![[Feedback]](/icons/banner/feedbackACT.gif){kind=icon}
![[For Subscribers]](/icons/banner/subscriberACT.gif){kind=icon}
![[Archive]](/icons/banner/archiveACT.gif){kind=icon}
![[Search]](/icons/banner/searchACT.gif){kind=icon}
![[Contents]](/icons/banner/tocACT.gif){kind=icon}
|
|
|
|
|
||
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Department of Physical and Metabolic Science (I.J.M., R.J.L., M.A.B., I.G.B., C.A.M., R.J.R.) and Department of Medicinal Chemistry (B.S.), AstraZeneca R&D Charnwood, Loughborough, United Kingdom
(Received January 9, 2006; accepted June 7, 2006)
 |
Abstract |
|---|
 Top Abstract Materials and MethodsResultsDiscussionReferences |
|---|
; Burchell and Coughtrie, 1989; Burchell et al., 1991; Clarke and Burchell, 1994), located principally in the hepatic endoplasmic reticulum. Many drugs are known to undergo glucuronidation either directly, or after phase I metabolism (Miners and Mackenzie, 1991) resulting in more hydrophilic species that are more easily excreted in bile or urine. Moreover, species differences in the stereo- and regios-elective glucuronidation of drugs have been documented, and these reactions may be catalyzed by UGT isoforms expressed in liver and other tissues, such as the kidney (Soars et al., 2001).
|
|
Materials and Methods |
|---|
|
TopAbstractMaterials and MethodsResultsDiscussionReferences |
|---|
Dog in Vivo. AZ11939714 was administered to an anesthetized male beagle dog as part of a pharmacological investigation into the antithrombotic properties of this compound. AZ11939714 was dissolved in dimethylacetamide to which was added polyethylene glycol 400 and distilled water (respective proportions 40:40:20) to yield a 3.33 mg/ml solution. This solution was administered as an intraduodenal bolus at 1.5 ml/kg (5 mg/kg). Bile was sampled from the gall bladder at 2 h postdose.
Microsome Preparation. Rat (male Sprague-Dawley) and dog (male beagle) hepatic microsomes were prepared according to standard methodology (Soars et al., 2002) and stored at –70°C before use. Human hepatic microsomes (pooled from 10 donors) were purchased from the International Institute for the Advancement of Medicine (Exton, PA).
Microsomal Oxidation Assay. AZ11939714 was added from a concentrated stock (0.3 mM in DMSO, 1% v/v) to provide a final concentration of 3 µM (1% v/v vehicle) in a suspension of microsomal protein (2 mg/ml) in a suitable vial. After a 2-min preincubation at 37°C, the cofactor NADPH was added (final concentration of 1 mM) and the reactions were allowed to proceed.
At appropriate time points up to 60 min, aliquots (100 µl) were removed from the incubation and added to 2 to 3 volumes of acetonitrile to terminate the reaction and denature the microsomal enzymes. Control incubations were also conducted in which NADPH or compound had been omitted. Once the incubations had been quenched, the samples were shaken for 5 min and then centrifuged for 15 min at 3000 rpm and 4°C. The supernatants were removed and analyzed by HPLC.
Microsomal Glucuronidation Assay. Rat and dog microsomes were thawed at room temperature and diluted with a solution of Brij 58 in 0.1 M phosphate buffer (pH 7.4 at 37°C) to give 5 mg/ml protein and 0.2 mg Brij 58/mg protein. A preliminary study (unpublished data) had established that the rate of glucuronidation of AZ11939714 in human hepatic microsomes was not enhanced by addition of this detergent. Human microsomes were therefore diluted to 5 mg/ml protein with phosphate buffer only. The diluted microsomes were left on ice for 20 min before incubation.
Incubations were carried out in glass vials in a shaking water bath at 37°C. To each vial was added 0.39 ml of phosphate buffer, 0.1 ml of magnesium chloride (100 mM; final concentration 10 mM), and AZ11939714 from a stock solution in DMSO to give a final concentration of either 20 µM (for metabolite identity work) or 3 µM (for intrinsic clearance measurements). The resultant concentration of DMSO was kept at or below 1% (v/v). The vials were incubated for 5 min before the addition of 0.4 ml of dilute microsomes (final protein concentration 2 mg/ml). After a further 5-min incubation, the reaction was initiated with 0.1 ml of UDPGA (60 mM in phosphate buffer; final concentration 6 mM). Samples (0.1 ml) were taken at various times up to 60 min and also 120 min, and quenched in 0.2 ml of ice-cold methanol. The quenched samples were stored on ice before centrifugation (3000g for 15 min at 4°C), and the supernatant was removed for direct analysis by HPLC and HPLC-MS. Control incubations were also performed from which UDP-glucuronic acid, compound, or microsomes were omitted.
For HPLC-NMR, a scaled-up incubation (5 ml total) was carried out for 120 min and quenched with 10 ml of ice-cold methanol. After centrifugation, as above, the supernatant was concentrated to approximately 1 ml by rotary evaporation. Dog metabolite samples were analyzed at this stage. For rat and human, aliquots (0.1 ml) of this concentrate were subjected to HPLC (see below), and the peaks corresponding to the glucuronide were collected and pooled. The pooled samples were taken to dryness by rotary evaporation, dissolved with sonication in ca. 500 µl of D2O, and injected onto the HPLC-NMR system. The amount of compound-related material on column was approximately 30 µg in each case.
Hepatocyte Assays. Hepatocytes were prepared from male Sprague-Dawley rats, beagle dogs, or an isolated lobe of human liver (obtained from local hospitals with ethical approval) by a two-step in situ collagenase perfusion method (McGinnity et al., 2004) and stored on ice, before incubation.
AZ11939714 was added from a concentrated stock (0.3 mM in DMSO) (1% v/v vehicle) to 1.5 ml of suspension buffer in a suitable vial. Cells (1.5 ml) at a concentration of 4 x 106 cells/ml (viability >85% by trypan blue exclusion) were placed in a separate vial, and both vials were preincubated in a water bath at 37°C. After 5 min of preincubation, the buffer and compound were added to the cells to give a final cell concentration of 2 x 106 cells/ml, and the reactions were allowed to proceed for up to 90 min.
At appropriate time points up to 90 min, an aliquot (100 µl) was taken out of the incubation mix and added to 2 to 3 volumes of acetonitrile to terminate the reactions and denature the hepatocytes. Control incubations were also conducted in which cells or compound had been omitted. Once the incubations had been quenched, the samples were shaken for 5 min and then centrifuged for 15 min at 3000 rpm and 4°C. The supernatants were removed and analyzed by HPLC.
Data Analysis. For NADPH-supplemented microsomal incubations and hepatocyte suspensions, the resultant peak areas of AZ11939714 were entered into an Excel spreadsheet and a plot of ln (residual concentration) versus time was produced. The treatment of the data was then akin to a one-compartment, pharmacokinetic model. As dose/C0 gives a term for the volume of the incubation (expressed in ml/mg microsomal protein or ml/106 cells) and the elimination rate constant k = 0.693/t1/2, an equation expressing CLint in terms of t1/2 could be derived: CLint = (Volume x 0.693)/t1/2.
For UDPGA-supplemented microsomal incubations, the concentration of the glucuronide in each sample (verified by mass spectrometry) was calculated by reference to the peak area of the parent compound in the t = 0 min sample (equivalent to the starting concentration of the incubation). It was assumed that the extinction coefficient of the glucuronide was similar to that of the parent compound. The linear portion of the concentration versus time plot for the glucuronide yielded a slope equivalent to the metabolic rate. This rate was divided by the starting concentration of the incubation to yield a value for CLint (i.e., CLint = v/S). The CLint estimates obtained were then used to project the in vivo clearance using the well stirred model as detailed previously (Soars et al., 2002; McGinnity et al., 2004).
HPLC Analysis. HPLC analysis of microsomal incubates was carried out using an HP1100 chromatography system (Agilent Technologies, Stockport, UK). The column was a Symmetry C8, 5 µm, 50 x 4.6 mm (Waters, Watford, UK) at 40°C fitted with a 15-mm guard column (same specification). Mobile phase A was methanol, and mobile phase B was 0.025% (w/v) ammonium acetate. A gradient was run commencing at 30% A, linear to 80% A over 11 min, switching to 100% A for 2 min, and then returning to initial conditions. The flow rate was constant at 1.5 ml/min. The reinjection time was 15 min. Detection was by UV absorbance at 285 nm.
HPLC-MS. HPLC-MS was carried out using an HP1100 chromatography system linked to a Finnigan MaT TSQ7000 mass spectrometer (Thermo Electron Corporation, Hemel Hempstead, UK) operating in atmospheric pressure chemical ionization mode (vaporizer temperature, 550°C; capillary temperature, 250°C). Q1 was scanned from 160 to 900 atomic mass units (scan time, 1 s).
For bile samples, a Symmetry C8, 5 µm, 75 x 4.6 mm column at 40°C was used. Mobile phase A was methanol and mobile phase B was 0.05 M ammonium acetate. A gradient was run commencing at 8% A from 0 to 12 min, linear to 80% A at 21 min, switching to 100% A at 24 min, holding at 100% A until 27 min, and then returning to initial conditions. The flow rate was constant at 1.5 ml/min. The reinjection time was 30 min. Detection was by UV absorbance at 285 nm.
For microsomal samples, a Symmetry C8, 5 µm, 50 x 4.6 mm column fitted with a C8 guard column at 40°C was used. Mobile phase A was methanol, and mobile phase B was 0.025% (w/v) ammonium acetate. The gradient was initially 5% A, linear to 90% A over 7 min, switching to 100% A at 7.1 min, and holding at 100% A until 8.5 min, and then returning to initial conditions. The reinjection time was 10 min. Detection was by UV absorbance at 285 nm.
HPLC-NMR. All samples and spectra were run on a 500 MHz Varian Unity Inova Spectrometer (Varian, Inc., Palo Alto, CA) fitted with an HPLC-NMR system comprising a Varian 9012 pump, 9050 UV detector, and HPLC-NMR Analyte Collector. A Jones Chromatography 7600 series solvent degasser and 7971 Column Heater (Grace Vydac, Lakewood, CA) were also used. The flow probe had a total volume of 110 µl and an active volume of 60 µl. Data were collected on the peaks of interest using the stopped-flow method and WET solvent suppression (Smallcombe et al., 1995) to suppress the resonances from acetonitrile and residual protonated water. Suppression was achieved using "seduce" pulses of approximately 20-ms duration. Referencing in all spectra was to acetonitrile at 2.00 ppm. Proton spectra were run using a 50° tip angle and 2.1 s repetition rate. Probe temperature (usually 10°C) was adjusted between 5 and 35°C to give optimum separation between the residual water signal and the protons of interest (in acetonitrile, the glucuronide anomeric proton and protons 1 and 2 on the cyclopentane ring resonate close to the position of water). Total acquisition times were typically several hours (see Results). TOCSY and ROESY spectra with WET solvent suppression were run using mixing times of 50 ms and 350 ms, respectively.
HPLC was performed using a Symmetry C8 column (4.6 x 75 mm, 3.5-µm packing) at 40°C. Mobile phase A was 80% acetonitrile in D2O; mobile phase B was 5% MeCN in D2O. Mobile phase C was 0.2 M phosphate buffer in D2O (pH 7.4). The gradient was 35% A from 0 to 1 min increasing linearly to 55% over the next 7 min, then increasing linearly to 90% A at 11.5 min, before returning to initial conditions. Mobile phase C was used at 10% throughout and the flow rate was 0.7 ml/min. The reinjection time was 15 min. Detection was by UV absorbance at 285 nm.
|
Results |
|---|
|
TopAbstractMaterials and MethodsResultsDiscussionReferences |
|---|
HPLC-NMR. Having identified a glucuronide of AZ11939714 in dog bile by HPLC-MS, HPLC-NMR of the untreated bile sample was used to identify the site of glucuronidation. A small volume of bile (10 µl) was injected directly onto the HPLC-NMR system and data were collected on the peak eluting at 6.36 min using the stopped-flow method. The amount of material on column was estimated as 10 µg (UV peak area). Proton (3.5-h) and COSY (47-h) spectra were acquired and compared with spectra of the parent compound under similar solvent conditions. The bile spectra showed contamination with bile acids, exhibited as proton signals between 0.5 and 2.0 ppm. Despite this interference, protons 1, 2, and 5 of the cyclopentane ring could be assigned with confidence, and protons 3 and 4, and the 3-hydroxymethyl proton were assigned tentatively (proton numbering shown in Fig. 1).
To have greater confidence in these assignments, a bile sample was partially purified by making three 15-µl injections on the same HPLC system, collecting the eluate at 6.36 min, and reinjecting the entire sample after reducing the organic content under a stream of dry nitrogen. The peak was again examined using the stopped-flow method, and proton (3-h) and TOCSY (20-h) spectra were acquired. This sample was cleaner and also contained a greater amount of metabolite, allowing the chemical shifts of all the key protons to be assigned with greater confidence.
Hepatic Microsomal Samples. HPLC and HPLC-MS. Under the conditions used, AZ11939714 eluted with a consistent retention time of 9.40 min. The chromatogram of the 60-min rat microsomal sample showed two closely eluting products (7.52 and 7.78 min) not present in the 0-min sample. Human microsomes produced a single metabolite with retention time identical to that of the earlier-eluting peak from rat microsomes. Dog microsomes produced a single metabolite the retention time of which (7.64 min) did not match either of the rat products. These differences, although small, were consistent over a number of incubations and suggested quite distinct routes of metabolism in the different species studied. The microsomal products from all three species were confirmed as glucuronides by HPLC-MS. All of the products gave clean atmospheric pressure chemical ionization spectra containing essentially two ions: the expected [MH+ + 176] for a glucuronide at m/z 633 and the aglycone fragment at m/z 457 corresponding to AZ11939714 itself.
|
The HPLC retention times of the rat and human microsomal glucuronides had suggested that these were different from that produced by the dog. For each sample, the HPLC peaks corresponding to the glucuronide metabolites were collected in the flow probe using the stopped-flow method, and proton (1 h) and COSY (8 h) spectra were acquired. Although the rat microsomes produced a mixture of two glucuronide metabolites, these were not separated under the HPLC and column loading conditions adopted, and the spectra showed a mixture of the two metabolites. Data from the proton and COSY spectra and comparison with parent allowed the chemical shifts of the protons around the cyclopentane ring and on the glucuronide residue to be identified. Partial proton spectra of the three glucuronides showing the key cyclopentane and glucuronide regions of the spectrum are shown in Fig. 2 and compared with the spectrum for parent. Comparison of the NMR chemical shifts (Table 1; also discussed below) suggested that the human microsomes formed the 3-hydroxymethyl glucuronide of AZ11939714. In contrast, the rat microsomes produced a mixture of the glucuronides at the 2-position and the 3-hydroxymethyl position in an approximate ratio of 2:1.
|
ROESY Confirmation of Sites of Glucuronidation. To confirm the sites of glucuronidation inferred from the proton spectra, ROESY spectra (60 h) were run on each sample using the stopped-flow method.
In each case, the ROESY spectrum showed a cross-peak between the anomeric proton of the glucuronide and the proton on the cyclopentane ring of AZ11939714 closest to the site of attachment. Given that we isolated the three possible glucuronide isomers, this experiment provided conclusive evidence of the close proximity of the various protons and confirmed the sites of glucuronidation inferred from proton spectra.
In Vitro-in Vivo Scaling. Incubations using hepatocytes from rat and dog were used to verify in vitro-in vivo scaling as described previously (Soars et al., 2002). Interestingly, for this class of lipophilic, neutral compounds, application of the well stirred model (blood/plasma = 1) and assuming fublood was equivalent to the unbound fraction in the incubation (fuinc) provided a reasonable estimate of clearance observed in pharmacokinetic studies in both rat and dog (Table 2). A more complete analysis has been summarized previously (Davis and Riley, 2004).
|
For this lead series, early observations indicated that for some compounds, a mixture of oxidative metabolism and glucuronidation was important. Therefore, routine analysis of metabolic stability in human in vitro systems relied on the use of separate microsomal incubations supplemented with cofactors for cytochrome P450 oxidation (NADPH) or glucuronidation (UDPGA). The application of the microsomal glucuronidation screen was similar to that described by Bouska et al. (1997). As shown in Table 3, these data were also scaled to provide a projection of the likely in vivo human CLb. Data generated in the glucuronidation assay were modeled as detailed earlier (Soars et al., 2002). These projections were later verified using freshly prepared suspensions of human hepatocytes when they became available. Table 3 also demonstrates the concordance between these test systems for AZ11939714.
|
|
Discussion |
|---|
|
TopAbstractMaterials and MethodsResultsDiscussionReferences |
|---|
), such as capillary NMR (Lewis et al., 2005) and HPLC-solid-phase extraction-NMR (Sandvoss et al., 2005), have brought improvements to sensitivity in metabolite identification. Despite this, much metabolite identification work is performed close to the sensitivity limits, and it is often only feasible to obtain proton NMR spectra or simple proton-proton correlations within a reasonable time frame. The degree of information present in a simple one-dimensional proton spectrum is often over-looked, and we were therefore keen to investigate whether proton chemical shift data alone were sufficient to determine the sites of glucuronidation. In the case of the three isomeric glucuronides, assignments of the proton spectra were made largely by a chemical shift and peak multiplicity comparison with the assigned spectrum for the parent molecule. COSY or TOCSY spectra served to confirm these assignments. Table 1 shows the chemical shifts determined this way, and Fig. 3 is a visual way of representing the changes in chemical shift (
) for the key protons in the cyclopentane ring compared with the parent molecule. The radius of the shaded circle is proportional to the absolute value (usually positive) of (for the diastereotopic protons, the largest shift change is shown). The results give a clear indication of the identity of the regioisomer, with an observed downfield shift of ca. 0.3 ppm at the site of attachment of the glucuronide, and smaller shifts (ca. 0.2 ppm) at adjacent positions. In the case of the 2-glucuronide (major metabolite from rat), the at the 2-position was somewhat smaller (0.18 ppm), but it was still the largest observed in the molecule and, coupled with the small shift (0.06 ppm) at the 5-position, confirms the identity. In the case of the human (3-hydroxymethyl) glucuronide, the diastereotopic separation of the two 3-hydroxymethyl protons was also increased significantly from ca. 0.05 ppm in the parent to 0.33 ppm in the glucuronide. For the other isomeric glucuronides, the separation of these protons did not change significantly on glucuronidation. This indicates the proximity of the additional chiral centers present in the glucuronide moiety and provides further evidence that the site of glucuronidation in human microsomes is at the 3-hydroxymethyl.
|
As anticipated, the metabolite spectra obtained from the hepatic microsome samples were cleaner than those obtained from bile because of reduced interference of the bile acids mentioned previously. Bile is a complex matrix, and most reported NMR studies on metabolites in bile involve multiple purification steps; for example, prepurification followed by preparative chromatography and NMR (Desmoulin et al., 2003) or HPLC-solid-phase extraction-NMR (Sandvoss et al., 2005). However, it was still possible to identify sufficient signals in the proton spectrum to be confident of the site of glucuronidation. This study therefore shows that useful information can still be obtained even without complete and exhaustive chromatographic separation.
Early elucidation of the sites most susceptible to glucuronidation of AZ11939714 in the species studied was pivotal to the optimization of metabolic stability in this class of compound. HPLC-NMR was essential in this respect. The site of glucuronidation could be predicted successfully from the proton chemical shifts alone, and later confirmed by observation of ROESY cross-peaks.
Implications of Species-Dependent Glucuronidation. Recent research in the area of glucuronidation has focused on issues of extrapolation to the in vivo situation (Soars et al., 2002). This topic is made complex, given the nature of the UGT enzyme isoforms and the requirement for diffusion or transport processes of the substrate, UDPGA, and glucuronide products (Lin and Wong, 2002).
The marked species-dependent regioselectivity in the glucuronidation of AZ11939714 is suggestive of different UGT isoforms being responsible in each species. Similar observations and conclusions were made for the glucuronidation of 2-hydroxyestriol in rat, dog, and guinea pig liver microsomes (Ohkubo et al., 1990), and luteolin and quercetin in rat and human microsomes from intestine and liver (Boersma et al., 2002). The glucuronidation of an antithrombotic thioxyloside drug (Pless et al., 1999) and denopamine (Kaji and Kume, 2005) was also found to exhibit species differences in regios-electivity. By studying glucuronidation in recombinant UGTs, these authors were able to confirm that the regioselectivity was due to specific isoforms expressed in rat and human. Jin et al., (1997) have also established the regioselectivity for some human UGT isoforms. Although the authors were aware that methodology to evaluate the contribution of individual UGT isoforms to a given drug's glucuronidation had been reported (Court et al., 2001), at the time of study, this approach was not considered sufficiently validated to be applied with confidence.
Extrapolation is clearly complicated if differences are observed in glucuronidation between the key species used in medical research, and further work on compounds related to AZ11939714 has indicated that the structure-metabolism relationships in rat, dog, and human are substantially different. Differences in the rates of glucuronidation between species are well established (e.g., Sharer et al., 1995) and, in some cases, have been linked to regioisomeric metabolism differences (e.g., Shimada et al., 1984; Soars et al., 2001; and references given previously). In the present study, the observed species differences precluded the establishment of a laboratory animal model entirely representative of humans and focused attention on human in vitro models. The in vitro-in vivo scaling methodology using hepatocytes was validated for rat and dog, thereby justifying the use of human in vitro models for compound optimization. An enhanced throughput screen (human hepatic microsomes, supplemented with UDPGA) was established with the aim of directing chemical synthesis toward lower rates of glucuronidation. The in vitro-in vivo extrapolation described was facilitated by the inclusion of clinical marker compounds and also by the determination of glucuronidation rates in human hepatocytes from several donors.
|
Footnotes |
|---|
Article, publication date, and citation information can be found at http://dmd.aspetjournals.org.
ABBREVIATIONS: AZ11939714, (1S,2R,3R,5R)-3-(hydroxymethyl)-5-[7-{[(1R,2S)-2-phenylcyclopropyl]amino}-5-(propylthio)-3H-[1,2,3]triazolo-[4,5-d]pyrimidin-3-yl]cyclopentane-1,2-diol; HPLC, high-performance liquid chromatography; MS, mass spectrometry; ROESY, rotating-frame Overhauser effect spectroscopy; UGT, UDP-glucuronosyltransferase; UDPGA, UDP-glucuronic acid; DMSO, dimethyl sulfoxide; WET, water suppression enhanced through T1 effects; TOCSY, total correlation spectroscopy; COSY, correlated spectroscopy; CLint, intrinsic clearance.
1 Current affiliation: Organon Laboratories Ltd., Newhouse, Lanarkshire, Scotland, UK. 
Address correspondence to: Dr. Richard J. Lewis, Department of Physical and Metabolic Science, AstraZeneca R&D Charnwood, Bakewell Road, Loughborough, LE11 5RH, UK. E-mail: Richard.J.Lewis{at}astrazeneca.com
|
References |
|---|
|
TopAbstractMaterials and MethodsResultsDiscussionReferences |
|---|
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
 |
 |
 |
|
 |
 |
 |
