Discovery of SHR9352: A Highly Potent G Protein-Biased μ-Opioid Receptor Agonist

Abstract

Recently, targeting the G protein-biased signaling has emerged as an attractive therapeutic strategy for treating severe acute pain with the potential to reduce the side effect of the traditional opioid drug. Herein, we describe the discovery of a highly potent G protein-biased μ-opioid receptor (MOR) agonist, SHR9352. This novel molecule exhibited excellent MOR activity and limited β-arrestin recruitment, as well as a high selectivity over κ-opioid receptor and δ-opioid receptor demonstrated robust in vivo efficacy and displayed favorable pharmacokinetic properties across species.

Introduction

μ-Opioid receptor (MOR) agonists, such as morphine and its derivatives, are among the most effective analgesics in clinical use, but their efficacies are always accompanied by opioid-related adverse effects, including respiratory repression, constipation, nausea, vomiting, and drug addiction.¹⁻⁴

In recent years, the concept of the biased G protein-coupled receptor (GPCR) ligands has been developed by several groups from both academia and industry.⁵⁻⁸ It was hypothesized that G protein-biased MOR agonists that selectively activate MOR G protein signaling with minimal or no recruitment of β-arrestin-2 would retain equivalent or better analgesic activity while decreasing the β-arrestin-2-dependent side effects. Pharmacological studies using β-arrestin-2 knockout mice treated with morphine had shown a significant reduction in the level of the respiratory repression and constipation, with the observation of enhanced analgesia.^9,10

To date, several ligands with different chemical scaffolds have been developed.^5,6,11,12 TRV130 (Oliceridine), the first example of this class, stimulates G protein activation by presumably changing the conformation of MOR upon its binding and results in little β-arrestin-2 recruitment.¹² TRV130 has demonstrated potent analgesia with reduced respiratory repression and constipation in both preclinical studies and clinical trials.^5,8,12−14 Manglik et al. recently developed another biased MOR agonist with a different chemotype, such as PZM21 (Figure 1). Compared with morphine, PZM21 was reported to be more efficacious, with reduced on-target liabilities.⁶

Figure 1.

Examples of typical structures of biased ligands.

In 2013, Trevena Inc. disclosed their study of the discovery of TRV130, including detailed structure–activity relationship (SAR) information. It was reported that methyl substitution at the benzylic position (compounds T-1, T-2, and T-3) resulted in 6–40 fold potency loss (Figure 2). “It was concluded that substitution on the benzylic position was not preferred and no further chemistry efforts were undertaken at this position”.¹⁴ Occasionally, we found that the benzylic position fused to the aromatic ring resulted in unexpected results and further optimization of the substitutions on this fused ring has profound influence on the β-arrestin-2 recruitment. Ultimately these efforts culminated in the discovery of a remarkably high potency and efficacious biased MOR agonist: SHR9352.

Figure 2.

SAR information of benzylic position substitutions.

Results and Discussion

Initially, the lead discovery efforts focused on fusing a saturated ring-afforded diastereomeric pyran amine 1 and 2. Unlike the reported benzylic isomers T2 and T3, which displayed similar MOR potency,¹⁴ the (S)-pyran amine 2 delivered a much higher potency, whereas the corresponding (R) isomer 1 showed very weak MOR activity (Figure 2). Inspired by this result, we focused on the (S) configuration to further explore SAR of the benzo-fused rings. Unfortunately, the hit compound 2 was found to exhibit very high β-arrestin-2 activity (E_max = 171%, EC₅₀ = 63 nM, Table 1). As for the low β-arrestin-2, recruitment was the key for a biased MOR agonist; more efforts were involved to modify that.

Table 1. SAR of the Benzylic-Fused Ring Series.

^aTRV130 was used as the positive control, and its EC₅₀ was 7.9 nM in this cAMP assay; the maximal response was defined as 100% at 10 μM TRV130.

^bMorphine was used as the β-arrestin assay positive control and its E_max was defined as 100% at 10 μM morphine. The β-arrestin E_max of TRV130 is 14% compared to that of morphine.

^cAll of the data were the mean value of three measurements.

The first round SAR was focused on the substitutions of the saturated ring and phenyl group. It was found that fluorine on the aromatic ring (compound 3) afforded the MOR cAMP potency similar to 2, whereas it exhibited higher β-arrestin recruitment (E_max = 341%). The six-membered carbon cycle amine 4 delivered the same level of MOR and β-arrestin potency compared to that of compound 2. Then, additional groups were introduced on the saturated ring to afford compounds 5–10 to modulate the potency. The hydroxyl group was installed on the saturated ring, as shown in Table 1. It was found that the cis substituted compound 6 has higher β-arrestin E_max than that of the trans isomer 5. Encouragingly, the trans-methoxyl 8 further decreased the β-arrestin recruitment (E_max = 30%). Ketone 9 was taken into consideration with a reduced chiral center, which had moderate β-arrestin activity but was less potent toward MOR activity. Then, vinyl cyanide 10 was obtained, which resulted in very high β-arrestin activity (E_max = 334%), whereas it resulted with decreased MOR cAMP potency.

Different ring sizes and aromatic groups were further investigated. The benzo-5-membered ring series 11 and 12 both afforded high MOR cAMP potency. Meanwhile, methyl ether 12 displayed much higher β-arrestin recruitment than that of 9, possibly due to the different angle of the methoxyl group. Other aromatic groups instead of the phenyl group were also prepared (13–16). It was observed that the benzene series had lower β-arrestin E_max (14 vs 9 and 16 vs 17). The result showed that thiophene 16 delivered less MOR potency than its analogue 14, although the structural difference among them was only the sulfur position.

In the next step, acetal group was introduced to replace carbonyl to afford compound 17, yielding an improvement of 2-fold better MOR potency and decrement of β-arrestin activity compared to that of its precursor 9. Considering that thio-acetal was more stable than acetal group, thioether 18 (hereafter named as SHR9352) was then prepared, which afforded a remarkably high MOR potency (EC₅₀ = 0.77 nM), with minimal β-arrestin recruitment (E_max = 18.8%).

Inspired by the biased activity seen in SHR9352, we would like to further understand its binding to the MOR receptor at the molecular level. On the basis of a published MOR agonist cocrystal structure (PDB code: 5C1M),¹⁵ the TRV130 molecule was modeled into the ligand binding site using MOE. As shown in Figure 3 and Scheme 1, the modeled TRV130 formed four key interactions with MOR as follows: first, the basic amine formed the signature ionic interaction with D147^3.32. Second, the pyridine ring formed π–π interaction with the histidine ring of the amino-terminal residue H54. Third, the spiro cyclopentyl ring was surrounded by a set of hydrophobic residues. These include S55, Q124^2.60, N127^2.63, W133, V143^3.29, I144^3.28, and C217. Last, the thiophene and pyridine rings were surrounded by nine residues, Y148^3.33, M151^3.36, V236^5.42, V300^6.55, H297^6.52, I296^6.51, W293^6.48, W318^7.35, I322^7.39, and Y326^7.43. As shown by the generic GPCR residue numbers,¹⁶ five of them were located in helix 5 and 6, whose conformational changes were reported in the literature as relevant to the biased GPCR signaling.¹⁷ The proposed binding model was also in agreement with the reported SAR of TRV130 in that modification in the thiophene ring and the inversion of the stereochemistry of the pyridine ring affected the G protein/β-arrestin selectivity.¹⁴ SHR9352 was modeled similar to the binding mode of TRV130 while maintaining all of the key interactions mentioned above. The 1,3-dithiolane group pointed to the same direction as the methoxy group on thiophene ring of TRV130, which packs even closer to W293^6.48 and I296^6.51, both of which were located on helix 6.

Figure 3.

Modeled binding mode of TRV130 (cyan) and SHR9352 (magenta) into the MOR cocrystal structure.

Scheme 1. Synthetic Route to SHR9352.

Reagent and conditions: (i) Boc₂O, Et₃N, CH₂Cl₂; (ii) KMnO₄, MgSO₄, acetone/H₂O, 52% two steps; (iii) HCl, CH₂Cl₂, 65%; (iv) NaBH(OAc)₃, CH₂Cl₂, 33%; (v) pyridinium p-toluenesulfonate, toluene, 110 °C, 96%.

The synthetic route of SHR9352 was straight forward. It started with (S)-1,2,3,4-tetrahydronaphthalen-1-amine 19, which could be obtained from a known procedure.¹⁸ The protection of benzylic amine with Boc group, followed by oxidation, delivered ketone 21 in 52% combined yield. The resulting amine from removal of the protecting group reacted with the reported intermediate 23(19) afforded compound 9. Finally, the thio-acetal functional group was installed under standard condition to furnish SHR9352 with quantitative yield.

Then, SHR9352 was moved for further in vitro study, which exhibited high potency in mouse and rat MOR cAMP assays consistently (Table 2). In addition, SHR9352 had potencies of 37 and 115 nM in human κ-opioid receptor (KOR) and human δ-opioid receptor (DOR) cAMP accumulation assays, respectively, achieving 48-fold (hKOR/hMOR) selectivity over KOR and 149-fold (hDOR/hMOR) selectivity over DOR.

Table 2. In Vitro Potency of SHR9352.

	mMOR	rMOR	hKOR	hDOR
nM	0.19	0.02	37	115

Table 3 summarized the pharmacokinetic (PK) data of SHR9352 across species. This compound exhibited low clearance in rats and dogs (CL was 17.0 and 16.1 mL/(min kg), respectively) and displayed linear dose-exposure relationship in rats from 0.05 to 0.5 mpk. In addition, the clearance was similar and showed comparable exposures in mice when compared intravenously (iv) with subcutaneous (sc) administrations. As shown in Table 3, the overall PK profile of SHR9352 was good across species, with an improved T_1/2 from rodent to nonrodent animals.

Table 3. Pharmacokinetics Profiles of SHR9352 in Mice, Rats, and Dogs^a.

species	mouse		rat			dog
administration	iv	sc	iv	iv	iv	iv
dose (mpk)	0.3	0.1	0.05	0.15	0.5	0.3
T1/2 (h)	0.78	0.67	1.37	1.13	0.76	7.12
CL (mL/(min kg))	81.9	86.6	17.0	18.2	18.5	16.1
AUC0–24h (ng/(mL h))	59.7	18.9	47.6	137	453	302

^aCL is clearance. T_1/2 is the half-life of the compound exposure in plasma. AUC is the area under the curve.

The rat incision assay was one of the tests used to measure the efficacy of analgesics.^5,20 The advantage of this model was its relatively high reproducibility. Therefore, this model was employed in the study of SHR9352. Consistent with high in vitro potency, SHR9352 displayed a remarkable efficacious character even at 0.01 mpk dosage. Meanwhile, a clear dose-dependent effect was observed from 0.01 to 0.03 mpk (90–128.7% effect). The 0.03 mpk dosage efficacy seems higher than that of the control blank group, possibly due to the added anesthesia effect of the drug. In another separate experiment, a higher dosage was also tested while saturated efficacy could be observed. It is worth mentioning that TRV130 was used as an internal reference and its 0.1 mpk dosage reached efficacy comparable to that of SHR9352 at 0.01 mpk dosage (Figure 4).

Figure 4.

In vivo rat incisional pain study.

Moreover, the β-arrestin associated side effect was investigated in the mouse charcoal meal model.²¹ TRV130 was also used as a reference control. It was found that with SHR9352, there was almost no display of the constipation effect after 0.02 mpk sc administration compared to that in the control group and TRV130 at 0.1 mpk. As shown in Figure 5, the percentages of gastrointestinal transit were 45.4, 44.5, and 42.2%, respectively. This result correlated well with the in vitro β-arrestin profile.

Figure 5.

In vivo mouse gastrointestinal function study-small intestinal transit test.

Because MOR expressed in brain played a key role in opioid mediated analgesia, the blood-brain barrier (BBB) penetration of ligands was one of the important parameters driving the efficacy. Thus, the K_p-unbound value was introduced to evaluate the BBB penetration ability.^22,23 As indicated in Figure 6 and Table 4, the brain concentration of SHR9352 was 403 ng/g and the K_p-unbound value was 0.334.a Both the favorable K_p-unbound value and high brain concentration of SHR9352 indicated good central-nervous system penetration, which supported the observed robust efficacy in the rat incision model.

Figure 6.

C_max of SHR9352 at 0.25 h after iv dosing.

Table 4. In Vivo Rat BBB Penetration Results.

cmpd	fu-rat plasma	fu-brain tissue	plasma (ng/mL)	brain tissue (ng/g)	Kp-unbound
SHR9352	0.55%	0.48%	1056	403	0.334

In addition, the Cytochrome P450 inhibitory potential of SHR9352 was investigated to assess the potential likelihood of drug–drug interactions (DDIs) of major CYPs, such as CYP1A2, CYP2C9, CYP2C19, CYP2D6, and CYP3A4. As highlighted in Table 5, SHR9352 had no CYP450 inhibitions in terms of 1A2, 2C9, and 2C19 but had moderate levels of inhibitions for CYP2D6 and CYP3A4 isozymes. However, due to very strong in vitro potency and low in vivo efficacious dosage (0.01 mpk), the CYP safety margin was considered to be sufficiently enough without DDI liability.b

Table 5. CYP Inhibition Data of SHR9352.

CYP450 isoform	1A2	2C9	2C19	2D6	3A4(m)a	3A4(t)b
IC50 (μM)	>30	34.0	24.3	3.24	3.27	4.84

^am: midazolam.

^bt: testosterone (FDA recommended 3A4 substrates).

Finally, we conducted a non-good laboratory practice 7 day acute rat toxicity study of SHR9352 at dosages of 0.05, 0.15, and 0.5 mpk. All of the dosages were well tolerated and no obvious abnormal findings were observed, except at the highest dose of 0.5 mpk for male rats, an elevation of monocytes percentage was found (mono %, p < 0.05). On the basis of a general toxic investigation, including body weight, hematology, blood chemistry, clinical observations, gross necropsy, and toxicokinetic analysis, it was concluded that no observed adverse effect level (NOAEL) and maximum tolerance dose were 0.15 mpk and over 0.5 mpk, respectively. The estimated safety margin was at least 15-fold on the basis of the efficacious dose of 0.01 mpk and the NOAEL at 0.15 mpk.c

In summary, SHR9352 demonstrated a highly potent G protein-biased MOR agonist. It exhibited excellent MOR cAMP activity and limited β-arrestin recruitment as well as a high selectivity over KOR and DOR. This novel molecule entity displayed favorable PK profiles across species and a robust in vivo efficacy with diminished β-arrestin associated side effects. Moreover, SHR9352 showed a sufficiently large safety window due to its very low efficacious dose and sub-nanomolar level in vitro potency. Additional tests and studies involving the rat constipation models were ongoing, and more relevant data will be disclosed in due course.

Glossary

Abbreviations

MOR

μ-opioid receptor

KOR

κ-opioid receptor

DOR

δ-opioid receptor

CNS

central-nervous system

BBB

blood-brain barrier

NOAEL

no observed adverse effect level

MTD

maximum tolerance dose

Supporting Information Available

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acsomega.7b01452.

• Synthesis and characterization data for new compounds and methods for in vitro, in vivo, and pharmacokinetic assays (PDF)

The authors declare no competing financial interest.