Elution, Partial Separation, and Identification of Lipids Directly from

Anal. Chem. 2010, 82, 8866–8874

Elution, Partial Separation, and Identification of Lipids Directly from Tissue Slices on Planar Chromatography Media by Desorption Electrospray Ionization Mass Spectrometry Justin M. Wiseman* and Jessica B. Li Prosolia, Inc., Indianapolis, Indiana 46202, United States A method for the examination of intact tissue sections for gangliosides and other lipids using desorption electrospray ionization (DESI) mass spectrometry (MS) is presented. In the present work, thin tissue slices (16 µm) taken from the rat brain are thaw mounted onto planar chromatographic media and the lipids are eluted, partially separated, and analyzed directly from the plate by DESIMS in the negative ion mode. With the lanes scanned parallel to the direction of the chromatographic separation in the full scan mode, the selected ion current associated with ions of separated lipid molecules is plotted in order of increasing Rf values. Distinctly different classes of lipids are detected using this method, including several ganglioside species (i.e., GQ1, GT1, GD1, and GM1) and sulfoglycosphingolipids. For the examination of gangliosides in the full scan negative ion mode from highperformance thin-layer chromatography (HPTLC) plates, the limit of detection (LOD) was determined to be approximately 3 pmol. Tandem mass spectrometry (MS/MS) using the linear ion trap was used to confirm the presence of selected gangliosides and other lipids directly from the HPTLC plate. DESI-MS/MS revealed the presence of both the GD1a and GD1b isomers. The simplicity of this approach where planar separations are relied upon for sample preparation and presentation to the MS should allow for the examination of a variety of complex samples including the rapid examination of foodstuffs, bacteria, whole blood, and needle biopsies for cancer diagnostics. Gangliosides are acidic glycosphingolipid (GSL) species having oligosaccharide head-groups containing one or more sialic acid residues attached to a hydrophobic ceramide chain that anchors in the outer monolayer of the plasma cell membrane of all vertebrate cells. The structures of gangliosides are diverse, comprising various combinations of saccharides and sialic acid. The ceramide moiety also varies with respect to chain lengths of the long chain base (sphingosine) and the fatty acyl tail. Gangliosides are particularly abundant in the central nervous system (CNS) and have a variety of cellular functions such as providing structural integrity to the cell membrane, serving as antigens, receptors for bacterial toxins, mediators of cell-cell adhesion, and * To whom correspondence should be addressed. Prosolia, Inc. 351 West 10th Street, Suite 316, Indianapolis, IN 46202.

8866

Analytical Chemistry, Vol. 82, No. 21, November 1, 2010

as mediators and modulators of signal transduction.1-4 Thus, the development of new, sensitive, and widely applicable analytical techniques for the analysis of GSL species is of particular interest. Thin-layer chromatography (TLC) is a simple and robust technique and is routinely employed for the separation and analysis of complex lipid mixtures. Currently, the most common approach for analyzing TLC plates by mass spectrometry (MS) involves scraping-off the identified bands and extracting the analytes from the silica gel, followed by further purification of the sample. This procedure is inherently time-consuming and poses the risk of losing material through the many sample transfer and extraction steps. As such, the direct analysis of separated analytes on TLC plates without further preparation is highly desirable.5,6 Direct mass spectrometric approaches for analyzing TLC plates in vacuo have been previously investigated and their reports date back to the early 1980-1990s with secondary ion mass spectrometry (SIMS)7,8 and fast atom bombardment (FAB).9 The capacity for direct examination of TLC plates by matrix-assisted laser desorption/ionization (MALDI) has been demonstrated in a numberofreportsutilizingbothinfraredandultravioletradiation.10-16 While this approach was successful, the method requires sophisticated instrumentation to preserve the native structures of labile (1) Hakomori, S. Curr. Opin. Hematol. 2003, 10, 16–24. (2) Hakomori, S.; Handa, K. Methods Enzymol. 2003, 363, 191–207. (3) Ramirez, O. A.; Gomez, R. A.; Carrer, H. F. Brain Res. 1990, 506, 291– 293. (4) Kotani, M.; Terashima, T.; Tai, T. Brain Res. 1995, 700, 40–58. (5) Busch, K. L. Thin Layer Chromatography Coupled to Mass Spectrometry. In Handbook of Thin Layer Chromatography; Sherma, J., Freid, B., Eds. Marcel Dekker: New York, 1990; pp 241-272. (6) Busch, K. L. J. Chromatogr., A 1995, 692, 275–290. (7) Unger, S. E.; Vincze, A.; Cooks, R. G.; Chrisman, R.; Rothman, L. D. Anal. Chem. 1981, 53, 976–981. (8) Dunphy, J. C.; Busch, K. L. Talanta 1990, 37, 471–480. (9) Chang, T. T.; Lay, J. O.; Francel, R. J. Anal. Chem. 1984, 56, 109–111. (10) Ivleva, V. B.; Sapp, L. M.; O’Connor, P. B.; Costello, C. E. J. Am. Soc. Mass Spectrom. 2005, 16, 1552–1560. (11) Dreisewerd, K.; Muthing, J.; Rohlfing, A.; Meisen, I.; Vukelic, Z.; PeterKatalinic, J.; Hillenkamp, F.; Berkenkamp, S. Anal. Chem. 2005, 77, 4098– 4107. (12) O’Connor, P. B.; Budnik, B. A.; Ivleva, V. B.; Kaur, P.; Moyer, S. C.; Pittman, J. L.; Costello, C. E. J. Am. Soc. Mass Spectrom. 2004, 15, 128–132. (13) Mehl, J. T.; Hercules, D. M. Anal. Chem. 2000, 72, 68–73. (14) O’Connor, P. B.; Mirgorodskaya, E.; Costello, C. E. J. Am. Soc. Mass Spectrom. 2002, 13, 402–407. (15) O’Connor, P. B.; Costello, C. E. Rapid Commun. Mass Spectrom. 2001, 15, 1862–1868. (16) Meisen, I.; Distler, U.; Muthing, J.; Berkenkamp, S.; Dreisewerd, K.; Mathys, W.; Karch, H.; Mormann, M. Anal. Chem. 2009, 81, 3858–3866. 10.1021/ac1016453  2010 American Chemical Society Published on Web 10/13/2010

oligosaccharide units and the addition of a matrix substance to enhance the desorption and ionization of the analyte molecules. Alternative methods developed independently by Van Berkel et al.17,18 and Luftmann et al.19,20 enabled solvent extraction of TLC bands directly from the TLC plate in which the effluent is subsequently directed into an electrospray ion source for introduction in the mass spectrometer. The advantage of these approaches is the absence of sample preparation, although carry-over from spot-to-spot can be problematic. Evidenced by these more recent examples and the introduction of commercial systems, the interest in coupling planar separations with mass spectrometry is rapidly expanding and may soon be mainstream. The ambient mass spectrometric desorption/ionization (DI) methods21 permit the analysis of surfaces at atmospheric pressure under ambient conditions. Recent reviews on the technical aspects of the various ambient or atmospheric pressure surface sampling techniques are available in the peer-reviewed literature.22,23 An intrinsic property of each of the ambient DI methods is their suitability to high-throughput chemical analysis, a characteristic that is derived from the ease of sample preparation. Desorption electrospray ionization (DESI)24 MS is one such method suitable for making direct measurements of surfaces under ambient conditions. Applications of DESI have been demonstrated in a variety of examples to date, including the analysis of intact tissues for targeted analytes such as drugs,25 alkaloids,26 lipids,27-30 and glycosides.31 The quantitative capacity of DESI when coupled to ion trap32 and triple quadrupole33 instruments has been demonstrated for select drug compounds in plasma33 and dried blood spots.34 The combination of TLC and DESI has also been demonstrated in several examples, including for the analysis of

(17) Van Berkel, G. J.; Sanchez, A. D.; Quirke, J. M. Anal. Chem. 2002, 74, 6216–6223. (18) Kertesz, V.; Ford, M. J.; Van Berkel, G. J. Anal. Chem. 2005, 77, 7183– 7189. (19) Luftmann, H. Anal. Bioanal. Chem. 2004, 378, 964–968. (20) Luftmann, H.; Aranda, M.; Morlock, G. E. Rapid Commun. Mass Spectrom. 2007, 21, 3772–6. (21) Cooks, R. G.; Ouyang, Z.; Takats, Z.; Wiseman, J. M. Science 2006, 311, 1566–1570. (22) Van Berkel, G. J.; Pasilis, S. P.; Ovchinnikova, O. J. Mass Spectrom. 2008, 43, 1161–1180. (23) Weston, D. J. Analyst 2010, 135, 661–668. (24) Cooks, R. G.; Ouyang, Z.; Takats, Z.; Wiseman, J. M. Science 2006, 311, 1566–1570. (25) Wiseman, J. M.; Ifa, D. R.; Zhu, Y.; Kissinger, C. B.; Manicke, N. E.; Kissinger, P. T.; Cooks, R. G. Proc. Natl. Acad. Sci. U.S.A. 2008, 105, 18120–18125. (26) Talaty, N.; Takats, Z.; Cooks, R. G. Analyst 2005, 130, 1624–1633. (27) Manicke, N. E.; Wiseman, J. M.; Ifa, D. R.; Cooks, R. G. J. Am. Soc. Mass Spectrom. 2008, 19, 531–543. (28) Wiseman, J. M.; Ifa, D. R.; Song, Q.; Cooks, R. G. Angew. Chem., Int. Ed. 2006, 45, 7188–92. (29) Dill, A. L.; Ifa, D. R.; Manicke, N. E.; Costa, A. B.; Ramos-Vara, J. A.; Knapp, D. W.; Cooks, R. G. Anal. Chem. 2009, 81, 8758–64. (30) Dill, A. L.; Ifa, D. R.; Manicke, N. E.; Ouyang, Z.; Cooks, R. G. J. Chromatogr., B: Anal. Technol. Biomed. Life Sci. 2009, 877, 2883–2889. (31) Jackson, A. U.; Tata, A.; Wu, C.; Perry, R. H.; Haas, G.; West, L.; Cooks, R. G. Analyst 2009, 134, 867–874. (32) Ifa, D. R.; Manicke, N. E.; Rusine, A. L.; Cooks, R. G. Rapid Commun. Mass Spectrom. 2008, 22, 503–510. (33) Kennedy, J. H.; Wiseman, J. M. Rapid Commun. Mass Spectrom. 2010, 24, 309–314. (34) Wiseman, J. M.; Evans, C. A.; Bowen, C. L.; Kennedy, J. H. Analyst 2010, 135, 720–5.

steroids,35 dyes,36 herbal supplements,37 natural products,38 and tryptic digests.39,40 More recently and concurrent with this research, DESI was successfully applied to the direct analysis of phospholipids separated in two dimensions on high-performance (HP) TLC plates.41 In this report, we demonstrate methodology for the desorption and ionization of ganglioside species, among other lipids, after partial separation on HPTLC plates by DESIMS using standard mixtures, tissue homogenates, and intact tissue sections. Most importantly, it is shown that gangliosides and other lipids can be eluted directly from thin, intact tissue sections mounted onto standard HPTLC plates, partially separated and analyzed by DESI-MS. The results of these studies are presented here. EXPERIMENTAL SECTION Chemicals. Methanol (MeOH) and acetonitrile (ACN) were purchased from Honeywell (Burdick & Jackson Morristown, NJ). Chloroform was purchased from EMD (an affiliate of Merck KGaA, Darmstadt, Germany). Formic acid (FA) was purchased from Mallinckrodt (Hazelwood, MO). Orcinol and sulfuric acid (H2SO4) were purchased from Sigma Aldrich (St. Louis, MO). All chemicals were used without further purification. Standards. Total ganglioside extract (porcine brain) was purchased from Avanti Polar Lipids (Alabaster, AL). Disialotetrahexosylgangliosides (GD1a) was purchased from Sigma Aldrich (St. Louis, MO) and was dissolved in chloroform-methanol (1:1, v/v). Male, Sprague-Dalley whole rat brain specimens were obtained from Bioreclamation (Long Island, NY). Each rat brain was snap frozen in liquid nitrogen following extraction and shipped to Prosolia on dry ice. High-Performance Thin-Layer Chromatography. HPTLC plates (5 × 5 cm2, glass-backed, 200 µm silica gel 60) were obtained from Whatman (Sanford, MA). A volume of 20 µL of a 1 mg/mL solution of the ganglioside standard porcine extract was loaded by depositing 5 µL in four replicates onto the HPTLC plate. The plate was developed in a 200 mL glass beaker lined with filter paper under saturated atmosphere using a mobile phase consisting of chloroform-methanol-2.5N ammonium hydroxide (60:40:9 v/v/v) for 10 min until the solvent front arrived at the line 1 cm from the top edge. Direct TLC. A 16 µm thick section was cut from whole rat brain using a cryostat (Leica Microsystems, Bannockburn, IL) at -18 °C and thaw mounted directly onto the HPTLC plate using the bristles of a cold artist’s brush. The tissue section was developed immediately using chloroform-methanol-2.5 N ammonium hydroxide (60:40:9 v/v/v) for 10 min. Staining. After analysis of the HPTLC plates by DESI MS and MS/MS, each plate was stained with 0.3% (w/v) orcinol-3 M (35) Janecki, D. J.; Novotny, A. L.; Woodward, S. D.; Wiseman, J. M.; Nurok, D. J. Planar Chromatogr. 2008, 21, 11–14. (36) Van Berkel, G. J.; Ford, M. J.; Deibel, M. A. Anal. Chem. 2005, 77, 1207– 1215. (37) Van Berkel, G. J.; Tomkins, B. A.; Kertesz, V. Anal. Chem. 2007, 79, 2778– 2789. (38) Kennedy, J. H.; Wiseman, J. M. Rapid Commun. Mass Spectrom. 2010, 24, 1305–1311. (39) Pasilis, S. P.; Kertesz, V.; Van Berkel, G. J.; Schulz, M.; Schorcht, S. J. Mass Spectrom. 2008, 43, 1627–1635. (40) Pasilis, S. P.; Kertesz, V.; Van Berkel, G. J.; Schulz, M.; Schorcht, S. Anal. Bioanal. Chem. 2008, 391, 317–324. (41) Paglia, G.; Ifa, D. R.; Wu, C.; Corso, G.; Cooks, R. G. Anal. Chem. 2010, 82, 1744–1750.


8867

Figure 1. Structures of common ganglioside species: monosialotetrahexosylgangliosides (GM), disialotetrahexosylgangliosides (GD), trisialotetrahexosylgangliosides (GT), and tetrasialotetrahexosylgangliosides (GQ).

H2SO4 solution. The plates were dipped for 10 s in the staining solution and then heated on a hot plate at 110 °C until a purple tint appeared on the spots. Images of stained chromatograms were taken with a commercial scanner (Konica Minolta C352). DESI-MS. The DESI experiments were carried out using an Omni Spray 2D Ion Source (Prosolia, Inc., Indianapolis, IN) coupled to a Thermo Fisher LTQ linear ion trap mass spectrometer. The DESI ion source is installed in place of the standard atmospheric pressure ion source on the mass spectrometer. The DESI spray solvent was optimized for ganglioside detection by loading 2 µL of 1 mg/mL ganglioside mixture in MeOHchloroform (1:1 v/v) solution onto an Omni Slide (Prosolia, Inc., Indianapolis, IN) and analyzed by DESI-MS using ACN-H2OFA 80:20:0.1 v/v/v, MeOH-H2O 70:30 v/v, MeOH-chloroform 1:1 v/v, MeOH-chloroform 3:1 v/v in the positive and negative ion mode (data not shown). Standard DESI ion source parameters were used as described elsewhere.42 The HPTLC plate was fixed onto the sample tray of the automated DESI source. The analysis of the tissue section began at the trailing edge of the tissue and each of three lanes was analyzed over a distance of 4 cm. (42) Wiseman, J. M.; Ifa, D. R.; Venter, A.; Cooks, R. G. Nat. Protoc. 2008, 3, 517–524.

8868


RESULTS The primary objectives of this research were twofold: (1) to establish DESI operating parameters and analytical performance for examining gangliosides partially separated on planar chromatographic media (in this case HPTLC plates) and (2) to determine the feasibility for performing planar chromatographic separations of intact tissue slices preceding direct analysis by DESI-MS and MS/MS. Figure 1 shows structures of common ganglioside species detected during this study. Ganglioside lipids present a hydrophobic ceramide core attached to a hydrophilic oligosaccharide headgroup. The presence of the sialic acid moiety makes the detection of the deprotonated forms of the molecules most efficient when the DESI source and the mass spectrometer are operated in negative ion mode. As such, all studies reported here were conducted in the negative ion mode. The solvent composition can have a significant effect on the desorption and ionization of analytes using DESI.43-45 The most commonly used solvent systems include binary and ternary (43) Badu-Tawiah, A.; Bland, C.; Campbell, D. I.; Cooks, R. G. J. Am. Soc. Mass Spectrom. 2010, 21, 572–579. (44) Takats, Z.; Wiseman, J. M.; Cooks, R. G. J. Mass Spectrom. 2005, 40, 1261– 1275. (45) Green, F. M.; Salter, T. L.; Gilmore, I. S.; Stokes, P.; O’Connor, G. Analyst 2010, 135, 731–737.

Figure 2. DESI mass spectrum of GD1a on an HPTLC plate in negative ion mode using a chloroform-methanol (1:3, v/v) spray solvent at 5 µL/min.

mixtures of methanol and water or acetonitrile and water with and without an acidic modifier (e.g., formic or acetic acid). As such, we found that a mixture of 3:1 v/v methanol-chloroform significantly enhanced the desorption of the analytes from the silica gel of the HPTLC plate. Figure 2 shows the DESI mass spectrum in negative ion mode of the GD1a standard recorded from the HPTLC plate. The mass spectrum shows ions related to the multiply deprotonated forms of homologous GD1a species, the (d36:1) and (d38:1), respectively. It is noteworthy that DESI is a soft desorption/ionization method and does not induce fragmentation of the labile oligosaccharide units of the ganglioside species, as evidenced by the lack of fragmentation in the mass spectrum. The limit of detection (LOD) for ganglioside detection using DESI-MS was evaluated by preparing serial dilutions of the GD1a standard and depositing 5 µL of each solution onto the HPTLC plate. Figure 3 shows the DESI mass spectra of 272, 27, and 2.7 pmol of GD1a standard, respectively. On the basis of these data, the resulting LOD was determined to be approximately 3 pmol of GD1a on the HPTLC plate. Notably, the absolute detection limits are substantially lower given the relatively small sampling area of the DESI spray when the solvent pump is operated at 5 µL/min (∼1 mm2) versus the diffuse 5 µL sample deposited onto the HPTLC plate (∼20 mm2). As a result, the fraction of the analyte sampled per unit area using DESI is significantly less than the 3 pmol limit reported. However, the chemical noise resulting from the desorption and ionization of constituents of the HPTLC plate further limit the practical detection limits, as evidenced by the high background levels in the DESI mass spectrum of the lowest GD1a standard (Figure 3). Notably, lower LODs are achieved for GD1 but higher LODs (30 pmol) for GM1 species, which is likely owing to the increased ionization efficiency due to the presence of more sialic acid residues, in the former case. HPTLC/DESI-MS of Gangliosides. HPTLC separations were performed using a mobile phase consisting of chloroformmethanol-2.5 N ammonium hydroxide (60:40:9 v/v/v) over 10 min of standard mixtures of gangliosides extracted from porcine brain. The 5 cm × 5 cm HPTLC plate was fixed to the sample tray of the automated DESI source and analyzed in full scan negative ion mode over the mass-to-charge range (m/z) 400-2000 using the linear ion trap. Each lane on the HPTLC plate was

Figure 3. Detection limit test in negative ion mode of GD1a deposited onto an HPTLC plate. DESI mass spectra of 270, 27, and 2.7 pmol of GD1a using chloroform-methanol (1:3, v/v) spray solvent at 5 µL/min. A total of 12 scans were averaged to produce each mass spectrum. The mass spectra are shown without background subtraction or peak smoothing.

analyzed by continuously moving the surface at a constant velocity beneath the DESI spray. Figure 4 shows the selected ion chromatograms (SICs) and the corresponding DESI mass spectra recorded in negative ion mode of lipids separated on the HPTLC plate. The SICs are presented in order of increasing retention factor (Rf). As shown in Figure 4b, the DESI mass spectrum of the HPTLC band over the time interval 0.3-0.55 min shows the ions at m/z 709.2 and 718.5, which correspond to the [M - 3H]3of homologous GT1 species differing in the fatty acid acyl chain length in the ceramide moiety, the (d36:1) and (d38:1), respectively. The inset shows the less abundant GQ1 species, (d36:1) and (d38:1), at m/z 604.6 and 611.5, respectively. The ions related to GD1, phosphatidylserine, phosphatidylinositol, GM1, and sulfoglycosphingolipids (“sulfatides”) species are shown in Figure 4c-g. Elution, Partial Separation, and DESI-MS of Lipids from Intact Tissue Slices. The examination of intact tissue using DESIMS does not readily produce ions related to the various ganglioside species (i.e., GM1, GD1, GT1, or GQ1). The identification of these lipids directly from tissue by DESI-MS is affected by the differences in desorption/ionization efficiency, relative concentrations of endogenous lipids, and the presence of isobaric ions in the m/z range 700-1000, where the multiply deprotonated forms of gangliosides are commonly detected. MS/MS may be employed to detect ganglioside lipids directly from intact tissue, but the sensitivity is not adequate to permit a comprehensive analysis under the current analytical conditions (Figure S1 in the Supporting Information). Alternatively, liquid extraction of lipids from tissue homogenates can be performed using common procedures; the Folch46 and Bligh and Dyer47 methods are two approaches. These methods usually require vigorous disruption of the tissue (46) Folch, J.; Lees, M.; Sloane Stanley, G. H. J. Biol. Chem. 1957, 226, 497– 509.


8869

Figure 4. Selected ion chromatograms (SICs) and the corresponding DESI mass spectra recorded in negative ion mode of an HPTLC separation of a standard extract of gangliosides: monosialotetrahexosylgangliosides (GM), disialotetrahexosylgangliosides (GD), trisialotetrahexosylgangliosides (GT), tetrasialotetrahexosylgangliosides (GQ). (a) SICs of GQ1(d36:1; d38:1), GT1 (d36:1; d38:1), GD1 (d36:1; d38:1), GM1 (d36:1), phosphatidylserine (38:1), and sulfatide (24:1). Each mass spectrum is an average over the time intervals (b) 0.3-0.55 min (inset shows GQ1 (d36:1; d38:1) species at m/z 604.6 and 611.5, respectively), (c) 0.55-0.85 min, (d) 1.1-1.4 min, (e) 1.2-1.6 min, (f) 2.25-2.5 min, (g) 2.852.95 min.

membranes by sonication in a chloroform-methanol mixture or pulverization in liquid nitrogen, which is followed by further purification. Moreover, these procedures are time-consuming and involve specialized glassware to minimize sample losses via adsorption to the walls of common laboratory vials and pipet tips. To circumvent these steps and enable enrichment of lipid fractions, we attempted to elute and separate gangliosides, among other lipids, directly from thin coronal sections of rat brain tissue while fixed to an HPTLC plate and to analyze the (47) Bligh, E. G.; Dyer, W. J. Can. J. Biochem. Physiol. 1959, 37, 911–917.

8870


separated fractions by DESI-MS and MS/MS. For direct TLC of tissue, a mobile phase consisting of chloroform-methanol2.5N ammonium hydroxide (60:40:9 v/v/v) was used to produce the data shown in Figure 5. Consistent with conventional solvent extraction of lipids from tissue, the chloroform-methanol mixture provided sufficient extraction efficiency which resulted in greater mass spectral response. Several factors are influencing the planar chromatography of intact tissue as opposed to the purified standards; these include the hydration of the tissue, tissue thickness, extraction efficiency by the passive flow of

Figure 5. (a) Chemically stained HPTLC plate with 0.3% (w/v) orcinol-3 M H2SO4 of a separation of a 16 µm rat brain section. (b) Selected ion chromatograms (SICs) of GQ1 (d36:1; d38:1), GT1 (d36:1; d38:1), GD1 (d36:1; d38:1), GM (d36:1), PS (38:1), and ST (24:1).

Table 1. Ions Detected in Negative Ion Mode from the Elution and Partial Separation of an Intact 16 µm Rat Brain Section Using HPTLC/DESI-MS and MS/MS lipid classa GQ1 GT1

GD1a/b GM1 ST

hST

PS

PI PlsEtn

PE

observed mass-to-charge ratio (m/z) 604.4 611.3 709.0 1063.6 718.3 1077.7 918.2 931.8 1545.0 1572.9 778.8 806.7 834.7 888.8 890.8 822.6 850.6 878.6 904.8 906.8 788.7 834.7 836.7 838.9 885.8 806.1 700.7 722.7 726.7 728.7 750.7 774.7 776.7 778.7 714.7 716.7 742.8 744.6 764.6

ion type [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 [M [M [M [M [M [M [M [M [M [M [M [M [M [M

-

4H]44H]43H]32H]23H]32H]22H]22H]2H]H]H]H]H]H]H]H]H]H]H]H]H]H]H]H]H]H]H]H]H]H]H]H]H]H]H]H]H]H]H]-

molecular weight (g/mol)

proposed lipid compositionb

2419.1 2447.1 2128.0 2128.0 2156.0 2156.0 1836.9 1864.9 1545.8 1573.9 779.5 807.7 835.7 889.8 891.8 823.6 851.6 879.6 905.8 907.8 789.5 835.8 837.7 839.9 886.6 807.1 701.7 723.7 727.7 729.8 751.8 775.6 776.5 779.8 715.7 717.7 743.8 745.6 765.6

d36:1 d38:1 d36:1 d36:1 d38:1 d38:1 d36:1 d38:1 d36:1 d38:1 16:0 18:0 20:0 24:1 24:0 h18:0 h20:0 h22:0 h24:1 h24:0 d36:1 d40:6 d40:5 d40:4 d38:4 d36:1 p34:1 p36:4 p36:2 p36:1 p42:0 p40:6 p40:6 p40:4 d34:2 d34:1 d36:2 d36:1 d38:5

a GQ1, Tetrasialohexsylgangliosides; GT1, trisialotetrahexsylgangliosides; GD1, disialotetrahexosylganliosides; GM1, monosialotetrahexosylganliosides; PS, phosphatidylserine; PI, phosphatidylinositol; hST, hydroxylated sulfatide; PE, phosphatidylethanolamine; PlsEtn, plasmenyl phosphatidylethanolamine. b (X:Y) represents the different number of carbon atoms and the different number of double bonds in the fatty acid chains. The notation (hX:Y) denotes a hydroxylated sulfatide species.


8871

Figure 6. DESI full scan negative ion mass spectra of selected lipids separated from a 16 µm coronal rat brain tissue section on an HPTLC plate using chloroform-methanol (1:3, v/v) spray solvent at 5 µL/min and a surface velocity of 200 µm/s. Each mass spectrum is an average over the time intervals (a) 0.2-0.4 min (inset shows GQ1 (d36:1; d38:1) species at m/z 604.4 and 611.3, respectively), (b) 0.65-0.9 min, (c) 0.85-1.25 min, (d) 1.4-1.6 min, (e) 1.55-1.75 min, and (f) 2.35-2.5 min.

solvent across the tissue section, solvent composition, and the effects of adsorption/desorption processes by other endogenous species on the active sites of the plate. We have not explored these processes in detail here, rather our purpose is to report the first results on this proof-of-concept study. The direct TLC method was performed by mounting a 16 µm thick coronal rat brain section onto the HPTLC plate approximately 1 cm from one edge as measured from the leading edge of the tissue section; the trailing edge or bottom of the tissue section marked the origin. The HPTLC plate was developed over 10 min, in a closed container, which resulted in only a partial separation of the various lipids but sufficient for direct detection by DESI-MS (Figure 5). Figure 5a displays the direct HPTLC separation of a rat brain section that has been chemically stained 8872


with orcinol and sulfuric acid after analysis by DESI-MS and MS/ MS. A separation of the ganglioside standard extract is shown on the same plate for comparison. Each of three lanes spanning a distance of 4 cm from the origin were analyzed. Figure 5b shows the SICs recorded from lane one in order of increasing Rf for several ganglioside and major lipid species detected by DESI-MS using the direct HPTLC method. (The results from the analysis of lanes 2 (Figures S4-5) and 3 (Figures S6-7) are shown in the Supporting Information). As shown in the SICs, GQ1, GT1, and GD1 species are not completely separated from each other but are clearly separated from the higher abundant phospholipids, enabling detection with minimal sample processing of the tissue section. The advantage of coupling TLC with DESI-MS is that optimal planar separations are not required given the high

Figure 7. DESI-MS/MS spectra of selected ganglioside lipids on the HPTLC plate recorded using the linear ion trap. The annotation of fragment ions is according to the nomenclature of Domon and Costello. (a) MS/MS spectrum of the precursor ions of m/z 708.7 recorded using a normalized collision energy (NCE) of 35% and an isolation m/z width of 5. The proposed cleavages are shown in the inset. (b) MS/MS spectrum of the precursor ions of m/z 918.2 from the time interval 1.4-1.6 min recorded using a normalized collision energy (NCE) of 35% and an isolation m/z width of 5. The fragmentation pattern indicates the presence of the GD1a isomer; the proposed cleavages are shown in the inset. (c) MS/MS spectrum of the precursor ions of m/z 918.2 from the time interval 0.85-1.25 min recorded using a normalized collision energy (NCE) of 35% and an isolation m/z width of 5. The fragmentation pattern indicates the presence of the GD1b isomer; the proposed cleavages are shown in the inset. (d) MS/MS spectrum of the precursor ions of m/z 603.8 recorded using a normalized collision energy (NCE) of 35% and an isolation m/z width of 5. The proposed cleavages are shown in the inset.

sensitivity, specificity, and spatial resolution of the DESI-MS system for making direct measurements of the TLC plate. The DESI mass spectra representative of each SIC shown in Figure 5b are displayed in Figure 6. In accordance with the HPTLC/ DESI-MS of the ganglioside standard mixture, the DESI mass spectra show ions related to the various ganglioside molecules (GQ, GT, GD, and GM), phosphatidylserines, phosphatidylinositols, phosphoethanolamines, and sulfatides. Table 1 displays a comprehensive list of the various lipid species detected in the rat brain tissue section from direct analysis by HPTLC/DESI-MS and MS/MS. Since full scan MS cannot reveal structural information, identification of the lipids was carried out by MS/MS using the ion trap and comparisons to various literature sources referenced here. MS/MS was performed either in the spot sampling mode or by continuous surface scanning, although continuous surface sampling at a constant velocity proved to be the best approach. Figure 7 shows several DESI-MS/MS spectra recorded from an HPTLC separation of rat brain tissue providing additional evidence toward the proposed assignments. The annotation of fragment ions in the product ion spectra is according to the nomenclature proposed by Domon and Costello.48 (48) Domon, B.; Costello, C. E. Glycoconjugate J. 1988, 5, 397–409.

Characteristically, the product ion spectrum in Figure 7a of GT1 (d36:1) recorded from the precursor ion [M - 3H]3- at m/z 708.7 shows intense fragment ions at m/z 917.7, 290.2, 1545.0, 581.3, and 1253.8, in order from most to least intense, corresponding to the [Y4R]2-, [C1β - H2O]-, [Y4R/B1β]-, [B2β - 2H2O]-, and [Y4R /B1β]- ions, respectively, which result from extensive desialylation. Notably, at this stage it is not possible to unequivocally differentiate between the GT1 isomers (i.e., GT1a and GT1b), although GT1b is most prevalent in vertebrate tissues, assuming that losing two sialyl substituents from either chain is equally probable. Figure 7b,c shows the MS/MS spectra of the precursor ions at m/z 918.2 assigned to GD1. The interrogation of the peaks in the SICs between 0.65 and 0.9 and 1.55-1.7 min using MS/MS shows two fragmentation patterns having subtle differences, which we attribute to the presence of the GD1 isomers, GD1a and GD1b/c. GD1a has two sialyl substituents with one bonded to each of the root sugar residues on the R- and β-chains whereas, GD1b/c have a single disialyl substituent bonded to the root sugar residue of the R-chain (Figure 1). Given that a single cleavage of the bond between Neu5Ac and Gal on either chain is most probable for GD1a, one would expect a more abundant peak associated with the loss of one sialyl substituent [M - Neu5Ac]- than of two sialyl substituents Analytical Chemistry, Vol. 82, No. 21, November 1, 2010

8873

[M - 2Neu5Ac]-. In comparison, because of the single disialyl substituent on GD1b/c, one would expect more abundant product ions associated with the loss of both substituents. In accordance with these assumptions, the product ion spectrum in Figure 7b shows intense product ions associated with the loss of two sialyl substituents located at m/z 1253.7 corresponding to the [M - 2Neu5Ac - H]- and at m/z 581.3 corresponding to the [2Neu5Ac - 2H2O - H]or [B2R - 2H2O]-. In contrast, Figure 7c shows intense product ions associated with the loss of a single sialyl substituent located at m/z 1545.0 and relatively little contribution from product ions associated with the loss of two sialyl substituents. Additionally, the elution order with GD1b/c at lower Rf than GD1a is consistent with the literature and further supports the proposed assignments of the GD1 isomers. The peaks in the SICs for m/z 918 and 932 located at 2.12 and 2.15 min, respectively, are attributed to isobaric ions. Finally, Figure 7d shows the fragmentation of the [M - 4H]4- of GQ1 confirming its presence in the tissue sample. Characteristically, the MS/MS spectrum shows product ions resulting from extensive desialylation of the parent ions. For comparison, we also performed biphasic liquid extractions of rat brain tissue using a modified Folch method and HPTLC/ DESI-MS of the purified material (Supporting Information). The results of these studies show good qualitative agreement with those obtained by direct elution and separation of the intact tissue section (Figures S2 and S3 in the Supporting Information). This fact is remarkable given the significant differences in the quantity of tissue used in each study, a few milligrams for direct elution versus 100 mg for homogenization and extraction and with a relatively simple procedure with which the direct elution experiment is performed.

fluids presents a challenge for direct analysis methods like DESI when coupled to mass spectrometers with only modest mass resolving power. The coupling of HPTLC and DESI-MS yields a simple and fast method for the analysis of gangliosides and other lipids without extraction of the identified bands from the silica gel. While the chromatographic separation is not optimized, a more comprehensive examination of complex mixtures can be achieved due to a reduction in mass spectral complexity. Importantly, our aim is to use rapid planar separations as tools for the presentation of complex samples into the DESI ion source. Furthermore, we demonstrated here that planar separations of lipids from intact tissue slices mounted on HPTLC plates can be achieved and in combination with DESI-MS results in a simplified method for the analysis of lipids in tissue. This procedure avoids tissue homogenization, solvent extraction, and further purification of the sample significantly reducing the complexity and time of the assay. Clearly, this procedure may be applied to other types of tissues and complex samples, such as those encountered in food safety, cancer diagnostics, and microbiological testing, resulting in a simplified method for the analysis of complex matrixes.

CONCLUSIONS The examination of complex lipid mixtures such as those encountered during the analysis of tissue sections or biological

Received for review June 22, 2010. Accepted October 4, 2010.

8874


ACKNOWLEDGMENT The authors acknowledge the assistance of Peter T. Kissinger, Brian Laughlin, and Joseph Kennedy of Prosolia for their review and helpful criticism of this manuscript. SUPPORTING INFORMATION AVAILABLE Additional information as noted in text. This material is available free of charge via the Internet at http://pubs.acs.org.

AC1016453

Elution, Partial Separation, and Identification of Lipids Directly from

Recommend Documents