In Vitro Reversible and Time-Dependent CYP450 Inhibition Profiles of Medicinal Herbal Plant Extracts Newbouldia laevis and Cassia abbreviata: Implications for Herb-Drug Interactions

This study evaluated the effects of Newbouldia laevis and Cassia abbreviata extracts on CYP450 enzyme activity. Recombinant CYP450 enzyme and fluorogenic substrates were used for evaluating inhibition, allowing the assessment of herb–drug interactions (HDI). Phytochemical fingerprinting was performed using UPLC-MS. The herbal extracts were risk ranked for HDI based on the IC50 values determined for each CYP enzyme. Newbouldia laevis inhibited CYP1A2, CYP2C9, and CYP2C19 enzyme activities with Ki of 2.84 µg/mL, 1.55 µg/mL, and 1.23 µg/mL, respectively. N. laevis exhibited a TDI (4.17) effect on CYP1A2 but not CYP2C9 and CYP2C19 enzyme activities. Cassia abbreviata inhibited CYP1A2, CYP2C9, and CYP2C19 enzyme activities showing a Ki of 4.86 µg/mL, 5.98 µg/mL, and 1.58 µg/mL, respectively. TDI potency assessment for Cassia abbreviata showed it as a potential TDI candidate (1.64) for CYP1A2 and CYP2C19 (1.72). UPLC-MS analysis showed that Newbouldia laevis and Cassia abbreviata possess polyphenols that likely give them their therapeutic properties; some of them are likely to be responsible for the observed inhibition. The observations made in this study suggest the potential for these herbal compounds to interact, especially when co-administered with other medications metabolized by these CYP450 enzymes.

mechanisms of Newbouldia laevis and Cassia abbreviata. The potential of these two herbal plants to cause HDI was also evaluated. Phytofingerprinting was performed in order to evaluate the profile of possible phytocompounds in both Newbouldia laevis and Cassia abbreviata.
The IC50 curve shift approach used to assess the TDI potency of the extracts was adopted according to Berry and Zhao [20]. The IC50 curve shift method employs a fold shift decrease in IC50 values as a result of pre-incubating with or without NADPH. The TDI curves of the extracts of NL and CA on CYP1A2, CYP2C9, and CYP2C19 are shown in Figure 3. A fold change, which leads to a decrease in IC50, is used to categorize a plant extract as a potential TDI. Plant extracts that give a ratio (IC50 from Single point concentrations of the plant extracts at 50 µg/mL were pre-incubated with or without NADPH and the percentage residual activity compared ( Figure 2). Medicinal plant extracts that were pre-incubated with NADPH showed a significant reduction (p < 0.05) in percent activity of CYP1A2, depicting its TDI potency without the possibility of enzyme recovery; this was not observed for CYP2C9 and CYP2C19 enzyme activities.
The IC 50 curve shift approach used to assess the TDI potency of the extracts was adopted according to Berry and Zhao [20]. The IC 50 curve shift method employs a fold shift decrease in IC 50 values as a result of pre-incubating with or without NADPH. The TDI curves of the extracts of NL and CA on CYP1A2, CYP2C9, and CYP2C19 are shown in Figure 3. A fold change, which leads to a decrease in IC 50 , is used to categorize a plant extract as a potential TDI. Plant extracts that give a ratio (IC 50 from pre-incubation without NADPH/IC 50 from pre-incubation with NADPH) of ě1.5 were classified as a positive TDI, as shown in Figure 4.
Extracts of NL and CA showed a TDI potency (dashed line) with an IC50 ratio of greater than 1.5 on CYP1A2 (Figure 4). The assay was validated with furafylline (a known CYP1A2 TDI compound) and α-naphthoflavone (no known TDI effect on CYP1A2). NL and CA extracts did not show any TDI effects on CYP2C9. NL did not show any TDI effects on CYP2C9; however, CA exhibited a TDI effect on CYP2C19, as shown in Figure 4.   Extracts of NL and CA showed a TDI potency (dashed line) with an IC50 ratio of greater than 1.5 on CYP1A2 (Figure 4). The assay was validated with furafylline (a known CYP1A2 TDI compound) and α-naphthoflavone (no known TDI effect on CYP1A2). NL and CA extracts did not show any TDI effects on CYP2C9. NL did not show any TDI effects on CYP2C9; however, CA exhibited a TDI effect on CYP2C19, as shown in Figure 4.   Extracts of NL and CA showed a TDI potency (dashed line) with an IC 50 ratio of greater than 1.5 on CYP1A2 (Figure 4). The assay was validated with furafylline (a known CYP1A2 TDI compound) and α-naphthoflavone (no known TDI effect on CYP1A2). NL and CA extracts did not show any TDI effects on CYP2C9. NL did not show any TDI effects on CYP2C9; however, CA exhibited a TDI effect on CYP2C19, as shown in Figure 4.

Phytofingerprinting by UPLC-MS
A non-targeted approach was used in order to provide information about all possible metabolites present in the analysed samples. The gradient parameters were adjusted by systematically changing the percentage of solvents to be able to determine as many compounds as possible. The chromatograms obtained ( Figure 8) showed the retention times of the various compounds that were found in the extracts. Using fragmentation data, and available databases, possible compounds that might be found in the extracts were proposed for CA (Table 1) and NL (Table 2). Selected structures of these polyphenols are shown in Figures 9 and 10 for Cassia abbreviata and Newbouldia laevis respectively.

Phytofingerprinting by UPLC-MS
A non-targeted approach was used in order to provide information about all possible metabolites present in the analysed samples. The gradient parameters were adjusted by systematically changing the percentage of solvents to be able to determine as many compounds as possible. The chromatograms obtained (Figure 8) showed the retention times of the various compounds that were found in the extracts. Using fragmentation data, and available databases, possible compounds that might be found in the extracts were proposed for CA (Table 1) and NL (Table 2). Selected structures of these polyphenols are shown in Figures 9 and 10 for Cassia abbreviata and Newbouldia laevis respectively

Prediction of in Vivo Herb-Drug Interaction for IC 50
It is becoming a common occurrence for in vitro data from HDI studies to be used to predict in vivo HDI risk [21][22][23]. Therefore, in this study, the putative GIT concentration that serves as the inhibitor concentration was calculated as shown in Table 3. Since the intestinal absorption and plasma concentration of each test compound are not known and with the knowledge that herbal extracts have different bioavailability [24], the bioavailable concentration was estimated using the % yield (weight extracted powdered material/weight of original starting material) to give the soluble extract available in the GIT tract. With this knowledge, differential bioavailable concentration was calculated as seen in Table 3. The IC 50 values were then compared to the calculated estimated bioavailable concentration as reported previously [23] and predictions made (Table 4). It was observed that all the IC 50 values obtained were three times lower than the estimated bioavailable concentration and therefore adequate amounts that may enter the hepatic portal vein can interact with CYP1A2, CYP2C9, and CYP2C19 enzymes. Note: HDI, herb-drug interaction, inhibitor concentration = estimated bioavailable concentration (µg/mL), * the likelihood of a clinically relevant interaction when these herbal extracts are taken is based on the assumption that the % yield serves as the bioavailable fraction, which was used in estimating the bioavailable concentration in the gut and also if there is complete absorption.

Discussion
The therapeutic efficacy of medicinal herbal plants is well acknowledged in both developed and developing countries. Medicinal plants from different countries mostly lack valid documentation on their safety, efficacy, manufacturing standards, and quality control. Given the varied amounts of compounds found in medicinal herbal plants, standardization parameters and development of marker profiles of commonly used medicinal plants are very important for maintaining the quality of herbal remedies and also provide knowledge on the optimal concentrations of the various bio-constituents [25,26]. Despite being natural, medicinal plant extracts are not free from risks and interactions with other drugs and herbs, which may lead to significant public health consequences including the possibility of being fatal [5,6]. Traditionally, whole plants or mixtures of medicinal plants are used rather than isolated compounds from these plants. Evidence indicates that crude plant extracts often have greater in vitro and/or in vivo therapeutic activity than isolated constituents at an equivalent dose. The aim of this study therefore was to evaluate the CYP450 inhibition activities of Newbouldia laevis and Cassia abbreviata extracts commonly used in West and Southern Africa for the treatment and/or management of common communicable and non-communicable diseases. It was observed that extracts from these two plants had differential inhibitory effects on CYP1A2, 2C9, and 2C19. This could be used for initial assessment of the potential risk of HDI, which would bring awareness and educate the public and/or patients on the dangers of co-use of herbal medicinal remedies and conventional medications.
Drug interactions can result in clinical fatalities; thus, in the drug discovery process, guidelines and opinions have been published by the FDA, EMA, and pharmaceutical industries [27,28] for the conducting of drug-drug interaction (DDI) studies to help target drugs that are likely to cause interactions at the early stages so that such entities can be either discontinued or modified before further processing. The potential interactions of herbal remedies with prescribed or over-the-counter drugs has been a major safety concern for clinicians and public health practitioners, as such interactions are difficult to predict and also, generally, there is a lack of available information on the composition of herbs and their pharmacological actions [29]. With the tremendous surge in the acceptance of and public interest in herbal medicinal remedies, researchers are adopting the guidelines and opinions proposed for DDI studies to study HDI. This is helpful with labelling purposes for commercially available herbal products and also to caution patients from combining herbal remedies with known conventional medications [30]. There is limited information currently available regarding the potential of Newbouldia laevis and Cassia abbreviata to interact with clinically prescribed drugs. This is a safety issue as these herbal plant remedies are commonly used by certain populations. This study therefore looked at the drug interaction potential of N. laevis and C. abbreviata extracts on three major cytochrome P450 enzymes, CYP1A2, CYP2C9, and CYP2C19. Typically, new chemical entities (NCE) are assessed for their potential to cause CYP-mediated drug interaction based on a single concentration screening and later a detailed classification involving IC 50 , TDI, and inactivation constants like K i and K inact .
Crude N. laevis and C. abbreviata extracts were used in this study to mimic the way patients take these herbal remedies and also the mode of extraction simulated to show the indigenous mode of extraction. Newbouldia laevis extract showed a strong reversible inhibitory effect on CYP1A2 and CYP2C9 enzyme activities whilst exhibiting a moderate reversible inhibition on CYP2C19. Cassia abbreviata extract showed a strong reversible inhibitory effect on CYP1A2, CYP2C9, and CYP2C19 enzyme activities. The implication is that there is a possibility of enzyme recovery should the herbal extracts inactivate these CYPs.
Time-dependent inhibition of CYP450 enzymes that is caused by NCEs is of great concern because such compounds can cause clinically relevant DDI or HDI [31]. The TDI of a CYP enzyme refers to the change in inhibition potency of a NCE as a result of a dosing period that leads to formation of inhibitory metabolites and mechanism-based inhibition [32]. The use of the IC 50 -shift approach as an initial step for TDI assessment has been recommended by the Pharmaceutical Research and Manufacturers Association (PhRMA) [27][28][29][30][31][32], where HLMs or recombinant CYPs are used to evaluate the increase in potency of a NCE to cause CYP-inhibition when the chemical entity is pre-incubated with NADPH [20,33]. This approach was used to assess the TDI potency of Newbouldia laevis and Cassia abbreviata extracts. Though the IC 50 -shift approach is recommended, there is a challenge with the interpretation of the criterion used to define a chemical as a potential TDI. PhRMA recommends the criterion of 1.2-to 3-fold decrease [32] for classification of potential TDI candidates, whilst Berry and Zhao [20] have adopted a criterion of ě1.5 as being significant for classification for TDI candidates. This study therefore adopted the Berry and Zhao approach of setting the TDI criterion at ě1.5. As seen in Figure 4, Newbouldia laevis and Cassia abbreviata showed TDI potencies of 4.08 and 1.64, respectively, on CYP1A2. TDI potency of 0.15 and 2.01 was observed for α-naphthoflavone and furafylline known non-TDI and TDI compounds respectively. Newbouldia laevis and Cassia abbreviata extracts did not exhibit any TDI potency on CYP2C9 enzyme with values of 1.18 and 1.41, respectively, implying they were not potential CYP2C9 TDI candidates. Although Newbouldia laevis extract exhibited a non-TDI effect (1.03) on the CYP2C19 enzyme, Cassia abbreviata extract exhibited a TDI effect on the CYP2C19 enzyme with a ratio of 1.72. Sulphaphenazole and miconazole, which are non-TDI compounds of CYP2C9 and CYP2C19, were used to validate the assay, respectively.
The differential reversible and irreversible inhibition of CYP1A2, CYP2C9, and CYP2C19 enzymes by Newbouldia laevis and Cassia abbreviata extracts implies that patients who are on prescribed drugs that have narrow therapeutic ranges (for example, theophylline, caffeine, imipramine, paracetamol, phenacetin-CYP1A2, warfarin, non-steroidal anti-inflammatory drugs (NSAIDs), tolbutamide, phenytoin, losartan-CYP2C9, diazepam, omeprazole, propranolol, amitriptyline-CYP2C19, or over-the-counter drugs that are metabolized by similar CYP450 enzymes) stand a risk of having HDI, which can lead to clinical consequences. The TDI effects of these herbals is felt when patients chronically use them because sufficient intermediary metabolites may be generated and this can be deleterious to consumers concomitantly taking conventional medications, as stated earlier.
Single point screening for the effects of pre-incubation with NADPH was performed on CYP1A2, CYP2C9, and CYP2C19. Data showed significant effects of pre-incubation of Newbouldia laevis extracts on CYP1A2 enzyme activity but not CYP2C9 and CYP2C19 enzyme activities. Cassia abbreviata extracts, however, exhibited significant effects on CYP1A2 and CYP2C19 enzyme activities after pre-incubation with NADPH but not CYP2C9 enzyme activity. These results combined with the TDI potencies imply that these herbs' inhibitory effects may either be increased or reduced after pre-incubation. The weakening inhibition potency could be due to recovery of enzyme activity since the inhibition potency of the herbs weakens with time regardless of the concentration applied, whilst the increase in inhibitory potency could be due to the interaction of the herb with CYP450 enzymes during pre-incubation.
Following the inactivation of CYP1A2, CYP2C9, and CYP2C19, the inactivation parameters were determined. The kinetic parameters for the inactivation of CYP1A2 by Newbouldia laevis and Cassia abbreviata extracts were: K i ; 2.84 µg/mL and 4.86 µg/mL, K inact ; 0.024 min´1 and 0.033 min´1, respectively. Inactivation kinetic parameters of CYP2C9 were: K i ; 1.55 µg/mL and 5.98 µg/mL, K inact ; 0.027 min´1 and 0.067 min´1, respectively. Kinetics parameters of inactivation for CYP2C19 were K i ; 1.23 µg/mL and 1.58 µg/mL, K inact ; 0.031 min´1 and 0.029 min´1, respectively. Determination of inactivation parameters gives more information on the type of inactivation, thus allowing patients taking these herbs concomitantly with conventional medication to be aware of the implications.
Herbal medicines consist of multi-phytochemical constituents with different physicochemical properties. The risk of HDI of orally administered herbal remedies depends on the bioavailable fraction of the phyto-constituents [34], which may or may not permeate through membrane barriers. Due to the multi-phyto nature of herbals, their bio-availabilities are variable depending on the constituents [24]. Prediction of in vivo HDI using in vitro results [21,23,35,36] gives preliminary information that can caution patients to be wary of combination therapies of herbals with conventional medications and also for pharmaceutical companies and researchers to get baseline information guiding further investigation. For herbal extracts that are already in use, HDI studies allow assessment of risk, helping with the design of in vivo HDI studies and revision of product labels to highlight the risk of co-use of these herbs with conventional medication, as in the case of Saint John 1 s Wort and the protease inhibitor indinavir [37]. This study assessed the risk of HDI on CYP1A2, CYP2C9, and CY2C19 activities and predicted these herbals can cause HDIs in vivo to a significant extent, with the assumption that at least one phyto-constituent is absorbed to cause HDI.
Profiling of herbal extracts for the phyto-constituents is a critical step in elucidating the possible mechanisms for HDI potential. Based on the UPLC-MS data obtained, the crude extracts of Newbouldia laevis and Cassia abbreviata are composed of several different phenolic compounds. The UPLC-MS profiles indicated the presence of at least 10 polyphenolic candidates including catechins and potent antioxidant flavonoids, which likely give these herbs their therapeutic effects. Epidemiological studies of polyphenols suggest that consumption of polyphenol-rich herbs and beverages may contribute to the prevention of diseases including cancers, cardiovascular diseases, osteoporosis, neurodegenerative diseases, and diabetes [38,39]. However, some of these polyphenols are known to inhibit CYP450 [40][41][42][43]. It is therefore possible that the interactions observed in this study may be related to some of the constituents that are in these herbs.

Plant Material Extraction
The leaves of Newbouldia laevis (UCC/BS/689) and Cassia abbreviata (UB/B/151) were obtained with the assistance of traditional herbal practitioners and authenticated by a botanist from the University of Cape Coast herbarium. Voucher specimens were kept at the University of Cape Coast herbarium. The plant material was air-dried at room temperature and ground into a fine powder. Extraction was performed using water to mimic the indigenous mode of extraction. Powdered plant material (10 g) was added to 100 mL of distilled water, heated for one hour at 60˝C, and allowed to extract for 72 h with 24 h decanting of supernatant and refilling with the same volume of distilled water on solid residue. Supernatants were pooled, centrifuged (14,000ˆg, 10 min), and filtered using a filter paper (8 µm, Whatman International LTD, Maidstone, UK). Filtrates were freeze-dried using a Virtis sentry freeze dryer (the Virtis Company, NC, Gardiner, NY, USA) and the dried powder stored in airtight containers at -20˝C until subsequent experiments.

CYP450 Inhibition
The determination of reversible inhibition was carried out as reported in a previous study with slight modification [23]. An initial study of two-point screening indicated that 100 µg/mL exhibited the maximum inhibitory effect on CYP1A2, CYP2C9, and CY2C19. Reconstitution of extracts in subsequent experiments were performed with 100 µg/mL concentration as the highest concentration. Briefly, serial dilutions (1:3) of the medicinal plant extracts from a concentration of 100 µg/mL were pre-incubated with a mixture of CYP 450 BACULOSOME ® (CYP1A2, CYP2C9, CYP2C19) plus reagent and regeneration system (consisting of glucose-6-phosphate and glucose-6-phosphate dehydrogenase) in Vivid ® reaction buffer in black costar 96-well plate for 20 min at 37˝C. The reaction was initiated by the addition of reconstituted substrate as seen in Table 5 and NADP + in Vivid ® reaction buffer and incubated for a further 30 min at 37˝C. The reaction was stopped by the addition of ice-cold 20% Tris base/80% acetonitrile (ACN). Enzyme activity was monitored by measuring the formation of fluorescent metabolite at excitation/emission wavelength of 405/460 nm. Fluorescence was measured using a Varian Cary eclipse (SSN instruments, Set Point Technology, South Africa) with Advanced reads software. Conditions for the experiment are shown in Table 5. Suggested standard inhibitors for the CYP enzymes were supplied by the manufacturer.

Determination of Time-Dependent Inhibition (TDI) Potency
The procedure followed was similar to that used for the screening of CYP450 inhibition with a slight modification that included employing the curve shift method by Berry and Zhao [20]. Plant extracts with a starting concentration of 100 µg/mL were serially diluted (1:3) in black costar 96-well plates. The black costar plates were divided into two halves. One half contained a reaction mixture made of BACULOSOME ® , regeneration system, NADP + , and Vivid ® reaction buffer, Table 1 and the other half contained BACULOSOME ® , a regeneration system, and Vivid ® reaction buffer ( Table 5). The mixture was pre-incubated for 30 min at 37˝C. After the pre-incubation step, reconstituted substrate (Table 5) and NADP + in Vivid ® reaction buffer was added to both halves of the plate and incubated for a further 30 min at 37˝C. The reaction was terminated by adding ice-cold 20% Tris base/80% ACN. The activity of the enzymes was monitored by measuring the formation of fluorescent metabolite at the same excitation and emission wavelength as in the CYP450 inhibition determination. Single point NADPH screening with a single concentration of 50 µg/mL was performed as in the TDI determination to assess the significance of the effect of pre-incubating with NADPH at a single concentration.

Estimation of Kinetics of Inactivation
The method used is a slight modification of the CYP450 inhibition and TDI determinations according to the method of Krippendorff et al. [44]. Briefly, serial dilutions of the extracts were added to a mixture of regeneration system, NADP + , and Vivid ® reaction buffer and pre-warmed. After pre-warming, the reaction was started by adding a mixture of substrate and BACULOSOME ® in Vivid ® reaction buffer. Fluorescence was measured for each well as indicated previously. However, with this method, instead of using end point measurements as done previously, CYP activity was determined every 5 min up to 30 min after the start of the reaction using an excitation/emission wavelength of 405 nm/460 nm.

UPLS-MS Analysis and Relative Quantification of Phenolic Compounds
Phytofingerprinting was performed on the extracts using a Waters Acquity UPLC system (Waters Corporation, Milford, MA, USA) with an Acquity BEH C18 column (2.1 mmˆ100 mm, 1.7 µm particle size) incorporating a binary pump, vacuum degasser, autosampler, column oven, and Micromass Xevo tandem quadrupole mass spectrometric detector (QTOF xevo G2; Waters micromass, Manchester, UK) equipped with ESI probe. Gradient elution was performed at a flow rate of 0.1 mL/min throughout at injection volumes of 10 µL. Gradient parameters were adjusted by systematically changing the percentage organic modifier at initial conditions, and/or the isocratic hold period at initial conditions, and/or gradient steepness. Electrospray mass spectra data were recorded on a negative ionization mode for a mass range m/z 100 to m/z 1500 at a collision energy of 50 V. Relative quantification of some constituents of the extracts was performed as stated previously [23]. Standards of epicatechin, catechin, chlorogenic acid, caffeic acid, and p-coumaric acid were used to generate calibration curves and quantities derived from the calibration curves. Calibration curve ranges for each reference compound ranged from 0.1 to 100 µg/mL.

IC 50 Determination
Data generated were exported into an Excel spreadsheet and the amount of metabolite formed at the various concentrations relative to the control (residual activity) was calculated using the equation below: % Residual activity " test´blank control´blankˆ1 00 (1) Percentage residual activity was plotted against the log-transformed concentrations of the extracts and positive inhibitors. Non-linear regression analysis was used for the sigmoid curve fitting and IC 50 values calculated using GraphPad ® Prism version 5.0 (GraphPad ® Software Inc., San Diego, CA, USA).

Inactivation Kinetics
The natural logarithm of the percentage remaining activity was plotted against the pre-incubation times at each extract concentration to obtain K obs (slope) as employed by Awortwe et al. [36]. K obs depicts the rate constant describing the inactivation at each inhibitor (extract) concentration. The K i and K inact values were then determined by a double reciprocal plot of the K obs values and [I] using the equation below: Kobs " KinactˆrIs Ki`rIs (2) where K inact is the maximal rate of inactivation; K i is the inhibitor concentration required for half-maximal inactivation; and [I] is the pre-incubation concentration of inhibitor (extract).

Statistical Analysis
All values are expressed as Mean˘SEM. Comparisons for significance were performed using unpaired t-test with p < 0.05 considered statistically significant.

Conclusions
Most patients in developed and developing countries consume herbal medicines for reasons already elucidated in addition to conventional medicines. This study has evaluated the inhibitory potencies of Newbouldia laevis and Cassia abbreviata on CYP1A2, CYP2C9, and CYP2C19 activities. It is observed that these two herbs caused differential inhibition of the CYPs and are likely to cause HDI. This study gives information on two herbs that have been used for decades and this may be used to caution patients to desist from combining them with other conventional medications.