Opioid signaling and design of analgesics

Abstract

Clinical treatment of acute to severe pain relies on the use of opioids. While their potency is significant, there are considerable side effects that can negatively affect patients. Their rise in usage has correlated with the current opioid epidemic in the United States, which has led to more than 70,000 deaths per year (Volkow and Blanco, 2021). Opioid-related drug development aims to make target compounds that show strong potency but with diminished side effects. Research into pharmaceuticals that could act as potential alternatives to current pains medications has relied on mechanistic insights of opioid receptors, a class of G-protein coupled receptors (GPCRs), and biased agonism, a common phenomenon among pharmaceutical compounds where downstream effects can be altered at the same receptor via different agonists. Opioids function typically by binding to an active site on the extracellular portion of opioid receptors. Once activated, the opioid receptor initiates a G-protein signaling pathway and/or the β-arrestin2 pathway. The proposed concept for the development of safe analgesics around mu and kappa opioid receptor subtypes has focused on not recruiting β-arrestin2 (biased agonism) and/or having low efficacy at the receptor (partial agonism). By altering chemical motifs on a common scaffold, chemists can take advantage of biased agonism as well as create compounds with low intrinsic efficacy for the desired treatments. This review will focus on ligands with bias profile, signaling aspects of the receptor and probe into the structural basis of receptor that leads to bias and/or partial agonism.

1. Introduction

Pain is classified as acute (severe and short-lived) and chronic (mild to severe with prolonged effect). Four major classes of analgesics are typically used for the treatment of pain, these include anti-inflammatory agents, antidepressants like selective serotonin reuptake inhibitors (SSRIs), anti-convulsants like gabapentinoids, and finally opioids.¹ Non-steroidal anti-inflammatory drugs (NSAIDs), such as aspirin and paracetamol (acetaminophen) treat pain by reducing inflammation. Despite the extensive therapeutic utility, NSAIDs are notorious for severe side effects such as gastrointestinal toxicities, cardiovascular complications, renal injuries, hepatotoxicity, and hypertension, whereas, acetaminophen is critically noxious to the liver.^2–4 SSRIs are notable candidates for the treatment of pain. Although they have a considerably better adverse effect profile, their application is typically limited due to the lack of efficacy. Gabapentin, an anticonvulsant, has been used to treat chronic pain. Nevertheless, due to the lack of uniformity in efficacy,^5,6 it often fails to provide sufficient analgesia in patients. Analgesic opioids (narcotics) operate by changing the brain’s perception of pain. Morphine, codeine, fentanyl, hydrocodone, methadone, and oxycodone are commonly used opioids for the treatment of acute pain. Opioids are very efficient drugs, but their long-term use to treat chronic pain can cause life-threatening side effects. In the last two decades, prescriptions of opioids have increased the risk of a fatal overdose by around four times.⁷

Opioid receptors are one of the most important targets in drug design.⁸ However, the lack of proper understanding of their physiology and mechanism of action provides an opportunity to find out mysteries in opioid receptor structure and its function to develop novel pain-relieving therapeutics with attenuated side effects.⁹ These receptors belong to the group of seven transmembrane domain G protein-coupled receptors (GPCRs). Four main subtypes of opioid receptors are reported, and they are referred to as the μ-opioid receptors (MOR), δ-opioid receptors (DOR), κ-opioid receptors (KOR), and the nonclassical nociceptin opioid receptor (NOP).^10–12 GPCRs are present at the cellular membrane with an orthosteric binding site on the cell’s surface. The signaling mechanism begins with binding to the orthosteric site which causes conformational changes along the intracellular portion of the receptor (Fig. 1A). Activation of the receptor allows for the recruitment of the heterotrimeric G-protein, which is comprised of 3 separate units, Gα, Gβ, and Gγ. All four types of opioid receptors can be activated by exogenous and endogenous opioids. This process stimulates GDP dissociation (the inactive heterotrimeric G-protein allows GDP to bind at Gα) and subsequent GTP attachment to Gα. This signaling sequence results in noticeable conformation changes to the G-protein unit. Detachment of both subunits promotes different downstream signaling process. Inhibitory Gα involves a reduction in cyclic adenosine monophosphate (cAMP) production by attenuating the activity of effector protein kinases (cAMP-dependent protein kinase) and reducing neuronal discharge. The Gβγ subunit stops the signal transduction by directly interacting with the ion channels and suppressing the Ca²⁺ influx. This causes a decrease in the impulsiveness of neurons and thus inhibiting the release of pronociceptive neuropeptides. Gβγ subunit regulates G protein-coupled phosphoinosite 3 kinase (PI3K) and/or mitogen-activated protein kinase (MAPK) pathways. Along with this “classical” G-protein pathway, another downstream signaling route has been reported which involves β-arrestins2 (Fig. 1A).^13,14 Upon GPCR activation, MOR and KOR undergo phosphorylation by G protein-coupled receptor kinase 2 (GRK2) leading to β-arrestin2 recruitment. This process regulates opioid receptor signaling through desensitization and clathrin-mediated internalization.¹⁴ Several structural studies reveal that receptors can preferentially interact and induce specific conformational changes by activating individual signaling pathways either via G-proteins or β-arrestin2.^15,16

Fig. 1.

Signaling pathway and reported chemical structures of MOR and KOR agonists. (A) MOR and KOR activation through external agonists and corresponding downstream signaling produces analgesia as well as adverse effects. (B) Reported small molecule chemical ligands behaving as biased and/or partial agonists at MOR and KOR.

The reported concept provides valuable insights regarding the selective binding mode of opioids involving specific residues on receptors, although more studies are required to fully understand the mechanism of action. The preferential stimulation of a certain signaling pathway is called biased agonism or functional selectivity.^17–19 However, recent studies indicate several ambiguities in the biased agonism hypothesis both at MOR as well as at KOR.^18,20–23 In contrast to previous reports, some adverse effects like respiratory depression remain intact in absence of the β-arrestin2 activation pathway.^22,24 Herein, the concept of low efficacy partial agonism introduces a suitable correlation between therapeutic activity and efficacy without bias factors.²⁵ The concept of functional selectivity along with partial agonism opens a new era for the development of safer opioid analgesics with fewer adverse effects while retaining efficacy. Herein, we will discuss the pharmacology of a few reported MOR and KOR agonists (Fig. 1B) and future directions to develop better drug candidates.

2. Mu opioid receptor

Opioids are clinically used for the treatment of pain, although their usage is not recommended for the treatment of central neuropathic pain anymore.¹ In the past few decades, the major problem with conventionally prescribed opioids is that at high doses they cause severe adverse effects, and at an overdose, they can be lethal.^8,26 Functional selectivity or biased agonism has been proposed as a phenomenon aimed at differentiating between the specific receptor activations (G-proteins over β-arrestins) followed by distinguishable downstream signaling pathways when activated by opioids to develop safer analgesia with reduced side effects.^27,28 At MOR it has been reported that opioid-related analgesia originates from G-protein dependent pathway whereas tolerance, addiction, and respiratory depression are dependent on the β-arrestin pathway.^28,29 These initial findings signify a preferential activation and signal transduction by G-protein-biased agonists over β-arrestin2 to diminish side effects. Phosphorylation-deficient MOR mutant mice that would not recruit β-arrestin2 exhibit promising analgesia but with induced respiratory depression, constipation, and hyperlocomotion. Along with this, morphine and fentanyl-induced respiratory depression is also preserved in β-arrestin2 knockout mice.^22,24 This contemporary knowledge suggests that the concept of bias agonism alone may not be enough to provide a probable explanation for reducing the side effects of MOR-mediated analgesia. Recent studies by Gillis et al.³⁰ highlight that the G-protein-biased agonists (TRV130, PZM21, and SR-17018) show lower intrinsic efficacies towards the receptor compared to agonists with higher intrinsic efficacies (fentanyl and morphine). In addition, the low intrinsic efficacy MOR agonists also show far reduced respiratory depression compared to agonists with higher intrinsic efficacy at the receptor. Hence, the concept of partial agonism suggests a novel mechanism with an advanced therapeutic profile for new opioid ligand development.^25,30–34 The next section covers the pharmacology of biased and partial agonists (see Fig. 1B for the chemical structures of these agents).

2.1. Putative G-protein biased ligands

The concept of G-protein-biased agonism creates new pillars to build analgesia with an advanced safety profile targeting MOR. TRV130 or Oliceridine is an FDA-approved drug, which has been developed from a high throughput screening (HTS) platform followed by chemical optimization to improve its pharmacology and selectivity towards MOR.^33–36 Since TRV130 is moderately effective in separating respiratory depression from analgesia in humans and other adverse effects in rodents, it allows for testing the biased agonism hypothesis in humans. PZM21 similarly was discovered through a structure-based platform.³⁵ It was proposed to be a G-protein-biased agonist with decreased respiratory depression.^30,35 In rodents, its rewarding properties were limited,^35,36 though intrathecally administered PZM21 still retained abuse liability.³⁷ SR17018, a substituted piperidine benzimidazole-based analog, showed a high agonistic affinity for the MOR with no respiratory suppression.^27,30 Several reviews have covered the pharmacology of these potent candidates in detail.^8,18

Recently a spirocyclic analog, LPM3480392 has been reported as a MOR selective biased agonist. This compound contains a structural modification of the marketed therapeutic TRV130 and it is a G-protein biased agonist at MOR like TRV130.^38,39 Although it functions as a partial MOR agonist like TRV130, LPM3480392 exhibits significant MOR agonistic activity (cAMP, EC₅₀=0.35 nM, E_max=91%) without β-arrestin2 recruitment (EC₅₀>30,000 nM, E_max=2%). In comparison, TRV130 in the same assay had cAMP, EC₅₀=1.8 nM, E_max=92% and β-arrestin2 EC₅₀=12 nM, E_max=5%). Along with this, LPM3480392 showed good brain penetration with a promising pharmacokinetic profile and produced a potent antinociceptive effect with reduced respiratory suppression as compared to morphine and TRV130. LPM3480392 did not show significant respiratory depression in mice at a dose of 0.3 mg/kg (SC) which is only 1.5 times greater than its antinociceptive ED₅₀ value. It showed antinociception at a lower dose (0.3 mg/kg) compared to morphine (7 mg/kg). LPM3480392, a partial agonist of MOR, is currently undergoing Phase I clinical trials (CTR20210370) as a potential pain-relieving candidate.⁴⁰

2.2. MOR partial agonists

While first-generation G-protein biased agonists have provided avenues to separate respiratory depression from analgesia. The concept has also led to scientists revisiting the concept of partial agonism.^21,34 The next section discusses two ligands, an established partial agonist and another new ligand that appears promising in preclinical models.

2.2.1. Buprenorphine

Buprenorphine (BUP), first synthesized and evaluated over 40 years ago, is the FDA-approved drug for the treatment of opioid use disorder (OUD).⁴¹ It is well characterized to blunt the adverse effects of many opioids, such as heroin. However, depending on the type of receptor buprenorphine shows multifunctional behavior. It acts as a partial agonist for MOR showing no β-arrestin recruitment,⁴² is an antagonist at DOR⁴² and KOR⁴², and a weak agonist at NOP⁴³. Buprenorphine is a potent analgesic⁴⁴ in rodents and humans but is known to show a ceiling effect in respiratory depression assays.⁴⁵ However, it is not free from other side effects, as it shows hyperlocomotion, constipation, and reward-like behavior in rodents.^42,46 The application of BUP is further limited by its active metabolites norbuprenorphine (norBUP), norbuprenorphine-3-glucuronide, and buprenorphine 3-glucuronide. The lower potency but higher efficacy of norBUP at MOR and unique biological activity of other glucuronides contribute strongly to the overall pharmacology of buprenorphine and thus making these metabolites^47–50 critical substrates in the overall pharmacology of buprenorphine. While buprenorphine is no doubt a safer analgesic, its metabolic liability and activity at opioid and NOP⁵¹ complicates its pharmacology. A metabolically stable buprenorphine-like analog is desirable to ideally test the low efficacy model.

2.2.2. SC13

Mitragyna speciosa plant (commonly known as Kratom) is enriched with biologically privileged natural products for the treatment of pain because of their remarkable opioid properties and stimulant-like effects.^52,53 Mitragynine, one of the major indole alkaloids isolated from kratom, along with semi-synthetic natural products, 7-OH mitragynine, and mitragynine pseudoindoxyl were reported as G-protein-biased agonists showing suitable analgesic properties.^53–61 As a follow-up study, Chakraborty et al.²⁵ have developed a structure-activity relationship (SAR) based on a new series of C9 derivatized 7-OH mitragynine analogs. Their research work came up with a suitable drug-like candidate, SC13, which has partial agonism (Fig. 2) with lower efficacy than DAMGO or morphine in heterologous G-protein assays (Fig. 2A) and synaptic physiology (Fig. 2D). SC13 has less efficacy relative to other potent MOR agonists such as DAMGO, fentanyl, and morphine (Fig. 2A) but lower efficacy than buprenorphine in assays with limited MOR reserve. In the BRET-based Nb33 recruitment assay³⁰ (an assay with limited receptor reserve discussed in detail in later sections) SC13 has an E_max=21% and 8% against hMOR (Fig. 2C) and mMOR respectively, which is completely relatable with the reported efficacy at MOR of synthetic small molecule partial agonists such as TRV130 (42%)³⁰, SR17018 (20%)³⁰, PZM21 (38%)³⁰ and buprenorphine (24%)³⁰; furthermore, SC13 has much lower efficacy than the full agonist morphine (72%)³⁰. SC13 also showed poor arrestin recruitment across a battery of assays for example TANGO (E_max=45%), Path hunter Discoverx (E_max%=<20%) and BRET (bioluminescence resonance energy transfer) assays based βarrestin1 (E_max=<10%) and βarrestin2 (E_max=<10%, Fig. 2B) when normalized against DAMGO. The authors also performed the BRET-based TRUPATH assay⁶² to study the activity of each of the Gα-subtypes. They determined that SC13 showed lower efficacy than DAMGO, morphine, or fentanyl at each Gi/o/z subtype and comparable intrinsic efficacy with buprenorphine at GoA, GoB, and Gz. This result attracts enormous interest as the most abundant Gα subunit in the brain is GoA (at GoA SC13: E_max=75%; Buprenorphine: E_max=65%).This advantageous MOR selective partial agonist demonstrated MOR-dependent antinociception and analgesic activity comparable to morphine while lacking conditioned place preference, constipation, respiratory depression, and hyperlocomotion at equianalgesic morphine doses. Moreover, SC13 was reported to be orally active as well as very potent when administrated subcutaneously under tested conditions in mice. This low-efficacy partial agonism hypothesis on the mitragynine template gives a novel approach to lead MOR agonists over buprenorphine with reduced side effects and preferable analgesia.²⁷ The limitations of SC13 include the presence of a metabolically labile furan ring which second-generation molecules may potentially aim to address.

Fig. 2.

G-protein signaling, arrestin signaling, whole cell electrophysiology of MOR ligands. (A) SC13 (blue) is MOR partial agonist with lower efficacy compared to morphine (purple), fentanyl (brown), and DAMGO (black) in Gi-1 BRET assay. (B) SC13 (blue) shows no β-arrestin2 recruitment compared to fentanyl (brown) and DAMGO (black) in BRET arrestin recruitment assay. (C) SC13 shows lower efficacy compared to DAMGO and morphine and similar efficacy to buprenorphine in Nb33 recruitment assays measured using BRET assays in hMOR. (D) Inhibition of electrically evoked IPSCs in VTA neurons in response to DAMGO (5 μM), morphine (10 μM) and SC13 (10 μM), where each circle is one neuron and horizontal bars indicate means. Reproduced with permission from Chakraborty S, DiBerto JF, Faouzi A, et al. A novel mitragynine analog with low-efficacy Mu opioid receptor agonism displays antinociception with attenuated adverse effects. J Med Chem. 2021;64(18):13873–13892.

3. Kappa opioid receptor

The KOR represents a promising target for pain attenuation. Because of its distribution across the peripheral and central nervous system, it attracts attention in a variety of novel areas including cardiovascular disease, pruritus, nausea, inflammatory diseases, spinal anesthesia, stroke, hypoxic pulmonary hypertension, multiple sclerosis, addiction, and post-traumatic cartilage degeneration.⁶³ KOR agonists do not activate the conventional pathway stimulated by morphine-like MOR agonists and they lack the side effects associated with MOR agonists, such as respiratory depression; therefore, they are considered to be promising non-addictive analgesics. Nevertheless, their use as therapeutics is inadequate due to other side effects such as sedation, motor incoordination, hallucinations, anxiety, and dysphoria-like states.⁶⁴ Understanding the mechanism of KOR activation to produce dysphoria is the key to the development of better analgesics and to defining how endogenous dynorphin opioids produce their depression-like effects. In addition, the aversive effects of KOR activation require arrestin-dependent p38 MAPK stimulation in dopamine neurons but do not require inhibition of dopamine release in the nucleus accumbens. This supports the hypothesis that functionally selective G-protein-biased KOR agonists might have a non-dysphoric, antipruritic, analgesic therapeutic value with low abuse potential.^65,66 Several laboratories have suggested that G protein–biased KOR agonists might be analgesic with fewer side effects. The next section will cover the pharmacology of biased KOR agonists and partial agonists which do not recruit βarrestin-2 (Fig. 1B).

3.1. Putative KOR biased agonists

Quite a few G protein-biased KOR full agonists have been synthesized in recent years. Triazole 1.1 was selected from the high-throughput screening (HTS) of nearly 300,000 compounds and identified as a G-protein-biased KOR agonist in cell lines.⁶⁷ Triazole 1.1 retained efficacy similar to conventional KOR agonists in antinociception and antipruritic assays but did not cause sedation and dysphoria/aversion at analgesic doses or higher compared to the balanced KOR agonist U50,488H.^68,69 Whereas, nalfurafine, another G-protein biased agonist⁷⁰, produced noteworthy antinociceptive and anti-scratch effects without causing dysphoria, psychotomimesis, sedation, motor incoordination, and conditioned place aversion (CPA) in the effective dose ranges.^71,72 Nalfurafine is also the only clinically used biased KOR agonist. This candidate is used in Japan for the treatment of hemodialysis and chronic liver diseases and is a moderately selective biased KOR agonist and partial MOR agonist.^73,74 Interestingly, RB-64²³ (Artasalvin or 22-thiocyanatosalvinorin A), a semi-synthetic derivative of centrally active salvinorin A (balanced KOR agonist), is G-protein biased KOR agonist and shows significant and long-lasting antinociception. RB-64, however, does induce CPA unlike triazole 1.1 and nalfurafine but lacks other KOR adverse effects like sedation and anhedonia-like actions. In summary, the results of KOR-biased full agonists are like MOR-biased agonists, mixed. It is possible that the promising results with nalfurafine are because of other mechanisms beyond bias (not activating the mTOR pathway for example⁷¹). The role of amplified assays in assessing the bias of the KOR full agonists needs to be examined more closely (discussed in sections later in this review). As additional reading on this topic, several reviews have covered the pharmacology of these KOR G-protein biased full agonists^18,75 in detail.

3.2. KOR partial agonists

KOR partial agonists have a potential advantage as antidepressants and anxiolytics over a KOR full agonist. Two ligands with KOR partial agonism that display reduced β-arrestin2 recruitment are discussed in this section.

3.2.1. 6′GNTI

6′-Guanidinonaltrindole (6′-GNTI) is reported to be a potent partial agonist at KOR for the G protein activation without recruiting β-arrestin2. It is a derivative of the highly potent DOR-selective antagonist Naltrindole (NTI) and the 6′-guanidine side chain is one of the major structural requirements to be selective at KOR over other opioid receptors. It also acts as an antagonist which blocks the β-arrestin2 pathway and KOR internalization stimulated by other unbiased KOR agonists (such as U50,488).^76,77 In endogenous striatal neurons as well the preservation of bias activity for G protein over β-arrestin2-mediated signaling induced by 6′-GNTI actions at KOR was observed.⁷⁸ Recent research indicates that 6′-GNTI is a potent anticonvulsant and an antiseizure candidate with reduced dysphoria-like side effects. The lack of CPA or motor impairment in vivo makes it a suitable, safer analgesic candidate over other balanced KOR agonists.⁷⁹ Among liabilities, it has poor brain penetration, and it is possible that, like MOR-biased agonists, some of its preferable properties arise because of its lower intrinsic efficacy.

3.2.2. HS666

HS666 is a chemical modification of Dopamine D2 receptor agonist RU24213, which is a moderately potent KOR antagonist.⁸⁰ HS666 (EC₅₀=449 nM and E_max=24%) showed a selective KOR partial agonism for β-arrestin2 recruitment when compared with U69,593 (EC₅₀=68 nM and E_max=100%) in DiscoveRx PathHunter β-arrestin 2 assay. In GTPγS assay, HS666 showed agonistic properties with G-protein: EC₅₀=36 nM and E_max=50% in comparison, with the balanced agonist, U69,693 (EC₅₀ 18 nM and E_max=100%).⁸¹ Remarkably, HS666 proved to have very promising effects on KOR-specific analgesia, when administered intracerebroventricularly, without showing motor impairment and CPP/CPA (Conditioned Place Preference/Conditioned Place Aversion).⁸¹ A few structural modifications have been performed on this scaffold by introducing bulky substitutes as well as enhancing hydrophobicity to develop very potent KOR selective analogs which exhibited analgesia while lacking sedation and motor impairment.⁸² In a follow-up study, HS666 was shown to not activate the mTOR pathway. Phosphoproteomic approaches in mouse brains have proposed that KOR-mediated CPA is dependent on this pathway.⁷¹ CPA, induced by balanced KOR agonists, is for example blocked by mTOR inhibitors like rapamycin. The lack of mTOR activation is also seen with another biased KOR agonist nalfurafine which also doesn’t cause CPA in rodents and lacks dysphoria in humans. More mechanistic insight is required to understand if the lack of CPA in rodents with HS666 is due to its partial agonistic property or because it does not activate the mTOR pathway. It must be noted that a close analog of HS666 namely HS665⁸¹ does show CPA in rodents possibly owing to its full agonism at KOR.

4. Structural basis of G-protein bias

While numerous MOR and KOR G-protein agonists do exist, the structural basis of receptor-mediated G-protein and arrestin signaling isn’t well understood. CryoEM structures of G-protein biased MOR ligands are now available⁸³ and previous work from Majumdar, Katritch, and Roth groups³² provide insights into sub-pockets that may drive arrestin disengagement. We will cover the structural basis of biased signaling using these concepts in this section.

4.1. Structure based design of MOR biased agonists

A ligand named PZM21 was discovered by computational docking at MOR followed by optimization.³⁵ As discussed previously, PZM21 was reported to be a G-protein biased partial agonist which showed analgesic actions with attenuated respiratory depression compared to the prototypic MOR agonist morphine.^30,35 A recent structure-based study by Georgios Skiniotis, Brian K. Kobilka, and Peter Gmeiner groups demonstrated that the binding specificity of biased ligands towards G-protein over arrestin recruitment is dependent on lipophilic vestibules formed by TM2, TM3, and ECL1 in MOR. The cryoEM structure of PZM21 (PZM21-μOR-Gi protein complex structure at 2.9 Å resolution) was recently solved by this group.⁸³ Further optimization and suitable substitution with lipophilic functionalities on the PZM21 scaffold led to an increase in G-protein bias compared to PZM21. The newly developed potential PZM21 analogs have higher activity with the extracellular vestibule, especially with V143^3.28, I144^3.29, W133^ECL1, and N127^2.63, and they bestow an increased G protein-biased activity in comparison with unbiased agonists such as DAMGO and BU72. The authors also determined another high-resolution cryoEM structure of FH210 (FH210-MOR-Gi protein complex at a global resolution of 3.0 Å), a naphthyl-substituted acryl amide analog of PZM21. FH210 retained PZM21-like binding modes such as the ionic interaction between the amine group of the ligand and D147^3.32 and a conventional hydrogen bond with Y326^7.43. Along with these conventional binding modes, FH210 showed increased Van der Waals interaction within the lipophilic pocket with D216^ECL2, C217^ECL2, W133^ECL1 and N127^2.63. In comparison to the thiophenyl moiety in PZM21; FH210 with a naphthyl moiety showed greater contact ~about 31 Ų higher within the lipophilic vestibule formed by TM2, TM3, ECL1, and ECL2. These recent findings provide a novel structure-based evolution approach for the development of potential therapeutic leads with functionally selective agonistic activity and possibly diminished side effects for MOR drug discovery.

4.2. TM5-ECL2 region as a subpocket for morphinan bias

The active state kappa-opioid receptor structure was solved by the Roth group in 2018⁸⁴ using a ligand named MP1104⁸⁵ based on the morphinan template. MP1104 is a mixed agonist at both MOR as well as KOR and robustly recruited β-arrestin2 at both subtypes. The 6β-amidophenyl arm of this compound was found to bind in the TM2-TM3 region of both MOR as well as KOR. The cyclohexene ring which is a part of ring C of the template prefers a boat conformation. Efforts were next made to design G-protein biased agonists on this template using the MP1104 bound KOR structure.³² Saturation of the cyclohexene ring C of the morphinan template led to the synthesis of MP1202 (Fig. 3Aa), an agonist which now showed G-protein bias at MOR but not KOR (Fig. 3Abc). Computational studies showed the 6β-amidophenyl arm in KOR binds in the TM2-TM3 region (Fig. 3Ae) and as a result, the compound retains arrestin recruitment. However, at MOR, the arm binds in the TM5-ECL2 region suggesting that this subpocket at the entrance of the orthosteric site may act as a hot spot to reduce arrestin signaling (Fig. 3Ad). To design KOR biased agonists, two acidic residues in the TM5-ECL2 region namely D223^5.35 and E209^ECL2 (Fig. 3Bd) were identified, and ligand design swapped the 3′-iodo group on the aryl ring with basic residues like 3′—NH₂ and 3′—NH—C=NH(NH₂) to force the 6β-amidophenyl arm to bind within the TM5-ECL2 region (Fig. 3Ba) and ensuring ring C takes a chair conformation at both receptors. Consistent with the design strategy, MP1207 and MP1208 showed vastly reduced arrestin recruitment at both MOR as well as KOR (Fig. 3Bbc). Analogs of MP1207/1208 where the basic residue was substituted with a polar group like −3′OH behaved as balanced agonists while positioning the basic substituent (—NH₂) at the 4′ position on the phenyl ring again led to a loss of bias compared to MP1207 as the salt bridge formation with the acidic residues was inhibited. The synthesis of MOR/KOR G-protein biased ligands targeting the TM5-ECL2 region using structure-based design is an example where chemists have taken advantage in rationally designing G-protein biased agonists, and this avenue could possibly be used at other targets as well. It must be noted that multiple hot spots at opioid receptors may exist beyond the TM5 ECL2 region which was exploited in the present case. It is also possible that this phenomenon of driving a reduction in arrestin efficacy is template-dependent.

Fig. 3.

Structural basis of G-protein bias. (A) MP1202 at KOR targets the TM2-TM3 region while at MOR targets the TM5-ECL2 region and shows distinct signaling properties (a) Chemical structure of MP1104 and 1202. (b) MP1202 (red) is partial agonist in Tango-arrestin recruitment assays compared to DAMGO (blue) at MOR. (c) MP1202 (green) is a full agonist in Tango-arrestin recruitment assays compared to U50,488h (purple) at KOR. (d) The docking poses of MP1202 (chair form, brown stick) and (boat form, green stick) at an active state of MOR, the saturated ring C in MP1202 leads to interaction of the ligand in the ECL2 and TM5 region leading to a preference of chair form shown by a red arrow. (e) The docking poses of MP1202 (chair form, brown stick) and (boat form, green stick) at an active state of KOR, the flip of ring C conformation from chair to boat is shown by a blue arrow. (B) MP1207 and MP1208 prefer the chair conformation and target the TM5-ECL2 region and are G protein biased agonists at KOR and at MOR show no measurable arrestin recruitment. (a) Chemical modification on MP1202. (b) MP1207 (orange) and MP1208 (green) are partial agonists in Tango-arrestin recruitment assays compared to U50, 488 (purple) at KOR. (c) MP1207 (orange) and MP1208 (green) show no arrestin recruitment in Tango-arrestin recruitment assays compared to DAMGO (blue) at MOR. (d) Docking of MP1202 (green sticks) and MP1207 (yellow sticks) in wild type KOR shows MP1207 chair interact with D223 and E209 residues in TM5-ECL2 region by salt bridge formation, while MP1202 boat form does not accommodate this region. Reproduced with permission from Uprety R, Che T, Zaidi SA, et al. Controlling opioid receptor functional selectivity by targeting distinct subpockets of the orthosteric site. Elife. 2021;10:e56519.

5. Assay limitations in correlating in vitro–in vivo function

Receptor reserve refers to the phenomenon in which an agonist produces the maximal response by activating only a fraction of the available receptor population present in the system. For second–messenger systems that measure downstream G-protein signaling, signal amplification may yield a similar maximal response making it more difficult to differentiate between a full agonist and a partial agonist.^86,87

The concept of using the right assay for assessing partial vs full agonists is important to distinguish between truly biased agonists and low efficacy G-protein biased agonists (see graphic presented in Fig. 4A–C). PZM21,³⁵ TRV130,³⁹ 7OH mitragynine^53,54,58 as well as SR17018⁸⁸ were previously described as G-protein biased agonists (described in this review under MOR biased agonists). However, when assessed in assays with minimum receptor amplification (assay was initially described by Stoeber et al.⁸⁹ and has recently been rigorously validated by Gillis et al.³⁰), these compounds were found to be low efficacy agonists.^30,33 This calls for the design and development of compounds with either greater G-protein bias and/or ones that are potency-biased instead of efficacy-biased and preferably with high efficacy in both G-protein and the arrestin pathway. Morphine, a well-characterized partial agonist shows an efficacy of 97–123%^30,32 normalized against DAMGO in an amplified assay, while in less amplified assays (e.g. in Nb33 recruitment assays), the efficacy is around 70% (Fig. 4D and E).^25,30,33 This scenario is seen with the previously defined biased agonists as well with PZM21 showing 84–86% efficacy³⁰ in an amplified assay compared to 38% efficacy in the same Nb33 assay and 7OH mitragynine, whose efficacy in BRET based assays and [³⁵S]-GTPγgS assays was 51%^33,54 and 77%⁵⁸ respectively, had an efficacy similar to buprenorphine in less amplified assays³³.

Fig. 4.

Signaling differences in conventional, biased and partial opioids (A) Signaling behavior of conventional opioids. (B) Signaling behavior of biased opioids in an amplified assay. (C) Decrease in G-protein and arrestin efficacy of ligands with limited receptor reserve. (D) Buprenorphine and 7-OH Mitragynine are partial agonist with lower efficacy compared to DAMGO and Morphine in Gi-1 BRET assay. (E) Intrinsic efficacy of buprenorphine, morphine as well as 7OH mitragynine is lowered in the NB33 recruitment assays measured using BRET assays in hMOR. Reproduced with permission from Bhowmik S, Galeta J, Havel V, et al. Site selective C–H functionalization of mitragyna alkaloids reveals a molecular switch for tuning opioid receptor signaling efficacy. Nat Commun. 2021;12(1):1–14; Chakraborty S, DiBerto JF, Faouzi A, et al. A novel mitragynine analog with low-efficacy Mu opioid receptor agonism displays antinociception with attenuated adverse effects. J Med Chem. 2021;64(18):13873–13892.

In addition to the Nb33 assay, other investigators have used irreversible antagonists⁹⁰ like β FNA, β-chlornaltrexone, and methcinnamox to evaluate the influence of this inactivation on agonist dose-response curves and again, arrive at an accurate estimation of true receptor intrinsic efficacy.³⁴

More recently, low efficacy agonism was assessed in the VTA region⁹¹ of the brain using electrophysiology as another ex-vivo measure of quantifying efficacy.²⁵ DAMGO was found to have a higher efficacy over morphine as expected from cell-based assays, while mitragynine agonists like SC13²⁵ were found to have lower efficacy over both morphine and the prototypic agonist DAMGO.

In addition, genetic strategies to decrease receptor levels have been looked at with the biased agonist TRV130 using the MOR +/− mice.⁹² TRV130 showed no tolerance in wild-type mice when administered for 10 days but this reduced tolerance was not observed in the homozygous MOR +/− mice where a tolerance developed within four injections.

Utilizing the appropriate in vitro assay along with ex vivo and in vivo approaches is critical for discovering novel entities which show pathway selectivity and distinguishing them from partial agonists. It is hoped that leads identified through iterative SAR approaches will be evaluated in a battery of these efficacy assays before administering them in animals for in vivo efficacy analysis and evaluating if functional selectivity can be obtained in vivo.

6. Future directions

The FDA approval of the first MOR-biased compound TRV130 (oliceridine, Olinvyk) will allow for testing the biased opioid hypothesis in humans.³⁸ However, the hypothesis of biased agonism needs to evolve beyond one G-protein (Gi-1) and one arrestin subtype (β-arrestin2) to include other Gα-subtypes. Specifically, opioids are known to couple to G_i1, G_i2, G_i3, G_oA, G_oB, and G_z in addition to G_usducin in the case of KOR.⁶² Although preliminary, opioids tend to have specificity for the G_z-isoform.^25,62,93 Interestingly, MOR agonists and KOR agonists show a preference for the Gz subtype with partial agonists (buprenorphine at MOR and pentazocine at KOR) showing higher potency as well as efficacy at this subtype.^25,94

While the exact role of Gz in opioid physiology is unknown, anti-conception of morphine in Gz KO mice is moderately sensitive to morphine antinociception.⁹⁵ Also, an increased tolerance and lethality⁹⁶ was observed with morphine in the same KO mice. As more compounds are screened across this paradigm, it is hoped that compounds with selectivity at other subtypes will be synthesized, and individual roles of these isoforms will be evaluated. Overall, this strategy may represent a novel avenue to developing safer analgesics.

G-protein-biased agonists, as well as partial agonists, are now appearing in the literature with structures, such as cryo-EM structures of PZM21, FH210,⁸³ and mitragynine pseudoindoxyl⁹⁷ at MOR are now available. It is expected that these structures will aid in the atomic level understanding of receptor efficacy, potency, and pathway selectivity. In addition, these structures will allow for the rational design of the next generation of highly biased tool compounds to probe opioid functional selectivity in the near future.

Abbreviations

β FNA

beta-funaltrexamine

BRET

bioluminescence resonance energy transfer

cAMP

cyclic adenosine monophosphate

CPA

conditioned place aversion

CPP

conditioned place preference

CryoEM

cryogenic electron microscopy

DAMGO

[DAla2, N-MePhe4, Gly-ol5]-enkephalin

DOR

delta opioid receptor

FDA

food and drug administration

GDP

guanosine diphosphate

GPCRs

G-protein coupled receptors

GRK2

G-protein coupled receptor kinase 2

GTP

guanosine triphosphate

HTS

high-throughput screening

KOR

kappa opioid receptor

MAPK

mitogen-activated protein kinase

MOR

mu-opioid receptors

mTOR

mammalian target of rapamycin

Nb33 assay

nanobody-based biosensor assay

NOP

nonclassical nociceptin opioid receptor

NSAIDS

non-steroidal anti-inflammatory drugs

OUD

opioid use disorder

PI3K

phosphoinosite 3 kinase

SAR

structure activity relationship

Sc

subcutaneous

SSRI’s

selective serotonin reuptake inhibitors

VTA

ventral tegmental area