In vitro hydrolysis of areca nut xenobiotics in human liver

Abstract

Areca nut (AN) is a substance of abuse consumed by millions worldwide, in spite of established oral and systemic toxicities associated with its use. Previous research demonstrates methyl ester alkaloids in the AN, such as arecoline and guvacoline, exhibit mood-altering and toxicological effects. Nonetheless, their metabolism has not been fully elucidated in humans. In the present study, an HPLC-UV bioanalytical method was developed to evaluate the hydrolytic kinetics and clearance rates of arecoline and guvacoline in human liver microsomes (HLM) and cytosol (HLC). The bioassay was capable of quantifying arecoline and guvacoline (and carboxylate metabolites arecaidine and guvacine, respectively) with good sensitivity, accuracy, and precision. Kinetics of arecoline and guvacoline hydrolysis best followed the Michaelis-Menten model. Apparent intrinsic clearance $\left({\text{Cl}}_{\mathrm{int}.\mathit{in}\mathit{vivo}}\right)$ of arecoline was 57.8 ml/min/kg in HLM and 11.6 mL/min/kg in HLC, a 5-fold difference. Unexpectedly, guvacoline was dramatically less hydrolyzed than arecoline in both HLM and HLC, with ${\text{Cl}}_{\mathrm{int}.\mathit{in}\mathit{vivo}}$ estimates of 0.654 ml/min/kg and 0.466 ml/min/kg, respectively. These results demonstrate, for the first time, arecoline undergoes significant hydrolysis with high clearance rates in the liver. Furthermore, differential tissue metabolic rates and utilization of specific esterase inhibitors unequivocally demonstrated arecoline is a substrate for CES1 and not CES2.

1. Introduction

Areca nut (AN), often referred as the betel nut, is a seed from the Areca catechu palm tree that grows indigenously in much of the tropical Pacific, Asia, and parts of east Africa [1]. AN has been used since antiquity as an herbal medicine for treating gastric and intestinal illness, along with disorders of the cardiovascular, pulmonary and central nervous systems [2]. The AN is also a substance of abuse, producing sensations of well-being, euphoria, warming of the body, and increased mental alertness. In fact, the AN is the fourth-most self-administered psychoactive substance globally, with an estimated several hundred million users worldwide [3].

The AN contains several nitrogenous, small molecule chemicals with biological activities collectively known as the areca alkaloids – arecoline, arecaidine, guvacoline (norarecoline) and guvacine [1]. Arecoline is a major psychoactive compound in the nut, acting at several sub-types of the nicotinic acetylcholine receptor (nAChR) that are closely related to nicotine’s addictive properties [4]. Lesser is known about guvacoline’s pharmacology, with current evidence suggesting it acts as a full muscarinic acetylcholine receptor (mAChR) agonist with less potency than arecoline [5]. Like arecoline, guvacoline increases locomotor activity in experimental addiction models probably due to interplay with mAChRs [6].

Arecoline is also a major toxicant found in the AN that damages, for example, hepatic [7] and oral [8] cells. Guvacoline and its metabolite, N-nitrosoguvacoline produced in situ during AN chewing, are toxic to human buccal epithelial cells and possibly carcinogenic [9]. Overall, AN consumption in humans is associated with an increased risk for dependence, addiction, respiratory disease, cardiovascular illness, hepatocarcinoma, hypothyroidism, precancerous lesions of the mouth, oro-pharyngeal cancers, and numerous other conditions/disorders [10].

In spite of their rampant abuse as major components of the AN and noted multiplex systemic toxicities, mechanisms of arecoline and guvacoline disposition in humans are surprisingly understudied [1,11]. Given the great importance of biotransformation in understanding the time course of medicinal compounds and their metabolites in the body and association with toxicities, further information about their metabolism is vitally needed. Arecoline and guvacoline are unique from other areca alkaloids by containing a labile methyl ester moiety in their chemical structure (Fig. 1). Other features of the arecoline molecule of significance to drug metabolism include N-methyl group and an alpha-beta unsaturated carbonyl system [1]. In vitro studies demonstrate that arecoline is converted to arecoline N-oxide primarily by human recombinant flavin-containing monooxygenase enzyme-1 (FMO1), with lesser contribution by FMO3 [12]. This appears to be a toxification reaction, as arecoline N-oxide like arecoline is mutagenic and carcinogenic [13]. Arecoline also undergoes rapid condensation with the endogenous tripeptide glutathione (GSH) via a 1,4-Michael addition reaction, spontaneously forming the arecoline-SG conjugate [14,15], which is further metabolized and excreted as a mercapturic acid conjugate [1].

Fig. 1.

Proposed metabolic schematics for hydrolysis of arecoline to arecaidine (A) and guvacoline to guvacine (B) in human liver microsomes (HLM) and human liver cytosol (HLC).

Pharmacokinetic studies in mice and humans have identified arecaidine as a major metabolite of arecoline in blood and urine [1]. Arecaidine (a carboxylate) is the likely metabolic by-product of arecoline (methyl ester) hydrolysis in humans, presumably occurring in the liver (Fig. 1). This presumption is based on previous in vitro metabolism experiments using mouse tissue homogenates [16]. Compared to in plasma, kidney and brain homogenates, arecoline was predominantly metabolized in mouse liver homogenate, with almost 100 % depletion occurring after a 30 min incubation [16]. Limitations to their work were only parent molecule (arecoline) concentrations were quantified with no detection or kinetic analysis of metabolite (arecaidine) formation, and a mechanism in human tissue was not explored.

Disposition of guvacoline is even less understood than arecoline with no enzymatic mechanisms defined at this time, to our best knowledge. Structurally, guvacoline (also known as norarecoline) differs only slightly from arecoline by lack of a methyl group at the pyridine nitrogen (Fig. 1). Guvacoline contains an alpha-beta unsaturated carbonyl system which explains its propensity to deplete cellular thiols and decrease cell survival in vitro [9]. Under strong alkaline conditions in vitro, guvacoline (methyl ester) is hydrolyzed to guvacine (carboxylate) in aqueous solutions [17].

Various exogenous and endogenous compounds, for example those containing ester bonds, are hydrolyzed into the corresponding carboxylic acid and alcohol products. Hydrolysis is a phase I drug metabolism process that can occur spontaneously but often enzymatically via hydrolases/esterases [18]. Hydrolysis serves several purposes in xenobiotic disposition, including bioactivation of prodrugs (e.g., ACE inhibitors [19]), deactivation of drugs of abuse (e.g., cocaine [20]), and toxification of drugs (e.g., flutamide [21]).

In the case of arecoline, the role of hydrolysis in toxicological sequalae is not fully understood but is presumed a detoxification step based on previous research [1,11]. With guvacoline, even less is known. Taken together, there is a pressing need to examine guvacoline and arecoline hydrolysis in human liver. Since drug metabolism experiments require precise and sensitive bioanalytical assays for measuring rates of substrate depletion and metabolite formation, the first objective of our study was to develop a high-performance liquid chromatography with ultraviolet detection (HPLC-UV) assay to quantify arecoline, arecaidine, guvacoline, and guvacine. To the best of our knowledge, there is no report about utilization of HPLC-UV to study in vitro guvacoline or arecoline metabolism. Secondly, we aimed to apply this assay to characterize, for the first time, the hydrolytic kinetics and clearance rates of arecoline and guvacoline in human liver microsomes (HLM) and human liver cytosol (HLC), and ascertain the specific hepatic esterase that catalyzes these reactions.

2. Materials and methods

2.1. Chemicals and reagents

Arecoline hydrobromide (purity, >99 %), arecaidine hydrobromide (purity, >99 %), guvacine hydrochloride (purity, >99 %), and guvacoline hydrobromide (purity, >99 %) were purchased from Toronto Research Chemicals (North York, ON, Canada). Imipramine hydrochloride (purity, ≥99 %), vinblastine sulfate (purity, ≥97 %), BW284c51 (1,5-Bis(4-allyldimethylammoniumphenyl)pentan-3-one dibromide) (purity, ≥97 %), ethopropazine (purity, ≥98 %), and BNPP (bis-p-nitrophenyl phosphate sodium salt) (purity, ≥99 %) were purchased from Sigma-Aldrich (St. Louis, MO, USA). Solvents for HPLC analysis included HiPerSolv Chromanorm^® HPLC-grade acetonitrile (ACN) and HPLC-grade water, both purchased from VWR (Radnor, PA, USA). HPLC-grade sodium phosphate dibasic anhydrous (purity, ≥99.9 %) and HPLC-grade sodium hydroxide solution (50 %) were both purchased from VWR. HPLC-grade phosphoric acid (85 %), ethanol, loperamide hydrochloride (purity, ≥98 %), and digitonin were purchased from Thermo Fisher Scientific (Waltham, MA, USA). Ultrapure water was generated by a Milli-Q^® purification system (MilliporeSigma; Burlington, MA, USA). Corning^® UltraPool^™ mixed gender human liver microsomes and human liver cytosol were purchased from Corning Gentest, BD Biosciences (Woburn, MA). Human intestinal microsomes, along with CES1 and CES2 Supersomes^™, were purchased from Discovery Life Sciences (Huntsville, AL, USA). All other chemicals, solvents and reagents were of HPLC grade or higher and procured from standard chemical suppliers.

2.2. Calibration standards and quality control

Stock solutions (calculated as free base) of arecoline, arecaidine, guvacoline and guvacine were prepared using Millipore water. A series of working standard solutions were prepared by diluting stock solutions to their final concentrations using sodium phosphate buffer (100 mM, pH 7.4). Quality controls consisted of samples prepared at low (0.1 μg/ ml), medium (1 μg/ml), and high (10 μg/ml) concentration of alkaloids fortified in the supernatant of liver microsomes or cytosol.

2.3. HPLC-UV instrument and chromatographic conditions

The HPLC assay was modified from a previous method in which AN alkaloids were detected in raw AN samples [22]. Analyses were conducted on a Waters e2695 Separations Module (Milford, MA, USA) consisting of a degasser, a quaternary pump, an autosampler, a column thermostat and a Waters 2998 PDA detector. The system was managed by Empower-3 software. The mobile phase consisted of 60 % acetonitrile and 40 % of 0.5 % ortho-phosphoric acid in water, with the pH adjusted to 4.5 using 50 % sodium hydroxide. Mobile phase was delivered isocratically at 1.2 mL/min, and eluent was monitored at 210 nm. Chromatographic separation was achieved on a Phenomenex (Torrance, Ca, USA) Luna^™ strong cation exchange (SCX) column (4.6 × 250 mm; 5 μm; 100 Å) protected by a Phenomenex guard cartridge of similar material. The injection volume was 25 μL.

2.4. Bioanalytical method development

The bioanalytical method for quantification of areca alkaloids in human liver subcellular fraction matrix was validated in general accordance with ICH guidelines [23]. Calibration curves were generated by injecting 8–9 concentrations of mixtures of standards and recording the resultant area (target peak area divided by the IS area). Assay linearity was evaluated by linear regression utilizing the methods of least squares, and correlation was expressed as the coefficient of determinant (R²). The limit of detection (LOD) and lower limit of quantitation (LLOQ) were determined based on a signal-to-noise ratio 3:1 for LOD and 10:1 for LLOQ.

Precision of the assay was expressed as percent RSD (relative standard deviation). Intra-day precision of the assay was determined by calculating the percent RSD for the analysis of three quality control samples (0.1, 1, 10 μg/ml) in triplicate on a single day, and inter-day precision of the assay was determined by analyzing 3 quality control samples in triplicate on 3 separate days. Accuracy was expressed as percent relative error (RE). Intra-day accuracy of the assay was determined by calculating RE following analysis of three quality control samples (0.1, 1, 10 μg/ml) in triplicate on a single day, and inter-day accuracy was determined by analyzing 3 quality control samples in triplicate on 3 separate days. To evaluate short-term stability, alkaloids (1 and 10 μg/ml) were prepared in sodium phosphate buffer at room temperature and concentrations were measured at pre-determined time points up to 6 h in triplicate. Assay selectivity and specificity were determined by comparing the chromatograms of the supernatant of fresh and thermally-inactivated HLM or HLC without analytes, to chromatograms with spiked supernatant; experiments were repeated a total of six times. Robustness of the assay was tested by comparing results after slight deliberate adjustments to the analytical conditions (detection wavelength, mobile phase, and flow rate).

2.5. Microsomal and cytosolic incubations

Corning^® UltraPool^™ HLM and HLC contain a donor pool of mixed sex (n = 150), simulating drug metabolism activity in an average individual. Pilot experiments were initially conducted to select an optimal incubation time and microsomal or cytosolic protein amount that resulted in metabolite formation in the linear range. Incubation mixtures (total volume = 100 μL) consisted of microsomes or cytosol (0.2 mg/ml protein content), 100 mM sodium phosphate buffer (pH 7.4), and arecoline or guvacoline. Control experiments included thermally inactivated HLM or HLC to correct for nonenzymatic hydrolysis. For the pre-incubation step, liver fraction was combined with sodium phosphate buffer in a 500 μL microcentrifuge tube, then placed into a Thermo Fisher Scientific Precision^™ shaking water bath maintained at 37 °C. Tubes were gently agitated at 65 rpm for 5 min, then substrate (arecoline or guvacoline) was added to initiate metabolic reaction. After 15 min, reactions were quenched with 100 μL ice-cold acetonitrile fortified with 10 μg/mL of imipramine (internal standard), vigorously vortexed mix, and then immediately placed on ice. Samples were centrifuged for 12 min at 21,100 relative centrifugal force (rcf) in a Sorvall Legend Micro 21R centrifuge (Thermo Fisher Scientific) kept at 4 °C. An aliquot of clear supernatant (100 μL) was transferred into an autosampler vial and 25 μL was injected into the HPLC system for quantification of arecoline or guvacoline and their corresponding hydrolytic metabolites (arecaidine or guvacine, respectively).

To compare enzymatic vs. non-enzymatic hydrolysis, three different arecoline and guvacoline concentrations (0.01 mM, 0.1 mM and 1.0 mM) were tested under the incubation conditions described above. Alkaloid concentrations were quantified at pre-determined time points (0, 5, 10, 20, 30, 60, and 90 min). For determining enzyme kinetics of the reactions, a range of arecoline (0.01–8 mM) or guvacoline (0.025–8 mM) concentrations were tested. All incubations were performed in at least triplicate.

2.6. Unbound fraction in plasma and microsomes

The fraction of arecoline and guvacoline unbound to plasma proteins $\left(f_{u.p}\right)$ was determined according to a published method [24]. In brief, a set of control samples containing only arecoline or guvacoline were prepared in sodium phosphate buffer (pH 7.4), while another set of samples contained alkaloids, buffer, and concentration of plasma protein corresponding to microsomal incubations. Samples were incubated at 37 ◦C for 15 min then transferred to Amicon-Ultra-0.5 centrifugal filter devices (molecular cut-off 30,000 Da) and subjected to centrifugation at 21,100 rcf for 30 min. Next, an aliquot of ultrafiltrate was transferred to an autosampler vial and 25 μl was injected into the HPLC system.

The $f_{u,p}$ was calculated from the analyte peak area ratio of arecoline or guvacoline contained in the plasma ultrafiltrate $\left({\text{Area}}_{\text{p}}\right)$ to the analyte peak area of arecoline or guvacoline contained in the buffer solution ultrafiltrate $\left({\text{Area}}_{\text{buf}}\right)$, according to the following equation:

$$f_{u.p}=\frac{{\text{Area}}_{\text{p}}}{{\text{Area}}_{\text{buf}}}$$

Unlike CYP450 enzymes and various other drug metabolizing enzymes, hydrolases do not require additional co-factors to initiate the reaction, complicating determination of the fraction of drug unbound in microsomes $\left(f_{u.m}\right)$ [25]. Therefore, the $f_{u.m}$ was calculated using the Hallifax and Houston equation [26] for weak bases as follows:

$$f_{u.m}=\frac{1}{1+\text{C}•{10}^{0.072•\log {\text{p}}^2+0.067•\log \text{P}-1.126}}$$

where C is the microsomal protein concentration (mg/ml). For this equation, a log P value is required. Chemaxon (Chemicalize) software was utilized to estimate the log P of arecoline and guvacoline.

2.7. Metabolic reaction phenotyping

The involvement of esterases in hydrolysis of arecoline was investigated by employing inhibitors of various human hydrolases. These include bis(4-nitrophenyl)-phosphate (BNPP) [27], digitonin [28], ethanol [29], loperamide [27], 1,5-Bis(4-allyldimethylammoniumphenyl)pentan-3-one dibromide (BW284c51) [30], ethopropazine [30], and vinblastine [27], which are inhibitors of CES (non-specific), CES1, CES1, CES2, acetylcholinesterase (AChE), butyrylcholinesterase (BuChE), and arylacetamide deacetylase (AADAC), respectively. The reaction conditions were generally similar to the kinetic assays with some modifications. Incubates contained sodium phosphate buffer (100 mM; pH 7.4), arecoline (25 μM), HLM (0.2 mg/ml) or HLC (0.2 mg/ml), and inhibitors pre-dissolved in DMSO (final DMSO concentration was kept at ≤1 %). Concentrations of inhibitors were approximately equal to their respective IC₅₀ reported in the literature: loperamide (1 μM), digitonin (20 μM), ethanol (41 mM; 0.19 g/dl), BNPP (10 μM); BW284c51 (25 μM), ethopropazine (1 μM), and vinblastine (10 μM). Components of control reactions were similar except DMSO was used in place of inhibitor. Inhibitors were pre-incubated with liver preparation for 5 min, prior to initiating reactions. After 15 min incubation at 37 °C, samples were processed and then analyzed on the HPLC-UV apparatus as stated above.

Since the tissue expression of CES1 is predominantly in the liver, whereas CES2 is primarily expressed in the intestine [20], the amount of arecaidine formed was compared between microsomal and cytosolic fractions of human liver and intestine. Furthermore, arecoline hydrolysis was compared between human recombinant CES1 and CES2, two major esterases that metabolize xenobiotics [20]. Two independent experiments were performed, in which incubates varied in respect to incubation time (5–40 min) and CES protein concentration (0.1–0.4 mg/ml).

2.8. Data analysis

The metabolism plots (rate of metabolite formation vs. substrate concentration) were subjected to curve fitting using the Enzyme Kinetics toolbox of SigmaPlot v13 software (Grafiti LLC; Palo Alto, CA, USA). The data best fit the Michaelis-Menten equation according to equation #1:

$$\text{v}=\frac{{\text{V}}_{\max }•\left[\text{S}\right]}{{\text{K}}_{\text{m}}+\left[\text{S}\right]}$$ (1)

in which ${\text{K}}_{\text{m}}$ = Michaelis–Menten constant representing the substrate (arecoline or guvacoline) concentration at half maximal velocity; $\text{v}$ = rate (velocity) of metabolite formation; $\text{S}$ = substrate (arecoline or guvacoline) concentration; and ${\text{V}}_{\text{max}}$ = maximal rate of metabolite formation. The units of ${\text{K}}_{\text{m}}$ and ${\text{V}}_{\text{max}}$ were μM and nmol/min per mg protein, respectively.

In vitro intrinsic clearance $\left({\text{Cl}}_{\text{int}.\mathit{in}vitro}\right)$ expressed in units of μl/min/ mg, was calculated similar to our previous work [31] according to equation #2:

$${\text{Cl}}_{\mathrm{int}.\mathit{in}vitro}=\frac{{\text{V}}_{\max }}{{\text{K}}_{\text{m}}}$$ (2)

The ${\text{Cl}}_{\text{int}.\mathit{in}vitro}$ was extrapolated to in vivo intrinsic clearance $\left({\text{Cl}}_{\mathrm{int}.\mathit{in}\mathit{vivo}}\right)$ using allometric scaling factors [32] according to equation #3:

$${\text{Cl}}_{\text{int}.\mathit{in}\mathit{vivo}}={\text{Cl}}_{\text{int}.\mathit{in}virro}•\frac{\text{mg}\text{of}\text{protein}}{\text{gram}\text{of}\text{liver}}•\frac{\text{gram}\text{of}\text{liver}}{\text{kg}\text{of}\text{body}\text{weight}}$$ (3)

where a microsomal protein per gram of liver (MPPGL) of 32 mg/g and liver weight of 20 g/kg were used. For enzyme kinetics experiments in HLC, a cytosolic protein per gram of liver (CPPGL) of 45 mg/g was used [33].

Total hepatic clearance $\left({\text{Cl}}_{\text{H}}\right)$ was calculated using the well-stirred model according to equation #4:

$${\text{Cl}}_{\text{H}}=\frac{\text{Q}•\frac{{\text{f}}_{\text{u}.p}}{{\text{f}}_{\text{u}.m}}•{\text{Cl}}_{\text{int}.\mathit{in}\mathit{vivo}}}{\text{Q}+\left(\frac{{\text{f}}_{\text{u}.p}}{f_{\text{u}.\text{m}}}•{\text{Cl}}_{\text{int}.\mathit{in}\mathit{vivo}}\right)}$$ (4)

where $Q$ is hepatic blood flow of 21 ml/min/kg [32], and $f_{u.m}$ and $f_{u.p}$ are fraction unbound in microsomes and plasma, respectively.

The hepatic extraction ratio $\left({\text{E}}_{\text{H}}\right)$ was calculated according to equation #5:

$${\text{E}}_{\text{H}}=\frac{\frac{{\text{f}}_{\text{u}.\text{p}}}{{\text{f}}_{\text{u}.m}}•{\text{Cl}}_{\text{int}.\mathit{in}\mathit{vivo}}}{\text{Q}+\left(\frac{{\text{f}}_{\text{u}.\text{p}}}{{\text{f}}_{\text{u}.\text{m}}}•C{l}_{\text{int}.\mathit{in}\mathit{vivo}}\right)}$$ (5)

Statistical differences between effects of chemical inhibitors on arecoline hydrolase activity compared to control was tested by ordinary One-Way ANOVA followed by Tukey’s multiple comparisons t-test using Prism 7.03 (GraphPad Software; Boston, MA, USA). The a priori level of significance was set at p < 0.05.

3. Results

3.1. Method and validation

The retention time of internal standard (imipramine), arecaidine and arecoline were approximately 4.1 min, 6.4 min, and 9.3 min, respectively (Fig. 2); whereas the retention time of guvacine and guvacoline were approximately 5.6 min and 7.6 min, respectively (Fig. 3). A total run time of 10 min was used to ensure complete removal of all analytes from the system. Similar retention times of analytes were found in the cytosol assays.

Fig. 2.

Representative HPLC-UV chromatograms following injection of (A) Thermally-inactivated (boiled) HLM with no analytes (blank); (B) Thermally-inactivated HLM fortified with 0.5 μg/ml of arecaidine and arecoline; (C) Thermally-inactivated HLM exposed to arecoline (10 μM) at 37 °C for 15 min; and (D) Fresh HLM exposed to arecoline (10 μM) at 37 °C for 15 min. The approximate retention times of imipramine (IS), arecaidine and arecoline were 4.1 min, 6.4 min and 9.3 min, respectively.

Fig. 3.

Representative HPLC-UV chromatograms following injection of (A) Thermally-inactivated (boiled) HLM with no analytes (blank); (B) Thermally-inactivated HLM fortified with 0.5 μg/ml of guvacine and guvacoline; (C) Thermally-inactivated HLM exposed to guvacoline (50 μM) at 37 °C for 15 min; and (D) Fresh HLM exposed to guvacoline (50 μM) at 37 °C for 15 min. The approximate retention times of imipramine (IS), guvacine and guvacoline were 4.1 min, 5.6 min and 7.6 min, respectively.

Calibration curves were best fit using linear regression with 1/concentration² (1/x²) weighted sum of squares. In the supernatant of liver microsomes, linear response over a concentration range of 0.025–10 μg/mL, 0.050–10 μg/mL, 0.039–100 μg/mL, and 0.039–10 μg/mL was observed for arecoline (R² = 0.9998), arecaidine (R² = 0.9997), guvacoline (R² = 0.9986), and guvacine (R² = 0.9992), respectively. The LOD, determined by a signal to noise ratio of at least 3:1, was 0.025 μg/mL, 0.050 μg/mL, 0.039 μg/mL, and 0.039 μg/mL for arecoline, arecaidine, guvacoline and guvacine, respectively. In addition, the LLOQ, determined by a signal to noise ratio of at least 10:1, was 0.050 μg/mL, 0.095 μg/mL, 0.060 μg/mL, and 0.054 μg/mL for arecoline, arecaidine, guvacoline and guvacine, respectively.

In the supernatant of liver cytosol, linear responses for concentration ranges of 0.024–10 μg/mL, 0.040–10 μg/mL, 0.038–10 μg/mL, and 0.038–10 μg/mL were observed for arecoline (R² = 0.9970), arecaidine (R² = 0.9987), guvacoline (R² = 0.9988), and guvacine (R² = 0.9997), respectively. The LOD was 0.024 μg/mL, 0.040 μg/mL, 0.038 μg/mL, and 0.038 μg/mL for arecoline, arecaidine, guvacoline and guvacine, respectively. Furthermore, the LLOQ was 0.074 μg/mL, 0.045 μg/mL, 0.049 μg/mL, and 0.044 μg/mL for arecoline, arecaidine, guvacoline and guvacine, respectively.

Results for intra-day and inter-day precision were deemed acceptable with RSD values ≤ 6.4 % for arecoline and arecaidine, and ≤12 % for guvacoline and guvacine, in the supernatant of microsomes. Accuracy was within acceptable limits with RE values ranging from −7.9 % to 6 % for arecoline and arecaidine, and −9.5 %–12.7 % for guvacoline and guvacine. The short-term stabilities of arecoline, arecaidine, guvacoline and guvacine after 6 h at room temperature were established. The amount of intact drug loss ranged from 1 to 6%, with %CV not exceeding 10 % in any samples. The results of intra- and inter-day precision and accuracy and short-term stabilities in supernatant of cytosol were also deemed acceptable, with values similar to those found in microsomes. To determine the specificity of the assay, the supernatant of fresh and thermally-inactivated liver fractions, without analytes, was analyzed for potential interferences at the retention time of the analytes. No coeluting peaks were observed in any samples. The results of the robustness test demonstrated that after conscious alterations of detection wavelength, mobile phase, and flow rate, the performance of the chromatographic system did not essentially change.

Overall, the validation tests demonstrate the bioanalytical method was suitable to evaluate the kinetics of arecaidine and guvacine formation in liver microsomes and liver cytosol through good selectively, precision, accuracy and sensitivity.

3.2. In vitro hydrolysis

The depletion of parent compound (arecoline or guvacoline) was monitored over 90 min in the presence of fresh HLM or thermally-inactivated HLM. Incubation samples containing alkaloid plus thermally-inactivated HLM exhibited ≤5.6 % loss, indicating minimal spontaneous (non-enzymatic) hydrolysis. Arecoline depletion in HLM best followed a monoexponential decay model with an elimination half-life (±SD) of 15.7 (±2.75) min (Fig. 4a), 16.6 (±3.77) min, and 17.2 (±2.71) min at starting concentrations of 10 μM, 100 μM and 1000 μM, respectively. Compared to arecoline, guvacoline exhibited only minimal depletion in HLM over time at all concentrations. For instance, at 90 min arecoline levels were reduced to 5.6 % control, whereas guvacoline concentrations remained at only 98 % control (Suppl Fig. 1). Consistent with these results, the amount of arecaidine formed over 90 min in HLM was substantially greater (approximately 30-fold) than guvacine (Suppl Fig. 1).

Fig. 4.

(A) Depletion of arecoline over 90 min in the presence of fresh HLM (▲) or HLC (); (B) Comparison of amount of arecaidine formed over 90 min in the presence of HLM (○) or HLC (). Each point represents the mean (±SD) of at least triplicate experiments.

Likewise, depletion of arecoline and guvacoline was monitored over 90 min in the presence of fresh or thermally-inactivated HLC. Incubation samples containing alkaloid plus thermally-inactivated HLC exhibited ≤7.2 % loss, indicating minimal non-enzymatic hydrolysis. Arecoline depletion plot over time in the presence of HLC or HLM at an initial arecoline concentration of 10 μM is included in Fig. 4a, demonstrating a notable greater loss of arecoline in HLM samples compared to HLC samples. For example, at 90 min there was approximately 12.8-fold greater concentrations of arecoline remaining in HLC samples compared to HLM samples (Fig. 4a). Consistent with these results, the amount of arecaidine formed over 90 min was approximately 12-fold greater in HLM samples vs. HLC samples (Fig. 4b).

Similar to in HLM, arecoline depletion in HLC best followed a monoexponential decay model with elimination half-lives (±SD) of 97 (±14) min (Figs. 4a), 99.8 (±15.7) min, and 126 (±26.5) min at starting concentrations of 10 μM, 100 μM and 1000 μM, respectively. At all tested concentrations, only minimal depletion of guvacoline in the presence of HLC was found (data not shown).

The enzyme kinetics of in vitro hydrolysis of arecoline and guvacoline were evaluated under linear conditions with respect to HLM or HLC protein concentration (0.2 mg/ml) and incubation time (15 min). Arecoline metabolism to arecaidine in HLM best obeyed Michaelis-Menten kinetics. A plot depicting rate of arecaidine formation as a function of arecoline concentration, along with the corresponding Eadie-Hofstee plot (inset), are included in Fig. 5a. For arecoline, modeling computed a ${\text{K}}_{\text{m}}$ of 8.68 mM (±1.64 mM) and ${\text{V}}_{\text{max}}$ of 783 nmol/min per mg protein (±91.4 nmol/min/mg protein) (Table 1). The apparent ${\text{Cl}}_{\text{int}.\mathit{in}vitro}$, calculated based on equation #2, was 90.3 μL/min/mg protein. Extrapolating to the human situation using scaling factors, the apparent ${\text{Cl}}_{\mathrm{int}.\mathit{in}\mathit{vivo}}$ was 57.8 mL/min/kg (Table 1). Using the fraction of arecoline unbound to plasma and microsomal proteins (0.978 and 0.983, respectively), the ${\text{Cl}}_{\text{H}}$ and ${\text{E}}_{\text{H}}$ calculations resulted in values of 15.4 mL/min/kg and 0.733, respectively (Table 1).

Fig. 5.

Non-linear regression fitting of the formation of arecaidine at varying arecoline concentrations in HLM (A) and HLC (B). The solid line represents the curve of the best fit; the inset is the corresponding Eadie–Hofstee plot. Each point represents the mean (±SD) of at least triplicate experiments.

Table 1.

Kinetic parameters of arecaidine and guvacine formation in HLM and HLC.

Human Liver Microsomes (HLM)
Parameter	Hydrolysis Reaction
	Arecoline → Arecaidine	Guvacoline → Guvacine
${\text{K}}_{\text{m}}$ (mM)	8.68	34.3
${\text{V}}_{\text{max}}$ (nmol/min/mg protein)	783	35
${\text{Cl}}_{\text{int}.\mathit{in}vitro}$ (μl/min/mg protein)	90.3	1.02
${\text{Cl}}_{\mathrm{int}.\mathit{in}\mathit{vivo}}$ (ml/min/kg)	57.8	0.654
ClH (ml/min/kg)	15.4	0.643
${\text{E}}_{\text{H}}$	0.733	0.0306
$f_{u.p}$	0.978	0.998
$f_{u.m}$	0.983	0.985
Human Liver Cytosol (HLC)
Parameter	Hydrolysis Reaction
	Arecoline → Arecaidine	Guvacoline → Guvacine
${\text{K}}_{\text{m}}$ (mM)	9.11	27.8
${\text{V}}_{\text{max}}$ (nmol/min/mg protein)	118	14.4
${\text{Cl}}_{\text{int}.\mathit{in}vitro}$ (μl/min/mg	12.9	0.518
protein) ${\text{Cl}}_{\mathrm{int}.\mathit{in}\mathit{vivo}}$ (ml/min/kg)	11.6	0.466
ClH (ml/min/kg)	n.c.	n.c.
${\text{E}}_{\text{H}}$	n.c.	n.c.

The kinetics of guvacoline hydrolysis to guvacine in HLM best obeyed the Michaelis-Menten model. Formation of guvacine at varying concentrations of guvacoline (along with corresponding Eadie-Hofstee plot) in the presence of HLM are depicted in Suppl Fig. 2. The estimated ${\text{K}}_{\text{m}}$ and ${\text{V}}_{\text{max}}$ were 34.3 mM (±13 mM) and 35 nmol/min per mg protein (±11.1 mM), respectively (Table 1). The apparent ${\text{Cl}}_{\text{int}.\mathit{in}vitro}$ and ${\text{Cl}}_{\mathrm{int}.\mathit{in}\mathit{vivo}}$ were 1.02 μL/min/mg protein and 0.654 ml/min/kg, respectively, both of which were dramatically less than the clearance estimates for arecoline. Plasma protein binding experiments demonstrated that $f_{u.p}$ was 0.998, and calculations determined an estimated $f_{u.m}$ of 0.985. Accordingly, guvacoline ${\text{Cl}}_{\text{H}}$ was 0.643 ml/min/kg with an ${\text{E}}_{\text{H}}$ of 0.0306.

In HLC, arecoline kinetic data best fit to the Michaelis-Menten model (Fig. 5b), with a ${\text{K}}_{\text{m}}$ of 9.11 mM (±1.89 mM) and ${\text{V}}_{\text{max}}$ of 118 nmol/min/ mg protein (±15.3 nmol/min/mg protein) (Table 1). The apparent Cl_{int. in vitro}, calculated based on equation #2, was 12.9 μL/min/mg protein. After extrapolation, the apparent ${\text{Cl}}_{\mathrm{int}.\mathit{in}\mathit{vivo}}$ was 11.6 mL/min/kg. For guvacine formation in HLC, the Michaelis-Menten graph with corresponding Eadie Hofstee plot (inset) are included in Suppl Fig. 3. Enzyme kinetic modeling computed a ${\text{K}}_{\text{m}}$ of 27.8 mM (±5.40 mM) and ${\text{V}}_{\text{max}}$ of 14.4 nmol/min/mg protein (±2.28 nmol/min/mg protein) (Table 1). The apparent ${\text{Cl}}_{\text{int}.\mathit{in}vitro}$ and ${\text{Cl}}_{\mathrm{int}.\mathit{in}\mathit{vivo}}$ were 0.518 μL/min/mg and 0.466 mL/min/kg, respectively. To our best knowledge, a standardized equation to estimate fraction of drug bound to cytosolic proteins is not readily available, therefore the ${\text{Cl}}_{\text{H}}$ and ${\text{E}}_{\text{H}}$ were not calculated for kinetics in cytosol.

3.3. Phenotyping arecoline hydrolysis reaction

Phenotyping of guvacoline hydrolysis reactions was not carried out since the in vitro and extrapolated in vivo intrinsic clearance of guvacoline in HLM and HLC was minimal in comparison to arecoline hydrolysis. Employing pharmacological inhibitors of esterases, the average percent activity (±SD) of arecaidine formation compared to control in HLM exposed to loperamide (CES2 specific inhibitor), digitonin (CES1 specific inhibitor), ethanol (CES1 specific inhibitor), and BNPP (non-specific CES inhibitor) was 100.2 % (±4.31 %), 70 % (±6.21 %; p < 0.0001), 44.6 % (±2.38 %; p < 0.0001), and 14.4 % (±0.77 %; p < 0.0001), respectively (Fig. 6a). Moreover, the mean percent activity (±SD) of arecoline hydrolase activity when exposed to BW284c51 (AChE specific inhibitor), ethopropazine (BuChE specific inhibitor), or vinblastine (AADAC specific inhibitor) was 103 % (±6.18 %), 102 % (±5.58 %), and 100 % (±2.14), respectively (Fig. 6a). The effects of esterase inhibitors on arecoline hydrolase activity in HLC exhibited a similar pattern as in HLM, with digitonin, ethanol and BNPP all showing significant reductions (p < 0.0001) in mean percent activity compared to control, whereas HLC exposed to loperamide, BW284c51, ethopropazine, or vinblastine demonstrated no significant change in percent activity (Fig. 6b).

Fig. 6.

Effects of esterase inhibitors on arecoline hydrolysis in HLM (A) and HLC (B). All assays were conducted in quadruplicate, and the data is expressed as mean ± SD. ****p < 0.0001

n.d. (not detected).

Differential hydrolysis of arecoline was tested in the presence of microsomes and cytosol of human tissues (liver and intestine), along with human recombinant CES1 and CES2. Results clearly indicate there was no arecoline hydrolysis in any incubate containing intestinal sub-cellular fractions (Fig. 7a). Furthermore, in both the time-dependent (Fig. 7b) and protein-dependent experiments (Fig. 7c), there was sub-stantial hydrolysis of arecoline by CES1 but no hydrolysis in any incubates containing CES2.

Fig. 7.

Reaction phenotyping assays of arecoline hydrolysis in microsomes and cytosol from human tissue (liver and intestinal) (A), and human recombinant CES enzymes (B; C).

All assays were performed in quadruplicate, with data expressed as mean ± SD.

n.d. (not detected).

4. Discussion

Our first hypothesis was the methyl ester AN xenobiotics, arecoline and guvacoline, are hydrolyzed in HLM and HLC to their corresponding carboxylate metabolites arecaidine and guvacine, respectively. To test this hypothesis, we conducted a sequence of studies with an initial focus on developing a bioanalytical assay utilizing HPLC-UV to carry out these experiments. For the first time, this study shows the quantification of areca alkaloids (arecoline, arecaidine, guvacoline and guvacine) following in vitro incubations in liver microsomes and cytosol. One of the challenges in chromatographic analysis of small molecule weak bases with minimal lipophilicity is achieving adequate separation from other substituents in the matrix while optimizing peak shape (minimizing peak tailing) and avoiding extended run times. Using traditional reversed phase HPLC methods, we routinely encountered these issues in our previous work with arecoline, arecaidine and synthesized metabolites, as well as cocaine analogs with similar chemical properties (unpublished data). In this situation ion exchange chromatography is a plausible alternative whereby weak bases are readily protonated by chemical reactions in the mobile phase, forming a cationic species that is retained on a negatively charged stationary phase. We were able to sufficiently separate arecoline and guvacoline from their hydrolytic metabolites on a strong cation exchange (SCX) column consisting of benzenesulphonic acid functional groups. Further validation tests (e.g., linearity, quantification limits, precision, accuracy) demonstrated the assay was suitable to accomplish our next goal of delineating hydrolytic kinetics of these two areca nut alkaloids in HLM and HLC.

Another primary finding of our work was rapid depletion of arecoline (elimination T_1/2 < 20 min) in the presence of HLM. This result was similar to findings from a pre-clinical study in rats and a human clinical trial. For example, subjects with Alzheimer’s disease treated with arecoline had an average elimination T_1/2 in plasma of 10.5 min [34] and pharmacokinetic studies in rats found a slightly shorter plasmatic half-life, ranging from 3.5 to 6 min [35]. In the present study, the ${\text{V}}_{\text{max}}$ of arecaidine formation was ~800 nmol/min per mg of microsomal protein in physiologically-simulated incubations. This data suggests, although does not prove, hydrolysis is a major driver of arecoline clearance in HLM and the liver is most likely a major organ involved in its metabolism in vivo.

In comparison to other kinetic models tested (e.g., Hill, Substrate Inhibition, Substrate Activation), arecoline conversion to arecaidine in HLM best fit the Michaelis-Menten equation with a ${\text{K}}_{\text{m}}$ in the millimolar range (8.68 ± 1.64 mM). To our best knowledge, the only other study to report arecoline enzyme kinetics in vitro was the work of Patterson and Kosh [16]. In mouse liver homogenate, the maximum rate of arecoline depletion was 4.7 nmol/min/mg protein, with a ${\text{K}}_{\text{m}}$ of 9.6 mM [16]. Their substrate depletion approach was limited by the lack of quantifying metabolite formation kinetics, but interesting to note that the ${\text{K}}_{\text{m}}$ between their work in mice tissue and ours in human tissue was very similar.

In vitro-in vivo extrapolation (IVIVE) is a transposition of quantitative data acquired in vitro to predict in vivo response in humans. For drug metabolism, kinetic data obtained in vitro (e.g., apparent intrinsic clearance) can be extrapolated to the human condition by using scaling factors to derive clearance rates and subsequently hepatic extraction ratio. Among the key parameters for these quantitative extrapolations is fraction of drug unbound to plasma and microsomal proteins, because it is widely thought that free drug, compared to bound drug, undergoes elimination and exerts pharmacological and toxicological effects (“free drug hypothesis”). Arecoline $f_{u.p}$ was 0.978, indicating a small amount of arecoline (~2 %) is predicted to bind plasma proteins such as albumin and α1-acid glycoprotein, leaving ~98 % available for distribution to peripheral tissues. This may explain, in part, arecoline’s prominent adverse toxicological profile involving multiplex organ systems. It is not unusual that arecoline is modestly lipophilic (log P = 0.65) and has minimal binding to plasma proteins since the degree of lipophilicity is directly proportional to amount of protein binding for weak bases [36].

Unlike CYP substrates, significant challenges exist in predicting $f_{u.m}$ of drugs metabolized by esterases because they do not require co-factors for activity. An alternative route is to estimate $f_{u.m}$ using published equations, such as the Hallifax and Houston [26] and Austin et al. [37] equations. In the present study, arecoline $f_{u.m}$ of 0.983 (calculated via the Hallifax and Houston equation) was used for human extrapolations. As an exploratory exercise, we calculated $f_{u.m}$ using the Austin equation, yielding 0.982, a virtually identical value to the $f_{u.m}$ calculated by the Hallifax and Houston equation.

Hepatic clearance can be defined as the volume of hepatic blood flowing through the liver that is cleared of drug per unit of time. Together, the ${\text{Cl}}_{\mathrm{int}.\mathit{in}\mathit{vivo}}$, $f_{u.p}$, $f_{u.m}$, and blood flow rate were used to estimate ${\text{Cl}}_{\text{H}}$. For arecoline, ${\text{Cl}}_{\text{H}}$ was equivalent to 15.4 ml/min/kg of body weight, with an ${\text{E}}_{\text{H}}$ of 0.733. Drugs with $E_H<0.3$, 0.3–0.7 and > 0.7 are considered low, intermediate and high extraction, respectively [38]. Therefore, our results indicate arecoline is a high extraction ratio drug, which means it is removed by the liver almost as rapidly as the liver is perfused with blood in which the drug is contained (“flow--dependent”). This suggests that disease states or drugs that impact blood flow may substantially influence the overall toxicokinetic and toxicodynamic profiles of arecoline.

Unexpectedly, guvacoline was only minimally hydrolyzed to guvacine in HLM under the experimental conditions reported herein. Compared to arecoline, guvacoline ${\text{V}}_{\text{max}}$, ${\text{Cl}}_{\mathrm{int}.\mathit{in}\mathit{vivo}}$, ${\text{Cl}}_{\text{H}}$, and ${\text{E}}_{\text{H}}$ in HLM were 96 %, 99 %, 96 %, and 96 % less, respectively. Structurally, guvacoline (norarecoline) is the N-demethylated derivative of arecoline (Fig. 1). There are examples in the literature of similar differences in hydrolysis. For instance, cocaine and its ethanolic metabolite cocaethylene are structurally similar in part to arecoline with N-methyl and ester functional groups. Interestingly, removal of the N-methyl group of cocaine and cocaethylene substantially decreases binding affinity to CES1 [39]. Furthermore, cocaine is hydrolyzed by a novel cocaine esterase to a far greater extent than norcocaine [40]. The authors of these cocaine studies did not further explore the mechanisms for the varying hydrolytic rates. We speculate the following chemical properties (individually or combination thereof) may explain the present findings of more efficient hydrolysis of arecoline than guvacoline in HLM and HLC: degree of basicity, substituent effects on electron withdrawal/donating properties, steric effects, and/or propensity for hydrogen bonding. Overall, this is an intriguing finding that warrants future research, such as molecular modeling.

Our second hypothesis was that arecoline hydrolysis is primarily driven by CES1, a major phase I drug metabolizing enzyme expressed in the liver. Reaction phenotyping studies determined that CES1-specific inhibitors, digitonin and ethanol, significantly reduced arecoline hydrolase activity, whereas the specific CES2 inhibitor loperamide exhibited no significant inhibition. Moreover, arecoline hydrolase activity was unimpeded by specific inhibitors of BuChE, AChE and AADAC. It is not surprising that arecoline is a substrate for CES1, since this enzyme prefers drugs with a small alcohol moiety and larger acyl group [20] that are present on the arecoline molecule. Contrarily, CES2 prefers substrates with a large alcohol moiety and smaller acyl group [20]. Additionally, reaction phenotyping in the present study found arecoline is not hydrolyzed in human intestinal subcellular fractions. This finding further confirms that arecoline is a substrate for CES1, and not CES2, because CES1 predominantly resides in the liver with marginal expression in the intestine, whereas CES2 is located primarily in the intestine with lower amounts in the liver [20].

Another intriguing finding of our work was arecoline, and to a lesser extent guvacoline, are also hydrolyzed in liver cytosol. Although CES1 and CES2 are primarily localized in the endoplasmic reticulum, they are also found in lesser amounts within the cytosolic fraction [41]. Consistent with these cellular localization patterns, hydrolysis rates were substantially lower in HLC compared to HLM in the present study. Reaction phenotyping experiments in HLC showed a similar pattern as in HLM, in which CES1 is a major driver in cytosolic hydrolysis of arecoline.

At present, it is unclear about the toxicological significance of arecoline hydrolysis. Most of the current data demonstrates arecoline is more toxic than arecaidine, suggesting hydrolysis is a detoxification step. For example, in cellular experiments arecoline-induced cytotoxicity can be mitigated by addition of pig liver esterase [42]. Furthermore, the electrophilic properties of arecaidine are substantially lower compared to arecoline, rendering arecaidine less likely to spontaneously react with GSH and N-acetylcysteine and less propensity to form covalent cysteine adducts on proteins [43]. Arecoline is more mutagenic than arecaidine in Salmonella typhimurium tester strains [44]. In addition, arecaidine is a more polar molecule than arecoline and presumably more readily excreted from the body. Finally, the median lethal dose (LD₅₀) of arecaidine is about 20-fold greater than arecoline [14].

Conversely, arecaidine demonstrated similar mitogenic properties as arecoline towards buccal and gingival fibroblasts [45]. Arecaidine is also a more potent stimulator of collagen synthesis than arecoline in human buccal mucosal fibroblasts [46], suggesting it plays a significant role in developing oral submucous fibrosis (OSF) – a major adverse effect of AN chewing. In addition, arecaidine has psychoactive properties via reducing the uptake of the γ-aminobutyric acid (GABA), a primary neurotransmitter that suppresses the CNS [10].

For guvacoline, the clinical toxicokinetic significance of hydrolysis remains unknown, with the present results showing very minor hydrolysis occurs in the liver. Guvacoline can readily deplete cellular thiols unlike guvacine [9], overall suggesting that hydrolysis of guvacoline, which inherently reduces the electrophilicity of its alpha-beta unsaturated carbonyl system, is a detoxification step. During AN chewing, guvacoline is converted to N-nitrosoguvacoline, a more toxic metabolite; therefore, hydrolysis would presumably counteract formation of this metabolite. However, data from the present data infers hydrolysis would compete little with nitroso conversion in the liver. The nitroso derivative of guvacine, N-nitrosoguvacine, has demonstrated little toxicological relevance in previous studies [9]. However, guvacine psychoactivity, like arecaidine, is mediated by competitive inhibition of GABA uptake in the CNS [10]. Whether the muscarinic-mediated effects of guvacoline are more clinically relevant than guvacine-mediated GABA effects (or vice versa) in vivo is unknown.

5. Conclusion

Presented herein, we describe a sensitive and precise HPLC-UV bioanalytical assay capable of rapidly quantifying arecoline and guvacoline, methyl ester alkaloids of the pernicious AN, and their carboxylate metabolites arecaidine and guvacine, respectively. The developed assay was successfully applied to deduce the in vitro kinetics and predict in vivo clearance rates of these compounds. This is the first report, to our best knowledge, demonstrating arecoline is extensively hydrolyzed (with high hepatic extraction ratio) to arecaidine in human liver. To our surprise, guvacoline was hydrolyzed at a substantially lower rate than arecoline. Taken together, our results provide scientific evidence that hydrolysis of arecoline occurs in human liver, predominantly via CES1, which is indicative of its likely in vivo fate in AN users. These findings set the stage for future work to improve our understanding of the systematic toxicokinetic and toxicodynamic profile of these areca alkaloids, predict drug interactions, and inform novel interventions to reduce their overall toxic burden in AN consumers.