Comparison of Drug Distribution Images from Whole-Body Thin

Anal. Chem. 2008, 80, 5168–5177

Comparison of Drug Distribution Images from Whole-Body Thin Tissue Sections Obtained Using Desorption Electrospray Ionization Tandem Mass Spectrometry and Autoradiography Vilmos Kertesz,*,† Gary J. Van Berkel,† Marissa Vavrek,‡ Kenneth A. Koeplinger,‡ Bradley B. Schneider,§ and Thomas R. Covey§ Organic and Biological Mass Spectrometry Group, Chemical Sciences Division, Oak Ridge National Laboratory, Oak Ridge, Tennessee 37831-6131, Drug Metabolism and Pharmacokinetics, Merck Research Laboratories, West Point, Pennsylvania 19486-0004, and MDS Analytical Technologies, Concord, Ontario, L4K 4V8, Canada Desorption electrospray ionization tandem mass spectrometry (DESI-MS/MS) and whole-body autoradiography (WBA) were used for chemical imaging of whole-body thin tissue sections of mice intravenously dosed with propranolol (7.5 mg/kg). DESI-MS/MS imaging utilized selected reaction monitoring detection performed on an AB/MDS SCIEX 4000 QTRAP mass spectrometer equipped with a prototype extended length particle discriminator interface. Propranolol images of the tissue sections using DESI-MS/MS were obtained at surface scan rates of 0.1, 0.5, 2, and 7 mm/s. Although signal decreased with increasing scan rate, useful whole-body images for propranolol were obtained from the tissues even at 7 mm/s, which required just 79 min of analysis time. Attempts to detect and image the distribution of the known propranolol metabolites were unsuccessful. Regions of the tissue sections showing the most radioactivity from WBA sections were excised and analyzed by highperformance liquid chromatography (HPLC) with radiochemical detection to determine relative levels of propranolol and metabolites present. Comparison of the DESI-MS/ MS signal for propranolol and the radioactivity attributed to propranolol from WBA sections indicated nominal agreement between the two techniques for the amount of propranolol in the brain, lung, and liver. Data from the kidney showed an unexplained disparity between the two techniques. The results of this study show the feasibility of using DESI-MS/MS to obtain useful chemical images of a drug in whole-body thin tissue sections following drug administration at a pharmacologically relevant level. Further optimization to improve sensitivity and enable detection of the drug metabolites will be among the requirements necessary to move DESI-MS/MS chemical imaging forward as a practical tool in drug discovery. At the present time, whole-body autoradiography (WBA) using radiolabeled drugs is a standard method for quantitative chemical imaging of the distribution of a drug and its metabolites in thin * Corresponding author. Phone: 865-574-4878. Fax: 865-576-8559. E-mail: [email protected]. † Oak Ridge National Laboratory. ‡ Merck Research Laboratories. § MDS Analytical Technologies.

5168

Analytical Chemistry, Vol. 80, No. 13, July 1, 2008

tissue sections.1,2 However, there are three main drawbacks to this technique that limit its utility. First, WBA requires a radiolabeled drug for the imaging experiment. Second, this approach does not provide molecular structure information. Thus, metabolites either cannot be distinguished from the parent drug or are not detected if the radiolabel is lost in a metabolic process. And third, development of a WBA image can require a 4-7 day exposure time depending on the radiolabel utilized. Information acquired by mass spectrometry compliments WBA because it can provide information about the particular molecular form of the drug-related material in the tissue. Mass spectrometrybased tissue imaging is generally label-free and can provide chemical images in 24 h or less. Multiple mass spectrometry-based surface sampling/ionization and chemical imaging techniques are currently being investigated for thin tissue section analysis and drug discovery applications.3 These techniques include, among others, secondary ion mass spectrometry (SIMS) and matrixassisted laser desorption ionization mass spectrometry (MALDI-MS).4,5 The latter technique has been particularly successful for the analysis of endogenous large-mass biomolecules such as proteins.6,7 MALDI-MS is also now being used for the analysis of drugs and their associated metabolites in tissues.8,9 Newer atmospheric pressure surface sampling/ionization techniques like electrospray-assisted laser desorption ionization (ELDI),10 atmospheric pressure (AP)-MALDI-MS,11,12 laser ablation (1) Solon, E. G.; Balani, S. K.; Lee, F. W. Curr. Drug Metab. 2002, 3, 451– 462. (2) Food and Drug Administration; 1005; 21CFR312.23. http://www.fda.gov. (3) Rubakhin, S. S.; Jurchen, J. C.; Monroe, E. B.; Sweedler, J. V. Drug Discovery Today 2005, 10, 823–837. (4) Todd, P. J.; Schaaff, T. G.; Chaurand, P.; Caprioli, R. M. J. Mass Spectrom. 2001, 36, 355–369. (5) McDonnell, L. A.; Heeren, R. M. A. Mass Spectrom. Rev. 2007, 26, 606– 643. (6) Chaurand, P.; Schwartz, S. A.; Reyzer, M. L.; Caprioli, R. M. Toxicol. Pathol. 2005, 33, 92–101. (7) Chaurand, P.; Cornett, D. S.; Caprioli, R. M. Curr. Opin. Biotechnol. 2006, 17, 431–436. (8) Khatib-Shahidi, S.; Andersson, M.; Herman, J. L.; Gillespie, T. A.; Caprioli, R. M. Anal. Chem. 2006, 78, 6448–6456. (9) Stoeckli, M.; Staab, D.; Schweitzer, A. Int. J. Mass Spectrom. 2007, 260, 195–202. (10) Huang, M.-Z.; Hsu, H.-J.; Wu, C.-I.; Lin, S.-Y.; Ma, Y.-L.; Cheng, T.-L.; Shiea, J. Rapid Commun. Mass Spectrom. 2007, 21, 1767–1775. (11) Creaser, C. S.; Ratcliffe, L. Curr. Anal. Chem. 2006, 2, 9–15. (12) Li, Y.; Shrestha, B.; Vertes, A. Anal. Chem. 2007, 79, 523–532. 10.1021/ac800546a CCC: $40.75  2008 American Chemical Society Published on Web 05/16/2008

electrospray ionization (LAESI)-MS,13 and desorption electrospray ionization (DESI)-MS14–19 are quickly developing, potentially complementary or alternative surface sampling/ionization approaches to tissue analysis and imaging. Some of the more important factors in the practical applicability of a mass spectrometry-based tissue imaging technique for drug discovery applications include detection levels, quantification, and analysis time. Khatib-Shahidi et al.8 reported imaging of olanzapine and its N-desmethyl and 2-hydroxymethyl metabolites in rat thin tissue sections following 8 mg/kg oral dosing using MALDI-MS/ MS. However, mass spectrometry data were not compared to WBA images. In that paper, 400 µm × 400 µm pixels were used to create the chemical image with each selected reaction monitoring (SRM) transition monitored for 4 s (4 s/pixel), corresponding to a surface scan rate of 100 µm/s if only one transition was monitored. At this speed, imaging of a 100 mm × 30 mm area with 200 µm lane spacing, which are approximately the size of a whole-body mouse sagittal tissue section and the vertical resolution used in our work, would take about 42 h. Stoeckli et al.9 used MALDI-MS imaging to measure distribution of a 14C-labeled compound in whole-body tissue sections from a rat dosed intratracheally (0.5 mg/kg) and compared the results to WBA images. Analysis time of a whole-body rat tissue section (approximately 174 mm × 43.5 mm in size, 500 µm × 500 µm pixel size) in full scan and MS/MS modes took approximately 4.8 and 44 h which equated to approximately 890 µm/s and 100 µm/s scan speeds, respectively. In their study, analyte-specific ionization efficiency, tissue-specific ion suppression, and matrix deposition were found to be issues affecting quantitative analysis. Good quantitative agreement was found between the MALDI-MS and WBA images of the same tissue section when using the appropriate calibration procedures. To date, DESI-MS has not been used to image whole-body thin tissue sections from small animals dosed with a drug. However, Wiseman et al.19 have presented data that demonstrates DESIMS imaging of a sagittal brain section from a rat dosed with 250 µg of clozapine via an intracerebral ventricular injection. The surface scan speed in that experiment was 200 µm/s. That is the same speed as the fastest published DESI-MS surface scan rate that has been used for imaging endogenous compounds in thin tissue sections.18,19 This imaging speed is very comparable to that reported for imaging of drugs and metabolites using MALDI-MS/MS.8,9 In this paper, we compare DESI-MS/MS and WBA methodologies for examining the distribution of intravenously dosed propranolol in whole-body mouse thin tissue sections. Propranolol is a well-studied β-adrenergic receptor antagonist, which was originally introduced for the treatment of angina pectoris and subsequently found wide use as an antihypertensive and (13) Nemes, P.; Vertes, A. Anal. Chem. 2007, 79, 8098–8106. (14) Taka´ts, Z.; Wiseman, J. M.; Gologan, B.; Cooks, R. G. Science 2004, 306, 471–473. (15) Taka´ts, Z.; Wiseman, J. M.; Cooks, R. G. J. Mass Spectrom. 2005, 40, 1261– 1275. (16) Wiseman, J. M.; Puolitaival, S. M.; Taka´ts, Z.; Cooks, R. G.; Caprioli, R. M. Angew. Chem., Int. Ed. 2005, 44, 7094–7097. (17) Wiseman, J. M.; Ifa, D. R.; Song, Q.; Cooks, R. G. Angew. Chem., Int. Ed. 2006, 45, 7188–7192. (18) Ifa, D. R.; Wiseman, J. M.; Song, Q.; Cooks, R. G. Int. J. Mass Spectrom. 2007, 259, 8–15. (19) Wiseman, J. M.; Ifa, D. R. Nat. Protoc. 2008, 3, 517–524.

antiarrhythmic agent.20 Because of its high lipophilicity, ability to penetrate the central nervous system, and attenuating effects on long-term memory potentiation, propranolol has also recently undergone extensive investigation for the treatment of posttraumatic stress disorder.21,22 The distribution of propranolol and its metabolites in rat23 and mouse24 has been characterized by autoradiography in the past but never by mass spectrometrybased imaging. The widespread distribution,25 the high levels of drug achieved in major organs such as the lung and brain,23,24,26,27 and its ionization efficiency by mass spectrometry make propranolol an ideal imaging candidate. For this study, mice were intravenously dosed (7.5 mg/kg) to eliminate extensive first-pass metabolism of propranolol in the liver that occurs after oral dosing.28 The results presented here demonstrate the potential of using DESI-MS/MS to obtain useful chemical images of the parent drug in whole-body mouse tissue sections in less than 1.5 h (7 mm/s scan speed). Qualitative distributions of parent drug levels were comparable with both techniques for most, but not all, organs examined. These discrepancies, low signal levels for the parent drug, and the inability to detect known metabolites present in the tissue will need to be addressed and overcome. EXPERIMENTAL SECTION Chemicals. HPLC grade methanol and water were purchased from Burdick & Jackson (Muskegon, MI). D,L-Propranolol hydrochloride and [D,L-propranolol-[4-3H] hydrochloride (in ethanol; 27 Ci/mmol) were purchased from Sigma-Aldrich (St. Louis, MO). A 0.1 µM propranolol test solution was prepared in 50/50 (v/v) methanol/water. Tissue Preparation for MS Tissue Imaging. Mice (male CD1; Charles River Laboratories) were administered propranolol intravenously via the tail vein at 7.5 mg/kg as an aqueous solution in 0.9% NaCl. At 20 or 60 min postdose, mice were euthanized with an isoflurane overdose and immediately frozen in dry ice/ hexane. The frozen mice were embedded/blocked in 2% aqueous carboxymethyl cellulose. Sagittal whole-body cryosections (40 µm thick) were prepared using a Leica CM3600 cryomacrotome. Frozen sections were transferred to 3 in. × 4 in., 1.2 mm thick glass slides (Brain Research Laboratories) using a tape transfer process (Macro-Tape-Transfer System, Instrumedics, St. Louis, MO), which utilizes a photoactivated polymer adhesive. After transfer to the glass slides, the sections were freeze-dried within the chamber of the cryomacrotome. All tissue sections were stored in a desiccator until analysis. Color images of the tissue sections were acquired using an HP Scanjet 4370 flat-bed scanner (HewlettPackard, Palo Alto, CA). (20) Hamer, J.; Grandjean, T.; Melendez, L.; Sowton, G. E. Br. Med. J. 1964, 2, 720–723. (21) Evers, K. Cambridge Q. Healthcare Ethics 2007, 16, 138–146. (22) Famularo, R.; Kinscherff, R.; Fenton, T. Am. J. Dis. Child. 1988, 142, 1244– 1247. (23) Schiff, A. A.; Saxey, A. Xenobiotica 1984, 14, 687–691. (24) Masuoka, D.; Hansson, E. Acta Pharmacol. Toxicol. 1967, 25, 447–455. (25) Evans, G. H.; Nies, A. S.; Shand, D. G. J. Pharmacol. Exp. Ther. 1973, 186, 114–122. (26) Schneck, D. W.; Pritchard, J. F.; Hayes, A. H. J. Pharmacol. Exp. Ther. 1977, 203, 621–629. (27) Bianchetti, G.; Elghozi, J. L.; Gomeni, R.; Meyer, P.; Morselli, P. L. J. Pharmacol. Exp. Ther. 1980, 214, 682–687. (28) Hayes, A.; Cooper, R. G. J. Pharmacol. Exp. Ther. 1971, 176, 302–311.


5169

Tissue Preparation for Quantitative Whole-Body Autoradiography (QWBA): Total Radiolabeled Drug-Related Material. The radiolabeled dosing solution was prepared by evaporating ethanol from a portion of D,L-propranolol-[4-3H] hydrochloride (in ethanol; 27 Ci/mmol; Sigma-Aldrich) under a stream of nitrogen followed by the addition of nonradiolabeled propranolol (D,L-propranolol hydrochloride, Sigma-Aldrich) and 0.9% NaCl. The specific activity of the resulting solution was 35.27 nCi/nmol. Mice (male CD-1; Charles River Laboratories) were administered the [3H]propranolol solution intraveneously via the tail vein at 7.5 mg/kg (1 mCi/kg radioactive dose) as an aqueous solution in 0.9% NaCl. At 20 or 60 min postdose, mice were euthanized by an isoflurane overdose and immediately frozen in dry ice/hexane. The frozen mice were embedded in 2% aqueous carboxymethyl cellulose. Holes 6 mm in diameter were drilled in the frozen blocks, and blood standards prepared with [3H]glucose (D-glucose-1-3H(N), Sigma-Aldrich) ranging from ∼100 to ∼15 000 nCi/g were added. Sagittal whole-body cryosections (40 µm thick) were collected onto 3M 810 adhesive tape (20 cm × 7.5 cm) using a Leica CM3600 cryomacrotome. The sections were freeze-dried within the chamber of the cryomacrotome. All tissue sections were stored in a desiccator until analysis. Color images of the tissue sections were acquired using a HP Scanjet 4370 flat-bed scanner. Selected sections were exposed to Fujifilm BAS-TR 20 cm × 40 cm phosphor imaging plates for 7 days in a lead-shielding box (Raytest, Straubenhardt, Germany). Plates were analyzed using a Fujifilm FLA-5100 phosphor imager with AIDA software for QWBA image analysis (Raytest). Areas of radioactivity on the resulting autoradioluminograph were quantitated in nmol · eq propranolol/g tissue using linear regression with blood standards present in each section with AIDA software. Propranolol concentration was then determined based on radioprofiling results for percent total radioactivity represented by unchanged parent drug. Radioprofiling of Drug-Related Material and Metabolite Identification from Brain, Kidney, Liver, and Lung Tissue Samples. Small samples of the tissues of interest were removed from 40 µm thick whole-body mouse sections (WBA sections on adhesive tape) using Miltex dermal biopsy punches of various diameters. Each tissue sample was extracted in 500 µL of 50/50 (v/v) acetonitrile/water. Samples were dried down under N2 and reconstituted in 200 µL of water. Samples were injected (50 µL) into an Agilent 1100 HPLC with a 96-well Foxy fraction collector, and 30 s fractions were collected into Wallac isoplates. The HPLC method was run using a Phenomenex Luna C18 150 mm × 2 mm column (pore size, 100 Å; particle size, 5 µm), flow rate 0.5 mL/ min, mobile phase A, 10 mM ammonium acetate, mobile phase B, acetonitrile, 7 min at 100% A, 7-45 min gradient to 85% B, and return to 100% A for 5 min. Samples were dried down under N2, and 200 µL of OptiPhase “SuperMix” liquid scintillation cocktail (Perkin-Elmer) was added to each well. A Perkin-Elmer MicroBeta TriLux 1450 liquid scintillation and luminescence counter was used to count radioactivity in the 96-well plate samples for 5 min each. Peak areas were integrated using Laura Lite radiochromatography software (LabLogic, Sheffield, England). Propranolol was confirmed in the brain, kidney, liver, and lung using an Advion NanoMate nanospray source with a 4000 QTRAP mass spectrom5170


eter (MDS SCIEX, Concord, Ontario, Canada) by infusing the reconstituted fractions. Attempts to confirm metabolites were unsuccessful. In a separate experiment, brain, kidney, liver, and lung were excised and rinsed in saline from a mouse (male CD-1; Charles River Laboratories) dosed intravenously via the tail vein at 7.5 mg/kg [3H]propranolol in 0.9% NaCl (1 mCi/kg radioactive dose), then sacrificed 60 min postdose. The radiolabeled dosing solution was prepared by evaporating ethanol from a portion of D,Lpropranolol-[4-3H] hydrochloride (in ethanol; 27 Ci/mmol; SigmaAldrich) under a stream of nitrogen followed by the addition of nonradiolabeled propranolol (D,L-propranolol hydrochloride, SigmaAldrich) and 0.9% NaCl. The specific activity of the resulting solution was 40.21 nCi/nmol. Organs were homogenized in three portions (w/v) water using a small tissue homogenizer. Samples were extracted in two portions (v/v) 50/50 acetonitrile/water (v/ v). Samples were dried down under N2 and reconstituted in 10 mM ammonium acetate. Radioprofile of the brain sample indicated the presence of only the parent drug. The identity of metabolites represented in the radioprofiles from lung, liver, and kidney extracts was determined using LC-MS/MS in conjunction with radiochemical detection. Lung, liver, and kidney extract samples were injected (100 µL) into an Agilent HPLC. The HPLC method was run using a Phenomenex Luna C18 150 mm × 2 mm column (pore size, 100 Å; particle size, 5 µm), flow rate 0.5 mL/min, mobile phase A, 10 mM ammonium acetate, mobile phase B, acetonitrile, 7 min at 100% A, 7-45 min gradient to 85% B, and return to 100% A for 5 min. The flow from the HPLC was split 4:1 between a radiochemical detector (IN/US Beta-RAM) and a Finnigan TSQ Quantum Ultra MS (ESI/positive ion mode). In the lung, liver, and kidney the major [3H] metabolites were identified as two different hydroxypropranolol glucuronides (m/z 452) with major product ions at m/z 276, 199, 173, and 116. Radioprofile of the liver sample of a mouse sacrificed 60 min postdose and the product ion spectra of the two different hydroxypropranolol glucuronides (m/z 452) are shown in Supporting Information Figure S1. DESI-MS System. All DESI-MS experiments were performed on a 4000 QTRAP mass spectrometer equipped with a particle discriminator interface (PDI).29–32 Evaluation of DESI-MS/MS signal level dependence on heated chamber length of the PDI was accomplished using 2.0, 4.0, 7.0, and 12 cm long chambers (Figure 1). Propranolol was monitored using SRM (m/z 260 f 116, collision energy ) 27 eV) (Scheme 1). Performance of the PDI was tested with each of the different chamber lengths by continuous infusion of the 0.1 µM propranolol test solution at 2.5, 5.0, and 10 µL/min in an on-axis position about 1 cm from the entrance of the PDI. The DESI sprayer, surface control and automation, and data analysis software have been described in detail elsewhere.33 (29) Schneider, B. B.; Baranov, V. I.; Javaheri, H.; Covey, T. R. J. Am. Soc. Mass Spectrom. 2003, 14, 1236–1246. (30) Schneider, B. B.; Lock, C.; Covey, T. R. J. Am. Soc. Mass Spectrom. 2005, 16, 176–182. (31) Corkery, L. J.; Pang, H.; Schneider, B. B.; Covey, T. R.; Siu, K. W. M. J. Am. Soc. Mass Spectrom. 2005, 16, 363–369. (32) Leuthold, L. A.; Mandscheff, J.-F.; Fathi, M.; Giroud, C.; Augsburger, M.; Varesio, E.; Hopfgartner, G. Rapid Commun. Mass Spectrom. 2006, 20, 103–110. (33) Van Berkel, G. J.; Kertesz, V. Anal. Chem. 2006, 78, 4938–4944.

Scheme 1. Structure and Mass-to-Charge Ratio of Propranolol and Origin of Major Product Ions

Figure 1. Schematics of a 4000 QTRAP mass spectrometer atmospheric sampling interface region equipped with (a) a normal nanospray source and (b) a modified PDI. (c) Photograph taken during imaging of a mouse thin tissue section showing the DESI emitter, surface, and sampling inlet. (d) Relative SRM signal recorded for propranolol (m/z 260 f 116) with the (filled bars) 1 cm long heated chamber (nanospray source) and with the (empty bars) 2 cm, (hatched bars) 7 cm, and (cross-hatched bars) 12 cm long heated chamber PDIs spraying a 0.1 µM test solution (1/1 water/methanol) with a typical on-axis electrospray configuration at 2.5, 5, and 10 µL/ min flow rates.

All chemical images were acquired using a unidirectional scanning mode34 and using 80/20 (v/v) methanol/water at 5 µL/min as the DESI spray solvent. At the beginning of an imaging experiment the spray impact plume was positioned at one corner of the area of interest, and the sampling end of the PDI heated chamber was positioned to just touch the tissue. The first lane was scanned by moving the surface parallel to the x-axis at a forward surface scan rate with the spray plume fixed in the optimal position in front of the sampling inlet. At the end of the first lane, the surface was lowered by 2 mm (34) Kertesz, V.; Van Berkel, G. J. Anal. Chem. 2008, 80, 1027–1032.

followed by moving the surface at a return speed back to the beginning of the first lane. With the surface lowered there was a 2 mm gap between the sampling capillary and the surface. As such, the desorption spray plume impacted the tissue surface outside the area of interest or at an already analyzed area preventing contamination of yet unanalyzed tissue regions. At the same time, the 2 mm gap between the sampling capillary and the surface eliminated ion signal as well. When the beginning of the first lane was reached, the surface was moved again parallel to the y-axis with the calculated lane spacing distance followed by raising the surface 2 mm so the sampling capillary touched the tissue again. The following lanes were scanned similar to the first lane. At scan rates below 2 mm/s, DESI-MS/MS data of all lane scans were acquired into a single data file. At scan rates of 2 mm/s and higher, DESI-MS/MS data of individual lane scans were stored in separate data files. Chemical images were created from these files by using a custom image analysis software (HandsFree Surface Analysis). Safety Considerations: The DESI emitter floats at the high ES voltage, and appropriate shields and interlocks should be used to avoid accidental contact with this component. Furthermore, extra precautions should be taken if DESI is used with tissues containing radioactive compounds. RESULTS AND DISCUSSION Evaluation of Signal Level Dependence on Heated Chamber Length of the PDI. The proper positioning of the surface to be analyzed with respect to the atmospheric sampling orifice/ capillary into the mass spectrometer is crucial for optimum DESI-MS/MS performance.34 When the surface is comprised of a large planar structure, it can be mechanically challenging to position the surface sufficiently close to a mass spectrometer inlet. For instruments equipped with a heated capillary atmospheric sampling inlet, proper positioning of the surface can be achieved by adding an extension tube to give the desired protrusion of the sampling inlet.14 A PDI can also provide a protruding inlet geometry for atmospheric sampling interfaces that normally use a simple orifice and curtain gas.29 The PDI exploits different mobilities and momentum characteristics of ions and large charged droplets and neutral particles for ESIMS. The original purpose of the device was to transmit the ions created at atmospheric pressure into the mass analyzer with high efficiency and deflect large charged and neutral species. Later it was shown that extending the heated chamber of the PDI provided an appropriate sampling geometry for a Analytical Chemistry, Vol. 80, No. 13, July 1, 2008

5171

MALDI,30 a chip-based nanospray,31 and a homemade DESI ion source.32 For this work, we modified the standard nanospray interface of a 4000 QTRAP mass spectrometer, already a PDI (1 cm heated chamber, Figure 1a), to use heated chambers of significantly extended length (Figure 1b and Supporting Information Figure S2). Performance was evaluated by measuring the continuous infusion ES-MS/MS signal (m/z 260 f 116) from a 0.1 µm propranolol solution at solvent flow rates of 2.5, 5.0, and 10 µL/ min (Figure 1d) with each of the modified configurations. For this test, an on-axis spray position, about 1 cm from the sampling aperture of the respective heated chamber, was used. The highest signal levels were achieved with a 2.0 cm long heated chamber at all flow rates, and signal decreased with increasing flow rate for each chamber length. Improvement in performance of the 2.0 cm long heated chamber over the normal nanospray PDI was probably related to improved desolvation. The normal PDI, with a 1.0 cm long heated chamber, was designed to operate with nanoliter per minute solvent flow rates rather than the much higher flows used here. Of course, the efficiency of desolvation with any heated chamber length was expected to decrease with increasing flow rate. The other observation was that the signal levels with heated chamber lengths of 7.0 and 12 cm were significantly attenuated compared to the shorter chambers as expected. It has been shown previously that diffusive losses can become very significant when residence times within the heated laminar flow chamber become long.29 Although we did not perform precise quantitative tests, we did observe a similar diminution of analyte signal in the DESI experiments with increasing heated chamber length. Therefore, for the imaging discussed here, we used the 7.0 cm heated chamber as this extension was the shortest chamber that allowed us to image a whole-body mouse thin tissue section in our current instrument configuration. DESI-MS/MS Imaging of Whole-Body Mouse Thin Tissue Sections. Figure 2a shows the scanned optical image of a wholebody mouse thin tissue section. The mouse was dosed with 7.5 mg/kg propranolol and euthanized after 20 min. Figure 2b is a spatial distribution plot of the SRM ion current for propranolol (m/z 260 f 116, 100 ms dwell time) obtained from the same section using DESI-MS/MS. This chemical image was constructed from 41 consecutive unidirectional lane scans at 100 µm/s with 500 µm spacing in three separate 20 mm × 20 mm regions of the thin tissue section. Total analysis time was 7.5 h. On the basis of the predicted metabolic pathway of propanolol,35 seven additional transitions were monitored in the same experiment: m/z 260 f 183 (propranolol), m/z 276 f 116 (hydroxypropranolol), m/z 292 f 116 (dihydroxypropranolol), m/z 436 f 116 (propranolol glucuronide), m/z 452 f 116 (hydroxypropranolol glucuronide), m/z 468 f 116 (dihydroxypropranolol glucuronide), and m/z 482 f 116 (propranolol glucuronic acid). Because actual standards were not available for optimization, the same ion source and collision cell conditions were used as for the parent drug. Out of these seven additional transitions, only the confirmatory transition

for propranolol (m/z 260 f 183, Scheme 1) provided signal (approximately 50% of that of the m/z 260 f 116 transition) above background levels. The inability to detect hydroxypropranolol glucuronide was unexpected, because this compound was identified as the major metabolite in lung, kidney, and liver extracts using LC-MS/ MS (see the Experimental Section). Furthermore, another surface sampling approach used in our laboratory, namely, the liquid microjunction surface sampling probe (LMJ-SSP)/electrospray ionization mass spectrometry system,36 clearly detected the parent drug (m/z 260 f 116 and m/z 260 f 183) and the hydroxypropranolol glucuronide metabolite (m/z 452 f 116) from these organs under the same instrumental conditions (see Supporting Information Figure S3). Importantly, we used the LMJ-SSP/ES-MS/MS system to obtain the product ion spectrum of the hydroxypropranolol glucuronide metabolite extracted from a liver tissue and to optimize its detection by SRM on our instrument (Supporting Information Figure S3). Under our experimental conditions, the product ion spectrum base peak was observed at m/z 276 (neutral loss of 176 Da, characteristic of glucuronides35), which had an abundance about 10 times that of the product ion we had been monitoring at m/z 116. Supporting Information Figure S3 shows the SRM signal levels for the drug (m/z 260 f 116) and the metabolite (m/z 452 f 116 and the new transition m/z 452 f 276) obtained with the LMJ-SSP/ES-MS/MS system from different positions on tissue from the dosed animal. Monitoring these transitions using the sampling probe system when examining the tissue of a control mouse did not produce signal above background levels proving their detection selectivity. Lung, liver, and kidney regions of the tissue from the dosed animal were analyzed again by DESI-MS/MS using this new transition for the hydroxypropranolol glucuronide metabolite (m/z 452 f 276), but no signal above background level was observed in that case either. For comparison to Figure 2, parts a and b, Figure 2c shows the scanned optical image of a whole-body tissue section of a mouse dosed intravenously with 7.5 mg/kg [3H]propranolol and euthanized 20 min after dose and its corresponding WBA image (Figure 2d). Simple visual comparison of the DESI-MS/MS image (Figure 2b) and the WBA image (Figure 2d) reveals that both methods confidently detected propranolol (and/or its metabolites in the case of the autoradioluminograph) in the brain, lung, stomach, and kidney regions. Supporting Information Figure S5 shows propranolol tissue concentrations determined by WBA in conjunction with radioprofiling. DESI-MS/MS images obtained using higher surface scan rates than used to acquire the image in Figure 2b reduced the analysis time but also somewhat compromised the signal levels for the drug and presented other issues that affected image quality. Imaging a tissue section at a 2 mm/s surface scan rate versus 100 µm/s resulted in a blurred propranolol chemical image (not shown) at the edges of organs (e.g., brain, lung). The distorted image was the result of incorrect spatial assign-

(35) Beaudry, F.; Le Blanc, J. C. Y.; Coutu, M.; Ramier, I.; Moreau, J-P.; Brown, N. K. Biomed. Chromatogr. 1999, 13, 363–369.

(36) Van Berkel, G. J.; Kertesz, V.; Koeplinger, K. A.; Vavrek, M.; Kong, A.-N., T. J. Mass Spectrom. 2008, 43, 500–508.

5172


Figure 2. (a) Scanned optical image of a 40 µm thick sagittal whole-body tissue section of a mouse dosed intravenously with 7.5 mg/kg propranolol and euthanized 20 min after dose. (b) Distribution of propranolol in 20 mm × 20 mm and 38 mm × 20 mm areas measured by DESI-MS/MS (SRM: m/z 260 f 116) using 80/20 (v/v) methanol/water as DESI solvent at a flow rate of 5 µL/min. Surface scan rate was 0.1 mm/s, dwell time was 100 ms, and the images were created from 41 lanes with 500 µm spacing. Pixel size was 84 µm (h) × 500 µm (v), and experiment times were 150 and 285 min for the 20 mm × 20 mm and 38 mm × 20 mm areas, respectively. (c) Scanned optical image of a 40 µm thick sagittal whole-body tissue section of a mouse dosed intravenously with 7.5 mg/kg [3H]propranolol and euthanized 20 min after dose. (d) Autoradioluminograph of [3H]propranolol-related material in the tissue section presented in (c).

ment of the mass spectrometric data scans. The location coordinates of the area being sampled were collected with an approximately 60 ms uncertainty due to the communication speed between the computer and the robotic stage. This uncertainty translated into the possibility of approximately 6, 30, and 120 µm spatial assignment error at 100, 500 µm/s and 2 mm/s surface scan rates, respectively. For this reason, another mass spectrometric data-to-location assignment approach was developed to significantly decrease spatial assign-

ment error. With this method each lane scan was stored as a separate mass spectral data file, and accurate alignment was achieved by using a propranolol containing calibration line pressed onto the slide holding the tissue section (see the Supporting Information for details). With this approach, as illustrated by the data in Figure 3, DESI-MS/MS imaging of propranolol in the tissue sections was accomplished at a surface scan rate of 7 mm/s, the maximum speed of the stage. At this speed, total acquisition time was 79 min for the whole-body Analytical Chemistry, Vol. 80, No. 13, July 1, 2008

5173

Figure 3. (a) Scanned optical image of a 40 µm thick sagittal whole-body tissue section of a mouse dosed intravenously with 7.5 mg/kg propranolol and euthanized 60 min after dose. (b) Distribution of propranolol in the 94 mm × 30 mm tissue section presented in (a) measured by DESI-MS/MS (SRM: m/z 260 f 116) using 80/20 (v/v) methanol/water as DESI solvent at a flow rate of 5 µL/min. Surface scan rate was 7 mm/s, dwell time was 20 ms, and the image was created from 151 lanes with 200 µm spacing. Pixel size was 140 µm (h) × 200 µm (v), and total experiment time was 79 min. (c) Scanned optical image of a 40 µm thick sagittal whole-body tissue section of a mouse dosed intravenously with 7.5 mg/kg [3H]propranolol and euthanized 60 min after dose. (d) Autoradioluminograph of [3H]propranolol-related material in the tissue section presented in (c).

mouse tissue (94 mm × 30 mm) using 151 lane scans with a 200 µm lane spacing. With the use of a 20 ms dwell time to monitor one SRM transition (m/z 260 f 116), one pixel in the image corresponded to an area of approximately 140 µm × 200 µm in size. 5174


Figure 3a is the optical image of the tissue section, and Figure 3b is the corresponding DESI-MS/MS image of propranolol for a mouse euthanized 60 min post dose using this high surface scan rate. Figure 3, parts c and d, shows a scanned optical image and corresponding WBA image, respectively, of a whole-body tissue

Figure 4. DESI-MS/MS signal of propranolol (SRM: m/z 260 f 116) at 100 and 500 µm/s and 2 and 7 mm/s surface scan rates used for analysis of a lung tissue of a mouse dosed intravenously with 7.5 mg/kg propranolol and euthanized 60 min after dose.

thin section of a different mouse dosed with [3H]propranolol and euthanized 60 min post dose. Visual inspection of the DESI-MS/ MS image (Figure 3b) and the WBA image (Figure 3d) indicates that propranolol and/or its metabolite (in the case of WBA) were detected in the brain, lung, and stomach regions by both techniques. In addition, the autoradioluminograph showed the drug and/or its metabolites in the salivary gland, kidney, and liver as well, whereas propranolol was detected with less confidence in these organs using DESI-MS/MS. Although Figure 3b demonstrates that useful information concerning drug distribution in the whole-body section was obtained at these fast surface scan rates, a drop in DESI-MS/MS signal level with increasing surface scan rate diminished the quality of the chemical image. Figure 4 shows relative DESI-MS/ MS signal levels of propranolol as a function of surface scan rates for the same lung tissue. The observed trend was in agreement with our previous study,37 where signal level was shown to decrease with increasing surface scan rate for analytes deposited on retaining/porous thin-layer chromatography (TLC) plates. Tissue and TLC phases both have three-dimensional structures that encapsulate the analytes. The droplet pickup desorption/ ionization mechanism of DESI38 requires the analyte to be first dissolved in the DESI spray solution at the surface then transferred to the gas phase upon sputtering by subsequent droplet and gas impact at the surface. This process will require some finite time to take place efficiently. Therefore, the lower signal levels with higher surface scan rate were most likely due to decreased effectiveness of the solid-liquid extraction step inherent to DESI. Quantitative Comparison of Propranolol Levels in Tissues using WBA and DESI-MS/MS. Figure 5, parts a and b, shows a comparison between radioactivity and DESI-MS/MS signal assigned to propranolol for kidney, lung, brain, and liver that were prepared from mice euthanized 20 and 60 min after drug administration, respectively. Only a relative comparison between the signal levels from the two techniques was possible. Signal levels for all organs were normalized to that of the brain tissue. These data were compiled for specific organs in the chemical (37) Pasilis, S. P.; Kertesz, V.; Van Berkel, G. J. Anal. Chem. 2007, 79, 5956– 5962. (38) Venter, A.; Sojka, P. E.; Cooks, R. G. Anal. Chem. 2006, 78, 8549–8555.

Figure 5. Relative signal of propranolol calculated from (filled bars) WBA and radioprofiling results and (empty bars) measured by DESIMS/MS (SRM: m/z 260 f 116) for kidney, lung, brain, and liver tissues of mice euthanized (a) 20 min (scanned at 100 µm/s, Figure 2b) and (b) 60 min (scanned at 7 mm/s, Figure 3b) after drug administration. Signals were normalized in all cases to the propranolol signal detected in the brain.

images shown in Figure 2, parts b and d, and Figure 3, parts b and d. We excluded quantitative examination of the stomach region from the study, because uneven levels of radioactive material were observed in the stomach after intravenous dosing. Radioactivity assigned to the parent drug was calculated using the results from WBA and radioprofiling HPLC experiments (see the Supporting Information for details). In general, there was a good correlation between the normalized radioactivity and DESI-MS/MS signals for the lung and liver, for both time points (20 and 60 min). In the case of the kidney, there was a significant difference between results provided by the two techniques. The exact cause of this observation was not clear. However, variations between the anatomical features of the kidney represented (e.g., renal medulla, renal cortex) for the tissue sections analyzed by DESI-MS/MS and WBA, as well as variability in rates of metabolism and excretion between animals, may account for at least some differences observed. In addition, it might be expected that desorption (extraction)/ionization efficiencies using DESI would differ among tissue types. Although this phenomenon would not explain the observed disagreement between the two techniques in the case of the kidney, it would affect DESI-MS/MS signal for different tissue types with equal drug concentration. These same matrix issues must be dealt with when using other mass spectrometry imaging techniques.5,9 Analytical Chemistry, Vol. 80, No. 13, July 1, 2008

5175

Prognosis for Imaging Drugs and Metabolites in Thin Tissue Sections with DESI-MS/MS. The data presented above indicate that effective and relatively rapid imaging of a parent drug in a whole-body thin tissue section, when dosed at a pharmacologically relevant level, is possible using DESI-MS/MS. However, the low signal level for the drug and the inability to observe the major metabolite are issues that must be addressed. On the basis of calculated propranolol concentration from WBA data (21.2 nmol/g tissue), tissue thickness (40 µm), and estimated volumetric density (1 g/mL) we estimate that in the kidney region in Figure 2, parts b and d, there was about 850 pmol propranolol/ mm2. This surface area concentration is significantly higher than the low picomole per square millimeter range for detection levels that we and others have reported for analytes with similar ionization efficiency that were simply deposited on surfaces and analyzed by DESI-MS.39,40 However, the maximum signal for propranolol in this region was only about 850 counts per second (cps). In this same region, the hydroxypropranolol glucuronide metabolite was estimated to be present at about the same concentration from radioprofiling and LC-MS/MS data (not shown), but no signal above background (about 50 cps) was observed for this analyte. The low signal levels could be due to one or more of the following phenomena: ion suppression, inefficient extraction of analyte from the matrix, inefficient desorption/ionization, and inefficient droplet/ion collection and transport into the mass spectrometer. Other experiments indicate that inefficient extraction maybe be one of the most important factors leading to the low signal levels. The use of an LMJ-SSP/ES-MS/MS system to analyze the same tissues accomplished the detection of propranolol and the major metabolite (Supporting Information Figure S3). Furthermore, similar signal levels were observed with the sampling probe system in a particular organ before and after analysis by DESI-MS/MS. These observations support the contention that only a small portion of the targeted compounds were removed from the tissue by the DESI process. As Figure 1d shows, signal level with the extended heated chamber (7.0 cm) PDI was attenuated by as much as 50% compared to the standard nanospray interface. Although this performance might be improved, we do not feel this signal loss had as large an influence on the DESI signals as the extraction issue. As an initial test of this assumption, we examined the same tissues using our DESI source on a Thermo Finnigan LTQ FT Ultra mass spectrometer. The LTQ uses the heated capillary type interface that is typically used for DESI-MS. Parent drug was detected only when using the pseudo-SRM scan function of the LTQ, and no metabolites were detected. A direct comparison of the PDI and the heated capillary interface performances on the same mass spectrometer is planned. The speed of analysis would appear to be ultimately limited by fundamental factors governing the time frame of the desorption process (dissolution/extraction/desorption). Even if we assume that our highest scan rate of 7 mm/s might be an upper limit for scan speed, there are still at least two experimental means to reduce overall imaging time significantly. One scenario is to (39) Van Berkel, G. J.; Tomkins, B. A.; Kertesz, V. Anal. Chem. 2007, 79, 2778– 2789. (40) Kauppila, T. J.; Talaty, N.; Salo, P. K.; Kotiaho, T.; Kostiainen, R.; Cooks, R. G. Rapid Commun. Mass Spectrom. 2006, 20, 2143–2150.

5176


eliminate the return scans by using raster scanning instead of unidirectional scanning. This effectively cuts in half the total imaging time for any area imaged. However, to provide the same quality image with raster scanning as with unidirectional scanning will require the ability to precisely align the spray capillary in the DESI sprayer assembly.34 The other possibility relates to the capabilities of the sample handling stage so that the return scans in unidirectional scanning are much faster. Increasing the maximum stage speed to 14 mm/s (a lead screw pitch exchange commercially available) would allow analysis of the mouse wholebody tissue section in Figure 3a in approximately 1 h with 200 µm lane spacing. If lane spacing was increased to 400 µm, which corresponds to the vertical resolution commonly reported with MALDI-MS,9 a chemical image of a whole-body tissue section of this size would be obtained in approximately 30 min using DESIMS/MS. CONCLUSIONS In this paper, we compared DESI-MS/MS and WBA for chemical imaging of whole-body thin tissue sections of mice intravenously dosed with pharmacologically relevant levels of the drug propranolol. DESI-MS/MS imaging utilized SRM detection performed on a triple quadrupole mass spectrometer equipped with an extended length heated chamber PDI to permit sampling from large planar surfaces. The PDI was able to efficiently sample ions from locations remote to the gas conductance limiting orifice; however, the observed signal levels decreased with increasing heated chamber length. Thus, the shortest length PDI practical for the current analysis configuration was used. Image quality was reduced as surface scan rate increased, due to reduced signal levels and spatial assignment errors. Nonetheless, chemically informative DESIMS/MS images of the parent drug were obtained at surface scan speeds up to 7 mm/s, the top speed for the stage used. That translated into an imaging time of just 79 min for a wholebody section of a mouse. We estimate that with simple changes in the surface scan scheme or use of a stage with faster movement capabilities, images of the same quality could be obtained in as little as 30 min. Nominal agreement for the relative distributions of propranolol in the brain, lung, and liver was shown between the DESI-MS/MS signal for propranolol and the radioactivity attributed to propranolol from WBA sections. In the case of kidney, the data exhibited an unexplained disparity between the two techniques. Unfortunately current performance characteristics of this imaging system were not sufficient to detect hydroxypropranolol glucuronide, the major metabolite known to be present in kidney and liver tissues at levels close to that of the parent drug. These discrepancies, low signal levels for the parent drug, and the inability to detect known metabolites present in the tissue will need to be addressed and overcome to make thin tissue section imaging by DESI-MS a practical drug discovery tool. ACKNOWLEDGMENT The PDI components as well as the microionspray II used to fabricate the DESI emitter were provided to ORNL through a Cooperative Research and Development Agreement with MDS Analytical Technologies (CRADA ORNL02-0662). Metabolite

characterization by nanospray MS was supported by Dr. Richard King, and the in-life portion of this study was supported by Anne Taylor and Scott Fauty, Merck Research Laboratories Drug Metabolism and Pharmacokinetics, West Point, PA. Dr. Richard King is thanked for the use of the Thermo Finnigan LTQ FT Ultra mass spectrometer. Study of the fundamentals of high-speed imaging by DESI-MS/MS was supported by the Division of Chemical Sciences, Geosciences, and Biosciences, Office of Basic Energy Sciences, United States Department of Energy. ORNL Technology Transfer and Economic Development Royalty Funds provided support for the development, modification, and application of the surface-scanning control

and data-processing software, as well as the use of the LMJSSP system for tissue sampling. ORNL is managed and operated by UT-Battelle, LLC, for the United States Department of Energy under Contract DE-AC05-00OR22725. SUPPORTING INFORMATION AVAILABLE Additional information as noted in text. This material is available free of charge via the Internet at http://pubs.acs.org. Received for review March 14, 2008. Accepted April 16, 2008. AC800546A


5177

Comparison of Drug Distribution Images from Whole-Body Thin

Recommend Documents