Limitations and potential of κOR biased agonists for pain and itch management

Abstract

The concept of ligand bias is based on the premise that different agonists can elicit distinct responses by selectively activating the same receptor. These responses often determine whether an agonist has therapeutic or undesirable effects. Therefore, it would be highly advantageous to have agonists that specifically trigger the therapeutic response. The last two decades have seen a growing trend towards the consideration of ligand bias in the development of ligands to target the κ-opioid receptor (κOR). Most of these ligands selectively favor G-protein signaling over β-arrestin signaling to potentially provide effective pain and itch relief without adverse side effects associated with κOR activation. Importantly, the specific role of β-arrestin 2 in mediating κOR agonist-induced side effects remains unknown, and similarly the therapeutic and side-effect profiles of G-protein-biased κOR agonists have not been established. Furthermore, some drugs previously labeled as G-protein-biased may not exhibit true bias but may instead be either low-intrinsic-efficacy or partial agonists. In this review, we discuss the established methods to test ligand bias, their limitations in measuring bias factors for κOR agonists, as well as recommend the consideration of other systematic factors to correlate the degree of bias signaling and pharmacological effects.

1. Introduction

κ-Opioid receptor (κOR) is a (Gi/o/z) protein-coupled receptor (GPCR) that is widely expressed and mediates important physiological functions such as antinociception, stress, anxiety, depression, and substance use disorder, through its unique signaling pathways (Al-Hasani et al., 2015, 2017; Liu et al., 2019a; Massaly et al., 2019). The side effects often associated with activation of μ-opioid receptor (μOR) agonists, such as respiratory depression and euphoria, have made κOR an attractive target for drug development in the treatment of pain and itch. However, a major limitation remains, as activation of κOR causes sedation, dysphoria, and hallucinations (Paton et al., 2020) (refer to Table 1).

Table 1.

Therapeutic and side effects of κOR agonists in preclinical and clinical studies (Due to large file, see separate Table 1.).

KOR-agonist	Preclinical effects	Side effects	Clinical effects		Bias factor/reference	Assays/model
			Beneficial effects	Side effects
Dyn 1–7 analogs KA204, KA305, KA311 (peptides)	↓ Withdrawal threshold (antinociception) of complete Freund adjuvant injected paw: Randall-Selitto test (0.3 mg/kg, i.pl., mice, M) 1	None	None	None	pERK, 2/ U50,488 1	ALPHA Screen cAMP and AlphaLISA SureFire Ultra pERK1/2 assays/ Intrinsic reactive activity (RAi)
(—) Spiradoline U-62,066E	↓ Nociception in Tail flick and Tail pinch (0.25 mg/kg, s.c. mice, M) ↓ Abdominal constriction (HCl) (0.18 mg/kg, s.c., mice, M) ↓ Spontaneously hypertension, socially isolated induced hypertension, normotensive (5 nmol/intrahippocampal microinjections) ↓ (moderate) Cocaine self-administration (0.0032–0.018 mg/kg/h; monkeys) 2	↑ Diuresis (0.32 mg/kg, i.p., rats, M) (63)2			G, 6/Salvinorin A3	GloSensor cAMP and Tango assays/ Transduction coefficient log(τ/K(A) 3
(+)-U50,488	↓ Nociception: tail-flick, tail pinch, writhing tests (2.5 mg/ kg, s.c., mice, M) 4 ↓ Nociception in warm plate (4.4 mg/kg, s.c., M), hot plate (7.0 mg/kg, s.c., M), tail flick (32 mg/kg, s.c., M) and air writhing (1.4 mg/kg, s.c. rat, M)) 4 ↓ Tumor cell growth (15.6–250 μM, NSCLC cell lines) 5 ↓ Scratching (1–10 mg/kg, p.o., mice) 6 ↓ Brain edema and neuronal injury (30 mg/kg, i.p.) in global cerebral ischemia in rat models (M) 7 ↑ Oligodendrocyte differentiation (0.5 μM) in vitro OPC cultures 8 ↑ Remyelination in the lysolecithin-induced demyelination mouse model (10 mg/kg/day, p.o., M & F) 8 ↑ Remyelination (1.6 mg/kg, i.p.) in EAE mice models (F)9 ↑ Remyelination (1.6 mg/kg, i.p.) in cuprizone-induced demyelination mice model (F) 9 ↑ Learning and memory (1.25 mg/kg, s.c.) in Morris water maze test in mouse model (M) of Alzheimer’s disease 10	↑ Aversion (10.0 mg/kg, i.p. rat, M) in cocaine induced place preference test 11 ↑ Sedation (10 mg/kg, i.p. mice, M) in CPP test 12 ↑ Anxiolytic actions (10 and 1000 μg/kg, i.p.) in elevated plus-maze test in rats (M) 8 ↓ Novel object recognition in mice (1.0, 10.0 mg/kg, i.p. M) 13	None	None	G, 8/ Salvinorin A3	GloSensor cAMP and Tango assays/ Transduction coefficient log(τ/K(A) 3
16-Bromo salvinorin A (16-BrSalA)	↓ Cocaine-induced reinstatement of drug-seeking behavior (0.3, 1.0 mg/kg, i.p., rats, M) ↓ Nociception: tail-flick (1 mg/kg, i.p., mice, M)14	No effect on motor coordination, locomotor in elevated plus maze test (1 mg/kg, i.p., mice, M) ↑ Aversion in the CPA test (1 mg/kg, i.p., mice, M) No effect on mobility or immobility time in the forced swim test (1 mg/ kg, i.p., mice, M) No effect in sucrose reinforcement test (1 mg/kg, i.p., mice, M)14 Sedative in rotorod performance assay (1 mg/kg i.p)15	None	None	G, 7.7/ U50,488 15	cAMP Accumulation (HitHunter™) and β-arrestin recruitment (PathHunter) assays/ Intrinsic reactive activity (RAi)
6′- Guanidinonaltrindole (6′-GNTI)	↓ PGE2-paw withdrawal latency (1 μg intraplantar, mice, M) 16	Needs both the κOR and δ-opioid receptor (δOR) for the antinociceptive effect16	None	None	G, 6/ Salvinorin A3	GloSensor cAMP and Tango assays/ Transduction coefficient log(τ/K(A) 3
Collybolide	↓ Nociception: tail-flick assay (2 mg/kg, i.p., mice, M). ↓ Depression: FST (2 mg/kg, i.p., mice, M) ↓ Pruritus: chloroquine- mediated itch (2 mg/kg, i.p., mice, M) 17	↑ Aversion in the CPA test (2 mg/kg, i.p., mice, M) ↑ Anxiogenic activity: open field test (2 mg/kg, i.p., mice, M)17	None	None	G, N.D./ Salvinorin A17	Phosphorylation of Akt and ERK1/2 phosphorylation assays/ N.D17
Dyn 1–8, Dyn 1–9, Dyn 1–11	↑ Spontaneous contralateral circling (20 nmol, ic.v mice, M) 18 Inhibiting dopamine in brain regions that are associated with drug dependence 19	None	None	None	G, 4/ Salvinorin (Dyn 1–8) 3 G, 16/ Salvinorin A (Dyn 1–9) 3 G, 44/ Salvinorin A (Dyn 1–11) 3	GloSensor cAMP and Tango assays/ Transduction coefficient log(τ/K(A) 3 GloSensor cAMP and Tango assays/ Transduction coefficient log(τ/K(A)3 GloSensor cAMP and Tango assays/ Transduction coefficient log(τ/K(A) 3
HS665	↓ Nociception: tail-flick (HS665 3.74 nM, HSS6 6.02 nM, i.c.v, mice, M)20	No effect on motor performance in rotarod test [HS665 (10 nM)]. No effect on CPA/CPP [HS665 (30 nM) mice, M, i.c.v)20	None	None	G, 389/ U69,59320	[35S]-GTPγS and the PathHunter β-arrestin 2 recruitment assays/ Transduction coefficient log(τ/K(A)20
HS666	↓ Nociception: tail-flick HSS6 6.02 nM, i.c.v, mice, M)20	No effect on motor performance in rotarod test [HS666 (30 nM)]. No effect on CPA/CPP [HS666 (150 nM, mice, M, i.c.v)20	None	None	G, 62/ U69,59320	[35S]-GTPγS and the PathHunter β-arrestin 2 recruitment assays/ Transduction coefficient log(τ/K(A)20
Isoquinolinone 2.1	↓ Nociception tail withdrawal assay (30,mg/kg i.p., mice, M)21	None	None	None	G, 31.4/ U69,59322 G, 7.2/ U69,59321	[35S]-GTPγS and β-arrestin 2 recruitment (using the EFC method) assays/Transduction coefficient log(τ/K(A)22 [35S]-GTPγS and β-arrestin 2 Imaging/ Transduction coefficient log(τ/K(A)21
LOR17	↓ Nociception: tail-flick (10.07 ± 0.36 mg/kg, i.p., mice, M) ↓ Nociception: writhing test (5.74 ± 0.46 mg/kg, i.p., mice, M) ↓ Thermal hypersensitivity: oxaliplatin-induced neuropathic nociception (10–20 mg/kg, s.c., mice, M)23	No effect on motor coordination, locomotor but depressive-like behavior (10 mg/kg, s.c., mice, M) 23	None	None	G, 853/ U50,48823	ELISA cAMP and PathHunter β-arrestin 2 recruitment assay/ Transduction coefficient log(τ/K(A)23
Salvinorin B Mesylate (Mesyl Sal B)	↓ Nociception: tail-flick (1 mg/kg, i.p., mice, M) 24 ↑ duration of action (Vs. SalA,1 mg/kg, i.p., mice, M) 24 ↓ Cocaine induced hypersensitivity (0.3 mg/kg, i.p., rat, M) 25 No effect on sucrose intake (0.3–1 mg/kg, i.p., rat, M) 25 No effect on memory impairment: novel object recognition (0.3–1 mg/kg, i.p., rat, M) 25	No effect on aversion, anxiety, or learning and memory (1–2 mg/kg, i.p., rat, M) 25 No effect on sedation (1–2 mg/kg, i.p., rat, M) 25	None	None	G, 0.61/ U50,48825	cAMP Accumulation (HitHunter™) and β-arrestin recruitment (PathHunter) assays/ Intrinsic reactive activity (RAi)25
RB-64	↑Anti-nociception: tail-flick (3 mg/kg, i.p., mice, β-arrestin Ko mice M) No anhedonia-like effect (3 mg/kg, i.p.,WT)26	No aversion at (1 mg/ kg) in WT β-arrestin-2 KO mice in CPA assay ↑ aversion at (3 mg/kg) in WT β-arrestin-2 KO mice in CPA assay no effect on rotorod performance (3, 10 mg/ kg, WT, β-arrestin-2 KO mice No effect on novelty induced loccomotion (3 mg/kg, WT, β-arrestin-2 KO mice26	None	None	G, 96 (m κOR)/ Salvinorin A 26 G, 35/ Salvinorin A26	GloSensor cAMP and Tango assays/ Transduction coefficient log(τ/K(A)26
Triazole 1.1	↓ Nociception (30 mg/kg, i.p., mouse, M) Tail flick assay 21 ↓ Pruritis (1–15 mg/kg, s.c., mice, M) in chloroquine phosphate-induced scratching responses 27 ↑ Anti-nociceptive effects: ICSS abdominal nociception model (24 mg/kg, i.p., rat) 27 ↓Oxycodone induced thermal antinociception (3.2–32 mg/ kg., rats M) 28	No effect on sedation, open field test No effect on dysphoria (3–30 mg/kg, i.p., rat) 27	None	None	[35S]-GTPγS and β-arrestin 2 Imaging/ Intrinsic reactive activity (RAi)29 G, 93/ U69,59329 G, 61.2/ U69,59321	G, 47/U69,59329 [35S]-GTPγS and ERK1/ 2 Phosphorylation/ Intrinsic reactive activity (RAi)29 [35S]-GTPγS and β-arrestin 2 recruitment (using the EFC method) assays/Transduction coefficient log(τ/K(A)21
G, 20/U69,59321	[35S]-GTPγS and β-arrestin 2 Imaging/Transduction coefficient log(τ/K(A)21
(−)-U50,488	N/A	N/A	None	None	Arr, 2/ Salvinorin A3	GloSensor cAMP and Tango assays/ Transduction coefficient log(τ/K(A) 26
GR89696	↑ Anti-nociception: spinal cord injury (0.32 μM, i.t., rat, M) and tail-flick (0.32 μM, i.t., rat, M) 30 ↑ Anti-nociception: tail-flick (0.01–0.1 g/kg, i.m., rhesus monkeys, M &F) 31 ↓ Itching: intrathecal morphine-induced scratching (0.1 μg/kg, i.m., rhesus monkeys, M & F) 31 ↑ Neuroprotection: global and cerebral ischemia (3–30 μg/ kg, s.c., gerbil, M & F) 32	↓ Locomotor recovery in a contusion model [0.32 μM, i.t., rat, M] 30	None	None	Arr, 5/ Salvinorin A3	GloSensor cAMP and Tango assays/ Transduction coefficient log(τ/K(A) 3
WMS-X600	None	None	None	None	Arr, 10/ U50,48833	GloSensor cAMP and Tango assays/ transduction coefficient log(τ/K(A)33
Compound 81	Not tested due to low potency	None	None	None	G, 6/ Salvinorin A34	GloSensor cAMP and Tango assays/ transduction coefficient log(τ/K(A) 34
Dyn 1–13	Improved the impairment of spontaneous alternation performance induced by DAMGO (10 ng) in a Y-maze (3 and 10 μg/mice)35	No effect on spontaneous alternation performance in a Y-maze (3 and 10 μg/ mice)35	None	None	G, 34/ Salvinorin A3	GloSensor cAMP and Tango assays/ Transduction coefficient log(τ/K(A) 3
Nalfurafine	↓ Pruritis: substance P and histamine induced itch (100 μg/kg, p.o., mice, M) [b] and compound 48/80-induced itch (2.5–30 μg/kg, s.c., mice, M) 36 ↓ Nociception: tail-flick (ED50 0.062 mg/kg, s.c., mice, M & F) 37 ↓ Neuroinflammation: EAE model of MS (0.03–0.01 mg/ kg, i.p., mice, F) 38 ↑ Remyelination: cuprizone model (0.01 mg/kg, i.p., mice, F) 38 ↓ Levodopa-induced dyskinesia, parkinsonism (10–30 μg/kg, s.c., rat, M) 39	No effect on anhedonia (sucrose preference test) (10 μg/kg, s.c.) and anxiety (elevated plus maze test) (10 μg/kg, s.c.) 40 ↑ Diuresis (5–10 μg/kg, s.c., rat, M) [h] No effect on CPA or CPP (5–20 μg/kg, s.c., M) 21, 41	(3–12 mg/kg, p.o.) n = 110; s.e., all; age 18–55 years [NCT05029401] Clinically approved (Japan) (2009) Remitch for medication-resistant pruritus in patients with hemodialysis Phase III: completed (9/ 2009): For the treatment of pruritus in patients receiving hemodialysis (5 μg, p.o.) n = 104; s.e., all; age 20 years and older [NCT01513161]. Phase II: completed (3/ 2018): Nalfurafine as a treatment for pruritus in patients with primary biliary cholangitis n = 44; s.e., all; age 18 years and older [NCT02659696]. Phase II: completed (12/ 2009): Pruritus in patients with chronic liver disease (2.5–10 μg, p.o.) n = 120; s.e., all; age 20 years and older [NCT00638495]. Phase III: recruiting patients (Exp. completion date 10/2021): Efficacy, safety and plasma concentration of nalfurafine for treatment of refractory pruritus (5 μg, p.o.) n = 133; s.e., all; age 18 years and older [NCT04728984]42.	Insomnia (≥3% (5 μg/kg, p.o.)43	G, 4.49/ U50,488H44 G, 7.2 (mκOR)/ U50,48845 G, 300 (hκOR)/ U50,48845 Arr1, 1.58/ U50,488H46 G, 1.32/ U50,488H46 G, 6/ U50,48833	GloSensor cAMP and Tango assays/Intrinsic reactive activity (RAi)44 (ERK1/2) phosphorylation and p38 MAPK/Intrinsic reactive activity (RAi)45 (ERK1/2) phosphorylation and p38 MAPK/Intrinsic reactive activity (RAi)45 [35S]-GTPγS and βgalactosidase complement assay/ Intrinsic reactive activity (RAi)43 [35S]-GTPγS and βgalactosidase complement assay/ Intrinsic reactive activity (RAi) GloSensor cAMP and Tango assays/ transduction coefficient log(τ/K(A)33
U69,593	↓ Reinstatement of extinguished amphetamine self-administration behavior (0.0 or 0.32 mg/kg, Rats, M) 47	None	None	None	Arr, 2.9 (hκOR)/ Dynorphin A (1–17)36 Arr, 627 (mκOR)/ Dynorphin A (1–17)36	[35S]-GTPγS and κOR receptor internalization/Intrinsic reactive activity (RAi)36 [35S]-GTPγS and κOR receptor internalization/Intrinsic reactive activity (RAi)36
Ethoxymethyl Ether Salvinorin B (EOM SalB)	↑ Remyelination: cuprizone model (0.01 mg/kg, i.p., mice,F) ↓ Neuroinflammation: EAE model of MS (0.03 mg/kg, i.p., mice, F) ↓ Clinical score: EAE model (0.01 mg/kg, i.p., mice,F)48	None	None	None	G, 2.53/ U50,48848	cAMP Accumulation (HitHunterTM) and β-arrestin recruitment (PathHunter)/Intrinsic reactive activity (RAi)48
ICI 204,448	↓ Body temperature (hypothermia) in rats exposed to 5 °C (5 and 10 mg/kg, s.c., rats, M)49	None	None	None	Arr, 2/ Salvinorin A3	GloSensor cAMP and Tango assays/ Transduction coefficient log(τ/K(A)36
ICI-199,441	↓ Nociception (0.05 mg/kg, s.c., mice, F) acetic acid (0.4%) induced abdominal constriction assay 50	Induced bradycardia (0.02 and 0.1 mg/kg s.c., rats,M) Sedation (0.78 mg/kg s.c., mice F) open field test51	None	None	G, 5.7 (hκOR)/ Dynorphin A (1–17) 36 Arr, 10 (mκOR)/ Dynorphin A (1–17)36 Arr, 4/ Salvinorin A3	[35S]-GTPγS binding and KOP receptor internalization/Intrinsic reactive activity (RAi)36 [35S]-GTPγS binding and KOP receptor internalization/Intrinsic reactive activity (RAi)36 GloSensor cAMP and Tango assays/ Transduction coefficient log(τ/K(A)36

¹Lohman, R.-J., Reddy Tupally, K., Kandale, A., Cabot, P. J. & Parekh, H. S. Design and development of novel, short, stable dynorphin-based opioid agonists for safer analgesic therapy. Front. Pharmacol. 14, (2023).

²Vonvoigtlander, P. F. & Lewis, R. A. Analgesic and mechanistic evaluation of spiradoline, a potent kappa opioid. J. Pharmacol. Exp. Ther. 246, 259–262 (1988).

³White, K. L. et al. Identification of novel functionally selective κ-opioid receptor scaffolds. Mol. Pharmacol. 85, 83–90 (2014).

⁴Vonvoigtlander, P. F., Lahti, R. A. & Ludens, J. H. U-50,488: A selective and structurally novel non-Mu (Kappa) opioid agonist. J. Pharmacol. Exp. Ther. 224, 7–12 (1983).

⁵Kuzumaki, N. et al. Effect of κ-opioid receptor agonist on the growth of non-small cell lung cancer (NSCLC) cells. Br. J. Cancer 106, 1148–1152 (2012).

⁶Kamei, J. & Nagase, H. Norbinaltorphimine, a selective κ-opioid receptor antagonist, induces an itch-associated response in mice. Eur. J. Pharmacol. 418, 141–145 (2001).

⁷Silvia, R. C., Slizgi, G. R., Ludens, J. H. & Tang, A. H. Protection from ischemia-induced cerebral edema in the rat by U-50488H, a kappa opioid receptor agonist. Brain Res. 403, 52–57 (1987).

⁸Privette, T. H. & Terrian, D. M. Kappa opioid agonists produce anxiolytic-like behavior on the elevated plus-maze. Psychopharmacology (Berl.) 118, 444–450 (1995).

⁹Du, C. et al. Kappa opioid receptor activation alleviates experimental autoimmune encephalomyelitis and promotes oligodendrocyte-mediated remyelination. Nat. Commun. 7, 1–10 (2016).

¹⁰Song, X., Cui, Z., He, J., Yang, T. & Sun, X. κ-opioid receptor agonist, U50488H, inhibits pyroptosis through NLRP3 via the Ca2+/CaMKII/CREB signaling pathway and improves synaptic plasticity in APP/PS1 mice. Mol. Med. Rep. 24, 529 (2021).

¹¹Suzuki, T., Shiozaki, Y., Masukawa, Y., Misawa, M. & Nagase, H. The Role of Mu- and Kappa-Opioid Receptors in Cocaine-Induced Conditioned Place Preference. Jpn. J. Pharmacol. 58, 435–442 (1992).

¹²Ehrich, J. M. et al. Kappa Opioid Receptor-Induced Aversion Requires p38 MAPK Activation in VTA Dopamine Neurons. J. Neurosci. 35, 12917–12931 (2015).

¹³Paris, J. J., Reilley, K. J. & McLaughlin, J. P. Kappa Opioid Receptor-Mediated Disruption of Novel Object Recognition: Relevance for Psychostimulant Treatment. J. Addict. Res. Ther. 2011 00 01, 1–10 (2011).

¹⁴van de Wetering, R. et al. The Kappa Opioid Receptor Agonist 16-Bromo Salvinorin A Has Anti-Cocaine Effects without Significant Effects on Locomotion, Food Reward, Learning and Memory, or Anxiety and Depressive-like Behaviors. Molecules 28, 4848 (2023).

¹⁵Paton, K., Atigari, D., Kaska, S., Prisinzano, T. E. & Kivell, B. M. Strategies for developing kappa opioid receptor agonists for the treatment of pain with fewer side-effects. J. Pharmacol. Exp. Ther. JPET-MR-2020–000134 (2020) https://doi.org/10.1124/jpet.120.000134.

¹⁶Dunn, A. D., Reed, B., Erazo, J., Ben-Ezra, A. & Kreek, M. J. Signaling Properties of Structurally Diverse Kappa Opioid Receptor Ligands: Toward in Vitro Models of in Vivo Responses. (2019) https://doi.org/10.1021/acschemneuro.9b00195.

¹⁷Gupta, A. et al. Collybolide is a novel biased agonist of κ-opioid receptors with potent antipruritic activity. Proc. Natl. Acad. Sci. U. S. A. 113, 6041–6046 (2016).

¹⁸Bruijnzeel, A. W. Kappa-opioid receptor signaling and brain reward function. Brain Res. Rev. 62, 127–146 (2009).

¹⁹Mysels, D. & Sullivan, M. A. The kappa-opiate receptor impacts the pathophysiology and behavior of substance use. Am. J. Addict. 18, 272–276 (2009).

²⁰Spetea, M. et al. Selective κ receptor partial agonist HS666 produces potent antinociception without inducing aversion after i.c.v. administration in mice. Br. J. Pharmacol. 174, 2444–2456 (2017).

²¹Zhou, L. et al. Development of functionally selective, small molecule agonists at kappa opioid receptors. J. Biol. Chem. 288, 36703–16 (2013).

²²Bohn, L. M. & Aubé, J. Seeking (and Finding) Biased Ligands of the Kappa Opioid Receptor. ACS Med. Chem. Lett. 8, 694–700 (2017).

²³Bedini, A. et al. Functional Selectivity and Antinociceptive Effects of a Novel KOPr Agonist. Front. Pharmacol. 11, 188 (2020).

²⁴Simonson, B. et al. Pharmacology and anti-addiction effects of the novel κ opioid receptor agonist Mesyl Sal B, a potent and long-acting analogue of salvinorin A. Br. J. Pharmacol. 172, 515–531 (2015).

²⁵Kivell, B. M. et al. Kappa Opioid Receptor Agonist Mesyl Sal B Attenuates Behavioral Sensitization to Cocaine with Fewer Aversive Side-Effects than Salvinorin A in Rodents. Mol. Basel Switz. 23, 2602 (2018).

²⁶White, K. L. et al. The G protein-biased κ-opioid receptor agonist RB-64 is analgesic with a unique spectrum of activities in vivo. J. Pharmacol. Exp. Ther. 352, 98–109 (2015).

²⁷Brust, T. F. et al. Biased agonists of the kappa opioid receptor suppress pain and itch without causing sedation or dysphoria. Sci. Signal. 9, ra117 (2016).

²⁸Zamarripa, C. A. et al. The kappa-opioid receptor agonist, triazole 1.1, reduces oxycodone self-administration and enhances oxycodone-induced thermal anti-nociception in male rats. Psychopharmacology (Berl.) 238, 3463–3476 (2021).

²⁹Lovell, K. M. et al. Structure-activity relationship studies of functionally selective kappa opioid receptor agonists that modulate ERK 1/2 phosphorylation while preserving G protein over βarrestin2 signaling bias. ACS Chem. Neurosci. 6, 1411–1419 (2015).

³⁰Aceves, M., Mathai, B. B. & Hook, M. A. Evaluation of the effects of specific opioid receptor agonists in a rodent model of spinal cord injury. Spinal Cord 54, 767–777 (2016).

³¹Ko, M.-C. & Husbands, S. M. Effects of Atypical κ-Opioid Receptor Agonists on Intrathecal Morphine-Induced Itch and Analgesia in Primates. J. Pharmacol. Exp. Ther. 328, 193–200 (2009).

³²Birch, P. J. et al. Neuroprotective actions of GR89696, a highly potent and selective kappa-opioid receptor agonist. Br. J. Pharmacol. 103, 1819–1823 (1991).

³³El Daibani, A. et al. Molecular mechanism of biased signaling at the kappa opioid receptor. Nat. Commun. 2023,141 14, 1–13 (2023).

³⁴Zheng, Z. et al. Structure based discovery of new antagonist and biased agonist chemotypes for the Kappa Opioid Receptor. J. Med. Chem. 60, 3070–3081 (2017).

³⁵Itoh, J., Ukai, M. & Kameyama, T. U-50,488H, a κ-opioid receptor agonist, markedly prevents memory dysfunctions induced by transient cerebral ischemia in mice. Brain Res. 619, 223–228 (1993).

³⁶DiMattio, K. M., Ehlert, F. J. & Liu-Chen, L.-Y. Intrinsic Relative Activities of Opioid Agonists in Activating Gα proteins and Internalizing Receptor: Differences between Human and Mouse Receptors. Eur. J. Pharmacol. 761, 235–244 (2015).

³⁷Zhou, Y. & Kreek, M. J. Combination of Clinically Utilized Kappa-Opioid Receptor Agonist Nalfurafine With Low-Dose Naltrexone Reduces Excessive Alcohol Drinking in Male and Female Mice. Alcohol. Clin. Exp. Res. 43, 1077–1090 (2019).

³⁸Denny, L. et al. Nalfurafine reduces neuroinflammation and drives remyelination in models of CNS demyelinating disease. 10, (2021).

³⁹Ikeda, K. et al. TRK-820, a selective kappa opioid receptor agonist, could effectively ameliorate L-DOPA-induced dyskinesia symptoms in a rat model of Parkinson’s disease. Eur. J. Pharmacol. 620, 42–48 (2009).

⁴⁰Endoh, T. et al. Potent antinociceptive effects of TRK-820, a novel kappa-opioid receptor agonist. Life Sci. 65, 1685–1694 (1999).

⁴¹Liu, S. S. et al. Kappa opioid receptors drive a tonic aversive component of chronic pain. J. Neurosci. 39, 4162–4178 (2019).

⁴²Inui, S. Nalfurafine hydrochloride to treat pruritus: a review. Clin. Cosmet. Investig. Dermatol. 8, 249–255 (2015).

⁴³Kumagai, H. et al. Effect of a novel kappa-receptor agonist, nalfurafine hydrochloride, on severe itch in 337 hemodialysis patients: A Phase III, randomized, double-blind, placebo-controlled study. Nephrol. Dial. Transplant. 25, 1251–1257 (2010).

⁴⁴Cao, D. et al. Comparison of Pharmacological Properties between the Kappa Opioid Receptor Agonist Nalfurafine and 42B, Its 3-Dehydroxy Analogue: Disconnect between in Vitro Agonist Bias and in Vivo Pharmacological Effects. ACS Chem. Neurosci. 11, 3036–3050 (2020).

⁴⁵Schattauer, S. S., Kuhar, J. R., Song, A. & Chavkin, C. Nalfurafine is a G-protein biased agonist having significantly greater bias at the human than rodent form of the kappa opioid receptor. Cell. Signal. 32, 59–65 (2017).

⁴⁶Liu, J. J. et al. Phosphoproteomic approach for agonist-specific signaling in mouse brains: mTOR pathway is involved in κ opioid aversion. Neuropsychopharmacology 44, 939–949 (2019).

⁴⁷Schenk, S. & Partridge, B. Effect of the kappa-opioid receptor agonist, U69593, on reinstatement of extinguished amphetamine self-administration behavior. Pharmacol. Biochem. Behav. 68, 629–634 (2001).

⁴⁸Paton, K. F. et al. The Salvinorin Analogue, Ethoxymethyl Ether Salvinorin B, Promotes Remyelination in Preclinical Models of Multiple Sclerosis. Front. Neurol. 12, 782190 (2021).

⁴⁹Rawls, S. M., Ding, Z., Gray, A. M. & Cowan, A. Peripheral κ-Opioid Agonist, ICI 204448, Evokes Hypothermia in Cold-Exposed Rats. Pharmacology 74, 79–83 (2005).

⁵⁰Barlow, J. J. et al. Structure/activity studies related to 2-(3,4-dichlorophenyl)-N-methyl-N-[2-(1-pyrrolidinyl)-1-substituted- ethyl]acetamides: a novel series of potent and selective kappa-opioid agonists. J. Med. Chem. 34, 3149–3158 (1991).

⁵¹POPOV, S. V. et al. Activation of Peripheral Opioid κ1 Receptor Prevents Cardiac Reperfusion Injury. Physiol. Res. 70, 523–531 (2021).

Agonist binding at κOR recruits inhibitory G-protein heterotrimers (Gαi/oGβGγ), leading to the dissociation of the heterotrimeric G-protein complex upon the exchange of GDP with GTP at the Gα subunit. This separation of the G-protein complex into Gα and GβGγ subunits initiates downstream signaling to kinases, which can be mediated by both β-arrestin–dependent and G-protein–dependent signaling pathways (Bruchas et al., 2006; McLennan et al., 2008; Schmid et al., 2013; Gesty-Palmer et al., 2006; Che and Roth, 2023). The Gαi/o complex inhibits adenylyl cyclase, reducing cAMP-dependent kinase activity. GβGγ activates G-protein-gated inwardly rectifying potassium (GIRK) channels and hyperpolarizes the cells (Yamada et al., 1998; Sadja et al., 2003). κOR also interacts with β-arrestin proteins (Appleyard et al., 1999; McLaughlin et al., 2003), which can promote activation of several kinases, including the mitogen-activated protein kinase (MAPK) p38 (Bruchas et al., 2006), c-Jun N-terminal kinase (JNK) (McDonald et al., 2000), and the extracellular signal–regulated kinases 1 and 2 (ERK1/2) (McLennan et al., 2008). Importantly, arrestin recruitment is mediated by GPCR kinases (GRKs) mediated receptor phosphorylation, which can be G-protein-independent (Che and Roth, 2023). Early research showed that G-protein signaling played a crucial role in mediating the antinociceptive and anti-pruritic effects of κOR agonists, as these therapeutic responses remain in mice lacking β-arrestin 2 (Morgenweck et al., 2015a). In recent years, there has been an increasing interest in selective biased agonists for κOR activation via the G-protein pathway rather than β-arrestin, not only for its analgesic and antipruritic effects effect, but as a safer alternative to μOR agonists (Mores et al., 2019; Dunn et al., 2019; Che and Roth, 2021).

Here, we discuss the current status of biased signaling at κOR, highlighting its potential for yielding novel safe and effective therapeutics for pain and pruritis. We also present an overview of several examples of G-protein-biased κOR agonists with promising therapeutic potential, comparing their levels of biased signaling and preclinical effects. Additionally, we highlight variability in biased factor values within the same agonists as well as other model limitations that hinder the development of many ligands. Finally, we provide some insight to help improve the correlation between the degree of signaling bias and specific behavior responses focusing on itch and pain. In addition to these measures, we also consider motor impairment, antinociception, and affective state. These factors can enhance the interpretability of pain data but also highlight the complexity of targeting the κOR and the potential side effects associated with their activation. The selection criteria for this review are limited primarily to mice and human studies, as not all κOR-biased agonists were tested in non-human primates. However, we include the extensive studies on nalfurafine and triazole 1.1 tested on non-human primates for pruritus. For a more comprehensive overview of the studies looking into non-human primates and kappa-biased agonists, please refer to work from Ko et al. and colleagues (Ko and Husbands, 2022).

2. Approaches for evaluating signaling bias at κOR

Functional selectivity or signaling bias has been defined as the capability of agonists to activate a selective subset of receptor signaling pathways (e.g., Gαi1 and β-arrestin 2). The bias factor is usually used to quantify the degree of ligand bias toward one signal. Several approaches have been developed to assess bias factors for full agonists or partial agonists by analyzing derived data from various in-vitro-based assays (Smith et al., 2018). Briefly, two different in-vitro based assays for two signaling pathways (G-proteins and β-arrestins) are performed for the ligand of interest, and a high-efficacy balanced agonist (for example, U50,488, U69,593, dynorphin, and salvinorin A for κOR), is usually included as a reference ligand. The obtained data are analyzed using either the operational model/transduction coefficient (Kenakin et al., 2012; Black and Leff, 1983) or the intrinsic reactive activity (RAi) method (Fig. 1) (Kenakin and Christopoulos, 2013; Black et al., 1985; Ehlert, 2005).

Fig. 1.

A systematic approach to select the appropriate model to quantify signaling bias for κOR agonists. Adapted from (Kenakin and Christopoulos, 2013). KA is the functional dissociation constant for the agonist; τ is the efficacy of the agonist in the given pathway; pEC₅₀ is the negative logarithm of the EC₅₀; and RA is the relative activity.

Typically, calculating the transduction coefficients represented by Δlog (τ/KA) values involves fitting the dose–response curve for both the reference ligand and the ligand of interest to the Black and Leff operational model. The log (τ/KA) for each signaling pathway that represents the affinity and efficacy of the ligand of interest are calculated and then compared to the (log(τ/KA) for the reference ligand. If an agonist exhibits bias compared to the reference agonist, then the value of antilog ΔΔlog(τ/KA) will deviate significantly from zero. This model integrates receptor density and coupling within a system, making it independent of receptor expression levels; however, weak partial agonists are difficult to analyze due to their ambiguous fitting to this model. To enhance accuracy of bias factor quantification for low efficacy agonists, such as weak partial agonists, it is recommended to increase the number of data points when creating dose-response curves using the “competitive” model (Stahl et al., 2015; Dunn et al., 2018). In this model, which was derived from “standard” operational model, the partial agonist is tested alongside a non-saturating concentration of the reference compound. For the intrinsic reactive activity (RAi) method Δ(Emax/EC₅₀), the log (Emax/EC₅₀) values for G-protein or β-arrestin signaling of ligand of interest are calculated and then compared to the log (Emax/EC₅₀) for the reference ligand. If an agonist demonstrates bias relative to the reference agonist, then the antilog of ΔΔlog (Emax/EC₅₀) will notably differ from zero. This model necessitates that the ligand dose-response curves display an ideal hill-slope.

In addition to each model’s limitations, it’s crucial to acknowledge the constraints when comparing multiple bias factors for the same κOR ligand (summarized in Table 1): 1) Inconsistent choice of reference ligand for quantifying the signaling bias for κOR ligands such as U50,488 and U69,593 (synthetic κOR selective agonists), dynorphin (an endogenous κOR opioid), and salvinorin A (natural occurring opioid); 2) Different cell culture based assays using overexpressed κOR in cell culture model systems may result in different expression levels, thus impacting the measuring of G-protein or β-arrestin signaling and bias factor values accordingly; 3) The predictive translational value of the method is not yet clearly established, given that these models do not consider ligand binding kinetics and the temporal dynamics of signaling processes, which can significantly impact the observed bias (Klein Herenbrink et al., 2016). Additionally, bias factors measured using κOR-expressing cell lines may not accurately correlate to the physiological conditions. A reasonable approach to tackle this latter issue could be achieved by characterizing κOR ligand properties in physiologically relevant cell types using the same neuronal cell lines or freshly isolated neurons that closely resemble the clinically relevant in-vivo site to measure both G-protein and β-arrestin signaling for a rigorous and robust assessment of ligand bias. It’s important to note that the expression levels of κOR vary across different brain regions in various species. Therefore, it is necessary to test κOR biased agonists in both mice and non-human primates to obtain more reliable translational results. For example, in-vitro work in HEK293-T cells transfected with human κOR and rodent κOR using the agonists pentazocine and butorphanol showed differences in p38 activation in downstream signaling, highlighting the differential ligand-directed downstream signaling between human and rodent κOR (Schattauer et al., 2012). Preclinical testing in mice helps with behavioral characterization of κOR in the specific brain regions, while non-human primates can improve translation of the biased κOR agonists, as their κOR expression is more similar to that of humans.

3. Representative examples of G-protein-biased κOR agonists

Nalfurafine.

Nalfurafine, also known as Remitch^®, is the first selective κOR full agonist clinically approved in Japan and South Korea to treat uremic pruritus and chronic pruritus in hemodialysis patients without producing dysphoria and hallucination at therapeutic doses; however, it has neither been approved in Europe nor tested in clinical trials in the U.S.A (Inui, 2015; Nagase et al., 1998). Although the same reference ligand, U50,488, is used to quantify bias, the extent of bias of this compound is questionable due to variations in cellular background, signaling assays, and the analytical approaches used to evaluate the degree of bias (i.e., operational model/transduction coefficient or the intrinsic reactive activity (RAi) method) (Schattauer et al., 2017; El Daibani et al., 2023; Liu et al., 2019b; Cao et al., 2020). Nalfurafine was reported to serve as a G-protein-biased agonist for both rat κOR and human κOR with a bias factor of 7 and 300, respectively, relative to U50, 488 by measuring the G-protein and β-arrestin signaling (Schattauer et al., 2017). This was done by using the early phase of Extracellular Signal-Regulated Kinase (ERK1/2) phosphorylation and p38 MAPK signaling, respectively. The obtained data were analyzed using the intrinsic reactive activity (RAi) method. However, when employing the same analytical approach, neither G-protein-dependent nor arrestin-dependent functional selectivity was observed in-vitro using [³⁵S]GTPγS binding in N2A-FmK6H cells and β-arrestin1 and 2 recruitment in HEK293 cells (Liu et al., 2019b). Another study reported that nalfurafine acts as a G-protein-biased κOR agonist with a bias factor of 4.49, as determined by conducting a GloSensor cAMP assay on HEK293-T cells stably expressing the human κOR, and a Tango assay on HEK293 cells stably expressing a tTA-dependent luciferase (Cao et al., 2020). Recently, another study reported a bias factor of 6 for human κOR relative to U50,488, using cAMP inhibition and Tango β-arrestin recruitment assays on HEK293-T cells to measure G-protein and β-arrestin signaling, respectively, by employing the transduction coefficient model (El Daibani et al., 2023). The discrepancies in nalfurafine’s bias factor can be attributed to differences in the assay methods for measuring G-protein activation and β-arrestin recruitment, cell lines, species origin of the receptor, levels of receptor expression, and the overlooking of ligand binding kinetics as well as the time-dependent dynamics of signaling processes. In the later study mentioned above, the crystal structure of the human κOR in complex with nalfurafine and an active state stabilizing nanobody Nb39 was determined at a 3.3 Å resolution, highlighting the locations that are responsible for the biased signaling mechanism at κOR. For example, the specific interactions with the furan ring of nalfurafine have been shown to be essential for its agonist activity. Furthermore, the occluded state was frequently observed with nalfurafine bound compared to the balanced agonist U50, 488 or the β-arrestin-biased WMS-X600 bound using molecular dynamic simulation, suggesting that ligands can achieve downstream biased signaling by promoting differential receptor conformational states and a nalfurafine-specific occluded state may favor the G-protein coupling and subsequent G-protein bias. The extensive detailed comparison of the signaling profile using the TRUPATH assay revealed differential transducer coupling preferences for nalfurafine to different Gα protein subtypes (Gi1, Gi2, Gi3, GoA, GoB, Gz, and Gustducin) and two β-arrestins (β-arrestin 1 and β-arrestin 2). Interestingly, nalfurafine has been shown to have the highest potency at Gz rather than Gi or Go subtypes, indicating that signaling preference occurs not only toward the G-protein signaling over the β-arrestin pathway (G-protein vs. β-arrestin) but also within G-protein subtypes signaling (G-protein vs. G-protein) (Han et al., 2023).

Preclinical studies assessing behavior in male CD-1 mice found that nalfurafine reduced 5′- GNTI and 48/80 compound-induced scratching behavior, as well as formalin-induced pain in a dose dependent manner without causing aversion (conditioned place aversion (CPA)), sedation (locomotor activities), motor incoordination, and anhedonia (intracranial self-stimulation) (Schattauer et al., 2017; Liu et al., 2019b). Interestingly, nalfurafine did not cause CPA at the therapeutic dose or ≤20 μg/kg dose which maintains effective analgesic and anti-pruritic effect (Liu et al., 2019b). However, at higher doses (40 μg/kg or higher), nalfurafine does cause CPA (though maintains effective analgesic and anti-pruritic effects), suggesting that aversion induced by nalfurafine is dose-dependent and may involve other targets or signaling pathways in rodents (Liu et al., 2019b; Cao et al., 2020; Chen et al., 2022). In non-human primates, systemic administration of nalfurafine (0.1–1 μg/kg) resulted in a dose-dependent decrease in morphine-induced scratching (Wakasa et al., 2004; Ko and Husbands, 2009). Nalfurafine also blocked oxycodone-induced scratching but caused sedation and motor impairment in non-human primates studies (Huskinson et al., 2022). Importantly, a single dose and repeated doses of nalfurafine administration, along with combinational administration of sub-threshold doses of nalfurafine and naltrexone decreased alcohol consumption in male and female C57BL/6 mice, supporting the potential role of κOR agonist for the treatment of alcoholism (Zhou and Kreek, 2019). Recent studies have revealed that nalfurafine enhances neuronal recovery and remyelination and reduces CNS immune-cell infiltration (CD4⁺ and CD8⁺ T cells) in experimental allergic autoimmune encephalomyelitis mouse model (Denny et al., 2021). These findings indicate the promising role of κOR agonists in the treatment of chronic inflammatory demyelinating diseases, such as multiple sclerosis.

RB-64.

RB-64 is a salvinorin A-derived ligand that was developed by molecular modeling studies. It was reported to act as a G-protein-biased agonist for κOR, showing a bias factor of 35 (human κOR) and 96 (mouse κOR) for G-protein signaling compared to salvinorin A using cAMP inhibition and Tango β-arrestin recruitment assays to measure the G-protein and β-arrestin signaling, respectively (White et al., 2014, 2015; Kroeze et al., 2015). It was shown to produce antinociception in the hot-plate assay to a similar degree as U69,593 and salvinorin A without causing locomotor incoordination, which was observed with the other two ligands, in the rotarod assay in both male and female C57BL/6 mice (White et al., 2015). However, the thiocyanate functional group in RB-64 (22-thiocyanatosalvinorin A) is proposed to form an irreversible covalent bond with κOR at C315^7.38(Chen et al., 2022), limiting its therapeutic potential (Yan et al., 2009). Studies of RB-64 also showed that the analgesic of κOR agonists are retained in β-arrestin 2 knockout mice, suggesting that these beneficial effects may not require β-arrestin (Wakasa et al., 2004). Several studies indicate that activation of κOR causes aversion through GRK-3/β-arrestin 2 recruitment and subsequent p38 phosphorylation (Bruchas et al., 2007; Bruchas and Chavkin, 2010; Ehrich et al., 2015). Contrary to the hypothesis that κOR-induced aversion is likely through β-arrestin 2, RB-64, salvinorin A, and U69,59 induced CPA in both wild-type or β-arrestin 2 knockout mice (White et al., 2015), suggesting that aversive effects may not be mediated by β-arrestin 2. RB-64 did not impair rotarod performance or inhibit novelty-induced hyperlocomotion in both wild-type and β-arrestin 2 knockout mice, suggesting that these effects may not require β-arrestin (White et al., 2015). Importantly, RB-64 caused aversion in both wild-type and β-arrestin 2 knockout mice (White et al., 2015), suggesting that aversion is not mediated by arrestin pathway or the arrestin pathway is compensated by the G-protein biased signaling mechanism in the β-arrestin 2 knockout mice. However, the sedative effects associated with salvinorin A and U69,593 may result in a misinterpretation of their aversive effects. This difference raises questions about our understanding of the underlying physiology and the utility of global knockout mice when attempting to identify a particular signaling pathway with a specific behavior.

Triazole 1.1.

A high throughput screening (HTS) campaign-based strategy led to the discovery of novel κOR agonists, such as triazole 1.1 (Frankowski et al., 2012). κOR activation selectivity was confirmed by conducting two assays, DiscoveRx β-arrestin PathHunter and an imaging based β-arrestin translocation assay. Triazole 1.1 was reported as a full G-protein-biased full agonist for κOR with a bias factor of 61 relative to U69,593 by measuring [³⁵S] GTPγS binding and κOR -β-arrestin 2 enzyme fragment complementation assays, while a bias factor of 20 was measured by [³⁵S] GTPγS binding and imaging of β-arrestin 2-GFP translocation to κOR assays (Zhou et al., 2013). Consistent with these results, triazole 1.1 was identified as a potent G-protein-biased agonist for κOR, exhibiting bias factors of 47 and 93 compared to U69,593. This was determined through G-protein coupling and β-arrestin 2 recruitment, as well as G-protein coupling and ERK 1/2 phosphorylation (Lovell et al., 2015).

Preclinical investigations examining behavior in C57BL/6 mice revealed that triazole 1.1 induced an antinociceptive effect in tail flick assay and antipruritic effects by suppressing chloroquine phosphate–induced scratching typically associated with conventional κOR agonists (Brust et al., 2016a). Importantly, at antinociceptive and anti-pruritic doses, triazole 1.1 (up to 24 mg/kg) did not induce dysphoria in intracranial self-stimulation assay. Furthermore, triazole 1.1 does not cause any sedation or reduction of brain dopamine in the nucleus accumbens (Brust et al., 2016a). In non-human primates, administration of triazole 1.1 resulted in reduced sedation and motor impairment compared to typical κOR agonists (Huskinson et al., 2020, 2022). Importantly, a comparative study on triazole 1.1 between cell lines and primary neurons (e.g., striatal neurons) demonstrates that G-protein-biased agonism at κOR is maintained in the neuronal environment, albeit with different degrees (Ho et al., 2018).

HS665 and HS666.

The first series of diphenethylamine derivatives, HS665 and HS666, were introduced as selective κOR ligands by chemical modification on Dopamine D2 receptor agonist (RU 24213) that exhibited moderate affinity to κOR, functioning primarily as a κOR antagonist (Fortin et al., 1991; Spetea et al., 2012). HS665 (full κOR agonist) and HS666 (partial κOR agonist) were reported to serve as selective G-protein-biased agonists for κOR with bias factors of 389 and 62, respectively, relative to U69,593 using the [³⁵S]-GTPγS binding and DiscoveRx PathHunter β-arrestin 2 assays (Spetea et al., 2017).

In male CD-1 mice, subcutaneous administration of HS665 produced antinociceptive effect comparable with U50,488 in a dose-dependent manner in the acetic acid-induced writhing test (Spetea et al., 2012). Further preclinical investigations evaluated behavioral antinociception, locomotor activity, and place preference of HS665 and HS666 after intracerebroventricular administration in C57BL/6J mice (Spetea et al., 2017). In a model of acute thermal nociception (warm water tail withdrawal assay), HS665 and HS666 exhibited dose-dependent potent analgesic effects with rapid onset in wild-type and μOR-knockout mice. Confirming the selectivity of these effects to κOR, no antinociception was observed in κOR-knockout mice. Neither HS665 nor HS666 significantly affected motor performance in rotarod assay. Additionally, HS666 did not induce CPP/CPA (conditioned place preference/conditioned place aversion). HS666 exhibited neither significant preference (CPP) nor aversion (CPA) at 25 times its analgesic dose ED₅₀, whereas HS665 produced significant place aversion when administered at 5 times its analgesic ED₉₀. Whether HS666’s lack of CPA was due to acting as a partial agonist, or its G-protein bias remains unclear. Expanding the structure–activity relationships on HS665 and HS666 by presenting chemical modifications led to generation of new diphenethylamines (Erli et al., 2017). Several analogs demonstrated high affinity and remarkable selectivity for κOR, functioning as either full or partial agonists. These new diphenethylamines exhibited enhanced antinociceptive effects with greater potency than U50,488, HS665, and HS666 in the writhing assay with a reduced liability profile, as evidenced by the absence of sedation and motor impairment in male CD1 mice. Although the functional activities for these new diphenethylamines were assessed by [³⁵S]GTPγS binding assay, β-arrestin 2 recruitment was not measured for any analogs in this series.

LOR17.

LOR17 (c[Phe-Gly-(b-Ala)-D- Trp]), a novel selective peptidic κOR agonist, was reported to act as a G-protein-biased agonist with a bias factor of 853 relative to U50,488 (Bedini et al., 2020). This was done by measuring the cAMP and PathHunter β-arrestin 2 recruitment assays and applying the transduction coefficient model. Preclinical behavioral assays have demonstrated that LOR17 effectively alleviates both pain in warm-water tail-withdrawal, as well as acetic acid-induced writhing tests and thermal hypersensitivity in a mouse model of oxaliplatin-induced neuropathic pain (Bedini et al., 2020). Additionally, LOR17 neither impacts motor coordination (rotarod test), locomotor and exploratory activities (hole-board test) nor induces depression-like behavior (forced swimming test) (Bedini et al., 2020). These findings make LOR17 a strong candidate for further development and clinical testing; however, comprehensive studies are needed to fully understand its pharmacological properties and confirm its therapeutic potential in humans.

16-Bromo salvinorin A.

16-Bromo salvinorin A is a G-protein-biased analogue of salvinorin A, a naturally occurring hallucinogen found in the leaves of Salvia divinorum, with a modified furan ring at the C-16 position (Riley et al., 2014). The reported bias factor for 16-bromo salvinorin A was 7.7 relative to U50,488, as determined by the cAMP Accumulation (HitHunterTM) assay in CHO cells and the β-arrestin recruitment (PathHunter) assay in U2OS cells, using intrinsic reactive activity (RAi) model (Paton et al., 2020). Administration of 1.0 mg/kg i.p. of 16-bromo salvinorin A effectively attenuated drug-seeking behavior in an animal model of drug-relapse in male Sprague–Dawley rats without causing sedation (Riley et al., 2014; van de Wetering et al., 2023); however, another study measuring motor coordination revealed that 16-bromo salvinorin A had sedative effects for up to 30 min using an accelerating procedure (Paton et al., 2020). This finding raised a question about whether the reduction in drug-primed responses is directly related to 16-bromo salvinorin A’s anticocaine effect or its sedative effect. In the warm water tail-withdrawal assay, 16-bromo salvinorin A demonstrated dose-dependent antinociceptive effects compared to U50, 488 (Paton et al., 2020; van de Wetering et al., 2023).When administered at a dose of 2 mg/kg, 16-bromo salvinorin A showed significant antinociceptive effects from 10 to 60 min, indicating a slower onset but longer duration of action compared to salvinorin A. This finding suggests an improved pharmacokinetic profile, likely due to structural modifications designed to reduce metabolism by phase I metabolizing enzymes, cytochrome P450 (Paton et al., 2020). In non-human primates, administration of 16-bromo salvinorin A at doses that did not cause strong sedative effects resulted in a dose-dependent elevation of prolactin, a neuroendocrine biomarker, similar to other κOR agonists (van de Wetering et al., 2023). Unlike salvinorin A, 16-bromo salvinorin A did not increase anxiety-related behaviors, aligning with the hypothesis that G-protein-biased agonists may have fewer side effects (Paton et al., 2020).

Mesyl salvinorin B.

Mesyl salvinorin B, a semi-synthetic analogue of salvinorin A, exhibits a G-protein-bias at the κOR with a bias factor of 0.61, using U50,488 as the reference ligand (Kivell et al., 2018). This bias was determined through cAMP Accumulation (HitHunter^™) assay in CHO cells and β-arrestin recruitment (PathHunter^™) assay in U2OS cells, using Intrinsic reactive activity (RAi) model. Preclinical studies were conducted to evaluate the potential therapeutic effects of mesyl salvinorin B (antinociceptive and anti-cocaine effects) and its possible side effects that are associate with κOR activation (anhedonia, sedation, aversion, depression, anxiety, and cognitive impairments) (Kivell et al., 2018; Simonson et al., 2015). In male Sprague Dawley rats, administering 0.3 mg/kg of mesyl salvinorin B 45 min prior to cocaine decreased cocaine-induced hyperactivity and sensitization without affecting sucrose self-administration or causing aversion, anxiety, or cognitive impairments, but it increased immobility in the forced swim test, suggesting pro-depressive effects (Kivell et al., 2018). In male C57BL6. SJL mice, the antinociceptive effects of mesyl salvinorin B were found to be less potent than salvinorin A using warm-water tail withdrawal and formaldehyde-induced inflammatory pain assays; however, mesyl salvinorin B did not cause the motor impairment observed with salvinorin A as assessed by the rotarod motor coordination test (Kivell et al., 2018). Additionally, mesyl salvinorin B, alone or in combination with naltrexone, was found to reduce and synergistically reduce excessive alcohol intake, respectively, in mice with no sex differences observed, suggesting potential in alcoholism treatment models (Zhou et al., 2017, 2018).

4. Expression and distribution of κOR

There are increasing numbers of G-protein-biased κOR agonists with promising side effect profiles in preclinical studies that are important for moving κOR agonists forward to clinical trials, regardless of their in-vitro agonist bias; however, it remains difficult to attribute the therapeutic potential solely to G-protein bias as opposed to other pharmacological properties. An important factor is the location of κOR in the central and/or peripheral nervous system. The distribution of κOR has been thoroughly investigated through various biochemical techniques, such as receptor autoradiography (Mansour et al., 1988), immunohistochemistry (IHC) (Mansour et al., 1996), and reporter protein (Cai et al., 2016; Chen et al., 2020) approaches in rodents. A detailed description is included in Fig. 2, which illustrates the different levels of κOR expression and their known function. κORs are majorly expressed in the ventral tegmental area, anterior cingulate cortex, endopiriform nucleus, hypothalamic nuclei, claustrum, amygdala nuclei, bed nucleus of stria terminalis, raphe nucleus, and nucleus accumbens, which are known to mediate negative affect, reward, motivation, pain, and addiction behaviors (Al-Hasani et al., 2015, 2017; Liu et al., 2019a; Massaly et al., 2019). Moderate levels of κOR are observed in the periaqueductal gray, parabrachial nucleus, thalamus, primary and secondary somatosensory cortices are known to be involved in pain processing and negative affect mediated by pain (Nguyen et al., 2022). κORs found in paraventricular nucleus and nucleus reunion of thalamus are known to suppress reward seeking behavior (Vollmer et al., 2022). Spinally expressed κORs are known to mitigate itch (Kardon et al., 2014; Inan et al., 2021). Peripherally, κORs are expressed on the dorsal root ganglia, mediating cold hypersensitivity and pain (Jamshidi et al., 2015; Ji et al., 1995; Madasu et al., 2021; Snyder et al., 2018); the trigeminal ganglia, mediating chronic pain (Jamshidi et al., 2015; Ji et al., 1995; Turnes et al., 2022); the gastrointestinal tract, ameliorating chronic abdominal pain (Holzer, 2009; Brust et al., 2016b); the respiratory system, mitigating pulmonary inflammation (Peng et al., 2012; Zeng et al., 2020); the skin, mitigating pruritus (Brust et al., 2016a; Snyder et al., 2018); and the heart, decreasing arrhythmias (Sobanski et al., 2014; Pugsley et al., 1998). Though these properties have made κOR an attractive target for drug development in the treatment of pain, κOR agonists also cause sedation, dysphoria, and hallucinations (Paton et al., 2020) (refer to Table 1). To date, it has not been possible to develop a drug targeting a designated location of the brain. In Table 1, we summarize the biased agonists which are either G-protein dependent or β-arrestin dependent along with the pros and cons preclinically.

Fig. 2.

Mice κOR expression adapted from Le Merrer et al., 2009) (Le Merrer et al., 2009). ARC arcuate nucleus, hypothalamus, AMY amygdala, BLA basolateral nucleus, amygdala (Knoll et al., 2011), BNST bed nucleus of the stria terminalis (Haun et al., 2020), CeA central nucleus, amygdala (Knoll et al., 2011; Haun et al., 2022; Yakhnitsa et al., 2022; Navratilova et al., 2019), CI claustrum, CL centrolateral thalamus, CM centromedial thalamus, COA cortical nucleus, amygdala, CPU caudate putamen, DMH dorsomedial hypothalamus, DMR dorsal and medial raphe, DR dorsal raphe (Land et al., 2009; Wright et al., 2018), DTN dorsal tegmental nucleus, EN endopiriform cortex, GP globus pallidus, HPC hippocampus (Shirayama et al., 2004), IC inferior colliculus, ICX insular cortex IP interpeduncular nucleus, LC locus coeruleus (Al-Hasani et al., 2013), LH lateral hypothalamus, ME median eminence, MeA median nucleus, amygdala, NAc nucleus accumbens (Al-Hasani et al., 2015; Massaly et al., 2019; Lorente et al., 2024; Coleman et al., 2021; Zhang et al., 2023), NST nucleus tractus solitarius, pituitary, NRGC nucleus reticularis gigantocellularis, OB olfactory bulb, OCX occipital cortex, PAG periaqueductal gray (Tejeda et al., 2013, 2015; Carr et al., 2010), PBN parabrachial nucleus, PCX parietal cortex, PNR pontine reticular, POA preoptic area (Cone et al., 2023), PV paraventricular thalamus, PVN paraventricular hypothalamus, RE reuniens thalamus, RM raphe magnus (Nguyen et al., 2022; Abraham et al., 2018), S septum, SC superior colliculus, SN substantia nigra, STN spinal trigeminal nucleus, TCX temporal cortex, Th thalamus, TU olfactory tubercle, VMH ventromedial hypothalamus, VP ventral pallidum, VR ventral raphe, VTA ventral tegmental area (Robble et al., 2020), Zi zona incerta. Human κOR expression adapted from (Massaly et al., 2019; Liu et al., 2019b; Peng et al., 2012; Hiller and Fan, 1996; Simonin et al., 1995).

In humans, κORs are expressed moderately in the substantia nigra, hippocampus, cerebral cortex, putamen, caudate nucleus, and nucleus accumbens (Massaly et al., 2019; Liu et al., 2019b; Peng et al., 2012; Hiller and Fan, 1996; Simonin et al., 1995) (Fig. 2). Additionally, they are sparsely distributed in the cerebellum, temporal lobe, and spinal cord (Peng et al., 2012). Human κORs are abundant in the cerebellum, deep layers (V and VI) of the cortex, and striosomes of the striatum, whereas, in contrast, rodents have a patchy distribution (Mansour et al., 1988; Quirion et al., 1987). Studies have emphasized the moderate expression of κOR in the periphery in the dorsal root ganglia (Peng et al., 2012; Moy et al., 2020), skin (Salemi et al., 2005) and sparsely in the lung, spleen, kidney, heart, skeletal muscle, liver, thymus, small intestine pancreas, and adrenal gland (Peng et al., 2012). κOR is differentially reported across species; human κOR expression is more similar to that in non-human primates than in rodent (Mansour et al., 1988; Quirion et al., 1987). In non-human primates, κOR is expressed highly in the limbic cortex, ventral striatum, caudate, substantia nigra putamen, claustrum, amygdala, hypothalamus, and globus pallidus. The expression of κOR in non-human primates was found to be similar to that in humans, suggesting a better translational model to test the ligands in the non-human primates. Overall, the differential κOR expression across species and neural substrate levels has posed translational limitations in the development of biased κOR agonists clinically.

5. Non-homogenous distribution of signal transducers

The limitation of translating κOR -biased ligands from in-vitro to in-vivo can be also attributed to the differential expression of κOR and its signal transducers, such as G-proteins and β-arrestins. Here, we focus on representative GNAi1 (Gαi1), ARRB2 (β-arrestin 2), as well as the non classical GNAZ (Gαz) in different brain regions and species. For example, data from the Human Proteome Atlas (Table 2) differentiates expression levels of κOR and downstream signaling messenger proteins β-arrestin 2, GNAi1, and GNAZ in different brain regions across humans and rodents (Sjöstedt et al., 2020). In humans, κOR and GNAZ are abundantly expressed in the cerebral cortex, β-arrestin 2 in the hypothalamus, and GNAi1 in the medulla oblongata (Sjöstedt et al., 2020). In mice, κOR is highly expressed in the midbrain (Table 2), but the downstream signaling proteins β-arrestin 2, GNAi1, and GNAZ are located in the cerebellum, basal ganglia, and cerebral cortex, respectively (Sjöstedt et al., 2020). In rodents, concomitant activation of the κOR and the presence of downstream signaling proteins together are known to mediate alcohol withdrawal (French et al., 2022; Robins et al., 2018) and itch (Ho et al., 2018; Morgenweck et al., 2015b) responses. In humans, β-arrestin 2 protein is expressed in κOR positive neurons within the synaptic function cluster, suggesting that both proteins cross-talk in neurons mediating synaptic function (Sjöstedt et al., 2020). Additionally, β-arrestin 2 pathway, along with p38 activation, is more known to mediate aversion (Bruchas and Chavkin, 2010). The finding that κOR agonists produce CPA in β-arrestin 2 knockout mice challenges the therapeutic potential of G-protein-biased κOR agonists (Liu et al., 2018, 2019b). Phosphoproteomic studies of the striatum following exposure to several κOR agonists revealed that the G-protein-biased agonists 6′-GNTI and HS666 did not activate mTOR signaling, unlike U50,488 (Liu et al., 2018). Inhibition of mTOR abolished CPA induced by U50,488, suggesting that mTOR signaling causes CPA. Additionally, 6′-GNTI and HS666 showed unique protein phosphorylation patterns compared to β-arrestin 2 recruiters. This indicates that G-protein-biased κOR agonists might offer a therapeutic window for pain treatment with fewer side effects (Liu et al., 2018). Interestingly, nalfurafine, a G-protein-biased κOR agonist, neither activates mTOR nor induces CPA (Liu et al., 2019b). Activation of mTOR did not explain κOR agonist-induced locomotor inhibition or affect the antipruritic and antinociceptive effects of U50,488. Therefore, the precise role of β-arrestin 2 in κOR agonist-mediated aversion remains unclear and the therapeutic versus side-effect profiles of G-protein-biased κOR agonists have yet to be determined. Recently, a mutant mouse line was generated in which all four residues in the C-terminal domain (S356, T357, T363, and S369) responsible for mouse κOR phosphorylation were mutated to alanine (Huang et al., 2022). This revealed that such mutations, with specific effects on κOR-mediated behaviors, unlike β-arrestin 2 knockout (KO) which affects many GPCRs, abolished κOR agonist-induced phosphorylation and subsequent β-arrestins-mediated signaling (Huang et al., 2022). The findings show that agonist-induced κOR phosphorylation affects tolerance and CPA in a sex-dependent manner without altering acute anti-pruritic and hypo-locomotor effects, highlighting the first sex differences in GPCR phosphorylation’s impact on GPCR-mediated behaviors (Huang et al., 2022).

Table 2.

Expression of κOR, GNAZ, GNAi1, and β-arrestin 2 in human and rodent brain. RNA seq data set values expressed as normalized transcripts per million (nTPM)in different brain regions were acquired from human protein atlas OPRK1 (κOR), GNAZ (Gαz), GNAi1 (Gαi1),and ARRB2 (β-arrestin 2) data available from v23.proteinatlas.org (Sjöstedt et al., 2020). (The expression of GNAZ, GNAi1, and β-arrestin 2 in non-human primate brain is not studied yet).

Neural substrate	Human				Rodent
	OPRK1	ARRB2	GNAi1	GNAZ	OPRK1	ARRB2	GNAi1	GNAZ
amygdala	18.9	58.1	21.5	87.1	6.3	30.6	115.6	37.1
basal ganglia	16	56.6	41	71.5	13.2	22.1	205.2	24.5
cerebellum	5.9	45.3	33.9	67.5	0.3	30.6	151	9.2
cerebral cortex	34.4	73.4	30.4	70.2	4.5	26	148.9	39.3
choroid plexus	0.7	48.2	11.9	71	–	–	–	–
hippocampus	4.2	78.7	25.4	65.1	1.9	31	139.3	32.2
hypothalamus	15.8	86.3	30.1	75.9	14	25.7	110.5	29.1
medulla oblongata	16.1	75	42.2	79.7	4.4	27.7	93.6	14.3
midbrain	6.7	68.9	33.8	73.4	7	27.6	105.9	18.4
pons	14.2	71	4.1	64.1	4.1	27.7	93.6	14.3
spinal cord	3.6	62.3	35.3	74.2	–	–	–	–
thalamus	18.4	70.4	32.4	65.2	5.4	25	101.1	18.1
white matter	9.2	85.5	76.9	66.1	2.1	18.3	196	14

Overall, factors such as differences in κOR expression among different species, κOR expression in different neural substrates, and the availability of κOR-mediated intracellular signaling proteins can contribute to κOR-mediated effects. Additionally, inadequate testing in the in-vitro assays, such as efficacy testing of biased agonists alongside full agonists and antagonists, and the lack of critical transducers, resulted in an incomplete understanding of the mechanism of action. These factors may also limit the translation of κOR biased ligands from preclinical to clinical studies. Therefore, it is important to consider these environmental factors when developing a biased ligand to target a specific behavioral phenotype. Due to lack of these factors of biased agonist testing, only nalfurafine (Remitch^®), a G-protein-biased agonist, is currently used for treating pruritus in hemodialysis patients. The differential expression of κOR and its signal transducers across different brain regions and species poses a challenge in translating κOR-biased ligands from in-vitro to in-vivo. Further research is required to determine the exact mechanism of action of κOR-biased ligands and their downstream signaling pathways to develop effective therapies for various disorders.

6. Limitations of preclinical studies for κOR biased ligands

Preclinical studies on rodent models enable the initial assessment and behavioral characterization of κOR-biased ligands; however, their limitations highlight the need for cautious interpretation of results. Despite the similarity in homology between human and rodent κOR, the expression levels of κOR vary across different brain regions, and there are key differences in the amino acid residues involved in GRK/arrestin signaling (Schattauer et al., 2012). For instance, human κOR has a tyrosine residue at position 369 instead of a serine in mouse κOR, making it not a GRK substrate compared to rodent κOR. Moreover, the residue responsible for arrestin recruitment and desensitization of human κOR is Ser-358, while in the rodent κOR, the corresponding position is an asparagine and not a GRK substrate (Schattauer et al., 2012). These findings suggest variations in biased signaling between humans and rodents and emphasize the importance of testing κOR-biased ligands in more reliable translational models such as non-human primates. However, the expression of κOR and its transducers, such as GNAZ, GNAi1, and β-arrestin 2 in the non-human primate’s brain remains unexplored. This underscores a significant gap in our understanding of downstream signaling processes, particularly when comparing non-human primate to human and rodent studies. Besides these limitations, in-vivo studies testing κOR -biased agonists often use different unbiased agonists for comparison. For example in rodent studies, triazole 1.1 mediated antinociception and suppressed chloroquine phosphate–induced scratching responses were compared to U50,488H (Brust et al., 2016a). In the case of nalfurafine mitigating the histamine-induced itch behavior, there was no comparison of nalfurafine’s antipruritic effect to that of U50,488 or salvinorin A (Togashi et al., 2002). Additionally, in a non-human primates experiment, the antipruritic effects of nalfurafine were studied alongside U50,488 and triazole 1.1, suggesting differences in testing ligands in-vivo across the biased ligands and also in between species (Huskinson et al., 2020, 2022). To bridge the gap between the effects of κOR-biased agonists in in-vivo studies and enhance the translational aspect, it is necessary to examine these agonists alongside an unbiased reference ligand. Additionally, the same reference ligand should be used when testing across different species to ensure the consistency and reliability of the results.

7. Conclusion and future directions

Establishing the predictive translational relevance for κOR ligand bias remains challenging, as the cell signaling pathways that regulate the therapeutic versus adverse effects of κOR activation are a subject of ongoing investigation. Also, the bias factor quantifying models described above do not take into account ligand binding kinetics, the temporal dynamics of signaling processes, or pharmacokinetics of ligands. Therefore, future research in bias signaling at κOR requires the consideration of neuronal cell lines or freshly isolated neurons that are highly expressed κOR in both rodents and human brains. Additionally, using the endogenous neuropeptide for κOR, dynorphin, as a reference ligand would be more persuasive for assessing bias factor. Instead of focusing only on the G-protein and β-arrestin signaling, particularly Gαi1 and β-arrestin 2, respectively, we recommend considering individual comprehensive analysis for other Gα proteins (Gi2, Gi3, GoA, GoB, Gz, and Gustducin) and β-arrestin 1 that are expressed differently in the brain.

Recent studies and discussions indicate that some insights from two decades of κOR research are now being acknowledged. These findings and suggestions lay the groundwork for future research into a successful understanding of κOR signaling, and thus may potentially lead to novel therapeutics with reduced side effects.

Data availability

Data will be made available on request.