). indifference to pain. PLoS ONE 9, e105895 (2014). block. Br. J. Pharmacol. 171, 438–451 (2014). 116. Zhang, N., Rogers, T. J., Caterina, M. & 140. Bagal, S. K., Marron, B. E., Owen, R. M., Storer, R. I. & 165. Virginio, C., MacKenzie, A., Rassendren, F. A., Oppenheim, J. J. Proinflammatory chemokines, such Swain, N. A. Voltage gated sodium channels as North, R. A. & Surprenant, A. Pore dilation of as C‑C chemokine ligand 3, desensitize mu‑opioid drug discovery targets. Channels (Austin) 9, 360–366 neuronal P2X receptor channels. Nat. Neurosci. 2, receptors on dorsal root ganglia neurons. J. Immunol. (2015). 315–321 (1999). 173, 594–599 (2004). 141. Sun, S., Cohen, C. J. & Dehnhardt, C. M. Inhibitors 166. Khakh, B. S., Bao, X. R., Labarca, C. & Lester, H. A. 117. Ye, D. et al. Activation of CXCL10/CXCR3 signaling of voltage-gated sodium channel Nav1.7: patent Neuronal P2X transmitter-gated cation channels attenuates morphine analgesia: involvement of Gi applications since 2010. Pharm. Pat. Anal. 3, change their ion selectivity in seconds. Nat. Neurosci. protein. J. Mol. Neurosci. 53, 571–579 (2014). 509–521 (2014). 2, 322–330 (1999).

562 | AUGUST 2017 | VOLUME 16 www.nature.com/nrd ©2017 Mac millan Publishers Li mited, part of Spri nger Nature. All ri ghts reserved. ©2017 Mac millan Publishers Li mited, part of Spri nger Nature. All ri ghts reserved.

REVIEWS

167. Yan, Z., Li, S., Liang, Z., Tomic, M. & Stojilkovic, S. S. 188. Marsh, B., Acosta, C., Djouhri, L. & Lawson, S. N. 212. Bagriantsev, S. N. et al. A high-throughput functional The P2X7 receptor channel pore dilates under Leak K+ channel mRNAs in dorsal root ganglia: screen identifies small molecule regulators of physiological ion conditions. J. Gen. Physiol. 132, relation to inflammation and spontaneous pain temperature- and mechano-sensitive K2P channels. 563–573 (2008). behaviour. Mol. Cell. Neurosci. 49, 375–386 ACS Chem. Biol. 8, 1841–1851 (2013). 168. Binshtok, A. M. et al. Coapplication of lidocaine and (2012). 213. Rodrigues, N. et al. Synthesis and structure-activity the permanently charged sodium channel blocker 189. Pollema-Mays, S. L., Centeno, M. V., Ashford, C. J., relationship study of substituted caffeate esters as QX‑314 produces a long-lasting nociceptive blockade Apkarian, A. V. & Martina, M. Expression of antinociceptive agents modulating the TREK‑1 in rodents. Anesthesiology 111, 127–137 (2009). background potassium channels in rat DRG is channel. Eur. J. Med. Chem. 75, 391–402 (2014). 169. Roberson, D. P., Binshtok, A. M., Blasl, F., Bean, B. P. cell‑specific and down-regulated in a neuropathic 214. Bocksteins, E. & Snyders, D. J. Electrically silent Kv & Woolf, C. J. Targeting of sodium channel blockers pain model. Mol. Cell. Neurosci. 57, 1–9 (2013). subunits: their molecular and functional into nociceptors to produce long-duration analgesia: 190. Zheng, Q. et al. Suppression of KCNQ/M (Kv7) characteristics. Physiology (Bethesda) 27, 73–84 a systematic study and review. Br. J. Pharmacol. 164, potassium channels in dorsal root ganglion neurons (2012). 48–58 (2011). contributes to the development of bone cancer pain 215. Bocksteins, E. Kv5, Kv6, Kv8, and Kv9 subunits: 170. Leffler, A. et al. The vanilloid receptor TRPV1 is in a rat model. Pain 154, 434–448 (2013). no simple silent bystanders. J. Gen. Physiol. 147, activated and sensitized by local anesthetics in rodent 191. Laumet, G. et al. G9a is essential for epigenetic 105–125 (2016). sensory neurons. J. Clin. Invest. 118, 763–776 silencing of K+ channel genes in acute‑to‑chronic 216. Tsantoulas, C. et al. Sensory neuron downregulation (2008). pain transition. Nat. Neurosci. 18, 1746–1755 of the Kv9.1 potassium channel subunit mediates 171. Leffler, A., Lattrell, A., Kronewald, S., Niedermirtl, F. & (2015). neuropathic pain following nerve injury. J. Neurosci. Nau, C. Activation of TRPA1 by membrane permeable 192. Lu, R. et al. Slack channels expressed in sensory 32, 17502–17513 (2012). local anesthetics. Mol. Pain 7, 62 (2011). neurons control neuropathic pain in mice. J. Neurosci. 217. Tsantoulas, C. et al. Kv2 dysfunction after peripheral 172. Roberson, D. P. et al. Activity-dependent silencing 35, 1125–1135 (2015). axotomy enhances sensory neuron responsiveness reveals functionally distinct itch-generating sensory 193. Lyu, C. et al. G protein-gated inwardly rectifying to sustained input. Exp. Neurol. 251, 115–126 neurons. Nat. Neurosci. 16, 910–918 (2013). potassium channel subunits 1 and 2 are down- (2014). 173. Zamponi, G. W., Striessnig, J., Koschak, A. & regulated in rat dorsal root ganglion neurons and 218. Bocksteins, E. et al. Kv2.1 and silent Kv subunits Dolphin, A. C. The physiology, pathology, and spinal cord after peripheral axotomy. Mol. Pain 11, 44 underlie the delayed rectifier K+ current in cultured pharmacology of voltage-gated calcium channels (2015). small mouse DRG neurons. Am. J. Physiol. Cell and their future therapeutic potential. Pharmacol. 194. Maljevic, S. & Lerche, H. Potassium channels: a review Physiol. 296, C1271–C1278 (2009). Rev. 67, 821–870 (2015). of broadening therapeutic possibilities for neurological 219. Stas, J. I., Bocksteins, E., Labro, A. J. & Snyders, D. J. 174. Zamponi, G. W. Targeting voltage-gated calcium diseases. J. Neurol. 260, 2201–2211 (2013). Modulation of closed-state inactivation in Kv2.1/ channels in neurological and psychiatric diseases. 195. Tsantoulas, C. & McMahon, S. B. Opening paths to Kv6.4 heterotetramers as mechanism for 4‑AP Nat. Rev. Drug Discov. 15, 19–34 (2016). novel analgesics: the role of potassium channels in induced potentiation. PLoS ONE 10, e0141349 In this Review, Zamponi discusses various chronic pain. Trends Neurosci. 37, 146–158 (2015). challenges and opportunities for using calcium (2014). 220. Knabl, J. et al. Reversal of pathological pain through channels as drug targets for neurological 196. Tsantoulas, C. Emerging potassium channel targets for specific spinal GABAA receptor subtypes. Nature 451, disorders. the treatment of pain. Curr. Opin. Support. Palliat. 330–334 (2008). 175. Abbadie, C. et al. Analgesic effects of a substituted Care 9, 147–154 (2015). 221. Bonin, R. P. & De Koninck, Y. Restoring ionotropic N‑triazole oxindole (TROX‑1), a state-dependent, 197. Wickenden, A. D. & McNaughton-Smith, G. Kv7 inhibition as an analgesic strategy. Neurosci Lett. voltage-gated calcium channel 2 blocker. channels as targets for the treatment of pain. Curr. 557, 43–51 (2013).

J. Pharmacol. Exp. Ther. 334, 545–555 (2010). Pharm. Des. 15, 1773–1798 (2009). 222. Klinger, F. et al. δ subunit-containing GABAA receptors 176. Patel, R. et al. Electrophysiological characterization of 198. Wu, Y. J. et al. Discovery of (S,E)-3-(2‑fluorophenyl)-N- are preferred targets for the centrally acting analgesic

activation state-dependent CaV2 channel antagonist (1-(3-(pyridin‑3‑yloxy)phenyl)ethyl)-acrylamide as a flupirtine. Br. J. Pharmacol. 172, 4946–4958 TROX‑1 in spinal nerve injured rats. Neuroscience potent and efficacious KCNQ2 (Kv7.2) opener for the (2015). 297, 47–57 (2015). treatment of neuropathic pain. Bioorg. Med. Chem. 223. Zeilhofer, H. U., Ralvenius, W. T. & Acuna, M. A.

177. Shao, P. P. et al. Aminopiperidine sulfonamide Cav2.2 Lett. 23, 6188–6191 (2013). Restoring the spinal pain gate: GABAA receptors as channel inhibitors for the treatment of chronic pain. 199. Zheng, Y. et al. Activation of peripheral KCNQ targets for novel analgesics. Adv. Pharmacol. 73, J. Med. Chem. 55, 9847–9855 (2012). channels relieves gout pain. Pain 156, 1025–1035 71–96 (2015).

178. Lipscombe, D. & Andrade, A. Calcium channel CaVα1 (2015). 224. Tegeder, I. et al. GTP cyclohydrolase and splice isoforms — tissue specificity and drug action. 200. Mathie, A. & Veale, E. L. Two-pore domain potassium tetrahydrobiopterin regulate pain sensitivity and Curr. Mol. Pharmacol. 8, 22–31 (2015). channels: potential therapeutic targets for the persistence. Nat. Med. 12, 1269–1277 (2006). 179. Bourinet, E. et al. Silencing of the Cav3.2 T‑type treatment of pain. Pflugers Arch. 467, 931–943 225. Latremoliere, A. et al. Reduction of neuropathic calcium channel gene in sensory neurons (2015). and inflammatory pain through inhibition of demonstrates its major role in nociception. EMBO J. 201. Wang, H. S. et al. KCNQ2 and KCNQ3 potassium the tetrahydrobiopterin pathway. Neuron 86, 24, 315–324 (2005). channel subunits: molecular correlates of the 1393–1406 (2015). 180. Choi, S. et al. Attenuated pain responses in mice M‑channel. Science 282, 1890–1893 (1998). 226. Chidley, C., Haruki, H., Pedersen, M. G., Muller, E. &

lacking CaV3.2 T‑type channels. Genes Brain Behav. 6, 202. Pan, Z. et al. A common ankyrin-G‑based mechanism Johnsson, K. A yeast-based screen reveals that 425–431 (2007). retains KCNQ and NaV channels at electrically active sulfasalazine inhibits tetrahydrobiopterin biosynthesis. 181. Jarvis, M. F. et al. A peripherally acting, selective domains of the axon. J. Neurosci. 26, 2599–2613 Nat. Chem. Biol. 7, 375–383 (2011). T‑type calcium channel blocker, ABT‑639, effectively (2006). 227. Chandrasekhar, S. et al. Identification and reduces nociceptive and neuropathic pain in rats. 203. Cooper, E. C. Made for “anchorin”: Kv7.2/7.3 (KCNQ2/ characterization of novel microsomal prostaglandin E Biochem. Pharmacol. 89, 536–544 (2014). KCNQ3) channels and the modulation of neuronal synthase‑1 inhibitors for analgesia. J. Pharmacol. Exp. 182. Wallace, M., Duan, R., Liu, W., Locke, C. & excitability in vertebrate axons. Semin. Cell Dev. Biol. Ther. 356, 635–644 (2016). Nothaft, W. A. Randomized, double-blind, placebo- 22, 185–192 (2011). 228. Jin, Y. et al. Pharmacodynamic comparison of controlled, crossover study of the T‑type calcium 204. Blackburn-Munro, G. & Jensen, B. S. The LY3023703, a novel microsomal prostaglandin E channel blocker ABT‑639 in an intradermal capsaicin anticonvulsant retigabine attenuates nociceptive synthase 1 inhibitor, with celecoxib. Clin. Pharmacol. experimental pain model in healthy adults. Pain Med. behaviours in rat models of persistent and Ther. 99, 274–284 (2016). 17, 551–560 (2015). neuropathic pain. Eur. J. Pharmacol. 460, 109–116 229. Wagner, K., Yang, J., Inceoglu, B. & Hammock, B. D. 183. Ziegler, D., Duan, W. R., An, G., Thomas, J. W. & (2003). Soluble epoxide hydrolase inhibition is antinociceptive Nothaft, W. A randomized double-blind, placebo-, 205. Li, H. et al. Antinociceptive efficacy of retigabine in the in a mouse model of diabetic neuropathy. J. Pain 15, and active-controlled study of T‑type calcium monosodium lodoacetate rat model for osteoarthritis 907–914 (2014). channel blocker ABT‑639 in patients with diabetic pain. Pharmacology 95, 251–257 (2015). 230. Inceoglu, B. et al. Endoplasmic reticulum stress in the peripheral neuropathic pain. Pain 156, 2013–2020 206. Rose, K. et al. Transcriptional repression of the peripheral nervous system is a significant driver of (2015). M channel subunit Kv7.2 in chronic nerve injury. neuropathic pain. Proc. Natl Acad. Sci. USA 112, 184. Francois, A. et al. State-dependent properties of a Pain 152, 742–754 (2011). 9082–9087 (2015).

new T‑type calcium channel blocker enhance CaV3.2 207. King, C. H., Lancaster, E., Salomon, D., Peles, E. & 231. Sisignano, M. et al. 5,6‑EET is released upon neuronal selectivity and support analgesic effects. Pain 154, Scherer, S. S. Kv7.2 regulates the function of activity and induces mechanical pain hypersensitivity 283–293 (2013). peripheral sensory neurons. J. Comp. Neurol. 522, via TRPA1 on central afferent terminals. J. Neurosci. 185. Xu, J. et al. A mixed Ca2+ channel blocker, 3262–3280 (2014). 32, 6364–6372 (2012). A-1264087, utilizes peripheral and spinal 208. King, C. H. & Scherer, S. S. Kv7.5 is the primary Kv7 232. Brenneis, C. et al. Soluble epoxide hydrolase limits mechanisms to inhibit spinal nociceptive transmission subunit expressed in C‑fibers. J. Comp. Neurol. 520, mechanical hyperalgesia during inflammation. in a rat model of neuropathic pain. J. Neurophysiol. 1940–1950 (2012). Mol. Pain 7, 78 (2011). 111, 394–404 (2014). 209. Feliciangeli, S., Chatelain, F. C., Bichet, D. & Lesage, F. 233. Lam, D. K., Dang, D., Zhang, J., Dolan, J. C. & 186. Zhu, C. Z. et al. Mechanistic insights into the The family of K2P channels: salient structural and Schmidt, B. L. Novel animal models of acute and analgesic efficacy of A-1264087, a novel neuronal functional properties. J. Physiol. 593, 2587–2603 chronic cancer pain: a pivotal role for PAR2. Ca2+ channel blocker that reduces nociception in rat (2015). J. Neurosci. 32, 14178–14183 (2012). preclinical pain models. J. Pain 387, e1–e14 210. Busserolles, J., Tsantoulas, C., Eschalier, A. & 234. Christianson, C. A. et al. Spinal matrix (2014). Lopez Garcia, J. A. Potassium channels in neuropathic metalloproteinase 3 mediates inflammatory 187. Scott, V. E. et al. A-1048400 is a novel, orally active, pain: advances, challenges, and emerging ideas. Pain hyperalgesia via a tumor necrosis factor-dependent state-dependent neuronal calcium channel blocker 157 (Suppl. 1), S7–S14 (2016). mechanism. Neuroscience 200, 199–210 (2012). that produces dose-dependent antinociception 211. Devilliers, M. et al. Activation of TREK‑1 by morphine 235. Berta, T. et al. Extracellular caspase‑6 drives murine without altering hemodynamic function in rats. results in analgesia without adverse side effects. Nat. inflammatory pain via microglial TNF-alpha secretion. Biochem. Pharmacol. 83, 406–418 (2012). Commun. 4, 2941 (2013). J. Clin. Invest. 124, 1173–1186 (2014).

NATURE REVIEWS | DRUG DISCOVERY VOLUME 16 | AUGUST 2017 | 563 ©2017 Mac millan Publishers Li mited, part of Spri nger Nature. All ri ghts reserved. ©2017 Mac millan Publishers Li mited, part of Spri nger Nature. All ri ghts reserved.

REVIEWS

236. Clark, A. K., Grist, J., Al‑Kashi, A., Perretti, M. & 257. Hill, R. NK1 (substance P) receptor antagonists — why 279. Luz, J. G. et al. Crystal structures of mPGES‑1 Malcangio, M. Spinal cathepsin S and fractalkine are they not analgesic in humans? Trends Pharmacol. inhibitor complexes form a basis for the rational contribute to chronic pain in the collagen-induced Sci. 21, 244–246 (2000). design of potent analgesic and anti-inflammatory arthritis model. Arthritis Rheum. 64, 2038–2047 258. Sorge, R. E. et al. Different immune cells mediate therapeutics. J. Med. Chem. 58, 4727–4737 (2012). mechanical pain hypersensitivity in male and female (2015). 237. Vicuna, L. et al. The serine protease inhibitor mice. Nat. Neurosci. 18, 1081–1083 (2015). 280. Leclerc, P. et al. Characterization of a human and SerpinA3N attenuates neuropathic pain by 259. Edwards, R. R., Dworkin, R. H., Sullivan, M. D., murine mPGES‑1 inhibitor and comparison to inhibiting T cell-derived leukocyte elastase. Nat. Med. Turk, D. C. & Wasan, A. D. The role of psychosocial mPGES‑1 genetic deletion in mouse models of 21, 518–523 (2015). processes in the development and maintenance of inflammation. Prostaglandins Other Lipid Mediat. 238. Landry, R. P., Jacobs, V. L., Romero-Sandoval, E. A. & chronic pain. J. Pain 17, T70–T92 (2016). 107, 26–34 (2013). DeLeo, J. A. Propentofylline, a CNS glial modulator 260. Smith, S. M. et al. Pain intensity rating training: 281. Sjogren, T. et al. Crystal structure of microsomal does not decrease pain in post-herpetic neuralgia results from an exploratory study of the ACTTION prostaglandin E2 synthase provides insight into patients: in vitro evidence for differential responses PROTECCT system. Pain 157, 1056–1064 diversity in the MAPEG superfamily. Proc. Natl Acad. in human and rodent microglia and macrophages. (2016). Sci. USA 110, 3806–3811 (2013). Exp. Neurol. 234, 340–350 (2012). 261. Dodd, S., Dean, O. M., Vian, J. & Berk, M. A. Review 282. Hellio le Graverand, M. P. et al. A 2‑year randomised, 239. Ji, R. R. Mitogen-activated protein kinases as of the theoretical and biological understanding of the double-blind, placebo-controlled, multicentre study potential targets for pain killers. Curr. Opin. Investig. nocebo and placebo phenomena. Clin. Ther. 39, of oral selective iNOS inhibitor, cindunistat Drugs 5, 71–75 (2004). 469–476 (2017). (SD‑6010), in patients with symptomatic 240. Ostenfeld, T. et al. A randomized, placebo-controlled 262. Paice, J. A. et al. AAPT diagnostic criteria for chronic osteoarthritis of the knee. Ann. Rheum. Dis. 72, trial of the analgesic efficacy and safety of the p38 cancer pain conditions. J. Pain 18, 233–246 187–195 (2013). MAP kinase inhibitor, losmapimod, in patients with (2017). 283. Wozniak, K. M. et al. The orally active glutamate neuropathic pain from lumbosacral radiculopathy. 263. Smith, S. M. et al. The potential role of sensory carboxypeptidase II inhibitor E2072 exhibits Clin. J. Pain 31, 283–293 (2015). testing, skin biopsy, and functional brain imaging as sustained nerve exposure and attenuates peripheral 241. Smith, M. T., Wyse, B. D. & Edwards, S. R. Small biomarkers in chronic pain clinical trials: IMMPACT neuropathy. J. Pharmacol. Exp. Ther. 343, 746–754

molecule angiotensin II type 2 receptor (AT2R) considerations. J. Pain http://dx.doi.org/10.1016/j. (2012). antagonists as novel analgesics for neuropathic pain: jpain.2017.02.429 (2017). 284. Vornov, J. J. et al. Pharmacokinetics and comparative pharmacokinetics, radioligand binding, 264. Edwards, R. R. et al. Patient phenotyping in clinical pharmacodynamics of the glutamate and efficacy in rats. Pain Med. 14, 692–705 (2013). trials of chronic pain treatments: IMMPACT carboxypeptidase II inhibitor 2‑MPPA show prolonged 242. Bevan, S., Quallo, T. & Andersson, D. A. TRPV1. recommendations. Pain 157, 1851–1871 (2016). alleviation of neuropathic pain through an indirect Handb. Exp. Pharmacol. 222, 207–245 (2014). 265. Viscusi, E. R. et al. A randomized, phase 2 study mechanism. J. Pharmacol. Exp. Ther. 346, 406–413 243. Girardin, F. Membrane transporter proteins: investigating TRV130, a biased ligand of the (2013). a challenge for CNS drug development. Dialogues mu‑opioid receptor, for the intravenous treatment of 285. Huang, J. L., Chen, X. L., Guo, C. & Wang, Y. X. Clin. Neurosci. 8, 311–321 (2006). acute pain. Pain 157, 264–272 (2016). Contributions of spinal D‑amino acid oxidase to 244. Huggins, J. P., Smart, T. S., Langman, S., Taylor, L. & 266. Lu, Z. et al. Mediation of opioid analgesia by a bone cancer pain. Amino Acids 43, 1905–1918 Young, T. An efficient randomised, placebo-controlled truncated 6‑transmembrane GPCR. J. Clin. Invest. (2012). clinical trial with the irreversible fatty acid amide 125, 2626–2630 (2015). hydrolase‑1 inhibitor PF‑04457845, which modulates 267. Moriconi, A. et al. Targeting the minor pocket of Acknowledgements endocannabinoids but fails to induce effective C5aR for the rational design of an oral allosteric The authors acknowledge generous funding support from the analgesia in patients with pain due to osteoarthritis of inhibitor for inflammatory and neuropathic pain relief. following US National Institutes of Health grants: NS039518 the knee. Pain 153, 1837–1846 (2012). Proc. Natl Acad. Sci. USA 111, 16937–16942 and NS038253 (US National Institute of Neurological 245. Keith, J. M. et al. Preclinical characterization of the (2014). Disorders and Stroke (NINDS) to C.J.W); NS072040 and FAAH inhibitor JNJ‑42165279. ACS Med. Chem. Lett. 268. Altun, A. et al. Attenuation of morphine NS036855 (NINDS to B.P.B.); and DA041912 (US National 6, 1204–1208 (2015). antinociceptive tolerance by cannabinoid CB1 and Institute on Drug Abuse to A.S.Y.). 246. Pawsey, S. et al. Safety, tolerability and CB2 receptor antagonists. J. Physiol. Sci. 65, pharmacokinetics of FAAH inhibitor V158866: 407–415 (2015). Competing interests statement a double-blind, randomised, placebo-controlled 269. Haugh, O., Penman, J., Irving, A. J. & Campbell, V. A. The authors declare competing interests: see Web version phase I study in healthy volunteers. Drugs R. D 16, The emerging role of the cannabinoid receptor family for details. 181–191 (2016). in peripheral and neuro-immune interactions. 247. Cajanus, K. et al. Effect of endocannabinoid Curr. Drug Targets 17, 1834–1840 (2016). Publisher’s note degradation on pain: role of FAAH polymorphisms 270. Deliu, E. et al. The lysophosphatidylinositol receptor Springer Nature remains neutral with regard to jurisdictional in experimental and postoperative pain in women GPR55 modulates pain perception in the claims in published maps and institutional affiliations. treated for breast cancer. Pain 157, 361–369 periaqueductal gray. Mol. Pharmacol. 88, 265–272 (2016). (2015). 248. Woolf, C. J., Safieh-Garabedian, B., Ma, Q. P., Crilly, P. 271. Kort, M. E. & Kym, P. R. TRPV1 antagonists: clinical & Winter, J. Nerve growth factor contributes to the setbacks and prospects for future development. Prog. DATABASES generation of inflammatory sensory hypersensitivity. Med. Chem. 51, 57–70 (2012). ClinicalTrials.gov: https://clinicaltrials.gov/ Neuroscience 62, 327–331 (1994). 272. Manitpisitkul, P. et al. Safety, tolerability and 249. Gimbel, J. S. et al. Long-term safety and effectiveness pharmacokinetic and pharmacodynamic learnings FURTHER INFORMATION of tanezumab as treatment for chronic low back pain. from a double-blind, randomized, placebo-controlled, Arena Pharmaceuticals press release on APD371: http:// Pain 155, 1793–1801 (2014). sequential group first‑in‑human study of the TRPV1 invest.arenapharm.com/releasedetail.cfm?ReleaseID=964598 250. Bramson, C. et al. Exploring the role of tanezumab as antagonist, JNJ‑38893777, in healthy men. BioTuesdays — Xenon still committed to TV‑45070 for a novel treatment for the relief of neuropathic pain. Clin. Drug Investig. 35, 353–363 (2015). neuropathic pain: http://www.biotuesdays.com/features/2015/ Pain Med. 16, 1163–1176 (2015). 273. Quiding, H. et al. TRPV1 antagonistic analgesic 7/21/xenon-still-committed-to-tv-45070-for-neuropathic-pain 251. Schnitzer, T. J. & Marks, J. A. A systematic review of effect: a randomized study of AZD1386 in pain Cara Therapeutics — Kappa Opioid Receptor Agonists: the efficacy and general safety of antibodies to NGF in after third molar extraction. Pain 154, 808–812 http://www.caratherapeutics.com/pipeline-technology/ the treatment of OA of the hip or knee. Osteoarthritis (2013). kappa-opioid-receptor-agonists/ Cartilage 23 (Suppl. 1), S8–S17 (2015). 274. De Petrocellis, L. & Moriello, A. S. Modulation of CNV2197944 phase II trial announcement press release: 252. Hochberg, M. C. et al. When is osteonecrosis the TRPV1 channel: current clinical trials and http://www.prnewswire.com/news-releases/convergence- not osteonecrosis? Adjudication of reported recent patents with focus on neurological pharmaceuticals-announces-start-of-phase-ii-study-for- serious adverse joint events in the tanezumab conditions. Recent Pat. CNS Drug Discov. 8, cav22-selective-blocker-cnv2197944-in-pain-associated- clinical development program. Arthritis Rheumatol. 180–204 (2013). with-diabetic-peripheral-neuropathy-dpn-214577061.html 68, 382–391 (2016). 275. Ghosh, S. et al. Full fatty acid amide hydrolase CR845 phase IIa results press release: http://ir. 253. Sorge, R. E. et al. Olfactory exposure to males, inhibition combined with partial monoacylglycerol caratherapeutics.com/releasedetail.cfm?ReleaseID=946233 including men, causes stress and related analgesia lipase inhibition: augmented and sustained CR845 phase III patient recruitment press release: http://ir. in rodents. Nat. Methods 11, 629–632 (2014). antinociceptive effects with reduced cannabimimetic caratherapeutics.com/releasedetail.cfm?ReleaseID=974348 254. Baron, R. et al. Peripheral neuropathic pain: side effects in mice. J. Pharmacol. Exp. Ther. 354, DEX‑IN phase II efficacy press release: a mechanism-related organizing principle based on 111–120 (2015). http://ir.recropharma.com/press-releases/detail/27/recro- sensory profiles. Pain 158, 261–272 (2016). 276. Grim, T. W. et al. Combined inhibition of FAAH pharma-announces-positive-top-line-results-for-phase 255. Mogil, J. S., Davis, K. D. & Derbyshire, S. W. and COX produces enhanced anti-allodynic effects DEX‑IN phase II side effects press release: The necessity of animal models in pain research. in mouse neuropathic and inflammatory pain http://ir.recropharma.com/press-releases/detail/29/recro- Pain 151, 12–17 (2010). models. Pharmacol. Biochem. Behav. 124, 405–411 pharma-announces-additional-information-for-phase-ii In this review, Mogil and colleagues describe (2014). US Centers for Disease Control and Prevention — the crucial role of animal models in the 277. Starowicz, K. et al. Full inhibition of spinal FAAH leads Overdose: http://www.cdc.gov/drugoverdose/ development of clinical analgesics and propose to TRPV1‑mediated analgesic effects in neuropathic Xenon and Genentech collaboration press release: https:// advancements in animal models of pain that could rats and possible lipoxygenase-mediated remodeling globenewswire.com/news-release/2016/03/15/819870/0/en/ improve the relevance and translatability of of anandamide metabolism. PLoS ONE 8, e60040 Xenon-and-Genentech-Extend-Collaborations-Focused-on- preclinical studies. (2013). Pain.html 256. Whiteside, G. T., Pomonis, J. D. & Kennedy, J. D. 278. Schlosburg, J. E. et al. Chronic monoacylglycerol Z944 phase Ib results press release: http://hpr-cro.com/ An industry perspective on the role and utility of lipase blockade causes functional antagonism of wp-content/uploads/2014/06/Press_release_HPR_170614.pdf Neurosci. Nat. Neurosci. 13 animal models of pain in drug discovery. the endocannabinoid system. , ALL LINKS ARE ACTIVE IN THE ONLINE PDF Lett. 557, 65–72 (2013). 1113–1119 (2010).

564 | AUGUST 2017 | VOLUME 16 www.nature.com/nrd ©2017 Mac millan Publishers Li mited, part of Spri nger Nature. All ri ghts reserved. ©2017 Mac millan Publishers Li mited, part of Spri nger Nature. All ri ghts reserved.

CORRECTION

CORRIGENDUM Breaking barriers to novel analgesic drug development Ajay S. Yekkirala, David P. Roberson, Bruce P. Bean & Clifford J. Woolf Nature Reviews Drug Discovery 16, 545–564 (2017) The compounds APD371,LY2828360, S-777469 and KHK6188 were incorrectly referred to as inhibitors of the cannabinoid receptors CB1 and CB2 in Table 1, when they are cannabinoid receptor agonists. In addition, KHK6188 is not currently in a Phase 2 clinical trial for neuropathic pain as stated in Table 1 and development of this agent has been discontinued. The error has been corrected in the html and pdf versions online.

NATURE REVIEWS | DRUG DISCOVERY www.nature.com/nrd ©2017 Mac millan Publishers Li mited, part of Spri nger Nature. All ri ghts reserved.