Distribution of Zeptomole-Abundant Doxorubicin Metabolites in

Doxorubicin (DOX) treatment of NS-1 mouse hybridoma cells results in the formation of ... In contrast, the total amount of other metabolites in a give...
1 downloads 0 Views 116KB Size
Download PDF
Anal. Chem. 2003, 75, 8-15

Distribution of Zeptomole-Abundant Doxorubicin Metabolites in Subcellular Fractions by Capillary Electrophoresis with Laser-Induced Fluorescence Detection Adrian B. Anderson, Chanda M. Ciriacks, Kathryn M. Fuller, and Edgar A. Arriaga*

Department of Chemistry, University of Minnesota, Minneapolis, Minnesota 55455

Doxorubicin (DOX) treatment of NS-1 mouse hybridoma cells results in the formation of zeptomole amounts of metabolites per cell that are difficult to determine by confocal microscopy or HPLC. The native fluorescence of DOX and its metabolites together with laser-induced fluorescence detection (LIF) has previously been used to detect a maximum of four components. In this study, we use capillary electrophoresis with postcolumn LIF (CELIF) to separate and detect 12 components attributed to DOX metabolism, resulting from treatment of NS-1 cells with 25 µM DOX for 8 h. The so-called metabolites 8 and 10 have been identified as doxorubicinone (DOXone) and 7-deoxydoxorubicinone (7-deoxyDOXone), respectively, by comigration with the corresponding synthetic standard. Due to comigration of DOX with doxorubicinol (DOXol), the presence of DOXol had to be determined separately by matrix-assisted laser desorption/ionization time-offlight mass spectrometry. The rest of the metabolites remain unidentified and are referred to by their number assignment. In comparison with the whole cell lysate, fractionation by differential centrifugation results in a better separation resolution of metabolites due to reduced amounts of metabolites in each fraction. This approach was chosen to compare the distribution of 13 metabolites in three subcellular fractions that form a pellet at 14000g and that generically are enriched in nuclei, organelles (mitochondria and lysosomes), and cytosolic components, respectively. The most abundant metabolite, DOXol, was estimated to be 90 ( 15, 18 ( 2, and 60 ( 12 amol/cell (n ) 5) in the nuclearenriched, organelle-enriched, and cytosole-enriched fractions, respectively. In contrast, the total amount of other metabolites in a given fraction varied from 0 to 1300 zmol. 7-DeoxyDOXone is the only metabolite that was present at similar levels in the three fractions. Other salient observations are metabolites 3, 7, and 11 are not detectable in the nuclear-enriched, organelle-enriched, and cytosole-enriched fractions, respectively; metabolite 9 and DOXone are more abundant in the nuclear-enriched fraction than in the other two fractions. The observations * To whom correspondence should be addressed. E-mail: chem.umn.edu. Fax: 612.626.7541.

8 Analytical Chemistry, Vol. 75, No. 1, January 1, 2003

arriaga@

presented here suggest that subcellular fractionation followed by CE-LIF could be a powerful diagnostic for monitoring drug distribution, which is highly relevant to DOX cytoxicity studies. Doxorubicin (DOX) is a widely used anthracycline that has proven to be effective against a variety of human malignancies,1 such as leukemia and breast cancer.2 When DOX reaches the targeted organ as either free drug administered intravenously or encapsulated in liposomes3,4 it enters the malignant cells, reaches its target in the nucleus, and halts cell proliferation.5 Unfortunately, DOX treatment is accompanied by the appearance of cardiac and liver toxicity6-10 and drug resistance6,11,12 that may result from cellular processes involving the parent compound or drug metabolites.13,14 The oxidant activity of DOX aglycon metabolites can lead to mitochondrial Ca2+ release, mitochondrial swelling, collapse of the mitochondrial membrane potential, modification of mitochondrial sulfhydryl groups, and superoxide (O2-) production, all of which may contribute to acute cardiac and liver toxicity.13,14 Two of these metabolites are measured in this report: doxorubicinone (DOXone) and 7-deoxydoxorubicinone (7-deoxyDOXone) (Figure 1). In addition to the aglycon metabolites, doxorubicinol (1) Hortobagyi, G. N. Drugs 1997, 54, 1-7. (2) Booser, D. J.; Hortobagyi, G. N. Drugs 1994, 47, 223-258. (3) Schwonzen, M.; Kurbacher, C. M.; Mallmann, P. Anticancer Drugs 2000, 11, 681-685. (4) Woessner, R.; An, Z.; Li, X.; Hoffman, R. M.; Dix, R.; Bitonti, A. Anticancer Res. 2000, 20, 2289-2296. (5) Bodley, A.; Liu, L. F.; Isreal, M.; Seshadri, R.; Koseki, Y.; Giuliani, F. C.; Kirschenbaum, S.; Silber, R.; Potmesil, M. Cancer Res. 1989, 49, 59695978. (6) de Bruijn, P.; Verweij, J.; Loos, W. J.; Kolker, H. J.; Planting, A. S. T.; Nooter, K.; Stoter, G.; Sparreboom, A. Anal. Biochem. 1999, 266, 216-221. (7) Jeyaseelan, R.; Poizat, C.; Wu, H.; Kedes, L. J. Biol. Chem. 1997, 272, 58285832. (8) Singal, P. K.; Deally, C. M.; Weinber, L. E. J. Mol. Cell. Cardiol. 1987, 19, 817-828. (9) Wadler, S.; Fuks, J. Z.; Wiernik, P. H. J. Clin. Pharm. 1986, 26, 491-509. (10) Zhou, S. Y.; Starkov, A.; Frober, M. K.; Leino, R. L.; Wallace, K. B. Cancer Res. 2001, 61, 771-777. (11) Baldini, N.; Scotland, K.; Serra, M.; Shikita, T.; Zini, N.; Ognibene, A.; Santi, S.; Ferracini, R.; Maraldi, N. M. Eur. J. Cell Biol. 1995, 68, 226-239. (12) Abbaszadegan, M. R.; Cress, A. E.; Futscher, B. W.; Bellamy, W. T.; Dalton, W. S. Cancer Res. 1996, 56, 5435-5442. (13) Sokolove, P. M. Int. J. Biochem. 1994, 26 (12), 1341-1350. (14) Licata, S.; Saponiero, A.; Mordente, A.; Minotti, G. Chem. Res. Toxicol. 2000, 13, 414-420. 10.1021/ac020426r CCC: $25.00

© 2003 American Chemical Society Published on Web 11/27/2002

Figure 1. Chemical structure of doxorubicin and metabolites. DOX ) doxorubicin; DOXol ) doxorubicinol; DOXone ) doxorubicinone; 7-deoxyDOXone ) 7-deoxydoxorubicinone.

(DOXol) (Figure 1) is reported to alter iron homeostasis thus leading to chronic toxicity.14 Subcellular localization is important in the identification of the metabolic path that DOX takes upon entering the cell. Fluorescence confocal microscopy and colocalization of DOX with specific organelle markers suggest that DOX accumulates in the nucleus,15-17 Golgi apparatus,17,18 lysosomes,18 mitochondria,17,19 cytoplasm,15 and microtubule network.15 However, this technique cannot distinguish between DOX and its metabolites because all of them do not have distinctive enough fluorescent spectra. Thus, confocal microscopy is not capable of providing either quantitative information about each DOX metabolite in the various subcellular compartments or any description of DOX metabolism.17, 20 An alternative approach is to combine a subcellular fractionation procedure with a highly sensitive analytical technique, such as capillary electrophoresis with postcolumn laser-induced fluorescence detection (CE-LIF), that then separates and detects the DOX metabolites present in the various subcellular environments. To that effect, subcellular fractionation has been done using either differential centrifugation or gradient centrifugation. Differential centrifugation is a technique by which the subcellular components of cells are separated based on their sedimentation coefficient via centrifugation.21,22 A typical fractionation will result in three subcellular fractions: one containing the nuclei (nuclear-enriched fraction), one containing mostly mitochondria and lysosomes (organelle-enriched fraction), and one containing cytosolic components such as cytoplasm and small microsomes (cytosole(15) Coley, H. M.; Amos, W. B.; Twentyman, P. R.; Workman, P. Br. J. Cancer 1993, 67, 1316-1323. (16) Meschini, S.; Molinari, A.; Calcabrini, A.; Citro, G.; Arancia, G. J. Microsc. 1994, 176, 204-210. (17) Beyer, U.; Rothen-Rutishauser, B.; Unger, C.; Wunderli-Allenspach, H.; Kratz, F. Pharm. Res. 2001, 18, 29-38. (18) Molinari, A.; Cianfrigilia, M.; Meschini, S.; Calcabrini, A.; Arancia, G. Int. J. Cancer 1994, 59, 789-795. (19) Serrano, J.; Palmeira, C. M.; Kuehl, D. W.; Wallace, K. B. Biochim. Biophys. Acta 1999, 44724, 1-5. (20) White, J. G.; Amos, W. B.; Fordham, M. J. Cell. Biol. 1987, 105, 41-48. (21) Hinton, R. H.; Mullock, B. M. In Subcellular Fractionation; Graham, J. M., Rickwood, D., Eds.; IRL Press: Oxford, 1996; pp 31-69. (22) Bronfman, M.; Loyola, G.; Koenig, C. S. Anal. Biochem. 1998, 255, 252256.

enriched fraction). This procedure allows for the isolation of subcellular environments in a relatively simple and gentle manner. A disadvantage to differential centrifugation is that the fractions obtained are enriched, not pure fractions.21,23 Higher purity fractionations can be accomplished by gradient centrifugation, which separates subcellular components based on their density by centrifugation through a density gradient.21 However, this technique is more involved, taking between 1 and 2 h as opposed to 20-40 min required for the differential centrifugation technique.21,24-27 Differential centrifugation offers two distinct advantages over gradient centrifugation. First, total centrifugation times for a typical differential centrifugation (20-40 min) are considerably shorter than for a typical gradient centrifugation (1-2 h).26,27 Second, the density gradient medium can threaten organelle integrity because of osmotic differences.21 Regarding metabolite separations, high-performance liquid chromatography (HPLC) has been used to analyze DOX and its metabolites from murine specimens,28 blood,28 and human plasma.6 Mixtures containing anthracyclines and their derived metabolites have also been separated by HPLC.29,30 Capillary electrophoresis has also been used to study DOX31,32 as well as DOX and its metabolites in biological fluids.33-35 We previously reported the use of CE-LIF to study DOX metabolites in single cells and in bulk cell preparations.36 The low limits of detection obtained with CE-LIF (30 zmol for DOX) are unmatched by the techniques employed to study DOX metabolism thus far. Here we monitor zeptomole-abundant DOX metabolites by taking advantage of the detection capabilities of CE-LIF and the native fluorescence of DOX and its metabolites. We compare the distribution of these metabolites in three subcellular fractions. Three of these metabolites are identified as DOXol, DOXone, and 7-deoxyDOXone. The ability to observe differences in the distributions of zeptomole-abundant metabolites between these fractions is a step toward establishing analytical strategies capable of detecting so far unidentified zeptomole DOX metabolites, defining their subcellular localization, and providing a more complete description of their metabolic significance. (23) Meijer, J.; Bergstrand, A.; DePierre, J. W. Biochem. Pharmacol. 1987, 36, 1136-1151. (24) Kowluru, A.; Tannous, M.; Chen, H. Q. Arch. Biochem. Biophys. 2002, 398, 160-169. (25) Latchoumycandane, C.; Chitra, K. C.; Mathur, P. P. Toxicology 2002, 171, 127-135. (26) Okado-Matsumoto, A.; Fridovich, I. J. Biol. Chem. 2001, 276, 38388-38393. (27) Rout, M. P.; Strambio-de-Castillia, C. In Cell Biology: A Laboratory Hanbook; Celis, J. E., Ed.; Academic Press: San Diego, 1998; Vol. 2, pp 143-151. (28) van Asperen, J.; van Tellingen, O.; Beijnen, J. H. J. Chromatogr., B 1998, 712, 129. (29) Fogli, S.; Danesi, R.; Innocenti, F.; Di Paolo, A.; Bocci, G.; Barbara, C.; Del Tacca, M. Ther. Drug Monit. 1999, 21, 367-375. (30) Maessen, P. A.; Pinedo, H. M.; Mross, K. B.; van der Vijgh, W. J. F. J. Chromatogr., A 1988, 424, 103-110. (31) Perez-Ruiz, T.; Martinez-Lozano, C.; Sanz, A.; Bravo, E. Electrophoresis 2001, 22, 134-138. (32) Reinhoud, N. J.; Tjaden, U. R.; Irth, H.; van der Greef, J. J. Chromatogr. 1992, 574, 327-334. (33) Andersen, A.; Holte, H.; Slordal, L. Cancer Chemother, Pharmacol, 1999, 44, 422. (34) Hempel, G.; Haberland, S.; Schulze-Westhoff, P.; Mohling, N.; Blaschke, G.; Boos, J. J. Chromatogr., B 1997, 698, 287-292. (35) Hempel, G.; Schulze-Westhoff, P.; Flege, S.; Laubrock, N.; Boos, J. Electrophoresis 1998, 19, 2939-2943. (36) Anderson, A. B.; Gergen, J.; Arriaga, E. A. J. Chromatogr., B 2002, 769, 97-106.

Analytical Chemistry, Vol. 75, No. 1, January 1, 2003

9

EXPERIMENTAL SECTION Chemicals and Reagents. Doxorubicin hydrochloride and doxorubicinol were gifts from Dr. A. Suarato (Pharmacia, Nerviano, Italy). Doxorubicinone and 7-deoxydoxorubicinone were purchased from Qvantas, Inc. (Newark, DE). Sodium borate decahydrate was purchased from EM Science (Gibbstown, NJ). SDS Ultrapure Bioreagent was obtained from J. T. Baker (Phillipsburg, NJ). Sucrose, HPLC ChromaAR grade water, and methanol (MeOH) were purchased from Mallinckrodt (Paris, KY). Digitonin, R-cyano-4-hydroxycinamic acid (CCA), acetonitrile (ACN), and chloroform were purchased from Aldrich (Milwaukee, WI). Trifluoroacetic acid (TFA) was purchased from Spectrum Quality Products (Gardena, CA). D-Mannitol, ethylenediaminetetraacetic acid (EDTA), phosphate-buffered saline (PBS), Dulbecco’s modified eagle medium (DMEM), and bovine calf serum were purchased from Sigma (St. Louis, MO). Dimethyl sulfoxide (DMSO) was purchased from Burdick and Jackson (Muskegon, MI). N-(2-Hydroxyethyl)piperazine-2′-(2-ethanesulfonic acid) (HEPES) was purchased from EM Science. CE buffer was 10 mM borate and 10 mM SDS (pH 9.4) (BS buffer). Cell washing and suspension was done with either PBS or 210 mM D-mannitol, 70 mM sucrose, 5 mM HEPES, and 5 mM EDTA (pH 7.4) (buffer M). All buffers were made using HPLC grade water (Mallinckrodt) that had been filtered through a 0.22-µm Nalgene filter and stored at room temperature for up to one month. The pH of all solutions was adjusted with either HCl or NaOH. Following pH adjustment, buffers were filtered through a 0.22-µm Nalgene filter and stored at room temperature for up to one month. Stock solutions of DOX and metabolites were prepared in 100% MeOH at the following concentrations: 3 × 10-3 M DOX; 1.72 × 10-3 M DOXol; 1 × 10-3 M DOXone; 1 × 10-3 M 7-deoxyDOXone. The stock solutions were stored at -20 °C and used up to one month postpreparation. On the day of analysis, a working solution for each analyte was prepared in BS buffer to prevent repeated freeze/thaw cycling of the entire stock solution between -20 and 25 °C. Emission spectra for DOX, DOXol, DOXone, and 7-deoxyDOXone were obtained with a FP-6200 spectrofluorometer (Jasco, Tokyo, Japan). The analytes were diluted to a concentration of 1 × 10-7 M in BS buffer for spectral analysis. The sample was placed in a 1-cm-path length quartz cuvette (Starna Cells, Atascadero, CA). Safety Considerations. Extra care was taken when DOX and its metabolites were handled because of the carcinogenic nature of the compounds. All samples were sequestered and disposed of according to the MSDS and University of Minnesota hazardous waste policy. Cell Incubation and Sample Preparation. NS-1 cells were incubated at 37 °C and 5% CO2 in DMEM with 25 µM DOX for 8 h 2 days after splitting. Clinically relevant concentrations of DOX are typically lower than those used in this study.14,37-39 However, the selection of NS-1 cells that tolerate relatively high doses of DOX was considered more adequate for the development of the (37) Noel, G.; Peterson, C.; Trouet, A.; Tulkens, P. Eur. J. Cancer 1978, 14, 363-368. (38) Zenebergh, A.; Baurain, R.; Trouet, A. Eur. J. Cancer Clin. Oncol. 1984, 20, 115-121. (39) McHowat, J.; Swift, L. M.; Arutunyan, A.; Sarvazyan, N. Cancer Res. 2001, 61, 4024-4029.

10

Analytical Chemistry, Vol. 75, No. 1, January 1, 2003

analytical strategies described here. Cell cultures were split every 3 days by diluting 5 mL of cell culture to a final volume of 30 mL with fresh DMEM. Following incubation, 44 mL of cells was concentrated by pelleting four tubes of 11 mL at 1000g for 10 min using a Beckman J2-21 centrifuge with JA-21 rotor (Beckman Coultier, Fullerton, CA) and then resuspending the cells in a minimal volume of cold PBS. The concentrated suspensions were combined and washed twice in 1 mL of cold PBS. The final pellet was resuspended in 1.2 mL of cold PBS to obtain an average cell density of 8.6 × 106 cells/mL as determined with a Fuchs Rosenthal ultra plane counting chamber (Hausser Scientific, Horsham, PA) using an inverted stage microscope (Nikon, Melville, NY). Following dilution, the cell suspension was divided into six 200-µL aliquots. Each aliquot was added to a siliconized microfuge tube. One tube was used for whole-cell extraction; the remaining five were treated as replicate samples for subcellular studies. Each sample was treated with 1 µL of digitonin working solution (diluted to 6.67 mg/mL with PBS from 100 mg/mL in DMSO stock) and stored on ice for 5 min. This treatment permeabilized the cell plasma membrane, facilitating disruption by nitrogen cavitation. The microfuge tubes were placed into the reservoir of a nitrogen cavitator (Parr Instrument Co., Moline, IL) that had been prechilled to 4 °C. The cavitator was then packed in ice and charged to 600 psi with nitrogen for 20 min. The pressure was quickly released by manually opening the release valve. Due to incomplete lysis, an additional 1 µL of digitonin working solution was added and the samples were subjected to a second nitrogen cavitation. Following the second cavitation, ∼80% cell lysis was observed. Lysis was characterized by disruption of only the plasma membrane, not the nuclear or organelle membranes. The cell lysate was centrifuged at 1400g for 6 min. This 1400g pellet was designated the nuclear-enriched fraction. The supernatant from the 1400g pellet was removed and centrifuged at 14000g for 20 min. The 14000g pellet was designated the organelle-enriched fraction. The supernatant from the 14000g pellet was removed and designated the cytosole-enriched fraction. The whole-cell lysate and subcellular fractions were then subjected to liquid-liquid extraction as described below. The entire sample preparation and cavitation process took less than 2 h from cessation of the incubation to injection of the sample. Membranes, proteins, and other subcellular components were solubilized by adding 450 µL of BS buffer36 to the nuclear and organelle fractions and adding 250 µL of BS buffer to the supernatant fraction samples. Following addition of the BS buffer, the samples were sonicated for 1 h to enhance disruption. Following lysis, 1.05 mL of CHCl3/MeOH (5:1, v/v) was added to each sample to extract DOX and metabolites. The samples were vortexed for 10 min to improve the extraction followed by centrifugation for 10 min at 1000g to separate the layers. The less dense aqueous layer and emulsification were removed with a pipet and discarded. The samples were dried at 45 °C under a stream of nitrogen. Following drying, 100 µL of 100% MeOH was added to the vial to reconstitute the residue remaining in the sample vial. The sample was sonicated for 15 min to facilitate reconstitution. Sample preparations were stored at -20 °C in polypropylene vials for up to one month after extraction. CE-LIF Analysis. A 6-µL aliquot of each replicate was diluted to 16.7% MeOH (v/v) with 30 µL of BS buffer prior to analysis.

Fresh dilutions were done on the day of the analysis, and diluted samples were not stored for reuse after more than 12 h. Each of the five replicates was analyzed once. In all cases, the DOX peak was off scale at the original sample dilution; therefore, the samples were further diluted (10 000-fold for the whole-cell extract and 100-fold for the subcellular fractions) to estimate DOX peak area in the original sample dilution. Capillary electrophoresis of subcellular localization and metabolite identification samples was performed using a home-built capillary electrophoresis instrument previously described40 with separation voltage supplied by a CZE1000R high-voltage power supply (Spellman, Hauppauge, NY). The 488-nm line of an argon ion laser (model 532R-BS-A04, Melles Griot, Carlsbad, CA) was the wavelength of excitation used for analysis. Data were collected at 50 Hz using a Labview (National Instruments, Austin TX) program written in-house. Samples were separated under positive polarity in an uncoated fused-silica capillary with an internal diameter of 50 µm and outer diameter of 150 µm with 2 mm of the polyimide coating burned off each end of the capillary. Capillary length was 40.6 cm for the subcellular localization experiments and 37.9 cm for the metabolite identification experiments. Samples were injected electrokinetically for 5 s at +100 V/cm and then separated under +400 V/cm. The capillary outlet is located in a sheath flow cuvette40 where the laser excites the analytes as they exit the capillary. The fluorescence signal is collected at 90° with respect to the laser beam by a 60× microscope objective (Universe Kogaku (America), Inc., Oyster Bay, NY) and then sent through a 505-nm long-pass filter (Omega Optical, Brattleboro, VT) to remove scattering at the wavelength of excitation, a 1.4-mm pinhole, and finally a 635 ( 27.5 nm bandpass filter (Omega Optical) to a photomultiplier tube (Hamamatsu, Bridgewater, NJ) biased at 1000 V. Prior to the first analysis, the capillary was flushed with 1% HCl, water, 0.1 M NaOH, and water followed by BS buffer for 15 min each at 15 psi to condition the capillary. An in-house nitrogen line attached to a buffer pressurizing chamber provided pressure for capillary flushing. Ten replicate injections of a 1 × 10-8 M DOX + 1 × 10-8 M DOXol solution were done to expose the capillary to the analyte prior to analysis in order to obtain reproducible migration times and peak heights. The capillary was flushed with BS buffer for 5 min at 15 psi followed by an electrokinetic flush for 5 min between runs. To identify the metabolites, the whole-cell extract was not used because it was found that the higher abundance of the multiple components made identification more challenging (e.g., compare traces B and C in Figure 2). Instead, a nuclear fraction extract was analyzed by CE-LIF and then sequentially spiked with 1 × 10-8 M 7-deoxyDOXone, followed by 1 × 10-8 M DOXone, and followed by 1 × 10-8 M DOXol. After the addition of a metabolite, the fraction was analyzed and compared to the original, unspiked nuclear fraction extract. To evaluate the capillary conditioning procedure, data collected on a P/ACE MDQ CE System equipped with a 488-nm LIF module (Beckman Coultier) were used in addition to the data collected on the home-built instrument described above. The capillary for this set of experiments was a bare fused-silica capillary (50-µm i.d.) with total length of 40.4 cm and effective length of 30 cm. The wavelength of excitation was 488 nm, and fluorescence (40) Wu, S.; Dovichi, N. J. J. Chromatogr. 1989, 480, 141.

Figure 2. Capillary electrophoresis separation of a whole-cell extract not treated with DOX (A), a nuclear fraction extract (B), and wholecell extract of cells treated with DOX (C). Expansion of the region from 275 to 290 s to show peaks 10-12 (D). The DOX peak (300 s) was off scale and therefore not labeled. Whole-cell and nuclear fraction extracts were prepared as described in the Experimental Section. Separations were performed in a 40.6 cm × 50 µm i.d. fusedsilica capillary at +400 V/cm following 5-s injection at 100 V/cm, BS buffer. The 488-nm line of an argon ion laser was used for excitation, and a 635 ( 27.5 nm band-pass filter was used for detection. Traces have been offset for clarity.

emission was collected by a 635 ( 27.5 nm band-pass filter (Omega Optical). All experiments were carried out under normal polarity by applying a voltage of +400 V/cm. Samples were injected electrokinetically for 5 s using +100 V/cm. Matrix-Assisted Laser Desorption/Ionization Time-ofFlight Mass Spectrometry (MALDI-TOF MS). Preparation and subcellular fractionation of DOX-treated NS-1 cells for the MALDITOF MS experiments were performed as described above with two exceptions. Digitonin (diluted to 10 mg/mL with buffer M from 100 mg/mL in DMSO stock) was added to each sample to a final concentration of 1.7 × 10-2 mg/mL, and only one cycle of digitonin treatment followed by nitrogen cavitation was required to achieve ∼80% cell lysis. Following the fractionation by differential centrifugation, sample lysis was done by three freeze/ thaw cycles followed by sonication for 15 min. MALDI samples were prepared by mixing 3 µL of the sample (subcellular fraction lysate, DOX standard solution, or DOXol standard solution) with 30 µL of MALDI matrix (9.5 mg/mL CCA in ACN/MeOH (1:1, v/v) plus 0.1% TFA) and spotting 1 µL of sample-matrix mixture onto a stainless steel target. MALDI was performed on a Bruker Reflex III mass spectrometer (Bruker, Billerica, MA) in reflectron mode. Data Analysis. Data manipulation and analysis were performed using IGOR Pro (Wavemetrics, Lake Oswego, OR). All electropherograms were smoothed using a binomial smoothing function set at 10 points. In addition, a Median filter was applied to eliminate peaks having a baseline width of 10 points or less. All MALDI mass spectra were imported as ASCII files into IGOR Pro for manipulation. Single-factor analysis of variance (ANOVA) was performed using a Microsoft Excel spreadsheet. Three ANOVA were performed, one comparing the presence of metabolites in the nuclear-enriched fraction to the organelle-enriched fraction, a second comparing the nuclear-enriched and cytosole-enriched fractions, and a third comparing the presence of metabolites in the organelle-enriched fraction to the cytosole-enriched fraction. For all ANOVA, the null hypothesis, “there is no difference in Analytical Chemistry, Vol. 75, No. 1, January 1, 2003

11

the distribution of DOX metabolites between subcellular fractions”, was tested at a confidence interval (p) of 0.05 or less. Successful rejection of the null hypothesis resulted from a p e 0.05. RESULTS AND DISCUSSION Electropherogram Reproducibility. Migration time reproducibility is crucial for identification and quantification of DOX metabolites. Migration time drift can cause errors in peak identification and is therefore not acceptable in such a complex, multicomponent system (see Figure 2 for representative electropherogram and metabolite numbering). For example, by using a typical capillary conditioning program (rinsing with BS buffer only for 5 min at 20 psi prior to analysis), the migration time of metabolite 9 from a whole-cell extract (Figure 2) drifted 10.9 s over 10 consecutive runs. While for metabolite 1 the migration time decreased 3 s over 10 consecutive runs. Using a more extensive rinsing procedure and exposing the capillary to a mixture of DOX and DOXol prior to the analysis of metabolite extracts (see Experimental Section) greatly improved migration time reproducibility from run to run. Using this conditioning procedure, the migration times for metabolites 1 and 9 were 176.7 ( 0.76 and 269.7 ( 1.3 s, respectively (Figure 2). In addition, there was no apparent drift in migration time for these two metabolites. Similarly, the extensive rinsing procedure described above improved peak area reproducibility. For example, while the relative standard deviation in the peak area for metabolite 1 was similar between the typical and extensive conditioning procedures mentioned above (11%, n ) 10, and 13%, n ) 5, respectively), for metabolite 9, there was a dramatic improvement in peak area reproducibility (77%, n ) 10 versus 15%, n ) 5, respectively). It is likely that adsorption of the anthracycline metabolites to silica surfaces41 was decreased by the pretreatment of the capillary walls with the mixture of DOX and DOXol. The extensive procedure was adopted for the rest of the experiments. Metabolite Identification. The analysis of standards of DOXone, DOXol, and 7-deoxyDOXone (Figure 1) confirmed that DOXone and 7-deoxyDOXone but not DOXol could be separated from the parent drug (data not shown). For identification, we compared the electropherograms of a nuclear extract (Figure 3A), the same nuclear extract spiked with 7-deoxyDOXone (Figure 3B), and then the same spiked with DOXone (Figure 3C). Following the addition of the 7-deoxyDOXone standard to the nuclear extract, an increase in the peak area of both metabolite 10 and metabolite 11 was observed (Figure 3B). Comparison of the migration time of the 7-deoxyDOXone standard in free solution (tm ) 234 ( 2.3 s) (data not shown) with the migration time of metabolite 10 from the nuclear extract (tm ) 232 ( 1.1 s) suggests that the identity of metabolite 10 is 7-deoxyDOXone. It may be that metabolite 11 is a 7-deoxyDOXone-lipid complex formed upon mixing of the standard with the cell extract. Similar complexes (i.e., DOXcardiolipin) have been previously reported,42 so it would not be surprising if a complex is formed with phospholipids that are coextracted during the liquid-liquid extraction procedure.36 Metabolite 11 may also be a degradation product formed during the synthesis of the 7-deoxyDOXone standard.43 The ratio of this (41) Tjaden, U. R.; de Bruijn, E. A. J. Chromatogr. 1990, 531, 235-294. (42) Cheneval, D.; Muller, M.; Toni, R.; Ruetz, S.; Carafoli, E. J. Biol. Chem. 1985, 260, 13003-13007.

12 Analytical Chemistry, Vol. 75, No. 1, January 1, 2003

Figure 3. Capillary electrophoresis separation of a nuclear fraction extract (A). Nuclear fraction extract spiked with 1 × 10-8 M 7-deoxyDOXone (B). Nuclear fraction extract spiked with 1 × 10-8 M DOXone and 1 × 10-8 M 7-deoxyDOXone (C). The DOX peak (250 s) was off scale and therefore not labeled. Nuclear fraction extract was prepared as described in Experimental Section. Separation performed in a 37.9 cm × 50 µm i.d. fused-silica capillary. Other conditions as in Figure 2. Traces have been offset for clarity.

degradation product to 7-deoxyDOXone may increase as a result of further degradation during handling. Previously we reported that the addition of DOX to a blank cell extract just prior to liquidliquid extraction results in the detection of an additional component that migrates out after DOX.36 Although this product is not expected to interfere with the analysis presented here, it may be possible that the DOX metabolites degrade in a similar manner leading to an increased number of detected components. Other strategies discussed previously36 that minimize degradation such as single-cell analysis may help in the future to distinguish between naturally occurring metabolites and artifacts of sample handling. The identification of the DOXone peak (metabolite 8) was very straightforward. After spiking with DOXone, an increase in the peak area of metabolite 8 identified this metabolite as such (Figure 3C). The observed migration times for DOXone and 7-deoxyDOXone are in agreement with the expected migration order and the structural features of DOX metabolites (Figure 1). Their structure affects their distribution between the micellar and the aqueous buffer phases present in a micellar electrokinetic chromatography (MEKC) separation. At the pH used in this study (pH 9.4), aglycon metabolites carry an additional negative charge due to the absence of the daunosamine sugar (Figure 1) and therefore partition less than DOX (that has a zero net charge) into the anionic SDS micelles. Since the anionic SDS micelles have negative electrophoretic mobility, they oppose bulk flow movement imposed by the electroosmotic flow present in separations using positive polarity voltages. Therefore, in an MEKC separation based on SDS micelles, negatively charged aglycons are expected to be retained less by the micelles and migrate out faster than DOX or DOXol. In contrast to the aglycons, the DOXol standard was not resolved from DOX standard in the MEKC separation. Therefore, (43) Personal Communication: Xie, M. Qventas, Inc., Newark, DE, 2002.

Figure 4. MALDI-TOF mass spectra of expected parent mass region for DOX standard (A), aglycon region of DOX standard in MeOH (B), aglycon region for DOXol standard in MeOH (C), nuclear fraction from DOX-treated cells (D), Organelle fraction from DOXtreated cells (E), and cytosolic fraction from DOX treated cells (F). MALDI matrix was R-cyano-4-hydroxycinamic acid plus 1% TFA.

to confirm the presence of DOXol in the subcellular fractions, we used MALDI-TOF mass spectrometry (Figure 4). Peaks were not detected at the molecular weight of the protonated species (M + 1) of either parent compound in the standards (544 amu for DOX and 546 amu for DOXol, Figure 4A). Panels B and C of Figure 4 show regions of the mass spectra of DOX and DOXol where peaks were detected at the molecular weight of the protonated aglycons (M +1) of DOX and DOXol (398 and 400 amu, respectively). These results indicated that the MALDI process cleaves the glycosidic bond between the aglycon and the daunosamine sugar resulting in detection of only the aglycon. On the basis of the peak intensities of the DOX and DOXol aglycons we were able to estimate the relative abundance for DOXol with respect to DOX to be 20% in the nuclear fraction, 50% in the organelle fraction, and 80% in the cytosolic fraction (Figure 4D-F, respectively). From the relative abundance and the amount of DOX calculated from CE-LIF experiments (n ) 5), we estimated the absolute abundance of DOXol in each fraction (Figure 5). Indeed, this value is subject to the low quantitative power of MALDI and should be used only as an estimate. In addition to these metabolites, nine metabolites were not identified at this time (metabolites 1-7, 9, and 12; Figure 2). On the basis of the structure of the metabolite standards and the relative migration time of the metabolites we can speculate about the identity of those metabolites otherwise unidentified. As discussed previously, that DOX aglycons migrate faster than DOX may result from a decreased partitioning of the aglycon into the SDS micelle resulting from a net negative charge in the aglycon structure (see Figure 1, for aglycon structure examples). Along the same line, metabolites 1-7 also migrate faster and are therefore likely to be more polar than DOX and the other identified aglycons. We can speculate that the identity of these metabolites may be either other aglycons or DOX metabolite-phospholipid complexes.44 Based on the migration order observed for the series DOXone (metabolite 8), 7-deoxyDOXone (metabolite 10), and DOX it is possible that metabolites 9 and 11 are the DOXol

Figure 5. Comparison of metabolite amount in three subcellular fractions. Data for the subcellular fractions corresponds to the average of five independent extractions and their corresponding CE-LIF analysis. Amount/cell was corrected for differences in fluorescence intensity of the metabolites. Unidentified metabolites are reported as if their fluorescent properties were those of DOX. Arrows indicate metabolites that are not present in all subcellular fractions.

analogues of DOXone (doxorubicinolone) and 7-deoxyDOXone (7-deoxydoxorubicinolone), respectively. It would be expected then that comigration of DOXol and DOX is not reflected by the corresponding aglycons because the aglycons do not retain the daunosamine sugar, which is expected to be the dominant factor in the comigration of DOXol and DOX. Standards for doxorubicinolone and 7-deoxydoxorubicinolone certainly would allow for clarification of these suggested identities. However, these standards were not available at the time of this analysis. Other efforts to conclusively identify these metabolites by CE-LIF-MS are underway. Spectroscopic Properties of DOX Metabolites. A DOX calibration curve was constructed using six concentrations from 5 × 10-8 M to 10-10 M. The equation, y ) (1.24 ( 0.05) × 10-9X - (0.30 ( 8.1) × 10-10 (r2 ) 0.9894), where y is molar concentration, X is the peak area (signal × time (V‚s)), and r is the correlation coefficient, predicts a linear dependence. However, the fluorescence emission spectrum and the sensitivity of the detection method for each metabolite standard and DOX may vary as a result of the differences in molecular structure. For DOX and DOXol, the integrated fluorescence intensities over the spectral band-pass of the 635(27.5 nm filter were nearly identical (18 fluorescence units (FU) versus 17 FU, respectively). Therefore, by using this filter in the CE-LIF analysis, no correction for quantification is necessary. On the other hand, there was a substantial difference in the integrated fluorescence intensities of DOXone and 7-deoxyDOXone versus the fluorescence intensity of DOX. The integrated fluorescence intensity was 72 and 11% of DOX for DOXone and 7-deoxyDOXone, respectively. The drastic decrease in relative fluorescence for 7-deoxyDOXone seems to result from the absence of an oxygen atom that is able to contribute to the π system of the aglycon ring, which confers the native fluorescence to DOX and its metabolites (Figure 1). When the identity of a metabolite is known, such as for metabolite 8 (DOXone) and metabolite 10 (7-deoxyDOXone), the ratio of the (44) Goormaghitgh, E.; Chatelain, P.; Caspers, J.; Ruysschaert, J. M. Biochim. Biophys. Acta 1980, 597, 1-14.

Analytical Chemistry, Vol. 75, No. 1, January 1, 2003

13

DOX fluorescence intensity to the metabolite fluorescence intensity can be used to compensate for the differences in fluorescence intensity in order to improve the quantitation of these metabolites. In this report, the peak area of DOXone was multiplied by a factor of 1.4 and the peak area of 7-deoxyDOXone was multiplied by a factor of 9.0 for this reason. Since no standards or specific structural information is available for the remaining unidentified metabolites, the fluorescence properties of these metabolites will be reported as those of DOX. This assumption facilitates the discussion of the abundance of the unknown metabolites in the following subsection and may facilitate a retroactive quantification once structural standards become available. Subcellular Localization of DOX Metabolites. The CE-LIF strategy described above was used to compare the amounts of metabolites in different subcellular fractions prepared by differential centrifugation (Figure 5). It is clear from these studies that the strategy was effective in determining differences in how each metabolite is distributed among the various subcellular fractions despite the large errors associated with handling microliter-size volumes in the centrifugation and extraction procedure. However, the interpretation of the metabolite distribution among these fractions needs to consider the composition of each subcellular fraction. Several independent reports indicate that the composition of subcellular fractions as determined by enzymatic activity assays is highly consistent (see lower part of Table 1).21,23,45 Based on these reports, the fraction pelleted at 1400g would contain nearly all of the nuclei (nu),23 and a small fraction of unbroken and partially broken cells (CL), mitochondria (mt), and large sheets of plasma membrane.21,23,45 The fraction pelleted at 14000g contains mainly mitochondria and lysosomes (ly) and also large microsomes (ms).21 The supernatant remaining after pelleting at 14000g is considered the cytosole-enriched fraction because it contains soluble components (cy), small membranebound structures, and microsomes found in the cytoplasm,21 although the presence of lysosomal activity has also been documented.23 To simplify the interpretation of the data, we chose to classify the metabolites according to the statistical significance of the differences between the different fractions. Table 1 summarizes the data for each of the five groups that are defined based on these criteria. Gray tones and shading have been selected to indicate the relative abundance for a given metabolite. Group I: Metabolite 10 (7-deoxyDOXone) showed no statistically significant difference in abundance between any of the fractions, suggesting that this metabolite either does not have a specific localization or that cell disruption induces redistribution of this metabolite. Group II: The abundances of metabolites 2 and 13 (DOXol) were found to be statistically different between all three subcellular fractions. Metabolite 2 was found to be most abundant in the nuclear-enriched fraction. DOXol was detected at significantly higher amounts in both the nuclear-enriched and cytosoleenriched fractions. However, the abundance of DOXol in the subcellular fractions was calculated from the relative abundances as determined in the MALDI-TOF-MS, a technique that is not considered quantitative. (45) Madden, E. A.; Storrie, B. Anal. Biochem. 1987, 163, 350-357.

14 Analytical Chemistry, Vol. 75, No. 1, January 1, 2003

Table 1. Comparison among Subcellular Fractionsd

a DOXone. b 7-DeoxyDOXone. c DOXol. d A statistically significant difference (p e 0.05) in the abundance of a given metabolite between fractions is indicated by a difference in the shading of the boxes. The shading from black, through three shades of progressively lighter gray, to white respectively corresponds to the following: (1) highest abundance; (2) intermediate abundance; (3) abundance was not statistically different from the abundance in either the nuclear-enriched or cytosole-enriched fraction (metabolites 4 and 5); (4) lowest abundance; (5) no metabolite detected. The chart below the table indicates within which subcellular fraction(s) the specified cellular components are found following the subcellular fractionation described in this report. Abbreviations: Nu-E, nuclear-enriched; Or-E, organelleenriched; Cy-E, cytosole-enriched; CL, whole cells; nu, nuclei; mt, mitochondria; ly, lysosomes; ms, microsomes; cy, cytoplasm.

Group III: Metabolites 3, 6, 8 (DOXone), 9, and 12 have similar abundances in the organelle-enriched and cytosole-enriched fraction but statistically different abundances in the nuclearenriched fraction. Metabolite 3 is distinct among this group in that it is not found in the nuclear-enriched fraction, suggesting that it is not localized in the nucleus. This interpretation is further validated since other impurities present in the nuclear-enriched fraction (plasma membrane, organelles, and whole cells) that may contribute to detection of this metabolite in this fraction were not sufficient to cause this effect. Group IV: Metabolite 7 was not detected in the organelleenriched fraction (mitochondria and lysosomes) making its abundance statistically different from the other two fractions. This could be the result of dilution of the organelle contents below the limit of detection of this method. Thus, it does not exclude the possibility that metabolite 7 is localized in either mitochondria or lysosomes. Group V: Metabolites 4, 5, and 11 are characterized by statistically significant differences between the nuclear-enriched and the cytosole-enriched fractions but not between the nuclearenriched and organelle-enriched fractions. Metabolite 11 is not present in the cytosole-enriched fraction but only in the other two fractions. As described earlier, metabolite 11 is related to 7-deoxyDOXone and possibly a phospholipid complex, which would suggest that it could be localized on the nuclear and at least one of either mitochondrial or lysosomal membranes. Although highly speculative, the distributions of metabolites 4 and 5 may indicate

that they are also preferentially localized in membranes including those of microsomes. CONCLUSIONS The results presented here illustrate that DOX metabolites that are present at extremely low levels can be detected by CE-LIF while the highly abundant DOXol can be detected by MALDITOF MS. This report describes 12 metabolites produced in cellulo, which is the largest number of DOX metabolism components described to date. Due to the novel nature of these metabolites and the lack of standards, only DOXone and 7-deoxyDOXone were identified and the possibility that some of the components could be artifacts cannot be ruled out. The strategy was further applied to DOX metabolites in three enriched subcellular fractions. The abundance of metabolites 1-12 varied from 0 to 1300 zmol. Some metabolites, such as DOXol, DOXone, and 7-deoxyDOXone, were found in all fractions. Metabolite 11, a 7-deoxyDOXone-related component, seems to be a complex with phospholipids found in the nuclear or organellar membranes. Overall, the CE-LIF method described here for the comparison of zeptomole-abundant metabolites could be extended to other analyses where differences

in abundance are key such as dosage, kinetics, and drug transport studies. Further work will focus on the identification of those yet unidentified metabolites and study of the subcellular distribution of DOX metabolites in other cell lines, tissues, and plasma relevant to cancer treatment. These experiments may provide a more complete description of metabolism and further our understanding of these metabolites in cytotoxicity, drug resistance, and mechanism of action. ACKNOWLEDGMENT Financial support from the National Institute of Health (R01GM61969) is greatly appreciated. We thank Nilhan Gunasekera, who cultured the NS-1 cells, Dr. Sally Palm, who provided the initial cell supply, and Dr. Antonio Suarato, Pharmacia, for his kind donation of DOX and 13-dihydrodoxorubicin (doxorubicinol). C.C. thanks the University of Minnesota Lando program.

Received for review July 1, 2002. Accepted October 9, 2002. AC020426R

Analytical Chemistry, Vol. 75, No. 1, January 1, 2003

15