Mouse brains used for tissue sections were flash frozen in liquid nitrogen immediately after dissection, and stored at −70 °C. The tissue was warmed to −15 °C, mounted onto a cutting chock with a mound of ‘Optimal Cutting Temperature Compound’ (OCT, Sakura Finetek, Torrence, CA) on the base of the tissue, sliced at 10 μm thickness with a cryostat (Leica CM 1900) , placed directly onto glass coverslips or MALDI plate inserts, and stored at −20 °C until matrix application and imaging.

A general description of the vacuum sublimation technique applied to this work is provided in an organic chemistry laboratory textbook [12]. The specific device used here to apply matrix to sectioned tissue samples was constructed over twenty years ago (Figure 1). A very similar device is commercially available (Sublimation apparatus 8023, Ace Glass, Vineland, NJ). A key component of the device was the flat bottom condenser to which the glass coverslips or MALDI plate inserts could be affixed using double sided thermally conductive tape (Product number 8865, 3M, St. Paul, MN). Matrix (300 mg) was added to the bottom section of the apparatus, the two pieces were assembled with an O-ring seal and connected to a direct drive vacuum pump (8 m³/hr). After 15 min at reduced pressure, the condenser was filled with cold water (15°C). After five additional min, heat was applied to the base of the sublimator via a heating mantle, the temperature was monitored using a thermocouple attached to the heating mantle, and time was measured from the onset of heat application. The sublimation of the solid matrix gradually progressed upon application of heat. The variable voltage regulator was set to 45 V and the temperature of the heating mantle increased gradually from 25 to 120 °C in 10 min. The deposition of matrix was visually observed to slowly move up the sides of the glassware to the level of the sample plate. Optimization of the amount of matrix that condensed onto the tissue was empirically determined by adjusting parameters of pressure, temperature at the condenser, amount of matrix, heat applied to matrix, and time of sublimation. A matrix layer that was too thin yielded very few ions and a matrix layer that was too thick yielded only matrix ions. Our optimal conditions were found to be 0.05 Torr pressure during sublimation, condenser temperature of 15°C, 45 V applied to the heating mantle, and 300 mg DHB as matrix added to the sublimator. A period of 10–11 min from the onset of heat application resulted in a thin uniform layer of matrix that yielded high quality images by MALDI mass spectrometry. After this time, the heat and condenser coolant were both removed, and the system was allowed to slowly come to room temperature while still under vacuum. Once equilibrated to room temperature, the system was brought to ambient pressure, opened, and the sample plate was carefully removed from the sublimation condenser and transferred to a MALDI plate holder for analysis by MALDI-TOF mass spectrometry.

The electrospray apparatus used for comparative matrix application consisted of a fixed position microspray needle, matrix solutions at 30 mg/mL made up in 1:1 H₂O: CH₃CN, 250 μL syringe, syringe pump set to flow at 1 μL/min, voltage of 1000 V applied to the spray needle. The sample plate was grounded, and moved mechanically at a rate of one mm/sec at a distance of 3 cm from the spray needle. Only plate regions with observably even electrospray deposition were used for imaging experiments.

Images were acquired using a quadrupole-TOF tandem mass spectrometer with an orthogonal MALDI source attached (QSTAR XL, Applied Biosystems/MDS Sciex, Thornhill, Ontario, Canada). MALDI mass spectra were obtained using standard conditions and 10 laser pulses from a nitrogen laser (337 nm) per image spot with a 0.5 sec accumulation time. Collisional activation of selected ions was carried out using relative collision energy of 40 volts with argon as collision gas. Mass spectrometric data were processed using software provided by Applied Biosystems/MDS Sciex at 10 units per amu, and images were visualized using BioMap software (Novartis, Basel, Switzerland).

The finely dispersed coating of the organic MALDI matrices that results from sublimation was visually evident on both the glass and metal surfaces used in this study. An example of this is shown in Figure 2A where DHB was applied by sublimation onto a sample section of mouse brain. Common matrices DHB, α-CHCA, and sinapinic acid were the only matrices tested at this time and all sublimed under reduced pressure (0.05 Torr) at temperatures between 120 °C and 140 °C without decomposition. An optimal coating of matrix was thick enough to yield abundant ions upon laser irradiation, yet sufficiently thin to utilize the spot overlap technique reported by Sweedler and coworkers [5]. The optimal amount of DHB (0.05 mg/cm²) was found by experiment to require 10 min of heating, at 0.05 Torr with the apparatus (Figure 1), and resulted in a layer between 4–5 μm thick. A close-up view of matrix deposited onto a glass coverslip is shown in Figure 2B. This image was captured with a confocal microscope using excitation light from an argon laser (488 nm) through a 63× objective.

A solution (100 μM) of a standard phospholipid compound 16:0/18:0-GPCho ([M+H]⁺ m/z 762.6) was applied to MALDI plates by robotic spotter. DHB was applied to one plate by ESI deposition and to another by sublimation. Images of these standard spots were acquired with 50μm × 50μm plate movement/pixel in positive ion mode and a comparison of these data is shown in Figure 3A–3D. The image of the GPCho standard with sublimed matrix (Figure 3A) was more uniform in appearance relative to the image of the GPCho standard with electrosprayed matrix (Figure 3C). The sublimed matrix standard also gave a significantly more intense signal (more ions upon laser irradiation), and fewer sodium adduct ions ([M+Na]⁺ m/z 784.6) when compared to the sample with ESI matrix deposition (Figure 3B and 3D).

The laser spot used in these imaging experiments was oval shaped with experimentally determined dimensions 400 × 200 μm. Sample plate movement at 50 μm × 50 μm/pixel with full ablation of the preceding spot resulted in a crescent shaped sampling area for each pixel with dimensions (assessed graphically) 250 μm × 150 μm at the edge of the laser spot where laser fluence was at fringe values. Furthermore, the pixelation of acquired data caused artifactual image broadening as in the example of a 50 μm analyte spot that might be analyzed with a single mass spectrum, or more likely from two consecutive spectra each with partial coverage of that 50 μm area. Therefore, although plate movement was controlled at 50μm × 50 μm, it was difficult to assess the image resolution based on applied instrument parameters.

A sagittal section (10 μm) of mouse brain was coated with DHB matrix by the sublimation method described and imaged in the positive ion mode by MALDI-TOF mass spectrometry. An average mass spectrum of the entire tissue region was calculated from the image database (Figure 4A). The mass range recorded in this image database was m/z 500–1100, but no ions with significant abundance appeared above m/z 900. In a separate experiment with this tissue slice, collisional activation of the most abundant ion observed at m/z 760.6 generated a product ion spectrum (Figure 4B) that supported the identity of this ion as a phosphocholine lipid, the molecular weight being consistent with 16:0a/18:1-GPCho, previously identified as a major glycerophosphocholine lipid component of rat and mouse brain [16,17]. The ion at m/z 826.6 in the region of the corpus callosum generated product ions upon collisional activation at m/z 767.6 (loss of 59 u, N(CH₃)₃) and m/z 163.1 (Figure 4C) indicative of a kaliated phosphocholine lipid adduct 18:1/18:1-GPCho [16]. Collisional activation of the ion at m/z 834.6 in the cerebellar region of the mouse brain tissue slice (Figure 4D) generated an abundant product ion at m/z 184.1 which supported its identity as 18:0a/22:6-GPCho.

A brain slice stained by Oil-Red-O (Figure 5A) taken 100 μm lateral to the imaged slice was used to evaluate the general anatomy of these sections analyzed by MALDI imaging along with a mouse brain atlas [18]. Mass spectrometric images of mouse brain (Figure 5B–D) appeared with high definition, texture, and discernable fine anatomical structure. The pattern delineated for the distribution of the ion at m/z 760.6 (Figure 5B) in a mouse brain sagittal section was clearly demarcated along the distinct subsections of the tissue when compared to Oil-Red-O stained tissue slice particularly in the region of the corpus callosum, striatum, and cerebellum. There was an appearance of mottling in the striatum region just beneath the corpus callosum, and underneath this was a long j-shaped region highlighting greater intensity of the ion at m/z 760.6. This region, the corpus callosum and the pyramidal tract at the base of the brain were highlighted specifically by the ion at m/z 826.6 (Figure 5C). These regions correspond to myelinated nerve fiber bundles connected one region of the brain to another. The prevalent appearance of specific phospholipids in defined regions was especially notable in the image generated by the ion at m/z 834.6 that is abundant in the cerebellum. The apparent increase of these phospholipid molecular species related to anatomical features of the mouse brain is under further investigation.

The sublimation technique for matrix application has several notable features. It was very reproducible, relatively inexpensive, and generated a high purity organic matrix directly applied onto the sample. How the gaseous matrix mixes with the phospholipids on the tissue surface is not clear and requires further study. The microcrystalline and highly pure nature of the matrix clearly enhanced the MALDI process for phospholipids and created good image definition when applying partial spot ablation with sequential plate movements. Also, the potential for spreading of analyte molecules due to solvents used in spray deposition was eliminated with this method. Matrix can be sublimated onto a variety of surfaces; glass coverslips, metallic tape, and stainless steel MALDI inserts were all used successfully in this initial work.