Abstract
Commercial formulations of 29 commonly used herbal supplements (HSs) and grapefruit juice were evaluated for drug interaction potential via quantification of their CYP3A inhibitory potential in two in vitro experimental models of human small intestine, cryopreserved human intestinal mucosa (CHIM), and cryopreserved human enterocytes (CHEs). Two CYP3A substrates were used—in the studies with CHIM, CYP3A activity was quantified via liquid chromatography tandem mass spectrometry quantification of midazolam 1’-hydroxylation, whereas in CHE, luciferin-IPA metabolism to luciferin was quantified by luminescence. Upon treatment of CHIM with the estimated lumen concentration of the HS upon each oral administration (manufacturers’ recommended dosage dissolved in 200 ml of culture medium), >80% CYP3A inhibition was observed for green tea extract, St. John’s wort, valerian root, horehound, and grapefruit juice. Less than 50% inhibition was observed for fenugreek, aloe vera, guarana, soy isoflavone, maca, echinacea, spirulina, evening primrose, milk thistle, cranberry, red yeast rice, rhodiola, ginkgo biloba, turmeric, curcumin, white kidney bean, garlic, cinnamon, saw palmetto berries, panax ginseng, black elderberry, wheat grass juice, flaxseed oil, black cohosh, and ginger root. The results were confirmed in a a dose-response study with HSs obtained from three suppliers for the four inhibitory HSs (green tea extract, horehound, St. John’s wort, valerian root) and three representative noninhibitory HSs (black cohosh, black elderberry, echinacea). Similar results were obtained with the inhibitory HSs in CHE. The results illustrate that CHIM and CHE represent physiologically relevant in vitro experimental models for the evaluation of drug interaction potential of herbal supplements. Based on the results, green tea extract, horehound, St. John’s wort, and valerian root may cause drug interactions with orally administered drugs that are CYP3A substrates, as was observed for grapefruit juice.
SIGNIFICANCE STATEMENT In vitro evaluation of 29 popular herbal supplements in cryopreserved human intestinal mucosa identified green tea extract, horehound, St. John’s wort, and valerian root to have CYP3A inhibitory potential similar to that for grapefruit juice, suggesting their potential to have clinically significant pharmacokinetic interaction with orally administered drugs that are CYP3A substrates. The results suggest that cryopreserved human intestinal mucosa can be used for in vitro evaluation of drug interactions involving enteric drug metabolism.
Introduction
Herbal medicines and supplements have been used for their putative health benefits and in the treatment of illnesses throughout human history. For the past decade, the use of herbal supplements (HSs) has gained wide acceptance by the general population for the purpose of health enhancement. The retail sales of herbal supplements in the United States have been reported to have increased from $4.8 billion in 2008 to $8.8 billion in 2018 (http://cms.herbalgram.org/herbalgram/issue123/files/HG123-HMR.pdf), with approximately 18–20% of the population acknowledging using herbal products (Bent, 2008; Peregoy et al., 2014). Potential HS-drug interactions are therefore a legitimate concern (Archer et al., 2014; Levy et al., 2017; Ziemann et al., 2019), especially with orally administered drugs. Because oral drug bioavailability is determined by a combination of the fraction absorbed and the fraction escaping metabolism by the intestinal mucosal epithelium, coadministration of drugs with HSs that are inhibitors and inducers of enteric drug metabolism may lead to significant changes in the drug concentration in the systemic circulation.
Because liver is the major organ for drug metabolism, evaluation of drug-drug interactions is traditionally focused on hepatic events, including metabolism- and transporter-mediated drug uptake and efflux. However, there is evidence that enteric drug metabolism can play a significant role in drug interactions involving orally administered drugs. This is illustrated clearly with clinically observed cases of grapefruit juice (GFJ)-drug interactions. Coadministration with GFJ is known to increase the Cmax and plasma AUC of drugs that are CYP3A substrates, including terfenadine, clopidogrel, saquinavir, cyclosporin, midazolam, nimodipine, triazolam, and verapamil (Schipperijn, 1997; van den Anker and de Wildt, 1997; Bailey et al., 1998b; Fuhr et al., 1998; Holmberg et al., 2014). In a clinical study, pretreatment of patients with GFJ was found to significantly increase the plasma Cmax and AUC of orally administered but not intravenously administered midazolam, demonstrating that GFJ inhibited enteric but not hepatic CYP3A, therefore establishing that pharmacokinetic drug interactions can occur via the inhibition of enteric drug metabolism (Kupferschmidt et al., 1995). Both clinical and in vitro studies show that the mechanism of GFJ-drug interactions is a result of inhibition of CYP3A-mediated metabolism, with furanocoumarins as the major inhibitors (Bailey et al., 1998a,b, 2003; Paine et al., 2006).
The findings with GFJ illustrate that inhibition of enteric drug metabolism by a perpetrator may lead to clinically significant drug interactions. In our laboratory, we have embarked upon a research program to identify HSs with potent inhibitory potential for enteric drug-metabolizing enzymes. We report here the applications of two in vitro enteric experimental models developed in our laboratory, cryopreserved human intestinal mucosa (CHIM) (Li et al., 2018) and cryopreserved human enterocytes (CHEs) (Ho et al., 2017), in the evaluation of enteric drug interaction potential of commercial formulations of HSs toward CYP3A, one of the most important enteric drug-metabolizing enzymes. Twenty-nine HSs (Table 1) that have been reported to be among the top 40–selling HSs in the United States (Table 2) were evaluated for their inhibitory potential toward enteric CYP3A activity in CHIM and CHEs. GFJ was also evaluated to demonstrate the ability of our in vitro enteric models to identify a complex natural product with proven clinically significant CYP3A inhibitory properties.
The common names, scientific names, brand names, and amounts per serving of the herbal supplements evaluated in this study
Each supplement was procured commercially and used directly by dissolution in culture medium without further extraction. The amount per serving represents the weight of herb extract in each recommended administration unit (e.g., per capsule) as indicated on the product label.
Ranking of the herbal supplements evaluated in this study based on US sales in years 2016, 2017, and 2018 fo
Materials and Methods
HSs.
The herbal supplements GFJ, orange juice, and cranberry juice used in the study were purchased commercially. The brands, commercial names, and active ingredients indicated in the product labels and the recommended dosages indicated in the product labels are shown in Table 1.
Isolation and Cryopreservation of CHEs and CHIM.
CHEs and CHIM were isolated and cryopreserved from human small intestines obtained from the International Institute for the Advancement of Medicine (Exton, PA) and stored in liquid nitrogen as previous described (Ho et al., 2017; Li et al., 2018). CHEs pooled from 10 individual donors (PCHE3107) and CHIM pooled from from five individual donors (PCHIM6031) were used in this study. The demographics for the donors of the human small intestines are shown in Table 3 (CHIM) and Table 4 (CHEs).
Demographics of the donors of the CHIM used in the preparation of the pooled donor CHIM (PCHIM6031) used in this study
Demographics of the donors of the enterocytes used in the preparation of the pooled donor CHEs (PHE3107) used in this study
Recovery from Cryopreservation.
Vials of PCHE3107 and PCHIM6031 (In Vitro ADMET Laboratories, Columbia, MD) were retrieved from liquid nitrogen storage, thawed in a 37°C water bath, and recovered in cryopreserved enterocyte recovery medium [In Vitro ADMET Laboratories (IVAL)]. The recovered CHE and CHIM pellets were resuspended in a protein-free cell culture medium, hepatocyte/enterocyte incubation medium (HQM) (IVAL). The CHE cell suspension was evaluated for cell concentration and viability via trypan blue exclusion, and medium was added to adjust to a cell concentration of 2 million viable enterocytes per milliliter (2× the final cell concentration). The CHIM pellet was resuspended with medium to a protein concentration of 1 mg/ml (2× the final protein concentration). Protein concentration was previously determined in CHIM using the Pierce BCA protein assay kit (Thermo Scientific, Rockford, IL).
Herbal Supplement–Drug Interaction Studies.
The HSs used were commercially formulated capsules or extracts. The daily recommended oral dose according to the product label of each HS was dissolved in 50 ml (4× the expected intestinal concentration based on the generally accepted intestinal fluid volume of approximately 200 ml. We inadvertently used a 200-ml volume in lieu of the generally accepted volume of 250 ml) of a protein-free medium, HQM (IVAL). The HS solutions were filtered to remove debris with a Nalgene 50-ml 0.2-µM aPES membrane (Thermo Scientific), and pH was adjusted to 7.0–7.2. The resulting filtrates were further diluted by 1:2 serial dilutions with HQM six times to yield dosing solutions representing 100%, 50%, 25%, 12.5%, 6.25%, 3.125%, and 1.56% after a 1:4 dilution when added to the reaction mixture. GFJ (and other fruit juices evaluated with CHE) was diluted similarly, with the undiluted juice serving as the 4× concentration based on the consumption volume of 50 ml. For the drug interaction studies, 25 μl of the 4× herbal supplements were added in triplicate to 96-well plates, and this was followed by 50 µl of CHEs or CHIM. These were preincubated at 37°C for 15 minutes, and this was followed by the addition of 25 µl of midazolam (Sigma-Aldrich Inc., St. Louis, MI) at 40 µM (4× the final concentration of 10 µM) for studies with CHIM and 25 μl of luciferin-IPA (Promega Inc., Madison, WI) at 12 µM (4× the final concentration of 3 µM) for studies with CHE to each well. The final concentrations of midazolam and luciferin-IPA are within 2× published Km values for CYP3A. The plates were then incubated at 37°C for 30 minutes. For studies with CHIM, at the end of the incubation period, 200 µl of ice-cold acetonitrile containing 250 nM of the internal standard, tolbutamide (Sigma-Aldrich), was added to each well followed by liquid chromatography tandem mass spectrometry quantification of 1′-hydroxy-midazolam as previously described (Li et al., 2018). For studies with CHE, 40 μl per well of the reaction mixture after incubation was transferred to two opaque white plates for the quantification of luciferin-IPA metabolism (plate 1) and cellular ATP contents (plate 2). Luciferin Detection Reagent (Promega) and ATP Detection Reagent (Perkin Elmer, Shelton, CT) at volumes of 40 µl per well were added to plate 1 and plate 2, respectively, for the quantification of luciferin and ATP, respectively. Luminescence was measured using a plate reader (PerkinElmer Victor3V Multichannel Plate Reader) as previously described (Li, 2009).
Data Analysis
Relative Activity.
CYP3A activity is expressed as relative CYP3A activity using the following equations:
in which CYP3A activity in CHIM was measured as picomoles per minute per milligram protein of 1′-OH midazolam formation from midazolam, and that in CHE was picomoles per minute per million cells of luciferin formation from luciferin-IPA.
Results with CHE were further normalized by the number of viable cells after HS incubation using the following equations:
IC50 Determination.
IC50 for CYP3A inhibition was determined using the Graphpad Prism 8.0 software (www.graphpad.com) via linear regression (log inhibitor concentration vs. response, variable slope, four parameters).
Results
CYP3A Inhibition Studies in CHIM
GFJ Inhibition of Enteric CYP3A Activity.
GFJ was used as a “positive control” in the investigation to evaluate whether its known CYP3A inhibitory effect could be observed in CHIM. GFJ at 100% oral concentration yielded a relative CYP3A activity of 4.69% (Fig. 1). To demonstrate dose-response relationship and to evaluate possible intraexperimental variation, GFJ was evaluated at concentrations of 1.56%, 3.13%, 6.25%, 12.5%, 25%, 50%, and 100% in four different plates in the same experiment. Dose-dependent inhibition of CYP3A activity was observed (Fig. 2) with IC50 values for GFJ in independent incubations ranging from 4.12% to 8.53% (Table 6).
Relative CYP3A4 activity (activity as percent of that for medium control) in CHIM in the presence of the recommended dosages of commercial formulations of 29 herbal supplements and grapefruit juice was plotted vs. log GFJ concentrations. The brands of the HSs are shown in brackets.
Effects of GFJ on CYP3A activity (midazolam 1′-hydroxylation) in CHIM. Relative CYP3A4 activity (activity as percent of that for medium control) was plotted vs. log GFJ concentrations. Results of four replicate incubations were shown.
IC50 and the 95% likelihood values for CYP3A inhibition in CHIM
Three different brands of noninhibitory (black cohosh, elderberry, echinacea) and inhibitory [green tea, horehound, St. John’s wort (SJW), valerian root] HSs and GFJ were evaluated for their inhibitory potential for CYP3A activity in CHIM. The GFJ results were from four replicate plates. The results were derived from nonlinear regression analysis of plots of CYP3A activity (midazolam 1′-hydroxylation) vs. log conc. of each substance using the Graphpad Prism 8.0 software. The different brands are shown in brackets.
IC50 values for the CYP3A inhibitory potential of selected inhibitory herbal supplements [horehound, St. John’s wort (SJW), and fruit juices], GFJ, cranberry juice (CBJ), and orange juice (OJ) in cryopreserved human enterocytes
The IC50 values are expressed as percent of the recommended oral dosage. The manufacturer of each of the herbal supplements is shown in parentheses. The results were derived from nonlinear regression analysis of plots of CYP3A activity (luciferin-IPA metabolism) vs. log conc. of each substance using the Graphpad Prism 8.0 software. The different brands are shown in parentheses.
CYP3A Inhibition by Commercial HS Formulations.
1. Treatment of CHIM with recommended oral dosages.
Relative CYP3A activity in CHIM treated with HSs at the recommended dosages are shown in Fig. 1. The HSs were classified based on their CYP3A inhibitory potential as follows:
a. Relative CYP3A activity <50% (>50% inhibition): green tea, St. John’s wort, valerian root, and horehound.
b. Relative CYP3A activity >50% (<50% inhibition): fenugreek, aloe vera, guarana, soy isoflavone, maca, echinacea, spirulina, evening primrose, milk thistle, cranberry, red yeast rice, rhodiola, ginkgo biloba, turmeric curcumin, white kidney bean, garlic, cinnamon, saw palmetto berries, panax ginseng, black elderberry, wheat grass juice, flaxseed oil, black cohosh, and ginger root.
2. Dose-response studies with HSs from three different manufacturers.
To evaluate whether the inhibitory effects were associated with the designated herb for the supplements and to confirm the results with the single concentration study, a dose-response study was performed with HSs obtained from three different brands of the four inhibitory HSs (green tea extract, horehound, St. John’s wort, valerian root) and three representative noninhibitory HSs (black cohosh, elderberry, echinacea). Dose-dependent inhibition was observed for the four inhibitory HSs (Fig. 3), with IC50 values for the HSs from the three suppliers expressed as percent of the daily recommended dosage, ranging from 27.5% to 49.5% for green tea, 16.1%–70.7% for horehound, 54.8%–86.3% for St. John’s wort, and 4.8%–71.3% for valerian root. The three noninhibitory HSs, except black cohosh from one of the brands, did not exhibit dose-dependent inhibition of CYP3A activity (Fig. 4; Table 5).
Dose-dependent inhibition of CYP3A activity (midazolam 1′-hydroxylation) by three brands each of the inhibitory HSs [green tea, horehound, St. John’s wort (SJW), and valerian root] in CHIM. Relative CYP3A4 activity (activity as percent of that for medium control) was plotted vs. log HS concentrations. The brands of the HSs are shown in brackets.
Relative CYP3A activity (midazolam 1′-hydroxylation) in CHIM upon treatment with three brands each of the noninhibitory HSs (black cohosh, elderberry, and echinacea). Relative CYP3A4 activity (activity as percent of that for medium control) was plotted vs. log HS concentrations. The brands of the HSs are shown in brackets.
Evaluation of Inhibitory Potential of Selected HSs on CYP3A Activity in CHEs.
To confirm the findings in CHIM, selected HSs found in CHIM to be CYP3A inhibitors (horehound, St. John’s wort, and valerian root), noninhibitors (echinacea, fenugreek, ginkgo biloba, guarana, red yeast rice), and three fruit juices (GFJ, cranberry juice, orange juice) were evaluated in CHE. Dose-dependent CYP3A inhibition was observed for horehound, St. John’s wort, valerian root, and grapefruit juice but not for orange juice, echinacea, fenugreek, ginkgo biloba, guarana, and red yeast rice (Fig. 5). The IC50 values expressed as percent of daily recommended dosage are 3.97% (GFJ), 52.24% (cranberry juice), 45.08% (horehound), 38.67% (St. John’s wort), and 14.26% (valerian root) (Table 6).
Relative CYP3A activity (luciferin-IPA metabolism to luciferin) in CHEs upon treatment with GFJ, cranberry juice (CBJ), and orange juice (OJ) (top left); the inhibitory HSs horehound, St. John’s wort (SJW), and valerian root (top right); and the noninhibitory HSs echinacea, fenugreek, ginko biloba, guarana, and red yeast rice (bottom). Relative CYP3A4 activity (activity as percent of that for medium control) was plotted vs. log HS concentrations. The brands of the HSs are shown in brackets.
Discussion
Drug interaction potential of HSs needs to be accurately defined because of the high probability of their coadministration with prescribed or over-the-counter drugs. HS-drug interaction can occur as a result of the inhibitory and inducing activities for drug-metabolizing enzymes and transporters, as observed with small-molecule drugs, with these events occurring both in the small intestine and in the liver. For inhibitory drug interactions, in which the perpetrator inhibits the metabolic clearance of a victim drug, enteric events may be particularly important. A major reason is that the concentrations of orally administered drugs in the intestinal lumen can be substantially higher than those in the plasma. For instance, oral administration of 500 mg of a drug with a molecular weight of 500 grams per mole will have a concentration of 4.16 mM after dissolution in the intestinal fluid of 240 ml. Also, as described earlier, enteric, not hepatic, events are responsible for clinically significant drug interactions with GFJ.
We report here the CYP3A inhibitory potential of 29 commercial formulations of HSs and GFJ in two in vitro enteric systems, CHIM and CHE. GFJ yielded the expected potent and dose-dependent CYP3A inhibition in both CHIM and CHEs, thereby substantiating that they represent appropriate in vitro experimental models for the evaluation of enteric CYP3A inhibition. Our results led to the identification of green tea extract, St. John’s wort, valerian root, and horehound as potent CYP3A inhibitors. It is notable that each HS was evaluated with the actual commercial product, thereby modeling a consumer’s enteric exposure upon its ingestion. However, because only the soluble portion of each commercial formulation was evaluated, the effects of components that are soluble in vivo but not in vitro may not be reflected. Our intention was to evaluate whether the commercial products that are readily available in the marketplace would inhibit enteric drug metabolism via exposure of the enterocytes to all the soluble ingredients in the commercial formulations. The correlation of our results with those reported by others (see below) for the herbal supplements evaluated and the finding of similar properties for supplements of different brands provide support for the herbal identity of the supplements evaluated.
Our findings with green tea are consistent with those previously reported. Green tea extracts and commercial formulations have been shown to inhibit CYP3A activity of human liver and intestinal microsomes (Wanwimolruk et al., 2009; Misaka et al., 2013a; Satoh et al., 2016). Pharmacokinetic drug interactions related to CYP3A inhibitory effects of green tea extract include increases in plasma AUC and/or Cmax of drugs that are CYP3A substrates, such as midazolam (Nishikawa et al., 2004; Misaka et al., 2013b), simvastatin (Misaka et al., 2013b), sildenafil (Werba, 2018 ), and tacrolimus (Vischini, 2011). Besides CYP3A inhibition, green tea extracts and the associated catechins have been found to be associated with inhibitory activities for UDP-glucuronosyltransferase (Mohamed and Frye, 2011; Tian et al., 2018) and efflux transporters (Zhou et al., 2004).
The pharmacokinetic drug interaction potential with St. John’s wort is mainly attributed to the CYP3A induction activity upon prolonged usage, leading to reduced plasma concentration and thereby compromised efficacy of the coadministered CYP3A substrate drugs (Markowitz et al., 2003; Adiwidjaja et al., 2019). Our findings on CYP3A inhibition by St. John’s wort are consistent with the reports by others that its major active constituent, hyperforin, has been reported to have both CYP3A-inducing (Hellum et al., 2007) and inhibitory (Obach, 2000; Komoroski et al., 2004) effects. Our results provide additional experimental evidence substantiating a previous report that St. John’s wort may have short-term inhibitory and long-term inductive pharmacokinetic drug interactions (Xie and Kim, 2005).
Our finding that valerian is a CYP3A inhibitor is consistent with a clinical study showing that the plasma Cmax for orally administered alprazolam, a CYP3A substrate, was significantly increased in healthy volunteers taking 1000 mg/day of valerian extract for 14 days (Donovan et al., 2004). Valerian extracts have been found to inhibit CYP3A (Lefebvre et al., 2004), CYP2C8 (Albassam et al., 2015), and UDP-glucuronosyltransferase (Alkharfy and Frye, 2007; Mohamed et al., 2010) activities in human liver microsomes.
We are surprised that there are no previous reports of the CYP3A inhibitory potential of horehound, which caused dose-dependent CYP3A inhibition in both CHIM and CHEs. Most reported studies on herb-drug interactions, both in vitro and in vivo, have not included this herbal supplement, which is a widely used HS (Bent, 2008; Peregoy et al., 2014) that ranked first in total sales in 2016, 2017, and 2018 (Table 2). Our finding that horehound was an inhibitor of CYP3A may explain the finding that aqueous extract of horehound attenuates the hepatotoxicity of cyclophosphamide (Ettaya et al., 2016), a protoxicant that is activated by P450 isoforms, especially CYP2B6 and CYP3A4, to hepatotoxic metabolites (Chang et al., 1997). In our laboratory, we plan to further evaluate horehound as well as its bioactive compounds to elucidate the physiologic relevance of the observed CYP3A inhibitory effects.
The observed CYP3A inhibitory potential of GFJ, green tea, St. John’s wort, and valerian root in CHIM and CHEs is therefore consistent with previous in vitro and in vivo findings, suggesting that the two intact cell models of the human small intestine represent physiologically relevant intact cell system for the identification of herbal supplements with drug interaction potential and for mechanistic evaluation of enteric herb-drug and drug-drug interactions. An accurate assessment of the drug interaction potential of herbal supplements requires multidisciplinary research including identification of the perpetrators of drug interactions, evaluation of their effects on enteric metabolism, determination of the amount of the perpetrators absorbed into the portal and systemic circulation, evaluation of interaction of the perpetrators in the plasma with hepatic and extrahepatic metabolism, and determination of the effects on victim drugs based on their fractions metabolized (fm) for the affected enteric, hepatic, and extrahepatic pathways. CHIM and CHEs thereby represent in vitro tools to provide information on the enteric events contributing to the drug interaction potential of herbal supplements. Although this report focuses on CYP3A, which is one of the most important enteric drug-metabolizing enzymes, a comprehensive evaluation of enteric drug interaction potential of HSs needs to include all drug-metabolizing enzymes as well as uptake and efflux transporters that are active in the small intestine.
Authorship Contributions
Participated in research design: Loretz, Li.
Conducted experiments: Loretz, Ho, Alam, Mitchell.
Performed data analysis: Loretz, Ho, Alam, Mitchell, Li.
Wrote or contributed to the writing of the manuscript: Loretz, Ho, Alam, Li.
Footnotes
- Received March 25, 2020.
- Accepted July 10, 2020.
No external funding was received for the research described in this manuscript.
Abbreviations
- AUC
- area under the curve
- CHE
- cryopreserved human enterocyte
- CHIM
- cryopreserved human intestinal mucosa
- GFJ
- grapefruit juice
- HQM
- hepatocyte/enterocyte incubation medium
- HS
- herbal supplement
- IVAL
- In Vitro ADMET Laboratories
- Copyright © 2020 by The American Society for Pharmacology and Experimental Therapeutics