The chemokine receptor, CCR5, acts as the major coreceptor involved in viral entry into the cell in the case of primary HIV infection. Therefore, blocking the CCR5 receptor prevents viral entry into host cells and should thus decrease viral load in HIV-1 infected individuals. This theory is supported by genetic evidence which shows that individuals lacking a functional CCR5 gene are highly resistant to HIV-1 infection (Liu et al., 1996; Samson et al., 1996; Chantry, 2004). UK-427,857 (4, 4-difluoro-N-{(1S)-3-[exo-3-(3-isopropyl-5-methyl-4H-1,2,4-triazol-4-yl)-8-azabicyclo[3.2.1]oct-8-yl]-1-phenylpropyl}-cyclohexanecarboxamide) is a novel CCR5 receptor antagonist designed through a rational drug discovery program and is currently undergoing clinical evaluation (Bayes et al., 2003). UK-427,857 has a molecular weight of 514 and is a moderately lipophilic (log D_7.4 = 2.1) and basic (pK_a = 7.3) molecule. The empirical formula is C₂₉H₄₁F₂N₅O and the structure is shown in Fig. 1.

Pharmacokinetic studies with UK-427,857 were undertaken in rat and dog during the discovery program leading to the identification of this compound as a development candidate. These studies were performed to characterize the general pharmacokinetic properties of the compound and to predict the potential pharmacokinetic profile in humans as part of the compound selection process. The animal pharmacokinetic studies and early in vitro permeability assessments indicated that the membrane permeability of UK-427,857 was restricted, leading to incomplete absorption. Additional in vitro studies in Caco-2 monolayers with inhibitors of P-glycoprotein and studies in P-glycoprotein null mice were undertaken to understand the absorption mechanism of the compound and to aid in the prediction of the likely clinical pharmacokinetic profile. Further pharmacokinetic data in animal species have been obtained as part of the general toxicology program and have allowed characterization of the dose-exposure relationships in both rat and dog. The dose-exposure relationship after oral administration in humans was studied over a wide dose range during early clinical trials. A major objective of this paper is to characterize the disposition properties of UK-427,857 in humans and to relate these to our knowledge of the disposition in animal species, to identify species differences and similarities. Such understanding provides valuable information for the prediction of human pharmacokinetic properties for future drug candidates. In addition, thorough characterization of the disposition and metabolic fate of UK-427,857 has been performed in the animal species used within the toxicology program and in humans to confirm the suitability of the animal species in the safety assessment of the compound. These studies used ¹⁴C-labeled compound and mass spectrometry to facilitate the identification of metabolic products.

Materials and Methods

Chemicals. UK-427,857; internal standards, UK-377,461 (structural analog) and UK-462,015 (D₅-UK-427,857); and authentic metabolite standards, UK-408,027 (metabolite 2) and UK-437,719 (metabolite 3), were synthesized at Pfizer Global Research and Development (Sandwich, UK). [¹⁴C]UK-427,857 (see Fig. 1) was prepared by Amersham Biosciences UK, Ltd. (Little Chalfont, Buckinghamshire, UK) with a radiochemical purity of 98% (by HPLC) and a specific activity of 33 μCi/mg. The chemical purity of all compounds was >95%.

Rat and Dog Pharmacokinetic Studies. All animal work was carried out in compliance with UK or French law and was approved by a local ethical review process as appropriate. Dose solutions were freshly prepared on the day of use. For rat pharmacokinetic studies, UK-427,857 was dissolved in saline (1 mg/kg for i.v. administration) or water (1 and 3.3 mg/ml for oral administration) containing 5% dimethyl sulfoxide (DMSO) and 0.5% 0.1 N HCl. The same i.v. dose vehicle was used for dog pharmacokinetic studies (0.5 mg/ml UK-427,857); however, a dose vehicle of 0.2 M lactate buffer (pH 3.9) was used for the oral doses (1 and 2 mg/ml UK-427,857).

For i.v. studies, two male Sprague-Dawley rats (∼250 g; Charles River, Margate, Kent, UK) were surgically prepared with jugular vein catheters at least 2 days before dose administration. Twenty-four noncannulated rats were used for oral dose studies. Two male and two female beagle dogs (Pfizer colony; 12-18 kg) were used in a crossover design study for i.v. and oral pharmacokinetics. The i.v. doses were administered at a dose level of 1 mg/kg via the tail vein to rats and at a dose level of 0.5 mg/kg as a 15-min infusion via the saphenous vein to dogs. Oral doses were administered by gavage tube. Twelve rats received doses of 3 mg/kg and 12 received doses of 10 mg/kg. Two dogs received doses of 1 mg/kg and two dogs received doses of 2 mg/kg. Oral doses to dogs were administered at least 1 week after the i.v. dose to the same animal. Blood samples (200 μl) were collected from the indwelling catheters of i.v. dosed rats. For the orally dosed rats, animals were terminally anesthetized with isoflurane, and blood samples (2 ml) were simultaneously collected from the hepatic portal vein and vena cava of two rats per time point for each dose level. Serial blood samples (2 ml) from dogs were collected from temporary indwelling saphenous vein catheters or by venipuncture of the cephalic vein. All blood samples were transferred to lithium heparin tubes, mixed, and centrifuged. Plasma samples were stored frozen before analysis. Urine samples (0-7 h) were collected from male dogs after i.v. doses by catheterization of the bladder and stored frozen before analysis.

Toxicology Studies in Rat and Dog. Within the toxicology program with UK-427,857, Sprague-Dawley rats (Charles River, Les Oncins, France) and beagle dogs (Marshall Farms USA, Inc., North Rose, NY) received multiple daily oral doses of UK-427,857 for 1 month. Toxicokinetic data were obtained from groups of five male rats receiving doses of 100, 300, and 1500 mg/kg and groups of six beagle dogs (three male and three female) receiving doses of 10, 50, and 250 mg/kg. Doses were prepared in 0.5% (w/v) aqueous methylcellulose containing 0.1% (w/v) Tween 80 and were administered at 10 ml/kg (rats) or 1 ml/kg (dogs) by oral gavage. Blood samples (0.5 ml) were collected at 1, 3, 7, and 24 h from rats on days 1 and 23 of the study. Blood samples (2.5 ml) were collected at 1, 4, 7, and 24 h from dogs on days 1 and 19 of the study.

Dose Escalation Study in Humans. Twenty-four male volunteers (aged 18-45 years, weight 60-100 kg) provided written informed consent and participated in a double-blind, placebo-controlled four-way crossover dose escalation study in two separate cohorts of 12 subjects. The study was conducted according to the Association of British Pharmaceutical Industry guidelines and the revised Declaration of Helsinki. The local ethics committee approved the clinical study protocol, and the study was performed at the Pfizer Clinical Research Unit and Kent and Canterbury Hospital. The doses administered were 1, 10, 100, and 900 mg for cohorts 1 and 3, 30, 300, and 1200 mg for cohort 2. Doses were administered in solution in 0.01 N HCl. Within each group of 12 subjects, nine received active drug and three received placebo. There was a minimum of 7 days between each dose. Each subject received three different doses of active drug over the entire study period. Samples of whole blood (7 ml) were collected up to 48 h postdose on each study day into heparinized tubes. Following centrifugation, plasma was removed and stored frozen before analysis.

Pharmacokinetic Study in P-glycoprotein Null Mice. Dose solutions were freshly prepared on the day of use. UK-427,857 (4 mg/ml) was dissolved in water containing 5% DMSO and 0.5% 1 N HCl. Male mice (25 g, 10 per strain, wild-type fvb and mdr1a/mdr1b knockout) received oral doses of 16 mg/kg by gavage. Blood samples (0.1 ml, n = 2 per strain per time point) were collected under terminal isoflurane anesthesia from the inferior vena cava. All blood samples were transferred to lithium heparin tubes, mixed, and centrifuged. Plasma samples were stored frozen before analysis.

Radiolabeled Studies in Animals. [¹⁴C]UK-427,857 was administered orally to CD1 male mice (Charles River, UK; n = 11, 25 g, 200 mg/kg, ∼2 μCi per mouse), male (n = 6) and female (n = 6) Sprague-Dawley rats (250 g, 100 mg/kg, ∼20 μCi per rat), and male (n = 1) and female (n = 1) beagle dogs (15 kg, 5 mg/kg, ∼100 μCi per dog). All doses were prepared in 0.2 M lactate buffer, pH 3.0, at nominal concentrations of 5 to 50 mg/ml. Daily urine and feces output was collected from mice (n = 3), rats (n = 2 per sex), and dogs (n = 2) for up to 7 days depending on recovery of radioactivity. Serial blood samples were collected from the two dogs (0-48 h). Blood samples were collected at individual time points (1, 3, 7, and 24 h) from four female and four male rats (n = 1 per time point) and eight mice (n = 2 per time point) under terminal anesthesia. Feces was homogenized with water (feces/water 1:1 for rat and mouse, 4:1 for dog). Urine and fecal homogenates were stored frozen before analysis. Aliquots of whole blood were retained and remaining samples centrifuged for preparation of plasma. Whole blood, plasma, and red blood cell samples were stored frozen before analysis. Radioactivity in urine, cage washes (approximately 1 ml), and plasma (0.1-1.0 ml) was measured by liquid scintillation counting of duplicate weighed samples in 4 ml of Starscint (Canberra Industries, Meriden, CT). Radioactivity in fecal homogenate and whole blood (approximately 0.5 g) was measured by combustion of triplicate weighed samples in an Oximate Sample Oxidizer (Canberra Industries) and liquid scintillation counting of evolved ¹⁴CO₂ in 10 ml of Permafluor E+ (Canberra Industries).

Biliary and Intestinal Secretion Study in Rat. Two male rats (Sprague-Dawley, 340 g) were anesthetized with isofluorane, and bile ducts were cannulated following laparotomy. Anesthesia was maintained with isofluorane for the duration of the experiment. [³H]UK-427,857 (3 mg/kg, ∼65 μCi per rat) was administered by i.v. injection into the caudal vein in a vehicle of 5 mM HCl in DMSO/saline (5:95). Total bile output was collected at intervals up to 6 h after dose administration, and the gastrointestinal tract and contents were collected at the end of the experiment. Total radioactivity in bile (10 μl) was measured by liquid scintillation counting of duplicate aliquots in Starscint. Radioactivity in gut contents and homogenized gastrointestinal tract (1:1 w/w with water) was measured by combustion of triplicate weighed aliquots (∼200 mg), and the evolved ³H₂O was measured by liquid scintillation counting in Monophase S (Canberra Industries). Metabolite profiling of bile and gastrointestinal tract contents was performed as described for animal excreta samples.

Radiolabeled Studies in Humans. Written consent and ethical approval were obtained as detailed above for this study, which was carried out at LCG Bioscience (Cambridge, UK). Healthy male volunteers (n = 3, aged 45-65 years, weight 80-90 kg) were given single 300-mg [¹⁴C]UK-427,857 (∼48 μCi per subject) oral solution doses. Urine and feces were collected daily for 7 days, and serial blood samples were collected at time points up to 120 h postdose. Urine, feces, and blood samples were treated, stored, and analyzed for total radioactivity as previously described for animal species.

Analysis of Excreta and Plasma Samples for Metabolites. For all species, excreta samples were pooled to represent >90% of the total administered dose. For the animal species, one fecal and one urine sample per sex (where applicable) were generated, representing >90% of the total dose eliminated in that matrix. In humans, single pooled samples for feces and urine were profiled for each subject separately. In animal species, plasma samples were pooled in a time-normalized manner (Hop et al., 1998) to generate a single sample representing >80% of the total plasma AUC. Fecal homogenates were extracted with methanol (20 ml), followed by a mixture of methanol (19 ml) and 0.1 M Tris buffer (pH 6; 1 ml) and, finally, a mixture of methanol (18 ml) and 0.1 M Tris buffer (pH 9; 2 ml). The combined extracts were reduced to dryness under nitrogen at 37°C on a Tubovap (Zymark). Urine samples were centrifuged before HPLC profiling. Plasma samples were treated with 3 volumes of methanol and centrifuged, and the supernatants were reduced to dryness under nitrogen at 37°C. HPLC profiling of excreta and plasma samples was achieved with an HIRPB column (250 × 7.76 mm; Hichrom, Theale, Berkshire, UK) using a linear binary solvent gradient involving methanol/0.1 M ammonium acetate at a flow rate of 2 ml/min. For excreta samples, on-line radiochemical detection (β-Ram; LabLogic, Broomhill, Sheffield, UK) was used to monitor the drug-related components, with LAURA 3 (LabLogic) for data analysis. The drug-related components were isolated separately for identification and reduced to dryness under nitrogen at 37°C. For plasma, fractions were collected into 96-well Scintiplates (PerkinElmer Life and Analytical Sciences, Boston, MA), which were dried on a DD4 vacuum centrifuge (Genevac, Ipswich, Suffolk, UK) and analyzed on a Microbeta 1450 scintillation counter (PerkinElmer Life and Analytical Sciences). The drug-related components were initially analyzed by direct infusion mass spectrometry at a flow rate of 5 μl/min following reconstitution in methanol/water (70:30 v/v; both containing 2 mM ammonium acetate) and appropriate dilution. Subsequently, liquid chromatography-tandem mass spectrometry was used to confirm assignments by comparison with authentic standards and resolve the components of mixed regions. Mass spectrometric identification used either an API4000 or 4000QTRAP instrument (both Applied Biosystems/MDS Sciex, Foster City, CA), using TurboIonSpray in positive ion mode. The mass spectrometer was optimized primarily by variation of the declustering potential, the collision energy, the TurboIonSpray temperature, and (for the QTRAP) the linear ion trap fill time.

Caco-2 Monolayer Studies. Caco-2 cells were seeded in 24-well Falcon Multiwell plates (polyethylene terephthalate membranes, pore size 1.0 μm) at 4.0 × 10⁴ cells per well. The cells were grown as previously described (Harrison et al., 2004) and permeability studies were performed after 15 to 17 days of culture. Cells were used between passage 34 and 41. The study was initiated by the addition of UK-427,857 [25 μM in Hanks' balanced salt solution, pH 7.4] to the donor well, 250 μl to the apical chamber, or 1 ml to the basolateral chamber, with control buffer in the opposing acceptor well. The effects of P-glycoprotein inhibitors on transport in both apical to basolateral (A to B) and basolateral to apical (B to A) directions were assessed by inclusion of verapamil (100 μM) or CP-100,356 (10 μM) in both donor and acceptor well solutions. Membrane integrity was assessed at the end of the 2-h incubation using Lucifer yellow added to the apical wells at a concentration of 100 μM. All liquid handling procedures were conducted using a Tecan Genesis robot (Tecan, Durham, NC). The concentrations of UK-427,857 in each sample were determined by a HPLC system with mass spectrometric detection. The analytical column was a 40-μm Opti-Lynx Reliasil C18 (3 × 15 mm; Optimize Technologies, Oregon City, OR). The Hanks' balanced salt solution was directly injected onto the column, and salts were washed to waste by back flushing with mobile phase consisting of 90:10 water/methanol containing 0.027% formic acid. Analyte was then eluted from the column with mobile phase comprising 10:90 water/methanol containing 0.027% formic acid. Flow rate was 1.5 ml/min for loading and 1.2 ml/min for elution. UK-427,857 was detected by an API 2000 mass spectrometer with an ionspray interface (PerkinElmerSciex Instruments, Boston, MA) using specific mass transition. Quantitation was performed against a calibration line constructed over the concentration range 0.5 to 5 μM from known standards. Apparent permeability (P_app) was calculated as previously described (Harrison et al., 2004).

P-gp Binding Affinity. The apparent binding affinity of UK-427,857 for recombinant human P-glycoprotein expressed in insect cell membranes was determined as previously described (Abel et al., 2001). Apparent kinetic parameters K_m and V_max were obtained by Michaelis-Menten analysis of the data.

Plasma Protein Binding Determinations. Samples of mouse, rat, dog, and human plasma from toxicology or clinical studies were used for the determination of plasma protein binding. Composite samples were prepared for mouse (n = 12) and rat (n = 9) by pooling C_max samples according to sex and dose level. Individual C_max samples were analyzed for dog (n = 54) and human (n = 54). [¹⁴C]UK-427,857 was added to samples at a concentration that allowed accurate radiochemical detection but did not exceed 10% of the measured concentration in any sample. Aliquots of plasma samples (0.15 ml) were dialyzed (Spectropor membrane strips, 120 × 22 mm, molecular weight cutoff 12,000-14,000; Spectrum Medical Industries, Rancho Dominguez, CA) against isotonic Krebs-Ringer buffer (0.15 ml, pH 7.4) for 4 h at 37°C on an oscillating platform in a 96-well equilibrium dialysis apparatus (HTDialysis, Gales Ferry, CT; Banker et al., 2003). Following dialysis, concentrations of drug in plasma and buffer were measured by liquid scintillation counting. The free fraction in plasma was calculated from the ratio of the concentration in buffer to plasma after dialysis.

Blood to Plasma Partitioning of UK-427,857. The partitioning of UK-427,857 between red blood cells and plasma in rat and dog was determined by the addition of approximately 100 ng/ml [³H]UK-427,857 (0.1 mg/ml solution in methanol) to triplicate 3-ml aliquots of freshly collected control rat and dog whole blood. Samples were incubated at room temperature for 20 min. Triplicate weighed aliquots (ca. 100 mg) of whole blood were removed and analyzed for radioactivity by liquid scintillation counting of evolved ³H₂O after combustion by an Oximate automatic sample oxidizer. Plasma was prepared by centrifugation of the remaining whole blood and radioactivity determined in duplicate weighed aliquots (ca. 100 mg) by liquid scintillation counting. The blood/plasma partition was expressed as a ratio of concentration in blood relative to plasma. The blood to plasma partitioning of UK-427,857 in human whole blood was similarly determined using fresh whole blood from six individual donors, spiked with approximately 100 ng/ml [¹⁴C]UK-427,857, with radioactivity in whole blood determined by liquid scintillation counting of evolved ¹⁴CO₂.

Analysis of UK-427,857 in Animal and Human Plasma. The analysis of UK-427,857 in plasma from mice, rats, and dogs was carried out by solvent extraction followed by HPLC with mass spectrometric detection. The analytical procedure was validated by the analysis of quality control samples. The method involved addition of internal standard (UK-377,461, 25 ng) to plasma samples (0.1-0.5 ml) followed by the addition of borate buffer (1 ml, 0.2 M, pH 10) and tert-butyl methyl ether (2 ml). After mixing and centrifugation, the organic layer was transferred using a Genesis robot (Tecan, Maennedorf, Switzerland) into a 96-deep well block (Porvair Filtronics Ltd., Shepperton, UK) and evaporated to dryness under a stream of nitrogen using a Techne Dri-Block DB.3A sample concentrator fitted with a 96-well needle array (Fischer Scientific Ltd., Loughborough, UK). Samples were reconstituted in 200 μl of 70% methanol (v/v) in water (containing 2 mM ammonium acetate). Aliquots of these samples (180 μl) were injected onto the HPLC system. The HPLC system comprised an HS100 C18 column (5 μm, 50 × 4.6 mm; Thermo Electron, Waltham, MA) and mobile phase 90% methanol (v/v) in water (containing 2 mM ammonium acetate) at a flow rate of 1 ml/min, split 50:1 postcolumn using an Acurate flow splitter (Presearch Ltd., Hitchin, Hertsfordshire, UK). Detection was by multiple reaction monitoring for the transitions m/z 514 → 389 (UK-427,857) and 464 → 216 (UK-377,461) using a Sciex API2000 mass spectrometer with ionspray interface (PerkinElmerSciex Instruments). The limit of quantitation was 2.5 to 20 ng/ml, depending on sample volume. Quality control samples were found to be within 15% of the nominal concentration throughout the calibration range, satisfying acceptance criteria for the method.

The analysis of UK-427,857 in human plasma was performed using protein precipitation followed by turbulent flow chromatography (Chassaing et al., 2001) with mass spectrometric detection. The analytical procedure was again validated by the analysis of quality control samples. The method involved addition of internal standard (UK-462,015, the D₅ analog of UK-427,857; 10 ng) to plasma samples (0.5 ml), followed by the addition of cold acetonitrile (1 ml, approximately 4°C) in a 96-deep well block. After mixing and centrifugation, an aliquot of the supernatant (0.89 ml) was transferred using a Multiprobe MPII robot (PerkinElmer Life and Analytical Sciences) into a separate 96-deep well block and evaporated to dryness under a stream of nitrogen at 50°C. Samples were reconstituted in 250 μl of 25% acetonitrile (v/v) in water. Aliquots of these samples (200 μl) were injected onto the HPLC system. The HPLC system comprised an HTLC turbulent flow system (model 2300; Cohesive Technologies, Franklin, MA) fitted with an HTLC Turbo C18 extraction column (50 μm, 50 × 1 mm; Cohesive Technologies) and an HTLC HiRes C18 analytical column (10 μm, 33 × 2.1 mm; Cohesive Technologies). Samples were loaded onto the columns using a mobile phase of 1.3 mM trifluoroacetic acid in water and eluted with a mobile phase comprising 2 mM ammonium acetate in 45% acetonitrile, 45% methanol, and 10% water; a flow reading of 1.0 was obtained on all lines. Analytical column eluate was split 50:1 postcolumn using an Acurate flow splitter. Detection was by multiple reaction monitoring for the transitions m/z 514 → 389 (UK-427,857) and 519 → 394 (UK-462,015) using a Sciex API2000 mass spectrometer with ionspray interface. The limit of quantitation was 0.25 ng/ml. Quality control samples were found to be within normal acceptance criteria for bioanalytical methods (Shah et al., 2000).

Results

Single-Dose Intravenous Pharmacokinetics of UK-427,857 in Rat and Dog, and Plasma Protein and Blood Binding. Single-dose intravenous pharmacokinetic data in rat and dog are shown in Table 1. Half-life values after intravenous administration are short in both rat and dog, at 0.9 and 2.3 h, respectively. The blood/plasma partition of UK-427,857 was determined at 1.1 and 0.9 for rat and dog, respectively. UK-427,857 exhibits high clearance in the rat, with a value close to liver blood flow (blood clearance calculated at 67 ml/min/kg). Clearance in the dog is moderate, at around 50% of liver blood flow in this species (blood clearance calculated at 23 ml/min/kg). UK-427,857 demonstrated large volumes of distribution in both rat and dog at 6.5 and 4.3 l/kg, respectively. UK-427,857 was moderately bound to plasma protein in all species, with mean values of 58.0% (S.D. = 4.3%, n = 12) in mouse, 51.0% (S.D. = 2.1%, n = 9) in rat, 63.7% (S.D. = 7.3%, n = 54) in dog, and 75.5% (S.D. = 4.3%, n = 54) in human. In human blood, the mean blood/plasma partition value (n = 6) was 0.59 (S.D. = 0.09), with a range from 0.51 to 0.65 in the individual samples.

Single-Dose Oral Pharmacokinetics of UK-427,857 in Rat, Dog, and Human. Single-dose oral pharmacokinetic data for rat, dog, and human are shown in Table 1. For comparative purposes, human pharmacokinetic data are presented on a mg/kg basis (with an assumed body weight of 70 kg) at dose levels of 30 and 300 mg. Systemic plasma concentrations were below the limit of detection (<20 ng/ml) of the analytical method in rat at an oral dose of 3 mg/kg, whereas a C_max of 55 ng/ml was observed at the higher dose of 10 mg/kg. Systemic bioavailability at this dose was estimated at 6% by comparison with dose-normalized intravenous AUC. Analysis of portal vein plasma samples from rats at doses of 3 and 10 mg/kg provided AUC values of 132 and 669 ng · h/ml, respectively. This allowed an estimation of the extent of absorption from the product of plasma clearance (after intravenous administration) and the portal vein AUC (Kwon and Inskeep, 1996), which indicated that 20 and 30% of the dose was absorbed unchanged from the gastrointestinal tract at doses of 3 and 10 mg/kg, respectively. In dogs, absolute bioavailability values of 42 and 40% were obtained after oral doses of 1 and 2 mg/kg. Absorption in dog was fairly rapid, with mean T_max values observed at 1.5 and 0.75 h. Elimination half-life values were similar to those obtained after intravenous doses. Pharmacokinetic data in humans at representative doses of 30 and 300 mg in Table 1 demonstrate a markedly superproportional increase in dose-normalized exposure over this dose range. For the 10-fold increase in dose, C_max increases 40-fold and AUC increases 20-fold. This is accompanied by an apparent increase in the rate of absorption, with T_max decreasing from 2.9 to 1.6 h at doses of 30 and 300 mg, respectively.

Multiple-Dose Pharmacokinetics in Rat and Dog. For both rats (males only) and dogs, AUC and C_max data were obtained after the first dose and at steady state (day 19 in dogs and day 23 in rats). No sex differences were observed in dogs; therefore, data from male and female animals were combined. In both species, systemic exposure was broadly similar on day 1 and at steady state. The relationship between dose and systemic exposure (AUC and C_max) is shown in Fig. 2 using data from day 1 of these studies.

Caco-2 Permeability and P-gp Affinity. Permeability of UK-427,857 across Caco-2 monolayers in the apical to basolateral (A to B) direction was limited, with P_app values of <1 × 10^-6 cm/s (n = 3). Markedly higher permeability was observed in the basolateral to apical (B to A) direction, with a mean P_app value of 12 × 10^-6 cm/s (n = 3), providing an efflux ratio (B to A/A to B) of >10, demonstrating polarized transport in the Caco-2 system. Such a profile is indicative of transporter-mediated efflux (Zhang and Benet, 1998). Verapamil and CP-100,356, which are both known inhibitors of P-glycoprotein (Doige and Sharom, 1992; Wandel et al., 1999), markedly reduced the efflux ratio of UK-427,857 transport in Caco-2 cells (Fig. 3). The apparent K_m for UK-427,857 in the ATP hydrolysis assay for P-gp binding was 37 ± 6.4 μM (n = 5), with V_max of 55 ± 3.4 nmol/mg/min.

Pharmacokinetics of UK-427,857 in P-gp Knockout Mice. After oral doses of 16 mg/kg UK-427,857 to wild-type fvb mice, the average C_max (n = 2) was 536 ng/ml and the AUC derived from average data was 440 ng · h/ml. The elimination half-life was 0.7 h. In mdr1a/1b knockout animals, following the same dose, the average C_max and AUC were increased by 108% and 183%, respectively, at 1119 ng/ml and 1247 ng · h/ml. The elimination half-life was 1.0 h.

Excretion Studies. The recoveries of radioactivity in urine and feces following single oral doses of [¹⁴C]UK-427,857 to animal species and humans are shown in Table 2. The predominant route of excretion in all species is fecal, accounting for between 72 and 94% of recovered radioactivity. The extent of urinary excretion ranged from 4.5% in male rat to 19.6% in human. Excretion of dosed radioactivity occurred rapidly in rodent species, with >95% of the recovered radioactivity obtained within the first 24 h. Excretion occurred more slowly in dog and humans, but in all cases, >90% of the recovered radioactivity was obtained within 96 h. In rats and mice, only 0.1 and 0.2%, respectively, of dosed radioactivity was recovered in the carcasses at 96 h postdose (by alkali digestion and scintillation counting).

Urinary and Fecal Metabolites. Methanolic extraction of fecal homogenates and centrifugation of urine resulted in the recovery of >80% of the drug-related material in all cases. HPLC profiling of excreta samples showed that in all species, a significant fraction of the dose was eliminated unchanged (see, for example, human feces and urine profiles, Fig. 4, panels A and B), with parent compound representing between 33% and 79% of the total excreted radioactivity (human, n = 3; and rat, n = 4, respectively). A high degree of commonality was observed in the metabolism of UK-427,857 across all species. The major metabolic pathways in humans were oxidation of the triazole moiety, oxidation in the difluorocyclohexyl ring, and N-dealkylation adjacent to the tropane ring, whereas in some animal species, aromatic hydroxylation at the para position of the phenyl ring was a major pathway (Fig. 5). The major excreted metabolites in humans were a product of hydroxylation of the methyl group of the triazole moiety (Met 8), accounting for 10% of the total dose (mean of three subjects), four products of mono-oxidation in the difluorocyclohexyl ring (Mets 4-7; together accounting for 29%), and the secondary amine resulting from N-dealkylation adjacent to the tropane ring (Met 2; UK-408,027; 7%) (Fig. 5). All were identified in the excreta of the toxicology species. The structures of Met 2 and Met 3 were further confirmed by comparison of mass spectral data with authentic standards.

Circulating Metabolites. Protein precipitation of plasma by treatment with methanol resulted in the recovery of >80% of the drug-related material in all cases. In human plasma, the major components were unchanged UK-427,857, accounting for 42% of the plasma AUC (mean of three subjects), the secondary amine resulting from N-dealkylation adjacent to the tropane ring (Met 2, UK-408,027; 22%), and an analog of the amine involving oxidation of the methyl group of the triazole ring (Met 1; 11%). A number of minor circulating components were also observed in humans (Fig. 4, panel C). Unchanged UK-427,857 was also the major circulating component in animal species (mean values, 58% mouse, 67% rat, and 58% dog), whereas the secondary amine resulting from N-dealkylation (Met 2) was a major plasma metabolite (>5% circulating radioactivity) in all species.

Biliary and Intestinal Metabolites in Rat. Following intravenous administration of [³H]UK-427,857 to bile duct-cannulated rats, 64% of the administered dose was excreted in bile in 6 h. A further 15% was associated with the gastrointestinal tract (3.5%) and its contents (11.5%). Parent compound was the major component present in bile, accounting for approximately 40% of dosed radioactivity. Profiling of the radioactivity in gastrointestinal tract contents showed only a single component present that corresponded to parent compound.

Discussion

UK-427,857 has high clearance in rat (estimated in excess of 80% liver blood flow) and moderate clearance in dog (estimated at around 50% of liver blood flow). In the dog, where oral bioavailability is in excess of 40%, absorption would appear to be near-complete, whereas in the rat, exposure in the hepatic portal vein indicates absorption of only around 20 to 30%. Such species differences in intestinal availability between rat and dog are quite common (van de Waterbeemd et al., 2001) and may reflect the larger aqueous pores present in the gastrointestinal tract of dogs (He et al., 1998) facilitating absorption of UK-427,857 via the paracellular route. Given the good aqueous solubility of UK-427,857, oral formulation changes were not considered to impact the extent of absorption. Aqueous pores in the gastrointestinal tract of humans are generally considered more similar in size to rat than to dog. At the time of candidate nomination, UK-427,857 was expected to have moderate absorption in human, similar to the rat, and moderate clearance, similar to the dog. Experience with previous compounds showing both variability in intestinal availability between species and apparent affinity for P-glycoprotein indicated the potential for nonlinear pharmacokinetics in humans.

The oral dose-exposure relationship in rat and dog is shown by the C_max/Dose and AUC/Dose values for day 1 toxicology data and single-dose data in Fig. 2, and contrasted with single oral-dose values for human. On this dose-normalized basis, exposure in rat is clearly much lower (approximately 1 order of magnitude) than in both dog and human, and is in keeping with the higher clearance observed in this species after intravenous administration. Excluding the lowest dose in rats (10 mg/kg), where only partial pharmacokinetic profiles were obtained, the dose-exposure relationship is broadly linear in terms of AUC but shows some subproportionality in terms of C_max at the highest dose (1500 mg/kg). In the dog, dose-normalized AUC and C_max show limited variability (2- to 3-fold) across the dose range 1 to 150 mg/kg. In contrast, the clinical dose-exposure relationship in humans shows marked nonproportionality in terms of both AUC and C_max, with superproportional increases observed up to doses of 300 mg. Dose-normalized exposure is approximately 10-fold higher at a dose of 300 mg compared with a dose of 3 mg. Based on an estimated fluid volume of 500 ml (Dressman et al., 1998) in the small intestine, a dose of 3 mg of UK-427,857 would result in a concentration of about 12 μM, which is below the P-glycoprotein apparent K_m of 37 μM and, hence, would not be expected to saturate P-glycoprotein-mediated efflux. This superproportional dose-exposure relationship is similar to that previously observed for a number of compounds including the NK2 antagonist, UK-224,671 (Beaumont et al., 2000), the α-antagonist, UK-294,315 (Harrison et al., 2004), and the phosphodiesterase 5 inhibitor, UK-343,664 (Abel et al., 2001). All of these compounds have arisen out of drug discovery programs that have either targeted enzyme inhibitors with subtype selectivity or are nonaminergic 7-transmembrane spanner receptor antagonists. The physicochemical requirements of such targets requires the incorporation of H-bonding functionalities, which, in itself, provides molecules with properties that are not considered ideal in terms of absorption and pharmacokinetics (Lipinski et al., 1997). Indeed, it would appear that these drug discovery programs tend to result in compounds of borderline membrane permeability that are subject to transport by P-glycoprotein and hence show nonproportional oral pharmacokinetics as a result of saturation of the efflux transport when given at relatively high doses (Harrison et al., 2004). The ability of P-glycoprotein to affect the membrane permeation of UK-427,857 has been demonstrated in the Caco-2 cell model using inhibitors of the transport protein. The role of P-glycoprotein is also supported by the studies in P-glycoprotein knockout mice, in which the absence of the transporter results in a modest increase in systemic exposure. It is interesting to note the similar physicochemical properties of this series of compounds that share these absorption characteristics (Table 3). All possess a number of H-bond donors or acceptors, are moderately lipophilic (Log D_7.4 values 1.4-2.6) and basic (calculated pKa 8.7 to 10.6), and have molecular weights above 500. Although other compounds of similar physicochemical properties have not shown nonlinear absorption profiles, it would appear that a high molecular weight results in an increased risk of this phenomenon. Indeed, a molecular weight above 500 has been shown to be associated with poorly absorbed compounds (Lipinski et al., 1997), and such relatively large molecules may be considered relatively poor membrane permeators and, hence, more liable to be affected by P-glycoprotein. The limited permeability of UK-427,857 through cell membranes is supported by the studies in Caco-2 cells where, even in the presence of transport inhibitors, the rate of permeation of the compound is relatively slow. The size of UK-427,857 and the other molecules cited (mol. wt. >500) would indicate that the transcellular pathway will not be a significant route of absorption in humans. It would appear that restricted membrane permeability per se is more critical in the sensitivity of these compounds to efflux than their affinity for the transport protein which, itself, appears to have a wide structure-activity relationship (Stouch and Gudmundsson, 2002) and is thus able to bind to a variety of xenobiotics. These moderately lipophilic bases meet the structural requirements previously assigned to P-glycoprotein substrates, possessing, as they do, multiple hydrophobic regions and H-bond acceptor functions (Ekins et al., 2002). The failure of the P-glycoprotein inhibitors to completely inhibit the polarized transport of UK-427,857 in Caco-2 cells may indicate a role of additional transport proteins in the efflux of this compound. Current knowledge on the clinical significance of such transporters prevents any conclusion as to their potential relevance at this time.

It would appear, therefore, that neither rat nor dog provides pharmacokinetic data that are predictive of the nonproportional oral pharmacokinetics of UK-427,857 in human. It is likely that multiple factors contribute to the species differences that are observed, including the generally higher permeability of dog gastrointestinal tract relative to other species, resulting in consistently high absorption throughout the dose range. It could be argued that the dose range in dog does not extend low enough to make this conclusion; however, such data are not available for UK-427,857, and previous experiences would indicate that no nonproportionality would be observed. In rat, incomplete bioavailability is apparent, based on hepatic portal vein exposures at relatively low oral doses. Gut wall metabolism is not thought to contribute significantly to the incomplete intestinal availability, given the overall low abundance of metabolites in the rat; incomplete absorption is therefore considered most likely. The high systemic clearance of compound in rat makes accurate assessment of bioavailability complicated, but there is no evidence that high oral doses saturate transporter (including P-glycoprotein) efflux, with pharmacokinetics appearing to remain reasonably proportional over an extended range. Overall, the most direct and valuable preclinical indicator of nonproportional clinical pharmacokinetics appears to be the in vitro data obtained in Caco-2 cells. Although neither rat nor dog is entirely predictive of the absorption of UK-427,857, it should be noted that the animal pharmacokinetic data in these two animal species were essential to demonstrate that the compound possessed oral bioavailability. In vitro data alone indicated very poor membrane permeability, which, in other drug discovery programs, might have been taken as an indication of no oral bioavailability potential (Walker et al., 2001). Only the combination of in vitro and in vivo data provide good representation of the bioavailability properties of UK-427,857.

UK-427,857 undergoes a degree of metabolism in animals or humans; however, parent compound is the major circulating and excreted component in all species. The metabolism of UK-427,857 that is observed involves oxidation at a number of positions in the molecule, resulting in hydroxylation or N-dealkylation. Additional studies (not reported here) have indicated a role for cytochrome P450 in the metabolism of UK-427,857. All biotransformation pathways observed in humans are also observed in the toxicology species reported here. Limited renal excretion of UK-427,857 is observed in all animal species (2-6% of the dose), with highest levels of unchanged drug in urine in humans accounting for 8% of the administered dose. High levels of unchanged drug in the feces of all species (25% human, 74% rat, 36% mouse, and 39% dog) may in part reflect unabsorbed dose passing through the gastrointestinal tract unchanged. Data from the bile duct-cannulated rat study demonstrate that a high amount of unchanged drug is secreted via the bile. This study also provides evidence for direct secretion of drug into the gastrointestinal tract by the presence of parent compound in the gut contents after intravenous administration when bile has been diverted. Secretion of xenobiotics from the systemic circulation into the gut lumen across the gut wall may be due to efflux by transporters including P-glycoprotein. The ability of the P-glycoprotein inhibitor, verapamil, to reduce the systemic clearance of the β-blocker, talinolol, has provided evidence of this phenomenon (Spahn-Langguth et al., 1998). In the case of talinolol, there is evidence to show that this process of small intestinal secretion also occurs in humans (Gramatté et al., 1996). Excretion of UK-427,857 in bile may again involve P-glycoprotein-mediated transport, since this protein has been shown to be associated with the canalicular membrane (Müller and Jansen, 1997). The high systemic clearance and extensive biliary excretion observed in rat would also indicate that uptake into the liver may be facilitated by carrier proteins to achieve the high hepatic extraction observed in this species. A number of transporter proteins have been identified that might perform this role; in particular, the sinusoidal membrane transporters Oatp1 and Oatp2 in the rat have been shown both in vitro and in vivo to be involved in active uptake of drug substrates into the liver (Ayrton and Morgan, 2001). It is interesting to speculate as to whether the incorporation of metabolic stability into drug design tends to result in molecules more susceptible to transporter-mediated processes or whether the property of relative metabolic stability makes the transporter-mediated processes more visible. In the case of UK-427,857, transporter-mediated processes form an important element of drug disposition and would appear to be a major factor in the species differences that are observed. The clinical consequences, in terms of drug interaction potential, at the level of drug absorption will depend upon the clinical dose and the associated degree of saturation of the transport protein.