Pharmacological Characterization of Dezocine, a Potent Analgesic Acting as a κ Partial Agonist and μ Partial Agonist

Dezocine is becoming dominated in China market for relieving moderate to severe pain. It is believed that Dezocine’s clinical efficacy and little chance to provoke adverse events during the therapeutic process are mainly attributed to its partial agonist activity at the μ opioid receptor. In the present work, we comprehensively studied the pharmacological characterization of Dezocine and identified that the analgesic effect of Dezocine was a result of action at both the κ and μ opioid receptors. We firstly found that Dezocine displayed preferential binding to μ opioid receptor over κ and δ opioid receptors. Dezocine, on its own, weakly stimulated G protein activation in cells expressing κ and μ receptors, but in the presence of full κ agonist U50,488 H and μ agonist DAMGO, Dezocine inhibited U50,488H- and DAMGO-mediated G protein activation, indicating that Dezocine was a κ partial agonist and μ partial agonist. Then the in intro results were verified by in vivo studies in mice. We observed that Dezocine-produced antinociception was significantly inhibited by κ antagonist nor-BNI and μ antagonist β-FNA pretreatment, indicating that Dezocine-mediated antinociception was via both the κ and μ opioid receptors. When co-administrating of Dezocine with U50,488 H or morphine, Dezocine was capable of inhibiting U50,488H- or morphine-induced antinociception. Finally, κ receptor activation-associated side effect sedation was investigated. We found that Dezocine displayed limited sedative effect with a ceiling effecting at a moderate dose. Thus, our work led to a better understanding of the analgesic mechanism of action of Dezocine in vivo.

Morphine, fentanyl and all related opioid drugs used for relief of moderate and severe pain in clinic are agonists at μ opioid receptors 1,2 . These prototypical μ agonists produced potent analgesia, and also mediated adverse effects including liability to addiction, respiratory depression, constipation and urinary retention, by action at μ opioid receptors 3 . An approach to limiting the undesirable side effects, but still have therapeutic potential as potent analgesics is to develop compounds that possess mixed agonist-antagonist or partial agonist activities at different opioid receptors, since activation of the different opioid receptors may have synergistic effects or may produce fewer side effects than acting at a single target 1,[4][5][6][7] . The well-known mixed agonist/antagonist analgesics include pentazocine, butorphanol, nalbuphine, buprenorphine and Dezocine 1,7 , among which, Dezocine, a synthesized bridged aminotetralin 4,[8][9][10] , currently is a prescribed painkiller in China for moderate to severe pain treatment. In 2016, sales of Dezocine reached more than $630 million nationally, with occupying over 45% of the nation's opioid analgesics market.
Early studies demonstrated that Dezocine produced more potent or equipotent analgesia than morphine in rodents, primates and humans, due to its μ agonist (morphine-like) activity 10 . The evidence came from both in vitro functional studies 11 and in vivo drug antinociceptive 12 and discriminative stimulus studies 13 . Meanwhile, Dezocine exhibited certain μ antagonist activity as showing by the fact that Dezocine produced antagonism on morphine-induced loss of righting reflex, although Dezocine did not significantly precipitate withdrawal jumping syndrome in chronically morphine-treatment mice 10 . Compared with opioid analgesics acting as μ full agonists such as morphine, Dezocine has a safe and tolerable side-effects profile, including relatively low analgesic tolerance 14 , lack of physical dependence capacity 10,15 and limited respiratory depression 10,15,16 . According to these observations, Dezocine seems to perform as a μ partial agonist, which can explain its unique profile. Indeed, Liu with his colleagues recently characterized molecular targets for Dezocine and provided reliable evidence supporting that Dezocine acted as a μ partial agonist 17 . In addition to μ opioid receptors, Dezocine also interacted with κ opioid receptors, however, the profile of Dezocine at κ receptors remained controversial. Due to its structurally related to Pentazocine, a κ agonist and μ partial agonist, Dezocine was initially identified as a κ agonist 4 . Subsequent studies from Gharagozlou et al. (2006) and Liu et al. (2014) by using in vitro functional assay demonstrated that Dezocine acted as a κ antagonist 17,18 . At present, in vivo data regarding to its activity at κ opioid receptor is lacking. Whether Dezocine-induced analgesia was partially a result of action at κ receptor and κ receptor-associated side effects were unknown.
Thus, the present work was undertaken to comprehensively address the pharmacological properties of Dezocine. We found that Dezocine displayed preferential binding to μ opioid receptor over κ and δ opioid receptors, and acted as a κ partial agonist and μ partial agonist, evaluated by functional [ 35 S]GTPγS binding assay. In vivo studies, Dezocine produced more potent antinociception than morphine, which was via activation of both the κ and μ opioid receptors. Dezocine was capable of inhibiting κ agonist-and μ agonist-induced antinociception. In addition, Dezocine displayed limited sedative effects with a ceiling effecting at a moderate dose.

Results
Affinity, selectivity and efficacy of Dezocine. Figure 1 22.01 nM and 398.6 nM for μ, κ and δ opioid receptors, respectively. The data indicated that Dezocine displayed preferential binding to μ opioid receptors over κ and δ opioid receptors.
To characterize the efficacy and potency of Dezocine, the [ 35 S]GTPγS binding assay was conducted. As shown in Fig. 3A-C and Table 2, DAMGO and U50,488 H, as reference compounds, exhibited full agonist efficacy for G protein activation by μ and κ opioid receptors, but Dezocine-stimulated [ 35 S]GTPγS was 33.6% and 45.8% of the maximal stimulation obtained with U50,488 H and DAMGO, respectively. As shown in Fig. 4 and Table 3, when co-incubated with U50,488 H, Dezocine caused a rightward shift in the concentration-response curves with a Ke value of 324.1 nM. When co-incubated with DAMGO, Dezocine also caused a rightward shift with a Ke value of 10.4 nM. These results suggested that Dezocine had more potent antagonist action against μ agonist than that against κ agonist. All together, the data obtained from in vitro functional assay indicated Dezocine acted as a κ partial agonist and μ partial agonist.
Antinociceptive effects of Dezocine. The antinociceptive property of Dezocine was characterized in different acute pain models. In the hot-plate test, a bell-shaped dose-response curve was obtained with Dezocine ( Fig. 5A). The maximal antinociceptive effect was around 60% at 7.5 mg/kg, and higher dose (10 mg/kg) was not able to further increase the antinociception. Thus, the estimated antinociceptive ED 50 of Dezocine in the hot-plate test was not obtained. To improve thermal antinociceptive efficacy, different administration routes, i.v., s.c., or p.o. were carried out. However, whatever the route of administration, Dezocine produced maximal antinociception around 55% (Fig. 5B). A fast onset of action was observed at 5 min, and the consistent high (close to peak) antinociceptive response up to 1 h was observed (Fig. 5C).
In the abdominal constriction test, Dezocine produced a full-dose response curve with an ED 50 value of 0.2 mg/kg (Fig. 5D, Table 3), which was 4-fold more potent than morphine (ED 50 : 0.8 mg/kg, Table 3). In the formalin test, Dezocine also produced a full dose-response curve in both phase I and phase II, with ED 50 values of 0.4 mg/kg (Fig. 5E, Table 3), which were approximately 13-fold more potent than morphine (ED 50 : 5.1 mg/kg) in the phase I and 9-fold more potent than morphine in the phase II (ED 50 : 3.1 mg/kg).
Agonist properties of Dezocine. Next, the agonist effect produced by Dezocine in the abdominal constriction test was determined by the use of selective antagonists. As shown in Fig. 6, one-way ANOVA assay revealed that the treatments induced significant changes [F(2,27) = 4.867, p < 0.05]. The following Tukey's comparison test showed that κ antagonist nor-BNI (p < 0.05) and μ antagonist β-FNA (p < 0.05) significantly Antagonist properties of Dezocine. Our in vitro data indicated that Dezocine exhibited partial agonist activity at μ and κ opioid receptors, suggesting that Dezocine could act as a κ and μ antagonist, in addition to its agonist activity. To assess the antagonist activity of Dezocine, Dezocine and κ agonist U50,488 H or μ agonist morphine were administered concomitantly. As shown in Fig. 7A, Dezocine pretreatment did not inhibit U50,488H-induced antinociception. However, in the presence of μ antagonist β-FNA, Dezocine significantly decreased the antinociceptive effects produced by U50,488 H (p < 0.05, Fig. 7B). We further found Dezocine, at the dose of 0.05 mg/kg, significantly decreased the antinociceptive effects produced by 0.4 mg/kg morphine (p < 0.05), but did not affect the action produced by 0.3 mg/kg morphine (Fig. 8). Co-administration of 0.125 mg/ kg Dezocine and 0.3 mg/kg morphine produced higher antinociception than morphine used alone (p > 0.05, Fig. 8).
Sedative side effect of Dezocine. Activation of κ opioid receptors causes severe side effects of sedation, which is undesirable for therapeutic use. To study the sedative effects of Dezocine, the mice ability to maintain their position on an accelerating rotarod was detected. As shown in Fig. 9, animals were injected with various doses of Dezocine (2.5-25 mg/kg), then the rotarod test was conducted. Dezocine, at a dose of 5 mg/kg, produced significant sedation (p < 0.05), when the dose increased (7.5 and 25 mg/kg), the sedative effect diminished. There was a trend that 2.5 mg/kg Dezocine caused sedation, but it did not reach significant difference (p = 0.0881).

Discussion
Our in vitro binding studies indicated that Dezocine produced high affinity at μ opioid receptor with Ki value of 1.46 nM, which was 15-and 273-fold selectivity for μ over κ and δ opioid receptors. These binding affinity results of Dezocine were consistent with the published data 17,19 . By using functional [ 35 S]GTPγS binding assay, we demonstrated that Dezocine acted as a μ partial agonist, which was also consistent with the well-known low activity of Dezocine to stimulate μ receptors 8 . We further demonstrated that Dezocine was a κ partial agonist, which was different from the existing literatures showing that Dezocine acted as a κ antagonist 17,18 . We found that Dezocine produced maximal stimulation of [ 35 S]GTPγS binding up to 33-45% mediated by κ receptors. Meanwhile, Dezocine also exhibited inhibitory effect on κ agonist U50,488H-stimulated [ 35 S]GTPγS binding. Thus, Dezocine was termed by us as a κ partial agonist because, on its own, it had a lower intrinsic activity than full κ agonist U50,488 H, but in the presence of U50,488 H, it acted as a weak antagonist by preventing access to the κ receptor. Existing literature indicated that the functional action of Aripiprazole, a partial agonist at Dopamine D2 receptors, might vary among agonist, partial agonist or antagonist depending upon the level of receptor expression, function examined, and assessment assay applied 20,21 .  Table 1.   Table 2.    Table 2.  Considering that in vitro assay may not accurately reflect the action of Dezocine in vivo, the behavioral experiments in mice were conducted. In the abdominal constriction and formalin tests, we found that Dezocine produced full dose-response curves and displayed more potent antinociceptive response than morphine. In agreement, Dezocine was proved substantially more potent than morphine and pentazocine in the rodents and monkey 4 . In contrast to this sigmoid curve, in the hot plate test (55 °C), Dezocine-induced antinociception exhibited a bell-shaped dose-response curve. The similar bell-shaped dose-response curve was displayed in the antinociceptive action of pentazocine and buprenprphine [22][23][24][25][26] . Buprenorphin acted as κ antagonist 27 and μ partial agonist 23,24,28 and pentazocine acts as a κ agonist and μ partial agonist 29 . Thus, it seems that the plausible explanation for the observed ceiling effect on antinociception is that these compounds behave as μ partial agonists. The exact rationale for a ceiling effect of Dezocine-produced antinociception in the hot-plate test was unclear. but the possibility was suggested by data developed by the Lutfy group 25 , showing that the bell-shaped dose-response curve of buprenorphine was eliminated by the ORL-1 receptor antagonist and speculated that activation of ORL-1 receptors compromised the receptor-mediated antinociceptive action of buprenorphine. Shu et al. (2011) found that nor-BNI enhanced the antinociception by pentazocine and turned the later descending portion of  Table 3. Data are presented as mean ± SEM from at least 10 mice.    We further found that the antinociceptive action of Dezocine was inhibited by κ antagonist nor-BNI and μ antagonist β-FNA, supporting our in vitro conclusion that Dezocine had κ and μ agonist activity. It is believed that Dezocine's clinical efficacy and little chance to provoke adverse events during the therapeutic process are mainly attributed to its partial agonist activity at the μ opioid receptors 30 . However, we demonstrated that the antinociceptive response of Dezocine was a result of action at both the κ and μ opioid receptors. Liu et al. (2014) demonstrated that in addition to opioid receptors, Dezocine also targeted norepinephrine and serotonin reuptake 17 . Wang et al. (2017) further found that Dezocine produced antihypersensitivity activity in neuropathic pain through its μ activation and norepinephrine reuptake inhibition (NRI) 31 . The activity of Dezocine at non-opioid sites in acute pain modulation was not investigated in the present work.
To assess the antagonist activity of Dezocine at the κ receptor in vivo, the animals were pretreated with Dezocine before giving a κ agonist U50,488 H. We noticed that Dezocine did not inhibit the antinociceptive effect of U50,488 H. A possible explanation for this finding was that Dezocine-suppressed U50,488 H action was compromised by its own μ agonist activity. Indeed, in the presence of μ antagonist β-FNA, Dezocine exhibited κ antagonist activity as demonstrated by its ability to decrease U50,488H-induced antinociception. On the other hand, we found that 0.05 mg/kg Dezocine significantly inhibited μ agonist morphine-induced antinociception, indicating its μ antagonist activity. Whereas co-administration of 0.125 mg/kg Dezocine with morphine produced higher antinociception than morphine used alone. This phenomenon can also be rationalized as resulting from Dezocine-induced κ receptor activation. Thus, our results may explain the controversial findings within the literature concerning the benefit of Dezocine and other opioid analgesic drugs in combination. Dezocine, as a μ partial agonist, was able to antagonize morphine analgesia 32 , but due to its κ agonist activity, combination of Dezocine and morphine might produce higher analgesic effects than morphine alone 15,33 . The action of Dezocine may vary depend on the dosage used. The accurate interactions between Dezocine and morphine or other opioids need further investigation.
The partial agonist activity is generally believed to support it safety, such as buprenorphine displays a ceiling effect on μ receptor-mediated respiratory depression 4,16 . Hence, we assume that Dezocine can produce lower κ receptor activation-associated sedative side effects. Indeed, in the present work, we found that Dezocine exhibited limited sedative effects with a ceiling effecting at 5 mg/kg. Moreover, attributed to the fact that the ED 50 values of Dezocine-induced antinociception was around 0.2-0.4 mg/kg, the dose of Dezocine that produced sedative side effects (5 mg/kg) was much higher than those to produce antinociception. Thus, the partial agonist activity of Dezocine supports it unique property for clinical application with less side effects.
Taken together, Dezocine was demonstrated as a κ partial agonist and μ partial agonist. In mice model of acute pain, Dezocine-produced antinociception was mediated by both the κ and μ receptor activation. Dezocine displayed limited sedative effects with a ceiling effecting at a moderate dose. Combined with other opioid drugs, the action of Dezocine may vary depend on the dosage used. In vitro studies. Cell culture. Chinese hamster ovary (CHO) cells stably expressing human κ-, rat μ, or rat δ-opioid receptors were maintained in F12 medium (Gibco) with 10% fetal serum and 0.25 mg/ml (G418) (Roche). Cells were incubated in a humidified atmosphere consisting 95% air and 5% CO 2 at 37 °C, and were seeded in 10 cm dishes. Cells were collected when reached 85% confluence, and the cell membrane was prepared.

Methods
Cell membrane preparation. Cells were washed with phosphate buffer saline (PBS), detached by incubation with PBS containing 1 mM EDTA, and centrifuged at 1000 g for 10 min. The cell pellet was suspended and homogenized in ice-cold homogenization buffer (50 mM HEPES, 1 mM MgCl 2 , 1 mM EGTA, pH 7.4,). After centrifugation at 40,000 g for 10 min (4 °C), pellets were resuspended, homogenized and centrifuged again. The final pellets were resuspended in Tris-HCl buffer (50 mM, pH 7.4). Protein concentration was detected and the aliquots were stored at −80 °C.
Receptor binding assay. Ligand binding assay was carried out according to our work 34,35   The potency of an antagonist is defined by its Ke (equilibrium dissociation constant) value, as determined by the "single" dose method. The agonists (U50, 488H and DAMGO) dose-response curves were generated in the absence and presence of Dezocine. Formalin test. The formalin test was performed according to our previous work 35 . In brief, the amount of time that mice spent licking or flinching the injected hind paw was recorded after injected with 20 μl 1.0% formalin at the plantar surface of the right hind paw. The mice were observed fro 60 min. The first 10 min was recorded as the phase I, and the rest of time was recorded as phase II. Percentage of analgesia was calculated as 100× (time of control -time of drug)/time of control.
Antagonist effect of Dezocine. To define the κ antagonist profile of Dezocine, mice were pretreated with various doses of Dezocine or saline. 5 min later, the mice were administrated with κ agonist U50,488 H (1.2 mg/kg, i.p). After 15 min, the antinociceptive effect was assessed. To avoid Dezocine's μ agonist activity contributing to its antinociception, mice were pretreated with μ antagonist β-FNA before Dezocine and U50,488 H co-administration. The μ antagonist effect of Dezocine was also determined by pretreatment with Dezocine followed by μ agonist morphine injection. After 15 min, the antinociceptive effect was assessed.
Rotorod test. The test was conducted in mice according to the method described previously with slight modification 31 . In brief, mouse was individually trained to maintain their position on a rotorod for 120 s or more, and the mice falling off within 120 s were excluded. 15 min after Dezocine injection, mice were put on the rotorod apparatus, the time duration for each mouse to fall off was recorded.
Statistical Analysis. Curve-fitting analysis was performed using GraphPad Prism 6.0 software. Data represents the mean ± SEM of at least three separate experiments. Statistical significance was determined by one-way ANOVA followed by post hoc comparison using Tukey's tests. When only two groups were compared, statistical significance was determined by unpaired student's t test.

Data Availability Statement
All data generated or analyzed during this study are included in this published article.