Interactions of mammalian cytochrome P450, NADPH-cytochrome P450 reductase, and cytochrome b(5) enzymes

An immobilized system was developed to detect interactions of human cytochromes P450 (P450) with the accessory proteins NADPH-P450 reductase and cytochrome b(5) (b(5)) using an enzyme-linked affinity approach. Purified enzymes were first bound to wells of a polystyrene plate, and biotinylated partner enzymes were added and bound. A streptavidin-peroxidase complex was added, and protein-protein binding was monitored by measuring peroxidase activity of the bound biotinylated proteins. In a model study, we examined protein-protein interactions of Pseudomonas putida putidaredoxin (Pdx) and putidaredoxin reductase (PdR). A linear relationship (r(2)=0.96) was observed for binding of PdR-biotin to immobilized Pdx compared with binding of Pdx-biotin to immobilized PdR (the estimated K(d) value for the Pdx.PdR complex was 0.054muM). Human P450 2A6 interacted strongly with NADPH-P450 reductase; the K(d) values (with the reductase) ranged between 0.005 and 0.1muM for P450s 2C19, 2D6, and 3A4. Relatively weak interaction was found between holo-b(5) or apo-b(5) (devoid of heme) with NADPH-P450 reductase. Among the rat, rabbit, and human P450 1A2 enzymes, the rat enzyme showed the tightest interaction with b(5), although no increases in 7-ethoxyresorufin O-deethylation activities were observed with any of the P450 1A2 enzymes. Human P450s 2A6, 2D6, 2E1, and 3A4 interacted well with b(5), with P450 3A4 yielding the lowest K(d) values followed by P450s 2A6 and 2D6. No appreciable increases in interaction between human P450s with b(5) or NADPH-P450 reductase were observed when typical substrates for the P450s were included. We also found that NADPH-P450 reductase did not cause changes in the P450.substrate K(d) values estimated from substrate-induced UV-visible spectral changes with rabbit P450 1A2 or human P450 2A6, 2D6, or 3A4. Collectively, the results show direct and tight interactions between P450 enzymes and the accessory proteins NADPH-P450 reductase and b(5), with different affinities, and that ligand binding to mammalian P450s did not lead to increased interaction between P450s and the reductase.     

Thermophilic cytochrome P450 enzymes

Thermophilic cytochrome P450 enzymes are of potential interest from structural, mechanistic, and biotechnological points of view. The structures and properties of two such enzymes, CYP119 and CYP175A1, have been investigated and provide the foundation for future work on thermophilic P450 enzymes.     

Rate-limiting steps in cytochrome P450 catalysis

Cytochrome P450 (P450) reactions are of interest because of their relevance to the oxidative metabolism of drugs, steroids, carcinogens, and other chemicals. One of the considerations about functional characterization is which steps of the catalytic cycle are rate-limiting. Detailed analysis indicates that several different steps can be rate-limiting with individual P450 reactions. N-Dealkylation of para-substituted N,N-dimethylanilines is a function of the electron withdrawing/donating properties of the substituent and the oxidation-reduction potential of the substrate, supporting a role in rate-limiting electron transfer from substrate to the high valent P450. In the oxidations of ethanol and acetaldehyde by human P450 2E1, a step following product formation must be the slow step (but not product release per se). Several oxidations catalyzed by human P450s 1A2 and 2D6 show slow C-H bond breaking, and apparent high-valent iron complexes accumulate in the reaction steady-state. Kinetic simulations were used to test the suitability of potential schemes and to probe the effects of changes in individual reaction steps.     

P450 reductase and cytochrome b5 interactions with cytochrome P450: effects on house fly CYP6A1 catalysis

The interactions of protein components of the xenobiotic-metabolizing cytochrome P450 system, CYP6A1, P450 reductase, and cytochrome b5 from the house fly (Musca domestica) have been characterized. CYP6A1 activity is determined by the concentration of the CYP6A1-P450 reductase complex, regardless of which protein is present in excess. Both holo- and apo-b5 stimulated CYP6A1 heptachlor epoxidase and steroid hydroxylase activities and influenced the regioselectivity of testosterone hydroxylation. The conversion of CYP6A1 to its P420 form was decreased by the addition of apo-b5. The effects of cytochrome b5 may involve allosteric modification of the P450 enzyme that modify the conformation of the active site. The overall stoichiometry of the P450 reaction was substrate-dependent. High uncoupling of CYP6A1 was observed with generation of hydrogen peroxide, in excess over the concomitant testosterone hydroxylation or heptachlor epoxidation. Inclusion of cytochrome b5 in the reconstituted system improved efficiency of oxygen consumption and electron utilization from NADPH, or coupling of the P450 reaction. Depending on the reconstitution conditions, coupling efficiency varied from 8 to 25% for heptachlor epoxidation, and from 11 to 70% for testosterone hydroxylation. Because CYP6A1 is a P450 involved in insecticide resistance, this suggests that xenobiotic metabolism by constitutively overexpressed P450s may be linked to significant oxidative stress in the cell that may carry a fitness cost.     

Extending the diversity of cytochrome P450 enzymes by DNA family shuffling

The cytochrome P450 enzymes involved in xenobiotic metabolism are an excellent starting point for the directed evolution of novel biocatalysts due to their wide substrate specificity. A shuffled library of three highly homologous mammalian genes (for P450 2C9, P450 2C11 and P450 2C19) was constructed by applying a modified DNA family shuffling procedure. The modifications made to the traditional DNA shuffling protocols involved non-random digestion via the use of different combinations of restriction enzymes (REs) followed by isolation of fragments under 300 bp by size-selective filtration. Shuffled cytochrome P450 mutants were co-expressed in Escherichia coli with their redox partner, NADPH-cytochrome P450 reductase (NPR). We report here how non-random fragmentation may help in chimeragenesis within the areas of low sequence similarity such as substrate recognition sites (SRSs) that are generally underrepresented in recombination using the random fragmentation process. Size-selective filtration was used to limit recovery of incompletely digested fragments and consequently minimize the chances for contamination of the shuffled library with parental forms. No parental forms could be detected in the shuffled library using restriction fragment length polymorphism (RFLP) analysis, suggesting the library was free of parental contamination. Sequencing of randomly selected mutants demonstrated a high level of chimeragenesis with on average of 8.0+/-2.2 crossovers and a low level of mutagenesis with 5.2+/-2.8 spontaneous mutations per approximately 1.5 kbp of the full-length P450 sequence. The proportion of properly folded protein as indicated by the observation of characteristic Fe(II).CO vs. Fe(II) difference spectra was 15% (4/27) of analysed mutants. Screening of the shuffled library for indole oxidation revealed four clones with similar or higher levels of indigo pigment production to those of the parental P450s and two clones with elevated P450 expression. In this paper we present a method for the effective family shuffling of cytochrome P450 enzymes, applicable to the creation of mutant libraries with expanded metabolic diversity and with a significant proportion of functional clones.     

Improved cloning and expression of cytochrome P450s and cytochrome P450 reductase in yeast

Combination of the pYeDP60 yeast expression system with a modified version of the improved uracil-excision (USER) cloning technique provides a new powerful tool for high-throughput expression of eukaryotic cytochrome P450s. The vector presented is designed to obtain an optimal 5' untranslated sequence region for yeast (Kozak consensus sequence), and has been tested to produce active P450s and NADPH-cytochrome P450 oxidoreductase (CPR) after 5' end silent codon optimization of the cDNA sequences. Expression of two plant cytochrome P450s, Sorghum bicolor CYP79A1 and CYP71E1, and S. bicolor CPR2 using the modified pYeDP60 vector in all three cases produced high amounts of active protein. High-throughput functional expression of cytochrome P450s have long been a troublesome task due to the workload involved in cloning of each individual P450 into a suitable expression vector. The redesigned yeast P450 expression vector (pYeDP60u) offers major improvements in cloning efficiency, speed, fidelity, and simplicity. The modified version of the USER cloning system provides great potential for further development of other yeast vectors, transforming these into powerful high-throughput expression vectors.     

Kinetic mechanism of cytochrome P450 reductase from the house fly (Musca domestica).

Recombinant house fly (Musca domestica) cytochrome P450 reductase has been purified by anion exchange and affinity chromatography. Steady-state kinetics of cytochrome c reductase activity revealed a random Bi-Bi mechanism with formation of a ternary P450 reductase-NADPH-electron acceptor complex as catalytic intermediate. NADP(H) binding is essential for fast hydride ion transfer to FAD, as well as for electron transfer from FMN to cytochrome c. Reduced cytochrome c had no effect on the enzyme activity, while NADP+ and 2'-AMP inhibited P450 reductase competitively with respect to NADPH and noncompetitively with respect to cytochrome c. The affinity of the P450 reductase to NADPH is 10 times higher than to NADP+ (Kd of 0.31 and 3.3 microM, respectively). Such an affinity change during catalysis could account for a +30 mV shift of the redox potential of FAD. Cys560 was substituted for Tyr by site-directed mutagenesis. This mutation decreased enzyme affinity to NADPH 35-fold by decreasing the bimolecular rate constant of nucleotide binding with no detectable effect on the kinetic mechanism. The affinity of the C560Y mutant enzyme to NADP+ decreased 9-fold compared to the wild-type enzyme, while the affinity to 2'-AMP was not significantly affected, suggesting that Cys560 is located in the nicotinamide binding site of the active, full-size enzyme in solution.     

High diversity and complex evolution of fungal cytochrome P450 reductase: cytochrome P450 systems

Cytochrome P450 reductase (CPR) is the redox partner of P450 monooxygenases, involved in primary and secondary metabolism of eukaryotes. Two novel CPR genes, sharing 34% amino acid identity, were found in the filamentous ascomycete Cochliobolus lunatus. Fungal genomes were searched for putative CPR enzymes. Phylogenetic analysis suggests that multiple independent CPR duplication events occurred in fungi, whereas P450-CPR fusion occurred before the diversification of Dikarya and Zygomycota. Additionally, losses of methionine synthase reductase were found in certain fungal taxa; a truncated form of this enzyme was conserved in Pezizomycotina. In fungi, high numbers of cytochrome P450 enzymes, multiple CPRs, and P450-CPR fusion proteins were associated with filamentous growth. Evolution of multiple CPR-like oxidoreductases in filamentous fungi might have been driven by the complexity of biochemical functions necessitated by their growth form, as opposed to yeast.     

Probing the world of cytochrome P450 enzymes

Variability in drug response can be attributed in part to variability in the activity of drug-metabolizing enzymes. One of the most important drugmetabolizing enzyme systems in humans is the cytochrome P450 (CYP) enzyme family, which is responsible for the oxidative metabolism of numerous endogenous compounds and xenobiotics. The clinical relevance of factors that influence CYP-mediated metabolism can be appreciated by estimating in vivo enzyme activity (i.e., the phenotype) through the use of "probe drugs," which are drugs predominately or exclusively metabolized by an individual CYP enzyme. Thus, the use of probe drugs alone or in combination (i.e., the cocktail approach) can provide an invaluable tool to explore the clinical relevance of genetic and nongenetic factors that affect CYP enzyme activity and thereby contribute substantially to variability in response to therapeutic drugs.     

Expression of bovine cytochrome P450c21 and its fused enzymes with yeast NADPH-cytochrome P450 reductase in Saccharomyces cerevisiae

Recombinant plasmids for expression of bovine cytochrome P450c21 (pA gamma 2), both P450c21 and yeast NADPH-cytochrome P450 reductase (pAR gamma 1), P450c21/yeast reductase fused enzymes (pAF gamma R1, pAF gamma R2, and pAF gamma R20), and yeast reductase/P450c21 fused enzymes (pAFR gamma 1 and pAFR gamma 2) were constructed by using expression vector pAAH5. The plasmids were each introduced into the yeast Saccharomyces cerevisiae AH22 cells. The recombinant yeast strains AH22/pA gamma 2 (Y21) and AH22/pAR gamma 1 (Y21R) produced 2-3 X 10(3) molecules of P450c21 per cell. The cultures of both strains converted progesterone and 17 alpha-hydroxyprogesterone into 11-deoxycorticosterone and 11-deoxycortisol, respectively. The 21-hydroxylase activity per cell of the strain Y21R was about three times higher than that of the strain Y21, probably due to overproduction of yeast reductase. The recombinant yeast strains AH22/pAF gamma R1 (Y21RF1), AH22/pAF gamma R2 (Y21RF2), and AH22/pAF gamma R20 (Y21RF20) produced about 1.1-2.0 X 10(4) molecules per cell of the corresponding P450c21/yeast reductase fused enzymes. The specific 21-hydroxylase activity toward 17 alpha-hydroxyprogesterone per cell of the strains Y21RF1, Y21RF2, and Y21RF20 was about 21, 28, and 49 times higher than that of the strain Y21, respectively. Thus, the fused enzymes were superior to P450c21 in the specific activity and in the expression level in the yeast. The Km values for 17 alpha-hydroxyprogesterone of P450c21 in the strains Y21 and Y21R, and of the fused enzymes in the strains Y21RF1 and Y21RF2 were 0.29, 0.30, 0.67, and 0.65 microM, respectively. The Vmax values of P450c21 in the strains Y21 and Y21R, and of the fused enzymes in the strains Y21RF1 and Y21RF2 were 28, 124, 151, and 222 moles/min.mole P450c21 or fused enzyme, respectively. These results indicated that the fused enzymes showed lower affinity for the substrate, probably due to structural modification and higher reaction rates through efficient intramolecular electron transfer as compared with those of P450c21. While the strain AH22/pAFR gamma 2 (YR21F2) produced about 3 X 10(4) molecules per cell of the reductase/P450c21 fused enzyme, the specific 21-hydroxylase activity of the fused enzyme toward 17 alpha-hydroxyprogesterone was extremely low, suggesting that the structure of the fused enzyme might not be suited for electron transfer in yeast microsomes.     

The role of cytochrome P450 lysine residues in the interaction between cytochrome P450IA1 and NADPH-cytochrome P450 reductase.

Cytochrome P450IA1 (purified from hepatic microsomes of beta-naphthoflavone-treated rats) has been covalently modified with the lysine-modifying reagent acetic anhydride. Different levels of lysine residue modification in cytochrome P450IA1 can be achieved by varying the concentration of acetic anhydride. Modification of lysine residues in P450IA1 greatly inhibits the interaction of P450IA1 with NADPH-cytochrome P450 reductase. Modification of 1.0 and 3.3 mol lysine residues per mole P450IA1 resulted in 30 and 95% decreases, respectively, in 7-ethoxycoumarin hydroxylation by a reconstituted P450IA1/reductase complex. However, modification of 3.3 mol lysine residues per mole P450IA1 decreased only cumene hydroperoxide-supported P450-dependent 7-ethoxycoumarin hydroxylation by 30%. Spectral and fluorescence studies showed no indication of global conformational change of P450IA1 even with up to 8.8 mol lysine residues modified per mole P450IA1. These data suggest that at least three lysine residues in P450IA1 may be involved in the interaction with reductase. Identification of lysine residues in P450IA1 possibly involved in this interaction was carried out by [14C]acetic anhydride modification, trypsin digestion, HPLC separation, and amino acid sequencing. The lysine residue candidates identified in this manner were K97, K271, K279, and K407.     

Inactivation of the hepatic cytochrome P450 system by conditional deletion of hepatic cytochrome P450 reductase

Cytochrome P450 (CYP) monooxygenases catalyze the oxidation of a large number of endogenous compounds and the majority of ingested environmental chemicals, leading to their elimination and often to their metabolic activation to toxic products. This enzyme system therefore provides our primary defense against xenobiotics and is a major determinant in the therapeutic efficacy of pharmacological agents. To evaluate the importance of hepatic P450s in normal homeostasis, drug pharmacology, and chemical toxicity, we have conditionally deleted the essential electron transfer protein, NADH:ferrihemoprotein reductase (EC, cytochrome P450 reductase, CPR) in the liver, resulting in essentially complete ablation of hepatic microsomal P450 activity. Hepatic CPR-null mice could no longer break down cholesterol because of their inability to produce bile acids, and whereas hepatic lipid levels were significantly increased, circulating levels of cholesterol and triglycerides were severely reduced. Loss of hepatic P450 activity resulted in a 5-fold increase in P450 protein, indicating the existence of a negative feedback pathway regulating P450 expression. Profound changes in the in vivo metabolism of pentobarbital and acetaminophen indicated that extrahepatic metabolism does not play a major role in the disposition of these compounds. Hepatic CPR-null mice developed normally and were able to breed, indicating that hepatic microsomal P450-mediated steroid hormone metabolism is not essential for fertility, demonstrating that a major evolutionary role for hepatic P450s is to protect mammals from their environment.     

Highly reactive electrophilic oxidants in cytochrome P450 catalysis

The cytochrome P450 enzymes effect a wide range of oxidations in nature including difficult hydroxylation reactions of unactivated C-H. Most of the high energy reactions of these catalysts appear to involve highly electrophilic active species. Attempts to detect the reactive transients in the enzymes have met with limited success, but evidence has accumulated that two distinct electrophilic oxidants are produced in the P450 enzymes. The consensus electrophilic oxidant termed "iron-oxo" is usually thought to be an analogue of Compound I, an iron(IV)-oxo porphyrin radical cation species, but it is possible that a higher energy electronic isomer of Compound I is required to account for the facility of the C-H oxidation reactions. The second electrophilic oxidant of P450 is speculative; circumstantial evidence suggests that this species is iron-complexed hydrogen peroxide, but this oxidant might be a second spin state of iron-oxo. This overview discusses recent studies directed at detection of the electrophilic oxidants in P450 enzymes and the accumulated evidence for two distinct species.     

Microsomal lipid peroxidation: the role of NADPH--cytochrome P450 reductase and cytochrome P450

The role of NADPH--cytochrome P450 reductase and cytochrome P450 in NADPH- and ADP--Fe3(+)-dependent lipid peroxidation was investigated by using the purified enzymes and liposomes prepared from either total rat-liver phospholipids or a mixture of bovine phosphatidyl choline and phosphatidyl ethanolamine (PC/PE liposomes). The results suggest that NADPH- and ADP--Fe3(+)-dependent lipid peroxidation involves both NADPH--cytochrome P450 reductase and cytochrome P450. Just as in the case of cytochrome P450-linked monooxygenations, the role of these enzymes in lipid peroxidation may be to provide two electrons for O2 reduction. The first electron is used for reduction of ADP--Fe3+ and subsequent addition of O2 to the perferryl radical (ADP--Fe3(+)-O2-), which then extracts an H atom from a polyunsaturated lipid (LH) giving rise to a free radical (LH.) that reacts with O2 yielding a peroxide free radical (LOO.). The second electron is then used to reduce LOO. to the lipid hydroperoxide (LOOH). In the latter capacity, reduced cytochrome P450 can be replaced by EDTA--Fe2+ or by the superoxide radical as generated through redox cycling of a quinone such as menadione.     

High diversity and complex evolution of fungal cytochrome P450 reductase: cytochrome P450 systems.

Cytochrome P450 reductase (CPR) is the redox partner of P450 monooxygenases, involved in primary and secondary metabolism of eukaryotes. Two novel CPR genes, sharing 34% amino acid identity, were found in the filamentous ascomycete Cochliobolus lunatus. Fungal genomes were searched for putative CPR enzymes. Phylogenetic analysis suggests that multiple independent CPR duplication events occurred in fungi, whereas P450-CPR fusion occurred before the diversification of Dikarya and Zygomycota. Additionally, losses of methionine synthase reductase were found in certain fungal taxa; a truncated form of this enzyme was conserved in Pezizomycotina. In fungi, high numbers of cytochrome P450 enzymes, multiple CPRs, and P450-CPR fusion proteins were associated with filamentous growth. Evolution of multiple CPR-like oxidoreductases in filamentous fungi might have been driven by the complexity of biochemical functions necessitated by their growth form, as opposed to yeast.     

Uncovering the role of hydrophobic residues in cytochrome P450-cytochrome P450 reductase interactions

Cytochrome P450 (CYP or P450)-mediated drug metabolism requires the interaction of P450s with their redox partner, cytochrome P450 reductase (CPR). In this work, we have investigated the role of P450 hydrophobic residues in complex formation with CPR and uncovered novel roles for the surface-exposed residues V267 and L270 of CYP2B4 in mediating CYP2B4--CPR interactions. Using a combination of fluorescence labeling and stopped-flow spectroscopy, we have investigated the basis for these interactions. Specifically, in order to study P450--CPR interactions, a single reactive cysteine was introduced in to a genetically engineered variant of CYP2B4 (C79SC152S) at each of seven strategically selected surface-exposed positions. Each of these cysteine residues was modified by reaction with fluorescein-5-maleimide (FM), and the CYP2B4-FM variants were then used to determine the K(d) of the complex by monitoring fluorescence enhancement in the presence of CPR. Furthermore, the intrinsic K(m) values of the CYP2B4 variants for CPR were measured, and stopped-flow spectroscopy was used to determine the intrinsic kinetics and the extent of reduction of the ferric P450 mutants to the ferrous P450--CO adduct by CPR. A comparison of the results from these three approaches reveals that the sites on P450 exhibiting the greatest changes in fluorescence intensity upon binding CPR are associated with the greatest increases in the K(m) values of the P450 variants for CPR and with the greatest decreases in the rates and extents of reduced P450--CO formation.     

Differential effect of copper (II) on the cytochrome P450 enzymes and NADPH-cytochrome P450 reductase: inhibition of cytochrome P450-catalyzed reactions by copper (II) ion

Inhibitory effects of Cu(2+) on the cytochrome P450 (P450)-catalyzed reactions of liver microsomes and reconstituted systems containing purified P450 and NADPH-P450 reductase (NPR) were seen. However, Zn(2+), Mg(2+), Mn(2+), Ca(2+), and Co(2+) had no apparent effects on the activities of microsomal P450s. Cu(2+) inhibited the reactions catalyzed by purified P450s 1A2 and 3A4 with IC(50) values of 5.7 and 8.4 microM, respectively. Cu(2+) also inhibited reduction of cytochrome c by NPR (IC(50) value of 5.8 microM). Copper caused a decrease in semiquinone levels of NPR, although it did not disturb the rate of formation of semiquinone. P450 reactions supported by an oxygen surrogate, tert-butyl hydroperoxide, instead of NPR and NADPH, were inhibited by the presence of Cu(2+). The results indicate that Cu(2+) inhibits the P450-catalyzed reactions by affecting both P450s and NPR. It was also found that the inhibition of catalytic activities of P450s by Cu(2+) involves overall conformational changes of P450s and NPR, investigated by CD and intrinsic fluorescence spectroscopy. These results suggest that the inhibitory effect of Cu(2+) on the P450-catalyzed reactions may come from the inability of an efficient electron transfer from NPR to P450 and also the dysfunctional conformation of NPR and P450.     

Coexpression in yeast of Taxus cytochrome P450 reductase with cytochrome P450 oxygenases involved in Taxol biosynthesis

To maximize redox coupling efficiency with recombinant cytochrome P450 hydroxylases from yew (Taxus) species installed in yeast for the production of the anticancer drug Taxol, a cDNA encoding NADPH:cytochrome P450 reductase from T. cuspidata was isolated. This single-copy gene (2,154 bp encoding a protein of 717 amino acids) resembles more closely other reductases from gymnosperms (approximately 90% similarity) than those from angiosperms (<80% similarity). The recombinant reductase was characterized and compared to other reductases by heterologous expression in insect cells and was shown to support reconstituted taxoid 10beta-hydroxylase activity with an efficiency comparable to that of other plant-derived reductases. Coexpression in yeast of the reductase along with T. cuspidata taxoid 10beta-hydroxylase, which catalyzes an early step of taxoid biosynthesis, demonstrated significant enhancement of hydroxylase activity compared to that supported by the endogenous yeast reductase alone. Functional transgenic coupling of the Taxus reductase with a homologous cytochrome P450 taxoid hydroxylase represents an important initial step in reconstructing Taxol biosynthesis in a microbial host.     

Differential effect of phenobarbital and beta-naphthoflavone on the mRNAs coding for cytochrome P450 and NADPH cytochrome P450 reductase

The induction in rat liver of a specific variant(s) of cytochrome P450 (PB-P450) by phenobarbital and its repression by β-naphthoflavone occur through corresponding changes in the levels of mRNA coding for the protein(s). The level of translatable mRNA coding for NADPH-cytochrome P450 reductase in rat liver increases on treatment with phenobarbital but not β-naphthoflavone.     

Characterization of human cytochrome P450 enzymes.

Many biochemical approaches have been applied to the human cytochrome P450 enzymes, and more than 20 different gene products have been characterized with regard to their properties and catalytic specificities. The complement of the various cytochrome P450 enzymes in a given individual varies markedly, and dramatic differences may be seen in drug metabolism, pharmacological response, and susceptibility to toxic effects. An understanding of the nature of the individual cytochrome P450 enzymes and their regulation should be useful in determining the most suitable animal models, ascertaining risk from chemicals, and in avoiding undesirable drug interactions.