High-Spatial Resolution Matrix-Assisted Laser Desorption Ionization

Apr 12, 2010 - Lysosomal Diseases Research Unit, SA Pathology at Women's and Children's Hospital, 72 King William Road,. North Adelaide, South Austral...
0 downloads 0 Views 3MB Size
Download PDF
Anal. Chem. 2010, 82, 3664–3670

High-Spatial Resolution Matrix-Assisted Laser Desorption Ionization Imaging Analysis of Glucosylceramide in Spleen Sections from a Mouse Model of Gaucher Disease Marten F. Snel and Maria Fuller* Lysosomal Diseases Research Unit, SA Pathology at Women’s and Children’s Hospital, 72 King William Road, North Adelaide, South Australia, Australia, 5006 MALDI mass spectrometric imaging (MSI) enables spatially resolved mass and intensity information to be obtained directly from tissue sections, thereby illustrating how analytes are distributed within these sections. Here we have used an oversampling technique on a commercially available MALDI orthogonal acceleration TOF mass spectrometer with ion mobility separation capability to produce high spatial resolution images of the glycosphingolipid, glucosylceramide (GC). To exemplify the biological application of our approach, GC was imaged in spleen sections from a conditional knockout mouse model of type 1 Gaucher disease in which the catabolism of this glycosphingolipid is impaired. The laser was continually fired at one position until no more ions were observed and then the sample was moved by 15 µm (laser diameter ∼150 µm). Ions were generated from only the unirradiated surface at each of these positions achieving an effective spacing of 15 µm. At 15 µm laser stepsize, it was possible to visualize macrophages enriched in GC, which could be distinguished from other cell types in the spleen. Current MALDI MSI spatial resolution is typically limited by the diameter of the laser spot-size, which is usually between 50 and 100 µm, covering an area equivalent to tens of mammalian cells. Mass spectrometric imaging (MSI) provides spatial as well as mass and intensity information on selective compounds in tissue sections.1 Matrix-assisted laser desorption ionization (MALDI) MSI has been used in situ for compounds such as intact proteins,2 ontissue enzymatic digestion of proteins on both fresh frozen sections and formalin fixed archive sections,3,4 small drug molecules and metabolites,5,6 and lipids.7,8 Routinely, tissue samples are cut into 10-15 µm sections which are evenly coated with MALDI matrix, * To whom correspondence should be addressed. Phone: 61 8 8161 6741. Fax: 61 8 8161 7100. E-mail: [email protected]. (1) Caprioli, R. M.; Farmer, T. B.; Gile, J. Anal. Chem. 1997, 69, 4751–4760. (2) Stoeckli, M.; Chaurand, P.; Hallahan, D. E.; Caprioli, R. M. Nat. Med. 2001, 7, 493–496. (3) Lemaire, R.; Desmons, A.; Day, R.; Salzet, M.; Fournier, I. J. Proteome Res. 2007, 6, 1295–1305. (4) Djidja, M. C.; Francese, S.; Loadman, P. M.; Sutton, C. W.; Scriven, P.; Claude, E.; Snel, M. F.; Franck, J.; Salzet, M.; Clench, M. R. Proteomics 2009, 9, 2750–2763. (5) Stoeckli, M.; Staab, D.; Schweitzer, A. M. Int. J. Mass Spectrom. 2007, 260, 195–202.

3664

Analytical Chemistry, Vol. 82, No. 9, May 1, 2010

typically by spraying an aerosol of matrix solution at the tissue surface. A pulsed laser desorbs matrix and analytes from the sample surface, and during the desorption process the ionization of analytes occurs. Ions generated are then measured using a mass analyzer. With the sample moved relative to the laser and recording data at each locus, spatial information is gathered. Experimental data are transformed into ion intensity maps, which display spatial location in an x, y and/or z coordinate system. A color scale is used to show ion signal intensity if one ion is displayed or ion type, when multiple ions are displayed in the one image. The spatial resolution at which data are acquired during an MSI experiment determines the level of detail of the images generated. MALDI imaging resolution is primarily limited by the diameter of the laser beam. Most MALDI imaging data report pixel sizes of 50-500 µm.2-9 For example, a laser step-size of 500 µm was used to determine organ distribution of β-peptides in whole body sections of mice.9 The advantage of low spatial resolution is to reduce data file size and acquisition time. At 100 µm laser step-size, it was possible to distinguish substructures in a rat brain, such as the hippocampus, medial geniculate body, and the interpeducular nucleus when imaging phosphatidylcholine (16:0, 20:4).8 However, compared to the size of a typical mammalian cell (5-20 µm), a laser step-size of 50-100 µm will only provide MS information on groups of cells rather than single cells. Animal tissues contain different cell types; therefore; high-spatial resolution MALDI MSI would enable the localization of ions to particular cell types. To obtain information on single cells, the spacing between observations would need to be at most the same as the diameter of the cell. Secondary ion mass spectrometry (SIMS) and MALDI ion sources have been used for high-resolution MSI.10 SIMS highresolution tissue images can be generated showing subcellular detail, e.g., changes in the level of phosphatidylcholine (PC) in (6) Trim, P. J.; Henson, C. M.; Avery, J. L.; McEwen, A.; Snel, M. F.; Claude, E.; Marshall, P. S.; West, A.; Princivalle, A. P.; Clench, M. R. Anal. Chem. 2008, 80, 8628–8634. (7) Burnum, K. E.; Cornett, D. S.; Puolitaival, S. M.; Milne, S. B.; Myers, D. S.; Tranguch, S.; Brown, H. A.; Dey, S. K.; Caprioli, R. M. J. Lipid Res. 2009, 50, 2290–2298. (8) Garrett, T. J.; Prieto-Conaway, M. C.; Kovtoun, V.; Bui, H.; Izgarian, N.; Stafford, G.; Yost, R. A. Int. J. Mass Spectrom. 2007, 260, 166–176. (9) Stoeckli, M.; Staab, D.; Schweitzer, A.; Gardiner, J.; Seebach, D. J. Am. Soc. Mass Spectrom. 2007, 18, 1921–1924. (10) Fletcher, J. S. Analyst 2009, 134, 2204–2215. 10.1021/ac902939k  2010 American Chemical Society Published on Web 04/12/2010

cell membranes during Tetrahymena mating have been shown by monitoring the 184 Da headgroup of (PC).11 Cluster-time-of-flight (TOF)-SIMS has been used to image the distribution of cholesterol, cholesterol sulfate, and vitamin E in skin samples from Fabry disease patients, but it was not possible to image individual globotriaosylceramide (Gb3) or digalactosylceramide (Ga2).12 Because of the limited sensitivity at higher m/z, SIMS experiments are often restricted to imaging ions below 500 Da and thus not suitable for high spatial resolution analysis of intact lipid species. On the other hand, MALDI MSI has a wide m/z range and has been used for the imaging analysis of many classes of lipids.7,8 There are two strategies for increasing MALDI MSI resolution, either by decreasing the diameter of the laser spot-size or using a standard spot-size and only exposing part of the sample to laser irradiation. Reducing laser spot-size was shown by advanced optics, producing laser spot-sizes of ∼1 µm in diameter. However, as yet, this technique has not been applied to the analysis of tissue sections.13 A similar MALDI ion source was also mounted on an axial TOF system, enabling the spatial distribution of two proteins with a scanning resolution of 10 µm.14 In a method for increasing spatial resolution without altering laser spot-size, tissue sections were deposited on a closely packed array of glass beads with a diameter of 38 µm.15 The beads were mounted in Parafilm, which was stretched after tissue deposition to separate the beads. Peptide mass spectra were recorded from individual tissue coated beads. The spatial resolution was determined by the size of the beads, which was ∼38 µm. The drawback of this approach was the poor reproducibility of the position of the glass beads after stretching which prevented automation of data analysis. A further method, using an oversampling approach with a conventional laser beam diameter on an unmodified MALDI mass spectrometer, achieved high spatial resolution.16,17 The ionizing laser was fired at a fixed position until it yielded no more ions, upon which the laser position was moved from the original position by a distance smaller than the laser diameter to a second position. At the second position, ions were only generated from the sample surface that had not previously been exposed to laser irradiation; therefore the effective area was reduced. This technique was used to produce ion images with a pixel size of 15 µm of peptide standards on an electron microscopy calibration grid and to detect peptides from an isolated Aplysia californica peptidergic neuron.16 An oversampling technique has also been demonstrated with an infrared laser (250 µm average spot size), using a sample of toluidine blue O which was masked by an electron microscope grid and imaged with a step-size of 40 µm. In the resultant ion image, the grid could clearly be discerned and a lateral scanning resolution of 40 µm was reported.17 Although the potential of oversampling as a means of improving spatial resolution in a MALDI imaging experi(11) Ostrowski, S. G.; van Bell, C. T.; Winograd, N.; Ewing, A. G. Science 2004, 305, 71–73. (12) Touboul, D.; Roy, S.; Germain, D. P.; Chaminade, P.; Brunelle, A.; Lapre´vote, O. Int. J. Mass Spectrom. 2007, 260, 158–165. (13) Koestler, M.; Kirsch, D.; Hester, A.; Leisner, A.; Guenther, S.; Spengler, B. Rapid Commun. Mass Spectrom. 2008, 22, 3275–3285. (14) Chaurand, P.; Schriver, K. E.; Caprioli, R. M. J. Mass Spectrom. 2007, 42, 476–489. (15) Monroe, E. B.; Jurchen, J. C.; Koszczuk, B. A.; Losh, J. L.; Rubakhin, S. S.; Sweedler, J. V. Anal. Chem. 2006, 78, 6826–6832. (16) Jurchen, J. C.; Rubakhin, S. S.; Sweedler, J. V. J. Am. Soc. Mass Spectrom. 2005, 16, 1654–1659. (17) Li, Y.; Shrestha, B.; Vertes, A. Anal. Chem. 2007, 79, 523–532.

ment has been depicted, there have been no reports of its biological relevance and, more importantly, whether it is possible to measure an endogenous compound in a conventional tissue section at this high resolution. Here we have used an unmodified commercially available MALDI mass spectrometer to achieve an effective ablation area of 15 µm using an oversampling technique. To test the biological feasibility of this approach, we analyzed glucosylceramide (GC) in tissue sections from a conditional knockout mouse model of type 1 Gaucher disease in which the catabolism of GC is impaired.18 We show that an endogenous glycosphingolipid, GC, was localized to macrophages in spleen sections. The presence of GC in tissue was confirmed by on-tissue MALDI MS/MS analysis using ion mobility separation (IMS) to remove interfering background ions. EXPERIMENTAL SECTION Sample Preparation. The spleens were removed from a 1 year old conditional knockout Gaucher type 1 mouse model19 and an unaffected control (both female) following euthanasia. The Women’s and Children’s Hospital Animal Ethics Committee (Adelaide, SA, Australia) approved the use of these mice for this study. Whole spleens were immediately frozen by immersion in dry ice cooled hexane, which had been precooled from room temperature for 5 min. Frozen spleens were cut in half transversely and mounted in optimal cutting temperature polymer (OCT) in such a way that half of the tissue protruded from the OCT. This made it possible to cut cross sections without the blade coming into contact with the OCT. Sections were cut in a Cryotome E (Thermo Fischer Scientific, Waltham, MA) at -18 °C at a thickness of 15 µm and were thaw mounted on microscope slides (Menzel-Gla¨ser, Braunschweig, Germany). Matrix Application. R-Cyano-4-hydroxycinnamic acid (CHCA) matrix was purchased from Sigma (Sigma-Aldrich, St. Louis, MO) and used without further purification. For each microscope slide, 30 mL of matrix solution was used at a concentration of 25 mg mL-1 in 700:300:1 v/v/v ethanol (Merck, Kilsyth, VA, Australia)-Milli-Q water-trifluoroacetic acid (Sigma-Aldrich, St. Louis, MO). Microscope slides were mounted on a cardboard support (35 cm × 20 cm) using sticky tape and placed against the back wall of a fume cupboard. Matrix solution was sprayed from a distance of 50 cm using an Eclipse HP-CS airbrush with a 0.3 mm nozzle (Anest Iwata, Yokohama, Japan) equipped with a smart jet pro air compressor (Anest Iwata, Yokohama, Japan) operating at 0.3 mPa. The matrix was applied in coats by passing the spray over the sample in a sweeping motion from left to right then right to left and then back again over 3 s. The coat was allowed to dry (∼20 s) before another coat was applied. This process was repeated until 30 mL of matrix solution had been sprayed onto the target. Mass Spectrometry. All mass spectrometric analyses were performed on a SYNAPT HDMS (Waters, Manchester, U.K.) orthogonal geometry time-of-flight mass spectrometer fitted with an intermediate pressure MALDI source. This instrument was equipped with an IMS device located between the quadrupole and (18) Kattlove, H. E.; Williams, J. C.; Gaynor, E.; Spivack, M.; Bradley, R. M.; Brady, R. O. Blood 1969, 33, 379–390. (19) Sinclair, G. B.; Jevon, G.; Colobong, K. E.; Randall, D. R.; Choy, F. Y. M.; Clarke, L. A. Mol. Gen. Metab. 2007, 90, 148–156.

Analytical Chemistry, Vol. 82, No. 9, May 1, 2010

3665

the TOF mass analyzer.20 Prior to all analyses, the mass spectrometer was mass calibrated over a m/z range from 100 to 1000 using a mixture of two poly(ethylene glycol) (PEG) standards (Sigma-Aldrich, St. Louis, MO), with average molecular weights of 600 and 1000 Da. The PEG standards were prepared at 1 mg mL-1 in a 1:1 v/v mixture of acetonitrile (Honeywell, Morristown, NJ)-water and combined with CHCA matrix solution (CHCA at 3 mg mL-1 in 1:1 v/v acetonitrile-water) and aqueous NaCl 1 mg mL-1 (Sigma-Aldrich, St. Louis, MO) at 5:5:1 v/v/v and spotted onto a stainless steel MALDI target plate in 1 µL aliquots and allowed to dry. PEG standard was mass analyzed on the mass spectrometer in positive ion V-mode of operation, and resultant data were used to mass calibrate. MALDI Imaging MS Data Acquisition on Spleen Sections. Following matrix application, slides were mounted on a MALDI sample holder (Waters, Manchester, U.K.). A digital image of the sample holder was made using a CanoScan LiDE 100 flat bed scanner (Canon, Tokyo, Japan). The image file was imported into MALDI Imaging Pattern Creator software (Waters, Manchester, U.K.), which was used to set the area to be imaged. Areas to be analyzed were marked in the software using a graphic drawing interface. The software was modified to allow minimum pixel spacing of 3.75 µm to be set (default limit was 25 µm). The areas to be analyzed covered the whole spleen for analysis at 50 µm step-size or an area of 3 mm × 3 mm for 15 µm step-size. Coordinate lists were exported, which were used in MALDI imaging methods in the instrument control software, MassLynx v4.1 (Waters, Manchester, U.K.). Data were acquired for 1 s at 200 Hz per laser position with a laser intensity setting of 250 (arbitrary units referring to the position of a variable neutral density filter in the laser beam path). The laser used was a frequency tripled Nd:YAG laser operating at 355 nm. Data were acquired over a mass range of 100-1000 Da. Postacquisition, mass spectral data were converted into BioMap (Novartis, Basel, CH) readable format using MALDI Imaging Convertor (Waters, Manchester, U.K.). In the conversion process, the m/z data dimension was condensed by averaging data into 30 mDa data bins. MALDI Imaging Convertor software was modified to handle data with pixel spacing smaller than 25 µm. Image analysis was performed using BioMap v. 3.7.5.5. To avoid smoothing artifacts and overinterpretation of imaging data, no image smoothing was applied. Intensities in ion maps were not averaged across several m/z values and no background subtraction was used. MALDI MS/IMS/MS of GC on Tissue. MS data from GCd3(16:0) standard (Matreya LLC, Pleasant Gap, PA) was compared with GC(16:0) present in the spleen tissue of the Gaucher mouse model, to confirm correct assignment of GC(16:0) in tissue. GC-d3(16:0) standard was prepared at 10 pmol µL-1 in methanol and mixed in a 10:10:1 v/v/v ratio with CHCA matrix solution and 20 mM aqueous NaCl (SigmaAldrich, St. Louis, MO) solution. A 1 µL aliquot of this mixture was then spotted on a stainless steel MALDI target (Waters, Manchester, U.K.) and allowed to dry. The GC-d3(16:0) was mass analyzed using MALDI MS/MS coupled with IMS. (20) Pringle, S. D.; Giles, K.; Wildgoose, J. L.; Williams, J. P.; Slade, S. E.; Thalassinos, K.; Bateman, R. H.; Bowers, M. T.; Scrivens, J. H. Int. J. Mass Spectrom. 2007, 261, 1–12.

3666

Analytical Chemistry, Vol. 82, No. 9, May 1, 2010

For the spleen sections, MALDI MS/MS data were acquired using automated acquisition in a raster pattern with a spacing of 15 µm between laser positions. For both GC-d3(16:0) and the spleen section, precursor ions were selected in the quadrupole with a precursor ion window of ∼3 Da; background ions were then separated from GC ions using IMS. The IMS cell gas pressure was 0.5 mbar nitrogen, IMS wave velocity was 300 ms-1, and IMS wave height was 10 V. Collision-induced dissociation (CID) was performed in the collision cell (transfer T-Wave) located after the IMS device (collision energy 80 V). Mass measurements were made using an orthogonal acceleration TOF (oa-TOF) system. For on-tissue analysis, the oa-TOF was operated in enhanced duty cycle mode (EDC). EDC exploits the fact that when ions are pulsed out of the collision cell region toward the TOF analyzer, some separation of ions according to their m/z is observed. This makes it possible to increase the efficiency with which selected ions are transmitted through the orthogonal TOF geometry by synchronizing the time at which they are pushed into the TOF analyzer with their arrival time in the TOF pusher region. Here, EDC was required, as signal intensity on tissue was very close to the limit of detection, and EDC was set to maximize the ion signal for m/z 264. Ions were detected between 100 and 350 Da. Optical Microscopy of Spleen Sections. An Olympus BX41 microscope (Olympus America, Melville, NY) equipped with an Olympus ColorView III camera (Olympus America, Melville, NY) was used to take digital micrographs of mouse spleen sections prior to matrix coating. After mass spectrometric analysis, MALDI matrix was washed from the spleen sections using gently running tap water, and the sections were stained with hematoxylin and eosin (H&E). RESULTS AND DISCUSSION Confirmation of GC in the Mouse Spleen Sections. Table 1 lists the theoretical masses of a number of naturally abundant GCs, the primary storage material in Gaucher disease,18 and highlights which masses could be matched to experimental data within ±10 ppm of the theoretical mass. Notably, only sodium and potassium adducts were matched. The absence of proton adducts is likely to be a function of the proton affinity and sodium and potassium cation affinity of GC compared to that of the MALDI matrix.21 Additional experiments to confirm the identification of GC were required, as small molecules (m/z < 1000 Da) can be incorrectly assigned using mass alone, owing to the complexity of the low mass MALDI spectra. In the m/z range of 700-850 Da, in which the compounds of interest were found, many other lipids, such as phospatidylcholines, as well as other background ions, are detected. In order to confirm the presence of GC in tissue, MS/ MS data obtained on tissue were compared with MS/MS data obtained from a GC-d3(16:0) standard, shown in Figure 1. The sodium adduct of GC(16:0) was chosen for MS/MS analysis, as it provided fragment ion spectra with a number of diagnostic fragment ions that were used to confirm the structure. The potassium adduct, although more intense in tissue than the sodium adduct, did not yield any fragment ions in the m/z range monitored. We speculate that under CID conditions, the loss of the K+ ion is the predominant fragmentation channel, resulting (21) Knochenmuss, R. Analyst 2006, 131, 966–986.

Table 1. Exact Masses of Adducts of Commonly Occurring Species of GC, Showing Which Adducts Were Observed in the Gaucher Samples Studied theoretical mass fatty acid 16:0 18:0 20:0 22:0 24:0 24:1

+

H

700.5707 728.6019 756.6331 784.6643 812.6955 810.6799

+

Na adduct 722.5527 750.5839 778.6151 806.6463 834.6775 832.6619

adducts observed in Gaucher tissue +

+

K adduct

H

738.5266 766.5578 794.5890 822.6202 850.6514 848.6358

-a -a -a -a -a a

Na+

K+

+ -a -a +b +b +b

++c +b +b ++c ++c ++c

b

a -: Not observed. b +: Observed at relative intensity up to 50% of base peak intensity. c ++: Observed at relative intensity greater than 50% of base peak intensity.

Figure 1. IMS-MS/MS spectra of sodiated (GC 16:0). (a) GC-d3(16:0) standard, acquired over full mass range and (b) GC(16:0) acquired from tissue using EDC acquisition with a set mass of 264 Da.

in the absence of diagnostic fragment ions. It was necessary to separate precursor ions using IMS, as background ions which differed in mass from the ions of interest by less than 170 mDa were also passed through the quadrupole, yielding fragment ions that complicated the fragment ion spectra. IMS has previously been used for on-tissue analysis to separate a phospholipid from a drug molecule,6 and lipids are known to typically have longer drift times than other common species found in tissue.22,23 In the experiments described here, background ions could be filtered out completely by IMS and all data shown here were isolated using this technique prior to CID. (22) Jackson, S. N.; Wang, H. J.; Woods, A. S.; Ugarov, M.; Egan, T.; Schultz, J. A. J. Am. Soc. Mass Spectrom. 2005, 16, 133–138. (23) McLean, J. A.; Ridenour, W. B.; Caprioli, R. M. J. Mass Spectrom. 2007, 42, 1099–1105.

For on-tissue analysis, optimum instrument sensitivity was required to detect low intensity fragment ions, hence the oa-TOF was operated with an enhanced duty cycle (EDC).24 The normal efficiency, with which ions are transferred from the collision cell into the oa-TOF system (the duty cycle) is between 10-20%, as ions cannot be transferred while a time-of-flight measurement is taking place. The duty cycle can be improved to near 100%, for a limited part of the m/z range, by exploiting the fact that ions are separated according to m/z as they are ejected from the collision cell. The time at which ions are pushed into the TOF system is synchronized with the arrival from the collision cell of ions of a particular m/z. It should be noted that only the selected ions and (24) Pringle, S. D.; Wildgoose, J. L.; Giles, K.; Worthington, K.; Bateman, R. H. Proceedings of the 52nd ASMS Conference on Mass Spectrometry and Allied Topics, Nashville, TN, May 23-27, 2004.

Analytical Chemistry, Vol. 82, No. 9, May 1, 2010

3667

ions of similar m/z are transferred and transmission of other ions is cut off. This effect is illustrated in Figure 1, where only the lower m/z part of the on-tissue MS/MS spectrum (recorded using EDC) is populated whereas the MS/MS spectrum of the standard (recorded using standard duty cycle) is complete. The m/z range from 100 to 350 was chosen, as many of the diagnostic fragment ions were found in this range. The two mass spectra depicted in Figure 1 show good agreement, confirming that the ion observed at 722.6 Da in the Gaucher spleen samples is indeed the sodium adduct of GC(16:0). A guide to aid full interpretation of CID spectra of glycosphingolipids generated on an oa-TOF mass spectrometer equipped with a MALDI source is given elsewhere.25 The fragment ions marked as B, C, Y, and Z confirm the presence of glucose as indicated in Figure 1. The ion with mass 264.3 is commonly used as a diagnostic fragment ion for the sphingosine backbone of sphingolipids.26,27 The fragment ions with m/z 280.3 and 304.3 Da (283.3 and 307.3 Da for the triply deuterated standard) are derived from the C16:0 palmitic acid residue as illustrated in Figure 1. Spatial Distribution of GC in Mouse Spleen. The distribution of GC(16:0) is shown in Figure 2, and all fatty acids (16:0, 18:0, 20:0, 22:0, 24:0, and 24:1) exhibited the same distribution pattern, which varied mainly in intensity (see Figures S1 and S2 in the Supporting Information). Therefore we used images from GC(16:0) to represent all GC data. MALDI MSI data were generated with two different laser step-size settings of 50 and 15 µm. Spleen tissues from the conditional knockout type 1 Gaucher mouse model and from an unaffected control of the same age and gender were compared. For each set of analyses, the Gaucher and control tissue were prepared, mounted on the same microscope slide, and coated with matrix all at the same time to minimize experimental variation. The effective analysis areas with diameters of 50 µm (area ≈ 2500 µm2) and 15 µm (area ≈ 225 µm2) were achieved using an oversampling approach. In this method, a laser position was maintained until no more signal was obtained from that spot, then the laser position was shifted by a distance smaller than the laser spot-size. At the second and subsequent laser positions, signal was only obtained from the area not previously irradiated, thus producing a smaller effective analysis area. This method was effective despite the fact that chosen increments were much smaller than the original laser spot, which was elliptical with dimensions of ∼150 µm × 75 µm (area ≈ 8836 µm2). Figure 2 shows representative mass spectra for each imaging experiment. Each mass spectrum was prepared by the summation of data from 10 adjacent laser positions, equivalent to 2000 laser shots. When comparing data acquired with 50 µm spacing between laser positions (parts b and d of Figure 2) with data acquired with 15 µm spacing (parts f and h of Figure 2), it can be seen that both signal intensity and signal-to-noise ratio are lower at 15 µm acquisitions. This observation is consistent with the difference in the area of the sample ablated, which was ∼11 times smaller when a 15 µm step size was used (area sampled has a squared relationship with the step size between laser (25) Hunnam, V.; Harvey, D. J.; Priestman, D. A.; Bateman, R. H.; Bordoli, R. S.; Tyldesley, R. J. Am. Soc. Mass Spectrom. 2001, 12, 1220–1225. (26) Pulfer, M.; Murphy, R. C. Mass Spectrom. Rev. 2003, 22, 332–364. (27) Merrill, A. H.; Sullards, M. C.; Allegood, J. C.; Kelly, S.; Wang, E. Methods 2005, 36, 207–224.

3668

Analytical Chemistry, Vol. 82, No. 9, May 1, 2010

positions, e.g., doubling the diameter of the sampled area will result in a quadrupling of the area). A square relationship is also observed between the laser step-size setting of microprobe imaging experiments, the number of data points per unit area, and acquisition time required. The large increase in instrument time and data storage resources that is required, as well as the large decrease in sensitivity at higher resolution, is an important consideration in experimental design and it is advisible to operate at the lowest spatial resolution acceptable. With the use of current technology, only highly abundant compounds can be detected at high spatial resolution. This can be advantageous when specific compounds of interest are highly abundant in small spatial volumes, e.g., one particular cell type only in a mixed cell population. In the experiments reported here, four prominent signals at m/z 738.5, 822.6, 848.6, and 850.6 were only observed in the Gaucher samples (cf. parts d and h of Figure 2). These signals stand out from other signals more clearly at the higher resolution, which can be explained based on the spatial distribution of these compounds. When a compound is confined to an area of tissue smaller than the area sampled at a laser position, the compound is effectively diluted relative to more evenly distributed species, therefore as the analysis area is decreased to a size similar to the size of the area to which the compound is confined, the compound becomes relatively more concentrated. Another effect that needs to be considered is that at each laser position a certain number of ions are required in order for a compound to be detected, as the area sampled is decreased; only compounds that are spatially confined still produce enough ions to be detected. As expected and previously reported,18 the concentration of GC was much higher in the conditional knockout type 1 Gaucher mouse model than in control samples, and in order to show an intensity distribution for the control samples, the intensity scale had to be rescaled by a factor of 10. With dependence on the fatty acid residue, LC-MS/MS experiments on the same samples revealed a 3.2-5.9-fold increase in the amount of GC in the diseased tissue (data not shown); with an uneven distribution across the tissue, the increase in individual cells was much greater than this. In the Gaucher model spleen sections, the distribution of GC was highly localized, whereas the ion intensity in the control samples was evenly distributed. This observation held true at both spatial resolution settings employed in these experiments. It was not clear from the MS data for the control sample whether GC was detected or whether the ion intensity at 822.6 Da was due to a shoulder on a peak at 822.4 Da. To answer this question, either higher mass resolution would be required or the use of an additional dimension of separation, such as MS/MS or IMS. Further work is planned to investigate both of the two latter options. The ability to produce effective laser desorption areas smaller than the laser beam using oversampling has been shown previously, and a 1:1 relationship between step-size and effective spot size has been demonstrated for step-sizes of 15 and 40 µm, respectively, using standard compounds overlaid with electron microscope grids as a reference pattern.16,17 It is important to consider what spatial resolution can really be achieved at a given laser step-size because the resolving power of a MALDI imaging experiment is affected by different factors, including instrument

Figure 2. MALDI MS images of control and Gaucher spleen sections showing false color ion intensity maps for m/z 738.6 (potassiated GC (16:0)) and representative MS spectra summed over 10 adjacent laser positions: (a and b) control sample, laser step-size 50 µm, (c and d) Gaucher sample, laser step-size 50 µm, prominent GC peaks are marked with asterisks, (e and f) control sample, laser step-size 15 µm, and (g and h) Gaucher sample, laser step-size 15 µm.

performance, sample preparation and matrix application, which can all result in loss of spatial information. Sample preparation and matrix application factors are not easily quantifiable not least because they are compound dependent. The instrumental aspect of spatial resolution can be more easily predicted and, assuming a perfect laser spot and absolute accuracy in the movement of the sample stage, would be predicted to be double the step-size, e.g., using a step-size of 15 µm, it should be possible to distinguish features that are 30 µm apart, with baseline separation. Here we have compared two different laser step-sizes, 15 and 50 µm, and

the advantage of using the smaller step-size can clearly be seen in a close-up view of the spleen tissue images from the Gaucher mice. The images are also compared with microscope images to show the size of feature that can be detected at each resolution setting (Figure 3). At 50 µm laser step-size, large features, such as a blood vessel, can be seen between areas of high GC ion intensity, which correspond to areas in the microscope image containing macrophages (Figure 3a,b). It is noteworthy that the distance between the areas of high GC bisected by the blood vessel is ∼100 µm, the ability to resolve this feature is in-line with Analytical Chemistry, Vol. 82, No. 9, May 1, 2010

3669

Figure 3. Comparisons of microscope images of Gaucher spleen sections with ion intensity maps for m/z 738.6 Da at different MALDI imaging resolutions: (a) microscope image of spleen section, H&E stained after imaging experiment, (b) MALDI MS image of tissue shown in part a, sampling step size 50 µm, (c) light microscope image of spleen section, prior to MALDI imaging analysis, and (d) MALDI MS image of tissue shown in part c, sampling step size 15 µm. The red squares shown in parts c and d are 30 µm × 30 µm and serve to illustrate experimental resolution. The red circles highlight a single GC containing macrophage cell.

the predicted instrumental resolving power limit of the imaging experiment with a 50 µm step-size. The resolving power of the 15 µm step-size imaging experiment would be predicted to be in the order of 30 µm. An image acquired at this step-size is shown in Figure 3d with a matching microscope image in Figure 3c. In order to illustrate the observed resolution, a red 30 µm × 30 µm box is shown in both pictures, positioned between two areas rich in GC containing macrophages. It is clearly possible to distinguish between features separated by ∼30 µm. This resolution makes it possible to see that storage of GC is primarily in macrophages. This observation was not unambiguously possible at the 50 µm step-size. It is worth noting that although a laser spot size of 15 µm is similar in size to macrophages in the Gaucher spleen (typical diameter of 20 µm), single macrophages cannot routinely be distinguished, when they are in close proximity to other macrophages. However, when a macrophage is bordered by other types of spleen cells, it can be imaged based on the difference in GC level between it and the surrounding tissue. One example of this is shown by the macrophage circled in Figure 3c,d. With the use of a laser step-size of 50 µm, it was possible to wash and H&E stain sections after imaging experiments had been performed on them. The stain was lighter in color than that of sections that had not been MALDI imaged, nevertheless it was possible to examine these sections under a light microscope (see Figure 3a). At a sampling step-size of 15 µm, tissues were degraded to such an extent that H&E staining was no longer possible; for this reason, microscope images acquired prior to MALDI MSI analysis had to be used for these sections (see Figure 3c).

spectrometer. We used an oversampling technique to obtain an effective sampling area of ∼2.5% of the laser spot size. With this reduced sampling area, high-quality MALDI images of GC, which is localized in macrophages in the conditional knockout Gaucher type 1 spleen tissue, were produced. Features greater than 30 µm could clearly be distinguished, including isolated macrophages. A phenomenon highlighted in this work is that chemical species that are locally abundant are more easily distinguished from less spatially confined species at higher spatial resolution. This occurs as they are only highly concentrated over a small part of the sample, and hence their signal is diluted by analysis at lower resolution.

CONCLUSIONS Here we demonstrate a method for high-resolution MALDI imaging using an unmodified commercially available MALDI mass

Received for review December 23, 2009. Accepted March 31, 2010.

3670

Analytical Chemistry, Vol. 82, No. 9, May 1, 2010

ACKNOWLEDGMENT We thank Dr. Lorne A. Clarke from University of British Columbia (Vancouver, Canada) for providing us with the conditional knock-out murine model of type 1 Gaucher disease. We also give our thanks to Kristian Brion, Helen Beard, and Sofie Hassiotis for their help with sample preparation. Also, we thank Richard Tyldesley-Worster from Waters Corporation (Manchester, United Kingdom) for modifying imaging software to facilitate acquisition of data at high resolution. SUPPORTING INFORMATION AVAILABLE Spatial distribution of GC species in the Gaucher spleen: laser step-size of MALDI imaging experiment is 50 µm (Figure S1, ac902393k_si_001.tif) and 15 µm (Figure S2, ac902393k_si_002.tif). This material is available free of charge via the Internet at http://pubs.acs.org.

AC902939K