Glycosylation of liver acute-phase proteins in pancreatic cancer and chronic pancreatitis

Transcription

1 432 DOI /prca Proteomics Clin. Appl. 2010, 4, RESEARCH ARTICLE Glycosylation of liver acute-phase proteins in pancreatic cancer and chronic pancreatitis Ariadna Sarrats 1, Radka Saldova 2, Eva Pla 1, Esther Fort 3, David J. Harvey 4, Weston B. Struwe 2, Rafael de Llorens 1, Pauline M. Rudd 2 and Rosa Peracaula 1 1 Unitat de Bioquímica i Biologia Molecular, Departament de Biologia, Universitat de Girona, Campus de Montilivi, Girona, Spain 2 Dublin-Oxford Glycobiology Laboratory, NIBRT, Conway Institute, University College Dublin, Belfield, Dublin, Ireland 3 Unitat de Digestiu, Hospital Universitari Dr Josep Trueta, Girona, Spain 4 Glycobiology Institute, Department of Biochemistry, Oxford University, Oxford, UK Purpose: Glycosylation of acute-phase proteins (APP), which is partially regulated by cytokines, may be distinct in disease and provide useful tumour markers. Thus, we have examined the glycosylation of major serum APP in pancreatic cancer (PaC), chronic pancreatitis (CP) and control patients. Experimental design: Using a specific anti-sialyl Lewis X antibody and N-glycan sequencing, we have determined glycosylation changes on a-1-acid glycoprotein (AGP), haptoglobin (HPT), fetuin (FET), a-1-antitrypsin (AT) and transferrin (TRF). Results: Increased levels of sialyl Lewis X (SLe x ) were detected on AGP in advanced PaC and CP and on HPT, FET, AT and TRF in CP. An increase in N-glycan branching was detected on AGP and HPT in the advanced stage of PaC and CP and on FET and TRF in the CP. A core fucosylated structure was increased on AGP and HPT only in the advanced PaC patients. Conclusions and clinical relevance: Changes in APP SLe x and branching are probably associated with an inflammatory response because they were detected in both advanced PaC and CP patients and these conditions give rise to inflammation. On the contrary, the increase in APP core fucosylation could be cancer associated and the presence of this glycoform may give an advantage to the tumour. Received: July 10, 2009 Revised: October 6, 2009 Accepted: December 18, 2009 Keywords: Acute-phase proteins / Core fucose / Liver / Pancreatic cancer / Sialyl Lewis X Correspondence: Professor Pauline M. Rudd, Dublin-Oxford Glycobiology Laboratory, NIBRT, Conway Institute, University College Dublin, Belfield, Dublin 4, Ireland pauline.rudd@nibrt.ie Fax: Abbreviations: 2-AB, 2-aminobenzamide; AGP, a-1-acid glycoprotein; APP, acute-phase proteins; AT, a-1-antitrypsin; ConA, concanavalin A; CP, chronic pancreatitis; FET, fetuin; FUT, fucosyltransferase; HC, healthy control; HCC, hepatocellular carcinoma; HPT, haptoglobin; IL, interleukin; NP, normal phase; PaC, pancreatic cancer; Q, quadrupole; RT, room temperature; SLe x, sialyl Lewis X; TBST, 0.1% Tween in TBS; TRF, transferrin; WAX, weak anion exchange 1 Introduction Pancreatic cancer (PaC) is the fourth leading overall cause of cancer death in the United States and it has the lowest 5-year survival rate (about 5%) [1]. This poor survival is mostly because of late diagnosis, frequently after metastasis. Early detection of PaC is currently not available because existing biomarkers for this disease are inadequate [2]. For example, the use of CA19-9, the most common tumour marker for PaC, is restricted by its false-positive results Additional corresponding author: Dr. Rosa Peracaula address: rosa.peracaula@udg.edu

2 Proteomics Clin. Appl. 2010, 4, [3, 4]. This marker is unable to differentiate PaC from benign pancreaticobiliary disorders such as chronic pancreatitis (CP). Thus, a great need exists for new sensitive and specific biomarkers for PaC. Liver plays an important role in the acute-phase response that is one of the key processes of the innate and natural immunity in response to tissue injury, infection and inflammation. Cytokines produced during inflammatory processes, including cancer, are the main stimulators for the synthesis of liver-positive acute-phase proteins (APP) the concentration of which in plasma increases [5]. At the same time, the production of a number of other liver proteins is reduced (negative APP). Some major plasma APP are glycosylated, and include a-1-acid-glycoprotein (AGP), haptoglobin (HPT), a-1-anti-chymotrypsin, a-1-antitrypsin (AT) and fibrinogen (positive APP); fetuin (FET) and transferrin (TRF) (negative APP). Altered glycosylation is a common feature of tumour cells and it can be reflected in their glycocalyx and secreted glycoproteins. Generally, N-glycans are more branched and more sialylated in tumours [6]. In addition, the Lewis structures sialyl Lewis X (SLe x ) and sialyl Lewis A (SLe a ), which carry an outer-arm fucose linked to N-acetylglucosamine and a sialic acid linked to a-2,3 to galactose, are overexpressed in carcinomas and have been related to increase the invasive and metastatic tumour potential [7]. In particular, Lewis X and related antigens, such as SLe x, have been found to be expressed in PaC tissues at higher rates than inflamed pancreas (CP) while they were barely detected in healthy pancreatic tissues [8 10]. PaC sera, but not sera from healthy subjects, inhibit E-selectin binding of pancreatic tumour cells [11], suggesting an overexpression of the sialylated Lewis antigens SLe x and SLe a in the PaC sera glycoproteins, which act as E-selectin ligands. Taken together, these data suggest that serum-specific proteins secreted by the tumour, which carry SLe x or aberrant glycans moieties, could be used as PaC tumour markers. In fact, using glycoprotein microarrays with multi-lectin detection techniques, an increase in both fucosylation and sialylation of some serum glycoproteins has been described for PaC patients compared with healthy controls and pancreatitis patients [12, 13]. Pancreatic adenocarcinoma is a tumour with a high inflammatory component [14] to which the liver responds by producing APP. In addition, factors released by tumour cells, such as cytokines, into the local microenvironment can have tumour-promoting effects. These include favouring the proliferation and survival of malignant cells, promoting angiogenesis and metastasis [15] and modulating the cellular glycosylation machinery. Proinflammatory cytokines, such as interleukin-1b (IL-1b), upregulate the biosynthesis of SLe x in hepatoma cells [16], and IL-6 and IL-8 increase the levels of SLe x and 6-sulfo-SLe x on bronchial mucins from cystic fibrosis patients [17]. Pancreatic adenocarcinoma cell lines and tissues produce proinflammatory cytokines [18, 19]. These released cytokines could alter the glycosylation of the tumour neighbouring cells and their secreted glycoproteins. Given the close proximity of pancreas and liver, the liver-secreted acute-phase-response glycoproteins could also alter their glycosylation in pancreatic adenocarcinoma. Therefore, not only tumour-secreted glycans but also tumour host-response glycans could provide useful biomarkers. Our aim was to identify glycosylation changes on the major APP (AGP, HPT, AT, FET and TRF) produced by the liver in the serum of PaC patients compared with healthy controls and CP patients. For that we used two approaches. One based on the use of specific antibodies against glycan epitopes, as they are known to be highly more specific than lectins, and if changes in their reactivity in the different groups of patients could be detected, they would provide more feasible approaches to develop PaC biomarkers. In particular, we have used antibodies against SLe x to highlight proteins with altered expression of this cancer-related epitope. As there are not available antibodies against all possible N-glycan modifications in tumours, it is necessary to determine the complete N-glycan structures of these potential altered glycoproteins in PaC. N-glycan sequencing is a well-established, direct strategy to evaluate glycan alterations [20]. Thus, in our second approach, we have used N-glycan sequencing, including HPLC, exoglycosidase digestion analysis and MS, in order to identify other specific glycosylation changes in different stages of PaC patients compared with healthy controls and CP patients. 2 Materials and methods 2.1 Serum samples Serum samples were obtained from five healthy controls (three females and two males; age range 44 72), nine PaC patients (six females and three males; age range 52 85, one each in stages I, II, III and IVA and five in stage IVB) and three CP patients (all males, age range 38 53) from the Hospital Universitari Dr J. Trueta (Girona, Spain) following the standard operating procedures of its Ethics Committee. Patients were diagnosed by biopsy or image examination by the Digestive and Pathology Units. 2.2 HSA and IgG depletion and concentration of serum samples Serum samples were subjected to HSA and IgG depletion according to the procedure described by Wang et al. [21], with modifications. Briefly, 600 ml of protein G sepharose fast flow (GE Healthcare, Uppsala, Sweden) was added to Spin-X centrifuge tube (Costar, Corning, NY, USA) and washed three times with MilliQ water. Polyclonal rabbit anti-human albumin antibody (Dako, Glostrup, Denmark) was added to the spin tube and incubated with protein G at

3 434 A. Sarrats et al. Proteomics Clin. Appl. 2010, 4, room temperature (RT) for 30 min with rotation, then washed three times with PBS for removal of unbound proteins. Thirty microlitre of human serum was diluted 1:10 in PBS, sonicated for 5 min and then incubated in a spin tube rotating for 30 min at RT. Afterwards, the spin tube was spun down for 2 min and the flow through was collected. The spin tube was washed three times with PBS and the flow-through fractions were pooled (final volume of approximately 1.8 ml). To concentrate and desalt the samples, they were placed on a MicroSep 3K tube (Pall, Ann Arbor, MI, USA) and centrifuged at 6000 g for 1 h. After that the MicroSep tube was washed four times with 2 ml of MilliQ water and centrifuged until the sample volume was about 300 ml. Protein concentration was determined by the Bradford protein assay using BSA as standard (Quick Start Bradford protein assay, BioRad, Hercules, CA, USA). 2.3 SLe x immunodetection in depleted serum samples After HSA and IgG depletion and concentration of serum samples, 40 mg of total protein was electrophoresed on a 10% polyacrylamide gel and transferred onto a PVDF membrane (Millipore, Bedford, MA, USA). The membrane was then washed with 0.1% Tween in TBS (TBST, 10 mm Tris-HCl ph 7.5, 100 mm NaCl, 0.1% Tween-20) for 15 min and blocked with 1% BSA in TBST for 1 h at RT. After blocking, it was washed with TBST for 5 min and incubated with a mouse monoclonal anti-sialyl Lewis X antibody (Clone KM93, Calbiochem, Darmstadt, Germany) diluted 1/50 in incubation buffer (0.5% BSA in TBST) for 2 h at RT. The membrane was then washed three times for 5 min with TBST and subsequently incubated with a goat polyclonal horseradish peroxidase-conjugated anti-mouse IgG1IgM1 IgA antibody F(ab)2 fragments (Abcam, Cambridge, UK) diluted 1/ in incubation buffer for 1 h at RT. After washing out the secondary antibody five times as described above, the membrane was incubated for 5 min with the horseradish peroxidase substrate solution West Dura (Pierce Biotechnology, Rockford, IL, USA). Chemi luminescence was visualized using the imaging system Fluorochem SP (AlphaInnotech, San Leandro, CA, USA) under non-saturating conditions DE Two different 2-DE procedures were performed depending on the type of strip (GE Healthcare) used in the first dimension. (i) For the identification of glycoproteins carrying increased SLe x, isoelectric focusing was performed on immobiline dry strips ph 3 10, 7 cm. Strips were subjected to active in-gel rehydration (12 h, 50 V, 201C) with 60 or 100 mg of depleted serum dissolved in 200 ml of rehydration buffer (8 M urea, 0.5% Triton X-100, 65 mm DTT, 1% pharmalyte 3 10 (GE Healthcare), traces of bromophenol blue). After rehydration, a water-damp electrode wick was placed between the strip and each of the electrodes. An extra 50 mm DTT wick was placed at 1 cm from the negative electrode. Isoelectric focusing was performed as follows: V in 30 min, 500 V for 1 h, from V for 30 min, 1000 V for 1 h, from V for 1.5 h, 2000 V for 1 h, V for 1.5 h, 4000 V until V h. For the second dimension, the strips were first equilibrated for 15 min in equilibration buffer (6 M urea, 30% v/v glycerol, 2% SDS, 50 mm Tris ph 8.8) containing 65 mm DTT and afterwards for 15 min in equilibration buffer containing 135 mm iodoacetamide. The strips were then placed on top of 10% polyacrylamide gels and sealed with a 0.5% agarose solution containing traces of bromophenol blue in running buffer (192 mm glycine, 25 mm Tris-HCl ph 8.5, 0.1% SDS). The second dimension was performed in a Miniprotean III unit (BioRad) under the following conditions: 10 ma per gel and an increase of 2.5 ma per gel every 15 min. (ii) For isolation of the APP to perform N-glycan sequencing, immobiline dry strips ph 3 7 non-linear, 24 cm were used. Strips were subjected to in-gel rehydration with 10 ml of serum dissolved in 450 ml of rehydration buffer 2 (8 M urea; 0.5% CHAPS; 0.2% pharmalyte 3 10; 0.2% DTT, trace of bromophenol blue) overnight at RT. Isoelectric focusing was performed in Ettan IPGphor II IEF system (GE Healthcare) as follows: V h, 8000 V for 10 min, 8000 V for 1 h. For the second dimension, the strips were first equilibrated for 15 min in equilibration buffer (50 mm Tris-HCl ph 8.8, 6 M urea, 30% v/v glycerol, 2% w/v SDS, traces of bromophenol blue) containing 1% w/v DTT and afterwards for 15 min in equilibration buffer containing 2.5% w/v iodoacetamide. The strips were then placed on top of 10% polyacrylamide gels and sealed with a 1% agarose solution in running buffer. The second dimension was performed in a Protean Plus Dodeca Cell (BioRad) under the following conditions: 1 W per gel overnight at 151C. Following electrophoresis, proteins separated in the 2-DE gels were either transferred to a PVDF membrane and subjected to SLe x immunodetection (after using procedure i) or were Coomassie blue stained (after using both the procedures i and ii). 2.5 High throughput release and processing of N-glycans from 2-DE gel spots Selected 2-DE gel spots were excised from the gel, cut into 1mm 3 gel pieces, transferred to a filter plate (protein precipitation plate, Whatman, Maidstone, Kent, UK) and stored at 201C. N-glycans were released from 2-DE gel

4 Proteomics Clin. Appl. 2010, 4, pieces and labelled according to the in-gel block method for human serum described earlier [20], with modifications. Briefly, gel pieces were treated with protein N-glycosidase F to release the N-linked glycans. Half of the extracted N-glycans were fluorescently labelled with 2-aminobenzamide (2-AB) by reductive amination (LudgerTag 2-AB labelling kit, Ludger, Culham, Oxfordshire, UK). The excess 2-AB was removed in this case with Whatman 3MM chromatography paper. The 2-AB-labelled N-glycans were then analyzed by normal-phase HPLC (NP-HPLC) and weak anion exchange (WAX)-HPLC. The unlabelled N-glycans were later analyzed by ESI-MS/MS and MALDI-TOF MS. 2.6 Simultaneous oligosaccharide sequencing by exoglycosidase digestions All enzymes were purchased from Prozyme (San Leandro, CA, USA). The 2-AB-labelled glycans were digested in 10 ml of 50 mm sodium acetate buffer, ph 5.5 for 18 h at 371C, (except in the case of jack bean a-mannosidase where the buffer was 100mMsodiumacetate,2mMZn 21, ph 5.0) using arrays of the following enzymes at the indicated concentrations: Arthrobacter ureafaciens sialidase (EC ), 1 U/mL; Streptococcus pneumoniae sialidase (EC ), 1 U/mL; bovine testes b-galactosidase (EC ), 1 U/mL; S. pneumoniae b-galactosidase (EC ), 0.1 U/mL; bovine kidney a-fucosidase (EC ), 1 U/mL; S. pneumoniae b-nacetylglucosaminidase, recombinant in Escherichia coli (EC ), 8 U/mL; jack bean b-n-acetylglucosaminidase (EC ), 10 U/mL; jack bean a-mannosidase (EC ), 50 U/mL; almond meal a-fucosidase (EC ), 3 mu/ ml; Xanthomonus sp. a-fucosidase (EC ), 0.1 U/mL. After incubation, enzymes were removed by filtration through a protein-binding EZ filters (Millipore) [22]. N-glycans were then analyzed by NP-HPLC or WAX-HPLC. NP-HPLC was performed using a TSKgel amide-80 5-mm ( mm) (Anachem) column on a Waters 2795 Alliance HT separations module. Solvent A was 50 mm formic acid adjusted to ph 4.4 with ammonia solution and solvent B was ACN. The column temperature was set to 301C. Conditions used were as follows: 60 min method a linear gradient of 35 47% solvent A over 48 min at a flow rate of 0.8 ml/min, followed by 1 min at % A and 4 min at 100% A, returning to 35% A over 1 min and then finishing with 35% A for 6 min [20]. Samples were injected in 65% ACN. The system was calibrated using an external standard of hydrolyzed and 2-AB-labelled glucose oligomers to create a dextran ladder, as described previously [22]. 2.8 N-glycan analysis by MS ESI-MS and MS/MS One microlitre of each sample was cleaned by allowing it to sit for approximately 1 h on the surface of a Nafion 117 membrane (Aldrich, Poole, Dorset, UK) [23] that was floating on water. The samples were then transferred to a small Eppendorf tube to which was added 6 ml of a 1:1 (v/v) mixture of water:methanol containing 0.2 M ammonium phosphate and the entire solution was then infused through Proxeon (Proxeon Biosystems, Odense, Denmark) borosilicate capillaries into a Waters (Waters MS Technologies, Manchester, UK) tandem quadrupole TOF (Q-TOF) mass spectrometer at a source temperature of 1201C for acquisition of electrospray (ESI) spectra. Both negative ion MS and MS/MS (argon collisions) spectra of glycan ions were recorded. Data acquisition and processing were conducted with Waters MassLynx version 4.1. Interpretation of the negative ion MS/MS spectra was according to published work [24 27]. 2.7 N-glycan analysis by WAX-HPLC and NP-HPLC WAX-HPLC was performed using a Vydac 301VHP mm column (Anachem, Luton, Bedfordshire, UK) on a 2695 Alliance separations module with a 474 fluorescence detector (Waters, Elstree Hertfordshire, UK). Solvent A was 0.5 M formic acid adjusted to ph 9.0 with ammonium solution, and solvent B was 10% v/v methanol in water. Gradient conditions were as follows: a linear gradient of 0 5% A over 12 min at a flow rate of 1 ml/min, followed by 5 21% A over 13 min and then 21 50% A over 25 min, % A over 5 min, and then 5 min at 100% A. Samples were injected in water. A FET N-glycan standard was used for calibration [22]. All HPLC units were equipped with Waters temperature control modules and Waters 2475 fluorescence detectors set with excitation and emission wavelengths of 330 and 420 nm, respectively [20] MALDI-MS One microlitre of an aqueous solution, cleaned with the Nafion membrane as above, was mixed on the stainless steel MALDI target plate with 0.5 ml of a saturated solution of 2,5-DHB in ACN and allowed to dry under ambient conditions. The sample spot was then recrystallized from ethanol. Spectra were acquired with a Waters Micro MX MALDI-TOF mass spectrometer in reflectron mode with an acceleration voltage of 12 kv. Data acquisition and processing were conducted with Waters MassLynx version Identification of the proteins in 2-DE spots by MS analysis Proteins contained in the 2-DE spots were in-gel digested with trypsin (sequencing grade modified, Promega Biotech

5 436 A. Sarrats et al. Proteomics Clin. Appl. 2010, 4, Ibérica, Madrid, Spain) in the automatic investigator ProGest robot from Genomic Solutions. Briefly, excised gel spots were washed with ammonium bicarbonate buffer (50 mm NH 4 HCO 3 ) and ACN. Proteins were reduced with 10 mm DTT solution during 30 min and alkylated with a 55 mm solution of iodoacetamide. After washing with buffer and ACN, proteins were digested overnight, at 371C with 0.27 nmol of trypsin. Tryptic peptides were extracted from the gel matrix with 10% formic acid and ACN; the extracts were pooled and dried in a vacuum centrifuge. Proteins were either analyzed in a MALDI-TOF- TOF (4700 Proteomics Analyzer, Applied Biosystems, IL, USA) or LC-ESI-Q-TOF (Q-TOF Global, Waters) mass spectrometer. For MALDI-MS and MS/MS analyses, the digests were redissolved in 5 ml of 0.1% TFA in 50% ACN. Typically, a ml of sample was mixed with the same volume of a matrix solution (5 mg/ml CHCA; Waters) in 0.1% TFA in 50% ACN and spotted to the MALDI plate. MS spectra were acquired in positive reflector mode (voltage of 20 kv in the source and a laser intensity that ranged from 5800 to 6200). Typically, 500 shots per spectrum were accumulated. MS/MS spectra were acquired using collision-induced dissociation with atmospheric air as the collision gas. An MS/MS 1 kv positive fragmentation mode was used. MS and MS/MS spectra from the same spot were merged in a single MASCOT (Matrix Science, London, UK) generic file prior to submission for database searching. For on-line LC-ESI-MS/MS analysis (Cap-LC-nano-ESI- Q-TOF) (CapLC, Waters), the tryptic digested peptide samples were resuspended in 25 ml of 1% formic acid solution and 4 ml was injected to a reversed-phase capillary C18 column (75 mm internal diameter and 15 cm length, PepMap column, LC Packings (Dionex Biogen, Madrid, Spain)). The eluted peptides were ionized via coated nano- ES needles (PicoTip TM, New Objective, Woburn, MA, USA). A capillary voltage of V was applied together with a cone voltage of 80 V. The collision energy for collisioninduced dissociation was ev, and argon was employed as the collision gas. Data were generated in PKL file format and submitted for database searching in the MASCOT server against a non-redundant Swiss-Prot database. The search parameters were: human taxonomy, two missed cleavages allowed, carbamidomethyl of cysteine as a fixed modification and oxidation of methionine as a variable modification. The peptide tolerance was 100 ppm and 0.25 Da for MS and MS/MS spectra, respectively. The significance threshold was set at po0.05. For the MASCOT generic file searches, only proteins with scores above significant MASCOT level (456) were considered as positive hits and protein summary was selected as the report format. For the PKL files, only peptides with scores above the significant MASCOT level were considered for protein identification (428) and peptide summary was selected as the report format. 3 Results 3.1 HSA and IgG serum depletion yields Depletion of HSA and IgG from serum samples reduced the quantity of serum proteins by about 70%. Coomassie staining of SDS-PAGE gels from the retained and flowthrough fractions showed that only albumin and IgG were removed (data not shown). To check that other minor serum proteins were not extracted along with albumin and removed from the sera, ELISA against ribonuclease 1, which is one minor serum protein, was performed with the serum and the corresponding serum-depleted fraction. The recovery yield for ribonuclease 1 was approximately 80%, giving some indication of the level of the losses of protein associated with the depletion method. After removal of albumin and IgG, depleted serum samples contained mainly serum liver proteins including APP and other minor proteins. 3.2 SLe x immunodetection in depleted serum samples Forty microgram of depleted serum samples from healthy controls (HC 1 3), PaC (PaC 1 9) and CP (CP 1 3) patients were electrophoresed, transferred onto a PVDF membrane and subjected to SLe x immunodetection. An immunoreactive band of approximately 50 kda was observed for stage IV PaC patients (PaC 1 5 and 8). Stages I, II and III PaC patients (PaC 6, 7 and 9) showed no SLe x immunodetection. Although this band was not detected in any of the healthy controls analyzed, it did appear in some of the CP patients CP 1 and 2 (two of three) (Fig. 1A). CP patients also showed upper bands of more than 100 kda that were positively stained for SLe x, which were not further analyzed. We focused the study on the 50 kda band, as it was increased in all advanced PaC patients. In order to identify the glycoproteins carrying the increased SLe x epitope in the 50 kda band, 60 mg of depleted PaC 8 sample was subjected to 2-DE followed by SLe x immunodetection. At approximately 50 kda, a single SLe x immunoreactive spot was detected and showed a pi of 3.8 (Fig. 1B). A preparative 2-DE gel was run under the same conditions but with 100 mg of depleted PaC 8 sample to isolate and identify the proteins of the SLe x immunoreactive spot. For that the spot was excised from the gel and subjected to the analysis of the peptide mass fingerprints. AGP1 (score: 102, sequence coverage: 26%, number of peptides: 9) and AGP2 (score: 119, sequence coverage: 25%, number of peptides: 10) were the only significant hits. AGP1 and AGP2 are two variants of the same protein, AGP, that comigrate in the 2-DE due to their high homology [28, 29]. Thus, AGP, one of the major members of the positive APP [30], is the main protein contributing to the increased SLe x detection in the approximately 50 kda band.

6 Proteomics Clin. Appl. 2010, 4, Figure 1. Sialyl Lewis X immunodetection on HSA-IgG-depleted sera of healthy controls, PaC and CP patients. (A) 40 mg of total protein was electrophoresed on a 10% polyacrylamide gel and transferred onto a PVDF to perform SLe x immunodetection. (B) 2-DE of 100 mg of total protein of sample PaC8. At approximately 50 kda, a SLe x immunoreactive spot (pi of 3.8) was detected and identified to be AGP. Figure 2. Example of a 2-DE gel of an undepleted serum (HC 2). Different protein spots corresponding to AGP, HPT, AT, FET and TRF were isolated by using 24 cm, ph 3 7 nonlinear immobiline dry strips and 10% polyacrylamide gels, followed by Coomassie staining. 3.3 N-glycan analysis of APP The increased SLe x found in AGP of stage IV PaC serum compared with healthy controls suggested that other APP may show an altered N-glycan moiety in tumour situations; therefore, we analyzed them in detail by N-glycan sequencing. 2-DE was used to isolate different APP including AGP, HPT, FET and AT (Fig. 2) from three healthy controls (HC 2, HC 4 and HC 5), four PaC patients from different stages (PaC 9 (stage II), PaC 7 (stage III), PaC 5 (stage IVa), PaC 8 (stage IVb)) and two CP patients (CP 1 and CP 2) that had been found SLe x positive for AGP. These protein spots were excised from the gel and screened for possible altered glycosylation by N-glycan analysis. After N-glycan release, the glycoproteins contained in each spot were identified using mass spectrometric analysis, as AGP (spot 1), FET (spots 2 4), AT (spots 5 7), HPT (spots 8 13) and TRF (spot 14) (Table 1). LC-ESI-Q-TOF analysis of spot 14 indicated that apart from TRF (score 642, 18 peptides matched), Ig mu chain (score 73, only two peptides matched) could also be present. Taking into account that Ig mu chain was not detected by MALDI-TOF-TOF, altogether indicates that this glycoprotein could be present in this spot, but in much minor proportion than TRF. N-glycans from each of the sample spots were identified using quantitative NP-HPLC combined with exoglycosidase digestions and the structural assignments were checked using Glycobase database matching ( ie/cgi-bin/public/glycobase.cgi). MALDI-TOF and negative ion ESI-MS N-glycan analyses were also performed to corroborate the N-glycan structures contained in each spot. In addition, N-glycans from AGP from a pool of the patients were subjected to WAX fractionation, prior to NP-HPLC, in order to assign the number of sialic acids in each peak. NP- HPLC of the N-glycan profiles of a representative spot of each APP is shown in Fig. 3. N-glycans identified in each protein spot are shown in Table 2. The relative percentages of the N-glycan structures for each individual protein spot were calculated in order to measure glycosylation changes between the different groups of patients. The N-glycans assigned in the AGP, HPT, AT and FET spots (spots 1 to 13) were all complex type, and presented bi-, tri- and tetra-antennary structures. TRF (spot 14) contained only bi- and tri-antennary glycans together with bisected

7 438 A. Sarrats et al. Proteomics Clin. Appl. 2010, 4, Table 1. Identification of the proteins in 2-DE spots by MS analysis Spot Equipment Identification a) Accession no. Protein score b) Sequence coverage (%) Peptides matched 1 MALDI-TOF-TOF AGP1 P MALDI-TOF-TOF AGP2 P LC-ESI-Q-TOF a-2-hs-glycoprotein/fetuin P LC-ESI-Q-TOF a-2-hs-glycoprotein/fetuin P LC-ESI-Q-TOF a-2-hs-glycoprotein/fetuin P MALDI-TOF-TOF a-1-antitrypsin P MALDI-TOF-TOF a-1-antitrypsin P MALDI-TOF-TOF a-1-antitrypsin P MALDI-TOF-TOF Haptoglobin P MALDI-TOF-TOF Haptoglobin P MALDI-TOF-TOF Haptoglobin P MALDI-TOF-TOF Haptoglobin P MALDI-TOF-TOF Haptoglobin P MALDI-TOF-TOF Haptoglobin P MALDI-TOF-TOF Transferrin P LC-ESI-Q-TOF Transferrin P LC-ESI-Q-TOF Ig mu chain C region P a) Only glycoproteins identified are listed (unglycosylated proteins are not listed). b) Only proteins with a significant MASCOT score (456) are listed for the MALDI analysis. For the ESI analysis, all the proteins identified with significant peptides scores (428) are listed. structures, which were not detected in the other APP. In addition, TRF had high mannose type N-glycans in a percentage that ranged from 3 to 11%, depending on the sample. High-mannose-type structures have not been described before for TRF. As spot 14 could include Ig mu chain, which has been reported to contain only high mannose structures [31], we cannot discard the possibility that these N-glycan structures originated from the Ig mu chain. The rest of the N-glycan structures of spot 14 can be attributed solely to TRF. The proportion of bi-, tri-, tetra-antennary, bisected and high-mannose structures were calculated for each APP spot. In order to estimate these changes for HPT, AT and FET, which are separated in several spots, the mean of the different spots of the same protein was calculated (Fig. 4A). Several N-glycans were outer-arm fucosylated and only one minor structure, a disialylated biantennary glycan F(6)A2G(4)2S2 in peak 7, present in all APP was core fucosylated. TRF contained in addition other three core fucosylated structures F(6)A2G(4)2S(6)1 in peak 4, F(6)A2BG(4)2S1 in peak 5 and F(6)A2BG(4)2S2 in peak 7. No N-acetylgalactosamine was detected. All galactoses were b-1-4-linked to N-acetylglucosamine (the same HPLC profile was obtained after digestion with bovine testes b-galactosidase, which hydrolyses non-reducing b-1-3- and b-1-4-linked terminal galactose, and with S. pneumoniae b-galactosidase, which hydrolyses non-reducing b-1-4-linked terminal galactose only). All outer-arm fucoses were determined to be a-1-3 linked, because they were digested by almond meal a-fucosidase (which releases a-1-3 and a-1-4 non-reducing terminal fucose) but not by Xanthomonus sp. a-fucosidase (which releases a-1-2 non-reducing terminal fucose). Substitution at position 4 was discarded because it was occupied by galactose. The summation of the relative peak areas containing sialylated structures with outer-arm fucose linked to N-acetylglucosamine, which is simultaneously attached to galactose (peaks 10, 13, 14 and 17 20), was taken as a measure of the SLe x antigen and was calculated for each spot. Although peak 13 contained a mixture of two structures with and without a-1-3 outer-arm fucosylation in AGP and HPT, the major one was always the outer-arm fucosylated tri-antennary glycan, A3F(3)1G(4)3S3 which was in a proportion of about 70% in AGP and 80% in HPT. In the rest of APP analyzed, FET, AT and TRF, peak 13 contained only the A3F(3)1G(4)3S3 structure. 3.4 Altered glycosylation of AGP The HPLC profiles of the N-glycans obtained from spot 1 (AGP) from controls, PaC and CP patient sera are shown in Fig. 3A. Most of the altered peaks were gradually increased or decreased compared with healthy controls according the following order: stages II III PaC, stage IV PaC and CP. Thus, stages II III PaC profiles were more similar to controls, while stage IV PaC and CP showed more marked differences compared with controls. Peaks 10, 13, 14 and were increased in PCa stage IV and pancreatitis and contained trior tetra-antennary structures with a-1-3 outer-arm fucosylation. The rest of the peaks (usually the non-fucosylated form of the increased ones) were either decreased or not altered. Taking into account that peak 13 of AGP contains a mixture of tri-antennary A3F(3)1G(4)3S3 (13a) and tetra-antennary A4G(4)4S3 (13b) in a proportion of about 70:30, the percentage of a-1-3 fucosylation (SLe x ) was calculated first by

8 Proteomics Clin. Appl. 2010, 4, Figure 3. NP-HPLC chromatograms of N-glycans released from (A) AGP spot 1, (B) HPT spot 8, (C) AT spot 5, (D) FET spot 2 and (E) TRF spot 14 from sera of healthy controls, PaC and CP patients.

9 440 A. Sarrats et al. Proteomics Clin. Appl. 2010, 4, Table 2. N-glycan structures identified on the APP Peak a) GU b) Structure c) Spot 1 (AGP) Spots 2 4 (FET) Spots 5 7 (AT) Spots 8 13 (HPT) Spot 14 (TRF) M5 d) M6 d) a A2G(4)2S1 3b M7 d) a A2G(4)2S(3)2 4b F(6)A2G(4)2S(6) a F(6)A2BG(4)2S1 5b M8 () d) a A2G(4)2S(6)2 6b A2G(4)2S(3,6) a F(6)A2G(4)2S2 7b F(6)A2BG(4)2S a A3G(4)3S(3)2 8b A3G(4)3S(3,6) A3G(4)3S(6) A3F(3)1G(4)3S A3G(4)3S A4G(4)4S a A3F(3)1G(4)3S3 70% 80%

10 Proteomics Clin. Appl. 2010, 4, Table 2. Continued. Peak a) GU b) Structure c) Spot 1 (AGP) Spots 2 4 (FET) Spots 5 7 (AT) Spots 8 13 (HPT) Spot 14 (TRF) 13b A4G(4)4S(3,?,?)3 30% 20% A4F(3)1G(4)4S a A4G(4)4S(6)3 15b A4G(4)4S (3,?,?,?)4 15c A3F(3)2G(4)3S A4G(4)4S(6) a A4F(3)1G(4)4S3 17b A4F(3)1G(4) 4S(3,?,?,?)4 17c A4 F(3)2G(4)4S a A4F(3)1G4S(6)4 18b A4F(3)2G4S A4F(3)2G4S4 20? A4F(3)3G4S4 a) Only peaks marked with a dot were detected on each protein. b) GU are mean values of the N-glycan peaks from all the glycoproteins analyzed. c) The N-glycan structures were assigned to each peak using quantitative HPLC, exoglycosidase digestion and Glycobase ( glycobase.nibrt.ie/cgi-bin/public/glycobase.cgi) combined with MALDI-TOF and ESI-MS. Structures not confirmed by MS are marked with an asterisk. Fetuin samples failed to give any spectra neither by ESI nor MALDI-TOF. d) High mannose structures could not be unambiguously assigned to transferrin. Minor structures are in brackets. adding the relative peak areas of A3F(3)1G(4)1, A4 F(3)1G(4)1, A4F(3)2G(4)2 and A4F(3)3G(4)3 generated after A. ureafaciens sialidase 1 bovine testes b-galactosidase digestion of each of the patients AGP N-glycans. Then, it was compared with that from the summation of the undigested peaks 10, 13, 14 and relative areas. The change in the proportion of SLe x

11 442 A. Sarrats et al. Proteomics Clin. Appl. 2010, 4, between the groups of patients was practically the same using the two methods, indicating that the structure 13b hardly influences the differences between the groups. Thus, in order to simplify the comparison with the results of the other APP, only the data of the undigested peaks are shown. The percentage of outer-arm fucosylation (SLe x ) was clearly increased in stage IV PaC ( ) and CP ( ) groups compared with controls ( ), whereas the stages II III PaC group showed a lower increase in outer-arm fucosylation ( ) (Fig. 4B). Branching was also increased in stage IV PaC and CP compared with controls. Thus, the percentage of tetra-antennary structures was increased in both stage IV PaC ( ) and CP ( ) compared with controls ( ). Interestingly, stage IV PaC showed a decrease of tri-antennary structures compared with controls ( versus ), whereas the CP group showed a decrease in biantennary glycans compared with controls ( versus ) (Fig. 4A). Peak 7 was the only peak containing core fucosylated structures (F(6)A2G(4)2S2). In spite of its low relative abundance, it was specifically increased in stage IV PCa (about 50%) compared with the other groups (Fig. 4C). 3.5 Altered glycosylation of HPT In all HPT spots (8 13), there was a gradual increase in the percentage of tri- and tetra-antennary structures with a-1-3 outer-arm fucosylation (peaks 10, 13 and 14), compared with controls according the following order: stages II III PaC, stage IV PaC and CP. However, these differences were more easily observed in spots 8 11, as spots 12 and 13 contain mainly biantennary structures, both mono- or disialylated and hardly any tri- or tetra-antennary structures. Generally speaking, the higher the isoelectric point of the protein spot, the lower the sialic acid content of the glycoprotein. In HPT spots, the most acidic spot (8) contained the lowest proportion of monosialylated structures and the highest proportion of tri- and tetra-sialylated structures, whereas the least acid spot (13) contained the highest proportion of monosialylated structures. N-glycan profiles of spot 8 are shown as an example of the changes observed (Fig. 3B). The percentage of outer-arm fucosylation (SLe x ) was increased in the CP group ( ) compared with controls ( ); however, it was just slightly increased in the stage IV PaC group ( ) and not increased for the stages I II PaC group (Fig. 4B). The changes in branching showed the same behaviour as for spot 1 (AGP): the percentage of tetra-antennary structures was increased in both stage IV PaC and CP compared with controls ( and versus ), and tri-antennary structures were decreased in stage IV PaC ( versus ), while biantennary ones where decreased in CP ( versus ) (Fig. 4A). Regarding core fucosylation, the low abundant peak 7 (F(6)A2G(4)2S2) was again increased in the stage IV PCa group compared with the rest of groups; however, the differences were not as pronounced as in AGP (Fig. 4C). 3.6 Altered glycosylation of AT In all AT spots, there was also a gradual increase in the size of peaks 13 and 14 (tri- and tetra-antennary structures with outer-arm fucosylation) compared with controls as follows: stages II III PaC, stage IV PaC and CP. However, these differences were more easily observed in spots 5 and 6. The N-glycan profiles of spot 5 are shown as an example of the changes observed (Fig. 3C). The percentage of a-13 outerarm fucosylation (SLe x ) was slightly increased in the CP ( ) compared with controls ( ) and it was not increased for any of the PaC groups (Fig. 4A). There were no evident changes of branching (Fig. 4A). Regarding core fucosylation, the low abundant peak 7 (F(6)A2G(4)2S2) was again not only increased in the stage IV PCa group ( ), but also in the CP group ( ) compared with controls ( ) (Fig. 4C). 3.7 Altered glycosylation of FET In all FET spots, a gradual increase in the size of peaks 13 and 14 (tri- and tetra-antennary structures with outer-arm fucosylation) was detected compared with controls as follows: stages II III PaC, stage IV PaC and CP. However, these differences were more easily observed in spots 2 and 3. The N-glycan profiles of spot 2 are shown as an example of the changes observed (Fig. 3D). The percentage of a-1-3 outer-arm fucosylation (SLe x ) was increased in the CP group around 2.5-fold compared with controls ( versus ) and it was not increased for any of the PaC groups. The percentage of tri- and tetraantennary structures was increased in CP compared with controls ( versus and versus ), while biantennary structures were decreased in CP ( versus ) (Fig. 4B). There were no significant changes regarding core fucosylation although an increased tendency in stage IV PaC was observed (Fig. 4C). 3.8 Altered glycosylation of TRF In TRF, only one peak (13) was found to contain a-1-3- fucosylated structures; thus, its relative percentage was taken as a measure of outer-arm fucosylation (SLe x ) and gradually increased compared with controls as follows: stages II III PaC, stage IV PaC and CP (Fig. 4B). Following the same gradation, the percentage of high mannose structures, bisected structures and tri-antennary structures gradually increased concomitant with a decrease in biantennary structures (Fig. 4A). Interestingly, and in a

12 Proteomics Clin. Appl. 2010, 4, Figure 4. Representation of branching (A), a-1-3 fucosylation (B) and core fucosylation (C) changes on AGP, HPT, AT, FET and TRF in healthy controls, stages II III PaC, stage IV PaC and CP patients. High mannose structures could not be unambiguously assigned to TRF. similar way to the other APP studied, peak 7 showed an increase in stage IV PaC compared with the rest of the groups, although, in this case, when comparing with CP, this increase was very small. When considering all peaks containing core-fucosylated structures (peaks 4, 5 and 7), CP was the group that showed the highest increase compared with controls (Fig. 4C). 4 Discussion 4.1 Altered glycosylation of APP in PaC and pancreatitis APP are plasma proteins secreted mostly by the liver and their plasma concentration increases (positive APP) or decreases (negative APP) by at least 25% in response to inflammation [5]. Most of these proteins are glycosylated and changes in their glycosylation pattern can be associated with disease [28, 32, 33]. Positive APP fulfil a variety of functions but principally are thought to help in the body response to infection and may also have a role in healing and repair [5, 34]. In our study, we have focused on the possible changes in the N-glycans of some of the major APP AGP, HPT, FET, AT and TRF in the serum of PaC patients. As controls, we have chosen serum from healthy people and also from CP patients that have a chronic inflammatory status, as changes of APP N-glycans have been described in inflammation [32]. Our results have shown several N-glycan modifications in stage IV PaC and in CP on the different APP analyzed. The main changes included increased a-1-3 fucosylation (SLe x ) and

13 444 A. Sarrats et al. Proteomics Clin. Appl. 2010, 4, branching in both stage IV PaC and CP and an increase of core fucosylation in stage IV PaC on most of the APP analyzed. 4.2 Changes in N-glycan branching An increase in the amount of tetra-antennary structures on AGP and HPT was observed in CP and in stage IV PaC. A concomitant decrease in tri-antennary structures was observed in stage IV PaC patients, while in CP patients, the percentage of biantennary glycans was decreased. FET showed the same changes in CP patients as AGP and HPT but more significantly, and in addition, an increase of triantennary structures was also observed. TRF showed a different repertoire of N-glycan structures than the other studied APP; apart from the main disialylated biantennary and minor tri-antennary N-glycans, it contained no tetra-antennary structures, but possessed bisected glycans, in agreement with other data previously reported [35 38]. In our work, we also found minor high mannose structures in each of the samples analyzed (from 3 to 11%); these probably originate from the Ig mu chain protein that appeared to contaminate the TRF sample. Interestingly, a change in branching pattern was also detected, namely a decrease in biantennary structures in both stage IV PaC and CP patients concomitant with an increase in bisected and tri-antennary glycans. Substantial increases in AGP glycoforms expressing biantennary glycans are apparent in the early phase of an acute-phase reaction. However, these are decreased compared with control levels after the second day following surgical trauma [39]. There is no agreement regarding branching changes observed in chronic inflammation. Some publications described an increase of biantennary glycans on AGP [40]. Others reported that AGP concanavalin A (ConA) reactivity showed a transition from initially elevated to decreased levels as disease became chronic, which indicates an increase in branching in chronic inflammation [41], consistent with the increased branching we observe in CP. HPT has previously been described as having an increased branching in ovarian cancer compared with controls [42] and in prostate cancer compared with benign prostatic hyperplasia [43]. Increase in TRF branching has been reported for HCC compared with healthy controls [38] and for chronic inflammation, in particular in different types of rheumatoid arthritis [44]. 4.3 Changes in SLe x expression AGP showed an increased SLe x content in pancreatitis and stage IV PaC patients compared with controls by two different approaches, Western blot and N-glycan sequencing. This APP has been described to carry an increased a-1-3 fucosylation in patients with acute inflammation [39, 40] and chronic inflammation such as rheumatoid arthritis and diabetes mellitus [40, 45, 46]. An increase in SLe x expression on AGP has also been reported in both acute and chronic inflammations [47]. a-1-3 fucosylation on AGP was proposed as a marker of progression and prognosis in different types of malignancies in a study with 214 patients [48]. In both advanced ovarian and breast cancers an increase of SLe x in serum AGP have been described and indeed proposed as a marker of disease progression in the case of breast cancer [49, 50]. The rest of APP, such as HPT, FET, AT and TRF, showed an increased a-1-3 fucosylation (SLe x ) for the CP patients compared with controls, which was much more pronounced on FET. Taking into account that the CP group showed also other immunoreactive bands for SLe x detected by Western blot, altogether these data suggest that this sialylated antigen is closely linked to an inflammatory condition of the pancreas. In the case of HPT, stage IV PaC patients also showed an increased SLe x compared with controls. Increases in fucosylated HPT have been observed in both acute (severe trauma) and chronic inflammations (rheumatic arthritis and inflammatory bowel disease) [42, 47] and in different types of cancers including advanced ovarian cancer [49, 51], mammary carcinoma [52], prostate cancer [43] and lung cancer [53]. An increase of total fucosylation on HPT has been described in PaC [50] and, in agreement with our results, the incidence of positive HPT fucosylation (Aleuria aurantia lectin affinity) increased in advanced stages [54]. In another work [55], the authors described that SLe x -type fucosylation on HPT also increased in pancreatitis patients but to a lower extent than in PaC patients. On the contrary, we have observed a higher HPT SLe x -type fucosylation in pancreatitis than in PaC patients. Interestingly, we observed a significant increase in the a-1-3 fucosylation of FET (2.5- fold) in the pancreatitis patients that should be further investigated with more serum samples to define a possible clinical implications. Although an increased fucosylation on AT has been described in other types of cancer (breast and ovarian) [56], we did not observe an increase in PaC patients and only pancreatitis patients showed a slight increase of a-1-3 fucosylation on AT. The increase in a-1-3 fucosylation of TRF has been detected in the tri-antennary structure present in peak 13. a-1-3 fucosylation increase has also been reported in the tri-antennary N-glycans of TRF for HCC patients using lectin and Bio-Gel P4 chromatography [38]. 4.4 Changes in core fucosylation While our results indicate that increase in branching and a-1-3 fucosylation on APP is a common feature of both advanced PaC and pancreatitis, increase of the minor-core-

14 Proteomics Clin. Appl. 2010, 4, Clinical Relevance PaC lacks any specific and sensitive tumour marker. As other carcinomas, it is associated with changes in cellular glycosylation and release of inflammatory cytokines. Serum APP, which are secreted by the liver, may modify their glycosylation in tumours and could be potential sensors of this disease. The glycosylation analysis of major APP in PaC, CP and control patients has shown that some N-glycan changes were associated to an inflammatory condition, as they were detected in advanced stages of PaC and in CP, while others (core fucosylation) could only be detected in advanced PaC patients and therefore could be regarded as cancer specific. The evaluation of a larger cohort of patients is required to define the potential clinical utility of increased core fucosylation of AGP and HPT as PaC markers. However, these results highlight the importance of determining the glycosylation changes of serum proteins, such as APP, as this post-translational modification is modulated in pathophysiological conditions and therefore it may be useful as a disease marker. fucosylated biantennary structure F(6)A2G(4)2S2 seems to be a more cancer-specific modification. Although the percentage of this glycan is increased in all the APP, it is only significantly augmented at advanced stages of PaC patients on AGP and HPT. AGP and HPT showed an increased core fucosylation in lung cancer patients compared with healthy controls [57]. In HCC, an increase of core fucosylation in AGP, AT, FET and TRF has been described compared with controls and other chronic liver diseases [38, 36, 58, 59]. An increase of core fucosylation on HPT has also been reported in HCC compared with chronic liver disease [60], although another studyshowedanoppositetendencywhencomparinghpt from HCC and control serum [58]. The increase in core fucosylation for most APP described in HCC has been usually detected using specific lectins, namely Lens culinaris agglutinin. Increase of core-fucosylated HPT has been described in PaC patients (Aspergillus oryzae lectin affinity) compared with controls [54] in agreement with our results. In our work, we have performed detailed N-glycan sequencing, which can discriminate between core and outer-arm fucosylation and gives more precise information than lectin-binding studies. In addition, we have also included the pancreatitis patients and have shown that increased core fucosylation is a cancerspecific modification of AGP and HPT. Other glycoproteins show an increased core fucosylation in cancer. Core fucosylated a-fetoprotein, called AFP-L3, is very specifically increased in HCC and was approved as a tumour marker by the food and drug administration in 2005 [61]. Serum ribonuclease 1 was also much more core fucosylated in PaC sera than in control patients [62] which suggests that quantification of core fucosylation of some PaC serum glycoproteins might be useful for diagnostic or prognosis purposes. 4.5 Regulation of glycosylation in APP Cytokines involved in the induction of the inflammatory reaction have been described to regulate both APP synthesis and glycosylation [5]. Their effect has been studied in vitro using hepatoma cell lines. Stimulation of the hepatic carcinoma cells HUH-7 with proinflammatory cytokines such as IL-1b and/or IL-6 for 2 days increased AGP production and ConA reactivity (biantennary glycoforms). However, there was a decrease on AGP ConA reactivity after 5 days of stimulation, probably linked to a decrease of biantennary structures (increase of branching) [63], consistently with the present study where an increase of branching is observed in chronic inflammation (pancreatitis). However, the mechanism by which these cytokines cause the branching modifications of APP has not been studied. Increases in tetra-antennary structures have been related to an upregulation of N-acetylglucosaminyltransferase V (GnT-V), the enzyme responsible for the b-1-6 branching [64]. Thus, this enzyme may be overexpressed in liver cells of both pancreatitis and stage IV PaC patients, as both illnesses show an increase of tetra-antennary structures in AGP and HPT. In stage IV PaC patients, a concomitant decrease in tri-antennary structures was observed while in pancreatitis biantennary glycans were decreased. These results suggest that N-acetylglucosaminyltransferase IV (GnT-IV), the enzyme responsible for the addition of the b-1-4 branching (third antennae), may be upregulated in CP. Stimulation of HUH-7 cells with IL-1b also increased SLe x on secreted AGP by enhancing the expression of the b-galactoside a-2,3-sialyltransferase IV (ST3 Gal IV) and fucosyltransferase VI (FUT VI) [16, 63]. Stimulation of hepatoma cells with IL-6 increase both HPT synthesis and fucosylation, which was linked to an enhanced expression of fucosylation-related genes such as FUT6 and FUT8, GDPfucose synthase and GDP-mannose-4,6-dehydratase [65]. Cytokines might play a similar role in vivo and modify APP glycosylation. The most important sources of cytokines are macrophages and monocytes at the inflammatory sites [5]. In a chronic inflammatory condition such as CP, these cytokines would be produced in the damaged pancreas and produce various systemic effects including liver stimulation to increase APP synthesis and specific glycosyltransferases responsible for theincreaseofappbranchingandsle x -type fucosylation. A similar mechanism may take place in cancer, as tumours and