Antinociceptive and sedative activity of Vernonia patula and predictive interactions of its phenolic compounds with the cannabinoid type 1 receptor

When tested in the acetic acid‐induced writhing and formalin‐induced paw‐licking tests, the ethanol extract of Vernonia patula (VP) aerial parts showed significant antinociceptive activity. In neuropharmacological tests, it also significantly delayed the onset of sleep, increased the duration of sleeping time, and significantly reduced the locomotor activity and exploratory behaviour of mice. Five phenolic compounds, namely gallic acid, vanillic acid, caffeic acid, quercetin and kaempferol, were detected in VP following HPLC‐DAD analysis. The presence of these phenolic compounds in VP provides some support for the observed antinociceptive and sedative effects. A computational study was performed to predict the binding affinity of gallic acid, vanillic acid, caffeic acid, quercetin and kaempferol towards the cannabinoid type 1 (CB1) receptor. Caffeic and vanillic acid showed the highest probable ligand efficiency indices towards the CB1 target. Vanillic acid displayed the best blood–brain barrier penetration prediction score. These findings provide some evidence for the traditional use of VP to treat pain.

. Here, we analysed the aerial parts V. patula for the presence of phenolic compounds, and tested its ethanol extract for antinociceptive and sedative activity in mice. We also conducted a molecular docking study to predict the interactions between the phenolic compounds detected in V. patula (VP) and the CB1 receptor in an effort to establish a putative mode of action for the phenolic compounds present in this plant.

| Plant collection and extraction
The aerial parts of V. patula (VP) were collected from Chittagong (Bangladesh) in 2015. The plant material was identified by M. A. Ali at the Bangladesh National Herbarium in Dhaka where a voucher specimen (DACB: 35107) is kept for future reference. The dried powdered aerial parts (500 g) were macerated in EtOH (1.5 L) for 5 days at 25 ± 2 C, with occasional shaking, and the resulting filtrate was concentrated under reduced pressure to yield the extract (66 g, 13.2% yield).

| HPLC-DAD analysis
Phenolic compounds in VP aerial parts were detected using a Dionex UltiMate 3000 LC system and Acclaim® C18 (4.6 × 250 mm; 5 μm) column (Sumi et al., 2016). A mixed solvent system of acetonitrile (solvent A), acetic acid solution at pH 3.0 (solvent B) and methanol (solvent C) was used as a mobile phase with the following gradient elution (0 min: 5% A/95% B; 10 min: 10% A/80% B/10% C; 20 min: 20% A/60% B/20% C and 30 min: 100% A). The extract (20 μL) was injected at a flow rate of 1 mL/min, which was maintained constant throughout the analysis. The diode array detector (DAD) was set from λ = 200 to 700 nm. Specific wavelengths for the detection of phenolic compounds were set up as follows: λ = 280 nm for 18.0 min (caffeic acid, catechin, vanillic acid, epicatechin and gallic acid), λ = 320 nm for the next 6 min (ellagic acid, p-coumaric acid and rutin), and finally λ = 380 nm for the rest of the analysis (myricetin, kaempferol and quercetin). A stock solution (100 μg/mL) of each of the aforementioned phenolic compounds was prepared and further diluted to 20 μg/mL, except for caffeic acid (8 μg/mL) and quercetin (6 μg/mL). A solution of VP (5 mg/mL) was prepared in ethanol. Spiked samples were also prepared by mixing the VP solution with each of the phenolic standards. All solutions were filtered through a 0.20 μm nylon syringe filter (Sartorius, Germany) prior to the HPLC analysis. The phenolic compounds present in VP were identified by comparing their retention times and absorbance profiles with the standards. Calibration curves were prepared with serial dilutions of each phenolic standard in methanol, ranging from 1.25 to 20 μg/mL (gallic acid, myricetin, vanillic acid, epicatechin, p-coumaric acid, kaempferol and ellagic acid), 0.5-8.0 μg/mL for caffeic acid, catechin and rutin), and 0.375-6.0 μg/mL for quercetin. Quantification of the peaks was performed with R 2 > 0.995. Data were obtained as means ± SD of three independent analyses (Table S1, Figure S1).

| Test animals
Swiss albino mice (20-25 g) of both sexes were collected from the animal research branch of the International Center for Diarrheal Disease and Research in Bangladesh (ICDDR, B) and kept under standard laboratory conditions (25 ± 2 C and 12/12 h light/dark cycle). The animals were fed with standard food and water ad libitum during the acclimatisation period. Before any experiments, the animals were kept fasted overnight. All animals were handled according to the ethical principles and guidelines for experiments on animals (Swiss Academy of Medical Sciences and Swiss Academy of Sciences, 2005) and approved by the BCSIR ethics committee (BCSIR/IAEC/01/13-14).

| Acetic acid-induced writhing test
This was used to evaluate peripheral antinociceptive behaviour (Whittle, 1964). Briefly, the mice were divided into five groups (n = 5).
The negative control group was treated with 1% Tween 80 in normal saline (10 mL/kg, p.o.). Diclofenac sodium (25 mg/kg) was administered intraperitoneally to the positive control group. Test groups were treated with the VP extract (100, 200 and 400 mg/kg respectively, p. o.) 30 min before intraperitoneal injection of 0.7% acetic acid. After a 5 min interval, the constriction of the abdomen and extension of hind legs (writhing) of the animals was determined for 10 min. The ability of the VP extract and diclofenac to reduce the number of writhing was calculated using Equation (S1.1).

| Formalin-induced paw licking test
This was performed using a previously published protocol (Dubuisson & Dennis, 1977). Briefly, overnight-fasted mice were distributed into six groups (n = 5). The negative control group was treated with 1% Tween 80 in normal saline (10 mL/kg, p.o.).
Diclofenac sodium (25 mg/kg) and morphine (5 mg/kg) were administered subcutaneously (s.c.) in the positive control group as phase 1 and phase 2 positive controls, respectively. The VP extract (100, 200 and 400 mg/kg, p.o.) was administered to the test group. A 1% formalin solution in saline (20 μL) was then injected s.c. to all groups. After the formalin injection, the licking or biting time of the injured paw was recorded as a nociceptive response, considering the first 5 min as phase-1 and the next 15-30 min as phase-2, respectively (Silva, Martins, Matheus, Leitao, & Fernandes, 2005).

| Hot plate test
The mice were placed on a hot metal plate (50 ± 0.5 C). The time period between the placement of the animal on the hot surface and its reaction-by lifting or licking its paws-to avoid thermal pain was recorded as the response latency (Eddy & Leimbach, 1953). Twenty seconds were used as a cut-off period to avoid any tissue damage of the paws. Mice were pre-treated with 1% Tween 80 in normal saline

| Tail immersion test
This was performed according to a previously published method (Silva et al., 2005). Five groups of mice (n = 5) were pretreated with VP (100, 200 and 400 mg/kg, p.o.) and with 1% Tween 80 in normal saline (10 mL/kg, p.o.) and morphine (5 mg/kg, i.p.) for the negative and positive control groups, respectively. Their tails were immersed in warm water (55 ± 1 C). The latency between tail submersion and deflection of the tail was recorded. Pre-treatment latency was recorded at 30, 60, 90 and 120 min. In a similar experiment, another five groups of mice (n = 5) were used and, to verify central analgesic activity, were administered naloxone (2 mg/kg, i.p.). The antinociceptive activity was evaluated based on the latency period of the tail withdrawal response. Equation (S1.2) was used to calculate the % MPE. The time period between the pentobarbital administration and the onset of sleep (i.e., when the animals lost their righting reflex) as well as the duration of sleep (i.e., time elapsed between the loss and recovery of the righting reflex) were recorded to evaluate the total hypnotic effect of VP in mice.

| Open field test
This assessed both locomotor and emotional activity in mice using an open field apparatus (Shilpi et al., 2006). The mice were placed in a box (1 mg/kg, i.p.), respectively. Locomotor activity was recorded as the number of crossed lines visited by each mouse for 3 min at 0, 30, 60, 90, 120, 180, and 240 min after oral administration of the test drugs. Different groups were used for different evaluation times. The percentage inhibition of movements (% MI) was calculated using Equation (S1.3).

| Hole cross test
This test was performed using a cage (30 cm × 20 cm × 14 cm in height) with a steel partition containing a hole (3 cm in diameter and 7.5 cm in height) in the centre (Takagi, Watanabe, Saito, 1971). The number of mice which passed from one chamber to the other through the hole was calculated after administering the test drugs for 3 min at 0, 30, 60, 90, 120, 180 and 240 min during the study period. Different groups were used for different evaluation times. The % MI was calculated using Equation (S1.3).

| Statistical analysis
One-way or two-way analysis of variance (ANOVA) followed by Dunnett's test was used when comparing the test samples with the negative control. Variances between different groups were measured to a significance at a near of p < .05. All statistical analyses were performed using the SPSS software v.11.5.

| Protein optimisation
The crystal structure of the cannabinoid receptor 1 (CB1) was retrieved from the PDB database (PDB ID: 5U09). Energy minimisation was performed using Swiss-PDB Viewer v. 4.1.0 (Guex & Peitsch, 1997). The heteroatoms and water molecules were removed from the crystal structure using PyMOL Molecular Graphics System v. 1.3 (https://pymol.org) prior to docking. The protein and ligand structures were saved in the PDBQT format.

| Determination of ligand-protein binding affinity and non-bonding interactions
The active binding pocket of CB1 was predicted by CASTp (v. 3.0) (Dundas et al., 2006). The protein (ID-J_5C2DB3AA670AB) showed the highest pocket area and volume at 652.65 Å 2 and 331.44 Å 3 , respectively. This predicted binding pocket was used for the generation of the grid box to dock the triterpenoids against CB1. The centre of the grid box was set at 12.5230, 7.2495 and 17.7743 Å and the box size was set at 81.5, 81.5 and 81.5 Å in x, y and z directions, respectively. AutoDock Vina (v. 1.1.2) was used to perform the molecular docking study (Trott & Olson, 2010). The docked pose of the lowest binding free energy con-

| Detection of phenolic compounds
Analysis of the ethanol extract of VP by HPLC-DAD revealed the presence of high amounts of caffeic acid and quercetin (123.75 and 96.35 mg/100 g dry extract, respectively). The extract also contained kaempferol, gallic acid, and vanillic acid (28.28, 17.82 and 8.91 mg/100 g of dry extract, respectively) (Table 1, Figure S2).

| Acetic acid-induced writhing test
VP demonstrated a significant dose-dependent decrease in the number of writhing responses. At 400 mg/kg, VP exerted a writhing inhibitory effect of 65.22% that was comparable to diclofenac (78.74%) ( Table 2).  (Table 3).

| Hot plate test
VP demonstrated a dose-dependent central antinociceptive effect with a maximum increase in the reaction time to the thermal stimulus at the dose of 400 mg/kg after 90 min (6.63 s) that was statistically significant compared to the negative control (p < .001). At that same dose, VP also showed a reaction time (5.15 s) that was higher than morphine. When naloxone was administered together either with VP or with morphine, the analgesic effect was overall reversed, but no statistical significance was observed (Table 4).

| Tail immersion test
All tested doses of VP (100, 200 and 400 mg/kg) significantly increased the latency period to the hot water-induced thermal stimulus, with the highest increase (7.11 s) seen after 45 min, and statistically significant with VP or with morphine, the analgesic effect was overall reversed, but no statistical significance was observed (Table 5). Values are mean ± SEM (n = 5). b In mL/kg. *p < .05, **p < .01, ***p < .001 vs control (two-way ANOVA followed by Dunnett's test).

| Pentobarbital-induced hypnosis
At the 400 mg/kg dose, VP significantly decreased the onset of sleep (7.96 min) compared to the control group (14.69 min) and significantly increased the total sleeping time (Table 6).

| Open field test
After 30 min of administration of VP, a noticeable decrease in locomotion in the murine model at all tested doses was observed, which was comparable to the standard diazepam (Table 7). Values are mean ± SEM (n = 7). b In mL/kg. *p < .05 vs control (one-way ANOVA followed by Dunnett's test).

| Hole cross test
VP at the doses of 200 and 400 mg/kg significantly suppressed both the motor activity and the exploratory behaviour at 90 min and continued up 240 min compared to the standard drug diazepam (Table 8).

| Computational study
Caffeic acid and vanillic acid displayed the highest ligand efficiency indices towards the CB1 receptor (0.0372 and 0.0357, respectively).
Both compounds engaged in H-bond and hydrophobic binding interactions with several amino acid residues of CB1 (Table 9; Figures 1 and 2).

| Prediction of the pharmacokinetic properties
All five phenolic compounds showed human intestinal absorption prediction scores similar to dronabinol. Only caffeic acid and vanillic acid showed a positive MLogP, albeit lower than that of THC. Caffeic acid and vanillic acid showed the highest blood-brain barrier (BBB) penetration prediction scores of all phenolics (0.71 and 0.84, respectively) (Table 10).

| DISCUSSION
VP showed significant antinociceptive activity in the acetic acidinduced writhing test used to evaluate peripheral pain (Gawade, 2012). It showed significant activity in phase 1 and 2 of the formalin-induced pain test (Tjølsen, Berge, Hunskaar, Rosland, & Hole, 1992). In the latter, phase-1 represents central pain (i.e., via the stimulation of sensory afferent C-fibres) whilst phase-2 represents inflammatory pain (i.e., via the production of prostaglandins and bradykinins in peripheral tissues) (Chen, Tsai, & Wu, 1995). VP also showed significant antinociceptive activity in the hot plate test (only at 200 and 400 mg/kg) and the tail immersion tests. These tests evaluate central analgesic activity linked to supra-spinal and spinal reflex, respectively, through the modulation of μ opioid (MOP) receptors (Jinsmaa et al., 2005). Co-administration of the opioid receptor antagonist naloxone, which reverses the action of morphine on MOP receptors, led to a reversal of the analgesic effect of VP. Interestingly, it has been suggested that the MOP and CB1 receptors mediate some overlapping pharmacological responses, including pain (Rios, Gomes, & Devi, 2006 (Howlader et al., 2017).
Their interaction with the cannabinoid receptors leads to a decrease in intracellular cAMP levels. Low concentrations of cAMP leads to a decrease in presynaptic entry of calcium ions, which results in a reduction in the release of neurotransmitters such as GABA, L-glutamate, noradrenaline, dopamine, serotonin and acetylcholine (Howlett et al., 2002;Lu & Mackie, 2016   Abbreviations: BBB, blood-brain barrier, MLogP, LogP oil/water ; NSP, negative score prediction; p.s., prediction score; TPSA, topological polar surface area.
facilities and to Mr. Hemayet Hossain, Senior Scientific Officer at CRD, for his help with the HPLC-DAD analysis.