Roles of Cytochrome P450 Enzymes in Pharmacology and Toxicology: Past, Present, and Future

Abstract

The development of the cytochrome P450 (P450) field has been remarkable in the areas of pharmacology and toxicology, particularly in drug development. Today it is possible to use the knowledge base and relatively straightforward assays to make intelligent predictions about drug disposition prior to human dosing. Much is known about the structures, regulation, chemistry of catalysis, and the substrate and inhibitor specificity of human P450s. Many aspects of drug-drug interactions and side effects can be understood in terms of P450s. This knowledge has also been useful in pharmacy practice, as well as in the pharmaceutical industry and medical practice. However, there are still basic and practical questions to address regarding P450s and their roles in pharmacology and toxicology. Another aspect is the discovery of drugs that inhibit P450 to treat diseases.

Introduction and History

The field of cytochrome P450 (P450 or CYP) research had its origin in studies on the metabolism of drugs, steroids, and carcinogens in the middle of the 20^th Century (Axelrod, 1955; G. C. Mueller & Miller, 1948; Ryan, 1959). However, the discovery of P450 as such did not occur until a few years later (Klingenberg, 1958; Omura & Sato 1962, 1964). The evidence for a role as the terminal oxidase in a hydroxylation developed with the 17α-hydroxylation of a steroid (Cooper, Levine, Narasimhulu, Rosenthal, & Estabrook, 1965). Studies on a bacterial P450 by Gunsalus developed independently (Hedegaard & Gunsalus, 1965; Katagiri, Ganguli, & Gunsalus, 1968), and that system (P450_cam, or CYP101A1) served as a useful model for many years (Mueller, Loida, & Sligar, 1995). Two papers in 1968 and 1969 by Lu and Coon established the identity of the liver microsomal P450 system involved in oxidations, consisting of three components: a P450, NADPH-P450 reductase (POR), and phospholipid (Lu & Coon, 1968; Lu, Junk, & Coon, 1969).

More about the historical development of the field was described in a recent review (Guengerich, 2019a). With considerable effort, many liver P450s (and some extrahepatic ones) were purified by conventional chromatography methods and characterized. Progress was also made in terms of mechanisms of catalysis and gene regulation. The introduction of recombinant DNA technology led to cloning of cDNAs, expression of P450s in heterologous systems, and ultimately a better understanding of the complexity of the P450 Superfamily with the completion of the Human Genome Project. Today the field of P450 research must be considered as mature, but that is not to say that all important questions have been answered. As a field matures, the background knowledge and the research tools improve and more important questions can be addressed.

The focus of this book is on pharmacology and the roles of P450 enzymes in the metabolism of drugs. However, P450s also have important roles in the metabolism of steroids (some of which are used as drugs), fat-soluble vitamins, fatty acids, chemical carcinogens, pesticides, industrial chemicals, food additives, and other chemicals. Collectively > 90% of all oxidations and reductions of chemicals known today are catalyzed by P450s (Rendic & Guengerich, 2015). This high percentage is also in part due to the preponderance of P450 reactions in the biosynthesis of natural products (Guengerich, 2021b), as well as drugs and industrial chemicals. P450s are found throughout nature, with the only current exceptions being some enteric bacteria (e.g., Escherichia coli, Salmonella typhimurium). The number of CYP genes in bacteria and plants probably exceeds the number in mammals, in large part because most plants have hundreds and sometimes > 1,000 CYP genes (e.g., wheat has 1,285).

Where is the P450 field today and what do we know?

This is an introductory chapter, and several other chapters will focus on some detailed aspects of P450 science. The focus here will be on human P450s, although the P450s in experimental animals are also still of great interest in the drug development process.

Roles of individual human P450s

One of the ways of grouping the 57 human P450s by function is presented in Table 1. Of these, it is not clear that CYP2A7 is expressed but CYP4F3 yields two proteins, so the number is still 57. The classification by substrates is not without caveats. Some P450s can be classified under multiple headings (e.g., 1B1 for steroids and xenobiotics, 27A1 for steroids and vitamins). Some of the P450s have moved from “orphan” status (Unknown in Table 1) but it is not clear how important these reactions are (e.g., 2U1, 2S1). P450 4X1 can slowly oxidize anandamide (Stark, Dostalek, & Guengerich, 2008) but has been left in the Unknown column. It is not clear how important most of the reactions of Xenobiotics and Fatty acids are to mammalian physiology. The point can be made that the P450s in the Xenobiotics column have a general function of clearing a wide variety of ingested natural products present in our food (e.g., terpenes, alkaloids) as a general protective mechanism, in the same way that export transporters do. Studies with transgenic mice have shown that the orthologs of many of the P450s listed under the headings of Xenobiotics and Fatty acids are not essential (Bissig et al., 2018; Gonzalez & Kimura, 2003). However, those involved in the metabolism of steroids, eicosanoids, and vitamins generally are essential.

Table 1.

Classification of human P450s based on major substrate class. This classification is somewhat arbitrary in some cases, e.g., P450s 1B1 and 27A1 could be grouped in either of two different categories.

Eicosanoids
Steroids	Xenobiotics	Fatty acids	Eicosanoids	Vitamins	Unknown
1B1*	1A1*	2J2	2U1	2R1*	2A7
7A1*	1A2*	2S1	4F2	24A1**	4X1
7B1	2A6*	2U1	4F3	26A1	20A1
8B1	2A13*	4A11	4F8	26B1
11A1*	2B6*	4A22	5A1	27A1
11B1*	2C8*	4B1**	8A1*	27B1
11B2*	2C9*	4F11		27C1
17A1*	2C18	4F12
19A1*	2C19*	4F22
21A2*	2D6*	4V2
27A1	2E1*	4Z1
39A1	2F1
46A1*	2W1
51A1*	3A4*
	3A5*
	3A7*
	3A43

^*Crystal structure available.

^**Crystal structure of animal orthologue available.

Abundance of P450s

Of the 57 P450s (Table 1), 50 are expressed mainly in the endoplasmic reticulum and seven are expressed by nuclear genes but transported (following proteolysis) to the mitochondria (11A1, 11B1, 11B2, 24A1, 27A1, 27B1, 27C1). Fractions of some of the endoplasmic reticulum (microsomal) P450s are cleaved and also enter the mitochondria (e.g., 1B1, 2D6, 2E1, 2C8) (Avadhani, Sangar, Bansal, & Bajpai, 2011). The microsomal P450s receive electrons (from NADPH) via the diflavin protein POR and sometimes cytochrome b₅ (b₅). Those in the mitochondria use a system involving the flavoprotein NADPH-adrenodoxin (Adx) reductase and Adx. Although the mitochondrial P450s clearly have important roles in the metabolism of steroids and vitamins (Table 1) (Guengerich, 2015), in some cases they can also be involved in the metabolism of drugs (Zhang et al., 2012) and other chemicals.

In mammalian liver the ratio of total P450 to POR has long been known to be 10–20:1 (Estabrook, Franklin, Cohen, Shigamatzu, & Hildebrandt, 1971). The concentrations of several P450 in human liver have been estimated using immunochemical (Shimada, Yamazaki, Mimura, Inui, & Guengerich, 1994) and, more recently, mass spectrometry proteomic approaches (Achour, Al Feteisi, Lanucara, Rostami-Hodjegan, & Barber, 2017). The results from several studies are summarized in Fig. 1. While it is clear that P450 3A4 and two P450 Subfamily 2C enzymes (2C8, 2C9) are the most abundant, there is a large amount of variability, even in cases where the same liver sets were analyzed (Guengerich, 2015). For instance, it is not clear whether P450s 2A6 and 2B6 should be considered minor or abundant enzymes (Fig. 1) (Guengerich, 2015).

Figure 1.

Percentages of total P450 in human liver samples accounted for by each P450. The data points were compiled (Guengerich, 2021a) from four sets with multiple liver samples (Achour, Russell, Barber, & Rostami-Hodjegan, 2014; Kawakami et al., 2011; Shimada et al., 1994) and one with a single liver sample high in P450 1A1 (Lang, Radtke, & Bairlein, 2019). The estimates were made immunochemically in one case (Shimada et al., 1994) and by LC-MS proteomic methods in the others (Achour et al., 2014; Kawakami et al., 2011; Lang et al., 2019). The value for P450 1A1 is a mean of measurements of 30 samples (Lang et al., 2019). The individual colors have no meaning but are added to facilitate visualization.

The composition of individual P450s in human small intestine has also been analyzed (Paine et al., 2006). In this organ, the dominance of P450 3A4 is even more striking and this has relevance in considering the disposition of only administered drugs and inhibition of drug metabolism. The total amount of P450 in the small intestine is only a few percent of that in liver, however, and this point needs to be considered in the context of first-pass clearance.

Regulation

Many of the P450s are subject to enzyme induction, as well as localization in different tissues due to the influence of tissue-specific promoters. A general scheme for induction (Fig. 2) involves binding of a ligand to a receptor, formation of a heterodimeric pair, nuclear transport, and binding to specific sites of the gene to cause (chromosome rearrangement and) enhanced transcription by RNA polymerase (Fig. 2). This is the general pattern seen for the AhR, PXR, PPARα, and RAR systems of gene regulation. With AhR the heterodimer partner is ARNT. With the bulk of the systems, which use receptors from the steroid nuclear receptor superfamily (PXR, PPARα,…), the partner is RXR, which is bound to retinoic acid or possibly another ligand. CAR, involved in regulation of P450s 2B6 and 3A4, is different in that while it can bind some ligands (e.g., 1,4-bis[2-(3,5-dichloropyridyloxy)]benzene, (TCPOBOP) (Maglich et al., 2003), in most cases the receptor is constitutively active and nuclear import is regulated by a phosphorylation cascade involving EGFR (Mutoh et al., 2013).

Figure 2.

General scheme for transcriptional regulation of P450s. L: ligand, R: receptor, Ŕ-heterodimeric partner, Coactiv: co-activator protein (e.g., hepatic nuclear factor (HNF) α in the case of P450 3A4), RNA pol: RNA polymerase (Guengerich, 2018a, 2021a).

Induction by drugs is important for several reasons: (1) It leads to changes in pharmacokinetics when the drug of interest is also an inducer. (2) Drug-drug interactions can be important clinically, as seen in the classic example of enhanced metabolism of 17α-ethnylestradiol (in oral contraceptives) by P450 3A4 inducers (Bolt, Kappus, & Bolt, 1975). (3) In some animal models, enzyme induction is correlated with development of certain cancers, particularly in rodent liver (e.g., barbiturates, PPARα inducers) (Lubet, Nims, Ward, Rice, & Diwan, 1989; Rao & Reddy, 1987). Although this is much less of a regulatory concern than in the past, the development of rodent tumors in the drug development scenario must be explained and regulatory agencies need assurance that this will not be an issue in humans.

The most thoroughly studied model of P450 induction is transcriptional control. However, regulation can also be at the post-translational level, including mRNA and protein stabilization, and epigenetic control. Examples of roles of gene methylation (i.e., 5-methyl deoxycytidine), histone modification (e.g., acetylation), and microRNA involvement are now known for P450s (Guengerich, 2015; Ingelman-Sundberg et al., 2013), although the significance in humans is still not established.

P450 genes can also be regulated by cytokines. Interferons can down-regulate P450s, and the suppression of drug metabolism by interferons has long been known to be associated with colds, flu shots, etc. (Mannering, Renton, el Azhary, & Deloria, 1980; Renton, 1981). Another phenomenon observed in rats is the down-regulation of some P450s by some of the common inducers, e.g. barbiturates and particularly Family 1 inducers, as seen particularly with P450s 2C11 and 2E1 (Dannan, Guengerich, Kaminsky, & Aust, 1983; Guengerich, Dannan, Wright, Martin, & Kaminsky, 1982; Thomas, Bandiera, Maines, Ryan, & Levin, 1987). This suppression has been shown to occur at the transcriptional level (Sawaya & Riddick, 2008) but its relevance in humans is unknown.

Rodents display dramatic sex effects with regard to P450 regulation (Waxman, Dannan, & Guengerich, 1985; Waxman & Holloway, 2009). The basis of this is complex and involves not only androgens and estrogens but also pulsatile patterns of growth hormone and JAK/STAT regulation (Waxman & Holloway, 2009; Wiwi & Waxman, 2005). Although there are some reports of sex differences in some P450s in humans (Wolbold et al., 2003; Zhang et al., 2011), the differences have not been seen by others (Yang et al., 2010) and, at the pharmacokinetic level, may be attributable to body fat. However, knowledge of sex differences in rodent P450s may be important in understanding the results of pre-clinical testing in drug development.

Catalytic mechanism

Much has been written about chemical mechanisms of catalysis by P450s elsewhere (Ortiz de Montellano, 2015; Guengerich, 2018b; Guengerich & Yoshimoto, 2018).

The catalytic cycle is shown in Fig. 3, where the P450 binds substrate (Step 1), the iron is reduced (Step 2), oxygen binds (Step 3), and the second electron is donated to the iron (Step 4). At this point the intermediates are unstable, and information about them has taken some time to accumulate. The Fe³⁺-O₂⁻ form (called Compound 0) becomes protonated (Step 5) and then H₂O is released to leave Compound I (after Step 6). In Step 7 the formal FeO³⁺ complex abstracts a hydrogen atom (or an electron from a heteroatom if the redox potential is low enough) to leave a “caged” radical (Step 7), which undergoes recombination with Compound II (FeOH³⁺) to generate the product in Step 8. Finally, the product (ROH) is released in Step 9.

Figure 3.

P450 catalytic cycle. The nine labeled steps show sequential (1) substrate binding, (2) 1-electron reduction, (3) oxygen binding, (4) second 1-electron reduction, (5) protonation of “Compound 0”, (6) loss of water to form “Compound I”, (7) hydrogen atom abstraction by Compound I, (8) oxygen rebound to form product, and (9) product dissociation. As indicated, ferrous P450 can also bind substrate (Yun, Kim, Calcutt, & Guengerich, 2005). In some cases, b₅ can provide the electron in step 2 or 4. In some sequential reactions, step 9 does not occur and a second oxidation of the initial product is observed (E. Gonzalez & Guengerich, 2017; Reddish & Guengerich, 2019).

The P450 can be reduced without having substrates present, at least with some P450s (Guengerich & Johnson, 1997; Johnston et al., 2011). In some cases there is evidence that b₅ provides the second electron (in Step 4) but in other cases b₅ can stimulate P450 reactions without electron transfer (Yamazaki et al., 2002). Another point is that the cycle in Fig. 3 relates only to the electronic changes that occur, but numerous changes in protein structure occur as well, and even binding of a substrate can involve a complex series of steps (Guengerich, Wilkey, Glass, & Reddish, 2019; Guengerich, Wilkey, & Phan, 2019; Isin & Guengerich, 2006).

An appreciation of the catalytic mechanism of P450 is important in understanding the kinds of reactions that P450s can do. In areas such as drug metabolism and natural product biosynthesis, products must be characterized and a knowledge of possible mechanisms is needed to discern possible pathways (Guengerich & Yoshimoto, 2018; Isin & Guengerich, 2007a).

Although possibilities have been raised of various other oxidant forms of P450 in various oxidations, almost all reactions can be explained by involving Compound I reactions. Some proposals for Compound 0 or other species have been re-valuated or analyzed further and re-interpreted in terms of Compound I (Groves, McClusky, White, & Coon, 1978; Guengerich & Yoshimoto, 2018; Huang & Groves, 2017; Krest et al., 2013; Rittle & Green, 2010; Yoshimoto & Guengerich, 2014). Only in a few cases has P450 Compound I been prepared directly and rigorously characterized (by reaction with a peracid) (Krest et al., 2013; Rittle & Green, 2010). Some bona fide Baeyer-Villiger-type oxidations may still prove to involve Compound 0 (Guengerich, 2021b).

Structures of P450s and binding of ligands

Although X-ray crystallography of P450s was limited to soluble bacterial P450s before 2000, the work of Johnson (Williams, Cosme, Sridhar, Johnson, & MeRee, 2000) and then others has led to a plethora of P450 structures. As of 2021 there were at least 260 structures of mammalian P450s in the Protein Data Bank, and 25 of the 57 human P450s have crystal structures available (plus apparent animal orthologues of P450 4B1 and 24A1).

All P450 structures to date have similar overall folds (Fig. 4). The interaction and movements among the I, F´, and G´ helices are important in modulating ligand specificity.

Figure 4.

A structure of P450 3A4 (Protein Data Bank (PDB) 1TQN), with major helices labeled (Yano et al., 2004). The heme prosthetic group is shown in gray.

Some of the human P450s have been crystallized in open, closed, and intermediate forms (Guengerich, Waterman, & Egli, 2016; Poulos & Johnson, 2015). A single structure of a P450 provides useful information about the bonding of a P450 with a substrate but it may not present a picture of how the P450 bound that substrate, i.e. the course of events leading to (productive) binding. Some of the P450s have been found to bind substrates in multiple ways and also to have multiple conformations in the absence of a substrate or other ligand (Ekroos & Sjögren, 2006; Hsu & Johnson, 2019; Porubsky, Battaile, & Scott, 2010).

One hypothesis about how enzymes such as P450s are able to bind so many substrates is that of induced fit (Fig. 5) (Koshland, Nemethy, & Filmer, 1966); i.e. binding of a substrate to an enzyme induces the enzyme to adopt a new conformation that is more favorable for productive catalysis. An alternative mechanism involves conformational selection (Fig. 5), where the enzyme exists in multiple conformations in the absence of ligand, one (or more) of which binds the substrate to yield a productive complex (Fig. 5). These are not necessarily completely distinct phenomena and may occur together. Discerning which course (Fig. 5) is dominant is usually difficult, in that the free energy involved in the route from E to a productive ÉS complex is identical regardless of the route (Chakraborty & Di Cera, 2017; Vogt, Pozzi, Chen, & Di Cera, 2014). One hallmark of the presence of complex binding pathways is slow kinetics, i.e. at less than diffusion-limited rates (Johnson, 2019). The two routes (Fig. 5) can be distinguished by measuring the kinetics of binding as a function of varying the concentration of ligand, enzyme, or both (Gianni, Dogan, & Jemth, 2014; Vogt & Di Cera, 2012). Such kinetic studies have been done with several human P450s and indicate the dominance of conformational selection pathways (Guengerich, Wilkey, Glass. & Reddish, 2019; Guengerich, Wilkey, & Phan, 2019). The binding of the preferred substrate camphor to bacterial P450_cam (P450 101A1) appears to be an exception (Guengerich, Child, Barckhausen, & Goldfarb, 2021), but the conformational selection mechanism appeared to be more dominant with alternate substrates of P450_cam.

Figure 5.

Hypotheses to explain complex substrate recognition data (Gianni et al., 2014; Vogt & Di Cera, 2012).

The binding of inhibitors to P450 3A4 has been shown to be a complex process, with multiple steps and spectrally detectable intermediates (Fig. 6) (Guengerich, McCarty, & Chapman, 2020; Isin & Guengerich, 2007b). Achieving full inhibition requires completion of the steps for P450 3A4 (i.e., the E*I complex in Fig. 6). With P450 17A1, multiple spectral intermediates are seen upon mixing but inhibition occurs immediately, before the spectral changes are completed (Fig. 6) (Child & Guengerich, 2020; Guengerich, McCarty, Chapman, & Tateishi, 2021). The difference may be due to the large size of the active site of P450 3A4 (~ 1400 ų (Yano et al., 2004)), which is able to accommodate two molecules of the inhibitor ketoconazole (Ekroos & Sjögren, 2006). No crystal structure of a P450 containing both a substrate and inhibitor has been published but is certainly feasible for P450 3A4 and probably some other P450s.

Figure 6.

Scheme summarizing interaction of P450 3A4 with inhibitors. The times of appearance of individual species are indicated in blue (Guengerich et al., 2020).

P450s and drug metabolism

In the early history of P450 research, little information was available about how many P450s existed, how many had major roles in drug metabolism, and which of these P450s metabolized individual drugs. Today the human P450s are all known (Table 1), with the completion of the human genome and recognition of the P450 signature sequence:

Phe-X-X-Gly-X-Arg-X_b-Cys-X-Gly

where the Cys is liganded to the heme iron atom and X_b is a basic residue. P450s are involved in the metabolism of ~ ¾ of (small molecule) drugs (Fig. 7), and about five P450s are involved with 90% of the drugs (Guengerich, 2015; Rendic & Guengerich, 2015). Those fractions have remained similar for new drugs, with P450 3A4 playing an even more dominant role (Fig. 7). This trend may be due, at least in part, to (i) a tendency towards larger molecules, in efforts to achieve selectivity and potency, and (ii) efforts to avoid dependence on the P450s showing more genetic polymorphism (e.g., 2C19 and 2D6).

Figure 7.

Fractions of small molecule drugs approved by US FDA in 2015–2020 metabolized by individual enzymes (Bhutani et al., 2021). UGT: uridine diphosphate glucuronosyl transferase; FMO, flavin-containing monooxygenase; AO, aldehyde oxidase. Reprinted from J. Med. Chem., Vol. 64, Bhutani, P., Joshi, G., Raja, N., Bachhav, N., Rajanna, P. K., Bhutani, H., Paul, A. T. and Kumar, R. US FDA approved drugs from 2015-June 2020: A perspective, pages 2339–2381, Copyright (2021), with permission from the American Chemical Society.

P450s and pharmacokinetic issues

One issue in drug development is prediction of sites of metabolism. Over the years there has been some progress in the in silico prediction of sites (Afzelius et al., 2007; Boyer et al., 2007; de Bruyn Kops, Sicho, Mazzolari, & Kirchmair, 2021; Ekins et al., 2005; Kirchmair et al., 2015; Martiny & Miteva, 2013; Wilson, White, & Mueller, 2003), especially if the “top three” sites are all predicted. Much of the success has been achieved with algorithms based on prior examples, as opposed to docking into X-ray structures. Nevertheless, there will probably always be some surprises regarding in silico predictions, e.g. testosterone is hydroxylated by P450 3A4 mainly at the 6β (as well as 2β, 1β, and 15β) carbon but 4,5-dihydrotestosterone is hydroxylated at the (chemically more inert) 18- and 19-methyl carbons (Cheng, Sohl, Yoshimoto, & Guengerich, 2012).

As molecules progress in the discovery/development process, they do require the use of analytical chemistry to define structures of metabolites. Progress in the past three decades in LC-MS and NMR has greatly improved the process, and there are novel techniques with possibilities, such as crystallization and X-ray diffraction of trapped compounds (Rosenberger et al., 2020).

What is more difficult is the prediction of rates of metabolism by P450s, although there are claims to be able to do this with artificial intelligence (Xiong et al., 2021). This is probably only realistic in situations where, for instance, rates are known for close analogs and the effects of adding substituents are subject to Hammett analysis or other linear free energy relationships (Burka, Guengerich, Willard, & Macdonald, 1985). Rates of (total) oxidative metabolism can be measured in relatively high throughput assays with liver microsomes and LC-MS, however. Such assays can be done with hepatocytes but not as rapidly or large-scale. The microsomal assays are a rapid means of stratifying for drug stability. However, if pharmacologically active products are formed, the results will be misleading regarding the value of a drug candidate.

Changing molecules to attenuate metabolism

When a lead drug is metabolized too rapidly, there may be possibilities for slowing the metabolism. To do this effectively, the site of oxidation should be known. If the P450 involved in oxidation is known, it is possible to dock the molecule to suggest changes that might prevent metabolism or bioactivation while maintaining pharmacological activity (Brodney et al., 2015). Strategies may involve (i) adding a moiety (at the site) that will resist oxidation or prevent binding to the P450, (ii) substituting deuterium for protium (Gant, 2014; Pirali, Serafini, Cargnin, & Genazzani, 2019; Stringer et al., 2014), or (iii) adding a “soft” site elsewhere in the molecule that “steer” oxidation there. Of these, the first option has been the most useful.

Variations in pharmacokinetics

In an ideal world, a new drug would have the same metabolites, half-life, and clearance in all individuals, and prescriptions would be easy to develop. However, there are several reasons for variable pharmacokinetics.

One issue is genetic inter-individual variability, i.e. genetic differences in the P450 enzymes. This issue is discussed in detail in another chapter (Daly: another chapter, cite in proof).

Other issues involve changes due to enzyme induction and inhibition. These can be due to the drug itself or to other drugs, or even chemicals found in foods (e.g., grapefruit) or societal habits (smoking, alcohol). When induction and inhibition are associated with the drug itself, the pharmacokinetics of the drug can be expected to change with time, even in the absence of other drugs.

Drug-drug interactions

Drug-drug interactions are an important issue and account for both a sizeable fraction of hospitalizations and hospital deaths (Montané, Arellano, Sanz, Roca, & Farre, 2018). These problems are seen with many diseases and therapeutic areas (Fig. 8A) (Yu, Zhou, Tay-Sontheimer, Levy, & Ragueneau-Majlessi, 2018). A large fraction of the pharmacokinetic drug-drug interactions are seen with P450 3A(4) and some of the drug transporters (Fig. 8B).

Figure 8.

Frequency of new molecular entities (NMEs, i.e. new drug candidates) in inhibition-based drug-drug interactions (DDIs) with drugs approved by the Food and Drug Administration (FDA) in the United States between 2013 and 2016 (Yu et al., 2018). A, Grouping by therapeutic class. B, Grouping by enzymes involved. Pgp and OAT1B1 are transporters. COMT: catechol O-methyl transferase.

The complexity of drug metabolism makes it hard to totally avoid drug-drug interactions and some other toxicity problems, as exemplified in the metabolism of phenacetin and acetaminophen (Fig. 9).

Figure 9.

Roles of P450s in the bioactivation and detoxication of chemicals: the complex example of phenacetin (Guengerich, 2019a). Acetaminophen (paracetamol, Tylenol ^®) is widely used as an analgesic, safe at low doses and hepatotoxic at high levels (S. S. T. Lee, Buters, Pineau, Fernandez-Salguero, & Gonzalez, 1996). Phenacetin has been classified as a carcinogen and withdrawn from use. The metabolism of acetaminophen has been investigated in detail (Dahlin, Miwa, Lu, & Nelson, 1984; Dahlin & Nelson, 1982; Guengerich, 2021a). Only in a few cases are the structures of the protein and DNA adducts known. Some of the indicated P450s have been identified in different species, including humans (Distlerath et al., 1985; S. S. T. Lee et al., 1996).

The analgesic phenacetin is no longer in use because it was associated with rat kidney cancers. It undergoes oxidation in several P450-dependent reactions, some of which can lead to the generation of reactive products that can covalently bind to proteins and DNA. The product of O-deethylation is acetaminophen, a drug used extensively for fever and pain. Acetaminophen is used therapeutically at least once per week by ~ 23% of the US population (Larson et al., 2005), with benefit. However, it is also involved in ½ of the cases of drug-induced liver failure.

The metabolism of phenacetin is induced (at least P450 1A2, the O-deethylase) by polycyclic hydrocarbons and other P450 Family 1 inducers (AhR agonists) (Conney et al., 1976; Pantuck et al., 1974). In humans, the oxidation of acetaminophen to a potentially toxic iminoquinone is catalyzed mainly by three P450s—2E1, 1A2, and 3A4 (Patten et al., 1993). P450 2E1 appears to dominate, and induction of P450 2E1 by ethanol is generally considered to be the basis of enhanced hepatotoxicity of acetaminophen in alcoholics (Lee & Kaplowitz, 2021).

Drug-drug interactions can either render a drug ineffective or exaggerate the pharmacology and make it toxic.

Induction

The most common problem with induction is the loss of drug efficacy due to enhanced metabolism of a drug. A classical example involves the induction of P450 3A4 by rifampicin or barbiturates and the ineffectiveness of oral contraceptives due to enhanced clearance of 17α-ethynylestradiol (Bolt, Bolt, & Kappus, 1977; Guengerich, 1988). This phenomenon continues to occur with other barbiturates (Wilbur & Ensom, 2000) and it is also seen with some herbal medicines (Hall et al., 2003), in that St. John’s wort contains a potential PXR inducer, hyperforin (Moore et al., 2000).

P450 inducers not only pose problems in the clinic but are also issues in experimental animals in the process of safety assessment. Some chemicals induce animal P450s and can confound pre-clinical pharmacokinetic studies or lead to toxicity problems. Even though the issues may not be relevant to humans, those issues need to be explained to regulatory agencies, and the testing may waste valuable resources. Moreover, some animal tumors are seen with certain modes of induction (e.g., PPARα), even if they are not recognized as being relevant to human medical situations. Overall, it is generally desirable to advance a lead drug that is not an inducer, it there is a choice and other factors are equal.

Inhibition

Modes of inhibition

Inhibition is a very important factor in pharmacokinetic drug-drug interactions. The subject has been treated extensively elsewhere, and only a brief treatise will be provided here.

A simple way of dividing P450 inhibitors is among (i) reversible inhibitors, (ii) quasi-reversible inhibitors, and (iii) irreversible inhibitors. Reversible inhibition is the most straight-forward situation. It follows the basic schemes generally taught in introductory biochemistry, i.e. competitive, non-competitive, uncompetitive, and mixed inhibition. In the simplest cases, two drugs are bound to the enzyme at either the same or at different sites. Reversible inhibition can be detected quickly with high throughput screening. The mechanisms may be more complicated than just simple competition for an active site, in that multiple ligand occupancy is possible for some P450s such as 3A4 (Ekroos & Sjögren, 2006). Other complex phenomena have been reported in our own laboratory (Bojić et al., 2014; Shinkyo & Guengerich, 2011).

Time-dependent inhibition

The reversible inhibition modes mentioned above occur quickly and are reversed as the inhibitor is removed, most generally by metabolism (sometimes by the same P450 enzyme) (Guengerich, 2019b). There are several modes of time-dependent inhibition, in which the catalytic activity of the P450 is involved in exacerbating the inhibition. Time-dependent inhibition can be divided into three general modes.

1. (i) Formation of quasi-reversible complexes. The most prominent examples are the oxidation of amines (primary) to C-nitroso compounds and of methylenedioxyphenyls to carbenes. These unstable products bind very tightly to ferrous P450 (Fe²⁺), yielding complexes with absorption maxima at 455 nm. The binding is very tight. A relevant example is troleandomycin (Pessayre et al., 1983). The rates of breakdown of the complex is slow and takes more than a day in vivo (Delaforge, Jaouen, & Mansuy, 1984).

2. (ii) Generation of reactive metabolites that bind to P450 covalently to inhibit. The list includes quinones, quinone imines, epoxides, and other species. These are chemical entities that bind irreversibly to P450s (and possibly to other proteins, including other P450s) (Fig. 9).

3. (iii) True mechanism-based inhibitors, which are compounds that are oxidized to enzyme intermediates (usually species with radical or carbocationic character) that react with groups in the protein—or the heme prosthetic group—to inactivate. Covalent products are formed.

In all three cases the inhibition develops with time, in the presence of POR, NADPH, and oxygen.

Drug candidates that behave in this way can be identified using in vitro screening paradigms, although they are more complex than for single reversible inhibitors and not as adaptable to high-throughput screening. With all three modes of quasi-reversible and irreversible inhibition, the nature of the kinetic constants is more complex than for simple reversible inhibition, and the most valuable parameters are k_inactivation (the maximum rate of enzyme inactivation at infinite inhibitor concentration) and K_i, the inhibitor concentration at which half the maximal rate of enzyme inhibition is obtained. In contrast to a K_m value, a K_i in this case is actually a parameter that reflects the dissociation constant (K_d) for the enzyme-inhibitor complex (before activation) (Johnson, 2019; Silverman, 1995).

Use of inhibitors to slow drug metabolism

Although the focus is usually on avoiding drug-drug interactions due to inhibition, there are situations in which inhibition of P450 metabolism is desirable. In some cases achieving metabolic stability of drugs is difficult, and the production of complex drugs (especially some natural products) may be expensive, e.g. cyclosporin. This has been an issue with some HIV therapies.

An anecdotal approach to slowing gut metabolism by P450 3A4 (and perhaps some other enzymes) is by ingestion of grapefruit juice (Bailey, Edgar, Spence, Munzo, & Arnold, 1990). The problem is that the content of bergamottin can vary, so the control of exactly how much inhibition occurs may be problematic. The viral protease inhibitor ritonavir is also a potent inhibitor of P450 3A4 (Greenblatt & Harmatz, 2015) and has been used as a “booster” in prescribing information with some P450 3A4 substrates. This is more powerful and, unlike grapefruit juice, inhibits hepatic P450 3A4 as well as intestinal. Rational design methods have been used with P450 3A4 crystal structures to pursue ritonavir analogs for this purpose (Sevrioukova & Poulos, 2014). The drug cobicistat, resembling ritonavir, was developed as a P450 3A4 inhibitor and is FDA approved as a booster drug.

Clinical issues

There are numerous examples of P450-based drug-drug interactions. One that is now classic is terfenadine (Fig. 10), the first non-sedating antihistamine on the market. More than 100 million people used the drug (as Seldane^®), and for most the drug was very helpful (Guengerich, 2014; Thompson & Oster, 1996). However, deaths were reported beginning in 1990, due to cardiac arrythmia, and eventually at lest 25 deaths were attributed to terfenadine (some estimates as high as 125 or more). A major factor was the use of ketoconazole or erythromycin at the same time. Tefenadine is oxidized mainly by P450 3A4 (Yun, Okerholm, & Guengerich, 1993), a fact which was unknown when the problems began. Normally terfenadine is oxidized rapidly and not found in the plasma; the product fexofenadine (which also has pharmacological activity) is the circulating active drug (Fig. 10). Blocking the metabolism of terfenadine causes it to accumulate, and it binds tightly to the hERG potassium channel receptor and causes torsades de pointes (long QT intervals) and leads to arrhythmias.

Figure 10.

Metabolism of terfenadine (Guengerich, 2014; D. Thompson & Oster, 1996; Yun et al., 1993). All steps are catalyzed primarily by P450 3A4. Oxidations of the antihistamine terfenadine catalyzed by P450 3A4. The oxidation of terfenadine was attenuated in individuals who have inherently low levels of P450 3A4 (Yang et al., 2010) or used P450 3A4 inhibitors (e.g., erthyromycin, ketoconazole) concomitantly with terfenadine (Guengerich, 2014; Yun et al., 1993).

With the development of knowledge about human P450 enzymes, it is now fairly routine to establish which P450(s) is involved in the metabolism of a drug (Fig. 7), and regulatory agencies have this expectation at the IND stage (filing of an “Investigational New Drug” application at the U. S. FDA or the equivalent elsewhere, when a new drug candidate is first administered to humans). Tables of known inhibitors of the major human P450s are available (Table 2), and predictions can be made about which drug interactions might be problematic.

Table 2.

Inhibitors of major P450s

1A2	2C9	2C19	2D6	3A4
Amiodarone	Amiodarone	Chloramphenicol	Amiodarone	Amiodarone
Cimetidine	Capecitabine	Cimetidine	Bupropion	Aprepitant
Ciprofloxacin	Clopidogrel	Citalopram	Celecoxib	Atomoxetine
Citalopram	Crisaborole	Esomeprazole	Chlorpheniramine	Boceprevir
Crisaborole	Efavirenz	Felbamate	Chlorpromazine	Chloramphenicol
Efavirenz	Fenofibrate	Fluoxetine	Cimetidine	Cimetidine
Fluoroquinolone	Fluconazole	Fluvoxamine	Cinacalcet	Ciprofloxacin
Fluvoxamine	Fluvastatin	Indomethacin	Citalopram	Clarithromycin
Furafylline	Fluvoxamine	Isoniazid	Clemastine	Delaviridine
Interferon	Isoniazid	Ketoconazole	Clomipramine	Diethyldithiocarbamate
Methoxsalen	Lovastatin	Lansopraxole	Cocaine	Diltiazem
Mibefradil	Metronidazole	Modafinil	Diphenhydramine	Erythromycin
Ribociclib	Paroxetine	Omeprazole	Doxepin	Esomeprazole
Rucaparib	Phenylbutazone	Oxcarbazepine	Doxorubicin	Fluconazole
Ticlopidine	Probenicid	Pantoprazole	Duloxetine	Fluvoxamine
	Rucaparib	Probenicid	Escitalopram	Gestodene
	Sertraline	Rucaparib	Fluoxetine	Grapefruit juice
	Sulfamethoxazole	Ticlopidine	Halofantrine	Idelalisib
	Sulfaphenazole	Ropiramate	Haloperidol	Imatinib
	Teniposide	Voriconazole	Hydroxyzine	Indinavir
	Voriconazole		Levomepromzaine	Itraconazole
	Zafirlukast		Methadone	Ketoconazole
			Metoclopramide	Lesinurad
			Mibefradil	Mibefradil
			Midodrine	Mifepristone
			Moclobemide	Nefazodone
			Palonosetron	Nelfinavir
			Panobinostat	Netupitant
			Paroxetine	Norfloxacin
			Perphenazine	Norfluoxetine
			Promethazine	Omeprazole
			Quinidine	Pantoprazole
			Ranitidine	Regorafenib
			Riclopidine	Ribociclib
			Ritonavir	Ritonavir
			Rolapitant	Saquinavir
			Rucaparib	Starfruit
			Sertraline	Telaprevir
			Terbinafine	Telithromycin
			Tripelennamine	Verapamil
				Voriconazole

Modified from Flockhart DA. Drug Interactions: Cytochrome P450 Drug Interaction Table. Indiana University School of Medicine (2007). “https//drug-interactions.medicine.iu.edu” Accessed 19 August 2021

Some P450 substrates are more sensitive to inhibition (by other drugs), and some have narrow therapeutic windows (Table 3), e.g. warfarin. If a new drug would be likely to be given in combination with one of these drugs, then a regulatory agency might well require a clinical interaction study.

Table 3.

Examples of sensitive in vivo P450 substrates and P450 substrates with narrow therapeutic range

P450 Enzymes	Sensitive substrates	Substrates with narrow therapeutic range
1A2	Alosetron, caffeine, duloxetine, melatonin, ramelteon, tacrin, tizanidine	Theophylline, tizanidine
2B6	Bupropion, efavirenz
2C8	Repaglinide	Paclitaxel
2C9	Celecoxib	Warfarin, phenytoin
2C19	Clobazam, lansoprazole, omeprazole, (S)-mephenytoin	(S)-Mephenytoin
3A (4,5)	Alfentanil, aprepitant, budesonide, buspirone, conivaptan, darifenacin, darunavir, dasatinib, dronedarone, eletriptan, eplerenone, everolimus, felodipine, indinavir, fluticasone, lopinavir, lovastatin, lurasidone, maraviroc, midazolam, nisoldipine, quetiapine, saquinavir, sildenafil, simvastatin, sirolimus, tolvaptan, tipranair, triazolam, ticagrelor, vardenafil	Alfentanil, astemizole, cisapride, cyclosporine, dihydroergotamine, ergotamine, fentanyl, pimozide, quinidine, sirolimus, tacrolimus, terfenadine
2D6	Atomoxetine, desipramine, dextromethorphan, metoprolol, nebivolol, perphenazine, tolterodine, venlafaxsine	Thioridazine, pimozide

The US FDA has ranges for the effects of inhibitors. The most general approach is to compare the pharmacokinetic “area-under-the-curve” (AUC) without and with inhibitor (AUC_R). If this value is in the range of 1.25–2.0, then the inhibition is considered “weak” and not expected to be a problem. A value of AUC_R in the range of 2–5 is considered “moderate,” and a value > 5 is “strong.” The latter two groups (AUC_R > 2) may require labeling in the form of a contraindication warning.

As mentioned earlier, one of the troublesome issues in drug development has been time-dependent inhibition, particularly with P450 3A4 substrates (Fig. 7) (Zimmerlin, Trunzer, & Faller, 2011). Eng et al. (Eng, Tseng, Cerny, Goosen, & Obach, 2021) have compared a large series of P450 3A4 substrates for in vitro inhibition parameters with clinical AUC_R values (Fig. 11).

Figure 11.

Boundary line for k_obs for time-dependent inhibition and relation to in vivo drug-drug interactions (DDI) (Eng et al., 2021). A, Fifty drugs were evaluated for P450 3A4 time-dependent inhibition in human liver microsomes (at 30 μM unless noted otherwise) and ranked by k_obs, the first-order rate of inactivation, as judged using midazolam 1´-hydroxylation (○), presented on a log₁₀ scale (right y-axis). The filled bars show the in vivo drug-drug interactions as judged by the AUC_R (AUC with the drug divided by the AUC without the drug, Clinical DDI magnitude). B, The study in Part A was repeated in human hepatocytes. The stippled line indicates a 2-fold in vivo difference. Also indicated are p < 0.05 statistical limits and a k_obs “boundary” of the lowest in vitro value with 2-fold in vivo difference. Reprinted from Drug Metab. Dispos., Vol. 49, Eng, H., Tseng, E., Cerny, M. A., Goosen, T. C. and Obach, R. S., Cytochrome P450 3A time-dependent inhibition assays are too sensitive for identification of drugs causing clinically significant drug-drug interactions: a comparison of human liver microsomes and hepatocytes and definition of boundaries for inactivation rate constants, pages 442–450, Copyright (2021), with permission from the American Society for Pharmacology and Experimental Therapeutics.

Several points are of note. (i) Many drugs show time-dependent inhibition, even drugs that have been on the market for some time, often without major issues. (ii) The assays with hepatocytes showed less-time-dependent inhibition (Fig. 11B) than the assays with microsomes (Fig. 11A). (iii) There is a rough correlation between in vitro rates of time-dependent inhibition of P450 3A4 and AUC_R but there are many outliers. (iv) The structures of many of the drugs that show time-dependent inhibition are not obvious as to why they should be (i.e., lack of acetylenes, cyclopropylamines, etc.). The results shown in Fig. 11 indicate that in vitro assays of time-dependent inhibition are still not completely reliable in predicting whether drug-drug interactions will be a problem. In the analysis of Novartis compounds by Zimmerlin et al. (Zimmerlin et al., 2011), it was noted that the incidence of time-dependent P450 3A4 inhibition—and of strong reversible inhibition—was much lower in drugs that reached the market than in drug candidates, indicative of the liability of time-dependent inhibition (Zimmerlin et al., 2011).

Toxicity issues

Slow metabolism

Rates of drug clearance are generally developed for the majority of patients, and it is the individuals who have unusually slow metabolism who are at the most risk, generally due to either genetic reasons (see other chapters in this treatise) or inhibition (vide supra). If an individual with slow metabolism is given the usual dose, then the buildup of drug can (i) yield an exaggerated pharmacological response of the normal target or (ii) alternatively, result in metabolism at a secondary site to produce a toxic product (Fig. 9). Either action can result in toxicity to the patient. For these reasons it is important to define the variability of pharmacokinetic parameters in clinical trials.

Bioactivation

The matter of the potential for generation of reactive metabolites has already been considered in regard to P450 inhibition, but the general issue of bioactivation involves the generation of products that can modify other proteins or nucleic acids to cause toxicity. As indicated with phenacetin and acetaminophen (Fig. 9), there is usually a balance of detoxication and bioactivation reactions occurring, and the net result determines whether a chemical is toxic or not—as well as the dose, of course. Some chemical moieties are more likely to cause problems. There are called “toxicophores,” and the list includes hydrazines and hydrazides, aryl acetic and aryl propionic acids, thiophenes, furans, pyrroles, anilines and anilides, quinones and quinone imines, medium chain fatty acids, halogenated hydrocarbons and some halogenated aromatics, nitroaromatics, thiols, thiono compounds and thiazolidinediones, and moieties that form α,β-unsaturated enol-like compounds (Michael acceptors) (Guengerich, 2021c). However, in some therapeutic programs one (or more) of these entities may be needed for the desired pharmacological activity. Also, there are numerous exceptions—e.g., atorvastatin (Lipitor^®), which has a masked aniline present but was the best-selling drug in the world for several years.

The finding of toxicity late in a drug development program is highly problematic and expensive, if a drug candidate must be dropped after spending considerable resources. Thus, there is considerable interest in reliably identifying bioactivation (and any other toxicity) issues early in the drug development process. The systems differ among pharmaceutical companies but the main elements follow.

1. (i) In silico screening. The most useful systems to date involve genotoxicity. The screens are based largely upon literature bases of results plus structural similarity. Tissue-selective toxicities are more complex and mechanisms are generally not well-described, so these are far less reliable.

2. (ii) Covalent binding screens. The extent of binding of drug candidates in vitro (microsomal and hepatocyte systems) and in vivo was proposed and used extensively in some pharmaceutical companies (Evans, Watt, Nicoll-Griffith, & Baillie, 2004). There was never a cut-off parameter but the approach was used to stratify compounds in making decisions as to which to advance. An issue is that compounds must be synthesized with radiolabels early in the process. There are correlations between the level of covalent binding and toxicity but many exceptions, even when accounting for the dose and body burden (Bauman et al., 2009; Dahal, Obach, & Gilbert, 2013; Nakayama et al., 2009; Obach, Kalgutkar, Soglia, & Zhao, 2008; Takakusa et al., 2008; Thompson et al., 2012).

3. (iii) General cellular toxicity. In some programs, toxicity is measured in cells in culture, particularly in some types of hepatocytes. However, all liver cells in culture have some type of metabolic deficiencies (regarding P450 expression). Even more problematic is the difficulty in using cellular assays to predict what will occur in a tissue.

4. (iv) Measurement of selective biomarkers. Some assays of interest include gross cell toxicity, mitochondrial toxicity, and inhibition of bile salt exporter protein (BSEP) and the transporter MRP2. Some pharmaceutical companies use a battery of assays, including covalent binding, in order to stratify their new chemical entities (Monroe et al., 2020; Thompson et al., 2012).

5. (v) Genotoxicity is a particular type of toxicity, and some of the assays are very well developed. The Ames bacterial mutagenicity tests (Ames, Durston, Yamasaki, & Lee, 1973) have been used extensively for 50 years and are almost universally used as the primary screen for genotoxicity, which is an indicator not only for potential carcinogenicity but also other maladies drive by mutation (e.g., teratogenesis). Bacterial mutation results are followed up by mammalian mutagenicity.

Human specific metabolites

Humans and experimental animals have different P450s, even if they are similar in their primary sequences and structures. One of the problems is human-specific (or “disproportionate”) metabolites and the MIST issue (Metabolites in Safety Testing) (Guengerich, 2006). The U.S. FDA and the International Commission on Harmonisation of Technical Requirements for Pharmaceuticals for Human Use (ICH) have agreed that if a metabolite is present at >10% of the in vivo metabolites of a drug, then it should be tested for safety in animals if the animal test species do not produce it (at the human level of exposure). The analysis generally involves the (in vivo) AUC, which is the most appropriate measure of the exposure of an animal or human to a chemical. Further, regulatory agencies are interested in having multiples (AUC) of exposure in animals compared to humans, in order to have more assurance of drug safety, particularly in vulnerable populations.

There are several relevant strategies to deal with the MIST requirements. One point is that in vitro comparisons of human and animal drug metabolism should be done very early in order to make decisions about which animal species in the most appropriate model. Another important early investigation is a human “mass balance” study (with radioactively-tagged drug) to be sure that a large fraction of the metabolites are accounted for.

If there are human-specific metabolites found, there are a few options. (i) If any of the disproportionate human metabolite is formed at all in a test animal, then it may be possible to increase the dose in order to produce the exposure (AUC) that is found in humans (if the drug is not toxic to the animals at the high dose). (ii) Transgenic (“humanized”) mice, expressing human P450s or other enzymes, may be used to produce the metabolite. (iii) The most straight-forward approach may be to synthesize the metabolite and administer it to animals, in order to achieve exposure. The synthesis might be difficult (e.g., with a macrolide antibiotic). If long-term testing is in order (e.g. cancer bioassay) then considerable time (and resources) may be required.

Human differences in regulation

One of the problems with the use of animals in risk assessment (or, more properly, risk characterization) is that humans and animals have differences in their regulation of P450s and other genes. As mentioned earlier, this variation makes P450 induction studies in animals less than predictive for humans. Accordingly, the most generally accepted induction assays are done with human hepatocytes. Although assays can be done with reporter constructs in other cells, there are issues regarding the need for co-activitors etc. (e.g., HNF-4α with PXR and P450 3A4) (Tirona et al., 2003).

Another issue is that the levels of some of the receptors and their actions are very different in animals. In particular, PPARα agonists can lead to liver tumors in rodents, as do barbiturates in rodents (Rao & Reddy, 1987). In the past, these were considered serious issues but are not today.

P450s as drug targets

Much of this chapter has dealt with concerns about avoiding P450 inhibition by drugs (with the exception of the P450 3A4 “booster” drugs). However, in several cases there are P450s that function in normal physiological processes but inhibition can be used therapeutically.

Current P450 inhibitors in use

Drugs are approved for the inhibition of at least four human P450 targets, namely 5A1, 11B1, 17A1, and 19A1 (Table 4).

Table 4.

P450s as drug targets

Currently in clinical practice

P450 5A1 (anti-platelet drugs, inhibit thromboxane production)

Pictamide

Riogrel

Ozagrel

Furegrelate

P450 11B1 (Cushing’s disease)

Mifepristone

P450 17A1 (prostate cancer)

Abiraterone

P450 19A1 (breast and other hormonal cancers)

Exemestane

Anastrozole

Letrozole

P450 51 (anti-fungal, inhibit fungal P450s)

Ketoconazole

Fluconazole

Itraconazole

Vorconazole

Posaconazole

Isavuconazole

Mifepristone

Discovery and development programs

P450 4A11 (hypertension)

P450 11A1 (prostate cancer)

P450 11B2 (hypertension)

P450 24A1 (increase vitamin D3 levels)

P450 26A1 (increase vitamin A levels)

P450 26B1 (increase vitamin A levels)

P450 5A1 is known by its more common name, thromboxane synthase. This is an unusual P450 that does not require electrons or oxygen; it rearranges prostaglandin H₂, an endoperoxide, to generate thromboxane. The inhibitors (including aspirin) are used to inhibit platelet aggregationin a variety of cardiovascular situations.

Mifepristone (RU-486) is approved for use in inhibiting P450 11B1 for treating Cushing’s Disease (Chu et al., 2001).

In castration-resistant prostate cancer, reducing the androgen load is an issue. The only approved drug for inhibiting P450 17A1, the androgen-synthesizing P450, is abiraterone acetate, a pro-drug ester that is cleaved to release abiraterone (Fig. 12A). Although there has been some speculation about how abiraterone works (Cheong et al., 2020), it appears to be simply a very tight-binding direct competitive inhibitor of P450 17A1 (K_d ~ 1 nM), the steroid 17α-hydroxylase/lyase (Guengerich, McCarty, Chapman, & Tateisihi, 2021). One long-standing goal with this P450 is the selective inhibition of the lyase reaction to block androgen production but not the synthesis of glucocorticoids, i.e. 17α-hydroxylation (Bird & Abbott, 2016; Burris-Hiday & Scott, 2021; Guengerich, McCarty, et al., 2021).

Figure 12.

Some multi-step steroid biosynthetic reactions catalyzed by human P450s that are targets for drugs. A, P450 17A1; B, P450 19A1.

P450 19A1, the steroid aromatase, is involved in estrogen synthesis (Fig. 12B), which is important in cancers of the breast, ovary, and uterus. At least three (third-generation) aromatase inhibitors have been successful and are in current use (Table 4). These are very tight-binding (K_i in low nM range). The major side effects are related to changes in calcium homeostasis, which is an inherent physiological response. Thus, it will probably be difficult to improve on aromatase inhibitors in the future.

Future prospects for P450 inhibition

Several other human P450s have been considered in the context of inhibition in relationship to other disease (Table 4). Of these, the most viable today may be P450 11B2. P450 11B2 produces aldosterone, a target in some forms of hypertension (Fig. 13).

Figure 13.

P450 11B2 oxidation of 11-deoxycorticosterone to aldosterone, a drug target (Hu, Yin, & Hartmann, 2014). Reprinted from J. Med. Chem., Vol.67, Hu, Q., Yin, L., & Hartmann, R. W. (2014). Aldosterone synthase inhibitors as promising treatments for mineralocorticoid dependent cardiovascular and renal diseases, pages 5011–5022, Copyright (2014), with permission from the American Chemical Society.

Although not listed in Table 4, there have been considerations given to the use of P450 inhibitors in cancer prevention, as a means of blocking the bioactivation of chemical carcinogens (Chun, Kim, Kim, Lee, & Guengerich, 2001). However, the approach is difficult in that many of the chemical carcinogens undergo both bioactivation and detoxication by P450s, sometimes even the same P450 (Ueng, Shimada, Yamazaki, & Guengerich, 1995). One target has been P450 2A6, with the goal of blocking the metabolism, of nicotine to decrease the desire to smoke more (Yano et al., 2006).

Pest control

P450s in microorganisms have also been drug targets. In particular, P450 Family 51 are involved in the 14α-demethylation of lanosterol and the equivalent sterol precursors in several species, and the final sterols (e.g., ergosterol) are needed for membrane integrity in these microorganisms. This is an important target not only for local conditions with troublesome yeasts (e.g., athlete’s foot) but serious systemic infections. In addition, CYP51 Family enzymes are targets in serious tropical diseases such as leishmania and sleeping sickness, which are endemic in parts of the world (Emami, Tavangar, & Keighobadi, 2017; Friggeri et al., 2018; Hargrove et al., 2017).

The complexity and difficult of developing better inhibitors is exemplified in the case of the antifungal posaconazole (Fig. 14A). During discovery and development his drug showed considerably better activity than a previous lead (SCH 51048) in several fungal species. The structure of the Candida albicans P450 51A enzyme crystal structure bound to posaconazole was solved later (Hargrove et al., 2017), after the drug entered the market (Fig. 14B). Although posaconazole had much better intrinsic anti-fungal activity than SCH 51048, the only difference is a hydroxyl on the side chain (Fig. 14A). However, in the crystal structure (Fig. 14B) the moiety containing the hydroxyl is positioned outside of the protein. Thus, rational design using SCH 51048 would almost certainly not have led to a decision to make the molecule now known as posaconazole.

Figure 14.

Posaconazole bound in the C. albicans CYP51 active site (Hargrove et al., 2017). A, An early lead compound in the program (SCH 51048) and posaconazole. B, X-ray crystal structure of C. albicans P450 51A1 bound to posaconazole. The arrow in Part B is pointed to the extra hydroxyl group in posconazole.

Targeting accessory enzymes

P450s use several accessory enzymes. The microsomal P450s use POR and sometimes b₅. The seven mitochondrial P450s use Adx and NADPH-Adx reductase, although this latter enzyme does not interact directly with P450s (Lambeth, Seybert, Lancaster, Salerno, & Kamin, 1982). Some drugs have been designed to block allosteric interactions (Busby et al., 2020; Sawyer, 2020), and one effort to screen inhibitors that selectively block interactions with individual P450s has appeared (Jensen et al., 2021) (see also Kim, Kim, McCarty, & Guengerich, 2021).

The future of P450 research

Predicting the future is always difficult. Following are some of my own thoughts; I realize that others may have different ones.

Recent developments

As alluded to earlier, there are extensive efforts to utilize artificial intelligence to better predict both metabolism and toxicology (Xiong et al., 2021). A nagging problem with high-throughput screening efforts has been the need to incorporate enzymes that will mimic metabolism in human liver.

The ability to use transgenic animals has developed. With the advent of CRISPR-Cas9 and newer gene technologies, it is possible to use transgenic rats (and other species) (Yasuda et al., 2021), not only mice. There are other approaches to “humanizing” mouse livers (Yamazaki, Suemizu, Mitsui, Shimizu, & Guengerich, 2016), although these animals are generally not as viable as wild-type mice. Rats offer a number of advantages over mice, particularly when addressing some questions.

There are also efforts to produce cell lines and model organelles that will better reflect mammalian physiology and human metabolism in vitro (Janssen et al., 2020; Park, Georgescu, & Huh, 2019; Underhill & Khetani, 2018).

Questions regarding basic research

Six questions are listed (Table 5). Some success has been achieved, although these are difficult topics and considerable resources have been spent. Most of these topics have already been discussed in the chapter and will not be elaborated further. It is noteworthy that the b₅ effect was first reported 50 years ago and many details still remain to be addressed.

Table 5.

Basic questions about P450 to be answered

• Is Compound I the only mechanism?

• How many conformational states exist & how do they relate to ligand recognition?

• Accessory enzymes: structures of binary complexes?

• Structures of the rest (32 more) of the human P450s

• Functions of the orphans (& quasi-orphans)?

• How does “allosteric” regulation work (including b5)?

Practical questions to be addressed

A list of questions regarding practical P450 issues is also presented (Table 6). The second item (predicting rates of metabolism) may never happen. The remainder will be commented on.

Table 6.

Practical questions about P450 to be addressed

• Predicting sites of metabolism

• Predicting rates a priori

• Predicting functions of SNVs

• Do SNVs affect disease incidence?

○ Cardiovascular

○ Hypertension

○ Cancer

• Can we use SNV data better in clinical practice?

• Better drugs for pests

• Veterinary applications

We know now that we are not dealing with only 57 human P450s—each has many SNVs and the list will grow as more human genomic data becomes available. Is it even realistic to try to express all of the SNVs and measure their functions in vitro? Are there prospects for using artificial intelligence? Is structural biology realistic? To date the only P450 SNV structures are of P450 SNV structures are of P450 2C9 variants (Parikh et al., 2020). The problem of understanding SNV effects is seen in our own work with P450 21A2 (Fig. 15) (Wang et al., 2017). The changes tend to be in certain regions. However, a single SNV can change the catalytic specificity constant (k_cat/K_m) a million-fold. However, crystallizing these mutants has been difficult, and several are even hard to express. Moreover, the Eyring equation

Figure 15.

P450 21A2 variants. Amino acid changes which give rise to the (A) salt-wasting (SW), (B) simple virile (SV), and (C) non-classical (NC) congenital adrenal hyperplasia phenotypes are mapped in the crystal structure of P450 21A2 (Pallan et al., 2015). Carbon atoms of wild type (*1) residues are highlighted in blue (SW), green (SV), and purple (NC). Reprinted from Molecular Endocrinology, Vol. 29, Pallan, P. S., Lei, L., Wang, C., Waterman, M. R., Guengerich, F. P., & Egli, M. (2015). Research Resource: Correlating human cytochrome P450 21A2 crystal structure and phenotypes of mutations in congenital adrenal hyperplasia, pages 1375–1384, Copyright (2015), with permission from The Endocrine Society.

k_obs = (k_BT/h) e^−ΔG‡/RT

(where k_B is the Boltzman contant, h is Planck’s constant, T is the absolute temperature, and R is the gas constant) indicates that a 10-fold change in activity only relates to a free energy (ΔΔG) change of 1.3 kcal/mol, less than one hydrogen bond. This makes the task of understanding the structural basis of a functional change difficult. Further, a coding region SNV can show different effects with different substrates (or reactions of the same substrate).

Although the effects of SNVs in P450s with roles in steroid biochemistry have been rather obvious in terms of phenotypic endocrinological problems (Auchus & Miller, 2015), these SNV effects have been more subtle in diseases such as hypertension and other cardiovascular problems. Although chemical carcinogenesis was one of the early reasons for emphasis on the study of P450s and there was much interest in SNVs and molecular epidemiology of cancer (Kirk, Bah, & Montesano, 2006; Vineis & Perera, 2007), it is still not very clear what most of the effects are or how important they are.

At the turn of the century, there was considerable enthusiasm for genomic medicine. Today the number of examples of application of P450 SNV knowledge to clinical practice is still small (actually the information has been more useful in drug development). Can this be improved?

The opportunity to use P450 targets in pests still seems enormous. Anti-fungals were discussed but there are also opportunities with tuberculosis (McLean, Dunford, Neeli, Driscoll, & Munro, 2007; Ouellet, Lang, Couture, & Ortiz de Montellano, 2009) and other maladies involving infectious microorganisms.

Finally, there is still considerable opportunity in veterinary medicine. Many of the questions where we have answers about drug-drug interactions are just beginning to be asked in veterinary practice.

Conclusion

In a sense, the field of P450 has become a mature one. However, that also means that we have accumulated a large knowledge base and also that we have the tools to cut deeper and address harder questions. In retrospect, the application of biochemical findings to problems in pharmaceutical science has been a true scientific success story. There are still more opportunities.

Non-standard abbreviations:

Adx

adrenodoxin

AhR

aryl hydrocarbon receptor

AO

aldehyde oxidase

ARNT

aryl hydrocarbon receptor nuclear transferase

AUC

area-under-the-curve

b ₅

cytochrome b₅

CAR

constitutively active receptor

COMT

catechol O-methyl transferase

DDI

drug-drug interactions

EGFR

epidermal growth factor receptor

FDA

(United States) Food and Drug Administration

FMO

flavin-containing monooxygenase

HNF

hepatic nuclear factor

IND

Investigational New Drug (application)

K _d

dissociation constant

K _i

inhibition constant

K _m

Michaelis constant

LC-MS

combined liquid chromatography-mass spectrometry

MIST

metabolites in safety testing

NME

new molecular entity

NMR

nuclear magnetic resonance (spectroscopy)

NC

non-classical (congenital adrenal hyperplasia)

P450 or CYP

cytochrome P450

PDB

Protein Data Bank

Pgp

P-glycoprotein

POR

NADPH-cytochrome P450 oxidoreductase

PPAR

peroxisome proliferator-activated receptor

PXR

pregnane X receptor

RAR

retinoic acid receptor

RXR

retinoid X receptor

SNV

single nucleotide variant

SV

simple virile (congenital adrenal hyperplasia)

SW

salt-wasting (congenital adrenal hyperplasia)

TCPOBOP

1,4-bis[2-(3,5-dichloropyridyloxy)]benzene

UGT

uridine diphosphate glucuronosyl transferase