Characterization and Significance of Human P450s

Anthony Lu was involved not only in the first separations of P450s (Lu and Coon, 1968; Lu et al., 1969) but also in subsequent work on the development of procedures that led to the isolation of purified enzymes (Ryan et al., 1975; West et al., 1979). Our own laboratory was involved in the characterization of rat liver P450s in the late 1970s and then began to turn attention to the characterization of human P450s. At that time this effort was even more challenging than the purification efforts with P450s of experimental animals, primarily because of the difficulty in obtaining useful samples. The approaches used in our laboratory and others were focused on the purification of the most plentiful P450s to electrophoretic homogeneity, without initial regard to particular catalytic activities. In retrospect, the first P450s we purified were probably P450 2C9 and P450 3A4 (Wang et al., 1980; Wang et al., 1983). Further efforts utilized the strategy of monitoring particular catalytic activities during purification, particularly reactions suspected of showing genetic polymorphism (Distlerath et al., 1985; Shimada et al., 1986; Guengerich et al., 1986). A surprising result was the great variety of substrates found for what is now known as P450 3A4, originally purified as the nifedipine oxidase (Guengerich et al., 1986).

The early purifications are mentioned because they followed from previous work by Anthony Lu and others (Lu and Coon, 1968; Lu et al., 1969) and led to the use of other methods that have been used to characterize human P450s, e.g., use of antibodies, cDNA cloning, heterologous expression, identification of selective chemical inhibitors (Newton et al., 1994). In retrospect, a few major findings have formed the basis for how drug discovery and development use P450 systems today. One point is that a relatively small number of the human P450s (approximately 6) are responsible for approximately 90% of the oxidations of drugs and carcinogens (Guengerich, 1995). Indeed, the single P450 3A4 is involved in the oxidation of at least half of the drugs used today. Thus studying human drug metabolism is far simpler than it would have been if all of the approximately 40 P450s (Nelson et al., 1996) were involved in drug metabolism. Another point is that although total levels of hepatic P450 do not vary considerably among individuals, there are major variations in the levels of the different P450s (Guengerich, 1995), which may be controlled by genetics and environment. These points were not really known 20 years ago but today form the basis of why P450s are studied regarding issues of bioavailability, drug-drug interactions, and genotoxicity (Guengerich, 1995).

Applications of Bacterial Expression of Human P450s

Although the human P450 purification studies were important in the development of approaches to drug development and discovery, recombinant expression methods were necessary for the widespread application of the methods. Bacteria have many advantages, particularly high yields and low cost, and have been the focus of attention in our own laboratory. Our own strategies to improve expression are based on several methods of others (Barnes et al., 1991) and are reviewed elsewhere (Guengerich et al., 1996).

The above-referenced approaches allowed us to express and purify a number of different human P450s successfully (Guengerich et al., 1996), but we needed better methods of facilitating electron transport to P450s within bacteria. We considered two approaches: coupling to endogenous flavodoxins (Dong et al., 1996) and the use of P450:NADPH-P450 reductase fusion proteins (Chun et al., 1996). In view of limited success with both approaches, we developed another system, based on early work by Henry Barnes, in which each human P450 was co-expressed along with human NADPH-P450 reductase (Parikh et al., 1997). The bicistronic vector uses a single promoter and yields a single RNA that is translated to two proteins at similar concentrations. Catalytic activity can be measured in isolated bacterial membranes or in the bacterial cells (Parikh et al., 1997).

In earlier collaborative work with Prof. David Josephy, we had found that human P450 expression vectors (pCW-based) could be used inSalmonella typhimurium TA1538 to activate pro-mutagens in the absence of added proteins (Josephy et al., 1995). A new system was developed in which the bicistronic vectors (Parikh et al., 1997) were used with a lac-based Escherichia coli system (Josephy et al., 1998). The system also contains another plasmid to express bacterialN-acetyltransferase and improve sensitivity toward arylamines and heterocyclic amines. This system was shown to be more sensitive to 2-aminoanthrecene than the parent strain devoid of the recombinant P450 system but fortified with rat liver post-mitochondrial supernatant (Josephy et al., 1998). More systems of this type can be developed by introducing other P450s and by switching the F′ factors to permit analysis of all possible base-pair and several frameshift mutations (Cupples et al., 1990). The system offers not only sensitivity as an advantage but also more relevance in that human P450s are utilized in the assays. In another line of investigation, we have utilized this system to screen libraries of P450s in which the putative substrate-binding regions have been subjected to random mutagenesis. Thousands of mutants can be screened within a relatively short period of time, and we have found a number of altered P450s that have decreased and also increased catalytic activity.

Another use of the bicistronic expression systems involves work with human P450 1B1, an enzyme of interest because of its roles in estrogen and carcinogen metabolism (Hayes et al., 1996). Expression in E. coli was accomplished by removal of residues 2 through 4 (Gly, Thr, Ser) from the wild-type sequence, and the protein was purified to homogeneity by ion-exchange chromatography (Shimadaet al., 1998a). Human P450 1B1 was also expressed along with NADPH-P450 reductase in bacterial membranes, using the bicistronic vector system. The recombinant enzyme was shown to activate several carcinogens and to catalyze 17β-estradiol 4-hydroxylation in preference to 2-hydroxylation.

Human P450s 1A1, 1A2, and 1B1 show considerable overlap of catalytic selectivity. P450 1A2 is expressed essentially only in liver but P450s 1A1 and 1B1 are expressed in many extrahepatic tissues, particularly some of considerable interest regarding cancers (Shimada et al., 1996). In collaboration with Prof. William Alworth, we examined 25 polycyclic compounds as selective inhibitors of human P450s 1A1, 1A2, and 1B1 (Shimada et al., 1998b). Of these, many were very potent inhibitors, with 50% inhibitory concentration values ≤10 μM, and four were selective for P450 1A1 or 1B1 (fig. 1). Although some of the compounds have acetylenic moieties, the inhibition of P450 1B1 appears to be due more to a competitive than a mechanism-based component. These chemicals may prove to be of use in distinguishing between the contributions of P450s 1A1 and 1B1 in reactions measured in extrahepatic cells and microsomes.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Structures of P450 1A1 and 1B1 inhibitors. The estimated 50% inhibitory concentration (IC₅₀) values are indicated. 2-Ethynylpyrene andα-naphthoflavone preferentially inhibited P450 1B1, with 30 and 5 nM concentrations needed for 50% inhibition; 1-(1-propynyl)pyrene and 2-(1-propynyl)phenathrene preferentially inhibited P450 1A1 with respective IC₅₀ values of 3 and 2 nM.³

Rate-Limiting Aspects of P450 Reactions

Studies on the interactions of the components of the P450 system and rate-limiting aspects began soon after the initial separation of P450, NADPH-P450 reductase, and phospholipid components from microsomes (Miwa and Lu, 1981; Miwa and Lu, 1984; Miwa et al., 1978). Although the basic aspects of the catalytic mechanism are generally agreed upon, many questions remain about component interactions and rate-limiting steps in catalysis. Much of the difficulty lies in the differences seen among P450s and the variation seen even among individual reactions catalyzed by a single P450.

Of the individual steps in the generalized P450 reaction cycle (fig.2), several have the potential to be rate-limiting (at least in particular situations) including steps1, 2, 4, 7, and9 in the scheme. With rabbit P450 1A2 in the presence or absence of substrate, reduction (step 2) is fast, and the measured rate of reoxidation of ferrous P450 upon mixing with saturating O₂ is close to the steady-statek_cat. However, in the case of ethanol oxidation (to acetaldehyde) catalyzed by human P450 2E1, we have presented pre–steady-state kinetic evidence that product release, or some other step after product formation, must be rate-limiting in the steady-state reaction (Bell and Guengerich, 1997). Under some conditions (but not others) the rate of reduction of ferric P450 3A4 reduction may be rate-limiting (Guengerich and Johnson, 1997).

• Download figure
• Open in new tab
• Download powerpoint

Figure 2

Generalized catalytic cycle for P450 reactions (Guengerich, 1991).

The complexity of the problem is seen in comparisons of the behavior of P450s 2E1 and 3A4. With P450 2E1, almost all reactions show dependence on cytochrome b₅ and the role appears to be transfer of the second electron (step 4). Some P450 3A4 reactions are insensitive to cytochrome b₅but others are dependent; in some situations apo-cytochromeb₅ can replace cytochromeb₅ and even stimulate ferric P450 3A4 reduction (step 2). Reduction of ferric P450 2E1 is insensitive to the presence of substrate (Bell and Guengerich, 1997); P450 3A4 reduction is usually but not always highly dependent on the presence of substrate (Guengerich and Johnson, 1997). Product release is rate-limiting for some P450 2E1 reactions but not all (Bell and Guengerich, 1997).

A further complication with P450 3A4 is the interactions among ligands, both substrates and “effectors” (e.g.,α-napthoflavone) that can stimulate or inhibit activity. Also, some reactions show sigmoidal plots (v vs. S) and there are unexplained patterns of cross-inhibition (Wang et al., 1997). Several possible explanations have been proposed, all of which involve “multiple” ligand-binding sites of unknown spatial relationship. Another open question is whether one ligand influences the oxidation of another by simply imposing steric constraints or by influencing particular microscopic rate constants in the catalytic cycle (fig. 2).

Oxidation of Non-Ionic Detergents by P450s

One major advance in the purification of hepatic P450s was the use of non-ionic detergents in chromatography (Ryan et al., 1975). However, these detergents have also been known to be P450 inhibitors (Lu et al., 1974), presumably acting to break up protein component interactions.

In the course of studies on ligand cooperativity with P450 3A4, we found new fluorescent high-performance liquid chromatography peaks whose appearance was NADPH-dependent and independent of substrate (Hosea and Guengerich, 1998). We found that the same products could be formed with microsomes to which the non-ionic detergent Triton N-101 had been added, suggesting that this product was formed from trace detergent remaining from the purification procedure.

The oxidation of Triton N-101 was catalyzed by several P450s, in the order 3A4 > 1A2 > 2C9 (P450s 2E1, 1A1, and 2D6 were inactive). With P450 3A4, the K_m was approximately 10 μM and thek_cat approximately 3 min⁻¹. Several other alkyl phenylethyleneoxy detergents (Tritons, Emulgens, Tergitol NP-10) were oxidized at similar rates. The extensive inhibition of human liver microsomal oxidation by ketoconazole (1 μM) suggests that P450 3A4 is a major catalyst of the reaction. The addition of Triton N-101 to P450 yielded typical “Type I” binding spectra (dissociation constant of enzyme-substrate approximately 10 μM). The presence of Triton N-101 inhibited the Type I binding spectra produced by testosterone and also inhibited testosterone 6β-hydroxylation activity, with the results suggesting a mixed mode of inhibition. This inhibition is presumably due to a competitive component (blocking substrate binding) at low concentrations and disruption of protein component interactions at high concentrations (Hosea and Guengerich, 1998).

Commercial detergents are mixtures of oligomeric components, and the chemical analysis of oxidation products is not straightforward. Oxidation of the ethyleneoxy side chain yields hemiacetals, which rearrange to yield shortened detergents that would be expected to migrate with the substrate (Hosea and Guengerich, 1998). Analysis of the substrate and product by positive ion electrospray mass spectrometry indicated that all major ions in the substrate were shifted 16 amu to higher mass, indicative of the hydroxylation of the nonyl side chain (fig. 3). Further, collision analysis indicated that the major site(s) of hydroxylation is not in the terminal 4 carbons. Evidence against phenyl ring hydroxylation was obtained with the lack of pH effects on the ultraviolet and fluorescence spectra of the Triton N-101 product. Thus the conclusion is that the site(s) of hydroxylation is in the methylene region shown in fig. 3.

• Download figure
• Open in new tab
• Download powerpoint

Figure 3

Postulated region of hydroxylation of Triton N-101 by P450 3A4 (Hosea and Guengerich, 1998).

In one sense the demonstrated hydroxylation of non-ionic detergents is the characterization of an interesting artifact. The work does make the point that care must be taken in the use of P450s purified with non-ionic detergents, particularly P450 3A4. However, the work also expands the repertoire of P450 substrates. Further, alkyl phenylethers are of interest as environmental contaminants. These compounds are also used as spermicides (Chvapil et al., 1980) and in that sense can be considered in the context of drugs because of human exposure. In other recent work we have found that short linear peptides are high-affinity ligands for P450 3A4; the physiological relevance of this observation is under investigation.

Concluding Remarks and Future Prospects

In the past 20 years we have gone from very limited knowledge about a few of the P450s in experimental animals to the point where biochemical knowledge about many human P450s is as sophisticated as available for most intrinsic membrane proteins. This is the result of important studies by many individuals. The work has already had dramatic implications in the pharmaceutical industry, and the investments of universities, government agencies, and industry in this field have certainly been profitable.

What are some areas in which mammalian P450 research will advance in the future? Major basic problems include crystallization and solving structures of these intrinsic membranes’ proteins, as well as a better understanding of the details of gene regulation. An improved grasp of structure-function relationships would be helped by the availability of structures as well as insightful mutagenesis studies. As pointed out in this article, several nagging problems remain concerning aspects of catalysis and component interactions. Ultimately several of these areas bear on the problem of catalytic selectivity.

The practical aspects of P450 research are enormous, and Anthony Lu touched on these in his recent symposium address (Lu, 1998). P450 problems in drug discovery and development involve better screens for new drugs at several stages. Rapid and economical assays of the P450 status of individuals could also be used to tailor drug therapy and avoid interaction problems. The development of P450s in areas such as fine chemical synthesis and bioremediation has barely been explored.

The next 20 years should continue to be good for P450 research.