Use of a Microplate Scintillation Counter as a Radioactivity Detector

Use of a Microplate Scintillation Counter as a. Radioactivity Detector for Miniaturized Separation. Techniques in Drug Metabolism. K. Olaf Boernsen,*,...
0 downloads 0 Views 68KB Size
Anal. Chem. 2000, 72, 3956-3959

Use of a Microplate Scintillation Counter as a Radioactivity Detector for Miniaturized Separation Techniques in Drug Metabolism K. Olaf Boernsen,*,† James M. Floeckher,‡ and Gerard J. M. Bruin†

Drug Metabolism and Pharmacokinetics, Novartis Pharma AG, CH-4002 Basel, Switzerland, and Packard Instrument Company, 800 Research Parkway, Meriden, Connecticut 06450

For in vivo and in vitro drug metabolism studies, the use of radioactively labeled drugs is one of the most important tools to develop the understanding of the fate of a drug in the body. In combination with separation methods, e.g., high-performance liquid chromatography (HPLC), radioactive labeling allows the highly selective, sensitive, and quantitative detection of unknown metabolites. Together with mass spectrometric and/or other spectroscopic data for structure elucidation of metabolites, completely elucidated biotransformation pathways for an administered drug can be obtained. In most studies, the radioactive isotopes 14C or 3H are used for labeling of a given drug. From the analytical point of view, the specific radioactivity should be as high as possible to detect the low metabolite concentrations in various biological matrixes, such as plasma, urine, bile, and feces, collected at different time points after dose administration. However, for various reasons, the amount of administered radioactivity should

be kept as low as possible (but still analytically acceptable) in most studies. This prerequisite puts high demands on the sensitivity of the radioactivity detector. With the introduction of a new radioactivity counting device in our laboratory, a microplate scintillation counter1 (here called the TopCount), it is possible to perform fast and sensitive analysis using low amounts of radioactivity. The key benefit of the TopCount counting system is the ability to count samples deposited on solid scintillators contained in low-cost microplates. The employed single photomultiplier (PMT), time-resolved scintillation counting technique1 uses pulse counting to distinguish between true decay events and background noise. It eliminates the requirement for two PMTs to count each sample, which facilitates close physical alignment of multiple PMTs for simultaneous counting of up to 12 samples directly in microplates. An attractive feature of the TopCount is the possibility to use it as a highly sensitive off-line radioactivity detector for miniaturized separation techniques, such as capillary liquid chromatography and capillary electrophoresis (CE). Due to the smaller amounts of radioactivity injected than in HPLC with normal-bore columns, and the smaller peak volumes, the use of conventional on-line radiomonitors is restricted because of their insufficient sensitivity. Off-line radioisotope sample counting, using “classical” liquid scintillation counting (LSC) techniques after fractionation of the eluent, offers higher sensitivity. However, LSC is labor-intensive and time-consuming. The TopCount system for microplate counting can significantly reduce laboratory work and increase sample throughput. Figure 1 shows a general setup for the fractionation of small volumes eluting from an HPLC column or CE capillary. After separation, the eluent fractions are automatically dispensed into special, opaque microplates (Deep-Well LumaPlate microplates) using a fraction collector. The use of opaque microplates reduces optical crosstalk with surrounding wells to negligible levels below 0.002% for 14C- and 3H-labeled compounds.2 The microplates contain an yttrium silicate-based solid scintillator, deposited on the bottom of each well. Due to the low flow rates in capillary LC, from approximately 5 to 50 µL/min for column diameters between 0.3 and 1 mm i.d.,3 a makeup flow must be added to

* Corresponding author: (e-mail) [email protected]. † Novartis Pharma AG. ‡ Packard Instrument Co.

(1) Floeckher, J. Application Note, TCA-003; Packard Instrument Co., 1991. (2) Floeckher, J. Application Note, TCA-002; Packard Instrument Co., 1991. (3) Chervet, J. P.; Ursem, M.; Salzman, J. P. Anal. Chem. 1996, 68, 15071512

In miniaturized separation techniques, such as capillary electrophoresis (CE) or capillary liquid chromatography (LC), conventional on-line radioactivity detection of labeled compounds is restricted, because of insufficient sensitivity. It will be shown that a microplate scintillation counter for 96-well plates (TopCount) can be used as a sensitive and easy-to-handle radioactivity detector for capillary LC and CE. The attractive combination of capillary LC, eluent fractionation, and subsequent off-line counting is described. The new method is applied for rapid and sensitive separation and detection of 3H-labeled parent drug and its metabolites at levels between 25 and 700 cpm in rat urine. The advantages of capillary LC coupled to the TopCount, and combined with LC-MS data, can be of benefit in many analytical areas, including the characterization of metabolites at low concentration within complex biological fluids. With the same setup, the fractionation with subsequent off-line counting is equally applicable to CE. This is demonstrated with electrophoretically separated 14C-labeled impurities, nicely resolved from a negatively charged main compound, at low levels.

3956 Analytical Chemistry, Vol. 72, No. 16, August 15, 2000

10.1021/ac000432s CCC: $19.00

© 2000 American Chemical Society Published on Web 07/14/2000

Figure 1. Experimental setup for fractionation of either eluent or background electrolyte into LumaPlate microplates after leaving an HPLC column or CE capillary, respectively.

obtain a sufficient droplet formation rate. By adapting this makeup flow rate, the wells are filled with only small volumes of liquid, between 30 and 100 µL, to shorten the time needed for the mandatory drying step (see below). After collecting the fractions in Deep-Well LumaPlate microplates, the plates are dried in a Speedvac centrifugal evaporator. The TopCount is then used to count the plates quickly (up to 12 wells at a time in parallel), and from the radioactivity in the wells, a radiochromatogram can be constructed. This note describes some first applications of the TopCount in combination with capillary LC and CE. EXPERIMENTAL SECTION (a) Capillary LC/Fractionation/Radioactivity Counting. Separations were performed with an HPLC instrument (HP1100, Hewlett-Packard, Waldbronn, Germany) and an HTS-PAL autosampler (CTC Analytics, Zwingen, Switzerland). A microflow processor (Accurate, LC Packings, Amsterdam, The Netherlands) placed between the pump and the injector was used to split the flow down to 12 µL/min. Gradient separations were performed at 50 °C. A similar column-switching system as described by Zell et al.4 was used. It consisted of a trapping column (4 mm i.d. × 2 mm column in a security guard cartridge (Phenomenex, Torrance, CA) and an analytical column (0.5 mm id. × 15 cm column, packed with 3-µm Luna C18, Phenomenex), connected by a microbore switching valve. Conditioning and rinsing of the trap were done by the HTS-PAL autosampler. An aliquot of urine was transferred by the autosampler onto the trapping column. The eluent used was water. During trapping, the switching valve diverts the effluent to waste to prevent urine salts and other matrix constituents from entering the analytical column. The gradient was from 10% B to 90% B in 25 min, whereby the mobile phase consisted of (A) 50 mM ammonium acetate in water and (B) acetonitrile. After passing through the UV detector, the HPLC fractions were collected into Deep-Well LumaPlate microplates (Packard Instrument Co., Meriden, CT) using a fraction collector (FC204, Gilson Inc., Middleton, WI) with a fraction collection time of 10 s. Because of the low (4) Zell, M.; Husser, C.; Hopfgartner, G. Rapid Commun. Mass Spectrom. 1997, 11, 1107-1114

flow rate, a makeup flow of 150 µL/min was delivered by a syringe pump via a T-piece as shown in Figure 1. The connection between the analytical column and the T-piece was made by a piece of 75-µm-i.d. fused-silica capillary. In this way, ∼24 droplets/min could be generated. The LumaPlate microplates, containing ∼30 µL/well after fractionation, were dried in a Speedvac evaporator (Savant Instruments Inc., Holbrook, NY) for 10 min at 43 °C. The LumaPlate microplates were then closed by a sealing film (TopSeal A, Packard Instrument Co.) and placed in the TopCount for counting. Prior to counting, the TopCount was calibrated, the 12 detectors were normalized, and the backgrounds of the photomultipliers were determined. Then the microplates were counted, using the 12 simultaneous counting detectors, for 8 min/sample. The counting results of the samples for the entire HPLC run were stored as ASCII files. The ASCII files were converted for integration and analysis using ASCIIFLO conversion software (Packard Instrument Co.) and then analyzed using the FLO-ONE analysis software (Packard Instrument Co.). This provided results in a manner similar to that performed on-line using the Packard Radiomatic model 505TR flow scintillation analyzer, which is an on-line detector for HPLC applications. (b) CE/Fractionation/Radioactivity Counting. The capillary electrophoresis separations were carried out on a Hewlett-Packard 3DCE instrument in the CE-MS mode. For CE with subsequent fractionation, the modified CE-MS capillary cassette was used to allow the capillary to exit the CE instrument and allow connection to the T-piece for introducing a makeup liquid. The capillary passed through a UV detection cell before it was guided into the T-piece. The capillary dimensions were 75-µm i.d., a total length of 86.3 and 19.8 cm from injection to UV detection point. The background electrolyte (BGE) used was 40 mM ammonium acetate, 0.1 mM EDTA, adjusted to a final pH of 7.5 with acetic acid. Before analysis the capillary was preconditioned by (1) a 4-min flush at 1 bar with a wash solution consisting of BGE plus 0.01% (w/v) hexadimethrine bromide (HDB) and (2) a 2-min flush at 1 bar with BGE. The positively charged surfactant HDB served as a dynamic coating at the fused-silica wall and reversed the electroosmotic flow. Injection was done at -8 kV for 10 s. The separation was performed at -18.5 kV with a generated current of 50 µA. The T-piece was connected to the common ground of the CE instrument. The fraction collection time was 3 s/well. The makeup liquid consisting of BGE/methanol (90/10, v/v) was added at a flow rate of 320 µL/min. With this flow rate, 1 droplet/ well could be collected. The drying and counting steps were as described above for the capillary LC system. RESULTS AND DISCUSSION To avoid loss of resolution, the fraction collection volume should be smaller than the peak volume of the separated compounds. However, the fraction collection time should not be chosen too small, because this would decrease sensitivity; i.e., the radioactivity in one peak would be distributed among too many wells. Therefore, a compromise between resolution and sensitivity had to be found. For instance, a peak eluting at 15 min with 7000 theoretical plates has a standard deviation σt of 10.7 s in time units, which is σv ) 2.1 µL in volume units at a flow rate of 0.2 µL/s for a 0.5-mm-i.d. column. The peak volume, into which the radioactivity migrates, is then equal to (2π)1/2σv ) 5.4 µL. Therefore, a fraction collection time of 10 s/well, which is equal to a fraction Analytical Chemistry, Vol. 72, No. 16, August 15, 2000

3957

Figure 2. Chromatogram of 5 µL of rat urine containing 3H-labeled drug and metabolites after injection of 5500 dpm, counted off-line with the TopCount system (8 min/well, 12 wells in parallel). The HPLC flow rate was 12 µL/min. Makeup liquid was ethanol at 150 µL/min. Inset: 15-fold expansion of y-axis scale

collection volume of 2 µL when using 0.5-mm-i.d. columns, at a flow rate of 12 µL/min, seems to be an appropriate starting condition. Figure 2 shows a chromatogram resulting from capillary LC/ fractionation/TopCount detection. A volume of 5 µL of a rat urine pool with a radioactivity amount of only 5500 dpm (92 Bq, or 2.48 nCi), containing 3H-labeled parent drug and metabolites, was injected into a 0.5-mm-i.d. column. The small peak at a retention time of 9.5 min with 25 cpm at its peak maximum represents the parent drug. The TopCount technology offers the possibility to measure lower levels of radioactivity than with LSC. Whereas the LSC background noise is typically 12-16 cpm, it is only 1-2 cpm for the TopCount. This implies that the small peaks at the retention times of 3.5, 9.5, and 11.5 min would be much more difficult to detect with LSC. The use of on-line radiodetection would be impossible in this application due to insufficient sensitivity. The main metabolite in this example, migrating at ∼6 min, eluted with 699 cpm in the peak maximum. This peak was reconstructed from six data points, which allowed an adequate description of the chromatographic peak. Despite the counting efficiency of only 40% for 3H-labeled compounds, resulting in much lower cpm than dpm values, all the metabolites and parent drug could be visualized in the chromatogram. It should be noted here that the TopCount counting efficiency is higher for 14C- than for 3H-labeled compounds (∼92% for 14C 1), leading to even better sensitivity for 14C-labeled compounds. When compared to classical LSC, the throughput of the TopCount/LumaPlate approach is much higher. The TopCount takes ∼1.7 h of counting time for the given chromatographic run of 25 min [150 fractions × 8/12 (counting time for 12 wells in parallel) ) 100 min)], as compared to 15 h of counting time for LSC [150 samples × 3 (counting time per sample) × 2 (measurement in duplicate) ) 900 min] to achieve the similar result. Another advantage of the TopCount method is the possibility to support LC-MS analysis of labeled metabolites in biological matrixes. It is well known that a reduction in column diameter produces a higher sample peak concentration in the detector.3,4 When the mass spectrometer behaves as a concentration-sensitive detector (signal independent of flow rate) or even shows higher responses at lower flow rates, the use of small-bore columns is a 3958 Analytical Chemistry, Vol. 72, No. 16, August 15, 2000

Figure 3. Electropherogram of a negatively charged compound and impurities. Detection was performed off-line by the TopCount system after fractionation of BGE in LumaPlate microplates. The makeup liquid was added at 320 µL/min. The BGE was fractionated between 6 and 10.5 min in 3-s fractions. Inset: 10-fold expansion of y-axis scale.

great advantage to correlate data from capillary LC-MS and capillary LC-TopCount detection. For example, with 0.5-mm-i.d. columns, the concentration in the peak maximum is ∼100 times higher than it would be after a separation on a 4.6-mm column for the same amount of sample injected. Due to the smaller amounts of sample injected into small-bore columns, the radioactivity signals are often too low for on-line detectors. The higher sensitivity of the TopCount now enables the use of one miniaturized chromatographic system for both MS and radioactivity detection, which makes peak assignment and correlation between data from MS and radioactivity detection much easier and faster. For instance, one HPLC run will be fractionated in the LumaPlate microplates, whereas the next run will be performed with MS detection with minimal effort to change from one detection system to the other. Of course, splitting of the eluent with one part directed into the MS and the other into the Lumaplates would be an even more elegant solution. As CE has been widely applied for the determination of charged molecules,5 we explored the combination of CE/ fractionation/TopCount detection for the analysis of a negatively charged drug. For CE, where much higher efficiencies are obtained than in HPLC, the fraction collection time was reduced to 3 s/well to avoid loss of separation power. The electropherogram of a 14C-labeled drug in water (0.11 mg/mL, 1.17 × 107 dpm/ mL), used as an application solution in an animal ADME study, is shown in Figure 3. On-column sample stacking6 was performed to increase the amount of radioactivity injected into the capillary. Due to electromigration dispersion, the main peak had a triangular peak shape, which compromised the separation power of CE somewhat in this example. From a comparison of peak areas in the electropherograms recorded with UV detection after hydrodynamic injection (50 mbar, 6 s, 26.8-nL injection volume) and electrokinetic injection (-8 kV, 10 s), it was calculated that ∼8800 dpm was injected. Liquid was collected between 6 and 10.5 min (5) Jorgenson, J. W.; Lukacs, K. D. A. Science 1983, 222, 266-273 (6) Burgi, D. S.; Chien, R. L. Anal. Chem. 1991, 63, 2042-2047

at 3-s intervals, which allowed the use of only one 96-well plate. Some of the radioactively labeled impurities were nicely resolved from the main peak and peak maximums; as low as 53 cpm could easily be detected. These impurities were not visible in the electropherogram recorded by UV detection because of insufficient sensitivity. The first example shown clearly demonstrates that the combination of capillary LC with subsequent fractionation and the Deep-Well LumaPlate/TopCount technology is very useful for drug metabolism studies. This method eliminates the laborintensive step of adding scintillation cocktail to the eluent fractions and the long serial counting times in LSC. LSC takes ∼10 times longer to count the same number of samples. Except for transferring the Lumaplates to the TopCount, everything can be done in an automated mode. Another advantage of this method is the high counting repeatability that was observed in our laboratory.7 The second example, describing CE/fractionation/TopCount detection, demonstrates that this combination can be a very valuable tool for purity control of radiolabeled, charged compounds. For the example shown here, i.e., the purity control of an application solution with low ionic strength, the injection of sufficient levels of radioactivity to allow detection will not be a problem in most cases. However, to investigate metabolite patterns in difficult biological matrixes, such as urine, bile, plasma, or feces extracts, a special sample pretreatment to desalt the samples will be a prerequisite. Otherwise, the use of on-column preconcen(7) Boernsen, K. O. Application Note, AN004-TC; Packard Instrument Co., 2000.

tration techniques will be difficult to realize. This topic is presently under investigation in our laboratory. A few limitations should be mentioned as well. Although no chemical quenching effects occur during counting, strongly colored fractions in the wells could reduce the counting efficiency. (A visual inspection of colored dry LumaPlate microplate wells can tell you which signals might be suppressed.) The samples in a LumaPlate microplate must be completely dry before counting. Besides that, the compounds of interest must not be volatile, because volatile compounds are likely to be lost during drying. It should also be kept in mind that apolar compounds might adsorb at the walls of the Luma plate, which would prevent them from interacting with scintillator at the bottom of the well. Therefore, it is highly recommended to dry the Luma plates in centrifugal vacuum evaporators rather than in shaking vacuum devices. Overall, the TopCount system combined with capillary LC can significantly speed up the analysis of radioactive samples. In addition, due to the lower background levels, lower detection limits, which are of great importance in drug metabolism studies, can be obtained. The ability to use the TopCount system as a radioactivity detector in combination with CE expands the capability of CE to be more frequently applied in drug metabolism studies. Received for review April 18, 2000. Accepted June 2, 2000. AC000432S

Analytical Chemistry, Vol. 72, No. 16, August 15, 2000

3959