Synthetic Studies of Neoclerodane Diterpenes from Salvia divinorum: Design, Synthesis, and Evaluation of Analogues with Improved Potency and G-protein Activation Bias at the μ Opioid Receptor

Abstract

Previous structure-activity relationship (SAR) studies identified the first centrally-acting, non-nitrogenous μ opioid receptor (MOR) agonist, kurkinorin (1), derived from salvinorin A. In an effort to further probe the physiological effects induced upon activation of MORs with this non-morphine scaffold, a variety of analogues were synthesized and evaluated in vitro for their ability to activate G-proteins and recruit β-arrestin-2 upon MOR activation. Through these studies, compounds that are potent agonists at MORs and either biased towards β-arrestin-2 recruitment or biased towards G-protein activation have been identified. One such compound, 25, has potent activity and selectivity at the MOR over KOR with bias for G-protein activation. Impressively, 25 is over 100x more potent than morphine and over 5x more potent than fentanyl in vitro and elicits antinociception with limited tolerance development in vivo. This is especially significant given that 25 lacks a basic nitrogen and other ionizable groups present in other opioid ligand classes.

Graphical Abstract

INTRODUCTION

The United States is currently experiencing an opioid and drug abuse epidemic, and drug overdose is the leading cause of accidental death in the United States.^1-2 Between 1999 and 2017, more than 700,000 Americans died from a drug overdose, and nearly 400,000 of those deaths involving an opioid.³ A discussion of the opioid epidemic is incomplete without discussing the prevalence of pain and prescription opioids, as roughly 218,000 people have died from overdoses related to prescription opioids between 1999 and 2017.⁴ Pain affects more American adults than cancer, heart disease and diabetes combined.⁵ The estimated costs in the US are over $600 billion, including healthcare, disability compensation and lost workdays; however, the personal costs in terms of suffering and quality of life cannot be measured.⁶

Opioid analgesics are the gold-standard for pain treatment and have been so for many years, despite their many adverse effects, including tolerance, dependence, and respiratory depression.⁷ The long-term use of opioids leads to analgesic tolerance, requiring higher doses to achieve adequate pain relief.⁸ As the opioid dose is increased, the patient is at higher risk for developing dependence and ultimately becoming addicted to the opioid therapy.⁹ Furthermore, the rates of prescription opioid misuse and addiction in chronic pain treatment have been estimated to be 21% - 29% and 8% - 12%, respectively.¹⁰ Additionally, it has been estimated that roughly 4% - 6% of people who misuse prescription opioids transition to heroin.^11-14 Even so, opioids are still the most prescribed drug class for pain.¹⁵ Conversely, patients who are suffering from chronic pain are often hesitant to use opioids, due to the associated risks and social and legal issues associated with using opioids, which results in frequently undertreated and inadequately treated pain.¹⁶ Therefore, the development of pain relievers without these stigmas and side effects is a critical medical need.

Studies have indicated that compounds that are functionally selective for the G-protein coupled pathway of the μ-opioid receptor (MOR) over the pathway for β-arrestin-2 recruitment may result in such compounds that elicit analgesia without the common side effects of current opioids.^17-20 Functional selectivity refers to the ability of ligands to stabilize different conformations of a single receptor subtype and thus differentially regulate the downstream signaling cascades.²¹ Early studies dosing mice that lack β-arrestin-2 with morphine demonstrated enhanced analgesia and reduced side effects associated with morphine.^{17-18, 22} These studies were the first to give insight into the possibility that the analgesic effects at MOR could be dissociated from the side effects through the development of functionally selective compounds.²¹ Conflicting evidence has shown that despite having bias for G-protein activation, TRV130²³ and PZM21²⁴ still induces drug liability potential and respiratory depression, respectively. Additionally the molecular mechanisms underlying respiratory depression are not fully understood. A recent study suggests that β-arrestin-2 does not mediate respiratory depression.²⁵ This highlights the need to develop additional biased ligands in order to elucidate how these adverse effects are mediated. Thus, these studies serve as a reminder that additional studies are necessary to further our knowledge of signaling bias, both in vitro and in vivo, for the understanding of adverse effects of MOR agonists.

We recently reported the identification of the first centrally-acting, non-nitrogenous MOR agonist (1).²⁶ Through investigation of the structure-activity relationships (SAR) of 1, we explored analogues with substitutions to the 2-, 3-, and 4-positions of the phenyl ring. Although this initial SAR campaign did not yield any compounds that were more potent at MORs than the unsubstituted 1, we did identify trends that we were able to use to guide the next set of analogues in the search for more potent MOR agonists with reduced side effect profiles. Notably, although no substituted phenyl compounds were more potent than the original, the substitutions at the 3- and 4-positions were generally more potent than the 2-position, and the most potent of the substituted analogues were methoxy and fluoro substituted derivatives. We also evaluated these compounds for their ability to recruit β-arrestin-2, and while 1 was somewhat biased towards activation of the G-protein pathway, all of the analogues evaluated were biased towards β-arrestin-2 recruitment.²⁶ Using these results to guide our analogue design, we report herein the 1) exploration of heterocyclic derivatives and the 2) exploration of binding pocket depth and hydrogen bond (H-bond) capabilities.

RESULTS AND DISCUSSION

Our current SAR study began with the conversion of salvinorin B (2) into a mixture 3a and 3b using previously developed methodology (Scheme 1).²⁶ In an effort to further probe MOR binding and activation by the neoclerodane scaffold, several heterocycles (4 – 15) were introduced as bioisosteres in place of the phenyl ring in 1. Analogues with varied linkers between the ester and phenyl moieties were also explored (16 – 18) to probe the depth and shape of the binding pocket. Additionally, due to the activity of the anisole derivatives, free and protected phenols (19 – 24) as well as benzylic alcohols (25 and 26) and a methylated derivative thereof (27) were sought in order to probe the H-bond characteristics in the MOR binding site and to manipulate the pK_a of the ligands. The structural and activity differences between morphine and codeine highlight the role that H-bonding plays at MORs, as the only structural difference between the two is a free phenol in morphine and a methylated phenol in codeine, with codeine being significantly less active at MORs than morphine.²⁷ However, the installation of free alcohols required protection methodology, which is not a straightforward task considering the complexity and lability of the neoclerodane diterpenoid core. Therefore, optimization of both protecting groups and deprotection methods was accomplished in order to synthesize a variety of the desired analogues.

Scheme 1.

Synthetic route to kurkinorin and derivatives ^a

^aReagents and conditions: a) Cu(OAc)₂, MeOH/CH₂Cl₂ (1:1); b) RCO₂H, DMAP, EDCI, CH₂Cl₂; c) TBAF (2.0 equiv.), THF, −10°C, 15 min.; d) KHSO₄; H₂O, MeOH, Acetone (1:1:1) ^b Alternative reaction conditions: Prepared from corresponding MOM-protected compounds, 21a – 21c, CBr₄, PPh₃, DCE, 40 °C, overnight.

Initially, TBS-protected phenolic acids were coupled to the mixture of 3a and 3b and then deprotected using common TBAF cleavage conditions. While this method was generally successful for the phenolic substrates that had only a single substitution on the phenyl ring, 19 and 20, it was low yielding and resulted in side products of similar retention times (t_R) during the preparative HPLC purification process. This issue was more evident in subsequent derivatives with halogen moieties ortho to the TBS-protected phenol. In fact, multiple rounds of preparative HPLC purifications were unsuccessful in purifying these compounds to greater than 90% purity. Therefore, other protection and deprotection methods were explored to provide more robust access to ortho-halogenated phenols. Ultimately, it was determined that the use of a methoxymethyl ether (MOM) protecting group (21a – 21c) allowed for mild removal using triphenylphosphine and carbon tetrabromide as reported by Peng, et. al.²⁸ These conditions resulted in better resolution of the HPLC peaks and ultimately allowed for a better purification process of compounds 22a – 22c.

Similar purification issues were seen in the synthesis of the benzylic alcohol substituents. The TBS-protected hydroxymethyl benzoic acids were coupled to the mixture of 3a and 3b, but TBAF deprotection again yielded low amounts of product with side products that complicated the purification process. This deprotection was successfully accomplished using previously reported, mild deprotection conditions with slight modifications to accommodate the insolubility of the derivatives in water.²⁹ A mixture of 1:1:1 H₂O/MeOH/acetone, 0.5 eq. KHSO₄, and the TBS-protected benzyl alcohol derivative stirring overnight resulted in the desired, free benzyl alcohol derivatives 25 and 26. Using these methods and the EDCI/DMAP coupling conditions described previously,²⁶ analogues 4 – 27 were prepared.

Previous SAR studies focused on neoclerodanes indicated that heterocyclic substitutions were well–tolerated and resulted in MOR agonists.^30-31 Utilizing a commercially available forskolin-induced cAMP assay, these compounds were evaluated for agonist activity at MOR.²⁶ Replacement of the phenyl ring of 1 with pyridines results in similar activity trends as seen for the phenyl ring substitutions with the 3- and 4- pyridines being more potent than the 2-pyridine (4 – 6, Table 1). Additionally, substituting the pyridine ring with a methoxy group, which was previously shown to maintain activity, does not substantially increase the potency of the pyridine derivative (5: EC₅₀ = 3.2 ± 0.8 nM; 7: EC₅₀ = 2.5 ± 0.3 nM). These data suggest that further substitutions off the pyridine ring will not significantly increase the potency of these compounds.

Table 1:

MOR and KOR pharmacology of SAR-driven analogues: Inhibition of forskolin-induced cAMP

Compound	R	MOR EC50 ± SEMa,b (nM)	KOR EC50 ± SEMa,b (nM)
DAMGO	N/A	0.6 ± 0.4	not determined
Morphine	N/A	5.04 ± 3	not determined
Fentanyl	N/A	0.26 ± 0.1	not determined
1	Phenyl	1.2 ± 0.2	>10 000
4	2-Pyridinyl	410 ± 50	not determined
5	3-Pyridinyl	3.2 ± 0.8	>10 000
6	4- Pyridinyl	6 ± 1	>10 000
7	6-CH3O-3-Pyridinyl	2.5 ± 0.3	>10 000
8	2-Benzofuranyl	30 ± 9	987.1 ± 267.5
9	2-Thianapthenyl	9 ± 2	>10 000
10	2-Thiophenyl	1.4 ± 0.8	>10 000
11	3-Thiophenyl	1.3 ± 0.9	>10 000
12	5-Oxazolyl	30 ± 2	>10 000
13	5-Thiazolyl	5 ± 2	>10 000
14	4-Oxazolyl	>10 000c	not determined
15	4-Thiazolyl	>10 000c	not determined
16	CH2C6H5	130 ± 50	187.4 ± 29
17	CH2CH2C6H5	400 ± 100	not determined
18	CHCHC6H5	30 ± 10	>10 000
19	3-OH-C6H4	1.3 ± 0.3	>10 000
20	4-OH-C6H4	0.83 ± 0.04	>10 000
21a	3-F,4-OCH2OCH3-C6H3	3.4 ± 0.6	>10 000
21c	3-Br,4-OCH2OCH3-C6H3	1300 ± 200	>10 000
22a	3-F,4-OH-C6H3	0.6 ± 0.2	>10 000
22b	3-Cl,4-OH-C6H3	4 ± 1	1630 ± 193 (110%)
22c	3-Br,4-OH-C6H3	17 ± 7	>10 000
23	6-Benzofuranyl	1.5 ± 0.6	>10 000
24	5-Benzofuranyl	9 ± 2	>10 000
25	4-CH2OH-C6H4	0.03 ± 0.01	>10 000
26	3-CH2OH-C6H4	2.42 ± 0.07	>10 000
27	4-CH2OCH3-C6H4	13 ± 3	>10 000

^aMean ± standard deviation; n ≥ 2 individual experiments run in triplicate.

^bE_max = 100%, unless noted otherwise.

^cE_max = 0 % up to 10 μM.

Heterocyclic compounds containing oxygen and sulfur atoms were also explored, as previous SAR indicated that these changes retained potency at MORs.³¹ 2-Benzofuroyl 8 was evaluated and found to have a potency of 30 ± 9 nM. Replacement of the oxygen atom by a sulfur atom (9) resulted in an increase in potency at MORs (EC₅₀ = 9 ± 2 nM). Because this compound was more potent than the benzofuran derivative, thiophene analogues were explored to determine if the fused benzene rings were crucial for activity or if they represented unnecessary steric bulk. The resulting 2- and 3-thiophene derivatives, 10 and 11 respectively, were the most potent heterocyclic analogues evaluated and had activities similar to 1. To probe the directionality of the possible H-bond acceptor capabilities of the sulfur and oxygen atoms of these analogues, the corresponding oxazole and thiazole derivatives were also synthesized. The 5-oxazole 12 and 5-thiazole 13 were active, with EC₅₀ values nearly identical to their larger benzofuran 8 and thianapthene 9 counterparts. Interestingly, both the 4-oxazole 14 and the 4-thiazole 15 were completely inactive up to the highest dose tested, 10 μM. These results indicate that while steric bulk of the 5-membered heterocyclic ring is tolerated, modifications to the directionality of the heteroatom’s H-bond accepting abilities are not. These results, as well as the reduced activity of 2-pyridyl analogue 4 suggests that there may be some steric repulsion between the carbonyl oxygen and an adjacent nitrogen atom creating an unfavorable binding pose. This provides another example of the influence of heterocycle topology and electronic properties on conformational preferences.³²

Analogues with differing linkers between the ester and phenyl moieties were also explored to further probe the depth of the binding pocket (16 – 18). Previous SAR studies indicated that potency could be modulated by altering the linker between the carbonyl and the aromatic ring.^33-34 The alkyl linkers 16 and 17 allow for increased flexibility of the phenyl ring and result in over 100-fold decrease in MOR activity. The ethylene linker in 18 is tolerated at the MOR but results in 25-fold loss in activity, with an EC₅₀ value of 30 ± 10 nM. These results indicate that the MOR binding pocket does allow for this extension of the phenyl ring, but these analogues do not produce any beneficial interactions with the receptor.

Some of the most active compounds that we initially evaluated were phenyl derivatives substituted with methoxy- or fluoro-substituents.²⁶ None of these compounds were more potent than 1. However, their activities warranted further probing of the H-bond characteristics in the receptor binding pocket. To allow for both H-bond donating and accepting, we evaluated the 3’- and 4’-phenolic derivatives (19 and 20, respectively). This strategy identified an analogue (20) with a comparable potency to 1 (20: EC₅₀ = 0.83 ± 0.04 nM, vs. 1: EC₅₀ = 1.2 ± 0.6 nM).

Having identified 20, we sought to further modulate the pK_a and H-bond donor capabilities of the compound by installing halogens adjacent to the phenol. Similar to our initial results with the mono-substituted derivatives, the 3’F, 4’OH-substitution (22a) was the most potent, with an EC₅₀ value of 0.6 ± 0.2 nM. The trend among these halogenated phenols follows the inverse of intramolecular halogen-hydrogen bond strength, with the 3’F, 4’OH- (22a) being the strongest, and the 3’Br, 4’OH- (22c) being the weakest of the three. These trends suggest that the halogens could be impacting interactions between the phenol and the receptor either by coordinating the phenolic hydrogen³⁵ or by lowering the pK_a of the phenol. Alternatively, the trend could also be explained by the increased bulk at the 3’-position. Further exploration of this interaction was accomplished by evaluating some of the MOM-protected phenols. Interestingly, 21a is just over 5-fold less active than its phenolic counterpart 22a, while 21c is significantly less active at MOR than 22c. These results indicate that the binding site may not accommodate the added mass of the MOM-protection well and that sterically smaller 3’-substituents are preferred (3’-F).

Comparison of the activities of the free phenols 19 and 20 to the corresponding methoxy analogues²⁶ suggested that the H-bond donating properties of the phenolic derivatives may be beneficial for MOR activity. In an effort to further probe the H-bond capabilities of the oxygen in this position, we evaluated benzofuran derivatives 23 and 24 as ring-locked derivatives of the methoxy analogues with locked conformations of H-bond acceptors.³⁶ Although no preference was seen between the 3’-methoxy (EC₅₀ = 11 ± 4 nM) and 4’-methoxy derivatives (EC₅₀ = 8.0 ± 7 nM)²⁶, the 4’OH-derivative was slightly more potent than the 3’OH-derivative (19: EC₅₀ = 1.3 ± 0.3 nM and 20: EC₅₀ = 0.83 ± 0.04 nM, respectively), and the opposite result was seen for their corresponding ring-constrained derivatives. The benzofuran-6-carbonyl 23 was essentially equipotent to its 3’OH-counterpart 19 at 1.5 ± 0.6 nM, but the benzofuran-5-carbonyl 24 was just over 10-fold less potent at 9 ± 2 nM than the 4’OH 20.

Based on the activity of the phenols, we decided to extend the alcohol moiety in order to probe the depth of the binding pocket through the addition of a methylene spacer. This derivatization also modified the analogues’ electronic properties by preventing the possible ionization present in the phenolic analogues. Strikingly, the 4’-CH₂OH derivative 25 was over 25-fold more potent than the corresponding phenol 4’-OH (20) (EC₅₀ = 0.03 ± 0.01 nM vs. EC₅₀ = 0.83 ± 0.04 nM). Due to this dramatic increase in potency, additional substitutions were probed to determine what functionalities are tolerated in the pocket, as well as what properties were mediating this increased potency. The 3’-benzylic alcohol 26 (EC₅₀ = 2.42 ± 0.07 nM) is not as potent as 20 or the 3’-phenolic compound 19 (EC₅₀ = 1.3 ± 0.3 nM), suggesting that the alcohol of the 4’-benzylic alcohol derivative is interacting with a residue in the back of the pocket and not in space that the alcohol of the 3’-derivative could reach. Additionally, methylating 25 (27) dramatically reduced potency 433-fold (EC₅₀ = 13 ± 3 nM vs. EC₅₀ = 0.03 ± 0.01 nM). This decrease in potency indicates that either the H-bond donating ability of the benzyl alcohol is important or the additional methyl is not tolerated due to steric hindrance.

Following the initial evaluation, analogues with MOR EC₅₀ values below 200 nM were also evaluated at KORs to determine their selectivity. Previous structural-activity relationship studies indicate that it is possible to gain appreciable activity at KORs. However, there are very few neoclerodanes that possess activity less than 100 nM at δ opioid receptors. The majority of tested compounds showed no activity at KORs with the exception of 8, 16, and 22b. Even with agonist activity at KORs, the activity of 8 and 22b were weak and they ultimately still maintain good selectivity for MORs over KORs. On the other hand, the introduction of the methylene linker between the C2 carbonyl and the aromatic ring (16) result in a loss of selectivity for MORs over KORs.

Analogues were additionally evaluated for their ability to recruit β-arrestin-2 through MOR activation using the previously described enzyme fragment complementation assay.²⁶ The functional selectivity of these compounds was evaluated by determining their bias factors (see Methods for details). As the standard to which all data is normalized, DAMGO has a bias factor value of 1, indicating no signaling bias, factors less than 1 indicate G-protein bias, and factors greater than 1 indicate β-arrestin-2 bias. Analogues were chosen for β-arrestin-2 recruitment analysis based upon the SAR seen in the G-protein mediated cAMP assay, and all analogues with potencies below 10 nM were evaluated. Additionally, compound 15 was evaluated due to its complete inactivity in the cAMP assay despite high structural similarity to potent analogue 13, and compound 18 was evaluated to determine if the ethylene linker between the phenyl and ester moieties resulted in a significant conformational change in the receptor to induce signaling bias.

The analogues evaluated in the cAMP assay recruited β-arrestin-2 to differing extents, while the inactive compound at MOR (15) failed to recruit β-arrestin-2. In contrast to the β-arrestin-2 biased 3’-unstubstituted phenolic compound 20, 3’-halogenated phenolic compounds gradually turned from slightly β-arrestin-2 biased (22a, 3’F, 4’OH) to slightly G protein biased (22c, 3’Br, 4’OH). This indicates the size and properties of the 3-substituent significantly affect the β-arrestin-2 recruitment of the analogues. The MOM-protected phenol 21a is only a weak recruiter of β-arrestin-2 and 21c does not recruit β-arrestin-2 at all, which indicates that the binding pocket of the receptor in the β-arrestin-2 recruitment conformation does not tolerate the steric bulk of both a 3’-substituent and the large 4’-OMOM. Conversely, 18 with the ethylene-linker has the highest bias factor of all analogues evaluated, at 22.5, suggesting that the spacer between the ester and phenyl moieties may allow the compound to adjust its orientation to allow for better accommodation by the different receptor conformation. The only other analogue with a similar bias factor to the MOM-protected phenols is 25, but due to its high potency in the cAMP assay, its bias factor is 0.14 despite its potent β-arrestin-2 recruitment of 14 ± 1 nM. Overall, the SAR trends between β-arrestin-2 recruitment and G-protein activation are different and indicate that compounds with larger 4’-substituents may have the potential to be biased activators of the MOR-associated G-protein pathway.

To determine the efficiency of the activity of these neoclerodanes, their physicochemical properties were calculated and the metrics of ligand efficiency (LE) and lipophilic-adjusted ligand efficiency (LELP) were calculated based on their EC₅₀ values in the G-protein pathway (Supplemental Table 1). All analogues had topological polar surface areas (tPSAs) over 100 Ų and cLogP values over the ideal 2.5,³⁷ highlighting the need for further probing of the physicochemical properties of these analogues. Analogue 25 was the most efficient analogue with an LE value of 0.38 and an LELP value of 9.49.

Previous work in vivo demonstrated the ability of 1 to induce centrally-mediated analgesia, equipotent to morphine with significantly reduced levels of tolerance, sedation, and reward compared to morphine.²⁶ Therefore, given the increase in potency, and G-protein bias of 25 at MOR in comparison to 1 in vitro, we selected 25 to evaluate analgesic effects in vivo.

To assess the induction of centrally-mediated nociception, we performed the hot water (50°C) tail-flick assay in male C57 Bl/6 mice with a 10 s time cutoff to prevent tissue injury. In this spinally-mediated reflex, 25 (5 mg/kg, i.p.) showed a rapid onset of action with significant antinociceptive effects observed at 5 min and maximal effects at 30 min, similar to morphine and 1. However, the overall duration of action of 25 (5 mg/kg, i.p.) was less than morphine (10 mg/kg, i.p.) and 1 (10 mg/kg) (Figure 1). Two-way repeated measures ANOVA revealed significant effects of time F(9,243) = 106.7 p < 0.0001, and drug F(5,27) = 180.3, p < 0.0001, with a significant interaction observed between the effects of drug and time F(45, 243) = 20.65, p < 0.0001. The antinociceptive effects of 25 were inhibited by prior administration of the selective MOR antagonist β-FNA (5 mg/kg, s.c.). Similarly, the antinociceptive effects of 1 were inhibited by prior administration of the opioid antagonist naloxone (10 mg/kg, s.c.) (Figure 1a and b). As predicted by its in vitro potency, 25 is more potent than 1 in vivo and are dependent upon MOR activation to produce antinociceptive effects.

Figure 1.

(a) Time course of analgesic effects in the hot water tail-flick assay of morphine (10 mg/kg, i.p.), 25 (5 mg/kg, i.p.), 25 following pre-administration of β-FNA (5 mg/kg, s.c), 1 (10 mg/kg, i.p.), and 1 following pre-administration of naloxone (10 mg/kg/s.c). (b) Area under the curve analysis, *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001, drug compared to vehicle, #p<0.05, ##p<0.01, ###p<0.001, ####p<0.0001 compared to morphine, ^p<0.05, ^^^^p<0.0001 25 or 1 compared to 25 or 1 + antagonist. Data shown as mean ± SEM (N = 5-7 per group).

The hot water tail-flick assay was further utilized to assess the dose-dependent antinociceptive effects in a cumulative, within animal design on days 1 and 9. Nonlinear regression analysis (four parameter) was used to calculate ED₅₀ values. One-way ANOVA followed by Bonferroni’s multiple comparisons test revealed significant differences in the ED₅₀ values between morphine, 25, and 1, with 25 (ED₅₀ = 2.35 ± 0.25 mg/kg) (mean ± SEM) significantly more potent than morphine (ED₅₀ = 6.43 ± 0.27 mg/kg) and 1 (ED₅₀ = 5 ± 0.51 mg/kg) on day 1 (Figure 2a and 2b, closed symbols). Antinociceptive tolerance is a major limiting factor of current opioid treatments, therefore, administration of 25 (5 mg/kg, s.c.), 1 (10 mg/kg, s.c.), or morphine (10 mg/kg, s.c.) daily on days 2-8 was used to induce chronic tolerance with a second dose response performed on day 9 (open symbols). On day 9, morphine showed a significant induction of tolerance (ED₅₀ 16.35 ± 0.75 mg/kg) compared to day 1, whereas 1 showed reduced tolerance compared to morphine (ED₅₀ 7.9 ± 0.11 mg/kg). However, 25 (ED₅₀ 3.60 ±0.19 mg/kg) showed no significant tolerance effect, demonstrating significant improvement in pharmacodynamics compared to 1.

Figure 2.

(a) (b) Antinociceptive dose-response effects of morphine (red), 25 (blue), and 1 (green) on day 1 (closed symbols) and day 9 (open symbols) following daily administration of morphine (10 mg/kg/s.c.), 25 (5 mg/kg/s.c.), or 1 (10 mg/kg/s.c) in the hot water tail-flick assay in mice. 25 was more potent than morphine on day 1, and a significant decrease in potency was seen with morphine between days 1 and 9. Analgesic effects of 25 on days 1 and 9 showed no significant differences in potency. 1 shared similar potency on day 1 with morphine. On day 9, 1 displayed a significantly reduced tolerance compared to morphine. (c) Onset of tolerance effects in the hot water tail-flick assay was observed with morphine from day 5, with no tolerance observed with 25, ####p<0.0001 morphine compared to 25. Data shown as mean ± SEM. (N = 8 per group).

The onset of tolerance was also evaluated on days 3, 5, 7 and 9 utilizing the hot-water tail-flick assay. Two-way repeated measures ANOVA revealed significant effects of drug F(2, 21) = 557.4, p < 0.0001, time F(2, 21) = 557.4, p < 0.0001, and interaction between time and drug F(8, 84) = 25.71 , p < 0.0001 (Figure 2c). Bonferroni’s multiple comparisons tests showed significant tolerance following morphine from day 5, whereas 25 showed no significant tolerance during the 9 days evaluated. Both 1 and 25 have reduced tolerance compared to morphine. Additional comparison of the behavioral effects of 1 and 25 is currently underway and will be reported in due course.

CONCLUSIONS

Kurkinorin’s desirable in vivo effects led us to explore additional structural modifications of this scaffold. Through the present investigation, we have identified structural trends required for activating MOR-associated G-proteins, as well as for recruiting β-arrestin-2 upon MOR activation. We have identified 25, a non-nitrogenous, non-ionizable compound that is a potent and biased activator of MOR-associated G-protein pathway with an EC₅₀ value of 0.03 ± 0.01 nM and a bias factor of 0.14. In vivo tests revealed the potent analgesic effects of 25, without significant tolerance effects. In all, these data suggest that additional structural investigations may identify analogues of 25 with more drug-like properties. Such studies are under way and will be discussed in due course.

METHODS

General Experimental Procedures.

Reactions performed in standard glassware were performed under an atmosphere of argon using glassware dried overnight in an oven at 120 °C and cooled under a stream of argon. Salvinorin A was isolated from the leaves of Salvia divinorum and converted to salvinorin B (2) and then 3a and 3b as previously described.^{26, 38} All other chemical reagents were purchased from commercial suppliers and used without further purification. All solvents were obtained from a solvent purification system in which solvent was passed through two columns of activated alumina under argon. Reactions were monitored by thin-layer chromatography (TLC) on 0.25 mm Analtech GHLF silica gel plates and visualized using a UV Lamp (254 nm) and vanillin solution. Flash column chromatography was performed on silica gel (4-63 mm) from Sorbent Technologies. ¹H and ¹³C NMR were recorded on a 500 MHz Bruker AVIII spectrometer equipped with a cryogenically-cooled carbon observe probe or a 400 MHz Bruker AVIIIHD spectrometer using tetramethyl silane as an internal standard. Chemical shifts (δ) are reported in ppm and coupling constants (J) are reported in Hz. High-resolution mass spectrum (HRMS) was performed on a LCT Premier (Micromass Ltd., Manchester UK) time of flight mass spectrometer with an electrospray ion source in either positive or negative mode. Melting points were measured on with a Thomas Capillary Melting Point Apparatus and are uncorrected. Preparative HPLC was carried out on an Agilent 1100 series HPLC system with diode array detection at 209 nm on a Phenomenex Luna C18 column (250 x 21.20 mm, 5 mm). Compounds were identified as ≥95% pure by HPLC before testing, unless noted otherwise, via one of three methods: A) Agilent 1100 series with diode array detection at 209 nm on a Phenomenex Luna C18 column (250 x 10 mm, 5 micron); B) Agilent 1100 series with diode array detection at 209 nm on an Phenomenex Luna C18 column (150 x 4.6 mm, 5 micron); or C) Waters Acquity UPLC with a photodiode array UV detector and an LCT Premiere TOF mass spectrometer using a Waters Acquity HSS T3 C-18 (2.1 x 50mm, 1.8um).

General coupling procedure.

An oven-dried flask was charged with 3a and 3b (mixture) (40 mg, 0.103 mmol), EDC·HCl (29.5 mg, 0.154 mmol), DMAP (18.8 mg, 0.154 mmol), and the appropriate benzoic acid (0.154 mmol). To the flask was added CH₂Cl₂ (8 mL). After stirring overnight at r.t. the reaction was quenched with HCl (1 M, 8 mL) and the organic layer rinsed sequentially with saturated NaHCO₃ (8 mL) and brine (8 mL) then dried over Na₂SO₄. The solvent was evaporated and the resulting residue purified by FCC (30-35% EtOAc/Pentane). Compounds <95% pure as indicated by HPLC were further purified by reverse phase semi-preparatory HPLC.

General procedure for deprotection of MOM-protected phenol coupling products.

To a dry flask under Ar was added MOM-protected coupling product (0.07 mmol), carbon tetrabromide (CBr₄) (0.035 mmol, 0.5 eq.), triphenylphosphine (PPh₃) (0.035 mmol, 0.5 eq.), and dichloroethane (3mL).²⁸ Reaction was heated to 40 °C overnight. If reaction was not complete as monitored by TLC the following day, a second addition of 0.5eq. of both CBr₄ and PPh₃ was added and the reaction was allowed to stir at 40 °C for a second night. Upon reaction completion as seen by TLC, the solvent was evaporated and the compound was purified via FCC (40% EtOAc/Pentane).

General procedure for deprotection of TBS-protected benzyl alcohol coupling products.

To a flask containing crude TBS-protected coupling product (0.2 mmol) was added a 1:1 mixture of MeOH and acetone (6 mL total) followed by a solution of KHSO₄ (0.1 mmol, 0.5 eq.) in 3 mL of H₂O.²⁹ Reaction is a white suspension after water addition. Reaction stirred overnight at r.t. Upon reaction completion, as monitored by TLC, the solvent was evaporated and the product was extracted from water (10 mL) into EtOAc (3x 10 mL). The combined organic layers were washed with brine, dried over Na₂SO₄, and solvent was removed. The compound was purified by FCC (50% EtOAc/Pentane).

In vitro pharmacology.

Cell lines and cell culture. The HitHunterTM Chinese hamster ovary cells (CHO-K1) stably expressing the human μ-opioid receptor (OPRM1, catalog # 95-0107C2), κ-opioid receptor (OPRK1, catalog # 95-0088C2), and the PathHunterTM Chinese hamster ovary cells stably expressing the human μ-opioid receptor β-arrestin-2 EFC cell line (catalog # 93-0213C2) were purchased from DiscoverX Corp. (Fremont, CA) and maintained in F-12 media with 10% fetal bovine serum (Life Technologies, Grand Island, NY), 1% penicillin/streptomycin/ ʟ-glutamine (Life Technologies), and 800 μg/mL Geneticin (Mirus Bio, Madison, WI). The media of the PathHunterTM cells was supplemented with an additional 250 μg/mL Hygromycin B (Mirus Bio). All cells were grown at 37 °C and 5% CO₂ in a humidified incubator.

Forskolin-induced cAMP accumulation.

Assays proceeded as previously described.³⁹ Briefly, the aforementioned HitHunter cell lines were plated at 10,000 cells/well in 384-well tissue culture plates and incubated at 37°C overnight. 5 mM stock solutions of all compounds were generated by dissolution in 100% DMSO (Alfa Aesar, Ward Hill, MA). These stock solutions were used to make 10 serial dilutions in 100% DMSO at a 100x concentration. Assay buffer [Hank’s Buffered Salt Solution (HBSS, Life Technologies) and 10 mM HEPES (Life Technologies)] and forskolin (DiscoverX) were used to dilute the serial dilutions to a working 5x concentration, with a final concentration of 100 μM forskolin and 5% DMSO. The HitHunter cAMP assay for small molecules (DiscoverX) was used according to the manufacturer’s directions using the 5x compound serial dilutions. Luminescence was quantified using the Cytation 5 plate reader and Gen5 software (BioTek, Winooski, VT). Data were normalized to vehicle controls and forskolin controls and were analyzed using nonlinear regression using GraphPad Prism 5.0 (GraphPad, La Jolla, CA).

β-arrestin-2 EFC recruitment assay.

Assays proceeded as previously described.²⁶ Briefly, the aforementioned PathHunter CHO-K1 OPRM1 β-arrestin-2 cell line was plated in 384-well tissue culture plates at 5,000 cells/well and incubated at 37°C overnight. 5 mM stock solutions of all compounds were generated by dissolution in 100% DMSO at a 100x concentration. The stock solutions were used to make 11 serial dilutions in 100% DMSO. Assay buffer was used to dilute the serial dilutions to a 5x working concentration, with a final concentration of 1% DMSO). The PathHunter Detection Kit (DiscoverX) was used according to the manufacturer’s directions using the 5x compound serial dilutions. Luminescence was quantified using the Cytation 5 plate reader and Gen 5 software (BioTex, Winooski, VT). Data were normalized to vehicle (1% DMSO in assay buffer) and the highest doses of DAMGO were used as 100% recruitment and converted to percentages. Bias factors calculated using the Equation 1, as previously described:²⁶

$$\mathit{log}(\mathit{Bias}\mathit{Factor})={\left(\log \left(\frac{Ema{x}_{test}\times EC{50}_{DAMGO}}{EC{50}_{test}\times Ema{x}_{DAMGO}}\right)\right)}_{\beta -arrestin}-{\left(\log \left(\frac{Ema{x}_{test}\times EC{50}_{DAMGO}}{EC{50}_{test}\times Ema{x}_{DAMGO}}\right)\right)}_{cAMP}$$ (Eq.1)

In Vivo Studies.

Adult male C57 BL/6 mice (20-29 g) were housed with a 12 h light/dark cycle, with experiments conducted during the light cycle. All experiments were approved by the Victoria University of Wellington Animal Ethics Committee and were carried out per their guidelines for animal care. Food and water were provided ad libitum outside of experimental procedures.

Hot Water Tail-flick Assay.

The hot water (50°C ± 0.5) tail-flick assay was carried out as previously described using a cut off time of 10 s.^{26, 40} Briefly, mice were habituated to the restrainer for 15 min over 3 days. On the test day baseline measurements were recorded in triplicate and the average tail-flick latency recorded. Latencies were measured at 1, 5, 10, 15, 30, 45, 60, 90, and 120 min post-intraperitoneal (i.p.) injection of 25, Morphine Sulfate (Hospira, New Zealand Limited), or vehicle (2:1:7; Dimethyl Sulfoxide (DMSO), Tween-80, Saline). To evaluate whether the analgesic effects of 25 were MOR mediated, mice were pretreated with the selective MOR antagonist beta-funaltrexamine (β-FNA) (Tocris) (5 mg/kg, s.c.) 24 hrs prior to administration of 25 (5 mg/kg, i.p.). The maximum possible effect (MPE) was calculated as follows: % MPE = 100 × (test latency − control latency) / (10 − control latency).

To determine dose-response effects and analgesic tolerance, a within animal cumulative dose-response tail-flick assay was carried out on day 1 and following 9 days of subcutaneous (s.c.) administration of morphine (10 mg/kg) or 25 (5 mg/kg) as previously described.^{17, 26} The induction of tolerance was also observed on days 3, 5, 7 and 9 by measuring tail-flick latencies 30 min post-injection.

Table 2:

MOR Pharmacology: MOR-mediated β-arrestin recruitment.

Compound	EC50 ± SEMa (nM) (% Efficacy)b	Bias Factorc
DAMGO	42 ± 5 (97)	1.0
Morphine	380 ± 40 (38)	0.36
Fentanyl	38 ± 2 (70)	0.34
1	140 ± 40 (96)	0.57
5	190 ± 20 (61)	0.76
6	120 ± 30 (71)	2.71
7	30 ± 10 (76)	4.42
9	700 ± 100 (90)	0.90
10	260 ± 30 (84)	0.33
11	46.2 ± 0.1 (84)	1.69
13	150 ± 10 (68)	1.64
15	>25 000 (0)	--
18	80 ± 10 (71)	22.5
19	40 ± 10 (87)	1.98
20	21 ± 4 (73)	2.13
21a	1300 ± 80 (80)	0.15
21c	>25 000 (0)	--
22a	24 ± 3 (79)	1.40
22b	180 ± 10 (69)	1.04
22c	1600 ± 300 (72)	0.57
23	40 ± 10 (90)	2.62
24	490 ± 80 (84)	1.21
25	14 ± 1 (81)	0.14
26	150 ± 30 (75)	0.87
27	360 ± 80 (74)	1.98

^aMean ± standard error of the mean; n ≥ 3 individual experiments run in triplicate.

^bMaximum efficacy values calculated based on DAMGO maximum stimulation.

^cBias factors were calculated using Eq. 1. Values <1 indicate bias towards the cAMP pathway and values >1 indicate bias towards the β-arrestin-2 pathway. DAMGO is the reference compound, with a bias = 1.

ABBREVIATIONS

E_max

maximum efficacy

KOR

kappa opioid receptor

MOM

methoxymethyl

MOR

mu opioid receptor

MPE

maximal possible effect

SAR

structure-activity relationship

SEM

standard error of the mean

TBS

tert-butyldimethylsilyl