Anal. Chem. 2010, 82, 4441–4447
Comparative Quantitation of Aberrant Glycoforms by Lectin-Based Glycoprotein Enrichment Coupled with Multiple-Reaction Monitoring Mass Spectrometry Yeong Hee Ahn,†,‡ Yong-Sam Kim,‡,§ Eun Sun Ji,†,| Ji Yeon Lee,†,⊥ Ji-Ae Jung,§ Jeong Heon Ko,§ and Jong Shin Yoo*,†,@ Division of Mass Spectrometry, Korea Basic Science Institute, Ochang-Myun, Cheongwon-Gun 363-883, Republic of Korea, Daejeon-KRIBB-FHCRC Research Cooperation Center, KRIBB, Yuseong-gu, Daejeon 305-806, Republic of Korea, Department of Chemistry, Hannam University, Daejeon 306-791, Republic of Korea, Department of Chemistry, Yonsei University, Seoul 120-749, Republic of Korea, and GRAST, Chungnam National University, Daejeon 305-764, Republic of Korea Lectin enrichment-coupled multiple-reaction monitoring (MRM) mass spectrometry was employed to quantitatively monitor the variation of aberrant glycoforms produced under pathological states. For this, aberrant glycoforms of the tissue inhibitor of metalloproteinase 1 (TIMP1) and protein tyrosine phosphatase K (PTPK), previously known target proteins for N-acetylglucosaminyltransferase-V (GnTV), were enriched by phytohemagglutinin-L4 (L-PHA) lectin and comparatively analyzed in the conditioned medium of the WiDr colon cancer cell line and its GnTV-overexpressing transfectant cells. Enriched glycoforms were digested, and the resultant peptides were comparatively quantified by MRM analysis. MRM quantitation data for the L-PHA-enriched samples revealed that the abundance of aberrant glycoforms of TIMP1 and PTPK was greatly increased (11.7- and 16.5-fold, respectively) in GnT-V-treated cells compared to the control cells, although the abundance of total TIMP1 and PTPK in GnT-V-treated cells was slightly different (1.1- and 0.5-fold, respectively) for unenriched samples compared to that in control cells. The dramatic variation in abundance of the aberrant glycoforms due to overexpressed GnT-V was confirmed quantitatively by comparative MRM analysis of lectinenriched samples. This method is capable of comparatively quantitating the abundance of a protein of interest and its aberrant glycoform and will be useful for studying pathological mechanisms of cancer or verifying biomarker candidates. * To whom correspondence should be addressed: Division of Mass Spectrometry, Korea Basic Science Institute, 804-1 Yangcheong-Ri, Ochang-Myun, Cheongwon-Gun 363-883, Republic of Korea. Phone: +82-43-240-5150. Fax: +8243-240-5159. E-mail:
[email protected]. † Korea Basic Science Institute. ‡ These authors contributed equally to this work. § KRIBB. | Hannam University. ⊥ Yonsei University. @ Chungnam National University. 10.1021/ac1001965 2010 American Chemical Society Published on Web 05/12/2010
Dynamic alterations in protein glycosylation, one of the most important posttranslational modifications, may be closely associated with the pathogenic processes of cells.1-3 Various abnormal glycosylation patterns, such as increased size and branching of N-glycans, have been discovered in cancer, together with features like increased levels of sialylation and fucosylation.2,4-7 Thus, the efficient analysis of these aberrant glycosylation states of proteins is important for studying the pathological mechanism of cancer or for developing cancer biomarkers. A variety of lectins have been employed for capturing specific glycoforms from a complex glycoproteome, providing the basis for detecting different types of glycoforms.8,9 Although the lectin blotting method is useful for comparative analysis of aberrant glycoforms of immuno-enriched glycoproteins, this gel-based method still has some unresolved problems before it can accurately compare quantitative variation in aberrant glycosylation states of glycoproteins because of the low capacity and high falsepositive rate of gel-based lectin blotting. An antibody-lectin sandwich array approach has also been introduced. This involves the efficient coupling of immuno-enrichment of a protein of interest using antibody microarrays, with the selective detection of a specific glycoform of the immuno-enriched protein by lectin microarrays, and is useful for visualizing lectin-binding glycoforms of purified proteins.10-12 Since the glycan levels of glycoproteins of interest can be probed for biological samples with high sensitivity and high throughput,13 this method can be a useful tool for profiling glycan microheterogeneity on multiple proteins.14,15 However, the number of antibody reagents that can be obtained (1) (2) (3) (4) (5) (6) (7) (8) (9)
Kim, Y. J.; Varki, A. Glycoconjugate J. 1997, 14, 569–576. Orntoft, T. F.; Vestergaard, E. M. Electrophoresis 1999, 20, 362–371. Hakomori, S. Adv. Exp. Med. Biol. 2001, 491, 369–402. Pierce, M.; Arango, J. J. Biol. Chem. 1986, 261, 10772–10777. Dennis, J. W.; Laferte, S.; Waghorne, C.; Breitman, M. L.; Kerbel, R. S. Science 1987, 236, 582–585. Barrabe´s, S.; Page`s-Pons, L.; Radcliffe, C. M.; Tabare´s, G. Glycobiology 2007, 17, 388–400. Zhao, J.; Qiu, W.; Simeone, D. M.; Lubman, D. M. J. Proteome Res. 2007, 6, 1126–1138. Dai, Z.; Zhou, J.; Qiu, S. J.; Liu, Y. K.; Fan, J. Electrophoresis 2009, 30, 2957–2966. Ito, S.; Hayama, K.; Hirabayashi, J. Methods Mol. Biol. 2009, 534, 195– 203.
Analytical Chemistry, Vol. 82, No. 11, June 1, 2010
4441
and employed for the immunoassay is very limited when compared to the vast number of glycoproteins newly discovered by a variety of biomarker developing processes, which need to be validated as credible biomarker candidates. The development of effective antibodies for a target glycoprotein is time-consuming and costly, and its success is dependent on the reproducible generation of a high-quality antibody. Instead, to preliminarily verify potential biomarker candidates for the plethora of newly discovered and unproven glycoproteins involving various diseases, a more general assay method featuring a simple, rapid, cost-effective, and multiplexing assay system is required; this is especially important for assaying proteins that have no proper antibody available. Mass spectrometry may provide an alternative method capable of assaying the quantitative variation of aberrant glycoforms of newly identified proteins that have yet to be proven as biomarker candidates. A multiple-reaction monitoring (MRM) method coupled with a stable isotope-coded internal standard is widely used for the quantitation of target peptides from complex tryptic digests of proteome samples.16-21 Through two-stage mass selections, the first stage selecting the precursor mass of an intact analyte and the second stage selecting a specific fragment ion of the precursor, the MRM approach is a powerful tool for specific and sensitive quantitation of target peptides from a protein of interest present in a very complex sample mixture. Thus, the high specificity and sensitivity of the MRM technique, including its quantitation ability, may be useful for assaying the quantitative variation in aberrant glycoforms of a glycoprotein comparatively between samples. However, note that although MRM is useful for the quantitative identification of a target peptide produced by enzymatic digestion of a glycoprotein, a correct assay of the quantitative variation in the aberrant glycoform cannot be achieved using only the MRM method due to the glycan microheterogeneity of the glycoprotein. Thus, enrichment of only the aberrant glycoform of the protein of interest should be conducted as a sample manipulation step prior to mass analysis.22 Recently, for the enrichment of aberrantly glycosylated proteins from breast cancer tissue or immuno-depleted colorectal cancer serum samples, an enrichment method using phytohemagglutinin-L4 (L-PHA) (10) Patwa, T. H.; Zhao, J.; Anderson, M. A.; Simeone, D. M.; Lubman, D. M. Anal. Chem. 2006, 78, 6411–6421. (11) Kuno, A.; Uchiyama, N.; Koseki-Kuno, S.; Ebe, Y.; Takashima, S.; Yamada, M.; Hirabayashi, J. Nat. Methods 2005, 2, 851–856. (12) Pilobello, K. T.; Krishnamoorthy, L.; Slawek, D.; Mahal, L. K. ChemBioChem 2005, 6, 985–989. (13) Forrester, S.; Hung, K. E.; Kuick, R.; Kucherlapati, R.; Haab, B. B. Mol. Oncol. 2007, 1, 216–225. (14) Wu, Y. M.; Nowack, D. D.; Omenn, G. S.; Haab, B. B. J. Proteome Res. 2009, 8, 1876–1886. (15) Yue, T.; Goldstein, I. J.; Hollingsworth, M. A.; Kaul, K.; Brand, R. E.; Haab, B. B. Mol. Cell. Proteomics 2009, 8 (7), 1697–1707. (16) Ahn, Y. H.; Lee, J. Y.; Lee, J. Y.; Kim, Y. S.; Ko, J. H.; Yoo, J. S. J. Proteome Res. 2009, 8, 4216–4224. (17) Anderson, L.; Hunter, C. L. Mol. Cell. Proteomics 2006, 5, 573–588. (18) Barnidge, D. R.; Goodmanson, M. K.; Klee, G. G.; Muddiman, D. C. J. Proteome Res. 2004, 3, 644–652. (19) Luna, L. G.; Williams, T. L.; Pirkle, J. L.; Barr, J. R. Anal. Chem. 2008, 80, 2688–2693. (20) Kuzyk, M. A.; Smith, D.; Yang, J.; Cross, T. J.; Jackson, A. M.; Hardie, D. B.; Anderson, N. L.; Borchers, C. H. Mol. Cell. Proteomics 2009, 8, 1860–1877. (21) McKay, M. J.; Sherman, J.; Laver, M. T.; Baker, M. S.; Clarke, S. J.; Molloy, M. P. Proteomics Clin. Appl. 2007, 1, 1570–1581. (22) Wang, Y.; Wu, S. L.; Hancock, W. S. Glycobiology 2006, 16, 514–523.
4442
Analytical Chemistry, Vol. 82, No. 11, June 1, 2010
lectin has been reported.23,24 Aberrant glycoproteins enriched selectively by L-PHA from total glycoproteins displaying complex glycan microheterogeneity could be detached from the lectin on a denaturation basis and digested directly for protein identification. An L-PHA lectin has been known to exhibit binding affinity for multiple antennary N-linked glycans having a β-1,6-N-acetylglucosamine (β-1,6-GlcNAc) moiety in an N-linked core glycan.25 It was also reported that the formation of these L-PHA-bound aberrant glycoforms is catalyzed abnormally by N-acetylglucosaminyltransferase V (GnT-V), which is a glycosyltransferase associated with pathogenic processes of cells and known to be upregulated in colon cancer cells.5,26,27 The resultant aberrant glycoforms of target proteins of GnT-V play important roles in invasive and metastatic cancer cells. In this study, proteomes obtained from the media of mock (control) and GnT-V-overexpressing WiDr cells were selected as model proteome samples for comparative monitoring of aberrant glycosylation induced by GnT-V. By coupling the lectin enrichment selective for the aberrant glycoforms with the MRM mass analysis of tryptic digests of the enriched glycoforms, we performed a comparative analysis for the quantitative variation of aberrant glycoforms of the tissue inhibitor of metalloproteinase 1 (TIMP1) and protein tyrosine phosphatase κ (PTPκ) which have recently been confirmed to be target proteins for GnT-V through various biochemical assays.26,27 EXPERIMENTAL SECTION Materials. Synthetic peptides (GFQALGDAADIR for TIMP1, GSGVSNFAQLIVR for PTPκ, and IVAISEDYPR for desmoglein2) and their stable isotope-coded counterparts were obtained from Anygen Co. (Kwangju, Korea). The stable isotope-coded peptides GFQALGDAADIR, GSGVSNFAQLIVR, and IVAISEDYPR (isotopically labeled site in italics; 13C and 15N incorporated) were detected in mass analyses by mass shifts of 4, 7, and 6 Da, respectively. Trypsin (sequencing grade) was from Promega (Madison, WI). A human protein set consisting of 48 proteins (catalog no. UPS2), biotinylated L-PHA, and avidin-agarose beads were obtained from Sigma-Aldrich (St. Louis, MO). All other reagents were purchased from Sigma-Aldrich, unless noted otherwise. Preparation of Samples from the Cancer Cell Line and Lectin Enrichment. Serum-free conditioned media of mock (control) and GnT-V-overexpressing WiDr colon cancer cells were collected and filter-concentrated to 2 mL, as previously described.27 The concentrated protein preparations were subjected to reduction in 14 mM β-mercaptoethanol followed by desalting using a Hiprep 26/10 desalting column (GE Healthcare, Milwaukee, WI). Each desalted protein (1 mg) was allowed to bind to 100 µL of L-PHA-avidin-agarose beads at 4 °C for 12 h in (23) Abbott, K. L.; Aoki, K.; Lim, J. M.; Porterfield, M.; Johnson, R.; O’Regan, R. M.; Wells, L.; Tiemeyer, M.; Pierce, M. J. Proteome Res. 2008, 7, 1470– 1480. (24) Kim, Y. S.; Son, O. L.; Lee, J. Y.; Kim, S. H.; Oh, S.; Lee, Y. S.; Kim, C. H.; Yoo, J. S.; Lee, J. H.; Miyoshi, E.; Taniguchi, N.; Hanash, S. M.; Yoo, H. S.; Ko, J. H. Proteomics 2008, 8, 3229–3235. (25) Cummings, R. D.; Kornfeld, S. J. Biol. Chem. 1982, 257, 11230–11234. (26) Kim, Y. S.; Kang, H. Y.; Kim, J. Y.; Oh, S.; Kim, C. H.; Ryu, C. J.; Miyoshi, E.; Taniguchi, N.; Ko, J. H. Proteomics 2006, 6, 1187–1191. (27) Kim, Y. S.; Hwang, S. Y.; Kang, H.-Y.; Sohn, H.; Oh, S.; Kim, J. Y.; Yoo, J. S.; Kim, Y. H.; Kim, C. H.; Jeon, J. H.; Lee, J. M.; Kang, H. A.; Miyoshi, E.; Taniguchi, N.; Yoo, H. S.; Ko, J. H. Mol. Cell. Proteomics 2008, 7, 1– 14.
Table 1. Selected Target Peptides Considering the Proteins for MRM Analysis protein, accesion number metal proteinase inhibitor 1 (TIMP1), P01033 protein tyrosine phosphatase κ (PTPκ), Q15262 desmoglein-2, Q14126
peptide
precursor ion, m/z (charge)
production, m/z
GFQALGDAADIR GFQALGDAaADIR GSGVSNFAQLIVR GSGVSNFAQLaIVR IVAISEDYPR IVAISEDYPaR
617.3 (2+) 619.3 (2+) 674.4 (2+) 677.9 (2+) 581.8 (2+) 584.8 (2+)
660.4 (y6), 717.4 (y7) 664.4 (y6), 721.4 (y7) 699.5 (y6), 1047.6 (y9) 706.5 (y6), 1054.6 (y9) 766.4 (y6), 950.5 (y8) 772.4 (y6), 956.5 (y8)
a A stable isotope-coded site. Among the two pairs of the selected transitions, the pair of the transition in bold was employed for MRM data processing.
phosphate-buffered saline (PBS). We prepared L-PHA-avidinagarose beads in advance by mixing biotinylated L-PHA with the avidin-agarose beads for 2 h at room temperature as previously described.24 The bound protein-bead conjugates were washed three times with PBS and 0.02% Tween 20. After the bound proteins were denatured in 6 M urea, the sample was diluted 10fold and then digested with 10 µg of trypsin in 50 mM ammonium bicarbonate and 1 mM CaCl2 at 37 °C overnight. The digested peptides were lyophilized and reconstituted with 100 µL of distilled water. Total samples without lectin enrichment were prepared separately by tryptic digestion of each 100 µg of desalted proteins collected from mock (control) and GnT-V-overexpressing WiDr cells. The desalted protein was denatured in 6 M urea and digested under the same conditions used for digestion of the lectin-enriched samples. Similarly, the digested peptides were lyophilized and reconstituted with 100 µL of distilled water. For the comparative MRM assay between the control and GnTV-overexpressing WiDr samples, each 10 µL portion of the total and L-PHA-enriched peptide samples was collected. After the stable isotope-coded internal standards had been spiked (heavy, 50 fmol each of GFQALGDAADIR, GSGVSNFAQLIVR, and IVAISEDYPR) to each solution, the volume of each sample was adjusted to 50 µL with an acidic buffer containing 0.5% acetic acid and 0.02% formic acid for MRM mass analysis. Preparation of the Standard Matrix Sample for MRM Analysis. To confirm the linear concentration range of target peptides in MRM mass analysis, tryptic digests of a set of human proteins were used as a standard matrix sample. Because we could not first establish an absolute abundance of endogenous proteins, TIMP1, PTPκ, and desmoglein-2 in media of control WiDr and GnT-V-overexpressing cells, a human protein set (lacking the corresponding proteins under consideration) was used to prepare a matrix peptide sample for the calibration of MRM analysis. The set of human proteins (10 µg) dissolved in 50 mM ammonium bicarbonate was reduced under 10 mM dithiothreitol and denatured at 95 °C for 10 min. After the solution had been cooled to room temperature, the solution was adjusted to 25 mM iodoacetamide and incubated for 10 min in the dark. The proteins were diluted 10-fold with 50 mM ammonium bicarbonate and then digested with 0.5 µg of trypsin at 37 °C overnight. Then, five equal aliquots of the proteins were spiked equally with the stable isotopecoded internal standards (heavy, 50 fmol each of GFQALGDAADIR, GSGVSNFAQLIVR, and IVAISEDYPR), and the unlabeled counterparts (light) were also spiked in increasing amounts (0, 0.5, 5, 50, and 500 fmol). Each sample volume was adjusted to 50 µL with an acidic buffer containing 0.5% acetic acid and 0.02% formic acid for mass analysis.
Liquid Chromatography. For mass analysis of the prepared tryptic peptides, a nano-LC/MS system consisting of a Surveyor high-performance liquid chromatography (HPLC) system (Thermo Finnigan, Austin, TX) and an LTQ-FT mass spectrometer (Thermo Finnigan) equipped with a nanoelectrospray ionization (ESI) source was employed. A portion (10 µL) of the prepared peptide solution was loaded onto a C18 trap column [5 µm, 300 µm (inside diameter) × 5 mm (LC Packings, Amsterdam, The Netherlands)] using an autosampler and washed with mobile phase A for 10 min at a rate of 20 µL/min. The peptides were transferred online onto a homemade C18 column [Aqua, 5 µm, 75 µm (inside diameter) × 100 mm (Phenomenex, Torrance, CA)], eluted under a gradient, and directly electrosprayed into an LTQ-FT mass spectrometer controlled by Xcalibur (Thermo Electron, Waltham, MA). Mobile phases A and B were composed of 0 and 80% acetonitrile, respectively, each containing 0.5% acetic acid and 0.02% formic acid. The gradient started at 5% B for 15 min; it was then increased to 20% for 3 min, 50% for 47 min, and 95% for 2 min; held at 95% for 5 min; and then held at 5% for 2 min. The column was equilibrated with 5% B for 8 min between each run. MRM Mass Analysis. From the MS/MS spectra of the synthetic target peptides (Sup. 1 of the Supporting Information), two transition channels to each of the target peptides were set for the MRM mass assay. The inclusion window was ±1 Da for each precursor ion and ±1 Da for the fragment ions. MRM analyses were conducted via comparison of the ratios of each chromatographic peak area of the selected transition pairs for each peptide sequence. Among two pairs of the selected transitions, only a pair of the most abundant transition, including an isotopically differentiated residue, was used for quantitation to minimize unwanted interference from co-eluting analytes or chemical background due to sample complexity. For the peptide sequence GFQALGDAADIR, transitions light (L) (m/z 617.3 f 717.4) and heavy (H) (m/z 619.3 f 721.4) were applied. For the peptide sequence GSGVSNFAQLIVR, transitions L (m/z 674.4 f 699.5) and H (m/z 677.9 f 706.5) were used. For the peptide sequence IVAISEDYPR, transitions L (m/z 581.8 f 950.5) and H (m/z 584.8 f 956.5) were employed (Table 1). MRM analyses were conducted in triplicate for each sample. RESULTS AND DISCUSSION Target Protein Selection and L-PHA-Enriched Sample Preparation. The purpose of this study was to establish a protocol for measuring quantitative alterations in the levels of only aberrant glycoforms of a protein of interest by rapid MRM-based quantitation technology without immuno-enrichment using an antibody. A biotinylated lectin-avidin-agarose system using the L-PHA Analytical Chemistry, Vol. 82, No. 11, June 1, 2010
4443
Figure 1. Scheme for the comparative quantitation of alterations in the level of an aberrant glycoform of a target protein TIMP1 due to overexpressed GnT-V. Table 2. MRM Analysis of Tryptic Digests of 48 Human Standard Proteins Spiked with Three Target Peptides To Determine the Linear Concentration Range peptide
peptide, L (fmol)
peptide, H (fmol)
log(peptide ratio) (L/H)
log(average area ratio) (L/H)
CV (%)
GFQALGDAADIR
0.1 1 10 100 0.1 1 10 100 0.1 1 10 100
10 10 10 10 10 10 10 10 10 10 10 10
-2 -1 0 1 -2 -1 0 1 -2 -1 0 1
-2.02 -1.01 -0.02 1.29 -2.14 -1.20 -0.12 0.98 -2.09 -1.07 -0.04 1.04
10.4 5.2 12.0 4.3 26.9 17.5 11.0 16.7 11.3 8.1 4.1 9.9
GSGVSNFAQLIVR
IVAISEDYPR
lectin was employed for enrichment of an aberrant glycoform of target proteins of GnT-V from media of control and GnT-Voverexpressing WiDr colorectal cancer cells. The culture medium of cancer cells contains many glycoproteins secreted from cells and shed from the cell membrane. As such, glycoproteins present in the media are assumed to be capable of playing roles like contributing to tumor progression via interacting and communicating between cells;28 the secreted and shed glycoproteins can also be important to the study of pathological processes and may be further applied for biomarker discovery. For a proof of concept of the MRM-based approach to aberrant glycoform quantitation, TIMP1 and PTPκ, which have been confirmed recently to be target proteins of GnT-V and fully characterized in terms of their aberrant glycosylation via immune and lectin blot analyses, were selected.26,27 Glycoprotein desmoglein-2, a component of intercellular desmosome junctions present in most tissues tested and involved in cell-cell adhesion tissue integrity,29,30 was selected as a potential control protein, negative (28) Tlsty, T. D. Semin. Cancer Biol. 2001, 11, 97–104.
4444
Analytical Chemistry, Vol. 82, No. 11, June 1, 2010
for L-PHA lectin enrichment. Using an enrichment technique with a L-PHA-avidin-agarose bead column instead of gel-based L-PHA lectin blotting, glycoprotein samples having a β-1,6-GlcNAc moiety at the N-linked glycan were prepared from the media of control and GnT-V-overexpressing WiDr cells. Total proteome samples were also prepared separately. Each sample was denatured using 6 M urea and digested with trypsin prior to MRM mass analysis (Figure 1). MRM Calibration of Target Peptide-Spiked Tryptic Digests of the Standard Sample. The absolute abundance of endogenous proteins, TIMP1, PTPκ, and desmoglein-2, in the media of control and GnT-V-overexpressing cells was unknown. Thus, in this study, we selected a set of human standard proteins lacking the corresponding proteins under consideration to prepare a matrix peptide sample for the calibration of MRM analysis. The matrix peptide sample prepared by tryptic digestion was partitioned (29) Green, K. J.; Simpson, C. L. J. Invest. Dermatol. 2007, 127, 2499–2515. (30) Uematsu, R.; Furukawa, J.; Nakagawa, H.; Shinohara, Y.; Deguchi, K.; Monde, K.; Nishimura, S. Mol. Cell. Proteomics 2005, 4, 1977–1989.
Figure 2. Calibration curves obtained by MRM mass analyses of three target peptides: (a) GFQALGDAADIR, (b) GSGVSNFAQLIVR, and (c) IVAISEDYPR. Each MRM analysis was conducted with an equal amount of stable isotope-coded internal standard (heavy, 10 fmol) and a different amount of synthetic target peptide (light, from 100 amol to 100 fmol).
equally into five tubes and then spiked equally with the stable isotope-coded internal standards (heavy, GFQALGDAADIR for TIMP1, GSGVSNFAQLIVR for PTPκ, and IVAISEDYPR for desmoglein-2), selected as peptide surrogates for the quantitation of the target proteins. After spiking increasing amounts of each target peptide (light) with the corresponding internal standard (heavy), we conducted mass analyses in MRM mode (Table 2). Figure 2 displays calibration curves obtained by MRM mass analyses of the three target peptides. For all three target peptides, good linearity was observed over 4 orders of magnitude for the spiked synthetic target peptides, from 100 amol to 100 fmol,
although the complexity of the matrix peptide sample obtained from tryptic digestion of the human standard protein set does not accurately represent that obtained from the digestion of colon cancer cells analyzed in this study. Because the amount of aberrantly glycosylated and subsequently lectin-enriched target proteins may be notably smaller than that of the overall target proteins, the high mass sensitivity and a broad linear concentration range are required for reliable MRM quantitation. Comparative Monitoring of Aberrant Glycosylation of TIMP1. TIMP1, a target protein for GnT-V in colon cancer cells, is a glycoprotein with two N-glycan moieties at residues Asn53 and Asn101; it has been reported to be abnormally GlcNAcylated in cancer cells and involved in cancer progression.27,31,32 Among the peptides consistently identified with a relatively strong signal intensity by ESI mass analysis in an earlier report,27 the peptide sequence GFQALGDAADIR, without a cysteine residue and posttranslational modification site reported within its sequence, was selected as a target peptide for MRM analysis. The comparative quantitation results obtained from triplicate MRM analyses of this target peptide of TIMP1 are presented in Table 3. From each sample prepared with or without lectin enrichment from the media of control and GnT-V-overexpressing WiDr cancer cells, the concentration of glycoprotein TIMP1 was quantitated at levels from hundreds of attomoles to several tens of femtomoles, with good reproducibility. Changes in the abundance of TIMP1 between the control and GnT-V-overexpressing cells were compared, as illustrated in Figure 3. The abundance of total TIMP1 in the GnT-V-overexpressing sample was quantitated with a slight increase compared to that in the control sample (Figure 3a), whereas the abundance of aberrant TIMP1 enriched by lectin from the GnT-V-overexpressing sample was quantitated to reveal a dramatic increase compared to that from a lectin-enriched control sample (Figure 3b). Because the sample amount used for the MRM analysis was an arbitrary portion from many media of WiDr cells, and thus the absolute values for protein abundance obtained by the MRM analyses may be meaningless, these results were expressed as changes in the ratio between the total TIMP1 and that in the lectin-enriched aberrant form, as shown in Figure 3c. Thus, the change in the abundance of aberrant TIMP1 between the control and GnT-V-overexpressing sample was very dramatic (11.7-fold) versus that (1.1-fold) of the total TIMP1 between samples that had not been enriched with the L-PHA lectin. This result indicates that the abundance of only the aberrant TIMP1 glycoform having a β-1,6-GlcNAc moiety of N-linked glycan and being specifically recognized by the L-PHA lectin increased remarkably in the GnT-V-overexpressing cancer cells, as compared to the increased abundance of total TIMP1, including a variety of glycoforms. These results were already confirmed qualitatively by immunoblotting and lectin blotting techniques in a previous report, where the aberrant TIMP1 glycoform induced by GnT-V plays important roles in invasive and metastatic colon cancer cells.27 The comparative monitoring method for substoichiometric aberrant glycosylation of glycoproteins presented here may also be applied to the analysis of more complex samples, like blood, and further for the verification of newly identified and unproven biomarker candidates involving altered protein glycosylation. (31) Egeblad, M.; Werb, Z. Nat. Rev. Cancer 2002, 2, 161–174. (32) Baker, E. A.; Bergin, F. G.; Leaper, D. J. Br. J. Surg. 2000, 87, 1215–1221.
Analytical Chemistry, Vol. 82, No. 11, June 1, 2010
4445
Table 3. Comparative Quantitation of the Target Proteins Obtained from the Control and GnT-V-Overexpressing WiDr Cancer Cells with and without L-PHA Lectin Enrichment WiDr, control protein
peptide
TIMP1
GFQALGDAADR
PTPκ
GSGVSNFAQLIVR
desmoglein-2
IVAISEDYPR
average (fmol) CV (%) average (fmol) CV (%) average (fmol) CV (%)
Comparative Monitoring for Aberrant Glycosylation of PTPK. The glycoprotein PTPκ has been reported to be a target protein for GnT-V in human colon cancer cells having 16 putative N-linked glycosylation sites on the protein backbone.26,33 In addition, this protein has also been reported to be involved in cancer progression via aberrant glycosylation by GnT-V in cancer cells. Among the peptides identified by ESI mass analysis in a previous report,26 the peptide sequence GSGVSNFAQLIVR was selected as a target peptide sequence for MRM analysis, considering its high mass sensitivity and the absence of cysteine and methionine residues. Comparative MRM analysis results for the target peptide of PTPκ are summarized in Table 3. The PTPκ present in each prepared sample was quantitated at levels from high (attomoles) to low (femtomoles), with good reproducibility. The abundance of total PTPκ was somewhat lower in GnT-V-overexpressing cells than in control cells (Figure 4a), indicating that the overall abundance of PTPκ can be influenced by overexpressed GnT-V. This tendency was also observed by immunocytochemistry in a previous study.26 Distinc-
WiDr, GnT-V-overexpressing
aberrant
total
aberrant
total
0.34 25.5 0.16 2.9 -
45.49 65.0 7.94 15.2 155.83 18.7
3.97 4.9 2.64 39.1 -
47.55 19.0 3.70 7.0 167.97 22.0
tively, Figure 4b shows that the abundance of aberrant PTPκ enriched by the L-PHA lectin column was remarkably higher in GnT-Voverexpressing cells than in control cells. This difference between the two samples following the lectin enrichment is more apparent in Figure 4c. Thus, the increase in abundance of aberrant PTPκ in the GnT-V-overexpressing sample to that in the control sample was very dramatic (16.5-fold), although a slight change in the abundance of total PTPκ between the two samples not enriched by the lectin column was also observed (0.5-fold). A similar tendency in the abundance of PTPκ was also reported in a study on human hepatocellular carcinoma cells using immunoblotting and lectin blotting techniques, in which GnT-V could activate epidermal growth factor receptor (EGFR) signaling by decreasing the level of PTPκ and promote cell migration through the aberrant glycosylation on the N-linked glycan of PTPκ.33 Thus, the decrease in the level of PTPκ and the increase in the level of the aberrant glycoform of the protein induced by overexpression of GnT-V are explained by these two processes.
Figure 3. Changes in the abundance of TIMP1 between the control and GnT-V-overexpressing WiDr colon cancer cells. Protein quantitation was conducted by MRM analysis of (a) total TIMP1 without L-PHA lectin enrichment and (b) aberrant TIMP1 obtained with lectin enrichment. (c) Effect of lectin enrichment on monitoring the change in abundance of aberrant glycosylation.
Figure 4. Changes in the abundance of PTPκ between the control and GnT-V-overexpressing WiDr colon cancer cells. Protein quantitation was conducted by MRM analysis of (a) total PTPκ without L-PHA lectin enrichment and (b) aberrant PTPκ obtained with lectin enrichment. (c) Clear effect of lectin enrichment on monitoring the change in abundance of aberrant glycosylation. 4446
Analytical Chemistry, Vol. 82, No. 11, June 1, 2010
Figure 5. Abundance of desmoglein-2 in each sample of control and GnT-V-overexpressing WiDr colon cancer cells. Desmoglein-2 was determined to be at similar concentrations in (a) two samples without L-PHA lectin enrichment but was not detected in (b) two samples with L-PHA lectin enrichment. NQ, not quantified.
In this study, changes in PTPκ involving GnT-V-mediated aberrant glycosylation in the colon cancer cells were monitored quantitatively by the MRM-based technique coupled with the lectin enrichment method, with high sensitivity and good precision from MRM mass analysis, which complemented the excellent structural selectivity for glycan microheterogeneity from lectin enrichment. Therefore, by conducting a comparative mass analysis of targetoriented samples enriched in advance by a specific lectin relative to that of total samples including all kinds of glycoforms, we can sophisticatedly monitor the quantitative change in a protein of interest between samples. It also shows that the comparative MRM-based monitoring coupled with lectin enrichment can be a useful method for complementing a variety of immunoblotting and lectin blotting techniques that are utilized routinely for analyzing protein glycosylation. Furthermore, it can be an alternative tool in cases where a suitable antibody for a glycoprotein of interest does not exist. Comparative Analysis of Desmoglein-2 Negative to L-PHA Lectin Enrichment. Desmoglein-2, a component of intercellular desmosome junctions, is a glycoprotein with repeating cadherin domains, and it involves an interaction of plaque proteins with intermediate filaments mediating cell-cell adhesion.29,30 Although desmoglein-2 is present universally in most tissues tested and in carcinomas and involves cell-cell adhesion tissue integrity, neither the glycan profile of the microheterogeneity of the protein nor its affinity for L-PHA lectin has been reported. As shown in Table 3, a target peptide, IVAISEDYPR, as a surrogate of desmoglein-2, was quantitated by the MRM analysis at levels of hundreds of femtomoles from each sample of the control and GnTV-overexpressing cells, in which the ratio of overall desmoglein-2 between the two samples was observed to reach ∼1:1.1 (Figure 5a). Thus, the overall abundance of the shed desmoglein-2 was essentially assumed to be unaffected by the overexpression of GnT-V in WiDr colon cancer cells. Desmoglein-2, however, was not quantitated even at levels of hundreds of attomoles from both lectin-enriched samples under the MRM analysis conditions that were used (Figure 5b). In general, lectin recognizes a specific glycan motif present on glycoproteins, whereas a protein antibody normally recognizes its (33) Wang, C.; Yang, Y.; Yang, Z.; Liu, M.; Li, Z.; Sun, L.; Mei, C.; Chen, H.; Chen, L.; Wang, L.; Zha, X. Arch. Biochem. Biophys. 2009, 486, 64–72.
target protein. The lectin-recognizable motif may present commonly over many kinds of glycoproteins. Thus, since lectin shows affinity for a specific glycan structure on glycoproteins rather than for a target protein itself, we suppose that no detection of desmoglein-2 from both the lectin-enriched samples is caused by the nonexistence of lectin-recognizable glycoforms with concentration levels monitored in this study. This can be an example for a protein not showing any difference between the lectin-captured samples. Glycoproteins are characterized by their heterogeneity mostly originating from a diverse glycan structure. This is why all molecules of a glycoprotein do not necessarily show an affinity for a specific lectin; instead, a subset of them is bound to it. Here, desmoglein-2 from both cell lines was resistant to enrichment by L-PHA, which may indicate that desmoglein-2 is indigenously a poor substrate for GnT-V for any structural reason. Although a further investigation is necessary for full glycoprofiling of desmoglein-2, we concluded that the level of the shed desmoglein-2 was not significantly different between both cells and also that desmoglein-2 shed from both cells does not show a difference in the affinity for L-PHA lectin at levels of hundreds of attomoles that can be observed in the MRM mass analysis used in this study. CONCLUSION Comparative, quantitative analysis of aberrant glycoproteins was achieved using MRM-based MS coupled with a lectin-specific glycoprotein-enrichment technique. Aberrant glycoforms of TIMP1 and PTPκ were enriched using an L-PHA lectin column from model proteome samples from WiDr cells and their GnT-V-overexpressing counterparts, and the enriched aberrant glycoforms were comparatively quantitated by MRM analysis relative to the analysis of total target proteins not enriched by the lectin column. This method provides a useful tool for monitoring changes in the abundance of aberrant glycosylation of proteins of interest in a comparative manner. Together with a gel- or array-based immunoblotting and lectinblotting approach generally used for the selective detection of a target glycoform having a specific glycan structure, the lectin-coupled MRM mass analysis approach presented here can be used as a complementary or alternative analytical method for the quantitative monitoring of protein glycosylation. In particular, this method will provide a more robust tool for assaying proteins in a high-throughput analysis, where a feasibility of antibody-based approaches is limited. The comparative MRM-based monitoring coupled with lectin enrichment may be further applicable for verifying potential biomarker candidates involving aberrant glycosylation by virtue of the rapid assay configuration, low cost, and effective multiplexing features of the MRM analysis. ACKNOWLEDGMENT This work was supported by the ‘Leading Foreign Research Institute Recruitment Program’ and ‘Convergence Research Center Program’ of the Ministry of Education, Science and Technology. SUPPORTING INFORMATION AVAILABLE Additional information noted in the text. This material is available free of charge via the Internet at http://pubs.acs.org. Received for review January 22, 2010. Accepted May 3, 2010. AC1001965 Analytical Chemistry, Vol. 82, No. 11, June 1, 2010
4447
