Structural Profiling of Individual Glycosphingolipids in a Single Thin

Anal. Chem. 2009, 81, 9481–9492

Structural Profiling of Individual Glycosphingolipids in a Single Thin-Layer Chromatogram by Multiple Sequential Immunodetection Matched with Direct IR-MALDI-o-TOF Mass Spectrometry Jamal Souady,† Jens Soltwisch,† Klaus Dreisewerd,† Jo¨rg Haier,‡ Jasna Peter-Katalinic´,† and Johannes Mu¨thing*,§ Institute of Medical Physics and Biophysics and Institute for Hygiene, University of Mu¨nster, D-48149 Mu¨nster, Germany, and Department of General Surgery, Mu¨nster University Hospital, D-48149 Mu¨nster, Germany The thin-layer chromatography (TLC) immunoenzyme overlay assay is a widely used tool for antibody-mediated identification of glycosphingolipids (GSLs) in mixtures. However, because the majority of GSLs is left unexamined in a chromatogram of a single assay, we developed a novel method that permits detection of various GSLs by sequential multiple immunostaining combined with individual coloring of GSLs in the same chromatogram. Specific staining was achieved by means of primary anti-GSL antibodies, directed against lactosylceramide, globotriaosylceramide, and globotetraosylceramide, in conjunction with alkaline phosphatase (AP)- or horseradish peroxidase (HRP)-conjugated secondary antibodies together with the appropriate chromogenic substrates. Triple coloring with 5-bromo-4-chloro-3-indolyl phosphate (BCIP)AP, Fast Red-AP, and 3,3′-diaminobenzidine (DAB)-HRP resulted in blue, red, and black precipitates, respectively, following three sequential immunostaining rounds. Structures of antibody-detected GSLs were determined by direct coupling of TLC with infrared matrix-assisted laser desorption/ionization orthogonal time-of-flight mass spectrometry. This combinatorial technique was used to demonstrate structural GSL profiling of crude lipid extracts from human hepatocellular cancer. This powerful technology allows efficient structural characterization of GSLs in small tissue samples and marks a further step forward in the emerging field of glycosphingolipidomics. Glycosphingolipids (GSLs), composed of a hydrophilic, highly variable carbohydrate chain and a lipophilic ceramide anchor, play pivotal roles in countless biological processes, including infectious diseases1 and the development of cancer.2-4 The GSLs can be * To whom correspondence should be addressed. Phone: +49-251-8355192. Fax: +49-251-8355341. E-mail: [email protected]. † Institute of Medical Physics and Biophysics, University of Mu ¨ nster. ‡ Mu ¨ nster University Hospital. § Institute for Hygiene, University of Mu ¨ nster. (1) Mu ¨ thing, J.; Schweppe, C. H.; Karch, H.; Friedrich, A. W. Thromb. Haemost. 2009, 101, 252–264. (2) Feizi, T. Nature 1985, 314, 53–57. 10.1021/ac901948h CCC: $40.75  2009 American Chemical Society Published on Web 10/20/2009

grouped into structural families, e.g., the globo- or neolacto-series,5 characterized by specific sequences of monosaccharides. In mammalian cells, the ceramide portion most often consists of the long chain aminoalcohol sphingosine (d18:1) that is linked with a fatty acid varying in chain length from C16 to C24. Because malignant cells show aberrant glycosylation, there is a growing interest to identify tumor-associated GSLs. They may serve as diagnostic markers or potential therapeutic targets.6 High-performance thin-layer chromatography (TLC) is widely used in glycosphingolipid (GSL) analysis because of its high resolving power, robustness, and easy handling.7 The TLC overlay assay, which has been introduced by Magnani and co-workers,8 offers a versatile tool for the localization and identification of GSLs on TLC plates. GSLs can be differentiated in overlay assays on the silica gel surface of the plate using anti-GSL antibodies that detect the lipid-bound oligosaccharides with strict structural specificity.9 The basic principle involves a primary GSL-binding antibody in conjunction with a secondary antibody, which is specific for the immunoglobulin (Ig) subclass of the primary antiGSL antibody. The secondary antibody is labeled with an enzyme that enables an amplified detection by use of a chromogenic substrate to visualize and localize the primary antibody on the plate. Alkaline phosphatase (AP) is commonly used in combination with 5-bromo-4-chloro-3-indolyl phosphate (BCIP) as a convenient substrate generating a blue precipitate.10 The sensitivity of this technique is generally high requiring only nanogram amounts of (3) Hakomori, S. Cancer Res. 1996, 56, 5309–5318. (4) Kannagi, R.; Yin, J.; Miyazaki, K.; Izawa, M. Biochim. Biophys. Acta 2008, 1780, 525–531. (5) Lochnit, G.; Geyer, R.; Heinz, E.; Rietschel, E. T.; Za¨hringer, U.; Mu ¨ thing, J. In Glycoscience: Chemistry and chemical biology; Fraser-Reid, B. O., Tatsuta, K., Thiem, J., Eds.; Springer-Verlag: Heidelberg, 2001; Vol. III, pp 21832249. (6) Distler, U.; Souady, J.; Hu ¨ lsewig, M.; Drmic´-Hofman, I.; Haier, J.; Denz, A.; Gru ¨tzmann, R.; Pilarsky, C.; Senninger, N.; Dreisewerd, K.; Berkenkamp, S.; Schmidt, M. A.; Peter-Katalinic, J.; Mu ¨ thing, J. Mol. Cancer Ther. 2008, 7, 2464–2475. (7) Mu ¨ thing, J. J. Chromatogr., A 1996, 720, 3–25. (8) Magnani, J. L.; Smith, D. F.; Ginsburg, V. Anal. Biochem. 1980, 109, 399– 402. (9) Kannagi, R. Methods Enzymol. 2000, 312, 160–179. (10) Mu ¨ thing, J. In Methods in Molecular Biology, Vol. 76: Glycoanalysis Protocols; Hounsell, E.F., Ed.; Humana Press Inc.: Totawa, NJ, 1998; pp 183-195.

Analytical Chemistry, Vol. 81, No. 22, November 15, 2009

9481

GSLs. Its high specificity also provides information on type of sugars and intrinsic glycosidic linkages of lipid-bound oligosaccharides. This nondestructive detection of GSLs has been recently coupled with direct infrared matrix-assisted laser desorption/ ionization orthogonal time-of-flight mass spectrometry (IR-MALDIo-TOF-MS) on the TLC plate.11,12 This method has been successfully applied for tracing tumor-associated GSLs in pancreatic cancer.6,13 However, immunodetection of a specific GSL in a single chromatogram of a complex GSL mixture leaves the majority of GSLs unexamined. Since only one GSL structure is principally investigated in one TLC run, an overlay procedure that allows multiple GSL detection by sequential use of different antibodies against distinct GSLs in the same sample would be highly desirable. This applies especially to samples that are only available in minute amounts, e.g., biopsy specimens obtained from preoperative tumor diagnostic. Multiple detection procedures based on enzyme-amplified detection using various substrates that are producing different colors have been developed for related techniques such as Western blotting and immunohistochemistry.14-18 AP and horseradish peroxidase (HRP) are two frequently used antibodyconjugated enzymes. Together with the appropriate chromogenic substrates such as BCIP (with or without nitro blue tetrazolium chloride) and Fast Red (in conjunction with naphthol phosphate) for AP, as well as 3,3′-diaminobenzidine (DAB) (doped with nickel chloride in H2O2 substrate) for HRP, differentially colored chromophores are produced. These chromophores are all insoluble in water and, thus, suitable for surface based immunoassays. In this study, we (1) developed a multiple immunoenzyme staining procedure targeting individual TLC-separated GSLs of the same sample on TLC plates and (2) merged this multicolor TLC immunostaining with IR-MALDI-o-TOF-MS for structural characterization of immunopositive GSLs directly on the TLC plate. For this purpose, we extended the conventional BCIP based TLC overlay assay, which usually uses a primary anti-GSL antibody and a secondary AP-conjugated antibody, by introducing two additional substrates: Fast Red for AP-conjugated antibodies and DAB for HRP-conjugated antibodies. Antibodies specific for lactosylceramide (LacCer) and neutral GSLs of the globo-series were employed for detection. Three sequential staining rounds resulted in blue, red, and black-colored bands, respectively, of individual immunopositive GSLs within a single chromatogram. In addition to the BCIP-stained GSLs, which are suitable for direct MS analysis as shown in previous studies,6,11 Fast Red- and DABcolored GSLs were demonstrated to be compatible with direct (11) Distler, U.; Hu ¨ lsewig, M.; Souady, J.; Dreisewerd, K.; Haier, J.; Senninger, N.; Friedrich, A. W.; Karch, H.; Hillenkamp, F.; Berkenkamp, S.; PeterKatalinic´, J.; Mu ¨ thing, J. Anal. Chem. 2008, 80, 1835–1846. (12) Mu ¨ thing, J.; Distler, U. Mass Spectrom. Rev. 2009; Epub ahead of print; DOI: 10.1002/mas.20253. (13) Distler, U.; Souady, J.; Hu ¨ lsewig, H.; Drmic´-Hofman, I.; Haier, J.; Friedrich, A. W.; Karch, H.; Senninger, N.; Dreisewerd, K.; Berkenkamp, S.; Schmidt, M. A.; Peter-Katalinic´ J.; Mu ¨ thing, J. PLoS ONE 2009, 4, e6813. (14) Shainoff, J. R.; Urbanic, D. A. Blood Coagul. Fibrinolysis 1990, 1, 479–484. (15) Lan, H. Y.; Mu, W.; Nikolic-Paterson, D. J.; Atkins, R. C. J. Histochem. Cytochem. 1995, 43, 97–102. (16) Speel, E. J.; Ramaekers, F. C.; Hopman, A. H. Histochem. J. 1995, 27, 833– 858. (17) Stern, C. D. Curr. Top. Dev. Biol. 1997, 36, 223–236. (18) Deininger, M. H.; Meyermann, R. Histochem. Cell Biol. 1998, 110, 425– 430.

9482


TLC-IR-MALDI-o-TOF mass spectrometry. To the best of our knowledge, this is the first report on sequential multiple immunostaining combined with multicoloring of distinct GSLs in the same sample, as well as the first study matching Fast Red- and DAB-mediated detection with in situ mass spectrometry on the TLC plate. This method was successfully applied to the identification of tumor-associated GSLs in human liver cancer. EXPERIMENTAL SECTION Reference GSLs. A mixture of neutral GSLs, comprising lactosylceramide (LacCer), globotriaosylceramide (Gb3Cer), and globotetraosylceramide (Gb4Cer) as main components, was isolated from human erythrocytes as previously described.19 The nomenclature follows the IUPAC-IUB recommendations.20 The symbolic representation system according to Varki21 and the Consortium for Functional Glycomics22 is used throughout this publication. All monosaccharides are in D-configuration and all glycosidic linkages originate from the C1 hydroxyl group. Antibodies. Primary polyclonal chicken IgY anti-LacCer, antiGb3Cer, and anti-Gb4Cer antibodies with well characterized binding specificities1,23,24 were used for overlay assays.10 Affinity chromatography-purified secondary rabbit anti-chicken IgY antibodies labeled with alkaline phosphatase (AP) or horseradish peroxidase (HRP) were purchased from Dianova (Hamburg, Germany). Primary and secondary antibodies were employed in a 1:2000 dilution in phosphate-buffered saline (PBS) supplemented with 1% (w/v) bovine serum albumin (BSA, no. 1930, fraction V; Serva Feinbiochemica GmbH & Co KG, Heidelberg, Germany) for the TLC overlay binding assays. Surgical Specimens. Tissue samples of a hepatocellular carcinoma and a corresponding control specimen were obtained from a patient who had undergone surgery for the primary tumor at an organ site without macroscopic tumor involvement and with a minimal distance to the tumor of 5 cm. The samples were snap frozen in liquid nitrogen immediately after removal and stored at -80 °C until use. Tissue wet weights of normal and cancerous tissues were 73.5 and 148.5 mg, respectively. The Local Ethical Committee of the Medical Council of Westfalen-Lippe and the Medical Faculty of the University Hospital Mu¨nster (Mu¨nster, Germany) approved the current study (reference number 1IXHai). All patients were informed and consented in writing. Lipid Extraction from Surgical Specimens. The tissue samples were homogenized using a rotor-stator dispersion system (Polytron PT 1200, Kinematica AG, Littau, Switzerland) and extracted three times using 4 mL of different chloroform/methanol mixtures with ratios (1/2), (1/1), and (2/1) (all v/v). The combined supernatants of the lipid extracts were dried by a rotary evaporator. Esterified non-GSL lipids, which were coextracted with GSLs, were saponified by incubation in 4 mL of 1 M aqueous (19) Meisen, I.; Friedrich, A. W.; Karch, H.; Witting, U.; Peter-Katalinic´, J.; Mu ¨ thing, J. Rapid Commun. Mass Spectrom. 2005, 19, 3659–3665. (20) Chester, M. A. Glycoconj. J. 1999, 16, 1–6. (21) Varki, A. Nature 2007, 446, 1023–1029. (22) Raman, R.; Venkataraman, M.; Ramakrishnan, S.; Lang, W.; Raguram, S.; Sasisekharan, R. Glycobiology 2006, 16, 82R–90R. (23) Duvar, S.; Peter-Katalinic´, J.; Hanisch, F. G.; Mu ¨ thing, J. Glycobiology 1997, 7, 1099–1109. (24) Schweppe, C. H.; Bielaszewska, M.; Pohlentz, G.; Friedrich, A. W.; Bu ¨ ntemeyer, H.; Schmidt, M. A.; Kim, K. S.; Peter-Katalinic´, J.; Karch, H.; Mu ¨ thing, J. Glycoconj. J. 2008, 25, 291–304.

NaOH at 37 °C for 1 h. After neutralization with 400 µL of 10 M HCl, samples were dialyzed against deionized water. The desalted samples were dried, and the extracts were dissolved in chloroform/ methanol (2/1, v/v) and stored at -20 °C. High-Performance Thin-Layer Chromatography. GSL mixtures were applied to glass-backed silica gel 60 precoated highperformance thin-layer chromatography plates (HPTLC plates, 10 cm × 10 cm, 200 µm thickness, no. 1.05633.0001; Merck, Darmstadt, Germany) with an automatic sample applicator (Linomat IV, CAMAG, Muttenz, Switzerland) and separated for 20 min in a solvent system composed of chloroform/methanol/water (120/ 70/17, each by volume) supplemented with 2 mM CaCl2. GSLs were visualized by dipping the dried plates in orcinol solution (0.3% (w/v) in 3 M sulfuric acid) followed by transfer onto a preheated heating plate (110 °C) or by TLC immunostaining. Conventional BCIP-AP-Mediated Single TLC Immunostaining. After separation of GSLs on the TLC plate, but prior to incubation with aqueous solutions, the silica gel layer was stabilized with plastic by means of chromatography in 0.5% (w/ v) Plexigum P28 (polyisobutylmethacrylate; Ro¨hm, Darmstadt, Germany) in n-hexane.10 Overnight preincubation of the TLC plate in PBS at 37 °C enhanced the sensitivity of the subsequent immunostaining procedure. All of the following steps were performed at ambient temperature unless otherwise stated. The TLC plate was blocked for 15 min using 1% (w/v) BSA in PBS to prevent unspecific antibody binding. The primary chicken IgY antiLacCer antibody was overlaid for 1 h, followed by 3-fold washing of the plate with PBS to remove unbound antibody. The APconjugated secondary antibody, which is directed against the IgYsubtype of the primary anti-LacCer antibody, was overlaid for 1 h. After 3-fold washing, the plate was equilibrated for 15 min in glycine buffer (0.1 M glycine, 1 mM ZnCl2, 1 mM MgCl2, pH 10.4). This is the substrate buffer for the ensuing AP-catalyzed reaction using 0.05% (w/v) of chromogenic 5-bromo-4-chloro3-indolyl phosphate p-toluidine salt (BCIP; Roth, Karlsruhe, Germany). BCIP coloring resulted in the generation of a blue precipitate at sites of phosphatase activity. The reaction was stopped by incubation with glycine buffer. After drying, the dried plate was stored at -20 °C under light protection. Sequential Multiple TLC Immunostaining Combined with Multicoloring. In order to detect unexamined GSL species from complex GSL mixtures, the chromatogram from the single assay was analyzed with anti-GSL antibodies possessing specificities different from that of the first anti-GSL antibody. To discriminate the individual staining assays, multicoloring was performed by introducing red and black chromogenic substrates for AP- and HRP-conjugated secondary antibodies, respectively, in addition to BCIP-AP-mediated blue coloring. Fast Red Coloring. The TLC plate was incubated overnight in PBS and then successively overlaid with the primary chicken IgY anti-Gb3Cer and secondary AP-conjugated rabbit anti-chicken IgY antibody, including the intermediate washing and substrate buffer equilibration steps (see Conventional BCIP-AP-Mediated Single TLC Immunostaining). For Fast Red staining, equal volumes of 2 mg/mL naphthol AS-MX phosphate (Serva) and 2 mg/mL Fast Red TR salt solution (Fluka Chemie GmbH, Buchs, Switzerland), both dissolved in 200 mM Tris pH 9.1 substrate buffer, were

filtered and mixed prior to incubation.25 The enzyme-catalyzed reaction of the AP-conjugated secondary anti-chicken IgY antibody yielded a red precipitate. Diaminobenzidine Coloring. For diaminobenzidine staining, primary chicken IgY anti-Gb4Cer antibody was detected with secondary HRP-conjugated rabbit anti-chicken IgY antibody, following the procedure described above for BCIP and Fast Red staining. Ten milligrams of 3,3′-diaminobenzidine (DAB, Fluka Chemie GmbH) was dissolved in 20 mL of 50 mM Tris pH 7.3 and supplemented with 1 mL of aqueous 1% (w/v) NiCl2 (hexahydrate salt) and 7 µL of 30% H2O2 prior to incubation. The nickel salt solution was added very slowly under stirring into the DAB solution to prevent precipitation. DAB coloring produced a black insoluble dye. For staining with HRP, exposure to azide was avoided to prevent loss of enzyme activity. Color development was stopped by washing with the substrate buffer as described for BCIP and Fast Red staining. Especially for DAB coloring, careful washing of the TLC plate with 50 mM Tris pH 7.3 is required to prevent background staining that may occur during drying on air. Triple Coloring Procedure. The three TLC-separated GSL species LacCer, Gb3Cer, and Gb4Cer were visualized by coloring in blue (BCIP-AP), red (Fast Red-AP), and black (DAB-HRP), respectively, in the same chromatogram by means of specific primary and corresponding secondary enzyme-labeled antibodies. Each assay of the three immunostaining rounds was done according to the above-described individual protocols. Since all three primary anti-GSL antibodies were derived from the same animal species (chicken), intermediate heat denaturation of bound antibodies after each round of immunostaining is essential to prevent cross-reactivity of the secondary antibodies with the primary antibodies of the preceding immunostain. In addition, heat treatment inactivated AP of secondary antibodies which were used in the first and second immunostaining round, respectively. Heat treatment was performed by incubating TLC plates for 90 min at 140 °C in a circulating air oven (ULE 500, Memmert GmbH, Schwabach, Germany) prior to the second and third immunostaining round. Colored chromatograms were dried and subjected to densitometry after each immunostaining round. Densitometry. Individual neutral GSLs of the reference mixture, prepared from human erythrocytes,19 were quantified as pinkish-violet orcinol stained bands with a CD60 scanner (Desaga, Heidelberg, Germany, software ProQuant, version 1.06.000) in reflectance mode at 544 nm with a light beam slit of 0.02 mm × 3 mm. Mixtures of 10 µg contained 0.3 µg and 0.6 µg of LacCer (d18:1, C24:1/C24:0) and LacCer (d18:1, C16:0), respectively, 1.7 µg and 0.4 µg of Gb3Cer (d18:1, C24:1/C24:0) and Gb3Cer (d18:1, C16:0), respectively, and 5.7 µg and 0.1 µg of Gb4Cer (d18:1, C24:1/C24:0) and Gb4Cer (d18:1, C16:0), respectively, as well as some minor, not further identified neutral GSLs. TLC immunodetected red, blue, and black-colored GSL bands were scanned at 520 nm (Fast Red), 650 nm (BCIP), and 750 nm (DAB), respectively. Serial dilutions of the reference mixture were employed to determine the detection limits of individual GSLs. In addition, absorption spectra of the Fast Red, BCIP, and DAB precipitates were recorded with the CD60 scanner covering a (25) De Jong, A. S.; Van Kessel-van Vark, M.; Raap, A. K. Histochem. J. 1985, 17, 1119–1130.


9483

wavelength range from 200 to 900 nm and were normalized to their maxima. Infrared Matrix-Assisted Laser Desorption/Ionization Orthogonal Time-of-Flight Mass Spectrometry (IR-MALDIo-TOF-MS). Coupling TLC with IR-MALDI-o-TOF-MS and specifications of the mass spectrometer have been described in detail previously.11,26-28 Immunostained TLC plates were incubated for 2 h in 10 mM ammonium acetate pH 3.6 prior to MS investigation. The Plexigum fixative was removed by dipping the dried plates three times for about 3 min in distilled chloroform. Since the Fast Red precipitate is soluble in chloroform and, therefore, was coextracted with the silica gel fixative, the red bands were encircled with a soft graphite pencil before dipping the plate in chloroform. In addition, the BCIP and DAB stained bands were also marked with a pencil to facilitate targeting of the laser beam in the ion source. TLC plates were cut into pieces of 30 mm × 40 mm to fit into the custom-made MALDI-MS target holder. Immunopositive bands were spotted with ∼1 µL of glycerol per band, acting as matrix for the IR-MALDI process. Signals acquired from 240 IR laser shots were combined for each spectrum. Serial dilutions of the reference mixture were employed to determine the detection limits of individual GSLs after multiple immunostaining. The instrument used was an orthogonal time-of-flight mass spectrometer equipped with an Er:YAG laser (Speser, Spektrum GmbH, Berlin, Germany) emitting pulses of ∼100 ns duration at a wavelength of 2.94 µm. The laser was operated with a repetition rate of 2 Hz, and the spot size on target was ∼200 × 350 µm2. Measurements were performed in positive ion mode with a TOF accelerator voltage of 10 kV and 1.2 mbar nitrogen pressure in the ion source. A mass resolution of ∼10 000 and a mass accuracy of 750 nm (Figure 2). Consequently, the three stainings could be detected specifically if the substrates were used in the order of (1) Fast Red, (2) BCIP, and (3) DAB followed by densitometric scanning at, respectively, 520, 650, and 750 nm after each immunodetection cycle. Sequential Multiple TLC Immunostaining Combined with Multicoloring of GSLs in Mixtures. The procedure aimed at the antibody-mediated specific detection of individual GSLs in mixtures was established by the use of neutral GSLs from human erythrocytes as references. The three major components of the mixtureareLacCer(Galβ4Glcβ1Cer),Gb3Cer(GalR4Galβ4Glcβ1Cer), and Gb4Cer (GalNAcβ3GalR4Galβ4Glcβ1Cer) as shown by orcinol staining of the chromatogram in Figure 3A. The neutral GSLs were separated first according to the number of monosaccharides and second according to the chain length of the fatty acid of the ceramide moiety. This resulted in the formation of double bands for each GSL species. The upper and lower bands mainly comprise GSLs with a ceramide moiety carrying a C24 fatty acid and a C16 fatty acid, each linked to sphingosine (d18:1). Individual GSLs within a mixture were localized in the same lane by sequential multiple TLC immunostaining with concomitant multicoloring in the order of (1) Gb3Cer with Fast Red and (2) LacCer with BCIP detection, both in conjunction with AP-conjugated secondary antichicken IgY antibodies, and (3) Gb4Cer by means of DAB detection obtained with HRP-conjugated secondary anti-chicken IgY antibody. The result of the three consecutive immunostaining rounds is shown in Figure 3A (lanes 1 to 3) in comparison to the reference orcinol sugar stain. The set of three differently colored substrates permitted specific and distinguishable detection of individual neutral GSLs and revealed clear densitometric scans, which were obtained after each immunostaining circle (Figure 3B). No interference in sequential densitometry of the differently stained bands could be observed when operating at the selected wavelengths. The color intensities of GSL bands remained stable through the three staining cycles with the exception of BCIP-AP mediated staining of LacCer which exhibited a slight decrease in color strength after visualization of Gb4Cer with DAB-HRP. Gb3Cer coloring with Fast Red indicated two Gb3Cer species with heterogeneity in the ceramide moiety (Figure 3A, lane 1). Moreover, the upper Gb3Cer band gave a stronger signal compared to the adjacent lower one indicating higher quantity of Gb3Cer with a long chain N-acyl group. In contrast to Gb3Cer, BCIP-stained LacCer showed higher intensities in the lower band pointing to prevalence of LacCer species with a short chain fatty acid (Figure 3A, lane 2). The DAB positive Gb4Cer spot appeared rather as one strong band than a double band suggesting the predominance of just one species with a long chain fatty acid (Figure 3A, lane 3). Multiple TLC Immunostaining and Concomitant Multicoloring Matched with IR-MALDI-o-TOF-MS. The immunodetected GSLs, LacCer, Gb3Cer, and Gb4Cer of the triply colored chromatogram of Figure 3A (lane 3) were characterized using direct IR-MALDI-o-TOF-MS. Mass spectra were recorded from the TLC plate in the positive ion mode and are displayed in the Figures 4-6. Neutral GSLs were predominantly detected as singly charged monosodiated adducts of the [M + Na]+ type and in some cases as doubly sodiated [M + 2Na - H]+ ions. All high Analytical Chemistry, Vol. 81, No. 22, November 15, 2009

9485

Figure 3. Multiple immunostaining combined with multicoloring of TLC-separated neutral GSLs from human erythrocytes. (A) Total GSL amounts of 8 µg were applied for orcinol staining (orc) and 4 µg for cumulative immunostain. Multiple TLC immunostaining was performed by three sequential immunostaining rounds on the same chromatogram (see Figure 1) with anti-Gb3Cer (1), anti-LacCer (2), and anti-Gb4Cer antibodies (3). Fast Red-AP, BCIP-AP, and DAB-HRP stains resulted in red-, blue-, and black-colored bands of Gb3Cer, LacCer, and Gb4Cer, respectively. (B) Red-colored Gb3Cer bands were scanned at 520 nm, blue-stained LacCer at 650 nm, and black-labeled Gb4Cer at 750 nm. The predominant fatty acid chain length of the GSLs reflected by the densitometric signals are indicated.

Figure 4. TLC-IR-MALDI-o-TOF mass spectra of BCIP-AP-mediated blue-colored LacCer from a multiply immunostained chromatogram of neutral GSLs from human erythrocytes. Mass spectra were acquired from the upper (A) and lower LacCer immunopositive band (B) after triple immunostaining of the same chromatogram (see Figure 3A, lane 3). Arrowheads in the insets point to the bands targeted by IR-laser irradiation. All ions assigned represent singly charged sodium adducts of the [M + Na]+ type. A comprehensive list of all detected GSL ions including low abundant species not assigned in the spectra is provided in Table 1. Crosses (+) indicate impurities from the silica gel.

and low abundant GSL ions obtained by TLC-IR-MALDI-o-TOFMS and the corresponding proposed structures are sum9486


marized in Table 1. The detection limits of the TLC-MS analysis and the densitometric scanning were estimated on the basis of dilution series and are summarized in Table 2. Proposed structures were confirmed by electrospray ionization quadrupole time-of-flight mass spectrometry (ESI-QTOF-MS) (Supporting Information, Figure S-1), followed by collision-induced dissociation (CID) of selected singly charged [M + Li]+ GSL ions. Low-energy CID of these lithiated adducts produced fragment ions characterizing the glycan as well as the ceramide moiety.29-31 Tandem mass spectra of selected Gb4Cer species are provided as examples in the Supporting Information (Figures S-2 and S-3). Details of the ESI-MS analyses are described in the experimental section of the Supporting Information. Mass Spectra of BCIP-AP-Mediated Blue-Colored LacCer. Figure 4 shows the mass spectra obtained by TLC-IR-MALDI-o-TOF-MS after immunodetection with the anti-LacCer antibody. The major LacCer species in the faintly stained upper band were LacCer (d18: 1, C24:1/C24:0) carrying a monounsaturated and a saturated fatty acid, detected at m/z 994.72/996.74, respectively, accompanied by minor LacCer (d18:1, C22:0) at m/z 968.71 (Figure 4A). The signals at m/z 884.59 obtained from the strongly stained lower band corresponded to LacCer (d18:1, C16:0), the only LacCer species with a C16 fatty acid in the ceramide moiety (Figure 4B). The blue indigo stain 5,5′-dibromo-4,4′-dichloro-indigo yielded from BCIP upon AP treatment32 was detected as molecular ions at m/z 485.81 (not shown) but did not interfere with MS analysis of the GSLs. The mass spectra displayed in Figure 4 were acquired from 105 and 224 ng of LacCer (d18:1, C24:1/C24:0) and LacCer (d18:1, C16:0), respectively. MS detection limits of LacCer with C24 and C16 fatty acids were 52.7 and 44.8 ng, respectively, compared to 10.5 and 11.2 ng obtained by densitometric scanning (see Table 2). (29) Hsu, F. F.; Turk, J. J. Am. Soc. Mass Spectrom. 2001, 12, 61–79. (30) Hsu, F. F.; Turk, J.; Stewart, M. E.; Downing, D. T. J. Am. Soc. Mass Spectrom. 2002, 13, 680–695. (31) Bennion, B.; Dasgupta, S.; Hogan, E. L.; Levery, S. B. J. Mass Spectrom. 2007, 42, 598–620. (32) Holt, S. J.; Sadler, P. W. Proc. R. Soc. London, Ser. B 1958, 148, 495–505.

Table 1. Monoisotopic m/z Values and Proposed Structures of Neutral GSLs from Human Erythrocytes Detected by Multiple TLC Immunostain and Coupled with TLC-IR-MALDI-o-TOF-MS origina

proposed structuresb

m/z (exp.)c

m/z (calc.)

upper band

LacCer (d18:1, C22:0) LacCer (d18:1, C24:1) LacCer (d18:1, C24:0) LacCer (d18:1, C16:0) Gb3Cer (d18:1, C22:1) Gb3Cer (d18:1, C22:0) Gb3Cer (d18:1, C23:0) Gb3Cer (d18:1, C24:1) Gb3Cer (d18:1, C24:0) Gb3Cer (d18:1, C25:1) Gb3Cer (d18:1, C25:0) Gb3Cer (d18:1, C26:1) Gb3Cer (d18:1, C26:0) Gb3Cer (d18:1, C16:0) Gb4Cer (d18:1, C22:1) Gb4Cer (d18:1, C22:0) Gb4Cer (d18:1, C23:0) Gb4Cer (d18:1, C24:1) Gb4Cer (d18:1, C24:0) Gb4Cer (d18:1, C25:1) Gb4Cer (d18:1, C25:0) Gb4Cer (d18:1, C24:1)* Gb4Cer (d18:1, C24:0)* Gb4Cer (d18:1, C26:1) Gb4Cer (d18:1, C26:0) Gb4Cer (d18:1, C16:0) Gb4Cer (d18:1, C16:0)* Gb4Cer (d18:1, h22:0) Gb4Cer (d18:1, h24:1) Gb4Cer (d18:1, h24:0)

968.71 994.72 996.74 884.59 1128.74 1130.75 1144.76 1156.77 1158.78 1170.77 1172.78 1184.80 1186.79 1046.65 1331.79 1333.82 1347.83 1359.83 1361.84 1373.84 1375.85 1381.80* 1383.82* 1387.86 1389.88 1249.73 1271.73* 1349.83 1375.82 1377.85

968.70 994.72 996.73 884.61 1128.74 1130.75 1144.77 1156.77 1158.79 1170.79 1172.80 1184.80 1186.82 1046.66 1331.82 1333.83 1347.85 1359.85 1361.86 1373.86 1375.88 1381.83 1383.85 1387.88 1389.90 1249.74 1271.72 1349.83 1375.84 1377.86

lower band upper band

lower band upper band

lower band

a Antibody-mediated detection in conjunction with BCIP-AP (LacCer, see Figure 4), BCIP-Fast Red (Gb3Cer, see Figure 5), and DAB-HRP (Gb4Cer, see Figure 6). b Major GSLs in bold print; ions repeatedly detected in the lower Gb4Cer positive band due to close separation with preponderant upper band Gb4Cer are not listed as lower band Gb4Cer species. c Ions detected were sodium adducts of [M + Na]+ type with the exception of minor Gb4Cer doubly sodiated [M + 2Na - H]+ ions (labeled with asterisks).

Mass Spectra of Fast Red-AP-Mediated Red-Colored Gb3Cer. The in situ IR-MALDI-o-TOF mass spectra obtained from the upper and the lower band of the anti-Gb3Cer Fast Red-colored TLC immunostain are shown in Figure 5A,B, respectively. Gb3Cer (d18: 1, C24:1/C24:0) were the most prevalent species detected in the strong immunopositive upper band. The corresponding [M + Na]+ ions were detected at m/z 1156.77 and 1158.78, respectively, flanked by less abundant ions at m/z 1130.75, indicative of Gb3Cer (d18:1/C22:0) (Figure 5A). Traces of Gb3Cer species with uncommon fatty acids in the ceramide portion were detected

as well. These are species with very long chain fatty acids, C26:1 and C26:0, detected at m/z 1184.80 and 1186.79, respectively, as well as those with an odd number of carbon atoms, C23:0, C25:1, and C25:0, evidenced by ions at m/z 1144.76, 1170.77, and 1172.78. The ions at m/z 1046.65 obtained from the less intensively stained lower band point to the presence of Gb3Cer (d18:1, C16: 0), the only Gb3Cer species with C16 short chain fatty acid in the ceramide moiety (Figure 5B). No signals were recorded from the silica gel which could be attributed to Fast Red. The precipitated azo dye of Fast Red is soluble in chloroform and is removed following chloroform extraction of the Plexigum prior to MS analysis. Mass spectra displayed in Figure 5 were acquired from 681 and 168 ng of Gb3Cer (d18:1, C24:1/C24:0) and Gb3Cer (d18:1, C16:0), respectively. MS detection limits of 34.1 and 33.6 ng for Gb3CerCer with long and short chain fatty acids, respectively, were obtained (Table 2). In comparison to MS, immunostain detection limits were 6.8 and 8.4 ng of Gb3Cer with C24 and C16 fatty acid, respectively. Mass Spectra of DAB-HRP-Mediated Black-Colored Gb4Cer. Figure 6 shows the mass spectra of the DAB-colored anti-Gb4Cer positive bands. The major Gb4Cer species in the upper prominent band were detected as monosodiated ions at m/z 1359.83/1361.84 indicating the structure Gb4Cer (d18:1, C24:1/C24:0) (Figure 6A). In addition, both variants appeared with very low intensities as doubly sodiated [M + 2Na - H]+ ions in the spectrum (m/z 1381.80/1383.82). Low abundant ions at m/z 1333.82 indicated the presence of Gb4Cer (d18:1, C22:0). Traces of Gb4Cer variations with uncommon fatty acid substitution also were observed. These are species with very long chain fatty acids C26:1 and C26:0 detected at m/z 1387.86 and 1389.88, respectively, and with odd numbered fatty acids C23:0, C25:1, and C25:0, allocated by ions at m/z 1347.83, 1373.84, and 1375.85, respectively. The highly abundant ions at m/z 1249.73 and the doubly sodiated counterparts at m/z 1271.73, which were determined in the lower DAB-colored band, could be assigned to Gb4Cer (d18: 1, C16:0) as shown in the spectrum of Figure 6B. These major ions were flanked by low abundant ions of Gb4Cer species with C22:0 and C24:1/C24:0 fatty acids derived from the slightly separated upper band (Figure 6A). Furthermore, rare species with hydroxyl groups in the ceramides, most likely carried by the C22:0 and C24:1/C24:0 fatty acids, were detected in the lower band of immunostained Gb4Cer at m/z 1349.83 and 1375.82/1377.85, respectively. In general, GSLs with long chain h24:1 or h24:0 hydroxylated fatty acids show a more pronounced retention in

Table 2. GSL Detection Limits Obtained by Multiple TLC Immunostaining and IR-MALDI-o-TOF-MS of Major Neutral GSLs from Human Erythrocytes GSLa

TLC immunostainb [ng]

IR-MALDI-o-TOF-MSc[ng]

coloringd

LacCer (C24) LacCer (C16) Gb3Cer (C24) Gb3Cer (C16) Gb4Cer (C24) Gb4Cer (C16)

10.5 11.2 6.8 8.4 11.4 2.9

52.7 44.8 34.1 33.6 22.8 28.7

BCIP-AP BCIP-AP Fast Red-AP Fast Red-AP DAB-HRP DAB-HRP

a Neutral GSLs separate in upper (C24) and lower (C16) bands by TLC (see Figure 3). b Detection limits achieved by densitometric scanning of multiply immunostained GSL dilution series. c Detection limits obtained by direct TLC-IR-MALDI-o-TOF-MS of a multiply immunostained GSL dilution series. d Procedures according to Figure 3.


9487

Figure 5. TLC-IR-MALDI-o-TOF mass spectra of Fast Red-APstained Gb3Cer from a multiply immunostained chromatogram of neutral GSLs from human erythrocytes. Mass spectra were acquired from the upper (A) and lower (B) Gb3Cer immunopositive band after triple immunostaining of the same chromatogram (see Figure 3A, lane 3). Arrowheads in the insets point to the bands targeted by IR-laser irradiation. All ions detected are singly charged sodium adducts of the [M + Na]+ type. A comprehensive list of all detected GSL ions including low abundant species not assigned in the spectra is provided in Table 1. Crosses (+) indicate impurities from the silica gel.

TLC separation than their nonhydroxylated C24:0 or C24:1 counterparts and, therefore, comigrate with GSLs containing short chain C16:0 nonhydroxy fatty acids. DAB is changed into a highly cross-linked phenazine polymer upon HRP reaction.33,34 This black-colored precipitate did not impede in situ IR-MALDI-o-TOF-MS of GSLs. The mass spectra displayed in Figure 6 were acquired from 2.3 µg and 57 ng of Gb4Cer (d18:1, C24:1/C24:0) and Gb4Cer (d18:1, C16:0), respectively. MS detection limits obtained for Gb4Cer with long and short chain fatty acids were 22.8 and 28.7 ng, respectively, compared to immunostain detection limits of 11.4 and 2.9 ng of the corresponding Gb4Cer species, respectively, when determined by TLC scanning (Table 2). Interestingly, the Gb4Cer species with the short chain fatty acid exhibited a considerably lower detection limit by immunostaining than the respective Gb4Cer variant with long chain fatty acid. Confirmation of TLC-IR-MALDI-o-TOF-MS Proposed GSL Structures by ESI-QTOF-MS1/MS2. Proposed structures were (33) Graham, B. C.; Karnovsky, M. J. J. Histochem. Cytochem. 1966, 14, 281– 302. (34) Seligman, A. M.; Karnovsky, M. J.; Wasserkrug, H. L.; Hanker, J. S. J. Cell Biol. 1968, 38, 1–14.

9488


Figure 6. TLC-IR-MALDI-o-TOF mass spectra of DAB-HRP-mediated black-colored Gb4Cer from a multiply immunostained chromatogram of neutral GSLs from human erythrocytes. Mass spectra were acquired from the upper (A) and (B) lower Gb4Cer immunopositive band after triple immunostaining of the same chromatogram (see Figure 3A, lane 3). Arrowheads in the insets point to the bands targeted by IR-laser irradiation. Ions were predominantly detected as singly charged [M + Na]+ and to a minor extent as doubly sodiated [M + 2Na - H]+ species (marked with an asterisk). Detected ions are listed in Table 1, in addition to low abundant ions not assigned in the spectra. The proposed structure of Gb4Cer (d18:1, C23:0) represented by low abundant ions at m/z 1347.83 in the upper band (A) was confirmed by ESI-QTOF-MS2 as shown in Figure S-2 of the Supporting Information. Open circles (O) indicate Gb4Cer variants with hydroxy fatty acids in the ceramide moiety (B). The proposed structure of Gb4Cer (d18:1, h24:0) indicated by low abundant ions at m/z 1377.85 was confirmed by ESI-QTOF-MS2 as shown in Figure S-3 of the Supporting Information.

substantiated by ESI-QTOF-MS1 and MS2. MS1 ions (Supporting Information, Figure S-1) and fragment ions obtained by lowenergy CID of selected GSL species provided information on the GSL structures. Diagnostic fragment ions characterized the ceramide moieties of individual LacCer, Gb3Cer, and Gb4Cer species, consistent with the proposed structures.35,36 An example for full structural characterization is shown in Figure S-2A of the Supporting Information for Gb4Cer (d18:1, C23:0). The list of ions arising from CID of this GSL species is provided in Table S-1 of the Supporting Information. The magnified view of the low m/z region between m/z 250 and 400 revealed the characteristic fragment ions arising from cleavage of Cer (d18:1, C23:0), as depicted in Figure S-2B and illustrated by the fragmentation (35) Domon, B.; Costello, C. E. Biochemistry 1988, 27, 1534–1543. (36) Domon, B.; Costello, C. E. Glycoconj. J. 1988, 5, 397–409.

Figure 7. Multiple immunostaining combined with multicoloring of TLC separated GSLs in crude lipid extracts from healthy liver and hepatocellular cancer tissues. (A) Extracts of normal (N) and tumor (T) tissues equivalent to 2 mg of tissue wet weight of a patient suffering from hepatocellular cancer were separated and probed with cumulative immunostains. Multiple immunostaining of both tissue extracts was performed simultaneously on the same TLC plate by three sequential immunostaining rounds with anti-Gb3Cer, anti-LacCer, and anti-Gb4Cer antibodies. Fast Red-AP, BCIP-AP, and DAB-HRP stains resulted in red-, blue-, and black-colored bands of Gb3Cer, LacCer, and Gb4Cer, respectively. (B) Densitometric scans of multiply stained chromatograms of crude lipid extracts from normal (N) and tumor (T) tissue (see panel A). Red-colored Gb3Cer bands were scanned at 520 nm, blue-stained LacCer at 650 nm, and black-labeled Gb4Cer at 750 nm. (C) Bar chart with GSL expression levels for short (C16) and long (C24) chain fatty acid species of LacCer, Gb3Cer, and Gb4Cer in normal (N) and tumor (T) tissues (see panel B).

scheme in Figure S-2C of the Supporting Information. As a second example, the low-energy CID spectrum of Gb4Cer (d18:1, h24:0) obtained from the [M + Li]+ precursor ions at m/z 1361.89 is shown in Figure S-3A, and the CID derived ions are listed in Table S-2 of the Supporting Information. The magnified view of the low m/z region between m/z 250 and 750 showing characteristic fragment ions from cleavages in the lipid moiety, Cer (d18: 1, h24:0), and the corresponding auxiliary fragmentation schemes are depicted in Figure S-3B and S-3C, respectively, in the Supporting Information. Multiple TLC Immunostaining Combined with Multicoloring of GSLs in Normal and Tumor Tissues. The procedure was applied to investigate the GSL profile of human hepatocellular cancer and associated normal liver tissue. To avoid laborious GSL isolation procedures, we probed crude lipid extracts from smallsized samples of normal and cancerous tissues. The applicability of such simple GSL preparations is demonstrated in Figure 7A for a malignant and adjacent healthy tissue sample. Extract aliquots corresponding to 2 mg of tissue wet weight of normal (N) and tumor (T) tissue, respectively, were applied to a TLC plate, separated, and analyzed by three sequential immunostaining rounds using the antibodies against LacCer, Gb3Cer, and Gb4Cer. The blue stained LacCer double bands, containing most likely LacCer species with long (C24) and short chain (C16) fatty acids in the upper and lower band, respectively, indicated a higher expression in the malignant tissue, suggesting LacCer as a cancerassociated GSL in this tumor sample (Figure 7A). Evident from the triple stain, the red-colored Gb3Cer doublet revealed a slightly decreased expression in the tumor tissue, particularly with respect to the upper Gb3Cer band, presumably carrying C24 fatty acid(s). Only a single Gb4Cer band, colored in black and representing most probably the long chain (C24) fatty acid variant of Gb4Cer, was detected in both healthy and cancerous liver tissue. Its low

abundance points to expression on a generally low level in both types of tissue with a tendency to a slight decrease in the tumor tissue (Figure 7A). The differently colored Gb3Cer, LacCer, and Gb4Cer bands were successively quantified after each immunostaining round by densitometric scanning. The merged densitometric scans which were performed according to the order of immunostain performance, i.e., first Fast Red, then BCIP, and third DAB, are shown in Figure 7B. Clearly separated peaks allowed comparative quantification of individual GSLs from normal (N) and tumor (T) tissue samples. The total LacCer signal intensity (i.e., the sum of upper (C24) and lower (C16) band signals) was 2.3 times higher in the neoplastic tissue, whereas a reduced expression of Gb3Cer and Gb4Cer by factors of 0.8 and 0.6, respectively, was characteristic for the malignant tissue, as demonstrated in the bar chart of Figure 7C. This synopsis of our investigation indicates tumor association of LacCer in this single sample of malignant liver tissue. All in all, the multiple staining procedure permits a comparative qualitative and semiquantitative determination of individual GSLs of a crude lipid tissue extract in the same chromatogram. Using the triple staining procedure, only minute quantities of GSLs, only one-third of that amount usually employed for three conventional single immunostains, was necessary. Proposed GSL structures were identified on the basis of (i) the use of specific antibodies suggesting GSL variants with hypothetical ceramides carrying C24 and C16 fatty acids and (ii) their chromatographic mobility in comparison to reference GSLs. TLC-IR-MALDI-o-TOF-MS of Multiply Immunostained and Multicolored GSLs from Normal and Tumor Tissue. The TLC immunodetected GSLs and their proposed structures deduced from the chromatograms of normal (N) and tumor (T) Analytical Chemistry, Vol. 81, No. 22, November 15, 2009

9489

Figure 8. TLC-IR-MALDI-o-TOF mass spectra of BCIP-AP-mediated blue-colored LacCer from a multiply immunostained chromatogram of crude lipid extracts from healthy liver and hepatocellular cancer tissues. Mass spectra were acquired after triple immunostaining from the upper (A) and lower (C) LacCer immunopositive band of normal (N) and from the upper (B) and lower (D) LacCer immunopositive band of tumor (T) tissue (see Figure 7A). Arrowheads in the insets pinpoint to the bands analyzed by IR-laser irradiation. The detected singly charged [M + Na]+ ions are listed in Table S-4 of the Supporting Information. Low abundant ions not annotated in the spectra are also listed in the Supporting Information. The ion abundances in the mass spectra are consistent with the signal intensities of the immunostaining, indicating higher LacCer expression in malignant tissue.

tissue (Figure 7) were analyzed by direct IR-MALDI-o-TOF-MS to obtain further structural data of the oligosaccharide and lipid moieties. For the anti-LacCer bands, low abundant [M + Na]+ ions of LacCer species with long and short chain fatty acids were detected in the upper and lower band of the normal (N) tissue as shown in Figure 8A,C, respectively. The spectrum of Figure 8A obtained from the upper band exhibits ions with similar signal intensities at m/z 994.70/996.71 and 968.68 that could be assigned to LacCer (d18:1, C24:1/C24:0) and LacCer (d18:1, C22:0). Low abundant ions at m/z 884.61 detected from the lower band were indicative for LacCer (d18:1, C16:0) (Figure 8C). In contrast to normal tissue, [M + Na]+ ions obtained from the immunopositive LacCer bands of the tumor (T) tissue appeared as very strong signals (Figure 8B,D). Ions with m/z values of 994.73/996.74 accompanied by low abundant signals at m/z 968.72 in the spectrum of Figure 8B were assigned to LacCer (d18:1, C24:1/C24:0) and LacCer (d18:1, C22:0), respectively. Strong signals of ions at m/z 884.62 of the lower band of tumor (T) tissue were indicative of LacCer (d18:1, C16:0) (Figure 8D). Furthermore, the mass spectra also reflected the increased expression level of LacCer species, consistent with the immunostaining results. All ions and the proposed structures obtained from TLC-IR-MALDI-o-TOF-MS analyses of normal and tumor tissue are listed in Table S-4 of the Supporting Information. The various Gb3Cer species detected by Fast Red-AP antiGb3Cer immunostaining (Figure 7) also were identified by direct TLC-IR-MALDI-o-TOF-MS. However, the mass spectra revealed very similar intensities of these species with only marginal qualitative and quantitative differences in the neoplastic and 9490


adjacent normal tissues (data not shown). Detected ions and proposed structures of Gb3Cer variants are summarized in Table S-4 of the Supporting Information. Anti-Gb4Cer staining of both normal (N) and tumor (T) tissue extracts showed less structural variability in the ceramide moiety of Gb4Cer species (Figure 7). The mass spectrum of Figure 9A recorded from the faint immunopositive Gb4Cer species of the normal (N) tissue provided evidence for the predominant expression of Gb4Cer (d18:1, C24:1/C24:0) in addition to the presence of low abundant Gb4Cer (d18:1, C23:0) and Gb4Cer (d18:1, C22: 0), respectively. A less complex mass spectrum was obtained for the Gb4Cer positive species of the tumor (T) tissue. Gb4Cer (d18:1, C24:1/ C24:0) ions at m/z 1359.86/1361.87 were detected as the prevalent Gb4Cer species in the extract of the cancerous tissue (Figure 9B). For a synopsis of all GSL ions obtained from TLC-IR-MALDIo-TOF-MS analyses of normal and tumor tissue and the proposed Gb4Cer structures, refer to Table S-4 of the Supporting Information. In summary, the detection and structural characterization of low GSL quantities directly from the TLC plate confirmed the high sensitivity of the combined TLC immunostaining MS technique. Futhermore, we demonstrated the robustness of this method by the use of it to detect differences in GSL profiles between crude lipid extracts from malignant vs healthy tissues. DISCUSSION Using different chromogenic substrates with distinct absorption characteristics the TLC immunoenzyme staining was improved with respect to the amount of information obtained from a single

Figure 9. TLC-IR-MALDI-o-TOF mass spectra of DAB-HRP-mediated black-colored Gb4Cer from a multiply immunostained chromatogram of crude lipid extracts from healthy liver and hepatocellular cancer tissues. Mass spectra were acquired from the Gb4Cer immunopositive band after triple immunostaining of single chromatograms of normal (N) and tumor (T) tissue tissue (A and B, respectively; see Figure 7A). Arrowheads in the insets mark the bands analyzed by IR-laser irradiation. Detected major singly charged [M + Na]+ and minor doubly sodiated [M + 2Na - H]+ ions are listed in Table S-4 of the Supporting Information. Crosses (+) and asterisks (*) indicate impurities from the silica gel and doubly sodiated ions, respectively.

TLC run. This method allowed the detection of different GSLs in a single chromatogram. The enzyme-substrate systems introduced to TLC overlay analysis were BCIP-AP, Fast Red-AP, and DAB-HRP. These reagents permitted the visual discrimination and specific densitometric detection of blue-, red-, and black-colored GSL bands, respectively, as shown for distinctly stained LacCer, Gb3Cer, and Gb4Cer. One technical issue that needed to be addressed was the use of polyclonal chicken IgY anti-LacCer, anti-Gb3Cer, and Gb4Cer antibodies for three consecutive immunostaining rounds. Their sequential employment in addition to the repetitive use of AP as a reporter enzyme raised the problem of carry-over effects in the second and third immunostaining circle. Due to the binding of the secondary antibody in a succeeding assay to the primary antibodies of a preceding immunostaining round, additionally colored precipitates would be generated at sites of previously detected GSLs. Therefore, heat denaturation of the primary antibody and associated secondary antibody of each preceding immunostaining round as well as concomitant heat inactivation of antibody-linked enzymes was crucial to inhibit cross-reactivity. This strategy permits the successive use of primary anti-GSL antibodies raised in the same species as shown for IgY chicken antibodies. However, sequential multiple TLC immunoenzyme staining with anti-GSL antibodies of different Ig-subtypes (IgG or IgM) and/or from different animal species (mouse, rat, rabbit, etc.) should be possible, too. In this case, it seems likely that heat treatment could be omitted, but this has not been demonstrated yet in the TLC overlay assay. As an alternative to heat treatment, we also examined the chelating agent EDTA for its capability to irreversibly inactivate the alkaline phosphatase and we screened the efficacy of proteolytic digestion of the antibodies. However, neither EDTA nor trypsin treatments inhibited “cross-coloring” and, moreover, caused the silica gel layer to peel off, even after plastic fixation. Most importantly, the heat treatment did not compromise the accessibility and structural integrity of GSLs. Comparing TLC overlay assays with and without heat treatment showed identical immunostain intensities, indicating reliable detection and quantification of GSLs after heat treatment. In addition, heat treatment stabilized the glass-backed silica gel layer; thus, no flaking of the silica could be observed even after overlong incubation of TLC plates in aqueous solutions. On the basis of

these observations, intermediate heat treatment after the first and the second immunostaining round was found to be the most suitable method to efficiently eliminate carry-over effects. The heat transfer of primulin-stained GSLs from the TLC plate to a PVDF membrane (“far-eastern blotting”) followed by secondary ion mass spectrometry of GSLs directly on the membrane has been previously introduced.37 This combined TLC blottingMS technique enabled purification of individual GSLs separated on a TLC plate and structural determination of about 1 µg of a GSL in limited biological samples when combined with direct MS analysis.38,39 The GSL transfer is performed with a household iron at 180 °C, and the GSL spots can be punched out followed by MS analysis. The suitability of this technique has been further demonstrated by several research groups40-42 indicating that heat transfer does not diminish the detection of GSL epitopes and, most importantly, does not influence their structural integrity as shown by immunological and MS analyses (also reviewed by Yu and Ariga43 as well as Mu¨thing and Distler12). These conclusions are in agreement with our findings and affirm the heat resistance of GSLs. One limitation of the TLC based procedure is that closely migrating GSLs, such as Gb3Cer and Lc3Cer, are poorly resolved using TLC standard solvents composed of chloroform, methanol, and water. Thus, if required, modifications of standard TLC might overcome this resolution problem by switching to automated multiple unidirectional44 or two-dimensional TLC using different separation solvents.45-47 (37) Taki, T.; Ishikawa, D.; Handa, S.; Kasama, T. Anal. Biochem. 1995, 225, 24–27. (38) Taki, T.; Ishikawa, D. Anal. Biochem. 1997, 251, 135–143. (39) Ishikawa, D.; Taki, T. Methods Enzymol. 2000, 312, 145–157. (40) Johansson, L.; Miller-Podraza, H. Anal. Biochem. 1998, 265, 260–268. (41) Johansson, L.; Johansson, P.; Miller-Podraza, H. Anal. Biochem. 1999, 267, 239–241. (42) Guittard, J.; Hronowski, X. L.; Costello, C. E. Rapid Commun. Mass Spectrom. 1999, 13, 1838–1849. (43) Yu, R. K.; Ariga, T. Methods Enzymol. 2000, 312, 115–134. (44) Mu ¨ thing, J. J. Chromatogr. B Biomed. Appl. 1994, 657, 75–81. (45) Sonnino, S.; Ghidoni, R.; Chigorno, V.; Masserini, M.; Tettamanti, G. Anal. Biochem. 1983, 128, 104–114. (46) Nakamura, K.; Ariga, T.; Yahagi, T.; Miyatake, T.; Suzuki, A.; Yamakawa, T. J. Biochem. 1983, 94, 1359–1365. (47) Yohe, H. C.; Wallace, P. K.; Berenson, C. S.; Ye, S.; Reinhold, B. B.; Reinhold, V. N. Glycobiology 2001, 11, 831–841.


9491

The specificity of the antibody and subsequent mass spectrometric analysis provide complementary information on the glycan as well as the ceramide structure of GSLs. Structural elucidation in such an extent would otherwise not be attainable by TLC overlay assay alone.48,49 Therefore, merging multiple TLC immunostaining and multicoloring with IR-MALDI-o-TOF-MS was a second significant innovation of our novel strategy. As already observed in previous investigations,11,12 the complex of primary anti-GSL antibodies together with the AP-conjugated secondary antibodies and the blue BCIP precipitate did not impede MS analysis. Similarly, MS analysis was possible after coloring GSL bands with AP- and HRP-labeled secondary antibodies using Fast Red and DAB staining, respectively. Furthermore, heat treatment did not compromise the accessibility and structural integrity of GSLs as confirmed by direct IR-MALDI-o-TOF-MS analysis of GSLs on the TLC plate after a triple immunostaining round (discussed above). The silica gel is a potential source of impurities such as salts, binding material from the production process, or other contaminants derived from the TLC overlay assay procedure. Signals of impurities were detected in mass spectra of immunostained GSLs that reflect the presence of polyethylene glycol based polymers. We suppose that traces of such hydrophilic contaminants in the incubation buffers could be the source and might tend to adhere to the silica gel over the repetitive incubation steps. Although chemicals with the highest purity were used and detergents, such as polysorbates (Tween) that contain polyoxyethylene side chains were omitted, signals of the contaminants appeared in the spectra but in a mass region that did not interfere with the mass region of GSL analysis. With our setup, MS detection limits were in the range between 20 and 50 ng for specific GSL species. This result is similar to those obtained by Nakamura et al.50 performing UV-MALDIquadrupole ion trap (QIT)-TOF-MS. BCIP-AP-, Fast Red-AP-, and DAB-HRP-mediated coloring resulted in similar MS detection limits upon immunoenzyme detection with a trend toward higher sensitivities for more complex GSLs with long glycan chains. We (48) Meisen, I.; Peter-Katalinic´, J.; Mu ¨ thing, J. Anal. Chem. 2003, 75, 5719– 5725. (49) Meisen, I.; Peter-Katalinic´, J.; Mu ¨ thing, J. Anal. Chem. 2004, 76, 2248– 2255. (50) Nakamura, K.; Suzuki, Y.; Goto-Inoue, N.; Yoshida-Noro, C.; Suzuki, A. Anal. Chem. 2006, 78, 5736–5743. (51) Vukelic´, Z.; Zamfir, A. D.; Bindila, L.; Froesch, M.; Peter-Katalinic´, J.; Usuki, S.; Yu, R. K. J. Am. Soc. Mass Spectrom. 2005, 16, 571–580. (52) 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. (53) Li, Y.; Teneberg, S.; Thapa, P.; Bendelac, A.; Levery, S. B.; Zhou, D. Glycobiology 2008, 18, 158–165. (54) Li, Y.; Zhou, D:; Xia, C.; Wang, P. G.; Levery, S. B. Glycobiology 2008, 18, 166–176. (55) Kirsch, S.; Zarei, M.; Cindric´, M.; Mu ¨ thing, J.; Bindila, L.; Peter-Katalinic´, J. Anal. Chem. 2008, 80, 4711–4722. (56) Kirsch, S.; Mu ¨ thing, J.; Peter-Katalinic´, J.; Bindila, L. Biol. Chem. 2009, 390, 657–672. (57) Hummel, I.; Klappe, K.; Kok, J. W. FEBS Lett. 2005, 579, 3381–3384. (58) Martin, S. F.; Williams, N.; Chatterjee, S. Glycoconj. J. 2006, 23, 147–157. (59) Chatterjee, S.; Pandey, A. Biochim. Biophys. Acta 2008, 1780, 370–382. (60) Chatterjee, S.; Kolmakova, A.; Rajesh, M. Curr. Drug Targets 2008, 9, 272– 281. (61) Levery, S. B. Methods Enzymol. 2005, 405, 300–369. (62) Feizi, T; Chai, W. Nat. Rev. Mol. Cell Biol. 2004, 5, 582–588. (63) Liu, Y.; Palma, A. S.; Feizi, T. Biol. Chem. 2009, 390, 647–656.

9492


are currently developing an IR-MALDI source coupled to tandem mass spectrometer such as a high resolution Fourier transform ion cyclotron resonance mass spectrometer.51 Such instruments would allow more precise structural characterization due to improved resolving power and mass accuracy as well as fragmentation capabilities.52-54 In comparison to HPLC coupled with MS,55,56 TLC is less easily automated. However, this method provides the opportunity to analyze many samples simultaneously. Furthermore, it is robust and crude lipid extracts can be analyzed without tedious purification of GSLs. Most importantly, in contrast to HPLC based techniques, TLC immunodetection matched with direct IR-MALDIo-TOF-MS allows some structural elucidation due to employment of oligosaccharide specific antibodies. The technique described here was successfully applied to the GSL compositional analysis of human tissue samples. Even minor GSL species of crude lipid extracts derived from minute amounts of tissue could be monitored. Comparison of the GSL composition of healthy and malignant tissue revealed an enrichment of LacCer and a decrease of Gb3Cer and Gb4Cer in cancerous liver tissue. However, additional studies are needed in order to elucidate the statistical significance of this finding. Interestingly, an upregulation of LacCer synthase in human liver tumor cells has been reported previously.57 Involvement of LacCer in the regulation of cell proliferation and cell adhesion58,59 implies that this GSL and its synthase could be potential drug targets.60 CONCLUSIONS Multiple TLC immunoenzyme staining in combination with multicoloring and IR-MALDI-o-TOF-MS is a powerful method that contributes to the emerging field of glycosphingolipidomics.61 Future studies should examine ways to automatize and shorten the procedure, e.g., by reducing incubation times of primary as well as secondary antibodies. Functional glycomics is now being explored through the use of carbohydrate microarrays that employ neoglycolipid technology in order to decipher the information content of the glycome.62,63 In this context, our technology may represent a key starting point for interfacing the glyco arrays with mass spectrometry. ACKNOWLEDGMENT This work has been funded by the Deutsche Krebshilfe (Grant DKH 106742 to J.M.) and was supported by the Deutsche Forschungsgemeinschaft (DFG, Grant DR416-5/1). Current financial support is provided by the EU GLYFDIS project (Grant LSHB-CT-2006-037661 to J.P.-K.). The authors thank Dr. Stefan Berkenkamp from Sequenom GmbH (Hamburg, Germany) for providing use of the o-TOF instrument and Dr. Joanne Yew for critical revision of the manuscript. SUPPORTING INFORMATION AVAILABLE Additional information as noted in the text. This material is available free of charge via the Internet at http://pubs.acs.org.

Received for review August 28, 2009. Accepted October 6, 2009. AC901948H

Structural Profiling of Individual Glycosphingolipids in a Single Thin

Recommend Documents