Hepatotoxicity of Herbal Supplements Mediated by Modulation of Cytochrome P450

Herbal supplements are a significant source of drug-drug interactions (DDIs), herb-drug interactions, and hepatotoxicity. Cytochrome P450 (CYP450) enzymes metabolize a large number of FDA-approved pharmaceuticals and herbal supplements. This metabolism of pharmaceuticals and supplements can be augmented by concomitant use of either pharmaceuticals or supplements. The xenobiotic receptors constitutive androstane receptor (CAR) and the pregnane X receptor (PXR) can respond to xenobiotics by increasing the expression of a large number of genes that are involved in the metabolism of xenobiotics, including CYP450s. Conversely, but not exclusively, many xenobiotics can inhibit the activity of CYP450s. Induction of the expression or inhibition of the activity of CYP450s can result in DDIs and toxicity. Currently, the United States (US) Food and Drug Administration does not require the investigation of the interactions of herbal supplements and CYP450s. This review provides a summary of herbal supplements that inhibit CYP450s, induce the expression of CYP450s, and/or whose toxicity is mediated by CYP450s.


Introduction
The aim of this review is to highlight the effects of herbal supplement use on xenobiotic metabolism and the development of hepatotoxicity or modulation of therapeutic efficacy, either alone or with concurrent use of other herbal supplements and Food and Drug Administration (FDA)-approved pharmaceuticals. Cytochrome P450 (CYP450) enzymes metabolize a large number of xenobiotics, which may lead to the formation of hepatotoxic compounds. Fortunately, pharmaceuticals are evaluated for the potential to inhibit or induce CYP450 enzymes before being marketed in the United States (US) [1]. However, regulations for herbal supplements in the United States do not require surveillance or the reporting of adverse events by the manufacturer to the FDA [2]. Thus, the data concerning the hepatotoxicity of herbal supplements are derived from case reports and series, retrospective databases, and progressive registries, such as the US Drug-Induced Liver Injury (DILI) Network and the Spanish DILI Registry [3,4]. Therefore, herbal supplements that inhibit CYP450 enzymes are problematic, because in addition to affecting the therapeutic efficacy of drugs that require bioactivation for effect, these interactions may inhibit the metabolism of toxic parent compounds to less toxic daughter compounds ( Figure 1) [5]. Schematic of the general mechanism; (B) a specific example discussed in the text [5].
The expression of CYP450 enzymes is induced by nuclear receptors that respond to xenobiotics. Many CYP450 enzymes are transcriptionally regulated by the xenobiotic receptors pregnane X receptor (PXR, also known as NR1I2, SXR, or PAR) and constitutive androstane receptor (CAR; NR1I3) [6][7][8][9]. We will discuss the involvement of PXR and CAR in herbal supplement hepatotoxicity later in this review. Many herbal supplements can activate these receptors leading to an increased expression of CYP450 enzymes and can affect the metabolism of other xenobiotics. In many instances of this can lead to hepatotoxicity, as well as deactivating active drugs and decreasing the therapeutic effect. The induction of CYP450 enzymes may result in hepatotoxicty by increasing the metabolism of less toxic parent compounds to much more toxic daughter compounds ( Figure 2) [10][11][12][13][14].  (A) Schematic of the general mechanism; (B) a specific example discussed in the text [5].
The expression of CYP450 enzymes is induced by nuclear receptors that respond to xenobiotics. Many CYP450 enzymes are transcriptionally regulated by the xenobiotic receptors pregnane X receptor (PXR, also known as NR1I2, SXR, or PAR) and constitutive androstane receptor (CAR; NR1I3) [6][7][8][9]. We will discuss the involvement of PXR and CAR in herbal supplement hepatotoxicity later in this review. Many herbal supplements can activate these receptors leading to an increased expression of CYP450 enzymes and can affect the metabolism of other xenobiotics. In many instances of this can lead to hepatotoxicity, as well as deactivating active drugs and decreasing the therapeutic effect. The induction of CYP450 enzymes may result in hepatotoxicty by increasing the metabolism of less toxic parent compounds to much more toxic daughter compounds ( Figure 2) [10][11][12][13][14]. Schematic of the general mechanism; (B) a specific example discussed in the text [5].
The expression of CYP450 enzymes is induced by nuclear receptors that respond to xenobiotics. Many CYP450 enzymes are transcriptionally regulated by the xenobiotic receptors pregnane X receptor (PXR, also known as NR1I2, SXR, or PAR) and constitutive androstane receptor (CAR; NR1I3) [6][7][8][9]. We will discuss the involvement of PXR and CAR in herbal supplement hepatotoxicity later in this review. Many herbal supplements can activate these receptors leading to an increased expression of CYP450 enzymes and can affect the metabolism of other xenobiotics. In many instances of this can lead to hepatotoxicity, as well as deactivating active drugs and decreasing the therapeutic effect. The induction of CYP450 enzymes may result in hepatotoxicty by increasing the metabolism of less toxic parent compounds to much more toxic daughter compounds ( Figure 2) [10][11][12][13][14].  Concurrent use of pharmaceuticals with herbal supplements, as well as the use of multiple herbal supplements, presents additional problems. Many patients do not report herbal supplement use to

Literature Search Methodology
To develop this review concerning the association of herbal supplement use with P450 modulation and hepatotoxicity, we used the search terms "herbal supplement" or "herb" combined with "hepatotoxicity" or "liver injury" to search the PubMed database. From this initial search, we selected compounds associated with hepatotoxicity in an in vivo model. We then searched for each of these supplements with "P450", "CYP450", or "cytochrome P450". Finally, we searched for "herbal supplement" or "herb" combined with "P450", "CYP450", or "cytochrome P450". Care is taken to include reports where the effect (P450 modulation or hepatotoxicity) is attributable to a single causative compound or when an extract of defined herbal ingredients is provided.

Herbal Supplements with Potential P450-Associated Hepatotoxicity
The most commonly used herbal supplements in the United States are echinacea, garlic, ginko biloba, saw palmetto, ginseng, grape seed extract, green tea, and St. John's wort [25]. This review will discuss herbal supplements that inhibit cytochrome P450 enzymes ( Figure 1) or induce P450 expression (Figure 2), and those whose toxicity is mediated by the P450 system ( Figure 3). Examples of these herbal supplements are discussed in the proceeding sections and are summarized in Tables 1-3, respective to figure numbering. These interactions can lead to HILI with the use of herbal supplements alone or when used in combination with traditional FDA-approved pharmaceuticals.

Literature Search Methodology
To develop this review concerning the association of herbal supplement use with P450 modulation and hepatotoxicity, we used the search terms "herbal supplement" or "herb" combined with "hepatotoxicity" or "liver injury" to search the PubMed database. From this initial search, we selected compounds associated with hepatotoxicity in an in vivo model. We then searched for each of these supplements with "P450", "CYP450", or "cytochrome P450". Finally, we searched for "herbal supplement" or "herb" combined with "P450", "CYP450", or "cytochrome P450". Care is taken to include reports where the effect (P450 modulation or hepatotoxicity) is attributable to a single causative compound or when an extract of defined herbal ingredients is provided.

Herbal Supplements with Potential P450-Associated Hepatotoxicity
The most commonly used herbal supplements in the United States are echinacea, garlic, ginko biloba, saw palmetto, ginseng, grape seed extract, green tea, and St. John's wort [25]. This review will discuss herbal supplements that inhibit cytochrome P450 enzymes ( Figure 1) or induce P450 expression (Figure 2), and those whose toxicity is mediated by the P450 system ( Figure 3). Examples of these herbal supplements are discussed in the proceeding sections and are summarized in Tables 1-3, respective to figure numbering. These interactions can lead to HILI with the use of herbal supplements alone or when used in combination with traditional FDA-approved pharmaceuticals.

Herbal Supplement Inhibition of P450s
Cytochrome P450 enzymes, predominantly hydroxylate xenobiotics and endobiotics, enable conjugation to additional chemical moieties to facilitate elimination from the body [26]. In the case of hepatotoxic compounds, these enzymes decrease the exposure of the compound to the liver ( Figure 1). Therefore, when patients are administered multiple xenobiotics, care should be taken that one compound does not increase the toxicity of another by inhibiting the metabolism of a toxic compound to its less-toxic daughter compounds. The ability of pharmaceuticals to inhibit P450 enzymes is evaluated during drug development and is usually well characterized before reaching market and general patient populations [1]. However, in the United States, herbal supplements are not required to be evaluated for this effect before reaching market. This would be problematic even if only the activation of inactive pro-drugs was prevented and the therapeutic efficacy was decreased. Inhibition of CYP450 enzymes may prevent the activation required for efficacious treatment by inactive prodrugs (e.g., tamoxifen) [27]. However, this is very concerning because of the potential for increasing hepatic and systemic exposure to toxic compounds, which would otherwise be metabolized by the P450 system. For example, catechins in green tea, terpenes in black cohosh and cranberry, geniposide and genipin in Gardenia, fucomarins in grapefruit juice, and Echinacea extract may inhibit the activity or decrease the expression of cytochrome P450 enyzmes (Table 1). Inhibition of activity in rat liver microsomes [31]. 3a

3A4
Root Extract (pill) Inhibition of activity as reflected by decreased caffeine metabolism in humans [41,42].

Sweet pepper (Capsicum annuum)
Dried and re-suspended in DMSO Inhibited activity of purified enzymes [46]. Herbs, compounds, or preparations marked with an asterisk have been associated with alteration of P450 metabolism in human subjects or have been associated with hepatotoxicity and may be clinically significant.

Green Tea
Green tea is traditionally made in China from the leaves of Camellia sinensis, and it is consumed to treat cancer, cardiovascular disease, dyslipidemia, inflammation, and weight loss [56][57][58][59][60]. Green tea use has been associated with hepatotoxicity at higher doses [61,62]. The hepatotoxicity of green tea in humans has been described as exhibiting a hepatocellular pattern of toxicity, and was evaluated by using the Roussel Uclaf Causality Assessment Method (RUCAM) causality assessment scale [62]. Additionally, using green tea in combination with other supplements was associated with liver injury that was shorter-onset and more-serious than that observed when patients were taking green tea alone [62]. This more-serious toxicity could be the result of interactions between the green tea and other components of the preparations. Whole green tea extract and the catechin (−)-epigallocatechin-3-gallate administered in a purified form inhibit the activity of multiple cytochrome P450 enzymes, including CYP2B6, CYP2C8, CYP2C19, CYP2D6, and CYP3A, in human liver and intestinal microsomes [30]. In rats that were administered commercially available green tea, the activities of hepatic microsomal cytochrome P450s were decreased, including those of CYP2C, CYP2E1, and CYP3A [63]. (−)-Epigallocatechin-3-gallate administered at non-lethal doses to mice decreased the levels of superoxide dismutase, catalase, and glutathione peroxidase. In mice, lethal hepatotoxicity was observed at higher doses [64]. Toxicity attributed to green tea extract has also been reported in rats [65]. Green tea extract administered to rats affected the pharmacokinetics of simvastatin and inhibited the hydroxylation of midazolam by CYP3a in liver microsomes [31]. Green tea extract and (−)-epigallocatechin-3-gallate pre-treatment reduced the area under the time-plasma concentration curve (AUC) and maximum plasma concentration (Cmax) of nadolol-a β-blocker that is not metabolized by cytochrome P450 enzymes but is reported to be a substrate for several drug transporters-in rats [5]. Thus, the consumption of green tea may inhibit the intestinal absorption of pharmaceutics taken concurrently. Additionally, green tea extract and a component of this extract, (−)-epigallocatechin-3-gallate, inhibit multiple P450 enzymes, and this inhibition may contribute to the toxicity that is associated with green tea use. Please see an earlier review concerning the hepatotoxicity of green tea for further information [66].

Black Cohosh
Black cohosh (Actae/Cimicfuga racemose) is native to the eastern region of North America. The roots and rhizomes of black cohosh are used as a hormone replacement and as an anti-inflammatory [67,68]. Hepatotoxicity has been reported in patients taking black cohosh [69,70]. Black cohosh has been reported as a probable causation of liver injury [69][70][71]. Although, a prior review discounted previous separate reports of hepatotoxicity that is attributed to black cohosh due to the difficulty in assigning causation as a result of using mixtures of multiple supplements, the temporal association of consumption and toxicity, and the presence of other confounding variables [72]. However, a recent case report found black cohosh to be a highly probable cause of hepatotoxicity [71]. In humans, the presence of 4HNE protein adducts in the hepatocytes of patients administered black cohosh was associated with acute liver necrosis and pathologic changes indicative of autoimmune hepatitis [70]. In mice, black cohosh administration increased liver weight and both the expression and activity of Cyp2b and Cyp3a [73]. However, this induction seems to be mediated by mouse but not by human PXR. Compounds isolated from the extract of black cohosh were triterpene glycosides, fukinolic acid, cimicifugic acid A, and cimicifugic acid B, which all inhibited multiple CYP450 enzymes (1A2, 2D6, 2C9, 3A4) [29]. Although these data suggest the potential for interaction with drugs that are metabolized by these enzymes, the authors found that these compounds were not directly toxic to the human hepatoma cell line HepG2, supporting the hypothesis that black cohosh causes an immune-mediated liver injury [29]. Black cohosh is hepatotoxic to humans, and components of black cohosh inhibit P450 enzymes; but, whether this toxicity is modulated by interactions with the P450-inhibitory compounds in the extract is unclear.

Grapefruit
The grapefruit (Citrus paradisi) is the result of an accidental cross between the pomelo (C. grandis) and the sweet orange (C. sinesis) that is believed to have originated in Barbados [77]. Capsules containing grapefruit extract are used to treat hypercholesterolemia, weight reduction, asthma, atherosclerosis, cancer, and depression [78][79][80][81][82]. Furocoumarins, dihydroxy bergamottin, and gergamottin found in grapefruit juice inhibit CYP3A4 activity in human liver microsomes [38,39]. Grapefruit juice also inhibits organic anion-transporting polypeptide [39]. In humans, the activity of clopidogrel, an antiplatelet drug used for coronary artery disease that requires activation by CYP2C19 and CYP3A4, is decreased by grapefruit juice [83].

Echinacea
Echinacea is a genus of flowering plants found in the Eastern and Midwestern US. It is used to treat colds, upper respiratory infections, and dermatologic issues, and contains cichoric acid, caftaric acid, and echinacoside [84][85][86]. Crude extracts of Echinacea activate hPXR to induce CYP3A4 gene transcription and increase the mRNA of CYP1A2, CYP3A4, and MDR1 in the HepG2 cell line [87]. However, Echinacea extract may inhibit the rifampicin-mediated induction of CYP3A4 transcription via hPXR [45] and has been reported to inhibit CYP1A2 and CYP3A [41,42].

Gardenia
Plants of the Gardenia genus are native to Africa, southern Asia, Australia, and Oceania. Geniposide is an iridoid glycoside found in the Gardenia jasminoides fruit. The fruit is used as a food coloring, an anti-inflammatory, an antithrombotic, and as an antidepressive [88][89][90]. Rats treated with geniposide or the crude extract of G. jasminoides had a decreased hepatic CYP3A4 expression and activity that was associated with an increase in glutathione [34]. Genipin, which is also found in G. jasminoides fruit, induces CYP2D6 and inhibits CYP2C19 and CYP3A4 mRNA, protein, and activity in the human hepatoma cell line HepG2. Genipin also decreased the expression of MDR1 (P-gp) [35]. Geniposide decreased hepatic CYP3a levels in rat livers [34]. The administration of gardenia yellow color caused hepatoxicity in rats [91]. Administering Gardenia yellow color, consisting of 30% geniposide (w/v), caused increased aspartate transaminase (AST) and alanine aminotransferase (ALT) in rats and histologic changes in rat liver [91]. The aqueous and alcoholic extract of Fructus gardeniae and geniposide all cause hepatoxicity in rats [92,93]. Cranberry, Grapefruit, Echinacea, and Gardenia alone may not be associated with toxicity in humans, but their effects upon the P450 system warrant concern.

Herbal Supplement Induction of P450 Enzymes
The hydroxylation of xenobiotics and endobiotics by P450 enzymes can produce daughter compounds that are more toxic than the parent compound ( Figure 2). This is problematic enough when the efficacy of pharmaceuticals is affected by decreasing plasma concentrations of drugs, especially in highly active anti-retroviral therapy regimens in which one component of the drug may increase the metabolism of another component and decrease the efficacy of anti-HIV treatment [94,95]. In the case of P450-mediated formation of toxic metabolites, this effect may be fatal. Fortunately, pharmaceuticals are evaluated for this potential. However, in the United States, herbal supplements are not required to be evaluated for their potential to induce the P450 system [1]. Hyperforin in St. John's wort, multiple compounds in Gingko biloba, piperine in black and white pepper, diallyl sulfide in garlic, and grapeseed extract increase p450 expression (Table 2). Table 2. Reports of herbal supplements with human Pregnane X Receptor (hPXR)-dependent or -independent induction of P450 expression.

Black pepper (Piper nigrum) * Piperine
Activation of CYP3A4 promoter via hPXR, increased mRNA and protein in intestinal cell lines and human hepatocytes [98].

Leaf extract
Increased activity of CYP2C19 reflected by decreased plasma concentrations of omeprazole and increased 5-hydroxyomeprazole in humans [103].
Herbs, compounds or preparations marked with an asterisk have been associated with alteration of P450 metabolism in human subjects or have been associated with hepatotoxicity and may be clinically significant.

The Pregnane X Receptor (PXR)
Activation of PXR by ligands such as rifampicin and many other compounds results in the transcription of genes that are involved in the transport and metabolism of xenobiotics. PXR is a nuclear hormone receptor (NHR), a class of proteins that are characterized by a DNA-binding domain, as well as a ligand-binding domain (LBD). The LBD of human PXR (hPXR) binds to many ligands with a wide range of different structures [108,109]. NHRs bind target DNA sequences in the promoters of target genes to induce their transcription. The LBD of PXR interacts with agonists to enable the recruitment of co-activating proteins to trigger the transcriptional activation [108][109][110][111]. PXR regulates the expression of cytochrome P450 enzymes (CYPs) CYP3A4, CYP2B6, CYP2C9, and CYP2C19; phase II enzymes, including UDP-glucuronosyltransferases and sulfotransferases; and transporters, including ATP-binding cassette transporter ABCB1 (also known as MDR1 or P-gp), multiple organic anion transporters, and multidrug-resistance protein 3 (MRP3) [6,7,[112][113][114]. The hPXR protein is expressed in the liver and intestines [115]. A mouse model in which mouse PXR is replaced with hPXR enables examination of the effect of hPXR-specific ligands on hPXR function in vivo [116,117].

The Constitutive Androstane Receptor (CAR)
CAR also controls the expression of xenobiotic metabolizing enzymes and transporters. Unlike PXR, CAR is constitutively active in the absence of ligand [118]. Agonist binding to CAR further activates CAR and results in the activation of such target genes as CYP2B6, the CYP2C subfamily, and CYP3A4 [9]. CAR also controls the expression of other genes that are involved in drug metabolism and transport [119,120]. Phenobarbital (used for refractory seizure conditions) induces CAR's dephosphorylation, which indirectly activates CAR and increases the transcription of target genes [9,121]. Although 1,4-bis[2-(3,5-dichloropyridyloxy)]benzene activates mouse CAR by direct binding, it does not affect human CAR [122]. Conversely, 6-(4-chlorophenyl)imidazo(2,1-b)(1,3)thiazole-5-carbaldehyde O-(3,4-dichlorobenzyl)oxime binds to and activates human CAR, but not mouse CAR [118,123]. Therefore, to study the in vivo effects of human CAR ligands, a humanized transgenic mouse model is used.

St. John's Wort
St. John's wort (Hypericum perforatum) is used to treat anxiety and depression. St. John's wort is native to Europe, but is now also cultivated in the US, Canada, and Australia [124]. A constituent of St. John's wort, hyperforin, increases the expression of CYP2C9 [99] and CYP3A4 [100] via activation of the hPXR. The concentration in commercial preparations of hyperforin can vary up to 10-fold between sources [100]. The use of St. John's wort can increase the metabolism of compounds and result in the formation of toxic products. Indeed, there has been a report of Hypericum perforatum induced liver injury that is associated with the use of copaiba, a herbal supplement that is used as an anti-inflammatory [125]. Whether this is due to hPXR activation is unclear. Although St. John's wort is not reported to be toxic alone, it may be responsible for increasing the toxicity of compounds that an organism is exposed to simultaneously. This toxicity may be mediated by the induction of the P450 system by St. John's wort constituents, such as hyperforin, activating hPXR.

Gingko Biloba
Gingko biloba is native to China but is now cultivated in many regions. It is used as an anti-hypertensive as well as to treat macular degeneration and tinnitus [126][127][128][129]. Gingko biloba extract increases hepatocyte DNA replication (an early indicator of hypertrophic change) and Cyp2b10, Cyp1a1, and Cyp3a11 mRNA in WT mice, but not in Car −/− mice [130]. This discrepancy suggests that the induction of the P450 system by mCar is associated with hepatic hypertrophy that is secondary to Ginkgo biloba consumption. Ginkgo biloba extract activates the transcription of a CYP3A4 promoter when an expression construct is overexpressed with either mouse or human PXR in HepG2 cells [101]. Ginkgo biloba extract also increases the expression of CYP3A4, CYP3A5, and ABCB1 (MDR1) mRNA in LS180 colorectal cancer cells [101]. Further characterization of specific components of Gingko biloba extract demonstrated that ginkgolide A and ginkgolide B induce the expression of CYP2B6, CYP3A4, UGT1A1, MDR1, and MRP2 mRNA in primary human hepatocytes [102]. Ectopically overexpressed hPXR is activated by both ginkgolide A and B in HepG2 [102]. Ginkgo biloba is associated with hepatotoxicity in humans [61]. Whole extract of Ginkgo biloba, as well as the isolated compounds gingkolide A and B themselves activate hPXR, thereby increasing the expression of CYP3A4, CYP3A5, and CYP2B6 and other genes controlled by hPXR. Whether the toxicity reported by humans taking Ginkgo biloba is a result of hPXR activation or of induction of the P450 system is unknown.

Ginseng
Ginseng is found in North America (Panax quinquefolius) and in eastern Asia (Panax ginseng). The roots and leaves are available as pills and in energy drinks, teas, and coffee beverages. Panax quinquefolius is used to reduce the incidence and severity of colds, whereas Panax ginseng is believed to enhance cognitive ability and to lower blood sugar levels; both contain ginsenosides and gintonin [97,131,132]. Wild Panax ginseng extract inhibits the induction of Cyp1a1 mRNA and protein expression induced by benzo-pyrene. Cyp1a1 metabolizes benzypyrene, and this activation is required for toxicity [133]. This activation is protective against benzo-pyrene-mediated toxicity in rats [133]. However, in clinical situations in which P450 activation results in the decrease of a toxic parent compound and the formation of a relatively non-toxic daughter compound, ginseng may contribute to toxicity (Figure 1). Such may have been the case in a patient administered imatinib [134]. The patient took imatinib daily for seven years with no complications, but after self-administering Panax ginseng as a constituent of an energy drink for three months, there may have been an interaction between imatinib and the ginseng that resulted in liver injury [135]. CYP3A4 and CYP3A5 metabolize 20(S)-protopanaxadiol, a ginseng sapogenin, in human liver and intestinal microsomes [97,136]. Additionally, several ginsenosides affect hPXR: Ginsenoside F2 and protopanaxadiol activate hPXR to induce the transcription of the CYP3A4 gene, whereas panaxotriol, Rg2, pseudoginsenoside F11, Rg1, ginsenoside, and Rb3 inhibt hPXR [97]. The potential for interactions and toxicity with ginseng administration may result from the ability of ginseng's components to affect the expression of P450 enzymes.

Piperine
Piperine is an alkaloid that is responsible for the flavor and smell of black and white pepper (Piper nigrum). White and black pepper are taken for GI distress, bronchitis, malaria, cholera, and cancer [137,138]. The piperamides in P. nigrum are thought to cause ROS production and to cause oxidative damage in cancer cells [139]. Piperine binds to and activates hPXR to induce the expression of CYP3A4 and MDR1 [98]. When administered simultaneously with CCl 4 , piperine increased the amounts of plasma liver enzymes, hepatic lipid peroxidation, and NADPH-cytochrome c reductase activity [10]. The toxicity of CCl 4 may be due to P450-mediated ROS production [140]. The increased activity of the P450 system due to hPXR activation by piperine may potentiate the toxicity associated with CCl 4 administration. Further studies of hPXR transgenic mice may fully elucidate the role of piperine in CCl 4 -mediated HILI.

Garlic
Garlic (Allium sativum) is used to treat a variety of medical conditions. Consumption of large amounts of garlic is believed to reduce the number of bites by insects, such as ticks [141]. Several linear sulfur-containing compounds are found in garlic oil, including diallyl sulfide, diallyl disulfide, and diallyl trisulfide [142]. Intraperitoneal injection of diallyl sulfide increased the expression of Cyp2b1 and Cyp2b2 mRNA in rat livers [107]. Diallyl sulfide induced Cyp2b10 mRNA in WT mice, and, to a lesser extent, in Car −/− mice [142]. Dially sulfide also increased Sult1e1 mRNA and protein expression as well as Car nuclear localization in the livers of WT mice but not in those of Car −/− mice [143]. Orally administered fresh garlic homogenate causes hepatotoxicity in rats [144,145]. Further studies using Car mouse models would reveal whether Car activation by garlic homogenate or diallyl sulfide is required for the observed toxicity.

Herbal Supplement Hepatotoxicity Mediated by the P450 System
In addition to affecting the toxicity of pharmaceuticals, the P450 system can mediate the toxicity of herbal supplements (Figure 3). Most of the compounds discussed here are terpenes, pyrrolizidine alkaloids, and ginsenosides (Table 3). These compounds are found in several herbal supplements that, in many instances, are combined with herbal supplements and pharmaceuticals that modulate the activity and expression of P450 enzymes. Pyrrolizidine alkaloids (PAs) are found in many different species that are consumed as herbal supplements [146]. Ginsenosides are found in ginseng [147]. Examples of terpenes include peppermint oil, menthol, camphor, germander, and pennyroyal. Table 3. Reported Hepatotoxicity of Herbal Supplements.
Ginseng (Panax ginseng and P. quinquefolius) Extract Possible liver injury in a patient after interaction with imatinib [134].
Single report of cholestasis in human [148].

Found in multiple species of plants
Monocrotaline CYP3A4-dependent hepatotoxicity in mice [168].

Characterization of Liver Injury by Pathophysiology
The clinicopathological characterization of DILI (reviewed in Wang et al., 2014) represents the clinical presentation of toxicity and the pathological effects of the injury and is a necessary step to assigning causation to a particular agent [172]. Cholestatic liver injury may be caused by the disruption of intracellular actin or transporter proteins ultimately resulting in bile stasis [173,174]. Hepatocellular liver damage is associated with compounds that are directly toxic to liver parenchyma and ALT elevation. ALT is expressed throughout the liver and is associated with diffuse hepatocellular liver injury [175]. Hepatotoxicity from herbal supplements is predominantly hepatocellular; 12% of patients experience severe, fatal DILI, 4% require liver transplant, and chronic injury develops in 10% [176,177]. Cholestatic liver injury is defined by elevated levels of serum alkaline phosphatase (ALP), γ-glutamyl transferase (GGT), and bilirubin [175]. This subtype is defined by an impeded bile flow and the deposition of bile acids within the liver lobules. Direct damage of biliary epithelium or impaired bile efflux transport can produce cholestasis [178]. Cholestasis can lead to fibrotic change and even hepatic necrosis. ALP is expressed in biliary epithelial cells and is released during biliary obstruction [179]. Mixed liver injury results in cases in which combinations of agents are used or when a single agent has a toxicity mechanism that damages multiple hepatic cell types [180].

Characterization of Liver Injury by Pathogenesis
Characterization of liver injury by pathogenesis considers intrinsic versus idiosyncratic types of injury [181]. Intrinsic liver injury is dose-dependent and in many cases is caused by the parent compound or a metabolite of the parent compound. Intrinsic liver injury associated with a measurable species in the blood or liver [182,183]. These reactions are predictable (i.e., occurring in animal models) and are usually detected in the preclinical development of pharmaceuticals. The pathogenesis of intrinsic liver injury is mediated by the compound and is usually reproducible in laboratory animals [181,[183][184][185][186]. Therefore, the pathogenesis of intrinsic liver injury is more easily studied than idiosyncratic liver injury. In the more common form of liver injury, idiosyncratic liver injury, the modification of compounds by the P450 system can result in the formation of protein adducts that are recognized as antigens [187][188][189]. In many cases, these adducts are not detectable in the blood and this effect is not reproducible in laboratory animals [190]. Idiosyncratic liver injury likely results from a combination of genetic and environmental factors that are not reproducible in laboratory animals [181,191]. Unlike intrinsic liver injury, the pathogenesis of idiosyncratic liver injury is not unique to the compound initiating the reaction, but is dependent upon an immune response [192,193]. This type of injury can only be detected in clinical development, but is more commonly only detected after the compound is marketed, due to a relatively low incidence hiding the effect in small study populations.

Assessment of Causality
Predicting the risk of liver injury due to herbal supplements requires both a standardized causality assessment method and a repository of this information reported in the clinic. The Council for International Organizations of Medical Sciences (CIOMS), Working Group IX, developed the Roussel Uclaf Causality Assessment Method (RUCAM), a list of tools and guidelines for researchers and clinicians to evaluate and manage the risk of medicinal products (including herbal supplements) [25,194,195]. These standardized tools were developed to define and assess causality of pharmaceutics in causing DILI. The guidelines were developed to standardize the evaluation of DILI during development and after the public release of pharmaceuticals. The application of this method requires classifying liver injury according to pathophysiology, time to onset, liver tests, historical risk factors, concomitant drug and herb use, patient history, and the response to re-exposure of the compound. The original RUCAM addressed the lack of defined items that had previously resulted in a lack of consensus between evaluators and also introduced the consideration of chronological data. This original version was plagued by intra-and inter-observer variability, as well as a lack of consideration of differential diagnoses due to non-drug/herb causes [195][196][197]. An update to RUCAM addressed these weaknesses and streamlined the evaluation of these cases. The updated definitions of the classification items reduces the variability of assessments and the dependence upon outside experts not involved with the case [195].

Peppermint Oil, Pennyroyal, and Menthol
Menthol is prepared from corn mint or peppermint, or is made synthetically. Menthol is administered topically to treat minor muscle and joint pain by causing the skin to feel cool and then warm, distracting patients from feeling deeper pain [198]. Pennyroyal (Mentha pulegium) is native to Europe, North Africa, and the Middle East. Another species (Hedeoma pulegioides) is native to North America. Pennyroyal is traditionally used in cooking and as an abortifacient. It is also used for colds, pneumonia, and dyspepsia [22,159,199]. Pulegone and menthol cause hepatotoxicity when orally administered to rats [164,167]. Pennyroyal oil causes hepatotoxicity in humans and in mice [159][160][161][162]. Menthone, a component of peppermint oil, caused hepatotoxicity in rats treated with it for 28 days by oral gavage [165]. Pugelone is metabolized to menthofuran by oxidation of the allylic methyl group, followed by an intramolecular cyclization reaction to from a hemiketal, and subsequent dehydration to form a furan [22]. Purified CYP2E1, CYP1A2, and CYP2C19 oxidized pulegone to menthofuran. Then subsequently oxidized menthofuran to 2-hydroxymenthofuran, an intermediate in the formation of mintlactone and isomintlactone [23].
Menthofuran is a monoterpene that is found in several species of mint plants and is oxidized by cytochrome P450 enzymes to hepatotoxic metabolites. The metabolism of menthofuran requires CYP1A2, CYP2E1, and CYP2C19 [22]. These cytochrome P450 enzymes oxidize menthofuran to 2-hydroxymenthofuran, which arises from a dihydrodiol produced from a furan epoxide and is an intermediate in the formation of mintlactone and isomintlactone [22]. The metabolites of menthofuran identified in human and rat liver microsomes that may be responsible for hepatotoxicity, are a γ-ketoenal and epoxides formed by oxidation of the furan ring, which form conjugates with hepatic proteins. CYP1A2, CYP2B6, and CYP3A4 are responsible for the formation of GSH conjugates in human liver fractions [24]. The proteins adducted in rat livers are serum albumin, aldehyde dehydrogenase (ALDH2), malate dehydrogenase (MDH1), and ATP synthase subunit d [166]. The activities of ALDH and ATP complex V were both decreased [166]. Inhibition of cytochrome P450 enzymes with SKF-525A, metyrapone, piperonyl butoxide, and carbon disulfide prevented or reduced the hepatoxicity of pugelone [200].
CYP2A6 is the major cytochrome P450, which is involved in the oxidative metabolism of menthol in human liver microsomes to form (+)-(1S,3S,4S)-and (−)-(1R,3R,4R)-trans-p-methane-3,8-diol derivatives [201]. (−)-Menthol inhibits CYP2A13 and CYP2A6 [32,33]. Additionally, the monoterpene (R)-(+)-pulegone causes increases in the marker of liver injury, glutamate pyruvate transaminase, in mice treated with pulegone that are mitigated by pretreatment with disulfiram and cimetidine, inhibitors of CYP2E1 and CYP1A2, respectively [163]. Additionally, menthol may affect the metabolism of pharmaceuticals. Pretreatment of mice with menthol increased the clearance of warfarin and the expression of Cyp2c protein, with a concomitant increase in Car nuclear translocation [202]. Peppermint oil, menthol, and methyl acetate reversibly inhibited the metabolism of nifedipine. Peppermint oil also increased the AUC of felodipine in human liver microsomes containing CYP3A4 [55]. In addition to direct toxicity resulting from pulegone metabolism mediated by the P450 system, components found that peppermint oil, pennyroyal oil, and menthol may directly affect the P450 system, leading to modulations and toxicity of concomitant herbal supplements and pharmaceuticals.

Camphor
Camphor was traditionally prepared from the wood of the camphor laurel (Cinnamomum camphora), which is native to Asia. Camphor is now predominantly produced from turpentine and is used in such products as Vicks VapoRub™. Camphor products are usually applied topically to reduce dermatologic complaints and the symptoms of respiratory tract disease [203,204]. (−)-Camphor is a bicyclic monoterpene that is found in the essential oil produced from dalmation sage and is metabolized by CYP2A6 of bacteria overexpressing the human protein [205]. Camphor may be hydroxylated by cytochrome P450 enzymes [206]. In fact, hepatotoxicity of camphor applied topically has been reported in an infant [158]; however, the toxicity may not be ascribable directly to topical administration and could have resulted from unobserved ingestion.

Germander
Germander (Teucrium chamaedrys and other species of the genus Teucrium) is used to treat gallbladder disease, fever, diarrhea, gout, and for weight loss [207][208][209]. Germander is banned in France, and its use is restricted in the US and Canada [210]. Germander is associated with hepatotoxicity in humans [78,124,125]. The furanoneoclerodane diterpene teucrin A is thought to be responsible for the toxicity of germander, the activation of which in rats [211] and isolated rat hepatocytes [212] led to covalent modification of hepatic proteins. Indeed, pretreatment with phenobarbital, a CAR activator, increased the hepatotoxicity of teurcin A [157], which is activated by cytochrome P450 enzymes and is required for hepatocellular damage in mice. This is concerning because the method of preparation of germander decoctions can affect the concentration of teucrin A [213]. Teucrin A and teuchamaedryn A were found to be hepatotoxic to isolated rat hepatocytes, and this toxicity was decreased by co-treatment with a CYP3A inhibitor [156]. The toxicity was associated with the formation of reactive metabolites that covalently bound hepatic proteins [156]. These proteins were further identified and were found to be mostly of mitochondrial and endoplasmic reticulum origin, and were abducted by the 1,4-enedial derivative of teucrin A to lysine and cysteine residues [211]. The hepatotoxicity of Germander, mediated by teucrin A-adduct formation may be mediated by the P450 system. This is concerning due to reports of hepatotoxicity in humans [61,153,154], as well as its potential use in combinations of other herbal supplements and pharmaceutics that may potentiate adduct formation and subsequent toxicity.

Pyrrolizidine Alkaloids
Pyrrolizidine alkaloids (PAs) and their N-oxides are found in approximately 3% of flowering plants. These plants are predominantly members of the Asteraceae, Boraginaceae, Compositae, and Fabaceae families, and can also be Senecio, Heliotropium, Crotalaria, and Symphytium species [214,215]. PAs contain pyrrolizidine, two fused penta rings with nitrogen at position 4. Pyrrole derivatives that react with DNA may cause injury to hepatic vasculature and liver parenchyma [216,217]. (±)-6,7-dihydro-7-hydroxy-1-hydroxymethyl-5H-pyrrolizine-derived DNA adducts cause hepatocarcinoma in rats and are formed from riddelline and monocrotaline by human or rat liver microsomes [218]. Senecionine, monocrotaline, adonifoline, and isoline can be conjugated to glucuronic acid by UGTs in human liver microsomes [219]. 9-Glutahionyl-6,7-dihydro-1hydroxymethyl-5H-pyrrolizine was detected as conjugated to glutathione in human liver microsomes and in the bile of rats treated that were with the pyrrolizidine alkaloids isoline, retrorsine, and monocrotaline [220]. Monocrotaline caused severe liver and moderate renal toxicity in mice; the degree of toxicity was abrogated by pretreatment with ketoconazole, a CYP3A4 inhibitor [168]. Correspondingly, the presence of N-oxide metabolites of monocrotaline was decreased by pretreatment with ketoconazole. Dehydromonocrotaline and dehydrorectronecine are metabolites of monocrotaline. Dehydromonocrotaline and dehydrorectronecine, but not monocrotaline, caused toxicity in human hepatic sinusoidal endothelial cells and HepG2 cells [168,169]. Retrorsine inhibits the activity of CYP3A4, as measured by testosterone hydroxylation [37]. The hepatotoxicity of PAs and their derivatives is mediated by the P450 system and care should be exercised when herbal supplements containing these compounds are used in combination with other herbal supplements and pharmaceuticals that induce the P450 system.

Discussion
Herbal supplement use can lead to HILI due to interactions with the P450 system. These interactions may not be predicted until a series of reports of HILI associated with the use of a particular herbal supplement leads to an investigation of potential interactions. Unlike prescription drugs regulated by the FDA (and international agencies), the potential of herbal supplements to induce or inhibit P450 enzymes is not investigated before the supplements are made available to the general population. This lack of information before a supplement is marketed is compounded by the difficulties in event-reporting and history-taking concerning herbal supplement use. Difficulty in assigning an etiologic designation to a particular xenobiotic in patients with positive histories of multi-pharmaceutical, multi-supplement, or mixed pharmaceutical and supplement regimens complicates what "post-marketing" reporting is available. Many networks for reporting exist, but are weakened by difficulties of self-reporting of herbal supplement use by patients and by poor history-taking. However, careful history-taking by physicians may prevent an interaction from taking place and identify the likely offending agent.
Previous articles have discussed herbal supplements and HILI, but this review has focused on herbal supplements that may interact with the P450 system to exacerbate the toxicity of other drugs or to cause toxicity themselves [221][222][223][224]. Herbal supplements may contribute to HILI by inhibiting P450 enzyme activity, inducing P450 gene transcription, or being metabolized by P450 enzymes without affecting their levels or activity. The inhibition of P450 enzymes may contribute to HILI by increasing the exposure of a toxic compound that is metabolized by P450 enzymes. Catechins in green tea, terpenes in black cohosh and cranberry, geniposide and genipin in Gardenia, fucomarins in grapefruit juice, and Echinacea extract may inhibit the activity or decrease the expression of cytochrome P450 enyzmes ( Table 1). The induction of P450 genes may contribute to HILI by increasing the exposure of a toxic parent compound that is formed by P450 metabolism or by the release of ROS. Hyperforin in St. John's Wort, multiple compounds that are found in Gingko biloba, piperine in black and white pepper, diallyl sulfide in garlic, and grapeseed extract increase P450 expression (Tables 1 and 2). The metabolism of herbal supplements by P450 enzymes may contribute to HILI by resulting in the production of a toxic parent compound. Most of the compounds discussed here are terpenes, pyrrolizidine alkaloids, and ginsenosides.
Careful analysis of existing literature combined with extensive history-taking concerning the use of herbal supplements can help to prevent herb-drug interactions and HILI. The recognition of and reporting of HILI events by physicians can lead to greater knowledge of potential interactions and hepatotoxicity, enabling them to better educate patients of the dangers of some combinations.