High-throughput approaches for evaluating absorption, distribution, metabolism and excretion properties of lead compounds.

Combinatorial chemistry methods and high-throughput screening for leads in industrial drug discovery have generated a potential bottleneck in the optimisation processes that seek to align potency with good pharmacokinetics in order to produce good medicines. This has resulted in the need for higher throughput methods of screening for absorption, distribution, metabolism and excretion properties. Significant progress has been made in throughput of in vivo pharmacokinetic studies, with the introduction of cassette, or multiple-in-one, protocols. In this technique, typically up to ten compounds are administered in one dose and analysed concomitantly on the mass spectrometer. High-throughput methods in in vitro absorption, distribution, metabolism and excretion are less well-developed as yet, and current approaches comprise automation of well-established methods for absorption using cell lines and metabolism using liver microsomes.

Genetic analysis of the human cytochrome P450 CYP2C9 locus.

Cytochrome P450 CYP2C9 metabolizes a wide variety of clinically important drugs, including phenytoin, tolbutamide, warfarin and a large number of non-steroidal anti-inflammatory drugs. Previous studies have shown that even relatively conservative changes in the amino acid composition of this enzyme can affect both its activity and substrate specificity. To date six different human CYP2C9 cDNA sequences, as well as the highly homologous CYP2C10 sequence have been reported suggesting that the CYP2C9 gene is polymorphic. Only nine single base substitutions in the coding region of CYP2C9 account for the differences seen between the CYP2C9 proteins. In this report we have developed polymerase chain reaction (PCR)-based assays to distinguish all seven sequences, and have determined their allele frequencies in the Caucasian population. Of the seven sequences studied in one hundred individuals only three appeared to be CYP2C9 alleles. These alleles termed CYP2C9*1, CYP2C9*2 and CYP2C9*3 had allele frequencies of 0.79, 0.125 and 0.085 respectively. The CYP2C10 gene could not be found in any of the samples studied. The assays developed here will allow the prediction of CYP2C9 phenotype, thus identifying those individuals who may exhibit different drug pharmacokinetics for CYP2C9 substrates.

A comparative study of 1-substituted imidazole and 1,2,4-triazole antifungal compounds as inhibitors of testosterone hydroxylations catalysed by mouse hepatic microsomal cytochromes P-450.

Three imidazole antifungal agents, ketoconazole, miconazole and tioconazole, and a group of structurally related 1-substituted imidazole and 1,2,4-triazole compounds were evaluated as inhibitors of the oxidative metabolism of testosterone catalysed by mouse hepatic microsomal cytochromes P-450. Spectroscopic studies showed that both imidazoles and triazoles interacted with ferric cytochrome P-450 in hepatic microsomes to produce type II difference spectra which could be distinguished by their different absorbance maxima; 429-430 nm and 425-426 nm respectively. Compound 4, which possesses both types of functional group, produced a spectrum which resembled that of imidazole compounds, indicating that the imidazole moiety had a higher affinity than the triazole for the haem of cytochromes P-450 present in microsomes. The test compounds differentially inhibited regio- and stereo-specific testosterone metabolism and the pattern of inhibition varied with the 1-substituent on the azole ring. Ketoconazole was a potent inhibitor of testosterone 6 beta-hydroxylation (IC50 0.08 microM) but was considerably less active against other hydroxylations and 17 beta-oxidation to androstenedione (IC50 range 13 to greater than 100 microM). In contrast, tioconazole (IC50 range 0.18 to 3.3 microM) and miconazole (IC50 range 0.15 to 10 microM) were relatively non-selective. Compounds 1 and 2, which differed from each other only in the type of azole ring, were most active against 16 beta-hydroxylation. The triazole analogue (compound 2) was a significantly more potent inhibitor of 16 beta-hydroxylation than the imidazole (compound 1), equipotent against androstenedione formation and less active against the other hydroxylations. Two relatively polar bis-azole analogues (compounds 3 and 4) were most active against androstenedione formation; however, in general they were less inhibitory than the lipophilic azoles. We conclude that azole antifungal agents of differing structure show different patterns of selective interaction with cytochromes P-450, a phenomenon primarily dependent on the 1-substituent on the azole ring, but also modulated to a lesser extent by the type of azole ring (imidazole or triazole).

Characterization of the enzyme responsible for the metabolism of sumatriptan in human liver.

Studies have been undertaken to investigate the enzymes responsible for the metabolism of [14C]sumatriptan in man. Oxidative deamination of sumatriptan to form the indole acetic acid derivative is the only phase 1 pathway evident in man and both cytochrome P450 (P450) and monoamine oxidase (MAO) are capable of catalysing this type of reaction. The metabolism of [14C]sumatriptan was therefore investigated in vitro in a preparation derived from human liver, which was shown, by the use of the probe substrates [14C]testosterone (P450), [3H]5HT (MAO-A) and [14C]benzylamine (MAO-B) to be a rich source of both enzyme systems. Incubation with clorgyline and deprenyl, probe inhibitors of MAO-A and MAO-B, respectively, showed that [14C]sumatriptan was metabolized by MAO-A; there was no evidence of P450 involvement in its metabolism. The data in this study therefore indicate that the enzyme MAO-A is the major enzyme responsible for the metabolism of sumatriptan in human liver.

Comparison of two azole antifungal drugs, ketoconazole, and fluconazole, as modifiers of rat hepatic monooxygenase activity.

The mechanism of action of azole antifungal agents is believed to involve inhibition of fungal cytochrome P-450, and, therefore, an investigation of the interaction of these drugs with mammalian cytochrome P-450 systems should provide some indication of their selectivity as antifungal agents. The ability of ketoconazole and fluconazole, the latter representing a new generation of triazole antifungal agents, to modify rat mixed function oxidase activity has been investigated in vitro with hepatic microsomes and in vivo using a N-methyl-[14C] antipyrine breath test. As a measure of selectivity the results have been compared with antifungal potency. Ketoconazole is more potent than fluconazole by an order of magnitude in inhibiting metabolism by O-dealkylation of ethoxycoumarin, methoxycoumarin and ethoxyresorufin (IC50 values of 6, 5 and 130 microM for ketoconazole respectively). The effects on the regio- and stereospecific hydroxylation of [14C] testosterone were also measured; the IC50 values for inhibition of total testosterone metabolism were 0.1 mM and greater than 3 mM for ketoconazole and fluconazole respectively. Marked selectivity differences were observed for the two drugs as indicated by ketoconazole being a potent inhibitor of 7 alpha-hydroxylation of testosterone (IC50 20 microM) while fluconazole did not inhibit this activity at 3 mM. In vivo investigations using a range of doses confirmed their ranking for inhibitory potency; the ED50 values for maximum demethylation rate were 17 mumol/kg and greater than 60 mumol/kg for ketoconazole and fluconazole respectively. Thus fluconazole has a lower propensity to interact with rat hepatic cytochrome P-450 and can be considered a more selective antifungal agent as its in vivo antifungal potency is an order of magnitude greater than ketoconazole.

Molecular modelling of the human cytochrome P450 isoform CYP2A6 and investigations of CYP2A substrate selectivity.

(1) The generation of a homology model of CYP2A6, the major catalyst of human hepatic coumarin 7-hydroxylase activity, involves the use of the recently published substrate-bound CYP102 crystal structure as a template. (2) A substantial number of structurally diverse CYP2A6 substrates are found to dock satisfactorily within the putative active site of the enzyme, leading to the formulation of a structural template (or pharmacophore) for CYP2A6 specificity/selectivity. (3) The CYP2A6 model is consistent with available evidence from site-directed mutagenesis studies carried out on CYP2A subfamily isoforms, and enables some explanation of species differences in CYP2A-mediated metabolism of certain substrates. (4) Quantitative structure-activity relationship (QSAR) analysis of CYP2A5 (the mouse orthologue) mutants yields statistically significant correlations between various properties of amino acid residues and coumarin 7-hydroxylase activity.

Molecular modelling of CYP1 family enzymes CYP1A1, CYP1A2, CYP1A6 and CYP1B1 based on sequence homology with CYP102.

Molecular modelling of a number of CYP1 family enzymes from rat, plaice and human is described based on amino acid sequence homology with the haemoprotein domain of CYP102, a unique bacterial P450 of known structure. The interaction of various substrates and inhibitors within the putative active sites of rat CYP1A1, human CYP1A2, a fish CYP1 enzyme CYP1A6 (from plaice) and human CYP1B1, is shown to be consistent with P450-mediated oxidation in each example or, in the case of inhibitors, mechanism of inhibition. It is reported that relatively small changes between the enzymes' active site regions assist in the rationalization of CYP1 enzyme preferences for particular substrate types, and a template of superimposed CYP1A2 substrates is shown to fit the putative active site of the human CYP1A2 enzyme.

Structural determinants of cytochrome P450 substrate specificity, binding affinity and catalytic rate.

The structural characteristics of cytochrome P450 substrates are summarised, showing that molecular descriptors can discriminate between chemicals of differing P450 isozyme specificity. Procedures for the estimation of P450 substrate binding interaction energies and rates of metabolism are described, providing specific examples in both individual compounds binding to P450s, including those of known crystal structure, and within series of structurally related chemicals. It is demonstrated that binding energy components are primarily hydrophobic/desolvation and electrostatic/hydrogen-bonded in nature, whereas electronic factors are of importance in determining variations in reaction rates. It is thus shown that the prediction of P450 substrate binding affinities and catalytic rates may be feasible, provided that sufficient structural information is available for the relevant enzyme-substrate complex.

Cytochrome P450 substrate specificities, substrate structural templates and enzyme active site geometries.

The structural characteristics of human cytochrome P450 substrates are outlined in the light of extensive studies on P450 substrate specificity. Templates of superimposed substrates for individual P450 isozymes are shown to fit the corresponding enzyme active sites, where contacts with specific amino acid residues appear to be involved in the interaction with each structural template. Procedures leading to the evaluation of likely P450 specificity, binding affinity and rate of metabolism are described in the context of key examples in which molecular modelling appears to rationalize experimentally observed findings.

The pattern of intramural veins of the left ventricle of the human heart.

A detailed injection, microradiographic, and histological necropsy study of 18 hearts showed that the pattern of myocardial veins differed from the arterial pattern. In "normal" hearts, large drainage veins began in the subendocardial zone and coursed fairly directly towards the epicardium, maintaining a comparatively even calibre throughout. Smaller, but similar, drainage veins also begain in the middle of the myocardial wall. In the outer myocardium small groups of these vessels converged to form a single vein which then entered, almost at right angles, the pericardial veins. In general, small veins within the myocardium entered directly into the large drainage veins without any extensive intermediate sized venous network. In "abnormal" hearts with left ventricular hypertrophy the myocardial veins appeared basically normal but were more widely separated and consequently overall less dense on microradiography. Interruption of main drainage veins and loss of normal pattern were seen in areas of myocardial fibrosis associated with coronary artery disease. In a heart with severe generalized coronary artery atheroma the normal pattern in the inner half of the wall was replaced by a "plexus" of small veins in which the majority of vessels coursed circumferentially. The possible significance of both the normal and abnormal vein pattern is discussed.

Molecular modelling of lanosterol 14 alpha-demethylase (CYP51) from Saccharomyces cerevisiae via homology with CYP102, a unique bacterial cytochrome P450 isoform: quantitative structure-activity relationships (QSARs) within two related series of antifungal azole derivatives.

The construction of a three-dimensional molecular model of the fungal form of cytochrome P450 (CYP51) from Saccharomyces cerevisiae, based on homology with the haemoprotein domain of CYP102 from Bacillus megaterium (a unique bacterial P450 of known crystal structure) is described. It is found that the endogenous substrate, lanosterol, can readily occupy the putative active site of the CYP51 model such that the known mono-oxygenation reaction, leading to C14-demethylation of lanosterol, is the preferred route of metabolism for this particular substrate. Key amino acid contacts within the CYP51 active site appear to orientate lanosterol for oxidative attack at the C14-methyl group, and the position of the substrate relative to the haem moiety is consistent with the phenyl-iron complexation studies reported by Tuck et al. [J. Biol. Chem., 267, 13175-13179 (1992)]. Typical azole inhibitors, such as ketoconazole, are able to fit the putative active site of CYP51 by a combination of haem ligation, hydrogen bonding, pi-pi stacking and hydrophobic interactions within the enzyme's haem environment. The mode of action of azole antifungals, as described by the modelling studies, is supported by quantitative structure-activity relationship (QSAR) analyses on two groups of structurally related fungal inhibitors. Moreover, the results of molecular electrostatic isopotential (EIP) energy calculations are compatible with the proposed mode of binding between azole antifungal agents and the putative active site of CYP51, although membrane interactions may also have a role in the antifungal activity of azole derivatives.

Pharmacokinetic evaluation of UK-49,858, a metabolically stable triazole antifungal drug, in animals and humans.

The pharmacokinetic profile of UK-49,858 (fluconazole), a novel triazole antifungal agent which is being developed for oral and intravenous use, was determined in mice, rats, dogs, and humans. Comparative data following oral and intravenous administration showed that bioavailability was essentially complete in all four species. Peak concentrations in plasma of drug normalized to a 1-mg/kg dose level following oral administration, were relatively high: 0.7, 0.6, 1.1, and 1.4 micrograms/ml in mice, rats, dogs, and humans, respectively. The volumes of distribution ranged between 1.1 liter/kg in mice and 0.7 liter/kg in humans, which are approximate to the values for total body water. Whole body autoradiography studies in mice following intravenous administration of [14C]UK-49,858 demonstrated that the drug was evenly distributed throughout the tissues, including the central nervous system and the gastrointestinal tract. Plasma protein binding was low (11 to 12%) in all species. Marked species differences were observed in elimination half-lives, with mean values of 4.8, 4.0, 14, and 22 h in mice, rats, dogs, and humans, respectively. The major route of elimination of the drug was renal clearance, with about 70% of the dose being excreted unchanged in the urine in each species. Studies with [14C]UK-49,858 on metabolism and excretion (intravenous and oral) in mice and dogs showed that about 90% of the dose was recovered as unchanged drug in urine and feces, confirming the metabolic stability of the drug. This pharmacokinetic profile is markedly different from that of imidazole antifungal drugs and undoubtedly contributes to the excellent efficacy of UK-49,858 in vivo.

Disposition of azole antifungal agents. II. Hepatic binding and clearance of dichlorophenyl-bis-triazolylpropanol (DTP) in the rat.

DTP (dichlorophenyl-bis-triazolylpropanol) was evaluated as a probe of drug-cytochromes P450 interactions in vitro and in vivo. Studies with rat liver microsomes demonstrate that DTP shows similar P450 binding affinity to its analog, ketoconazole, as determined by P450 difference spectra and inhibition of the metabolism of methoxycoumarin. As a more polar azole, DTP shows less affinity for rat plasma albumin (fraction unbound 0.56) than ketoconazole (fraction unbound 0.037). DTP metabolism is simpler than that of ketoconazole, with only one pathway, N-dealkylation which removes a triazole ring to yield DTP glycol. This primary metabolite is further metabolised to a carboxylic acid, a glycol glucuronide and a third unknown secondary metabolite (probably an acid glucuronide). Over a dose range of 0.1-24mg/kg there is complete mass balance recovery in urine via the five metabolites and unchanged drug. However DTP metabolism is dose dependent and while the affinity of DTP for the cytochromes P450 carrying out the initial dealkylation is high (1.5 microM based on unbound blood concentration), the capacity of the reaction is low (1 nmole/min). Under linear conditions, metabolic clearance is low (19ml/h), but ten-fold higher than renal clearance. The liver is the major distribution site for both DTP and ketoconazole. At low DTP concentrations, a specific high affinity process dominates the hepatic binding of DTP resulting in a liver:blood partition coefficient of approximately 30. Hepatic binding is concentration dependent and the progressive decrease in partition coefficient observed as the dose of DTP is escalated is coincident with a decrease in volume of distribution. The two saturable processes involved in the disposition of DTP result in an unusual concentration dependency in the blood concentration-time profile of this azole. Following administration of a high dose (10mg/kg) of DTP the log concentration-time profile is sigmoidal. At high concentrations (above 1mg/L) both the N-dealkylation and the hepatic binding of DTP are saturated, but as concentrations fall to approximately 0.05mg/L the former process becomes linear and the time profile is convex over this concentration range. At later times as DTP concentrations decline further, the tissue binding also reaches the linear region and the time profile becomes concave. Only at low concentrations (below 0.05mg/L) do both processes become first order and the true half life is evident.

Treatment of cryptococcal meningitis in mice with fluconazole.

Fluconazole is a recently developed triazole with activity in vitro against Cryptococcus neoformans, water solubility, and excellent oral absorption. We compared fluconazole in murine cryptococcosis with ketoconazole and amphotericin B. Fluconazole was highly effective in suppressing cryptococcosis in mice challenged by the intravenous and intranasal routes, and was comparable with the other two drugs in its protective capacity. However, fluconazole was superior to ketoconazole and comparable with amphotericin B after intracerebral challenge. Fluconazole may warrant clinical evaluation in cryptococcosis.

Molecular modelling of CYP3A4 from an alignment with CYP102: identification of key interactions between putative active site residues and CYP3A-specific chemicals.

1. A structural model of CYP3A4 is reported on the basis of a novel amino acid sequence alignment between the CYP3 family and CYP102, a bacterial P450 of known crystal structure. 2. Construction of the CYP3A4 model from CYP102 is facilitated by the relatively high sequence homology between the two protein (52% homology; 27% identity) with many conservative amino acid changes, yielding a structure of low internal energy. 3. A considerable number of specific substrates, and some specific inhibitors, are shown to occupy the putative CYP3A4 active site via interactions with the same amino acid residues in almost all cases investigated. 4. The CYP3A4 model rationalizes the known positions of metabolism for many substrates of this major human P450 such that the route of metabolism in novel development compounds can be predicted.

Molecular modelling of cytochrome P4502D6 (CYP2D6) based on an alignment with CYP102: structural studies on specific CYP2D6 substrate metabolism.

1. A molecular model of CYP2D6 has been constructed from the bacterial form CYP102 via a homology alignment between the CYP2D subfamily and CYP102 protein sequences. 2. A number of typical CYP2D6 substrates are shown to fit the putative active site of the enzyme, as can the specific inhibitor quinidine. 3. Some of the allelic variants in CYP2D6, which give rise to genetic polymorphisms in 2D6-mediated metabolism, can be rationalized in terms of their position within the active site region. 4. The results of site-directed mutagenesis experiments are consistent with the CYP2D6 model generated from the CYP102 crystal structure. 5. The possibility of an alternative orientation within the active site may explain the CYP2D6-mediated metabolism of relatively large-sized substrates.

The disposition and metabolism of [14C]fluconazole in humans.

The disposition and metabolism of 14C-labeled fluconazole (100 microCi) was determined in three healthy male subjects after administration of a single oral capsule containing 50 mg of drug. Blood samples, total voided urine, and feces were collected at intervals after dosing for up to 12 days post-dose. Pharmacokinetic analysis of fluconazole concentrations showed a mean plasma half-life of 24.5 hr. Mean apparent plasma clearance and apparent volume of distribution were 0.23 ml/min/kg and 0.5 liter/kg, respectively. There was no evidence of any significant concentrations of metabolites circulating either in plasma or blood cells. Mean total radioactivity excreted in urine and feces represented 91.0 and 2.3%, respectively, of the administered dose. Mean excretion of unchanged drug in urine represented 80% of the administered dose; thus, only 11% was excreted in urine as metabolites. Only two metabolites were present in detectable quantities, a glucuronide conjugate of unchanged fluconazole and a fluconazole N-oxide, which accounted for 6.5 and 2.0% of urinary radioactivity, respectively. No metabolic cleavage products of fluconazole were detected.