Automated Ambient Desorption− Ionization Platform for Surface

Sep 17, 2009 - Pharmacy, University of Helsinki, P.O. Box 56, FI-00014, University of Helsinki, ... A fully automated atmospheric pressure ionization ...
0 downloads 0 Views 5MB Size
Download PDF
Anal. Chem. 2009, 81, 8479–8487

Automated Ambient Desorption-Ionization Platform for Surface Imaging Integrated with a Commercial Fourier Transform Ion Cyclotron Resonance Mass Spectrometer Jaroslav Po´l,†,‡ Veronika Vidova´,†,§ Gary Kruppa,† Va´clav Kobliha,† Petr Nova´k,† Karel Lemr,†,§ Tapio Kotiaho,‡,| Risto Kostiainen,‡ Vladimı´r Havlı´cˇek,†,§ and Michael Volny´*,† Laboratory of Molecular Structure Characterization, Institute of Microbiology, Academy of Sciences of the Czech Republic, Vı´denˇska´ 1083, CZ-142 20 Prague, Czech Republic, Division of Pharmaceutical Chemistry, Faculty of Pharmacy, University of Helsinki, P.O. Box 56, FI-00014, University of Helsinki, Finland, Department of Analytical Chemistry, Faculty of Science, Palacky´ University, Tr´. Svobody 8, CZ-771 46 Olomouc, Czech Republic, and Laboratory of Analytical Chemistry, Department of Chemistry, University of Helsinki, P.O. Box 55, FI-00014, University of Helsinki, Finland A fully automated atmospheric pressure ionization platform has been built and coupled with a commercial highresolution Fourier transform ion cyclotron resonance mass spectrometer (FTICR-MS) instrument. The outstanding performance of this instrument allowed screening on the basis of exact masses in imaging mode. The main novel aspect was in the integration of the atmospheric pressure ionization imaging into the current software for matrix-assisted laser desorption ionization (MALDI) imaging, which allows the user of this commercial dual-source mass spectrometer to perform MALDI-MS and different ambient MS imaging from the same user interface and to utilize the same software tools. Desorption electrospray ionization (DESI) and desorption atmospheric pressure photoionization (DAPPI) were chosen to test the ambient surface imaging capabilities of this new ionization platform. Results of DESI imaging experiments performed on brain tissue sections are in agreement with previous MS imaging reports obtained by DESI imaging, but due to the high resolution and mass accuracy of the FTICR instrument it was possible to resolve several ions at the same nominal mass in the DESI-MS spectra of brain tissue. These isobaric interferences at low resolution are due to the overlap of ions from different lipid classes with different biological relevance. It was demonstrated that with the use of high-resolution MS fast imaging screening of lipids can be achieved without any preseparation steps. DAPPI, which is a relatively new and less developed ambient ionization technique compared to DESI, was used in imaging mode for the first time ever. It showed promise in imaging of phytocompounds from plant leaves, and selective ionization of a sterol lipid was achieved by DAPPI from a brain tissue sample. Atmospheric pressure desorption ionization mass spectrometry techniques, often referred to as ambient mass spectrometry,1 allow * To whom correspondence should be addressed. E-mail: [email protected]. Phone: (420) 296-442-796. † Academy of Sciences of the Czech Republic. ‡ Faculty of Pharmacy, University of Helsinki. § Palacky´ University. | Department of Chemistry, University of Helsinki. 10.1021/ac901368q CCC: $40.75  2009 American Chemical Society Published on Web 09/17/2009

for the direct analysis of surfaces in the open atmosphere. The introduction of the desorption electrospray ionization (DESI) in late 20042,3 ignited the interest in ambient surface sampling in the mass spectrometry community, and since DESI many other techniques for the direct analysis of surfaces by mass spectrometry have been introduced. Each of them has been assigned a new acronym, but often there is significant overlap between these techniques4 because on the more fundamental level there are only a limited number of available desorption and ionization processes.5 The future of this new but rapidly emerging field of mass spectrometry is still debated, and thorough reviews were recently published.4,5 A logical extension of the atmospheric pressure surface sampling mass spectrometry techniques is their utilization for twodimensional (2D) imaging analysis. With the use of imaging MS a 2D distribution can be created from the individual mass spectra, and an image of a particular compound on a surface can be obtained.6 Mass spectrometry imaging is theoretically possible with any surface desorption ionization (DI) technique, although the quality and practicality of the obtained data will differ between individual DI-MS techniques. Desorption and ionization from surfaces under vacuum is well-established in mass spectrometry, and currently, two desorption-ionization techniques dominate the field of mass spectrometry imaging: secondary ion mass spectrometry (SIMS)7-9 and matrix-assisted laser desorption ionization (1) Cooks, R. G.; Ouyang, Z.; Takats, Z.; Wiseman, J. M. Science 2006, 311, 1566–1570. (2) Takats, Z.; Wiseman, J. M.; Cooks, R. G. J. Mass Spectrom. 2005, 40, 1261– 1275. (3) Takats, Z.; Wiseman, J. M.; Gologan, B.; Cooks, R. G. Science 2004, 306, 471–473. (4) Van Berkel, G. J.; Pasilis, S. P.; Ovchinnikova, O. J. Mass Spectrom. 2008, 43, 1161–1180. (5) Venter, A.; Nefliu, M.; Cooks, R. G. TrAC, Trends Anal. Chem. 2008, 27, 284–290. (6) McDonnell, L. A.; Heeren, R. M. A. Mass Spectrom. Rev. 2007, 26, 606– 643. (7) Aoyagi, S. Surf. Interface Anal. 2009, 41, 136–142. (8) Debois, D.; Bralet, M. P.; Le Naour, F.; Brunelle, A.; Laprevote, O. Anal. Chem. 2009, 81, 2823–2831.

Analytical Chemistry, Vol. 81, No. 20, October 15, 2009

8479

(MALDI).10-13 They are supplemented by laser ablation inductively coupled plasma mass spectrometry (LA-ICPMS)14,15 for elemental imaging. Each of these techniques has a different area of application, and all are commercially available. A new addition to mass spectrometry imaging is recent utilization of DESI in imaging mode.16-22 Here we present a modular platform for imaging by different atmospheric pressure surface mass spectrometry techniques. This platform mounts on the atmospheric pressure inlet of a dual-source high-resolution commercial mass spectrometer. Only a very few reports exist about the DESI coupling with Fourier transform ion cyclotron resonance mass spectrometer (FTICR-MS),23,24 and we are not aware of any previously reported surface imaging experiments in this configuration. Due to a custom modification of the software, the platform can be controlled through the regular user interface and the data analyzed by a standard MALDI imaging software, including statistical and other program tools. We selected two different atmospheric pressure desorption/ionization techniques to demonstrate the capabilities of our setup. The first one is DESI,3 which is used mainly to analyze small and medium size polar molecules, and it has been used to address a variety of analytical problems.25-30 DESI imaging overcomes some of the disadvantages of the vacuum imaging MS techniques although in other respects it too has limitations. DESI, which is conducted outside the mass spectrometer at the atmospheric environment, offers mainly the advantage of little or no sample (9) Takahashi, L. K.; Zhou, J.; Wilson, K. R.; Leone, S. R.; Ahmed, M. J. Phys. Chem. A 2009, 113, 4035–4044. (10) Dekker, L. J. M.; van Kampen, J. J. A.; Reedijk, M. L.; Burgers, P. C.; Gruters, R. A.; Osterhaus, A.; Luider, T. M. Rapid Commun. Mass Spectrom. 2009, 23, 1183–1188. (11) Mikawa, S.; Suzuki, M.; Fujimoto, C.; Sato, K. Neurosci. Lett. 2009, 451, 45–49. (12) Murphy, R. C.; Hankin, J. A.; Barkley, R. M. J. Lipid Res. 2009, 50, S317– S322. (13) Pevsner, P. H.; Melamed, J.; Remsen, T.; Kogos, A.; Francois, F.; Kessler, P.; Stern, A.; Anand, S. Biomarkers Med. 2009, 3, 55–69. (14) Becker, J. S.; Dietrich, R. C.; Matusch, A.; Pozebon, D.; Dressier, V. L. Spectrochim. Acta, Part B 2008, 63, 1248–1252. (15) Zoriy, M. V.; Mayer, D.; Becker, J. S. J. Am. Soc. Mass Spectrom. 2009, 20, 883–890. (16) Ifa, D. R.; Gumaelius, L. M.; Eberlin, L. S.; Manicke, N. E.; Cooks, R. G. Analyst 2007, 132, 461–467. (17) Ifa, D. R.; Manicke, N. E.; Dill, A. L.; Cooks, G. Science 2008, 321, 805– 805. (18) Ifa, D. R.; Wiseman, J. M.; Song, Q. Y.; Cooks, R. G. Int. J. Mass Spectrom. 2007, 259, 8–15. (19) Kertesz, V.; Van Berkel, G. J. Rapid Commun. Mass Spectrom. 2008, 22, 2639–2644. (20) Kertesz, V.; van Berkel, G. J. Anal. Chem. 2008, 80, 1027–1032. (21) Kertesz, V.; Van Berkel, G. J.; Vavrek, M.; Koeplinger, K. A.; Schneider, B. B.; Covey, T. R. Anal. Chem. 2008, 80, 5168–5177. (22) Wiseman, J. M.; Ifa, D. R.; Venter, A.; Cooks, R. G. Nat.Protoc. 2008, 3, 517–524. (23) Wu, S.; Zhang, K.; Kaiser, N. K.; Bruce, J. E. J. Am. Soc. Mass Spectrom. 2006, 17, 772–779. (24) Bereman, M. S.; Nyadong, L.; Fernandez, F. M.; Muddiman, D. C. Rapid Commun. Mass Spectrom. 2006, 20, 3409–3411. (25) Bereman, M. S.; Muddiman, D. C. J. Am. Soc. Mass Spectrom. 2007, 18, 1093–1096. (26) Pasilis, S. P.; Kertesz, V.; Van Berkel, G. J. Anal. Chem. 2007, 79, 5956– 5962. (27) Van Berkel, G. J.; Ford, M. J.; Deibel, M. A. Anal. Chem. 2005, 77, 1207– 1215. (28) Van Berkel, G. J.; Kertesz, V. Anal. Chem. 2006, 78, 4938–4944. (29) Wiseman, J. M.; Puolitaival, S. M.; Takats, Z.; Cooks, R. G.; Caprioli, R. M. Angew. Chem., Int. Ed. 2005, 44, 7094–7097. (30) Zhao, M. X.; Zhang, S. C.; Yang, C. D.; Xu, Y. C.; Wen, Y. X.; Sun, L. S.; Zhang, X. R. J. Forensic Sci. 2008, 53, 807–811.

8480

Analytical Chemistry, Vol. 81, No. 20, October 15, 2009

preparation. In DESI, pneumatically assisted electrospray produces charged droplets that are directed at a surface. As they collide with the surface they wet it, and analyte is extracted into the liquid film (this is one of the factors that influence spatial resolution in DESI imaging experiments); impact of subsequent primary droplets releases secondary droplets that reach the inlet of mass spectrometer (“droplet pickup” mechanism). Ions are formed from the secondary droplets by an electrospray mechanism. The detailed mechanism of DESI has been investigated by simulations and experiments.2,31-34 DESI imaging has been used mainly to obtain chemical images of biomedically important surfaces,35,36 although other applications were published too.16,17 The second technique, desorption atmospheric pressure photoionization (DAPPI), was first reported in 200737 and has been applied in analysis of drugs and forensic samples.38-40 DAPPI is a surface analogy of the atmospheric pressure photoionization (APPI) and uses a microchip-heated nebulizer to deliver a solvent vapor jet at the sample surface, followed by thermal desorption and ionization in the gas phase from photoinitiated ion-molecule reactions using a UV lamp. In a certain sense DAPPI is similar to different corona and plasma ionization techniques except that the reagent ion population is initiated by a photoionization process directly or through a dopant rather than by a corona/plasma discharge. In the Haapala et al.37 DAPPI configuration, which is used here as well, rapid thermal desorption is the dominant surface sampling process. The ionization process is likely similar as that in APPI used with liquid introduction ion sources. The desorption-ionization process of DAPPI has been studied by Luosuja¨rvi et al.41 The choice of solvent dopant vapor dramatically affects the ionization of analyte, and DAPPI, which is still an unexplored desorption-ionization technique, is considered to be complementary to DESI in terms of detectable analytes (DAPPI is especially useful for nonpolar aromatic compounds). This is the first report on DAPPI imaging and one of only very a few examples of ambient MS imaging other than DESI. EXPERIMENTAL SECTION Mass Spectrometer. An atmospheric pressure ion source platform constructed in our laboratory was mounted on a dualsource (MALDI/ESI) high-resolution mass spectrometer APEX (31) Costa, A. B.; Cooks, R. G. Chem. Phys. Lett. 2008, 464, 1–8. (32) Venter, A.; Sojka, P. E.; Cooks, R. G. Anal. Chem. 2006, 78, 8549–8555. (33) Volny, M.; Venter, A.; Smith, S. A.; Pazzi, M.; Cooks, R. G. Analyst 2008, 133, 525–531. (34) Benassi, M.; Wu, C. P.; Nefliu, M.; Ifa, D. R.; Volny, M.; Cooks, R. G. Int. J. Mass Spectrom. 2009, 280, 235–240. (35) Wiseman, J. M.; Ifa, D. R.; Song, Q. Y.; Cooks, R. G. Angew. Chem., Int. Ed. 2006, 45, 7188–7192. (36) Dill, A. L.; Ifa, D. R.; Manicke, N. I.; Ouyang, Z.; Cooks, R. G. J. Chromatogr., B 2009, 877, 2882–2889. (37) Haapala, M.; Pol, J.; Saarela, V.; Arvola, V.; Kotiaho, T.; Ketola, R. A.; Franssila, S.; Kauppila, T. J.; Kostiainen, R. Anal. Chem. 2007, 79, 7867– 7872. (38) Luosujarvi, L.; Arvola, V.; Haapala, M.; Pol, J.; Saarela, V.; Franssila, S.; Kostiainen, R.; Kotiaho, T.; Kauppila, T. J. Eur. J. Pharm. Sci. 2008, 34, S29–S29. (39) Luosujarvi, L.; Laakkonen, U. M.; Kostiainen, R.; Kotiaho, T.; Kauppila, T. J. Rapid Commun. Mass Spectrom. 2009, 23, 1401–1404. (40) Kauppila, T. J.; Arvola, V.; Haapala, M.; Pol, J.; Aalberg, L.; Saarela, V.; Franssila, S.; Kotiaho, T.; Kostiainen, R. Rapid Commun. Mass Spectrom. 2008, 22, 979–985. (41) Luosujarvi, L.; Arvola, V.; Haapala, M.; Pol, J.; Saarela, V.; Franssila, S.; Kotiaho, T.; Kostiainen, R.; Kauppila, T. J. Anal. Chem. 2008, 80, 7460– 7466.

Figure 1. Coupling of the atmospheric ionization platform to FTICR-MS instrument. (A) Detail of the ionization platform. (B) The ionization platform mounted on the FTICR-MS instrument. (C) Schema shows how the pulse signals are transmitted from software layers to the stepper motors of the ionization stage. Technical details about the ionization platform design can be found in ref 42.

Ultra 9.4T FT-MS (Bruker Daltonics, Germany and U.S.A.). The platform mounts on the atmospheric pressure inlet instead of the commercial nebulizer. The instrument was operated in positive API mode with API voltages on (to allow entry of ions from the inlet side), but the MALDI functions were also enabled to utilize imaging tools. The acquisition data set size was set to 512K points with the mass range of m/z 250-1800 (m/z 100-650 in DAPPI). Spectra averaging mode, which consisted of three scans, was used. The instrument was externally calibrated using arginine clusters resulting in mass accuracy below 2 ppm over the m/z range reported above. After the analysis the data were processed and interpreted using commercially available softwares DataAnalysis and FlexImaging (Bruker Daltonics). Ion Source Platform. The ion source platform, which was reported previously,42,43 consisted of an aluminum frame holding two independently controlled stages. The first one was for mounting the sample, and its position was adjustable in three dimensions by three stepper electromotors (allowing positioning with micrometer accuracy). The second stage controlled the sprayer in three axes and its angle also by four stepper electromotors. The whole device was connected to a lab-made control console with two joysticks (the first one controlling the sample stage, the second one controlling the sprayer) and display for the position of each axis. The device was designed so that it could be mounted on Bruker/Agilent MS ion sources using the same hinge system, and it can accept the Bruker/Agilent electrospray ionization (ESI) sprayer directly (DESI mode) or compatible lab-made interface for the nebulizer microchip (DAPPI mode). In 2D imaging mode the cables that connect the x, y stepper motors of the sample stage with the remote control were disconnected and connected to the custom-made signal decoder instead. This decoder acts as an interface between the Bruker console and the (42) Takats, Z.; Kobliha, V.; Sevcik, K.; Novak, P.; Kruppa, G.; Lemr, K.; Havlicek, V. J. Mass Spectrom. 2008, 43, 196–203. (43) Pol, J.; Novak, P.; Volny, M.; Kruppa, G. H.; Kostiainen, R.; Lemr, K.; Havlicek, V. Eur. J. Mass Spectrom. 2008, 14, 391–399.

stepper motors of the ion source platform (Figure 1 shows pictures of the platform and schema of the setup). Unused channels on the console were used to send electrical pulses generated by the modified pulse program from the console to the decoder. The number of these pulses is based on the operator’s instructions that can be input by the standard user interface thanks to the modification of the default pulse program. These pulses are then converted by the decoder into dc voltages that are understandable for stepper motors. Thus, number of pulses sent by the mass spectrometer to the console is directly converted into the distance. This allows controlling the x and y movement of the sample stage through the default instrument’s software (Figure S1 shows the FlexImaging window with an acquired DESI-MS image of a brain tissue section; see the Supporting Information). The schema of the arrangement is in Figure 1C, and the software details of the implementation can be found below. FlexImaging Software. The APEX Ultra FT-MS instrument used in these experiments includes a MALDI source and FlexImaging software that allows the user to set up MALDI imaging experiments. The same software was used here for ambient imaging. For MALDI-MS imaging, which is the standard usage of the system, a user would define the region of measurement and its shape, raster pattern, and the laser spot size, and the Hystar software then works with the FlexImaging software to control the automated MALDI-MS imaging of the desired region. For the ambient imaging experiments described here, the desired image was always defined to be a rectangle matrix of points (Figure S2, Supporting Information). The raster pattern was defined to be linear across the sample, with a move in the y-direction after each line across the sample was complete, followed by movement along a line in the opposite x-direction. The scanned image always had to be a rectangle, although the user could define its size, number of acquired spots, and their distance (which defines spatial resolution). Although the MALDI target did move during the ambient imaging experiments, the MALDI laser was switched off Analytical Chemistry, Vol. 81, No. 20, October 15, 2009

8481

so that no MALDI ions were generated. Instead, the ESI source was switched on, and spectra of ions generated by the atmospheric pressure source were acquired. The sample stage target was controlled by modifying the standard pulse program (section below) used for MALDI-MS imaging experiments. Automation through Hystar software, the format of the data, and their handling, were the same as in standard MALDI imaging experiments. Pulse Program Modification and Synchronization of the Surface Stage Movement and FT-MS Data Acquision. A full description of the pulse programming language for the instrument used in these experiments is beyond the scope of this paper, and vendor-specific. For interested readers, the complete modification to the pulse program is included in the Supporting Information for this paper (Figure S3, Supporting Information). Briefly, the pulse programming language included with the Bruker Daltonics APEX Ultra FT-MS has a “C”-like syntax and includes commands that control the timing of all events, including accumulation of ions in the external ion source, transfer of the ions into the FTMS analyzer cell, excitation of the ions into coherent cyclotron motion, and acquisition of the digitized signal from the analyzer cell by the ADC. The pulse programming language also allows for user modifications to the pulse programming language, and there are a number of so-called TTL channels through which the user can control external devices, such as lasers, triggered power supplies, or signal generators etc., in synchronization with the acquisition of FT-MS data. We used three of these TTL pulse channels to control movement of the DESI sample stage in the positive x-direction, y-direction, and negative x-direction, with one TTL pulse corresponding to one step of the stepper motors (Figure 1C). The pulse programming language also includes 32 loop counters whose values can be incremented, decremented, and checked against conditions in conditional statements while a pulse program is running. To control the movement of the DESI stage, we added TTL trigger outputs that were conditional on loop counter values after the acquisition of each FT-MS spectrum. Four loop counters were used: a loop countered defined the number of TTL pulses in the x-direction per pixel in both the positive xand negative x-direction; another loop counter defined the number of TTL pulses per pixel in the y-direction; another loop counter defined the number of pixels per line in the x-direction; a final loop counter was used as a Boolean to determine whether movement should take place in the positive x-direction (0, initialized to this value) or negative x-direction (1). The user set the first three of these loop counters to the desired values in the graphical user interface of the FT-MS data acquisition software. The logic of the pulse program modifications was that after each spectrum was acquired the software output TTL pulses on the positive x-channel causing the stepper motor to moved by the desired number of steps per pixel. This was repeated after each FT-MS spectrum was acquired until the pixel loop counter reached the desired number of pixels per line. The pixel loop counter was then reset to 0, TTL pulses were output on the y-channel causing the stepper motor to move one pixel in the y-direction, and the Boolean controlling the x-direction was reset to its opposite value. This was repeated until the Hystar software controlling the entire process completed the run covering the entire desired image. By 8482

Analytical Chemistry, Vol. 81, No. 20, October 15, 2009

acquiring data in this way, not only could the data acquisition be synchronized with the FT-MS data acquisition, but the imaging data could then be displayed in FlexImaging (Figure S1, Supporting Information) and all its functions can be potentially utilized. Extension of the Inlet Capillary (Flared Inlet Adaptor). An externally heated extension was custom-built and tightmounted on the original inlet capillary. The length of the adaptor was 60 mm, and the internal diameter was 0.5 mm. This inlet adaptor was bent by 90°, and its flared inlet ended approximately 3 mm above the imaged surface, which slid under the adapter. Thus, the desorbed species were taken in from the above by the flared adaptor and immediately transported further into the normal inlet transfer capillary by the means of pressure difference. Although single-point ambient sampling is usually compatible with any atmospheric pressure inlet, 2D scanning requires the above (or similar) modification of the inlet so the surface can freely move under the inlet. Desorption Electrospray Ionization (DESI). In our DESI experiments, a micro-ESI sprayer was used (Agilent, U.S.A.) with nitrogen as a nebulizer gas at 3-5 L/min. High voltage (3.0 kV) delivered by an external high-voltage power supply (Spellman, U.S.A.) was connected to the sprayer, which thus had to be insulated from the rest of the metal frame of the platform. The solvent (50:50 methanol/water) was delivered by a standard syringe pump (Cole-Parmer, U.S.A.) at a speed of 5 µL/min. Standard microscope glass slides were used as substrates for analyzed tissues. Tissue Samples. The collected tissue samples were frozen in liquid nitrogen and stored at -80 °C and subsequently cut in 30 µm sections using a cryomicrotome (CM1950, Leica, Germany). The thin slices were thaw-mounted onto microscope glass and analyzed by DESI-MS. Desorption Atmospheric Pressure (DAPPI). DAPPI arrangement was in principle the same as reported before.37 DAPPI utilized lab-made heated nebulizer microchip that produced confined vapor jet of mixed nebulizer gas and organic solvent. An external power supply was used for heating the microchip; the power was set to 4.2 W, which resulted in jet temperature of 320 °C. The desorption/ionization process was accomplished by the heated plume aiming at a surface that thermally desorbed the analytes followed by photoionization of the desorbed gas-phase analytes with a photoionization lamp (rf photoionization lamp for APPI, Agilent, U.S.A.). The desorbed analytes were then transferred into the heated capillary extension adaptor and further into the mass spectrometer. The position of the microchip with respect to the sample surface was adjusted according to the previous setup;37 the angle was 45°, and the distance was 4 mm. The solvent, in this study toluene or acetone (both Sigma-Aldrich, for trace analysis), acted as a dopant that assisted the ionization and was delivered via standard syringe pump. The nebulizer gas was taken directly from the atmospheric pressure interface in the MS, and the flow rate was 0.2 L/min. Chemicals. All solvents (methanol, water, acetone, toluene) used were obtained from Sigma-Aldrich and were of the HPLC or trace analysis purity. Cholesterol standard was from SigmaAldrich. Tissues and plant leaves were obtained from on campus collaborators.

Figure 2. DESI-MS spectrum of mouse brain tissue section obtained from a single point using an average of three scans. Detail of the m/z 700-900 region with dominant phospholipid peaks. Inset: Overall spectrum of the whole m/z 250-1800 region.

Lipid Nomenclature and Database Search. Lipid search was performed through the online database, www.lipidmaps.org, maintained by the Lipid Metabolites and Pathways Strategy (LIPID MAPS) and Nature Publishing Group (the same database has already been utilized to interpret high-resolution DESI-MS spectra42). The limiting conditions for the database search were set to 0.01 Da for mass accuracy (the tightest available) and above 106 for ion intensity for phospholipids and above 105 for sphingolipids. The nomenclature used was taken from the reference recommended by the database.44,45 RESULTS Desorption Electrospray Ionization (DESI). DESI in an imaging mode gives chemical information about a surface while maintaining the spatial distribution of the analytes. Figure 2 shows a typical positive DESI-MS spectrum of mouse brain tissue. The spectrum shows a pattern of peaks between m/z 700 and 900 attributed mostly to phospholipid (and some other lipid classes) species and a characteristic peak at nominal m/z 369 (exact m/z 369.3516), which is known to be a fragment ion formed by the loss of water from protonated cholesterol molecule ([M + H H2O]+) in several other soft ionization techniques.12,46-48 We also confirmed the origin of the peak by running DESI of a pure cholesterol standard dried on a Teflon surface. Because of this (44) Fahy, E.; Subramaniam, S.; Brown, H. A.; Glass, C. K.; Merrill, A. H.; Murphy, R. C.; Raetz, C. R. H.; Russell, D. W.; Seyama, Y.; Shaw, W.; Shimizu, T.; Spener, F.; van Meer, G.; VanNieuwenhze, M. S.; White, S. H.; Witztum, J. L.; Dennis, E. A. J. Lipid Res. 2005, 46, 839–861. (45) Fahy, E.; Subramaniam, S.; Brown, H. A.; Glass, C. K.; Merrill, A. H.; Murphy, R. C.; Raetz, C. R. H.; Russell, D. W.; Seyama, Y.; Shaw, W.; Shimizu, T.; Spener, F.; van Meer, G.; VanNieuwenhze, M. S.; White, S. H.; Witztum, J. L.; Dennis, E. A. Eur. J. Lipid Sci. Technol. 2005, 107, 337– 364. (46) Johnson, D. W.; ten Brink, H. J.; Jakobs, C. J. Lipid Res. 2001, 42, 1699– 1705. (47) Palmgren, J. J.; Toyras, A.; Mauriala, T.; Monkkonen, J.; Auriola, S. J. Chromatogr., B: Anal. Technol. Biomed. Life Sci. 2005, 821, 144–152. (48) Murata, T.; Takahashi, S.; Takeda, T. Anal. Chem. 1975, 47, 577–580.

undoubted identification, this peak can be used for a single-point internal mass correction. Cholesterol, a sterol lipid, is found predominantly in the plasma membrane bilayer, and high abundance of its in-source fragment ion in the DESI-MS spectrum relative to phospholipids is due to cholesterol being a single molecule, whereas the signal of phospholipids is divided into many different molecular species that are present in the same cellular compartment as cholesterol (it was suggested by a reviewer that higher ionization efficiency of cholesterol can also contribute to the higher abundance). The actual distribution of lipid classes within the tissue regions and the distribution individual lipids within the tissue compartments is not well understood.12 The base peak is at the nominal m/z 760, and it has been previously identified as protonated molecule [PC(16/18) + H]+, molecular formula C42H82NO8P, and can be potentially used for internal mass postcorrection too. The complexity of the DESI-MS tissue spectrum is enormous, and even with mass accuracy and resolution of the FTICR instrument, which dramatically reduces isobaric interferences, there are still many isomers for each exact monoisotopic mass that cannot be distinguished by mass. Table S1 included in the Supporting Information summarizes the major phospholipids identified through the database search. The database search resulted in large number of matches in known phospholipidom, but only combinations with even number of carbons on fatty acid chains (normal in mammals) were included in Supporting Information Table S1. The strong presence of sodiated and potassiated adducts, as shown in Supporting Information Table S1, is undoubtedly due to the high natural concentrations of these alkali cations in tissues. Supporting Information Table S1 lists exact masses that matched the database search and number of potential structures that correspond to each molecular formula. The high resolution of the FTICR-MS instrument allowed discovering that many peaks in DESI-MS spectrum of tissue sections have the same nominal mass, and thus two (or more) different species with different molecular formula are present for each nominal m/z (e.g., peaks at nominal m/z 732, 746, 756, 764, 782, 792, 797, 810, 832). Interestingly, Wiseman et al.35 previously reported (using a low-resolution instrument) that lipid peaks at 810 and 832 are among those that were more dominant in DESI-MS spectra of metastatic human liver adenocarcinoma tissue relative to healthy liver tissue. The highresolution spectra show that the peak at m/z 810 is formed by [PC(16/22) + H+] or [PC(18/20) + H]+ (exact mass 810.6008) as well as by sphingolipid N-(15Z-tetracosenoyl)-1-β-glucosylsphing-4-enine (exact mass 810.6818) and the peak at m/z 832 is formed by [PC(18/22) + H+] or [PC(20/20 + H]+ (exact mass 832.5851) and sphingolipid N-(15Z-tetracosenoyl)-1-βglucosyl-sphing-4-enine (exact mass 832.6643). Figure 3 shows the baseline-resolved peaks at nominal m/z 832 as an example (spectra zoomed on nominal m/z 810 are included in the Supporting Information, Figure S4). Sphingolipids are known to be unexplored surface markers for many cell disorders, and their elevation and changes relative to the healthy state of a cell is often a result of oncology illnesses.49 Table 1 summarizes all sphingolipid candidates found by high-resolution DESI-MS. These findings clearly demonstrate that, although tandem mass spectrometry methods (or even non-MS-based techniques) are needed (49) Ogretmen, B.; Hannun, Y. A. Nat. Rev. Cancer 2004, 4, 604–616.

Analytical Chemistry, Vol. 81, No. 20, October 15, 2009

8483

Figure 3. Baseline-resolved lipid ions at nominal m/z 832. The peak at 832.5847 is a phosphocholine lipid; the peak at 832.6630 is sphingolipid N-(15Z-tetracosenoyl)-1-β-glucosyl-sphing-4-enine.

for isomer identification, exact mass measurement with high resolution are useful for quick probing of the lipid composition and for addressing lipid changes due to biological factors on real samples. Sphingolipids and phospholipids cannot be distinguished at low mass resolution, and separation of lipid classes has been previously achieved by DESI coupled to a thin-layer chromatography (TLC) on a low-resolution linear ion trap instrument.50 As we demonstrate here, the chromatographic step can be avoided if sufficient resolution and mass accuracy is used. In imaging mode, the spatial resolution of the mass spectrometry image can be decided by the operator when setting up the experiment in FlexImaging software, but there are principal and practical limitations for DESI resolution. Usually, the useful resolution is in the order of hundreds of micrometers, although much lower values were reported,19 and in this particular proofof-principle experiment the system was set up to acquire spectra every 500 µm and scan an area 10 mm × 10 mm, which resulted in matrix of 400 spectra (20 × 20) that occupied roughly 800 MB of disk space (broadband 512K points, m/z range of 250-1800). The distribution of “laser” spots in FlexImiging software, which is identical with the DESI sprayer target spots in the modified setup, can be seen on Supporting Information Figure S2. The actual spatial resolution, defined by the distances between these spots, was 500 µm. Figure 4 shows the 2D distribution of selected lipids (with ±1 mDa accuracy) in a brain horizontal section and their combinations. Panels A and B in Figure 4 show distributions of peaks at m/z 760.5850 and 810.6006. The suggested identities of individual m/z obtained from the database search are in Supporting Information Table S1 (phospholipids) and Table 1 (sphyngolipids). The power of the accurate mass and highresolution instrument is demonstrated in Figure 4C, where combined distributions of three peaks with nominal m/z 814 can be seen. The three peaks in the figure have exact m/z 814.5160, (50) Ifa, R. I. Presented at the ASMS Annual Conference on Mass Spectrometry, Philadelphia, PA, May 31-June 4, 2009.

8484

Analytical Chemistry, Vol. 81, No. 20, October 15, 2009

814.6317, and 814.6878. As can be seen in Supporting Information Figure S5 they are baseline-resolved by the spectrometer, and they can be imaged separately within the same scan because they will overlap spatially on the surface but not by m/z in the mass analyzer. The knowledge of the surface distribution provides additional information useful for analyte identification. For instance, the isotopic peaks must have the same 2D distribution. If not, another specie interferes at the same m/z. Similar should apply for the protonated molecule and alkali adducts of the same analyte, although one could theoretically imagine a situation where differences in alkali metal concentration in tissue will result in differences between distributions of proton and alkali adducts. The contaminants from the environment, solvents, or surface of the platform are immediately obvious because they are equally distributed everywhere in the square image, even outside the sample (Supporting Information Figure S6). Desorption Atmospheric Pressure Photoionization (DAPPI). DAPPI is a surface utilization of APPI, which was originally developed in order to broaden the group of analytes that can be analyzed by LC-MS toward less polar compounds. DAPPI has a similar relation to APPI as DESI has to ESI, and it allows surface analysis of less polar compounds directly from surfaces. In this report DAPPI has been selected as the second ambient MS technique demonstrated with the integrated imaging platform on FTICR-MS. The ionization of analytes in DAPPI occurs through photoionization and gas-phase ion-molecule reactions. Recent studies51,52 performed on an ion trap and orbitrap instruments revealed the advantages of ambient ionization methods for the direct screening of plant and fungal materials. An available parched leaf of Salvia (sage) that dried out in the laboratory environment was thus selected as a proof-of-principle sample for the high mass resolution DAPPI-MS imaging. For imaging mode, spatial resolution was set up in FlexImage to be the same as in DESI experiments. However, the thermal-shock desorption, which is believed to be the dominant desorption mechanism in DAPPI, limits the maximum resolution that can be obtained to approximately 1 mm (with the heated nebulizer microchip used in this work and with the glass surface used, because thermal conductivity of the sample surface may affect the spatial distribution of the desorbed analyte). Supporting Information Figure S7 shows the DAPPI-MS spectrum obtained from the Salvia leaf. One immediately noticeable feature is the absence of peaks in the region below m/z ) 180, where well-known monoterpenic components of different Salvia oils can be found by standard analysis of Salvia extract.53,54 It is presumed that the volatile compounds have been evaporated from the leaf during drying although known poorer ion transmission of the used mass spectrometer in the low m/z region could contribute as well. On the other hand, the spectrum is especially rich between m/z 300 and 600, where mostly aromatic compounds, esters, and higher terpenoids can be found. Peaks above m/z 600 were due to surface and lab-environment contamination, which (51) Jackson, A. U.; Tata, A.; Wu, C. P.; Perry, R. H.; Haas, G.; West, L.; Cooks, R. G. Analyst 2009, 134, 867–874. (52) McEwen, C.; Gutteridge, S. J. Am. Soc. Mass Spectrom. 2007, 18, 1274– 1278. (53) Kivilompolo, M.; Oburka, V.; Hyotylainen, T. Anal. Bioanal. Chem. 2007, 388, 881–887. (54) Ollanketo, M.; Peltoketo, A.; Hartonen, K.; Hiltunen, R.; Riekkola, M. L. Eur. Food Res. Technol. 2002, 215, 158–163.

Figure 4. Two-dimensional distributions of selected lipid-originating ions in mouse brain tissue section: (A) m/z 760.5850; (B) m/z 810.6006; (C) distribution of three peaks with the same nominal mass 814 (red, 814.5160; blue, 814.6317; yellow, 814.6878).

Table 1. Possible Sphingolipid Candidates Obtained from a Database Search input mass

matched mass

δ

abbreviation

formula

648.629 703.5745 729.5914 731.6063 731.6063 733.6129 759.6367 810.6808 813.6839 753.5868 753.5868 808.6666 832.663 834.6835 834.6835 604.5063 604.5063 766.5603 766.5603 766.5603 769.5616 769.5616 848.6378 850.6581 850.6581

648.6289 703.5749 729.5905 731.6062 731.6062 733.6218 759.6375 810.6818 813.6844 753.5887 753.5887 808.6643 832.6643 834.6799 834.6799 604.5071 604.5071 766.5517 766.5599 766.5599 769.5626 769.5626 848.6382 850.6538 850.6538

0.0001 0.0004 0.0009 0.0001 0.0001 0.0089 0.0008 0.001 0.0005 0.0019 0.0019 0.0023 0.0013 0.0036 0.0036 0.0008 0.0008 0.0086 0.0004 0.0004 0.001 0.001 0.0004 0.0043 0.0043

Cer(d18:1/24:1(15Z)) SM(d18:1/16:0) SM(d18:1/18:1(9Z)) SM(d18:0/18:1(9Z)) SM(d18:1/18:0) SM(d18:0/18:0) SM(d18:1/20:0) GlcCer(d18:1/24:1(15Z)) SM(d18:1/24:1(15Z)) SM(d18:0/18:1(9Z)) SM(d18:1/18:0) GlcCer(d18:0/22:0) GlcCer(d18:1/24:1(15Z)) GlcCer(d18:0/24:1(15Z)) GlcCer(d18:1/24:0) Cer(d18:0/18:1(9Z)) Cer(d18:1/18:0) CerP(d18:1/24:1(15Z)) GlcCer(d18:0/18:1(9Z)) GlcCer(d18:1/18:0) SM(d18:0/18:1(9Z)) SM(d18:1/18:0) GlcCer(d18:1/24:1(15Z)) GlcCer(d18:0/24:1(15Z)) GlcCer(d18:1/24:0)

C42H81NO3 C39H79N2O6P C41H81N2O6P C41H83N2O6P C41H83N2O6P C41H85N2O6P C43H87N2O6P C48H91NO8 C47H93N2O6P C41H83N2O6P C41H83N2O6P C46H91NO8 C48H91NO8 C48H93NO8 C48H93NO8 C36H71NO3 C36H71NO3 C42H82NO6P C42H81NO8 C42H81NO8 C41H83N2O6P C41H83N2O6P C48H91NO8 C48H93NO8 C48H93NO8

ion [M [M [M [M [M [M [M [M [M [M [M [M [M [M [M [M [M [M [M [M [M [M [M [M [M

+ + + + + + + + + + + + + + + + + + + + + + + + +

H]+ H]+ H]+ H]+ H]+ H]+ H]+ H]+ H]+ Na]+ Na]+ Na]+ Na]+ Na]+ Na]+ K]+ K]+ K]+ K]+ K]+ K]+ K]+ K]+ K]+ K]+

Analytical Chemistry, Vol. 81, No. 20, October 15, 2009

8485

Figure 5. (A) Distribution of an ion at m/z 430.3810 on sage leaf; (B) combined distributions of peaks at m/z 301.2166 (green) and 315.0863 (red). Whereas 301 is located more on the side of the leaf, 315 is centered. The inset shows the part of the leaf that was imaged.

was also ionized by DAPPI. The high resolution, together with external and internal mass calibrations, allowed the determination of exact masses with errors below 1 ppm, which is usually sufficient for determination of molecular formula. However, due to the fact that phytochemistry is very rich in isomers there are many different phytocompounds with the same molecular formula, which makes it impossible to characterize the actual structure without comparing tandem spectra with standards. Although the object was to demonstrate 2D chemical imaging with DAPPI we still tried to identify some of the interesting analytes by comparing them with available literature reports. For instance, the peak at nominal m/z 430 corresponds to tocopherol radical cation with mass error 0.9 ppm (for the 2D distribution on the leaf see Figure 5A). Tocopherol, a compound with vitamin E activity and common food additive, was previously identified in Salvia leaves.55 Another study reported that in ESI and APCI tocopherol predominantly ionizes as [M - 2H + H]+ (m/z 429), whereas in APPI (DAPPI and APPI spectra are often similar) it forms the molecular ion as well as this fragment.56 Both peaks have overlapping 2D distribution, which makes it possible that the 429 originates from tocopherol in accordance with the previous studies. The peak at m/z 331 corresponds to carnosol protonated molecule (-0.09 ppm), and the peak at m/z 301 corresponds to methyl carnosic acid (-0.3 ppm), which are all known to be present in Salvia.57 The peak at m/z 283 has close nominal mass (-0.7 ppm) as reported carnosol fragment,58 and its surface distribution overlaps with carnosol mass. Carnosic acid and its derivates are known as strong antioxidizing factors. It is thus interesting that maximum intensities of unknown peak at m/z 315 were obtained from areas where no carnosic derivates were detected (Figure 5B). The DAPPI imaging platform was also tested on brain sections. Interestingly, when acetone was used as a dopant the DAPPI (55) Abreu, M. E.; Muller, M.; Alegre, L.; Munne-Bosch, S. J. Sci. Food Agric. 2008, 88, 2648–2653. (56) Cai, Y. X.; Kingery, D.; McConnell, O.; Bach, A. C. Rapid Commun. Mass Spectrom. 2005, 19, 1717–1724. (57) Schwarz, K.; Ternes, W. Z. Lebensm.-Unters.-Forsch. 1992, 195, 99–103. (58) Maillard, M. N.; Giampaoli, P. Talanta 1996, 43, 339–347.

8486

Analytical Chemistry, Vol. 81, No. 20, October 15, 2009

ionization became extremely selective and the cholesterol fragment at nominal mass 369 is practically the only peak in the spectrum with an intensity 108 (Supporting Information Figure S8), which is 2 orders of magnitude higher than in the DESI-MS spectrum obtained on the same instrument. CONCLUSION Ambient mass spectrometry is a very quickly emerging field of mass spectrometry which is now subject of research in different mass spectrometry groups. One of the potential problems of the ambient MS concept is the complexity of the spectra that are measured directly without the time-consuming isolation and separation steps involved. In this work we demonstrated that the combination of high resolution and mass accuracy and integrated 2D imaging mode can help to identify analytes even in extremely complex samples by ambient MS approaches. We showed that more biologically relevant information is obtained by DESI-MS of tissues in imaging mode if high resolution is used. As an example we were able to distinguish two different lipid classes. We also demonstrated that DAPPI can be utilized for imaging of analytes from complex matrixes. We are aware that direct single MS ambient measurement (without a preseparation step and without tandem MS) suffers many limitations when dealing with really complex matrixes. However, our examples demonstrate that in some cases a known in-source fragmentation pattern (example of cholesterol) together with careful comparing of 2D distribution of suspected precursor-product ion pairs (tocopherol) can help confirm, or rule out, candidate compounds even without measuring the collisional tandem MS spectra. ACKNOWLEDGMENT The work was supported by the Ministry of Education, Youth, and Sports of the Czech Republic (LC07017, MSM 6198959216) and by an Institutional research concept of the Institute of Microbiology, Academy of Sciences of the Czech Republic (AV0Z50200510). M.V.’s research was supported by a Marie Curie International Reintegration Grant within the seventh European

Community Framework Program. J.P. thanks the Academy of Finland for financial support. We also thank Martin Strohalm for the help with image processing and for useful discussion of the results. Karel Sˇefcˇ´ık is gratefully acknowledged for machining and other engineering work. Sami Franssila and Ville Saarela are acknowledged for the fabrication of the heated nebulizer microchip.

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 June 24, 2009. Accepted August 28, 2009. AC901368Q

Analytical Chemistry, Vol. 81, No. 20, October 15, 2009

8487