Localization and imaging of gangliosides in mouse brain tissue sections by laserspray ionization inlet

Abstract

A new ionization method for the analysis of fragile gangliosides without undesired fragmentation or salt adduction is presented. In laserspray ionization inlet (LSII), the matrix/analyte sample is ablated at atmospheric pressure, and ionization takes place in the ion transfer capillary of the mass spectrometer inlet by a process that is independent of a laser wavelength or voltage. The softness of LSII allows the identification of gangliosides up to GQ1 with negligible sialic acid loss. This is of importance to the field of MS imaging, as undesired fragmentation has made it difficult to accurately map the spatial distribution of fragile ganglioside lipids in tissue. Proof-of-principle structural characterization of endogenous gangliosides using MSⁿ fragmentation of multiply charged negative ions on a LTQ Velos and subsequent imaging of the GD1 ganglioside is demonstrated. This is the first report of multiply charged negative ions using inlet ionization. We find that GD1 is detected at higher levels in the mouse cortex and hippocampus compared with the thalamus. In LSII with the laser aligned in transmission geometry relative to the inlet, images were obtained in approximately 60 min using an inexpensive nitrogen laser.

Gangliosides are a class of glycosphingolipids (GSL) composed of an oligosaccharide chain anchored to a hydrophobic ceramide base. They are identified by the presence of at least one sialic acid in their sugar chain (Fig. 1) (1, 2). Gangliosides are found in all eukaryotic cells, primarily as plasma-membrane constituents in which they can segregate into lipid rafts (3, 4). The central nervous system contains high concentrations of gangliosides that participate in cell-cell interactions and regulate cell proliferation (5), differentiation (6, 7), recognition (8) and signaling (9). In biological systems, gangliosides comprise a variety of related structures (10). Heterogeneity comes from their oligosaccharide content, linkage, and other modifications, and from variations of the ceramide moiety, including chain length and degree of unsaturation and hydroxylation (11). The structure of both the ceramide and oligosaccharide can vary with species and environment and can change with development (12–14), aging (15, 16), and disease, making them important biomarkers (17, 18) of certain diseases, including Alzheimer's (19, 20), Tay-Sachs (1, 21, 22), and cancer (23–25). Because ganglioside species may have only minor structural variations and may be present in low abundance, sensitive methods are required for their analysis for both identification of their structures and location within normal and diseased tissue.

Fig. 1.

Structures of ganglioside species analyzed. GM1 (one sialic acid), GD1_a, GD1_b, and GD3 (two sialic acids), GT1_b (three sialic acids), and GQ1_b (four sialic acids).

Although the sensitivity of mass spectrometry (MS) is ideal for profiling of gangliosides, the fragile glycosidic bond linking sialic acid to their oligosaccharide chain, especially with multiple sialic acids, breaks during vacuum matrix-assisted laser desorption/ionization (MALDI), complicating accurate identification of these gangliosides (26–28). Dehydration and decarboxylation have also been reported (27). To compensate, derivitization of carboxylic acid has been used to stabilize sialic acid, making it less amenable to fragmentation (29, 30). Matrix selection has also been shown to affect fragmentation (31–33). High-pressure, vibrationally cooled MALDI reduces in-source fragmentation but requires instrument modifications (10, 28, 34). “Softer” vacuum MALDI techniques have been used, such as infrared (IR)-MALDI, which allow the analysis of GD1 with negligible fragmentation or salt adduction (35, 36). Atmospheric pressure (AP)-MALDI decreases the fragmentation and matrix cluster formation common to vacuum MALDI, although sialic acid loss is present in more complex gangliosides (37). Electrospray ionization (ESI) provides soft ionization, but is limited to analyzing gangliosides in solution (38, 39).

Several methods are available to determine the distribution of gangliosides in tissue. Monoclonal antibodies can detect differences in the hydrophilic oligosaccharide chain on the cell surface but not in the fatty alcohol or N-acyl carbon chains embedded in the lipid bilayer (40). MALDI imaging MS is an effective technique for visualizing the distribution of metabolites (41, 42), lipids (43–45), and proteins (46) in tissue, but MALDI imaging of gangliosides, as noted above, is limited because in-source or metastable fragmentation of polysialylated gangliosides results in the production of GM1, making differentiating the GM1 from fragment ions difficult and thus limiting the ability to assign the correct ganglioside species (31–33, 47–49). MALDI analysis and imaging of gangliosides containing multiple sialic acids is further complicated because these gangliosides form potassium and sodium adducts, reducing sensitivity and producing very complex mass spectra, which also adds difficulty in determining the correct ganglioside distribution (31). Despite these limitations, several groups have demonstrated the differential distribution of gangliosides in mouse brain tissue (31–33, 48, 49), offering precedence for imaging of gangliosides using MALDI as a desorption/ionization method. Mono-sialylated gangliosides, including GM2 and GM3, have been imaged by MALDI without fragmentation or salt adduction (32, 50). Other surface methods, including desorption electrospray ionization (DESI), are applicable to imaging, but the achievable spatial resolution is lower than the spatial resolution commonly attainable with MALDI imaging (51).

Laserspray ionization inlet (LSII) is a novel method (supplementary Fig. I-A) that combines a soft, ESI-like ionization with the high spatial resolution surface sampling capabilities of MALDI. In LSII, solid-state matrix/analyte mixtures are laser ablated at AP into the heated inlet capillary of a mass spectrometer. With the assistance of thermal energy and vacuum, ions are created inside the inlet capillary as the analyte moves through the pressure gradient from AP to vacuum. The observed positively charged ions are highly charged, similar to ESI, and examples include peptides, proteins, and synthetic polymers (52–55). LSII is a subset of matrix assisted inlet ionization (MAII; supplementary Fig. I-B), which operates independent of a voltage or a laser (56). MAII and LSII depend on the appropriate matrix (57–59), similar to MALDI (60–63), but do not require absorption at the laser wavelength employed. In LSII, the laser ablation only acts as the method to transfer a matrix/analyte sample into the inlet of the mass spectrometer, permitting the use of highly focused lasers to improve spatial resolution. Additionally, LSII often employs transmission geometry (TG) in which the laser is aligned 180° to the sample/orifice and ablates the analyte from the backside (supplementary Fig. I-C) (52, 64–66). This technique offers further improvements to spatial resolution and high-throughput acquisitions (65, 66), as only a single laser shot is required to obtain a mass spectrum. This is especially important for imaging mass spectrometry, in which LSII has recently been developed by Richards et al. (65) to image phospholipids and sulfatides from mouse brain tissue using solvent-free sample preparation, inlet ionization (supplementary Fig. I) at 450°C, and detection of the singly charged negative ions. Here, we show MSⁿ analysis and tissue imaging of fragile gangliosides producing multiply charged negative ions using LSII with inlet temperatures at 400°C or higher without the fragmentation or metal adduct formation common to MALDI.

MATERIALS AND METHODS

Materials

Matrixes 2,6-dihydroxyacetophenone (2,6-DHAP), 2,4,6-trihydroxyacetophenone (THAP), and 2,5-dihydroxybenzoic acid (2,5-DHB), and ganglioside standards from bovine brain GM1, GD1_a, GD1_b, and GT1_b were purchased from Sigma Aldrich Inc. (St. Louis, MO). Ganglioside standards from bovine brain GD3 and GQ1_b were purchased from Santa Cruz Biotechnology (Santa Cruz, CA). 2,5-dihydroxyacetophenone (2,5-DHAP) matrix and solvents acetonitrile, ethanol, and methanol were purchased from Fisher Scientific Inc. (Pittsburgh, PA). Matrixes and ganglioside standards were used without further purification.

Ganglioside standard sample preparation

Ganglioside standards GM1 and GD1 were dissolved in methanol to 1 pmol μl⁻¹. Standards GD3 and GT1_b were dissolved in methanol to 5 pmol μl⁻¹. Standard GQ1_b was dissolved in methanol to 10 pmol μl⁻¹. Saturated 2,5-DHAP, 5 mg ml⁻¹ 2,5-DHB, and 20 mg ml⁻¹ 2,4,6-THAP matrix solutions were prepared in acetonitrile-water (1:1; v/v). Saturated 2,6-DHAP matrix solution was prepared in ethanol-water (1:1; v/v) and in acetonitrile-water (1:1; v/v). For LSII MS analysis, 0.5 μl of ganglioside standard solution was spotted on to a glass microscopy slide (Gold Seal, Portsmouth, NH), followed by 1.5 μl of matrix solution. The ganglioside and matrix were then mixed on the glass slide while still wet. For MALDI MS analysis, ganglioside standards and matrix were premixed in a 1:1 v:v ratio. The dried droplet method (61) was used to spot 1 μl of the matrix/standard mixture onto the sample plate.

Tissue preparation

Animal care was in accordance with the US Public Health Service Policy on Humane Care and Use of Laboratory Animals. The mouse experimental studies were approved by the Institutional Animal Care and Use Committee of the University of Indiana, Bloomington. Mouse brain tissue was obtained from C57 B1/6 mice, 20 weeks old, that were euthanized with an overdose of isoflurane anesthesia and transcardially perfused with ice-cold 1× phosphate-buffered saline (150 mM NaCl, 100 mM NaH₂PO₄, pH 7.4) for 5 min to remove red blood cells. The brains were frozen at −22°C and sliced into 10 μm thick sections in sequence using a Leica CM1850 cryostat (Leica Microsystems Inc., Bannockburn, IL). The tissue sections were placed on prechilled glass microscopy slides that were precoated with 2,5-DHB matrix. 2,5-DHB was preground in a glass vial containing 1.3 mm chrome beads for 30 min at 15 Hz using a TissueLyser II ball mill device (Qiagen, Valencia, CA). The preground matrix was then applied to the glass slide using a previously published procedure (67). Briefly, 2,5-DHB was placed in the bottom compartment of the TissueBox with approximately 30 beads. The glass slide was separated from the matrix by 5 μm mesh. The TissueBox was placed in the TissueLyser II, where the movement of beads forced matrix through the mesh and onto the slide. The precoated slides were briefly warmed with a finger from behind to allow sections to relax and attach. To avoid water condensation, the tissue was stored at −20°C and transported under dry ice in an airtight, desiccant-containing box where it were kept until use. Before imaging, 300 mg of 2,5-DHAP was dissolved in 9 ml of acetonitrile-water (1:1; v/v) and applied to the tissue using an artistic airbrush.

Mass spectrometry and data processing

LSII-MS, MSⁿ, and imaging experiments were performed on a Thermo Fisher Velos LTQ (Thermo Fisher Scientific, Bremen, Germany) linear ion trap mass spectrometer. The instrument was operated in negative ion mode using a mass range mass-to-charge (m/z) 500–2400, depending on the molecular weight of the ganglioside. Similar to previous work (65), the source housing was removed and the interlocks were overridden to allow unobstructed access to the ion entrance orifice. A 337 nm nitrogen laser (Newport Corporation, Irvine, CA; VSL-337ND-S) beam aligned 180° from the entrance of the mass spectrometer and focused on the ion transfer capillary using a 102 mm focal length lens (CVI Melles Griot, Albuquerque, NM) was used to ablate the sample directly into the mass spectrometer. The laser fluence per pulse was approximately 0.5–1.0 J cm⁻².

For LSII analysis of ganglioside standards, samples were placed approximately 1–2 mm from the mass spectrometer vacuum entrance (inlet) and moved manually through the laser beam. Each spectrum represents one laser shot. The ion trap fill time was set at 50 msec. The sheath gas flow rate, auxiliary gas flow rate, and sweep gas flow rate were all set to 0 (arbitrary units). Spray voltage was set to 0 kV. The capillary temperature was 400 to 450°C when using 2,5-DHB. A capillary temperature of 400°C was used with 2,5-DHAP matrix. Peak assignments and structural information were obtained using collision-induced dissociation (CID) LSII MSⁿ. For ganglioside standards, an isotopic width of 0.7 and collision energies of 25–33 eV were used. For tissue, an isotopic width of 1.0 and a collision energy of 40 eV were used.

For imaging experiments, the glass slide mounted with mouse brain tissue was attached to a computer-controlled xyz stage (Newmark Systems, Mission Viejo, CA) using a custom-made sample holder and moved through the laser beam in TG mode. Lanes were spaced at 100 μm intervals, and each lane was 10.2 mm long. The tissue was moved through the laser at a speed of 100 μm sec⁻¹. The laser was fired only once at each position on the tissue; each pixel of the image represents a single laser shot. The ion trap fill time was set to 125 msec. Gangliosides were detected using 2,5-DHAP matrix with inlet temperatures of at least 400°C. Tissue images were created using BioMAP 3.7.5.6 (Novartis Institutes for BioMedical Research, Basel, Switzerland). Thermo XCalibur .RAW files were converted to .IMG files using customized imaging software.

Intermediate pressure (0.16 Torr) MALDI MS was performed on a SYNAPT G2 mass spectrometer (Manchester, UK) equipped with a Nd:YAG laser (355 nm). The instrument was operated in negative ion sensitivity mode. A laser fluence of 250–300 (arbitrary units) and a 200 Hz firing rate were used. The sensitivity mode settings were 0 V for the sample plate, 10 V “extraction,” 10 V “hexapole bias,” and 5 V “aperture 0.” Approximately 20 laser shots were summed for each spectrum. LSII-MS was performed on the SYNAPT G2 by converting the nano-ESI source to a LSII source according to published procedure (54–59, 68) using the 337 nm nitrogen laser employed for LSII-MS on the LTQ Velos.

RESULTS AND DISCUSSION

Method development of LSII-MS and MSⁿ using ganglioside standards

Standard gangliosides were used to establish and optimize the LSII MS, MSⁿ, and imaging conditions for detection of gangliosides directly from tissue. Matrixes, 2,5-DHB and 2,5-DHAP, were tested on purchased samples of GM1, GD1_a and GD1_b, GD3, GT1_b, and GQ1_b. As previously reported, the temperature of the ion entrance orifice to the mass spectrometer is an important LSII parameter in both positive and negative ion mode detection (65). Ion abundance in LSII is increased by heating the ion transfer capillary (53, 68–70). Using 2,5-DHB on a LTQ Velos mass spectrometer, GD1 could only be analyzed at capillary temperatures between 400°C and the maximum setting of 450°C in the negative ion mode. The negatively charged ions at m/z 1836.8 (d18:1-18:0) and 1864.8 (d20:1-18:0) were identified as [GD1-H]^- (Fig. 2A). Low abundance fragment ions corresponding to the loss of one sialic acid are observed at m/z 1544.9 and 1572.9. To confirm the negligible fragmentation with LSII, a mass spectrum of GD1 was acquired from m/z 150 to 2000. An inset of the lower mass region (supplementary Fig. II) reveals predominantly matrix clusters. The sialic acid fragment, which would appear at m/z 290, is not observed. At capillary temperatures below 400°C, only matrix aggregates are present, and GD₁ is not detected (supplementary Fig. III). The more complex GT1_b and GQ1_b, which contain up to four sialic acids, could not be analyzed using 2,5-DHB. We speculate this is due to the high temperature requirements of 2,5-DHB when used as a LSII matrix (53, 65, 70, 71).

Fig. 2.

Negative mode MS analysis of GD₁ ganglioside. Mass spectra: (A) LSII with 2,5-DHB matrix; (B) LSII with 2,5-DHAP matrix; and (C) MALDI with 2,5-DHB matrix.

A unique feature of LSII is the ability to selectively produce multiply charged ions or singly charged ions depending on the sample preparation and voltage application (52, 55, 72). 2,5-DHB produces only singly charged negative ions of GD1 (Fig. 2A), whereas 2,5-DHAP primarily produces doubly charged negative ions (Figs. 2B, 3). Interestingly, 2,5-DHAP produces negligible or no fragment ions. The absence of fragmentation with 2,5-DHAP matrix may be the result of lower thermal requirements in LSII (53, 72). We are currently investigating mechanistic arguments that may assist in providing an explanation for these observations (58, 59). The mass spectrum of purchased GD1 using 2,5-DHAP as matrix is shown in Fig. 2B obtained using 400°C inlet capillary temperature. However, molecular ions were observed between 100°C and 450°C for GD1 using this matrix (supplementary Fig. IV). The signals at m/z 917.4 and m/z 932.0 represent the doubly charged intact deprotonated GD1 (d18:1-18:0) and GD1 (d20:1-18:0), respectively. Singly deprotonated ions were also detected for GD1 at m/z 1836.9 (d18:1-18:0) and 1865.0 (d20:1-18:0). Further, although GD1 is present as the doubly deprotonated ion, the signals corresponding to sialic acid loss occur as low abundant singly deprotonated ions at m/z 1,545.5 and 1,573.5. The sialic acid fragment at m/z 290 is not present (supplementary Fig. V), again confirming minimal fragmentation with LSII.

For comparison, the commercial intermediate pressure MALDI source on the SYNAPT G2 was used for the analyses of a number of ganglioside standards. Fig. 2C shows the MALDI results for GD1 using 2,5-DHB as a matrix. The intact singly deprotonated ions were detected at m/z 1836.06 and 1864.06, but they are accompanied by adduction and fragmentation peaks, most of which were in higher abundance than the deprotonated intact ion. Salt adduction is present at m/z 1859.06, 1874.04, and 1886.04. The dehydrated ions at m/z 1818.02 and 1847.05 tentatively identified as [GD1(d18:1-18:0)-H₂O-H]⁻ and [GD1(d20:1-18:0)-H₂O-H]⁻, respectively, are more abundant than the parent ions. The dominant ions at m/z 1544.92 and 1572.96 arise from sialic acid loss, and they appear at similar m/z values as those obtained for MALDI of GM1 (supplementary Fig. VI). MALDI acquisitions were also obtained with 2,5-DHAP as the matrix, and the resulting mass spectra contained identical m/z signals to those obtained with 2,5-DHB but with overall lower abundances (supplementary Fig. VII-A). To demonstrate that the observed fragmentation was a consequence of MALDI and not simply caused by instrumentation differences relative to the LTQ Velos, supplementary Fig. VII-B shows an LSII acquisition of GD1 with 2,5-DHAP on the SYNAPT G2 using the nano-ESI source converted to a LSII source (54, 55, 57–59, 68). Abundant [M-2H]²⁻ ions are observed without adduction or fragmentation.

The soft LSII analysis of gangliosides other than GD1 is demonstrated in Fig. 3. In Fig. 3A, the intact singly deprotonated GM1 ions at m/z 1545.1 (d18:1-18:0) and m/z 1573.1 (d20:1-18:0) were the only abundant signals. GD3 analysis in Fig. 3B resulted in intact singly deprotonated ions at m/z 1442.5 (d16:1-18:0), 1470.8 (d18:1-18:0), and 1498.8 (d20:1-18:0), as well as singly and double acetylated GD3 gangliosides at m/z 1512.8, 1540.8, and 1554.8, respectively. The corresponding doubly deprotonated ions were observed at m/z 720.9, 734.9, 748.9, 755.9, 769.9, and 777.0. In Fig. 3C, analysis of the trisialylated GT1_b resulted in intact doubly deprotonated ions at m/z 1063.0 (d18:1-18:0) and 1077.0 (d20:1-18:0). Additionally, sodium and potassium replacement of an acidic hydrogen was detected at m/z 1088.0 and 1095.9. Finally, the analysis of GQ1_b, a ganglioside with four sialic acids, produced the intact doubly deprotonated ions at m/z 1208.5 (d18:1-18:0) and m/z 1222.4 (d20:1-18:0). The species with sodium and potassium replacing and acidic hydrogen were observed at m/z 1227.4, 1233.4, and 1241.4, respectively. A low abundance, triply deprotonated signal was present at m/z 814.9. According to the supplier information, GT1_b and GQ1_b are provided as salts (73), and it is our belief that the metal cation adduction was not from the LSII process but from the extract itself, as no other adduction is present.

Fig. 3.

Negative mode LSII MS analysis of gangliosides with 2,5-DHAP matrix. Mass spectra of (A) GM1; (B) GD3; (C) GT1_b; and (D) GQ1_b.

Astonishing differences were observed between LSII detecting intact parent ions in high abundance (Fig. 2A, B) and MALDI detecting a notable degree of fragment ions (Fig. 2C). MALDI, using the matrix 2,5-DHB and the commercial intermediate pressure source of the SYNAPT G2, of GD3 (supplementary Fig. VIII), GT1_b (supplementary Fig. IX), and GQ1 (supplementary Fig. X) produced intact parent ions, along with extensive fragmentation and adduction patterns. Only GM1 (supplementary Fig. VI) did not display fragmentation in MALDI. As expected, the desialylated lipids appeared at the same m/z as the MALDI signal from GM1. MALDI fragmentation caused the intensity of GM1 to be artificially high and hindered any attempts at relative or spatially distributed quantification of this lipid. 2,6-DHAP, which has been used as a matrix for MALDI analysis of gangliosides (32, 33), and 2,4,6-THAP were also tested for LSII analysis of gangliosides, but they yielded poor results (supplementary Fig. XI). Some fragmentation was observed with LSII but at intensity levels greater than an order of magnitude less than the parent ions. We conclude that the “softness” of LSII circumvents fragmentation issues similar to ESI.

To assess the potential for structural assignments of gangliosides, LSII CID was performed. Fragments were assigned according to the convention set by Domon and Costello (74) and Ann and Adams (75). Supplementary Fig. XII shows the MS² to MS⁶ spectra of the doubly deprotonated GD1_b ion at m/z 917.7 with an isolation window of 0.7 and collision energy of 25–33 eV. GD1 exists as a mixture of structural isomers, GD1_a and GD1_b, which differ in the location of their sialic acids (Fig. 1). CID can differentiate between the isomers based on fragmentation patterns that are similar to ESI from these lipids from solution (76, 77). For example, in the MS² to MS⁴ spectra, major signals came from the sequential loss of oligosaccharide chain moieties at m/z 1544.8, 1253.7, 1091.9, 888.7, 726.6, and 563.2. The abundant signal at m/z 581.2 was indicative of the GD1_b isomer, as it corresponds to the loss of two attached sialic acids not present in the GD1_a isomer. MS⁵ and MS⁶ scans were able to reveal information about the ceramide chain.

LSII imaging of gangliosides was examined using a model system of GM1 and GD1 ganglioside standards (supplementary Fig. XIII) that were mixed with 2,5-DHAP matrix and spotted side by side on a glass slide, with GM1 on the left and GD1 on the right. The spots were imaged using LSII with the laser aligned in TG using the homebuilt source and by MALDI with the laser aligned in reflection geometry using a commercial intermediate pressure MALDI source on the SYNAPT G2 mass spectrometer. The images were created for m/z 1572.8, corresponding to GM1 (d20:1-18:0) and also the undesired fragment ion of GD1. In the LSII image, m/z 1572.8 is located almost exclusively on the sample spot containing the GM1 standard (supplementary Fig. XIII-B, left side of the images), suggesting usefulness for accurate ganglioside imaging. With MALDI, m/z 1572.8 appears on both sides of the image, although GM1 was only spotted on the left side (supplementary Fig. XIII-A), making the distinction between GM1 and the fragment of GD1 difficult. The performance of 2,5-DHAP matrix in negative mode has been noted (37, 62, 65), but because it sublimes in vacuum MALDI (78), its use in imaging has been restricted. The lack of fragmentation and sodiation in LSII using this and other matrixes relative to MALDI analysis of purchased standards provides a basis for tissue-bound ganglioside analysis.

Structural characterization and imaging of endogenous gangliosides directly from mouse brain tissue using LSII-MS and MSn

In the initial structural characterization and imaging studies of endogenous gangliosides directly from tissue, 2,5-DHAP was used, as less fragmentation is observed with this matrix compared with 2,5-DHB. Because ablation is at AP with LSII, matrix sublimation is not an issue. A single LSII acquisition obtained from one laser shot penetrating the mouse brain tissue in TG shows many different lipids in the mass spectrum (Fig. 4A). Abundant signals at m/z 700.5, 766.6, 790.6, and 862.7 were assigned to various phospholipids and sulfatides. Doubly charged GD1 ions were detected at m/z 917.5 and 931.5. The isotope distributions were consistent with a doubly charged ion as shown for the m/z 917.5 ion in Fig. 4B. Fig. 4C shows low abundant signals at m/z 1544.8 and 1572.8 that were assigned to the singly charged [GM1-H]⁻ ions.

Fig. 4.

Negative mode LSII analysis of gangliosides in mouse brain tissue. (A) Single laser shot mass spectrum from mouse brain tissue with 2,5-DHAP matrix. (B) Inset of the doubly deprotonated GD1 ganglioside at m/z 917.5 (d18:1/C18:0). (C) Inset shows GM1 ganglioside present in low abundance, suggesting negligible fragmentation.

Although GD1 is the most abundant ganglioside in the brain (79, 80), in MALDI mass spectra of tissue, the signal corresponding to [GM1-H]⁻ is frequently the most abundant ganglioside signal, possibly due to GD1 fragmentation. With LSII, the GM1 species is present in much lower abundance than the GD1 species (Fig. 4), suggesting that the distribution of tissue gangliosides revealed by LSII more likely reflects their authentic distribution. Although LSII is able to reduce undesired in-source fragmentation, lower abundant gangliosides, including the O-acetylated species and the GQ1 species imaged by Colsch et al., are yet not detected in tissue (32). This may be the result of selective ionization in LSII using 2,5-DHAP or may be because the sensitivity of LSII, similar to AP-MALDI, is currently lower than vacuum MALDI.

LSII CID MSⁿ can be used to confirm the ganglioside structures ablated directly from mouse brain tissue. Because of the abundant multiply charged ions produced by LSII using 2,5-DHAP, linkage information can be obtained from gangliosides from tissue without employing extraction techniques. For example, LSII MS² was used to identify the GD1_b isomer directly from tissue (supplementary Fig. XIV-A). The doubly charged molecular ion at m/z 917.5 selected for CID fragmentation shows signals at m/z 1544.5, 1253.5, 1091.8, 888.8, and, in low abundance, 726.6 and 563.5, which correspond to the loss of successive sugar groups from the oligosaccharide chain and are in agreement with the results obtained from the GD1_b standard (supplementary Fig. XII-A). The characteristic [sialic acid-H]⁻ fragment is present at m/z 290. These signals are in agreement with CID of GD1 using ESI (32, 81–83). Several fragments identify the GD1_b isomer, particularly m/z 581.3, which corresponds to the loss of two attached sialic acids. Lower intensity fragment ions, including m/z 1382.9, the loss of the two end group sugars (sialic acid and galactose), and m/z 1161.8, the loss of the end group sialic acid, and galactose and the attached GalNAc, also suggest the GD1_b isomer. For further confirmation, MSⁿ was compared against results obtained for the GD1_b standard (supplementary Fig. XII). Not only were the tissue-bound lipids identified by identical MS/MS fragmentation compared with purchased standards, single laser ablation events determined the existence of the GD1_b isomer based on the presence of signals corresponding to a sialic acid dimer, the end group sugars, and the GalNac group (82). For comparison, CID of GD1_a from tissue is included in supplementary Fig. XIV-B. Targeted tissue analysis is complicated by isobaric and isomeric species that are often remedied with FT-ICR-MS and CID analysis (84, 85). However, LSII can detect gangliosides with enough abundance to confirm identification using MSⁿ with less expensive and simpler mass analyzers as is shown here for the linear ion trap on the LTQ-Velos.

In a proof-of-principle imaging study (supplementary Fig. I-C), LSII was used to image GD1 species directly from mouse brain tissue. Because of the low resolution of the LTQ Velos and to ensure only the doubly deprotonated GD1 species were imaged, imaging experiments were conducted in zoom mode to provide higher mass resolution. 2,5-DHAP was used as matrix and applied on top of the tissue using the spray coating approach (86). Improved results were obtained by combining spray coating with the precoated approach (64, 66, 87), in which the matrix was also placed below the tissue. The enhanced ionization is rationalized by the extraction of lipids into the matrix from both surfaces of the 10 μm thick tissue section. The laser aligned in TG ablated both matrix layers for ionization. As is the case with MALDI spray coating approaches (88, 89), the slightly wet matrix may assist in extracting the analyte from the ablated tissue into the matrix. Matrixes such as 2,5-DHB and 2,5-DHAP showed usefulness for precoating the mouse brain tissue. Images of [GD1-2H]²⁻ are shown in Fig. 5 for m/z 917.5 (d18:1-18:0) and m/z 931.5 (d20:1-18:0). In the tissue imaging experiments, each pixel of the image represents a single laser shot. The single shot per pixel allows high spatial resolution images (∼20 μm) obtained in approximately 1 h. Although both species of GD1 are located throughout the brain, they are more abundant in the hippocampus and cortex than in the underlying thalamus. The distribution of the two forms differed slightly, with GD1 at m/z 917.5 (d18:1-18:0) present in higher abundance than m/z 931.5 (d20:1-18:0), as also observed in Fig. 4.

Fig. 5.

TG LSII imaging mass spectrometry of (A) section from the Allen Brain Atlas (93) corresponding approximately to the experimental section. The scale bar is 1001 μm, upper-right corner. The red highlighted area corresponds approximately to the molecular layer. (B) Experimental section of a 10 μm thick mouse brain slice. (C) [M-2H]²⁻ ion images of GD1 at (C.1) m/z 917.5 and (C.2) m/z 931.5.

Negative ion MSⁿ successfully characterized fragile ganglioside lipids directly from mouse brain tissue aided by the “softness” of ionization using LSII and the propensity to form multiply charged ions. The potential for imaging these lipids in mouse brain tissue is demonstrated for doubly charged negative ions for GD1. The near absence of fragmentation of fragile gangliosides and the lack of metal adduct ions when using LSII aids analyses directly from tissue in which time-consuming separation methods are not possible. This is in contrast to MALDI analyses of these compounds in which special conditions are necessary to reduce fragmentation and adduction. Multiply charged ions produced by ESI can be detected with high sensitivity (90). Our hypothesis is that with proper thermal and vacuum conditions, equivalent or better sensitivities can be achieved in inlet ionization, as is supported by recent work using solvent-assisted inlet ionization (supplementary Fig. I-D) of 0.7 ng of protein digest on column (91). We further hypothesize that shorter acquisition times will simplify convoluted mass spectral data relative to sample complexity and charge state distributions associated with the direct analysis of tissue material (66) and that deconvolution programs will assist in the interpretation of LSII ions similar to those used in ESI (92). Besides addressing pressing needs for enhanced desolvation of the matrix in inlet ionization through instrument modifications, current research aimed at imaging gangliosides includes the search for improved matrix materials for ease in desolvation/ionization at lower inlet temperatures; improving sample preparation techniques, particularly matrix preparation and application, to visualize lower abundance gangliosides such as GT1 and GQ1 from tissue while enhancing sensitivity and spatial resolution; shortening analyses time (<10 min for one mouse brain tissue section) (57); and enhancing dynamic range by expanding applications to LSII-ion mobility spectrometry-MS (54, 55, 59, 68) and MSⁿ imaging for the differentiation of ganglioside isomers.