Isolation and Pharmacological Characterization of Six Opioidergic Picralima nitida Alkaloids

Abstract

The seeds of the akuamma tree (Picralima nitida) have been used as a traditional treatment for pain and fever. Previous studies have attributed these effects to a series of indole alkaloids found within the seed extracts; however, these pharmacological studies were significantly limited in scope. Herein, an isolation protocol employing pH-zone-refining countercurrent chromatography was developed to provide six of the akuamma alkaloids in high purity and quantities sufficient for more extensive biological evaluation. Five of these alkaloids, akuammine (1), pseudo-akuammigine (3), akuammicine (4), akuammiline (5), and picraline (6) were evaluated against a panel of >40 central nervous system receptors to identify that their primary targets are the opioid receptors. Detailed in vitro investigations revealed 4 to be a potent kappa opioid receptor agonist and three alkaloids (1-3) with micromolar activity at the mu opioid receptor. The mu opioid receptor agonists were further evaluated for analgesic properties but demonstrated limited efficacy in assays of thermal nociception. These findings contradicted previous reports of the antinociceptive properties of the akuamma alkaloids and the traditional use of akuamma seeds as analgesics. Nevertheless, their opioid-preferring activity does suggest the akuamma alkaloids provide distinct scaffolds from which novel opioids with unique pharmacologically properties and therapeutic utility can be developed.

Graphical Abstract

For centuries, morphine and its semi-synthetic derivatives have been integral components of effective pain management.¹ Opioid analgesics produce their powerful pain-killing effects through the activation of mu opioid receptors (μOR), one of three opioid receptor subtypes found throughout the central and peripheral nervous system.^2,3 Despite the effectiveness of opioid analgesics in acute and subacute settings, chronic pain remains an escalating and poorly managed health concern, affecting approximately 20% of adults worldwide.^4,5 Over the past two decades, opioid prescriptions for the treatment of chronic pain have risen dramatically despite their reduced effectiveness against chronic pain states.^{3, 6} Unfortunately, the prolonged use of opioid analgesics elicits numerous adverse effects including respiratory depression, tolerance, and dependence.^1–3 Increased prescription and duration of use of short and long-acting/extended-release opioids, combined with the side effect profile of opioid medications, has led to the current opioid epidemic characterized by >40,000 opioid overdose deaths per year since 2016.⁷ To counter this trend, in 2016 the Center for Disease Control and Prevention provided new guidelines for the use of opioids in patients suffering from chronic pain with a focus on reducing and replacing opioids when possible.⁸ This abrupt change has been effective in decreasing opioid prescriptions, but may have come at the expense of patients that were benefitting from their current opioid therapy and are now left undertreated.^9,10 It is within this setting that patients search for non-prescription alternatives such as Mitragyna speciosa (kratom) and Picralima nitida (akuamma) to self-medicate their pain and opioid withdrawal symptoms.^11,12

Historically, natural products have served as an excellent source of novel scaffolds to initiate drug discovery efforts.¹³ This is particularly true in the arena of pain and other disorders of the nervous system. Beyond the aforementioned analgesics derived from morphine, naturally occurring salicylic acid, capsaicin, and tetrahydrocannabinol have all been exploited for their pain killing effects.^14,15 For centuries, the akuamma tree has been used by natives of Western Africa to treat a variety of ailments including malaria, dysmenorrhea, and gastrointestinal disorders.¹⁶ The seeds in particular have been used for their analgesic and antipyretic properties.¹⁶ Notably, anecdotal reports indicate that, unlike traditional opioid analgesics, akuamma does not elicit euphoria, tolerance, or dependence. The analgesic effects of P. nitida seeds have been generally attributed to a class of indole alkaloids known as the akuamma alkaloids composed of akuammine (1), akuammidine (2), pseudo-akuammigine (3), akuammicine (4), akuammiline (5), and picraline (6).¹⁷ In standard nociception assays, both the ethanolic extract of P. nitida and isolated 3 demonstrated antinociceptive properties. While their potency is lower than morphine, the effects of 3 in these assays appeared to be longer lasting.^18,19

Previous investigations indicated the antinociceptive effects of the akuamma alkaloids are produced through their interaction with the opioid receptors.^16,17 However, the scope of these studies was limited to the opioid receptors, excluded other nervous system receptors, and the rigor of those findings is linked to the suboptimal tools available at the time to study the pharmacology of the akuamma alkaloids. Moreover, several alkaloids found in relatively high abundance in P. nitida have remained uninvestigated. Nevertheless, the structural differences between the akuamma alkaloids and traditional opioid analgesics, paired with the reported mild side effect profile, suggests these alkaloids may exhibit unique signaling properties at the opioid receptors, and, therefore, may be promising starting points for the development of new pain management drugs. Thus, to provide additional insight into the effects of akuamma, studies were initiated to identify an effective preparative purification strategy that permits the resolution of the six main alkaloids found within akuamma seeds to provide a more thorough investigation of their pharmacological activity in vitro and in vivo.

RESULTS AND DISCUSSION

Extraction and Isolation of Akuamma Alkaloids.

To initiate studies of the akuamma alkaloids, an isolation process capable of providing six alkaloids in high purity and quantities sufficient for in vitro and in vivo studies was needed. Initial efforts revealed 1 and 2 could be easily isolated in >95% purity through liquid-liquid extraction and selective crystallization. However, isolation of the other major alkaloids in this manner proved difficult due to their similar solubilities in organic solvents and their general tendency to form critical pairs, particularly at the preparative scale. Previous studies have employed combinations of normal-phase column chromatography, preparative TLC, high performance liquid chromatography (HPLC) and recrystallization to purify alkaloids present in P. nitida extracts.^17,20–22 In our hands, normal phase chromatography resulted in poor separations of the alkaloids due to their remarkably similar polarities. Furthermore, irreversible adsorption of the alkaloids to the stationary phase severely decreased yields and was particularly problematic for compounds present in minor quantities.²³ Although semi-preparative reversed-phase HPLC proved to be more effective in terms of compound resolution, severe limitation remained in overall efficiency in terms of time and resources for generating the quantities required for our studies.

The shortcomings of standard chromatography techniques led to the investigation of countercurrent separation (CCS) to purify the P. nitida alkaloids. By eliminating the use of a solid stationary phase in favor of a continuous flow liquid-liquid partitioning system, both major forms of CCS, high speed countercurrent chromatography (HSCCC) and centrifugal partition chromatography (CPC), avoid the irreversible adsorption of compounds observed in solid phase-based liquid chromatography and allow for quantitative sample recovery.^24–26 Specifically, pH-zone-refining countercurrent chromatography (pHZR-CCC), which exploits the acid-base interactions of the two immiscible phases, is perfectly matched for the separation of basic alkaloids.^24–30 The method provides the benefit of a high loading capacity and produces highly concentrated fractions with minimal compound overlap.^24–26 Okunji et. al. have applied pHZR-CCC to the fruit rind of P. nitida, thus it stood to reason that their method could be adapted to isolate alkaloids found in the ground seeds.²³

As with any liquid chromatography method, the selection of a suitable solvent system is critical to the outcome of the isolation. Solvent systems for pHZR-CCC require two immiscible solvents and typically incorporate the addition of co-solvents to modulate the partition coefficients of the compounds of interest. To select an applicable solvent system, a straightforward partitioning experiment originally developed by Ito was employed.²⁴ Five solvent systems comprised of 2-4 solvents, were modified with the addition of either acid (10 mM HCl) or base (10 mM TEA) and the partition coefficients of 3-6 were determined via HPLC to give K_acid and K_base, respectively (Table 1). Suitable K values for basic compounds such as the akuamma alkaloids should fit within the parameters K_acid << 1 and K_base>> 1.

Table 1.

Partition Coefficients of Akuamma Alkaloids in CCS Solvent Systems. ^a

Solvent System		Pseudo-akuammigine (3)	Akuammicine (4)	Akuammiline (5)	Picraline (6)
1:1:1:1 Hex/EtOAc/MeOH/H2O	Kacid	<<0.01	<<0.01	<<0.01	<<0.01
	Kbase	2.71	1.63	0.23	0.82
3:7:3:7 Hex/EtOAc/MeOH/H2O	Kacid	<<0.01	<<0.01	<<0.01	0.01
	Kbase	32.3	28.9	2.53	7.29
1:1 EtOAc/H2O	Kacid	0.10	0.02	0.03	0.03
	Kbase	51.9	222	28.2	16.09
2:2:3 MTBE/CH3CN/H2O	Kacid	2.22	0.61	0.27	0.56
	Kbase	87.0	27.7	5.61	11.4
3:1.5:4 MTBE/CH3CN/H2O	Kacid	0.25	0.14	0.05	0.10
	Kbase	30.9	64.1	6.82	17.4

^aThe K_acid and K_base for each akuamma alkaloid were calculated by taking AUC_{upper phase}/AUC_{lower phase} as observed by HPLC for respective pH conditions.

Generally, pHZR-CCC has relied heavily on hexane-ethyl acetate-methanol-water (HEMWat) and methyl tert-butyl ether (MTBE)-acetonitrile-water solvent systems. In particular, these solvent systems have been successfully employed in the separation of several classes of structurally similar indole alkaloids.^{23, 27–30} Initial investigation revealed that the less polar HEMWat systems were incompatible with the akuamma alkaloids due to the strong retention of compounds in the acidic aqueous phase and unexceptional base partition values for 5 and 6. Turning to the 2:2:3 MTBE/CH₃CN/H₂O system previously employed by Okunji, K_acid>1 was observed for 3.²³ Further alterations to the solvent composition and rations revealed 1:1 EtOAc/H₂O and 3:1.5:4 MTBE/CH₃CN/H₂O systems both proved to be viable solvent systems providing K_acid << 1 and K_base>> 1 for 3-6. Of these two, the 1:1 EtOAc/H₂O system was selected as the most appropriate solvent for pHZR-CCC based on the lower acid partition coefficient values and more consistent base partition coefficient values displayed, in addition to exhibiting a larger difference between the K_acid and K_base of the four alkaloids.

Having identified a promising solvent system, separation of the dichloromethane fraction with pHZR-CCC was first attempted in descending mode using an acidic aqueous mobile phase. Using this method, 4 was first to elute, followed by 3 and subsequently mixed fractions of 5 and 6. As expected, when performing the run in ascending mode with the basic organic layer as the mobile phase, the order of elution was observed to be the exact opposite of the descending method. Reversing the elution mode facilitated concentration of the fractions due to the use of lower boiling point organic solvent and allowed the alkaloids of interest to elute much earlier in the run.

A plot of the pH values for each fraction produced a series of alternating zones of increasing pH and plateaus which is characteristic of the pHZR-CCC. When overlaid onto the UV-Vis chromatograms, the elution of major alkaloids was observed to coincide with a plateau on the pH curve, presumably at the points where the pH is roughly equal to their isoelectric point (Figure 1). While initially fractions were collected in 15 mL volumes, reduction of the volume to 7.5 mL led to significant increases in alkaloid purity. From a 1.2 g sample of the dichloromethane fraction, this process directly provided 130 mg of 3 and 145 mg of 4 in high purity. Despite significant attempts to optimize the solvent system and pHZR-CCC conditions, 5 and 6 consistently co-eluted as ~1:1 mixture. Fortunately, this mixture could be easily separated on silica gel via flash chromatography to provide 61 mg of 5 and 90 mg of 6. Notably, attempts to directly purify the dichloromethane fraction via flash chromatography were unsuccessful due to considerable co-elution of multiple alkaloids, thus highlighting the necessity to first simplify the fraction via pHZR-CCC.

Figure 1.

pH-zone-refining countercurrent chromatography chromatogram of akuamma alkaloid DCM extract. Elution of the alkaloids akuammiline (5) and picraline (6) occurred between high pH 3 to low-mid pH 4 range. Elution of pseudo-akuammigine (3) occurred between mid pH 4 to low pH 5. Elution of akuammicine (4) occurred between low to high pH 5.

Once isolated, the purified alkaloids 1-6 were identified by comparison of the ¹H and ¹³C NMR spectra to literature values (Tables S1–12, Supporting Information).^31–33 In particular, the comparison of the ¹³C NMR chemical shifts to literature values revealed an average absolute difference of 0.24 ppm, with major differences arising from subtle solvent-dependent changes in chemical shifts. In addition to this agreement with literature values, the spectral data are consistent with previously reported structures.

Identification of Major Drug Targets Through the Psychoactive Drug Screening Program (PDSP).

With the akuamma alkaloids in hand, five of the isolated alkaloids were evaluated via the Psychoactive Drug Screening Program (PDSP) to determine possible receptor targets for their purported biological effects. Alkaloids 1 and 3–6 were first assessed at a single concentration (10 μM) for their ability to displace radiolabeled ligands from a diverse panel of human G-protein coupled receptors, ion channels, and transporters. In this primary screen, all five akuamma alkaloids inhibited [³H]-SCH-23390 from the dopaminergic D₅ receptor and [³H]-U69,593 from the κOR (Figure 2). 1, 3, and 4 also displaced [³H]-DAMGO from the μOR, whereas 5 and 6 produced significantly less displacement in these assays. All five of the tested alkaloids produced minimal inhibition of [³H]-DADLE binding at the δOR. Although additional displacement was noted for several of the serotonergic and the histaminergic H₃ receptors these were generally low levels of inhibition (<60%). In contrast to 1, 3, 5, and 6, which appear to be moderately selective for the opioid and D₅ receptors, the primary binding data indicate 4 is considerably more promiscuous.

Figure 2.

Receptor binding profiles of akuamma alkaloids. The akuamma alkaloids akuammine (1), pseudo-akuammigine (3), akuammicine (4), akuammiline (5), and picraline (6) were assessed at 10 μM for their ability to displace radiolabeled ligands from membranes expressing individual receptors. The heatmap represents mean displacement of radioligand from four replicates.

To validate these potential receptor targets, secondary binding experiments were carried out by the PDSP to determine binding affinities (K_i) for each receptor-ligand pair demonstrating >50% inhibition in the primary screen. Notably, all five alkaloids possess a K_i >10 μM at the D₅, indicating a false positive in the primary screen data (Table S13, Supporting Information). Conversely, 1 and 3 possess considerable affinity at the μOR (K_i = 0.76 μM and 1.0 μM, respectively) while 4 and 5 bind with sub-micromolar affinity to the κOR (K_i = 0.17 μM and 0.40 μM, respectively). These data generated from the cloned human opioid receptors are in good agreement with those reported by Menzies et. al. using guinea pig brain homogenates.¹⁷ Furthermore, the data strongly support the hypothesis that any observed biological effects of akuamma likely occur through interactions with the opioid receptors.

Characterization of in vitro Pharmacology of Akuamma Alkaloids.

Having established that the alkaloids target opioids receptors, the affinity and potency of akuamma alkaloids for binding, activation of G-protein and β-arrestin 2 recruitment at the μOR, κOR, and δOR were assessed in cellular assays. In general, the alkaloids had higher affinity and activity at the μOR and κOR, relative to δOR (Figure 3A–I), confirming the results from the PDSP. For all cellular characterization assays, the alkaloids had weaker affinity, potency and efficacy when compared to reference ligands DAMGO, U50,488, and leu-enkephalin (Figure 3A–I). More specifically, at the μOR, 1–3 had the highest binding affinities with K_is of 0.30, 0.32 and 0.59 μM, respectively (Figure 3A, Table 2). The binding affinity for these compounds was reflected in their increased potency and efficacy in the cAMP inhibition assay at the μOR, relative to the other alkaloids, with 1 and 2 producing IC₅₀s of 2.6 and 3.14 μM (Figure 3D, Table 2). Compounds 4-6 exhibited minimal cAMP inhibition at μOR, which is reflective of their relatively lower binding affinity at the receptor (Figure 3A, 4D). The alkaloids had non-determinable β-arrestin 2 recruitment at the μOR, but 2 did show minimal recruitment at the highest concentration tested (Figure 3G). At the κOR, 4 had the highest binding affinity with a K_i of 89 nM, which mirrors its potency in the cAMP assay with an IC₅₀ of 240 nM (Figure 3B, 3E, Table 2). Compounds 1, 3, 5 and 6 all had similar binding affinities at the κOR, while 2 had the least affinity (Figure 3B, Table 2). Notably, 1 did not inhibit cAMP production, suggesting it possesses antagonistic or inverse agonistic properties at the κOR. Within the tested dose-range, the alkaloids minimally recruited β-arrestin at the κOR but followed the general trend that β-arrestin 2 recruitment was most apparent in alkaloids that display the strongest binding affinity (Figure 3H, Table 2). Compared to the μOR and the κOR, binding affinity, as well as potency and efficacy of the compounds in the cAMP inhibition assay, was lower at the δOR (Figure 3C,3F, Table 2). Similar to the other receptors, there was non-determinable β-arrestin recruitment by the alkaloids at the δOR, although 2 did show minimal recruitment at the highest concentration tested (Figure 3I).

Figure 3.

Pharmacological characterization of akuamma alkaloids at μOR, κOR, and δORs. The akuamma alkaloids akuammine (1), akuammidine (2), pseudo-akuammigine (3), akuammicine (4), akuammiline (5), and picraline (6) were characterized for binding affinity using [³H]DAMGO, [³H]U69,593 and [³H]DPDPE (A, B, C), inhibition of forskolin-induced cAMP in a Glo-sensor assay in transfected HEK-293 cells (D, E, F) and the ability of the alkaloids to recruit β-arrestin 2 in a PathHunter assay. (G, H, I) at μOR (A, D, G), κOR (B, E, H), and δOR (C, F, I). All curves are representative of the averaged values from a minimum of 3 independent assays.

Table 2.

Summary of Akuamma Alkaloids In Vitro Characterization at Opioid Receptors.

Compounds	Binding		cAMP			β-arrestin 2
μOR	pKi	Ki (μM)	pIC50	IC50 (μM)	α	pEC50	α
DAMGO	−9.5 ± 0.1 (1)	0.00035	−8.2 ± 0.1 (4)	0.0066	100	−6.7 ± 0.1 (3)	100
1	−6.5 ± 0.1 (3)	0.30	−5.6 ± 0.2 (4)	2.60	62 ± 6	ND (3)	ND
2	−6.5 ± 0.1 (3)	0.32	−5.5 ± 0.1 (3)	3.14	94 ± 6	ND (3)	ND
3	−6.2 ± 0.1 (3)	0.59	−5.3 ± 0.1 (4)	5.24	82 ± 7	ND (3)	ND
4	−5.5 ± 0.1 (3)	3.31	−5.1 ± 0.2 (3)	8.24	45 ± 7	ND (3)	ND
5	−4.5 ± 0.1 (3)	30.7	−4.7 ± 0.8 (3)	18.7	50 ± 40	ND (3)	ND
6	ND (3)	132	ND (3)	45.0	ND	ND (3)	ND
κOR	pKi	Ki (μM)	pIC50	IC50 (μM)	α	pEC50	α
U50,488	−10.0 ± 0.1(1)	0.000094	−8.9 ± 0.1 (6)	0.0015	100	−7.5 ± 0.1 (3)	100
1	−5.8 ± 0.1 (3)	1.68	ND (4)	0.073	ND	ND (3)	35 ± 6
2	−4.8 ± 0.1 (3)	14.2	ND (5)	ND	ND	ND (3)	ND
3	−5.6 ± 0.1 (3)	2.25	−5.2 ± 0.2 (6)	6.46	69 ± 8	ND (3)	20 ± 20
4	−7.1 ± 0.1 (3)	0.089	−6.6 ± 0.1 (4)	0.24	84 ± 4	−4.4 ± 0.4 (3)	50 ± 10
5	−6.0 ± 0.1 (3)	1.11	−5.6 ± 0.1 (4)	2.71	92 ± 6	ND (3)	30 ± 12
6	−5.6 ± 0.1 (3)	2.38	−5.7 ± 0.1 (4)	1.97	92 ± 5	ND (3)	ND
δOR	pKi	Ki (μM)	pIC50	IC50 (μM)	α	pEC50	α
Leu-Enk	−8.9 ± 0.1 (1)	0.0012	−8.9 ± 0.1 (4)	0.0014	100	−7.9 ± 0.1 (3)	100
1	−5.0 ± 0.1 (3)	10.4	−4.7 ± 0.8 (3)	20.3	50 ± 35	ND (3)	ND
2	−4.8 ± 0.1 (3)	15.0	−4.8 ± 0.4 (3)	15.4	90 ± 30	ND (3)	ND
3	−5.1 ± 0.1 (3)	8.37	ND (3)	95.6	ND	ND (3)	ND
4	−4.6 ± 0.1 (3)	23.2	−4.9 ± 0.4 (3)	12.4	60 ± 20	ND (3)	ND
5	−4.2 ± 0.6 (3)	60.3	−4.6 ± 0.5 (3)	24.4	90 ± 50	ND (3)	ND
6	−4.0 ± 0.9 (3)	98.8	ND (3)	ND	ND	ND (3)	ND

Figure 4.

Effects of pseudo-akuammigine in mouse models of thermal nociception. Antinociception by pseudo-akuammigine (3) was tested at doses of 5 mg/kg (p.o., n=16) (A-B) and 5 and 10 mg/kg (s.c., n=8) (C-D) in C57BL/6 mice via the tail flick assay (A and C) and the hot plate assay (B and D) at various time points. Morphine (6 mg/kg, s.c., n=8) served as a positive control (A-B). All data is expressed as maximum possible effect (%MPE) normalized to a saline baseline (treatment – saline baseline). For the 5 mg/kg doses, ^P < 0.05 vs. vehicle and ^^P < 0.01 vs vehicle. For morphine, ***P < 0.001 vs. vehicle. Morphine and 10 mg/kg alkaloid 3 data was analyzed with a paired t-test. Data for 5 mg/kg alkaloid 3 (p.o. and s.c.) was analyzed with one-way, repeated measures ANOVA followed by Sidak’s multiple comparisons post-test.

The comparable binding results reported in Menzies et al., in the PDSP screen shown in Figure 2, and in the radioligand binding assays shown in Figure 3A–C confirm that the akuamma alkaloids interact with opioid receptors.¹⁷ The results from the in vitro signaling assays in Figure 3D–I further demonstrate that the alkaloids not only bind to opioid receptors but can elicit intracellular, inhibitory G-protein activity. Furthermore, at the highest concentrations tested the alkaloids induced β-arrestin 2 recruitment, although the potency for the akuamma alkaloids in the β-arrestin 2 recruitment assay was too weak to calculate bias factors. Inspection of the functional responses at κOR suggests that the efficacy of β-arrestin 2 recruitment correlates with the potency for G-protein mediated cAMP inhibition; 4 is the most efficacious recruiter, while 2 is the weakest. Based on their inherent opioid activity, moving forward, more potent and selective opioids may be discovered using the akuamma alkaloids as a scaffold for drug design.

In vivo Characterization of Antinociceptive Effects of Akuamma Alkaloids.

Given the ability of the akuamma alkaloids to bind to and activate the μOR, we hypothesized that the reported analgesic efficacy of the akuamma plant may be primarily exerted by these μOR-activating akuamma alkaloids. As 3 has previously been demonstrated to be antinociceptive in Wistar rats when administered per os (p.o.; 5 mg/kg), experiments were conducted to reproduce these findings in mice (Table S14, Supporting Information).¹⁹ Compound 3 did not produce antinociception in mice at 5 mg/kg dose (p.o.) in the tail flick and hot plate assays of thermal nociception at any of the timepoints tested (Figure 4A–B). In this experiment, subcutaneously (s.c.) administered morphine (6 mg/kg) served as a positive control and produced significant antinociception at the 30 min timepoint. However, when 5 mg/kg 3 was administered subcutaneously, minimal yet statistically significant antinociception was measured at 30 minutes in both nociception assays, as well as at 60 minutes in the tail-flick assay (Figure 4C–D, Table S15, Supporting Information). Notably, a 10 mg/kg dose of 3 (s.c.) also failed to produce antinociception at 30 minutes (Figure 4C, D).

Because 1 and 2 had slightly higher potencies than 3 at the μOR, they were also tested for antinociceptive properties. Alkaloid 1 was tested at 3, 10, 30, and 60 mg/kg doses (s.c.). In the tail flick assay, minimal yet statistically significant antinociception was measured at 110 minutes for the 3 mg/kg dose, and at 30 minutes for the 60 mg/kg dose (Figure 5A, Table S16, Supporting Information). In the hotplate assay, minimal yet statistically significant antinociception was measured at 110 minutes for the 3 mg/kg dose, at 60 minutes for the 30 mg/kg dose, and at 30 minutes for the 60 mg/kg dose (Figure 5B, Table S16, Supporting Information). Alkaloid 2 was tested at 3, 10, and 30 mg/kg doses (s.c.). In the tail flick assay, minimal yet statistically significant antinociception was measured at 50 minutes for the 10 mg/kg dose, and at 30 minutes for the 30 mg/kg dose (Figure 5C, Table S17, Supporting Information). In the hotplate assay, minimal, yet statistically significant, antinociception was measured at 110 minutes for the 3 mg/kg dose (Figure 5D, Table S17, Supporting Information). For both 1 and 2, dose-dependent increases in antinociception were not observed reproducibly between nociception assays, and there was no general trend in the time-course of the antinociceptive effect. To explore whether the route of administration for 1 and 2 would influence the antinociceptive effect, oral dosing was also examined. However, no antinociception was measurable, indicating that metabolism of the compounds is unlikely to contribute to potential effects as has been previously proposed (data not shown).¹⁹

Figure 5.

Effects of akuammine and akuammidine in mouse models of thermal nociception. Antinociception by akuammine (1) (3, 10, 30, and 60 mg/kg (s.c.)), A-B) and akuammidine (2) (3, 10, and 30 mg/kg (s.c.)), C-D) was assessed in C57BL/6 mice (n=8, per alkaloid) via the tail flick assay (A and C) and the hot plate assay (B and D) at various time points. All data is expressed as maximum possible effect (%MPE) normalized to a saline baseline (treatment – saline baseline). For the 3 mg/kg doses, ^P < 0.05 vs vehicle and ^^P < 0.01 vs vehicle. For the 10 mg/kg doses, statistical significance is indicated as ^∂P < 0.05 vs vehicle. For the 30 mg/kg doses, ^#P < 0.05 vs vehicle. For the 60 mg/kg doses, *P < 0.05 vs vehicle and **P < 0.01 vs vehicle. Data was analyzed with one-way, repeated measures ANOVA followed by Sidak’s multiple comparisons post-test.

In a previous study in rats, the antinociceptive effects of 3 differed kinetically from those produced by morphine, with antinociceptive activity for 3 peaking at 180 minutes when administered p.o..¹⁹ To account for potential delayed onset in antinociception for akuamma alkaloids, nociception was tested in mice at 50 and 110 minutes following s.c. administration with 1 and 2. However, with continued testing and after failing to detect convincing levels of antinociception, testing was adjusted to higher doses of 30 and 60-75 minutes in an attempt to capture either a rapid or delayed peak in antinociceptive efficacy. To ensure that the s.c. route of administration was not contributing to the lack of observable antinociception, nociception was measured following p.o. administration of 1-3, yet still convincing levels of antinociception were not observed as was demonstrated by Duwiejua et al.¹⁹ The lack of antinociception with p.o. administration suggests that metabolism of the akuamma alkaloids does not greatly contribute to their purported antinociceptive effects. The incongruous antinociceptive findings between this study and previous research may be explained by species differences: in this study, C57BL/6 mice were used, whereas Wistar rats were used in the previous study.¹⁹

Conclusion.

Natural products, and in particular plant alkaloids, are a well-known source for medicinal compounds with analgesic potency. Most familiar are the opioids found in Papaver somniferum, but more recently Mitragyna speciosa has gained recognition in Western society as an alternative source of naturally occurring analgesics. This study provides a detailed investigation of the isolation of six abundant, yet chromatographically very similar, alkaloids from the seeds of Picralima nitida, a traditional plant with reported analgesic properties. Using high-purity isolates, this study undertook a detailed characterization of the pharmacology of the isolated akuamma alkaloids in mammalian cells and evaluated their antinociceptive effects in mice.

The cellular signaling characterization of the akuamma alkaloids at the opioid receptors agrees with previous findings but enhances the previous assessment by providing an analysis of intracellular signaling properties, particularly β-arrestin recruitment, at the opioid receptors as well as the binding capability to non-opioid receptors.¹⁷ It was observed that the akuamma alkaloids’ potency was too weak to accurately determine β-arrestin recruitment, however, several alkaloids display significant recruitment (>25%) at the highest dose that could be tested. The pharmacological profile of the akuamma alkaloids stands in contrast to the kratom alkaloids, particularly 7-hydroxymitragynine, which can be classified as a highly potent and G-protein-biased μOR agonist.³⁴ Of all the akuamma alkaloids investigated, 4 stands out as being relatively potent at the κOR. The promiscuity of 4 for other receptors is conceivably problematic; however, it has the potential for serving as a scaffold for developing novel κOR agonists.

Analysis of akuamma’s antinociceptive properties revealed limited antinociceptive efficacy of three akuamma alkaloids: 1-3. While the observed limited antinociceptive efficacy does not support akuamma’s traditional use for pain relief and is not congruous with a previous report of potent antinociception by 3 in rats,¹⁹ this apparent contradiction is by far not unusual for investigations of ethnomedicinally used plants; as we and others have demonstrated, compound abundance and pharmacological relevance are not necessarily correlated,^35–37 and even very minor components or impurities can be responsible for the observed biological activity.³⁸ Thus, while these studies provide detailed insight into the pharmacology of six highly abundant alkaloids present in the P. nitida extracts, other components present in lower abundance may possess potent antinociceptive activity. Furthermore, it is possible that pharmacokinetic and/or pharmacodynamic differences between rats and mice may account for the discrepant antinociceptive responses for the akuamma alkaloids in Wistar rats and C57BL/6 mice. Future studies should explore the pharmacokinetics of the akuamma opioids and akuamma metabolites in mice, confirm the reported antinociceptive effect in rats, and investigate potential synergistic interaction of the akuamma alkaloids in vivo. Additionally, despite the limited antinociceptive efficacy reported here, the unique structural features of the akuamma alkaloids provide opportunities to study the opioid receptors. By exploring the structure-activity relationships of these scaffolds and developing synthetic analogs, particularly those with improved pharmacokinetic and pharmacodynamic properties, the akuamma alkaloids may be transformed into useful pharmacological probes of the opioid receptors and to gain utility in treating pain and other disorders.

EXPERIMENTAL SECTION

General Experimental Procedures.

All solvents and reagents were purchased from commercial sources and used directly without further purification. Akuamma seed powder was purchased from Relax Remedy. ¹H and ¹³C NMR spectra were recorded on Bruker 400 MHz spectrometer and referenced to the residual solvent peaks (CHCl₃: ¹H δ=7.26, ¹³C δ=77.16 ppm; D₂HCOD: ¹H δ= 3.31, ¹³C δ=49.00 ppm) High-resolution mass spectra were obtained on a Shimadzu LCMS-IT-TOF and observed values are within 5 ppm of calculated exact masses of the indicated ions. High-performance liquid chromatography was conducted on an Agilent 1260 Infinity II fitted with a DAD detector and a Phenomenex Luna Omega PS-C18 column (100 x 4.6 mm). A gradient of acetonitrile/water (20-45%) each containing 0.1% formic acid with a flow rate of 1 ml/min was used. The purity of all compounds was determined to be >95% as determined by HPLC.

Drugs.

Leu-enkephalin, forskolin, and morphine sulfate pentahydrate were purchased from Sigma Aldrich (St. Louis, MO, USA). (2S)-2-[[2-[[(2R)-2-[[(2S)-2-Amino-3-(4-hydroxyphenyl)propanoyl]amino] propanoyl] amino]acetyl]-methylamino]-N-(2-hydroxyethyl)-3-phenylpropanamide (DAMGO), and 2-(3,4-dichlorophenyl)-N-methyl-N-[(1R,2R)-2-pyrrolidin-1-ylcyclohexyl]acetamide (U50,488) were purchased from Tocris Bioscience (Bio-techne Corporation, Minneapolis, MN, USA). [³H]DAMGO (49.2 Ci/mmole, lot#2573313), [³H]U69,593 (60 Ci/mmole, lot#2367921), and [³H]DPDPE (53.7 Ci/mmole, lot#2376538) were purchased from Perkin Elmer (Waltham, MA, USA).

Preparation of Dichloromethane Fraction, Akuammine (1), and Akuammidine (2).

Akuamma seed powder (250 g) was allowed to stir for two hours in methanolic hydrochloride solution (400 mL). Subsequently, the seed powder was filtered, and the filtrate was evaporated to dryness under reduced pressure. The resulting extract was dissolved in aqueous hydrochloric acid (400 mL, 2N), washed with hexanes (3x400 mL), and extracted with dichloromethane (3x400 mL). The combined dichloromethane layers were evaporated to dryness under vacuum to provide the crude dichloromethane fraction (3.356 g). The aqueous layer was brought to pH=12 with 28% ammonium hydroxide, washed with hexanes (1x400 mL), and extracted with ethyl ether (3x400 mL). The combined ethereal layers were dried over magnesium sulfate and concentrated under vacuum to provide a mixture of 1 and 2. This mixture was treated with cold acetone to precipitate 1 as a white solid (472 mg). The acetone filtrate was concentrated under vacuum and the resulting residue recrystallized in dichloromethane to yield crystalline 2 (15.0 mg).

pH-Zone Refining Countercurrent Chromatography of the Dichloromethane Fraction.

The pH-Zone Refining Countercurrent Chromatography was performed on a SCPC-250 (Gilson Incorporated, Middleton, WI USA) chromatograph equipped with a 266 mL rotor. The rotation speed could be adjusted from 500 to 3000 rpm. Samples were injected through a 20 mL sample loop. The detection was performed by a UV-Vis DAD detector. Fractions were collected with a Gilson-Armen Fraction Collector LS-5600. Chromatographic data were acquired by using the Gilson-Armen Glider CPC Control Software V2.9.2.9 and then transferred to an Excel worksheet for further processing.

The optimal solvent system was determined by evaluation of the acid and base partition coefficients of the alkaloids of interest using guidelines described by Ito.¹⁸ Five different solvent system formulations were tested, with triethylamine (TEA) and hydrochloric acid (HCl) added to the upper (organic) and lower (aqueous) phases, respectively. Partition coefficients were determined by comparing the area under the curve for the HPLC peaks produced by akuammicine, akuammiline, pseudo-akuammigine, and picraline (Table 1).

The pH-zone-refining countercurrent chromatography separation was prepared by thoroughly mixing equal volumes of ethyl acetate and water in a 2 L separatory funnel and allowing the layers to separate. The ethyl acetate layer was basified with TEA to a final concentration of 10 mM to be used as the upper phase. The aqueous lower phase was adjusted to a final concentration of 8 mM with hydrochloric acid. The dichloromethane extract (1.2 g) was dissolved in 10 mL of upper phase with less than 1 mL of the lower phase to aid solubility and loaded into a 20 mL sample loop. An additional 5 mL of upper phase was used to rinse the sample vial and added to the sample loop. The instrument column was filled with the lower phase at a rotation speed of 500 rpm. The rotation speed was increased to 3000 rpm and the sample was introduced into the column. The basified upper layer was pumped through the coil at a flow rate of 10 mL/min with elution in ascending mode. Elution was monitored at 254 nm, 284 nm and 330 nm. Fractions were collected in 7.5 mL quantities. After elution, the pH of each fraction was measured using a benchtop pH meter (Mettler Toledo) and fraction contents were evaluated using TLC. All fractions were dried with sodium sulfate, concentrated under vacuum, and analyzed by ¹H NMR. Fractions containing purified, individual alkaloids were combined separately to provide 3 (130 mg) and 4 (145 mg). Fractions containing a mixture of 5 and 6 were combined and further purified by silica gel flash column chromatography eluting with 0-2% MeOH/CHCl₃ containing 1% TEA to yield pure samples of 5 (61 mg) and 6 (90 mg).

Primary and Secondary Receptor Screening.

Compounds 1 and 3-6, but unfortunately not 2, were submitted to the Psychoactive Drug Screening Program (UNC-Chapel Hill) as dry powders to be evaluated using standard protocols. Primary screening was conducted at 10 μM (DMSO) against the “Comprehensive Screen” panel consisting of 37 different GPCR, ion channel, and transporter targets. Assays producing >50% inhibition of radioligand binding were further investigated in secondary binding assays using a 12-point concentration-response curve to determine binding affinity (Ki).

Cell Culture.

HEK293 cells (RRID:CVCL_0045, Life Technologies, Grand Island, NY, USA) were maintained in DMEM supplemented 10% FBS. CHO-K1-human δ opioid receptor (δOR PathHunter β-arrestin 2 cells and CHO-K1-human μ opioid receptor (μOR) PathHunter β-arrestin 2 cells stably expressing the δOR or μOR and β-arrestin 2 (RRID:CVCL_KY70, RRID:CVCL_KY68, DiscoverX, Fremont, CA, USA) were maintained in F12 media supplemented with 10% FBS and containing 800 μg/mL geneticin and 300 μg/mL hygromycin. U2OS-human κ opioid receptor (κOR) PathHunter β-arrestin 2 cells stably expressing the κOR and β-arrestin 2 (RRID:CVCL_LA97, DiscoverX, Fremont, CA, USA) were maintained in McCoy’s 5A media supplemented with 10% FBS and containing 500 μg/mL geneticin and 250 μg/mL hygromycin. All cell lines were maintained in T75 flasks under sterile conditions and kept at 37 °C and 5% CO₂. During passaging, cells were dislodged from the flask following a 3-minute incubation with 0.25% trypsin, and sub cultivated at ratios of 1:10 (HEK293), 1:5 (CHO) and 3:10 (U2OS).

Competitive Radioligand Binding Assay.

Binding assays were performed on membranes isolated from CHO cells stably expressing the δOR or μOR and from U2OS cells stably expressing the κOR (DiscoverX) as previously described using tritiated radioligands ([³H]DAMGO, [³H]U69,593, [³H]DPDPE for μOR, κOR, δOR, respectively).³⁹

GloSensor cAMP Inhibition Assay.

cAMP inhibition assays were performed as previously described in HEK293 cells transiently transfected with pGloSensor22F-cAMP (Promega, Madison, WI, USA) and either FLAG-mouse δOR, HA-mouse μOR, or FLAG-mouse κOR.⁴⁰

PathHunter β-arrestin2 Recruitment Assay.

β-arrestin recruitment assays were performed as previously described using CHO or U2OS cells (CHO-K1-human δOR, CHO-K1-human μOR, or U2OS-human κOR PathHunter β-arrestin 2 cells, DiscoverX).⁴⁰

Animals.

Wildtype C57Bl/6N mice (24 male, 24 female; 7-8-weeks old) were purchased from Envigo (Indianapolis, IN) and were acclimated to the facility and to handling for 1 week prior to any experimental procedures. See Supporting Information for details on subject groups for drug testing (Table S14, Supporting Information). All mice were housed on a 12-hour light (21:30-9:30)/12-hour dark cycle under controlled temperature (21-23 °C) with ad libitum food access. All experiments were conducted between 10:30-15:00 in a well-lit room. At a minimum, mice were given 2 days between experiments to recover from thermal stimuli. All experimental procedures were approved by the Purdue Animal Care and Use Committee of Purdue University under protocol #1605001408.

Tail Flick Thermal Nociception Assay.

Antinociception was measured as previously described.⁴¹ On the first day of the experiment, mice were habituated to handling restraint; a black washcloth was used to restrain the mice during the experimentation. On the following days of drug testing, a radiant heat tail-flick apparatus (Columbus Instruments, Columbus, OH, USA) was set to a beam intensity of 7-9 as this intensity yielded reproducible responses between 2-3 seconds. On each test day, a baseline tail flick response was first obtained for each mouse. The cutoff time for testing was calculated as 3 times this baseline response time. A saline injection was then administered (s.c. or p.o.) and after 30 minutes, tail flick responses were collected again. Drugs were then administered (s.c. or p.o.), and tail flick responses were collected at various time points following administration. All measurements were collected in duplicate by testing two different regions on the mouse’s tail.

Hot Plate Thermal Nociception Assay.

On the first day of the experiment, mice were habituated to the hotplate apparatus (Columbus Instruments, Columbus, OH, USA) for 1-2 minutes (while the hotplate was turned off). On the following days of testing, the hot plate was maintained at a temperature of 55 ± 0.5 °C. On each test day, a single baseline time for latency to demonstrate nociceptive behavior was first obtained for each mouse. Behavior considered a positive nociceptive response was fore or hind paw licking, jumping, or non-explorative rearing. Upon demonstrating this behavior, the mouse was immediately removed from the apparatus. The cutoff time for testing was calculated as 3 times the baseline response time. A saline injection was then administered (s.c. or p.o.) and after 30 minutes, hot plate latency responses were collected again. Drugs were then administered (s.c. or p.o.), and hot plate latency responses were collected following administration at various time points. All measurements were collected only once to avoid damage to paws.

Statistics.

All data were analyzed using GraphPad 8 (GraphPad Prism software, La Jolla, CA) and is presented as means ± SEM. For in vitro findings, composite figures consisted of one curve averaged from three, independent assays. In these independent assays, PathHunter β-arrestin recruitment and radioligand binding assays were run in duplicate, and GloSensor cAMP assays were run in triplicate. Data from each independent signaling assay was normalized to a positive control before being averaged and added to the composite figure. For nociception assays, significance was calculated via one-way, repeated measures ANOVA with Sidak’s multiple comparison’s test to compare saline treatment with drug treatment at multiple time points. For any nociception assays where only one time point was tested, a paired t-test was used to assess significance between saline and drug treatment. Nociception data is represented as percent maximal possible effect (%MPE) (calculated as % MPE= (treatment response time - baseline response time)/(cutoff time - baseline response time) * 100) and is normalized (drug treatment %MPE - saline treatment %MPE). Statistical measures and values for all nociception assays are summarized in Tables S15–17 in the Supporting Information.