Simultaneous Screen for Microsomal Stability and Metabolite Profile

Turbulent-Laminar Flow LCrLC and Automated. Tandem Mass Spectrometry. H. K. Lim,* K. W. Chan, S. Sisenwine, ... ing of drug candidates during lead opt...
0 downloads 8 Views 104KB Size
Download PDF
Anal. Chem. 2001, 73, 2140-2146

Simultaneous Screen for Microsomal Stability and Metabolite Profile by Direct Injection Turbulent-Laminar Flow LC-LC and Automated Tandem Mass Spectrometry H. K. Lim,* K. W. Chan, S. Sisenwine, and J. A. Scatina

Drug Safety and Metabolism, Wyeth-Ayerst Research, 9 DeerPark Drive, Building 3, Monmouth Junction, New Jersey 08852

A LC-LC/MS/MS method has been developed that significantly increases the throughput in metabolism screening of drug candidates during lead optimization in discovery. This was accomplished by the reduction of sample preparation time through an on-line extraction of a drug and its metabolites from microsomal proteins using turbulent flow chromatography. Following its injection onto a column at turbulent flow, the drug and its metabolites are backwashed onto a reverse-phase column via online column switching and resolved chromatographically at a laminar flow of 2 mL/min. This tandem turbulentlaminar flow chromatographic system in a total cycle time of 8 min can achieve adequate separation of isomeric metabolites of venlafaxine, haloperidol, or adatanserin. Further improvement in throughput can be achieved by multiplexing both microsomal stability assessment and metabolite profiling into a single analysis. This is made possible by the ability of the ion-trap mass spectrometer to perform simultaneously multiple-reaction monitoring for microsomal stability and data-dependent multiplestage mass spectrometric analysis for metabolite profiling within a single LC analysis. Such a LC-LC/MS/MS approach can dramatically shorten the time for providing metabolism feedback to the drug discovery process. The high cost and lengthy process of pharmaceutical research and development are largely due to the high attrition rate in drug development. This has stimulated pharmaceutical companies to look for new technologies that can increase the productivity and efficiency in their drug discovery efforts. With the implementation of enabling technologies such as combinatorial chemistry, highthroughput screening (HTS), molecular biology, and genomics, a large number of compounds can be generated and screened against multiple disease targets in a very short time. This strategy of increasing the number of lead candidates in discovery to compensate for the high attrition rate will not be fruitful without knowing the reasons for failure of the selected leads in development. As a result, a survey was conducted of seven British pharmaceutical companies, which indicated that 63% of the * To whom correspondence should be addressed. Tel: 732-274-5407. Fax: 732-274-5465. E-mail: [email protected].

2140 Analytical Chemistry, Vol. 73, No. 9, May 1, 2001

compounds failed in development due to toxicity and poor biopharmaceutical properties1. Consequently, metabolism and pharmacokinetic properties of lead compounds are being characterized increasingly early in the discovery phase. As a result, larger numbers of compounds are being screened for microsomal stability, and their metabolites are characterized to identify metabolic softspots so that these vulnerable sites can be blocked or designed out of the drug candidates. To keep pace with the number of lead candidates, higher throughput metabolism and pharmacokinetic screens have been developed throughout the pharmaceutical industry; however, the throughput in these areas still lags behind that of activity screening and may become one of the bottlenecks in advancing drug candidates. In general, the bottlenecks of HTS in metabolism and pharmacokinetic studies are attributed to sample preparation, analysis, and data processing. An approach to increase throughput is parallel sample processing, such as automated solid-phase extraction in plates containing either 96 wells or a multiple thereof2. A further improvement in sample throughput has recently been reported that uses direct analysis by turbulent flow liquid chromatography/tandem mass spectrometry in which approximately 4 h is required to analyze a 96-well plate3. This technique relies on the selectivity of selected-reaction monitoring to offset the need for extensive chromatographic separation. The recent introduction of parallel analysis of samples by multiplexing several LCs via a switching valve can further reduce analysis time to produce even higher throughput.4 All of the strategies mentioned so far have dramatically increased the throughput for quantitative bioanalyses. In general, the throughput achieved to date for pharmacokinetic screening has surpassed that of metabolic profiling because of the need to chromatographically separate the drug and its metabolites for structural characterization. Superimposed upon this is the complexity and the number of mass spectrometric experiments needed for structural elucidation, which pose a (1) Venkatesh, S.; Lipper, R. A. J. Pharm. Sci. 2000, 89, 145-154. (2) Simpson, H.; Berthemy, A.; Buhrman, D.; Burton, R.; Newton, J.; Kealy, M.; Wells, D.; Wu, D. Rapid Commun. Mass Spectrom. 1998, 12, 75-82. (3) Ayrton, J.; Dear, G. J.; Leavens, W. J.; Mallet, D. N.; Plumb, R. S. Rapid Commun. Mass Spectrom. 1997, 11, 1953-1958. (4) Korfmacher, W. A.; Bryant, M. S.; Cox, K. A.; Ng, K. K.; Veals, J.; Clarke, N.; Hsieh, Y. S.; Palmer, C. A.; Lin, C. C. Proceedings of the 47th ASMS Conference on Mass Spectrometry and Allied Topics, Dallas, TX, 1999; p 1135. 10.1021/ac001112b CCC: $20.00

© 2001 American Chemical Society Published on Web 03/30/2001

considerable challenge to automation. Furthermore, the lower sensitivity of full-scan mass spectrometric experiments that is needed for structural elucidation often requires work-up of larger sample volumes. Irrespective of the sample work-up procedure, the larger sample volume would require longer sample preparation time, which would impact the throughput. Consequently, there is considerable interest in developing an on-line sample preparation procedure to overcome the need for the time-consuming evaporation and reconstitution steps typical of off-line sample work up. On-line direct analysis of biological matrixes has been reported using restricted access columns5 and solid-phase extraction cartridges.6 Another approach is to employ turbulent flow chromatography3,7,8. To date, all of the analytical applications of turbulent flow chromatography have been limited to fast quantitative bioanalysis using a very steep gradient3,7,8 and with little emphasis on chromatographic separation capability; therefore, it was of interest to investigate the separation capability of gradient turbulent flow chromatography for automated structural elucidation of a drug and its metabolites. This report describes the development of a tandem LC/tandem MS (LC-LC/MS/MS) method whereby, within a single injection, the proteins in the microsomal incubates are removed by turbulent flow chromatography prior to separation of a drug and its metabolites by fast-gradient laminar flow chromatography. In this single LC analysis, the disappearance of the drug is then detected by multiple-reaction monitoring experiments, and the metabolites are characterized by data-dependent full-scan product ion experiments. Such an approach both eliminates the need for sample extraction and multiplexes both metabolic stability and metabolite characterization in one analysis. This approach maximizes the information content from one experiment and significantly increases throughput for in vitro metabolism studies. EXPERIMENTAL SECTION Materials. Glucose-6-phosphate monosodium salt, 1 M magnesium chloride, 0.5 M ethylenediaminetetraacetic acid disodium salt, glucose-6-phosphate dehydrogenase, β-nicotinamide adenine dinucleotide phosphate sodium salt, haloperidol, and disodium phosphate heptahydrate were obtained from Sigma Chemical Co. (St. Louis, MO). Sodium chloride and 1 N sodium hydroxide were of Baker analyzed grade (J. T. Baker, Phillipsburg, NJ). Ammonium acetate (BioChemika grade) and formic acid (Chemika grade) were obtained from Fluka (Ronkonkoma, NY). Monosodium phosphate monohydrate was purchased from Fisher Scientific (Atlanta, GA). HPLC-grade water, acetonitrile, and potassium phosphate were from EM Science (Gibbstown, NJ). Rat liver microsomes obtained from saline- and phenobarbital-treated male Sprague-Dawley rats were purchased from Xenotech (Kansas City, KS). Adatanserin and venlafaxine were obtained from WyethAyerst Research (Princeton, NJ). In Vitro Rat Liver Microsomal Incubations. Incubations consisted of 2 mM ethylenediaminetetraacetic acid, 10 mM magnesium chloride, 21-27 µM drug (free base), 2 mg microso(5) Needham, S. R.; Cole, M. J.; Fouda, H. G. J. Chromatogr., B 1998, 718 (1); 87-94. (6) Bowers, G. A.; Clegg, C. P.; Hughes, S. C.; Harker, A. J.; Lambert, S. LCGC, 1997, 15 (1), 45-53. (7) Quin, H. M.; Takarewski, J. J. International Patent No. WO 97/16724. (8) Jemal, M.; Qing, Y.; Whigan, Y. Rapid Commun. Mass Spectrom. 1998, 12, 1389-1399.

Figure 1. Schematic representation that corresponds to the 2-column switching configuration and its flow path during direct extraction under turbulent flow (A) and followed by fast-gradient chromatographic separation under laminar flow (B).

mal protein, 3.5 mM glucose 6-phosphate, 0.4 unit glucose 6-phosphate dehydrogenase and 1.3 mM β-nicotinamide adenine dinucleotide phosphate. Potassium phosphate buffer pH 7.4 (0.1 M) was added to bring the volume to 1 mL. Incubations were carried out at 37 °C for 1 h, and the reaction was terminated by placing the tubes in an ice bath and acidifying the incubate to pH 2 with diluted phosphoric acid. For simultaneous investigation of microsomal stability and metabolite profile of adatanserin, 200µL aliquots were removed after 0, 5, 10, 15, 20, 30, 40, 50, or 60 min of incubation at 37 °C. The reaction was quenched by adding to a screw-capped test tube containing 30 µL of internal standard (haloperidol, 10 µg/mL) and 20 µL of 10% (v/v) phosphoric acid while standing in an ice-bath. The tube was immediately handvortexed and then centrifuged at 790g at 4 °C for 10 min. The supernatant was transferred to autosampler vials, and an aliquot of 50-200 µL of sample was analyzed by turbulent-laminar flow LC-LC/MS/MS. Appropriate control samples without added NADPH regenerating system, drug, or microsomes were also included in the study. Turbulent-Laminar Flow Chromatography. All chromatographic separations were carried out using an HP1090M liquid chromatograph (Hewlett-Packard, Palo Alto, CA) equipped with two 6-port Rheodyne switching valves under the control of ChemStation. The LC was replumbed with 0.01-in.-i.d. PEEK tubing to minimize backpressure. The schematic representation for the tandem LC is shown in Figure 1. An aliquot of microsomal incubate (50, 100, or 200 µL) was injected onto a 50 × 1 mm i.d., 50-µm porous particles HTLC C8 column (Cohesive System, Inc., MA). The mobile phase for on-line extraction was 4 mL/min of 0.1% (v/v) acetic acid in water. During the extraction step, the Analytical Chemistry, Vol. 73, No. 9, May 1, 2001

2141

eluant from the HTLC column was diverted to waste. After 0.5 min, the column was washed at the same flow rate with 10 mM ammonium acetate for 0.3 min. Subsequently, the flow rate was reduced in 0.2 min to 2 mL/min. At this point, both of the valves were switched so that the flow direction through the HTLC column was reversed and the compounds trapped at the front of the HTLC column were backflushed onto the 20 × 4 mm i.d., 3-µm Aquasil C18 analytical column (Keystone Scientific, Bellefonte, PA). The eluant from the analytical column was directed to the mass spectrometer, and the channel in the Rheodyne valve to the waste outlet was flushed by flow from the auxiliary pump (LKB, model 2150, Sweden). The drug and its metabolites were separated using a 5-min linear gradient of 10 mM ammonium acetate and acetonitrile. The columns were held at 95% acetonitrile for 1 min prior to ramping down in 0.1 min to 100% of 0.1% acetic acid in water. At this time, both of the valves were switched to restore forward flow to the HTLC C8 column to waste and the flow from auxiliary pump to the analytical column to the mass spectrometer. The HTLC C8 column was equilibrated at 4 mL/ min with 100% of 0.1% (v/v) acetic acid in water for 0.9 min before the next injection. At the same time, the analytical column was also equilibrated for 0.9 min with 95:5 10 mM ammonium acetate: acetonitrile at 1 mL/min. Both the extraction and the analytical columns were kept at 50 °C during analysis. The total cycle time was 8 min. Mass Spectrometry. On-line liquid chromatography/mass spectrometric (LC/MS) analyses were performed using a Finnigan MAT ion-trap mass spectrometer (LCQ, Finnigan MAT, San Jose, CA) operated in the positive ion electrospray ionization mode. The entire LC eluate was split postcolumn to about 0.45 mL/min prior to spraying into the mass spectrometer at +4.5 kV. Desolvation of the droplets was aided by the heated capillary set at 230 °C and by the auxiliary and sheath gases set to 30 and 90, respectively. Ions were sampled into the mass spectrometer with an injection time set at 50 ms and with automatic gain control. Ultrahigh-purity helium was used as the buffer gas. For datadependent MSn analysis, each analytical scan from m/z 150 to 600 consisted of 2 µscan. The most abundant ion exceeding the threshold of 100 000 was isolated using an isolation width of 2 Da, and the predominantly monoisotopic peak was selected for collision-activated dissociation (CAD) experiments. Up to MS3, data were obtained with a relative collision energy of 26%. The multiple-reaction monitoring (MRM) experiments for quantifying adatanserin and haloperidol monitored the transitions of precursorto-product ions at m/z 370 f 206 and 376 f 165, respectively. The product ions were isolated as described previously. Ejected ions were detected by the electron multiplier set at a gain of 4 × 105. Data acquisition and reduction was carried out using Navigator software, version 1.2. RESULTS AND DISCUSSION It has been reported that the plate height of a column is reduced at turbulent flow, which is contrary to prediction by the van Dempter plots.7 This turbulent flow is achieved with acceptable column backpressure using a small-inside-diameter column packed with large particles.7 However, the use of large particles to achieve this low column backpressure will decrease the resolution efficiency. Therefore, whether this opposing effect by large particles can be offset by the gain in reduced plate height at turbulent flow 2142 Analytical Chemistry, Vol. 73, No. 9, May 1, 2001

Figure 2. Direct injection of a phenobarbital-induced SpragueDawley rat liver microsomal incubate of venlafaxine onto an HTLC C8 column (50 × 1 mm i.d., 50 µm) flowed at 4 mL/min of 0.1% (v/v) acetic acid. This was followed by separation using a gradient volume of 79.7 column volumes of 10 mM ammonium acetate and acetonitrile at 4 mL/min prior to detection by full-scan product ion scans (A). The same sample was analyzed by direct injection under turbulent flow, followed by fast-gradient chromatographic separation under laminar flow using half the gradient volume prior to automated data-dependent multiple-stage mass spectrometric detection (B). A control microsomal incubate (without drug) was immediately analyzed as in (B) after the sample was checked for carryover (C). Other details are as described in the Experimental Section.

will impact upon the resolution capability of this column, which has not been addressed before. Figure 2A depicts a chromatogram that was obtained from a direct injection and separation of a phenobarbital-induced rat liver microsomal incubate of venlafaxine by turbulent flow chromatography using a gradient volume of 79.7 column volumes. There was no separation of venlafaxine and its demethylated and hydroxylated metabolites, even though the gradient that was used was about 8-fold greater than the recommended minimum that is needed for chromatographic separation.9 This absence of separation can, perhaps, be explained by the low plate height achieved at turbulent flow being offset by the larger particles (50 µm) used in the column. The unresolved components made it difficult to detect the demethylated and hydroxylated metabolites by automated data-dependent tandem mass spectrometry, as illustrated by detection of only the largest peak (venlafaxine). This may be explained by a reduction of ion injection time through automatic gain control by the larger coeluting component. Although fast-gradient turbulent flow chromatography can separate the drug and its metabolites from microsomal proteins, it does (9) Snyder, L. R.; Kirkland, J. J.; Glajch, J. L. Practical HPLC Method Development, 2nd ed.; John Wiley & Sons: New York, 1997.

not have sufficient resolving power to separate the drug and its metabolites for structural elucidation by automated data-dependent tandem mass spectrometric analysis. To improve chromatographic separation, we used turbulent flow chromatography only for on-line extraction and performed the analytical separation by conventional reverse-phase HPLC. However, the need to operate the extraction and analytical columns at different flow rates with gain in overall analysis time and without sacrifice in chromatographic resolution presented an analytical challenge. It had been reported that fast-gradient chromatographic separation could be achieved using short columns packed with small particles at high laminar flow;10,11 therefore, the coupling of turbulent flow chromatography for online extraction with fast-gradient chromatographic separation at high laminar flow offered the gain in analysis time without a sacrifice in resolution. The 2-column configuration depicted in Figure 1 differed from that previously reported8 in which the extraction HTLC column served as an “injector” for the analytical cartridge column. Essentially, the drug and its metabolites are separated from microsomal proteins under a turbulent flow of 4 mL/min using 0.1% (v/v) formic acid (see Figure 1A). Figure 1B depicts elution from the extraction column in the reverse direction at 2 mL/min onto the analytical column. The analytical column separated the drug and its metabolites prior to its mass spectrometric analysis. Such a two-column-switching configuration in Figure 1 is routinely used in our laboratory for the direct injection of plasma or serum for the quantitation of lead drug candidates to support discovery pharmacokinetic studies. However, the chromatographic separation for pharmacokinetic studies was carried out using a much-reduced gradient volume, resulting in a total cycle time of 5 min. Analysis of the microsomal incubate of venlafaxine using the 2-column system clearly shows a significant improvement in separation (see Figure 2B), and there is no carryover observed in a subsequent injection of a control microsomal incubate (see Figure 2C). The rationale for selecting an Aquasil C18 column (20 × 4 mm i.d., 3µm) that flows at 2 mL/min is that the faster and better separation is achieved with the larger inside diameter column using the same gradient volume (see Figure 3). This may be due to less peak broadening because of the smaller column pressure drop when switching from turbulent to laminar flow at 2 mL/min. This separation of venlafaxine and its metabolites (Figure 2B) was accomplished using half of the gradient volume (39.8 column volumes), as employed by gradient turbulent flow chromatography described in Figure 2A, which clearly indicated the importance of particle size in resolution. The fast-gradient chromatographic separation obtained using a 20 × 4 mm i.d. column packed with 3-µm particles can perhaps be explained by the van Dempter plot determined experimentally for this column (Figure 4). The van Dempter plot indicated that increasing the flow to 0.46 (2 mL/min) from 0.09 cm/sec (0.4 mL/min) resulted in only doubling the plate height but a gain of at least 3-fold in analysis time. This loss in resolution is greatly offset by the large gradient volume used in the separation. This gradient volume is more than adequate for separation, as indicated by no further (10) Lim, H. K.; Stellingweif, S.; Sisenwine, S.; Chan, K. W. J. Chromatogr., A 1999, 831, 227-241. (11) Lee, H. W.; Li, L.; Kyranos, J. Proceedings of the 47th ASMS Conference on Mass Spectrometry and Allied Topics, Dallas, TX, 1999; p 2715.

Figure 3. Comparison of a fast-gradient chromatographic separation of adatanserin and its metabolites in Sprague-Dawley rat liver microsomes from a 60-min incubation using (A) an Aquasil C18 column (20 × 2 mm i.d., 3µm) at 0.5 mL/min and (B) an Aquasil C18 column (20 × 4 mm i.d., 3µm) at 2 mL/min using a gradient volume of 39.8 column volumes. Other details are as described in the Experimental Section.

Figure 4. van Dempter plot determined experimentally for the Aquasil C18 column (20 × 4 mm i.d., 3µm) by plotting the plate heights at various linear velocities. The plate height was determined by injecting 50 ng of haloperidol on the column at each isocratic flow rate of 10 mM ammonium acetate and acetonitrile (30:70). Detection was by full-scan product ion scan of the protonated molecule ion of haloperidol. The linear velocity was calculated using the equation, u ) 4F/60πd2u (12) where F is the flow rate (mL/min), u is the external porosity (estimated to be 0.58 by manufacturer), and d is the column inside diameter (cm).

improvement in resolution upon increasing the gradient volume to 55.8 column volumes (data not shown). The current system permits the injection of e200 µL of microsomal incubates without deterioration of the chromatographic separation and with acceptAnalytical Chemistry, Vol. 73, No. 9, May 1, 2001

2143

able column backpressure for routine analysis (data not shown). This ability to load more samples onto the column is advantageous, especially in the structural elucidation of minor metabolites. In addition, the availability of different LC sorbents for extraction, such as polar end-capped HTLC (Cohesive Technologies, Inc.) and the polymer-based Oasis HLB (Waters), offers flexibility in the extraction of metabolites from both phase 1 and phase 2 metabolism. The 2-column configuration setup in Figure 1 allowed the extraction column to handle more injections because it was only used to trap the drugs and their metabolites. The utilization of a ternary mobile phase system built in flexibility for separation of diverse polarity of compounds. Furthermore, the solvent system for separation can be different from that used for extraction. A mobile phase that is optimized for analytical separation and for mass spectrometric ionization can be selected independently from the extraction solvent. Last, the deployment of two 6-port switching valves reduced the overall analysis time to 8 min by allowing an equilibration of the analytical column during the extraction phase. The improved chromatographic separation of venlafaxine and its metabolites facilitated their detection by automated datadependent tandem mass spectrometry. Figure 2B shows the reconstructed ion chromatogram (RIC) that corresponds to the base peak from data-dependent MS/MS analysis of microsomal incubates of venlafaxine. Venlafaxine and four metabolites can be easily detected. An additional filter corresponding to plotting the RIC of the common product ion instead of the base peak can be used to facilitate the identification of drug-derived peaks. The need for good chromatographic separation in structural elucidation is succinctly illustrated by the mass spectral data of isomeric metabolites, such as N-demethylvenlafaxine (Figure 5A) and O-demethylvenlafaxine (Figure 5B). In each case, the identity of the product ion from the loss of the neutral aliphatic amine by β-cleavage of the dehydrated protonated molecule ion and the mass shift of the substituted benzyl cation in the spectrum helped to localize the metabolic softspot. If these two demethyl metabolites were not chromatographically resolved, a composite mass spectrum from the two species would make the interpretation difficult. Similarly, chromatographic separation of the isomeric hydroxylated metabolites (VM1 and VM3) is also necessary for the location of their metabolic softspots (data not shown). To demonstrate the general utility of this methodology, fast microsomal metabolite profiling of haloperidol was investigated by direct analysis using a dual column-switching configuration operated under turbulent-laminar flow. A total of four drug-related peaks were detected, including the presence of the potentially neurotoxic pyridinium metabolite of haloperidol (data not shown). In general, the in vitro metabolic profiles obtained for venlafaxine and haloperidol by direct microsomal incubate injection are consistent with those previously that were reported using conventional sample preparation procedures13,14 Simultaneous Microsomal Stability and Metabolite Profiling. The feasibility of multiplexing microsomal stability and (12) Niessen, W. M. A.; van der Greef, J. In Chromatographic Science Series; Cazes, C., Ed.; Marcel Dekker: New York, 1992; Vol. 58, Chapter 1. (13) Fogelman, S. M.; Schmider, J.; Venkatakrishnan, K.; von Moltke, L. L.; Harmatz, J. S.; Shader, R. I.; Greenblatt, D. J. Neuropsychopharmacology 1999, 20 (5), 480-490. (14) Gorrod, J. W.; Fang, J. Xenobiotica ,1993, 23 (5), 495-508. (15) Obach, R. S. Drug Metab. Dispos. 1999, 27 (11), 1350-1359.

2144 Analytical Chemistry, Vol. 73, No. 9, May 1, 2001

Figure 5. Product ion mass spectrum corresponding to (A) Ndemethylvenlafaxine (VM4) and (B) O-demethylvenlafaxine (VM2), which were identified in a Sprague-Dawley rat liver microsomal incubate of venlafaxine using turbulent-laminar flow LC-LC prior to detection by data-dependent tandem mass spectrometry.

metabolite profiling within a single analysis is shown in Figure 6, which was obtained from the analysis of a 60-min microsomal incubate of 27 µM adatanserin. The first two reconstructed ion chromatograms (RIC) show the results from multiple-reaction monitoring of the protonated molecule ions of adatanserin (Figure 6A) and haloperidol (as internal standard, Figure 6B); the other three (Figure 6C-E) are RIC from data-dependent multiple-stage mass spectrometric experiments. The peak area ratios from adatanserin and haloperidol in Figure 6A,B, respectively, were used to calculate the microsomal stability of adatanserin (see Figure 7). Adatanserin is a highly metabolized drug, as indicated by its 93% turnover at 60 min incubation. The metabolism of adatanserin is linear up to 30 min of incubation (see insert of Figure 7). The linear relationship obtained under a semilogarithmic plot (see insert of Figure 7) suggests that the in vitro microsomal metabolism of adatanserin followed first-order reaction kinetics, with an average enzyme activity of 0.64 nmol/min/mg protein. Hence, it is probable that hepatic metabolism may contribute significantly to the clearance of this drug candidate. The complimentary piece of information needed for lead optimization is the identification of the metabolic labile site(s). This can be available from the same analysis of the microsomal incubates by data dependent MSn experiments. The RICs in Figure 6C-E provided molecular weight information and product ion mass spectra for structural elucidation of the metabolites. A total of 8 metabolites were identified as mono- or dihydroxylation forms of adatanserin and further oxidation to their corresponding keto metabolites. Additionally, information is available on the time-

Figure 6. Simultaneous determination of the microsomal stability and metabolite profile by direct analysis of a 60-min Sprague-Dawley rat liver microsomal incubate of adatanserin at turbulent flow followed by fast-gradient chromatographic separation. Reconstructed ion chromatograms [RIC] (A,B) are from MRM analysis of adatanserin and haloperidol (as internal standard), and RICs (C-E) are from automated data-dependent multiple-stage mass spectrometric analysis. The labeled peaks (M1-M8) correspond to the following in vitro metabolites of adatanserin: M1, dihydroxyadatanserin; M2 and M5, ketoadatanserin; M3 and M6, hydroxyketoadatanserin; M4, M7, and M8, hydroxyadatanserin.

Figure 7. Microsomal stability was determined by plotting the amount of adatanserin remaining in Sprague-Dawley rat liver microsomes as a function of incubation time. The insert is the semilogarithmic plot of the microsomal stability of adatanserin.

dependent formation of these metabolites. The product ion mass spectrum from third-stage mass analysis of the major metabolite in the microsomal incubate of adatanserin indicated that the

Figure 8. Product ion mass spectrum from third-stage mass analysis of adatanserin (A) and the major metabolite (B) detected in a 60-min Sprague-Dawley rat liver microsomal incubate of adatanserin.

metabolic softspot is the adamantane moiety (see Figure 8). This is suggested by the product ion at m/z 151 (Figure 8B), which corresponds to a mass shift of 16 Da from the adamantane ion at m/z 135 (Figure 8A). Hence, the adamantane moiety needs to be replaced with either a more metabolically stable bioisoteric group or a blockage of the susceptible carbon on the adamantane ring to produce a more metabolically stable drug candidate. An alternative would be exploring the development of the metabolite as a drug candidate. Thus, this LC-LC/MS/MS approach can simultaneously provide both microsomal stability and metabolite profile information to aid the selection and refinement of lead drug candidates. CONCLUSIONS A turbulent-laminar flow LC-LC chromatographic method has been successfully developed for the direct analysis of microsomal incubates. On-line extraction of drugs and their metabolites from microsomal proteins by turbulent flow chromatography is achieved by flowing aqueous mobile phase at high flow rate through a short, small-i.d. column packed with large particles. As demonstrated in the present study, this method has significantly reduced the time for sample preparation. Large sample volumes up to 200 µL can be injected to increase the amount of minor metabolites for structural elucidation. Direct analysis by turbulent flow chromatography coupled with fast-gradient chromatographic separation on a short-cartridge column, using a laminar gradient flow of 40 column volumes was demonstrated to be adequate for the resolution of a drug and its regioisomeric metabolites. Such a chromatographic setup is ideally suited for use in high-throughput Analytical Chemistry, Vol. 73, No. 9, May 1, 2001

2145

screenings for metabolites in discovery. Further improvement in throughput is demonstrated through a simultaneous analysis of a microsomal incubate for stability and a profile. This is accomplished by performing multiple-reaction monitoring and datadependent multiple-stage mass analysis within a single run. Such an approach was successfully applied to the simultaneous investigation of both microsomal stability and metabolite profiling of adatanserin. Although the ion-trap mass spectrometer used in our study has exceptional full-scan sensitivity that permitted the acquisition of a product ion mass spectrum of a low-nanogram amount of a compound that was injected on a column, it had difficulties in detecting the metabolites when the incubation was carried out at a substrate concentration of 1 µM. It is anticipated that the new

2146

Analytical Chemistry, Vol. 73, No. 9, May 1, 2001

instruments will have a 10-fold improvement in sensitivity and will facilitate the carrying out of these experiments at lower substrate concentrations. This will allow the calculation of intrinsic clearance from the microsomal stability data and will permit an allometric scaling to human levels.15 ACKNOWLEDGMENT This paper was presented at the 47th ASMS Conference on Mass Spectrometry and Allied Topics, Dallas, TX, June 13-17, 1999. Received for review September 18, 2000. Accepted February 7, 2001. AC001112B