Sensitive Liquid Chromatography-Electrospray Mass Spectrometry

Dec 15, 2008 - The sensitive and specific method was applied to the analysis of the O-glycosylation of MUC1 in breast, prostate and gastric cancer, in...
0 downloads 0 Views 1MB Size
Sensitive Liquid Chromatography-Electrospray Mass Spectrometry Allows for the Analysis of the O-Glycosylation of Immunoprecipitated Proteins from Cells or Tissues: Application to MUC1 Glycosylation in Cancer Malin Ba¨ckstro ¨ m,* Kristina A. Thomsson,* Hasse Karlsson,* and Gunnar C. Hansson* Institute of Biomedicine, Department of Medical Biochemistry, University of Gothenburg, Sweden Received September 5, 2008

We have analyzed the structures of the glycans on immunoprecipitated proteins from small amounts of cell or tissue lysates, by liquid-chromatography electrospray mass spectrometry (LC-ESI-MS) and MS/MS. The sensitive and specific method was applied to the analysis of the O-glycosylation of MUC1 in breast, prostate and gastric cancer, including analysis of a patient tumor specimen. The method will be applicable for the glycosylation analysis of individual proteins. Keywords: Mucin • MUC1 • CD43 • Leukosialin • oligosaccharide • O-glycans • immunoprecipitation • breast cancer • gastric cancer • prostate cancer

Introduction The functions of many proteins are dependent on posttranslational modifications, where glycosylation is among the most common. The structural analysis of glycans has in recent years received increasing biological and medical attention, as carbohydrates have been implicated in several biological processes and important diseases, including in the immune system1,2 and in cancer.3 It is therefore of increasing importance to be able to analyze the structure of both N- and O-glycans on different proteins from small amounts of biological material. N-glycans have been more frequently studied, due to the ease of their tagging with fluorescent dyes, which has enhanced their detection. In contrast, the analysis of O-glycans has been particularly challenging, due to the requirement of chemical release in reducing conditions, blocking further derivatization at the reducing end. The amounts needed for the structural determination of oligosaccharides have decreased drastically in recent years, much thanks to technical developments within both liquid chromatography (LC) and mass spectrometry (MS). In cases where several milligrams of protein were needed ten years ago, it is now possible to get good structural information of the glycans from microgram amounts of purified glycoproteins using LC-ESI-MS.4 Sensitive MS analyses of glycans are also often performed in the positive ion mode after permethylation of the hydroxyl groups. In this way it is possible to make a total glycome analysis of low numbers of whole cells.5,6 In this study, we have analyzed nonderivatized O-linked oligosaccharides with LC-ESI-MS in negative ion mode, an approach originally developed by Karlsson and coworkers.7 By this we avoid the potential problems associated with the derivatization step and we can easily analyze nega* To whom correspondence should be addressed: Institute of Biomedicine, Department of Medical Biochemistry, University of Gothenburg, Box 440, 405 30 Gothenburg, Sweden; [email protected]

538 Journal of Proteome Research 2009, 8, 538–545 Published on Web 12/15/2008

tively charged oligosaccharides with multiple sialic acid and sulfate groups. The sensitivity of our approach is in the high attomole/low femtomole range and here we show that this is sufficient to analyze the structures of oligosaccharides on specific, endogenous proteins that have been purified by immunoprecipitation (i.p.) from a limited number of cells. Immunoprecipitated proteins from lysates of cells or tissues were separated by gel electrophoresis and blotted onto PVDF membranes. Individual protein bands were cut out and the O-glycans chemically released from the blots and analyzed in negative ion mode with LC-ESI-MS. We first developed and verified the method for the analyses of the O-glycans of the mucin-like membrane glycoprotein CD43. It was then applied to the O-glycosylation of MUC1 mucin, a membrane glycoprotein with a variable number of tandem repeats, each with five potential O-glycosylation sites.8 MUC1 has been implicated in cancer development, both through adhesive and antiadhesive properties caused by its glycans and also through involvement in signaling processes through its cytoplasmic tail.9 We analyzed the O-glycosylation of MUC1 from several cancer cell lines, and also from a gastric cancer surgical specimen, showing the applicability of the method also for patient samples.

Materials and Methods Cell Lines and Lysates. The breast cancer cell line T47D, the nontumor mammary epithelial cells line MCF-10A,10 the prostate carcinoma cell line DU-145 and the chronic myelogenous leukemia cell line K562 were all obtained from ATCC. Cell lines were cultured in Iscove’s modified Dulbecco’s Medium (Lonza Biologicals) supplemented with 10% fetal bovine serum (Lonza Biologicals), except MCF-10A which was cultured in Dulbecco’s Modified Eagle’s Medium:F-12 (1:1) supplemented with 5% horse serum, 20 ng/mL epidermal growth factor, 0.1 µg/mL cholera toxin, 10 µg/mL insulin and 10.1021/pr800713h CCC: $40.75

 2009 American Chemical Society

Glycosylation Analysis of Immunoprecipitated Proteins

research articles

Figure 1. Presentation of the combined immunoprecipitation-glycan release-LC-MS method used in this paper. (A) Schematic presentation of the work-flow. (B) Western blot of immunoprecipitated CD43 from K562 cells. The middle part of the membrane, with a molecular mass marker and parts of the neighboring lanes, was developed with anti-CD43-4D2 and goat-antimouse-Ig-AP. The outer parts of the membrane were left unstained and were used to release O-glycans. (C) Capillary LC-ESI-MS combined mass chromatogram (for m/z 675-, 966- and 4832-) of O-glycans released from CD43 from cut-out membrane pieces shown in B. I.p. of CD43 was from lysates of either 106, 5 × 105 or 2 × 105 K562 cells. The relative abundance is in all three mass chromatograms relative to the peak height in the top chromatogram. (D) LC-ESI-MS combined mass chromatogram (for m/z 675-, 966- and 4832-) for O-glycans released from membrane pieces from the lane without cell lysate shown in B. Lower mass chromatogram shows O-glycans released from Ig H chain from anti-CD43-4D2 (50 kD) and the upper chromatogram shows O-glycans released from around 100 kD. The relative abundance is in both mass chromatograms relative to the peak height in the lower chromatogram. The structures of the glycans were elucidated as described in Materials and Methods. The 6-linked branch is shown in bold.

0.5 µg/mL hydrocortisone. All cell lines were free from mycoplasma, as detected with MycoAlert (Lonza Biologicals).

Cell lysates were prepared from confluent 15 cm Petri dishes using 1 mL lysis buffer containing 50 mM Tris-HCl pH 8.0, 150 Journal of Proteome Research • Vol. 8, No. 2, 2009 539

research articles

Ba¨ckstro ¨ m et al.

Figure 2. O-glycans released from MUC1 from the normal mammary epithelial cell line MCF-10A. (A) NanoLC-ESI-MS base peak (top panel) and individual mass chromatograms for the m/z indicated to the right. (B) MS/MS spectrum showing fragmentation of the oligosaccharide alditol observed at m/z 8472-. The structures of the glycans were elucidated as described in Materials and Methods. The 6-linked branch is shown in bold.

mM NaCl, 1% Triton X-100 and Complete, EDTA-free protease inhibitor cocktail (Roche) according to the manufacturer’s instructions. Lysed cells were then centrigfuged at 16 000× g for 10 min and the supernatants stored at -20 °C for later analysis. Tumor Sample. A tumor surgical specimen was obtained from a 72 year old male with a low-differentiated tubular gastric 540

Journal of Proteome Research • Vol. 8, No. 2, 2009

adenocarcinoma in the antrum, which was of intestinal type, with informed consent and with permission from the regional ethical committee in Western Sweden. A tumor lysate was prepared by homogenizing the tumor in lysis buffer as above, using an Ultra-Turrax T8 (Ika-Werke, Staufen, Germany), followed by incubation with stirring at 4 °C for 30 min and centrifugation as above.

research articles

Glycosylation Analysis of Immunoprecipitated Proteins

Figure 3. O-glycans released from MUC1 from the breast cancer cell line T47D. NanoLC-ESI-MS base peak (top panel) and individual mass chromatograms for the m/z indicated to the right. Nonoligosaccharide peaks are depicted with *. The structures of the glycans were elucidated as described in Materials and Methods. The 6-linked branch is shown in bold.

Antibodies. The glycosylation-independent anti-CD43 monoclonal antibody 4D2 was previously raised in our laboratory.11 Monoclonal antibodies against MUC1 were: CT2 (a gift from Dr Sandra Gendler, Mayo Clinic, Scottsdale, AZ), 214D4 (a gift from Dr John Hilkens, The Netherlands Cancer Institute, Amsterdam, The Netherlands) and HMFG-1 (a gift from Dr Joyce Taylor-Papadimitriou, Cancer Research UK, London, UK). Immunoprecipitation and Immunoblotting. Cell lysates containing 0.3-0.9 mg of total protein, as determined by the BCA assay (Pierce Biotechnology), or a tumor lysate corresponding to 24 mg wet weight of tumor, were precleared with 20 µL Protein G-PLUS agarose beads (SantaCruz Biotechnology) for 1 h at 4 °C. Two µg antibody was then added to the cleared lysates and incubated for 2-3 h at 4 °C. 20-50 µL Protein G-PLUS agarose beads were then added and the incubation continued for 16-20 h. Beads were washed 3-4 times with 200 µL 50 mM Tris-HCl pH 7.4, 150 mM NaCl containing 0.09% NP40 and finally resuspended in 30-50 µL 2× Laemmli sample buffer with 200 mM dithiothreitol. Samples were heated for 5 min at 95 °C before loaded into wells of sodium dodecyl sulfatepolyacrylamide gels (SDS-PAGE). After the separation, proteins were blotted to Immobilon-P membranes (Millipore) using semidry blotting in a Transblot SD (Bio-Rad) in 48 mM Tris, 39 mM glycin, 1.3 mM SDS and 10% methanol at 110 mM for 1 h. Lanes with samples were divided by cutting with a pair of scissors leaving a part of the lane to be stained with antibodies for immunolocalization of the protein of interest. Membranes were incubated with anti-CD43-4D2 or anti-MUC1-214 D4 antibody in PBS+5% milk+0.1% Tween-20 for 16-20 h at 4 °C, washed with PBS+0.1% Tween-20 and then incubated with goat-antimouse Ig-alkaline phophatate (AP) (Southern Biotech, Birmingham, AL) for 1 h at 20 °C. After washing, blots were developed with 5-bromo-4-chloro-3-indolyl-phosphate/ nitro blue tetrazolium (BCIP/NBT, Promega). The main part of the lane was kept unstained for release of carbohydrates.

Release and Purification of Glycans. Membrane pieces containing the protein of interest, or control membrane pieces, were cut out from PVDF membranes, cut into small pieces and put in wells of 96-well plates. O-glycans were released by β-elimination using KOH/NaBH4 and analyzed as oligosaccharide alditols, as previously described.7 HPLC Columns. HPLC columns were packed in-house into fused silica capillaries (Polymicro Technologies, Phoenix, AZ, USA) with 5 µm porous graphite particles (Hypercarb, ThermoHypersil, Runcorn, UK). Capillary-LC-MS and MS/MS. The capillary-LC-MS method is described elsewhere,12 with minor changes. Briefly, the oligosaccharides (2 µL) were injected in the capillary column (20 cm × 180 µm i.d) and eluted at a flow rate of around 3 µL/min using a linear acetonitrile gradient containing 8 mM NH4HCO3. The mass spectrometer was a LTQ linear quadrupole ion trap mass spectrometer (Thermo Electron, San Jose´, CA). NanoLC-MS and MS/MS. A binary HPLC pump (Agilent 1100, Agilent Technologies, Palo Alto, CA) delivered a flow rate of 200 µL/min which was splitted down in a T (Valco) before the injector, allowing 400 nl/min through the analytical column. The sample (1 µL) was injected onto the nano column (20 cm × 75 µm i.d) and eluted with a gradient of 0-50% B (A: 0.04% NH3 in H20; B: 100% acetonitrile). The nanoLC-MS interface was modified in-house and is similar to the one described by Shen et al.13 The heated capillary was kept at 200 °C, the capillary voltage -50 V and the electrospray voltage -1.6 kV. Full-scan (m/z 380 - 2000) and data dependent MS/MS scans were acquired of the most abundant ions in each scan. Normalized collision energy was 35%. The data were processed using the Xcalibur version 2.0 (Thermo Electron, San Jose´, CA). Elucidation of the Structures of Glycans. The structure of each glycan was elucidated based on the molecular mass and the MS/MS fragment ions in combination with the retention time compared to previously characterized glycans4,14 and previous knowledge of commonly found core structures in cancer-associated MUC1.15-17 Hexoses were always interpreted as Gal and N-acetylhexosamines as GlcNAc, unless at the inner core position where it is interpreted as GalNAc. Deoxyhexoses were interpreted as fucose. The 6-linked branch is always written in bold.

Results and Discussion Sensitivity and Specificity of the O-Glycan Analysis of Immunoprecipitated CD43. To make the structural analysis of the glycans present on individual proteins possible, we used a combination of i.p. of specific proteins, release of glycans from protein bands of interest and LC-ESI-MS analysis of the released glycans. The approach is schematically described in Figure 1A. To test the sensitivity and the specificity of the method, we first analyzed endogenous CD43 in K562 cells. As shown in Figure 1B, immunoprecipitated CD43 was separated in SDS-PAGE and blotted to a PVDF membrane. The middle part of the membrane, containing the marker and smaller parts of the adjacent lanes, were stained for CD43, while the rest of the adjacent lanes were kept unstained and used to cut out the CD43-band at around 100 kD (left lane, i.p. with K562 lysate) or a membrane piece at the same migration serving as a negative control (right lane, i.p. with no cell lysate). In order to estimate the number of cells needed for this type of analysis, CD43 was immunoprecipitated from decreasing numbers of K562 cells. LC-ESI-MS analysis of the O-glycans released from CD43 immunoprecipitated from a lysate containing 106 cells Journal of Proteome Research • Vol. 8, No. 2, 2009 541

research articles

Ba¨ckstro ¨ m et al.

Figure 4. O-glycans released from MUC1 from the prostate cancer cell line DU-145. (A) Western blot of immunoprecipitated MUC1. (B) NanoLC-ESI-MS base peak (top panel) and individual mass chromatograms for the m/z indicated to the right. Contaminating components depicted with * are of noncarbohydrate origin as their MS/MS spectra showed no typical carbohydrate fragment ions. (C) Combined mass chromatograms for all the oligosaccharides detected in B, after i.p. with the two different anti-MUC1 antibodies CT2 and HMFG-1. The structures of the glycans were elucidated as described in Materials and Methods. The 6-linked branch is shown in bold.

showed three core 1 O-glycans: Galβ3(NeuAcr6)GalNAc-ol (m/z 675-), NeuAcR3Galβ3GalNAc-ol (m/z 675-) and NeuAcR3Galβ3(NeuAcr6)GalNAc-ol (m/z 966-, 4832-) (Figure 1C, top panel). Also for 5 × 105 cells, the same three oligosaccharide alditols were observed, but with lower intensity (Figure 1C, middle panel). When even fewer cells were used, no oligosaccharide components could be detected (Figure 1C, lower panel), indicating that the detection limit had been reached. The lack of oligosaccharides from 2 × 105 cells is most likely explained by a low efficiency in the i.p. step, rather than in the analytical step. Immunoprecipitation is a rather inefficient technique that needs optimization and our experience is that less than 50% of the protein is actually pulled down by the antibody, leaving a lot of protein in the supernatant. Improvements can probably be made at this step, to increase the sensitivity even further and allow for analysis of even fewer cells. To ascertain that the detected O-glycans originated from CD43, and not from the antibody used for the i.p., a membrane piece at 50 kD, containing the immunoglobulin (Ig) heavy (H) chain of the anti-CD43 antibody, was also cut out for analysis. It was found that O-glycans were released also from the Ig H chain and that the oligosaccharides were the same as those found on CD43 (Figure 1D, lower panel). O-glycans are sometimes present on IgG and we have detected O-glycans on several, but not all, monoclonal antibodies tested. This has to be taken into account when using i.p. in combination with sensitive LC-MS analyses, as the antibody can tail over long distances in the gel. One has to be very careful to make sure that the glycans detected are not released from the i.p. antibody instead of the protein of interest. However, in this case the glycans released from the membrane piece cut out at around 542

Journal of Proteome Research • Vol. 8, No. 2, 2009

100 kD in the same lane (at the CD43-corresponding height) none or very small amounts of glycans could be detected (Figure 1D, top panel). The O-glycans on Ig H chain will thus not disturb the glycan analysis of other proteins when they are sufficiently well separated in the gel. O-Glycans on MUC1 in Normal Mammary Epithelium and Breast Cancer. The alteration of MUC1 O-glycosylation in breast cancer compared to that of normal mammary MUC1 is well-established and consists of a switch toward shorter, more sialylated core 1 structures in the cancer.18 The source for determining the glycans on normal MUC1 has been human milk, or the noncarcinoma mammary cell lines MTSV1-7.19,20 We used another normal mammary cell line, MCF-10A, and compared its glycosylation of MUC1 to that of the breast cancer cell line T47D, using i.p. from a lysate containing as little as 0.33 mg of total proteins. In MCF-10A, we found 4-5 components in the base peak LC-MS chromatogram, with the dominating one corresponding to the core 2 structure NeuAcR3 Galβ3(NeuAcr3Galβ3/4GlcNAcβ6)GalNAc-ol (m/z 1331-, 6652-, Figure 2A, top panel) and only smaller amounts of core 1 structures. The structures of these oligosaccharides, and also several others that were not detectable in the main chromatogram, were all deduced by a combination of the fragmentation pattern observed in MS/MS, retention time and previous knowledge of the biosynthetic pathways of O-glycans. In total, we could detect twelve different oligosaccharides, ranging from tri- to octasaccharides, where several were isomers. The largest oligosaccharide detected in MCF-10A was a disialylated octasaccharide with the molecular mass of 1696 (m/z 1695-, 8472-), the MS/MS spectrum of which is shown in Figure 2B. The dominance of core 2 structures in MCF-10A was in contrast to the breast cancer cell line T47D, where the vast

Glycosylation Analysis of Immunoprecipitated Proteins

research articles

Figure 5. O-glycans released from MUC1 from a gastric tumor. (A) Capillary LC-ESI-MS base peak chromatogram of oligosaccharide alditols released from MUC1 (300 kD) immunoprecipitated from the tumor lysate (large chromatogram) or from a membrane piece at 300 kD in a lane with no lysate. Contaminating components depicted with * are of noncarbohydrate origin as their MS/MS spectra showed no typical carbohydrate fragment ions. (B) Individual mass chromatograms for the m/z indicated to the right. (C) Oligosaccharide alditols that were present in lower amounts and confirmed by MS/MS. The structures of the glycans were elucidated as described in Materials and Methods. The 6-linked branch is shown in bold.

majority of oligosaccharide alditols were of core 1 type (m/z 675- and m/z 966-, 4832-, Figure 3). We could however also detect small amounts of core 2 glycans in T47D, including one fucosylated oligosaccharide of Lewis type (m/z 1477-, 7382-, Figure 3), that was also detected in MCF-10A. T47D has, in contrast to other breast cancer cell lines, been reported to have only core 1-structures,21-23 but it is obvious that by using more sensitive methods, some core 2-based oligosaccharides can also be detected. By using a mobile phase with a high pH (ammonium hydroxide) for the nanoLC-MS, we get a very high sensitivity for negatively charged oligosaccharides (K.A. Thomsson et al., to be published elsewhere). This made it possible to detect the core 2 oligosaccharides in T47D, which are most likely present in very low amounts. O-Glycans on MUC1 in Prostate Cancer. The expression and glycosylation of MUC1 in prostate cancer has not, in contrast to breast cancer, been extensively studied. There are several indications that MUC1 is overexpressed also in this cancer form,24,25 with one report on increased sialylation of MUC1 O-glycans that has a role in tumor progression,26 but data are limited and only obtained by antibody staining. We have now analyzed the structures of the O-glycans of MUC1 in the prostate cancer cell line DU-145, which is derived from a brain metastasis. This cell line shows a strong MUC1 expression in flow cytometry analysis with several anti-MUC1 antibodies (data not shown). In the base peak LC-MS chromatogram of the O-glycans released from MUC1 from this cell line, we found that core 2 structures dominated, with NeuAcR3Galβ3(NeuAcr3Galβ3/ 4GlcNAcβ6)GalNAc-ol being the most prominent (m/z 1331-, 6652-, Figure 4B, top panel). A total of ten different oligosaccharides could be detected in individual mass chromatograms and were verified by MS/MS fragmentation. The majority of the compounds were sialylated core 2 structures. In addition, one fucosylated oligosaccharide, present in low amounts, was also detected (m/z 1477-, 7382-). This pattern of oligosaccharides was similar to that for MUC1 in normal mammary epithelium, represented by MCF-10A, and maybe suggests that there is less alterations in glycosylation in prostate cancer in contrast to that seen for breast cancer. More information is however needed, both from other prostate tumors and also from the normal prostate, to allow any conclusions on the glycosylation changes on MUC1 in prostate cancer. We also compared the carbohydrates released from MUC1 that had been immunoprecipitated from DU-145 with two different antibodies, CT2 and HMFG-1. CT2 is directed against the cytoplasmic tail of MUC1 and is thus noncarbohydrate dependent in its reactivity. HMFG-1 is on the other hand reacting with an epitope in the tandem repeat of MUC1 which is affected by glycosylation.27 Figure 4C shows the combined mass spectra for all the detected oligosaccharides from the i.p. with the two different antibodies. The chromatograms are completely overlapping, which shows that HMFG-1 has a reactivity that is broad enough to recognize all the major glycoforms of MUC1 in this cell line. Similar results were obtained with the anti-MUC1 antibody 214D4, which is an even more broadly reacting antibody specific for an epitope in the tandem repeat27 (data not shown). Analysis of MUC1 O-Glycans from Gastric Cancer Tissue. MUC1 was also immunoprecipitated from a lysate of a tumor from a patient with gastric adenocarcinoma and the structures of the O-glycans determined. This tumor showed a strong overexpression of MUC1 compared to the normal mucosa of Journal of Proteome Research • Vol. 8, No. 2, 2009 543

research articles the same patient as detected in Western blots (data not shown). I.p. with CT2-anti MUC1 was however unsuccessful, indicating that most of the MUC1 was devoid of the cytoplasmic tail. By using HMFG-1 for the i.p., a large variety of oligosaccharides were detected, including both sialylated and fucosylated core 1 and core 2 structures (Figure 5). In this analysis, some major components of nonoligosaccharide nature were also detected, but these were not derived from the tumor as the same components were also detected in an analysis from an i.p. without the tumor lysate (small inset in Figure 5A). The eight most abundant oligosaccharides are shown in individual mass chromatograms in Figure 5B. An additional five oligosaccharides were also detected in lower amounts (Figure 5C), making the total number of components detected 24. A high proportion of oligosaccharides with R6-linked sialic acids were observed. Significant amounts of (NeuAcr6)GalNAcol (m/z 513-) was found and there was also significantly more of the branched glycan Galβ3(NeuAcr6)GalNAc-ol than the linear NeuAcR3Galβ3GalNAc-ol variant (both with m/z 675-). In addition, small amounts of the oligosaccharide GalGlcNAc GalGlcNAc(NeuAcr6)GalNAc-ol, which is of core 3, was also detected (m/z 1243-, Figure 5C). Two isomers of the structure NeuAcR3Galβ3(NeuAcr3Galβ3/ 4GlcNAcβ6)GalNAc-ol (m/z 1331-, 6652-) were detected and the difference must reside in the configuration of the glycosidic bond between Gal and GlcNAc (type 1 or type 2). A mixture of type 1 and type 2 chains was also found in the monosialylated core 2 structure with m/z 1040-, where the four components most likely could be explained by the presence of both type 1 and 2 extensions and by the location of the terminal sialic acid to either the 3- or the 6-linked branch. Four isomers of the oligosaccharide with m/z 1186- were also detected. Here, the different oligosaccharides can differ in the location of both the sialic acid and the fucose as well as in the type 1 or 2 chain. The most abundant component (with retention time 30 min) was determined by MS/MS to have the fucose on the 3-linked branch (by the presence of the fragment ion 530-), leading to the conclusion that the glycan has the structure FucGalβ3(NeuAcr3Gal3/4GlcNAcβ6)GalNAco-ol (data not shown). The other isomers could however not be determined from their MS/MS spectra. The successful analysis of the MUC1 O-glycans from a cancer patient sample shows the applicability of this methodology also for tissue samples. It will be important to continue and analyze more gastric cancer samples and samples from other cancer forms, to further elucidate the O-glycans on MUC1 in cancer and to relate this information to the biological roles of different glycosylations. Moreover, the number of different O-glycan structures and isomers was much higher in the patient sample compared to that found in the cell lines, indicating that a higher complexity may be found also in other patients. This, in combination with the fact that there will probably be individual variations, makes it necessary to analyze many patients to draw any conclusions on MUC1 glycosylation in different cancer forms.

Conclusions We have used a combination of i.p., gel electrophoresis, blotting, glycan release and LC-ESI-MS and LC-ESI-MS/MS to determine the oligosaccharide structures on CD43 and MUC1 from complex protein mixtures such as cell or tissue lysates. In this way, we could analyze glycans from as little as a few hundred thousand cells. We have more recently also 544

Journal of Proteome Research • Vol. 8, No. 2, 2009

Ba¨ckstro ¨ m et al. applied this methodology to the analyses of proteins with fewer O-glycans than the mucins (IgA1) and also to a glycoprotein with three N-glycans (CD83), (to be published elsewhere). The method is thus not limited to the analysis of the O-glycosylation of mucins as it can be widely used for analyses of glycans on individual both O- and N-glycoproteins. The sensitivity for detecting individual glycans will of course depend on the number of glycans in that protein, but also on the heterogeneity of the glycosylation of that protein, as a larger complexity will mean a lower abundance of each glycan. We have found that the most critical steps, with low recoveries of protein, are the i.p. and the blotting. Improvements in these areas will further advance the usability of the method.

Acknowledgment. We thank Ms Gisela Nilsson and Dr Dan Baeckstro¨m for the culture of MCF-10A cells. This work was supported by the Swedish Science Council (#7461 and #342-2004-4434), the Swedish Cancer Foundation and IngaBritt and Arne Lundberg’s foundation to G.H. and the Magn. Bergvall’s foundation, Assar Gabrielsson’s foundation for cancer research, The Lars Hierta’s minne foundation, Jubileumsklinikens Research foundation and Serena Ehrenstro¨m’s and Torsten och Sara Jansson’s foundation to M.B. The work was supported by the MIVAC Swedish Foundation for Strategic Research Center (The center for mucosal immunobiology and vaccines). References (1) Daniels, M. A.; Hogquist, K. A.; Jameson, S. C. Sweet ‘n’ sour: the impact of differential glycosylation on T cell responses. Nat. Immunol. 2002, 3 (10), 903–10. (2) Bax, M.; Garcia-Vallejo, J. J.; Jang-Lee, J.; North, S. J.; Gilmartin, T. J.; Hernandez, G.; Crocker, P. R.; Leffler, H.; Head, S. R.; Haslam, S. M.; Dell, A.; Van Kooyk, Y. Dendritic cell maturation results in pronounced changes in glycan expression affecting recognition by siglecs and galectins. J. Immunol. 2007, 179 (12), 8216–24. (3) Fuster, M. M.; Esko, J. D. The sweet and sour of cancer: glycans as novel therapeutic targets. Nat. Rev. Cancer 2005, 5 (7), 526–42. (4) Olson, F. J.; Ba¨ckstro¨m, M.; Karlsson, H.; Burchell, J.; Hansson, G. C. A MUC1 tandem repeat reporter protein produced in CHOK1 cells has sialylated core 1 O-glycans and becomes more densely glycosylated if coexpressed with polypeptide-GalNAc-T4 transferase. Glycobiology 2005, 15 (2), 177–91. (5) Haslam, S. M.; North, S. J.; Dell, A. Mass spectrometric analysis of N- and O-glycosylation of tissues and cells. Curr. Opin. Struct. Biol. 2006, 16 (5), 584–91. (6) Babu, P.; North, S. J.; Jang-Lee, J.; Chalabi, S.; Mackerness, K.; Stowell, S. R.; Cummings, R. D.; Rankin, S.; Dell, A.; Haslam, S. M. Structural characterisation of neutrophil glycans by ultra sensitive mass spectrometric glycomics methodology. Glycoconj. J. 2008, . in press. (7) Schulz, B. L.; Packer, N. H.; Karlsson, N. G. Small-scale analysis of O-linked oligosaccharides from glycoproteins and mucins separated by gel electrophoresis. Anal. Chem. 2002, 74 (23), 6088–97. (8) Gendler, S. J.; Lancaster, C. A.; Taylor-Papadimitriou, J.; Duhig, T.; Peat, N.; Burchell, J.; Pemberton, L.; Lalani, E. N.; Wilson, D. Molecular cloning and expression of human tumor-associated polymorphic epithelial mucin. J. Biol. Chem. 1990, 265 (25), 15286– 93. (9) Gendler, S. J. MUC1, the renaissance molecule. J. Mammary Gland Biol. Neoplasia 2001, 6 (3), 339–53. (10) Tait, L.; Soule, H. D.; Russo, J. Ultrastructural and immunocytochemical characterization of an immortalized human breast epithelial cell line, MCF-10. Cancer Res. 1990, 50 (18), 6087–94. (11) Sikut, R.; Andersson, C. X.; Sikut, A.; Fernandez-Rodriguez, J.; Karlsson, N. G.; Hansson, G. C. Detection of CD43 (leukosialin) in colon adenoma and adenocarcinoma by novel monoclonal antibodies against its intracellular domain. Int. J. Cancer 1999, 82 (1), 52–8. (12) Andersch-Bjo¨rkman, Y.; Thomsson, K. A.; Holmen Larsson, J. M.; Ekerhovd, E.; Hansson, G. C. Large scale identification of proteins, mucins, and their O-glycosylation in the endocervical mucus during the menstrual cycle. Mol. Cell. Proteomics 2007, 6 (4), 708– 16.

Glycosylation Analysis of Immunoprecipitated Proteins

research articles

(13) Shen, Y.; Zhao, R.; Berger, S. J.; Anderson, G. A.; Rodriguez, N.; Smith, R. D. High-efficiency nanoscale liquid chromatography coupled on-line with mass spectrometry using nanoelectrospray ionization for proteomics. Anal. Chem. 2002, 74 (16), 4235–49. (14) Sewell, R.; Ba¨ckstro¨m, M.; Dalziel, M.; Gschmeissner, S.; Karlsson, H.; Noll, T.; Ga¨tgens, J.; Clausen, H.; Hansson, G. C.; Burchell, J.; Taylor-Papadimitriou, J. The ST6GalNAc-I sialyltransferase localizes throughout the Golgi and is responsible for the synthesis of the tumor-associated sialyl-Tn O-glycan in human breast cancer. J. Biol. Chem. 2006, 281 (6), 3586–94. (15) Lloyd, K. O.; Burchell, J.; Kudryashov, V.; Yin, B. W.; TaylorPapadimitriou, J. Comparison of O-linked carbohydrate chains in MUC-1 mucin from normal breast epithelial cell lines and breast carcinoma cell lines. Demonstration of simpler and fewer glycan chains in tumor cells. J. Biol. Chem. 1996, 271 (52), 33325–34. (16) Mu ¨ ller, S.; Goletz, S.; Packer, N.; Gooley, A.; Lawson, A. M.; Hanisch, F. G. Localization of O-glycosylation sites on glycopeptide fragments from lactation-associated MUC1. All putative sites within the tandem repeat are glycosylation targets in vivo. J. Biol. Chem. 1997, 272 (40), 24780–93. (17) Storr, S. J.; Royle, L.; Chapman, C. J.; Abd Hamid, U. M.; Robertson, J. F.; Murray, A.; Dwek, R. A.; Rudd, P. M. The O-linked glycosylation of secretory/shed MUC1 from an advanced breast cancer patient’s serum. Glycobiology 2008, 18 (6), 456–62. (18) Taylor-Papadimitriou, J.; Burchell, J.; Miles, D. W.; Dalziel, M. MUC1 and cancer. Biochim. Biophys. Acta 1999, 1455 (2-3), 301– 13. (19) Brockhausen, I.; Yang, J. M.; Burchell, J.; Whitehouse, C.; TaylorPapadimitriou, J. Mechanisms underlying aberrant glycosylation of MUC1 mucin in breast cancer cells. Eur. J. Biochem. 1995, 233 (2), 607–17. (20) Hanisch, F. G.; Uhlenbruck, G.; Peter-Katalinic, J.; Egge, H.; Dabrowski, J.; Dabrowski, U. Structures of neutral O-linked polylactosaminoglycans on human skim milk mucins. A novel type

of linearly extended poly-N-acetyllactosamine backbones with Gal beta(1-4)GlcNAc beta(1-6) repeating units. J. Biol. Chem. 1989, 264 (2), 872–83. Hanisch, F. G.; Stadie, T. R.; Deutzmann, F.; Peter-Katalinic, J. MUC1 glycoforms in breast cancer--cell line T47D as a model for carcinoma-associated alterations of 0-glycosylation. Eur. J. Biochem. 1996, 236 (1), 318–27. Mu ¨ller, S.; Hanisch, F. G. Recombinant MUC1 probe authentically reflects cell-specific O-glycosylation profiles of endogenous breast cancer mucin. High density and prevalent core 2-based glycosylation. J. Biol. Chem. 2002, 277 (29), 26103–12. Ba¨ckstro¨m, M.; Link, T.; Olson, F. J.; Karlsson, H.; Graham, R.; Picco, G.; Burchell, J.; Taylor-Papadimitriou, J.; Noll, T.; Hansson, G. C. Recombinant MUC1 mucin with a breast cancer-like Oglycosylation produced in large amounts in Chinese-hamster ovary cells. Biochem. J. 2003, 376 (Pt 3), 677–86. Cozzi, P. J.; Wang, J.; Delprado, W.; Perkins, A. C.; Allen, B. J.; Russell, P. J.; Li, Y. MUC1, MUC2, MUC4, MUC5AC and MUC6 expression in the progression of prostate cancer. Clin. Exp. Metastasis 2005, 22 (7), 565–73. Garbar, C.; Mascaux, C.; Wespes, E. Expression of MUC1 and sialylTn in benign prostatic glands, high-grade prostate intraepithelial neoplasia and malignant prostatic glands: a preliminary study. Anal. Quant. Cytol. Histol. 2008, 30 (2), 71–7. Arai, T.; Fujita, K.; Fujime, M.; Irimura, T. Expression of sialylated MUC1 in prostate cancer: relationship to clinical stage and prognosis. Int. J. Urol. 2005, 12 (7), 654–61. Sikut, R.; Sikut, A.; Zhang, K.; Baeckstro¨m, D.; Hansson, G. C. Reactivity of antibodies with highly glycosylated MUC1 mucins from colon carcinoma cells and bile. Tumour Biol. 1998, 19 (Suppl 1), 122–6.

(21)

(22)

(23)

(24)

(25)

(26)

(27)

PR800713H

Journal of Proteome Research • Vol. 8, No. 2, 2009 545