OUP user menu

Journal of Integrated Pest Management


Laurence J. Egan, Ireland

Associate Editors

Maria T. Abreu, USAShomron Ben-Horin, IsraelSilvio Danese, ItalyPeter Lakatos, HungaryMiles Parkes, UKGijs van den Brink, The NetherlandsSéverine Vermeire, Belgium


Published on behalf of

Autoantibodies against exocrine pancreas in Crohn's disease are directed against two antigens: The glycoproteins CUZD1 and GP2

Lars Komorowski, Bianca Teegen, Christian Probst, Karola Aulinger-Stöcker, Christian Sina, Klaus Fellermann, Winfried Stöcker
DOI: http://dx.doi.org/10.1016/j.crohns.2012.10.011 780-790 First published online: 1 November 2013


Background: Autoantibodies against exocrine pancreas (PAb) have been reported to be pathognomonic markers of Crohn's disease (CD). Recently, the glycoprotein GP2 has been proposed as the exclusive target for PAb but two equally prevalent binding patterns can be observed in the indirect immunofluorescence test (IIFT) using cryosections of human pancreas: a reticulogranular and a droplet pattern.

Aim: To identify autoantigens corresponding to the staining patterns.

Methods: Different lectins were screened for their ability to immobilize PAb-reactive glycoproteins from cell free human pancreas. The glycoproteins were then purified via UEA-I affinity chromatography and identified by mass spectrometry. The two candidate autoantigens were separately expressed in HEK293 cells, and the recombinant cells applied as substrates in IIFT to analyze sera from 96 patients with CD, 89 controls and hybridoma supernatants during the generation of murine monoclonal antibodies.

Results: The UEA-I eluate was able to neutralize PAb reactivity of both patterns in IIFT. It contained two major constituents which were identified as the glycoproteins CUZD1 and GP2. With the recombinant cells, 35.4% of the CD patients exhibited positive reactions (CUZD1 alone 19.8%, GP2 alone 9.4%, and both antigens 6.2%). The reaction with the CUZD1 expressing cells was strictly correlated to the reticulogranular pattern, whereas the antibodies causing the droplet pattern stained the GP2 expressing cells. Antigen-capture ELISA using the newly generated monoclonal antibodies against CUZD1 and GP2 verified this relationship.

Conclusions: The concordant reactivities of the different platforms can be regarded as a proof for the authenticity of the two identified autoantigens.

  • Crohn's disease;
  • Autoantibodies;
  • Exocrine pancreas;
  • CUZD1;
  • GP2
inflammatory bowel disease
ulcerative colitis
Crohn's disease
matrix-assisted laser desorption/ionization trap-time of flight mass spectrometry
quadrupole ion trap
receiver-operating characteristics
area under the curve
sodium dodecylsulfate
polyacrylamide gel electrophoresis
fluorescein isothiocyanate
indirect immunofluorescence test
horse radish peroxidase
alkaline phosphatase
Ulex europaeus agglutinin I

1 Introduction

Crohn's disease (CD), one form of chronic inflammatory bowel disease (CIBD), affects up to 0.15% of the population in industrialized countries.13 The cause of the disease is unknown, immune mechanisms may be involved in the pathogenesis.

Research on the humoral immunity in CD revealed the presence of autoantibodies against acinar cells of the exocrine pancreas (PAb) in the sera of CD patients.2,48 PAb have been reported to be pathognomonic markers with a prevalence of 39% in the indirect immunofluorescence test2,4,8 and can be used to differentiate CD from ulcerative colitis (UC; prevalence 2%). Two different staining patterns were observed: a reticulo-granular and a droplet pattern. Apart from PAb, other serological CIBD markers are established: A) antibodies against the cell wall mannan of Saccharomyces cerevisiae (ASCA) with 67% prevalence in CD (immunoglobulin classes IgA and IgG together), which are rarely found in UC, but also in 25% of patients with celiac disease9,10; B) autoantibodies against the cytoplasm of neutrophil granulocytes (perinuclear pattern, P-ANCA), which appear in 67% of UC but in only 7% of CD patients and are mainly directed to DNA-bound lactoferrin11; and C) autoantibodies against intestinal goblet cells, which exclusively appear in 28% of UC patients.2,12,13

Next to their high value as CD-specific yet non-invasive markers in the diagnostic work-up of CIBD, particularly in the differential diagnosis of CD and UC,14 PAb have also been proposed as markers for clinical presentations of the disease, e.g. microbial load, severity of penetration, site of inflammation, age at disease onset, duration of the disease, co-appearance of other diseases etc.4,1517 Nevertheless, none of these links have so far been confirmed by independent studies such that they cannot be exploited for the clinical practice, yet.

More recently, the pancreatic major glycoprotein GP2 of the zymogen granule membrane has been described independently by us and another group as a target for the associated autoantibodies.1822 Anti-GP2 represent, however, only a proportion of CD associated PAb.2224 In this work, it was possible to identify the two CD-relevant target antigens of PAb, combining lectin-based affinity chromatography with mass spectrometry. For verification, the antigens were recombinantly expressed in a human cell line and used in a cell-based immunofluorescence test for the serological diagnosis of CD.

2 Materials and methods

2.1 IBD patients and healthy controls

Serum samples were obtained consecutively from 96 patients with CD (65 female, median age 40, range 18–73 years) and 39 patients with UC (24 female, median age 42, range 17–83 years) attending the Medical Clinic I of the University of Lübeck between August 2005 and March 2007. Most of them were established patients and had already started therapies of different regimens. They met internationally accepted clinical, endoscopical, radiological and histological criteria compatible with either CD or UC.25,26 Patients with unclear diagnosis and with features suggestive of other coexistent intestinal diseases were excluded. In case of multiple patient appearances, only the sample from the first admission was selected. Presence of clinical signs and symptoms and all relevant information were recorded in electronic databases. Relationships between clinical features and the appearance of autoantibodies against exocrine pancreas (PAb) were, however, not evaluated because A) the study was focussed on the definition of PAb target antigen, B) the number of sera containing PAb was too low for the generation of statistically significant results, and C) the heterogeneity of the patient cohorts with respect to treatment regimens.

A control collective consisted of 50 healthy blood donors (32 females, median age 26, range 18–52 years). The study was designed in December 2004 when the prototypic immunoassays (see below) were established for autoantibody testing, making this a retrospective cohort study.

The experiments conducted were institutionally approved by the ethics committee at the medical faculties of the University of Lübeck (institutional board projects 05-112). In adherence to the Helsinki principles, informed consent was obtained from all patients whose material was used in this study.

2.2 Indirect immunofluorescence test

Autoantibodies in human sera were determined by IIFT using frozen sections of unfixed human pancreas according to the manufacturer's instructions (Euroimmun) as well as wild-type or genetically modified HEK293 cells (generated as stated below). In some cases, monoclonal murine antibodies were used in the first step of IIFT, followed by incubation with Cy2 or Cy3 anti-mouse IgG (Jackson Research Europe, United Kingdom). In neutralization experiments, samples with potential antigen content were mixed with the diluted PAb serum 30 min prior to the IIFT as described.2 In all cases, the incubated slides were evaluated by two experts.

2.3 Cell-free pancreas supernatant

Pieces of human pancreas were cut to small cubes with scissors and then minced thoroughly in 4 volumes 50 mmol/L ethanolamine pH 10, 150 mmol/L sodium chloride, 5 mmol/L EDTA, 0.1% (w/v) Triton X-100 containing as proteinase inhibitors Complete Mini, EDTA free (Roche, Switzerland) on ice with a Miccra D-8 (ART Labortechnik, Germany). The suspension was filtered through glass wool and then centrifuged for 60 min at 100,000 ×g, 4 °C. The supernatant was exchanged against TNE (50 mmol/L Tris–HCl pH 7.4, 150 mmol/L sodium chloride, 5 mmol/L EDTA, 1 mmol/L PMSF) by dialysis. Aliquots were stored at -70 °C until further use.

2.4 Search for lectins reacting with CD-related pancreatic autoantigens

Streptavidin-coated microplates (Euroimmun) were incubated with 0.5 μg/mL biotinylated lectins (Vectorlabs, USA) in PBS (50 mmol/L sodium dihydrogen phosphate pH 7.4, 150 mmol/L sodium chloride) for 60 min and washed with PBS, containing 0.05% (v/v) Tween-20. Subsequently, cell-free pancreas supernatants at varying dilutions in a sample buffer (0.5% w/v BSA in PBS, containing 0.05% v/v Tween-20) was applied to the wells for 60 min. Bound antigen was detected by incubation with PAb positive CD sera, 100-fold diluted in a sample buffer for 60 min, followed by 30 min of anti-human IgG HRP conjugate (Euroimmun) and 15 min of tetramethylbenzidine substrate (Euroimmun). The indicator reaction was stopped by the addition of 1 volume 1 mol/L sulfuric acid. The OD was read at 450 nm using an automated spectrophotometer (Tecan, Germany). Individual blank values (same lectin and serum samples, no pancreas fraction) were subtracted to compensate for direct binding of human antibodies via their carbohydrate moieties.

2.5 Lectin chromatography

Ulex europaeus agglutinin I (UEA I) conjugated to agarose (Vectorlabs, USA) was equilibrated with 5 volumes of TNE supplemented with 1 mmol/L fucose (Sigma, Germany) to prevent unspecific binding. After the application of the supernatant, the matrix was washed with 10 volumes of equilibration buffer. Finally, bound glycoproteins were eluted with 100 mmol/L fucose in the TNE. Fractions of the eluate were concentrated by 10 kDa MWCO ultrafiltration, dialysed extensively against TNE and stored at − 70 °C until further use.

2.6 ELISA detection of autoantibodies against exocrine pancreas

For a combined antigen (CUZD1 + GP2) ELISA, microplates (Nunc, Denmark) were coated with 0.8 μg/mL UEA I-purified pancreas proteins in PBS for 60 min and then blocked with 0.1% (w/v) casein in PBS. Human sera were diluted 100-fold in a sample buffer (1% w/v casein in PBS, containing 0.05% v/v Tween-20), and the dilution applied to the wells for 30 min followed by the anti-human IgG HRP conjugate for 30 min, and tetramethylbenzidine substrate (Euroimmun) for 15 min. The optical density was read at 450 nm using an automated spectrophotometer (Tecan, Germany).

Autoantibodies against CUZD1 and GP2 were selectively analyzed by two antigen capture ELISAs: Microplates were coated with 1 μg/mL monoclonal antibodies against either CUZD1 or GP2 in PBS for 60 min, blocked with 0.1% (w/v) casein in PBS, and then loaded separately with CUZD1 and GP2 in PBS obtained by UEA I-affinity chromatography (0.8 μg/mL glycoprotein) for 30 min. Incubations with human sera, anti-human IgG HRP conjugate and tetramethylbenzidine substrate were performed as in the combined antigen ELISA.

2.7 Cloning and eukaryotic expression of CUZD1 and GP2

The cDNA coding for mature GP2 (mGP2-coli) without its N- and C-terminal signal peptides according to EMBL acc. no. AB035541 was amplified from human pancreas mRNA (TAKARA, Germany) employing the one step RT-PCR system (QIAGEN, Germany) according to the manufacturer's instructions. Primers are given in Table 1. The amplification products were digested using NcoI and XhoI and ligated into the pET24d expression vector (Novagen, Germany). The protein was then over-expressed in Escherichia coli and purified essentially as described by Probst et al. for envoplakin fragments.27

View this table:
Table 1

Primer sequences for RT-PCR amplification of cDNAs of CUZD1 and GP2. F, forward primer; R, reverse primer.

ProteinRestriction sitesPrimer sequences (5′-->3′)

DNA codings for full-length CUZD1 according to EMBL acc. no. BC136755 and full-length alpha GP2 according to acc. no. AB035541 were amplified from human pancreatic mRNA (TAKARA, Germany) employing the one step RT-PCR system (QIAGEN, Germany) according to the manufacturer's instructions. Primers are given in Table 1. The amplification products were digested using HindIII and XhoI, and ligated into a modified pCEP4 expression vector (Invitrogen, Germany) from which the EBNA1 coding sequence had been removed. Glycosylation deficit variants of recombinant CUZD1 and GP2 were generated by exchanging codons for asparagine against codons for glutamine residues in putative sequons (site directed mutagenesis with suitable primer pairs) to prevent attachment of N-glycans at the respective sites. Similarly, in one variant of GP2, the codon for asparagine505 was mutated to prevent attachment of the GPI anchor. In one other GP2 variant, the C-terminal signal sequence was deleted.

Both proteins and their variants were individually expressed in the human cell line HEK293: For transfection, ExGen500 (Fermentas, Germany) was used according to the manufacturer's instructions. For the IIFT, HEK293 were grown on glass slides, transfected separately, and allowed to express the recombinant proteins for 48 h. Slides were washed with PBS, fixed in acetone for 10 min at room temperature, air-dried and stored at − 20 °C until use.

2.8 Generation of antibodies against CUZD1 and GP2

Polyclonal antibodies were generated by immunizing rabbits with mGP2-col following a standard 87-day program (Eurogentec, Belgium). Serum reactivity was controlled by Western blot using the immunogen.

Monoclonal antibodies were raised by immunization of mice with UEA I-purified pancreatic proteins at Eurogentec, using a standard immunization protocol. Mouse sera as well as supernatants of hybridoma cultures were screened for the presence of antibodies against CUZD1 and GP2 by IIFT using frozen sections of human pancreas and transfected HEK293 as substrates. Antibodies from positive clones reacting with either CUZD1 or GP2 were purified by protein A sepharose affinity chromatography (GE Healthcare, Germany) according to the manufacturer's manual, dialysed against PBS, concentrated to 0.5 mg/mL, and stored at − 20 °C until use. Clones HL1b (anti-CUZD1) and HL2a (anti-GP2) were selected for further experiments.

2.9 SDS-PAGE and Western blots

SDS-PAGE and electrotransfer onto nitrocellulose membranes (Pall, United States) were conducted with the NuPAGE system (Invitrogen) following the manufacturer's instructions. Some proteins were also analyzed by MALDI-TOF fingerprinting and MALDI-QIT-ToF tandem mass spectrometry after SDS-PAGE and tryptic cleavage.28 Protein concentrations were determined by bicinchoninic acid assay (Sigma).

2.10 Statistical analysis

All statistical analyses were performed using the MedCalc V9.6.2.0 software. In general, Fisher's exact test was used to calculate statistical significances of independence in contingency tables.

3 Results

All sera used in this study were initially screened for the presence of PAb by IIFT using human pancreas as substrate (Table 2). Both patterns could be observed alone or in combination in the CD sera, the reticulogranular pattern as well as the droplet pattern. None of the healthy blood donors exhibited antibodies against exocrine pancreas.

View this table:
Table 2

Prevalence of PAb in 96 patients with Crohn's disease as determined by indirect immunofluorescence using human pancreas sections and transfected HEK293 cells, respectively. Sera from 50 healthy blood donors did not show PAb reactivity on any of the substrates.

IgG onlyIgA onlyBothΣ
Human pancreasReticulogranular pattern only9.4%3.1%3.1%15.6%
Droplet pattern only5.2%0.0%3.1%8.3%
Both patterns3.1%1.1%2.1% a5.3%
Transfected HEK293CUZD19.4%8.3%2.1%19.8%
Both3.1%0.0%3.1% a6.2%
  • a One serum contained IgA against CUZD1 (reticulogranular pattern) and IgG against CUZD1 and GP2 (reticulogranular and droplet patterns) but no IgA against GP2.

3.1 Reactivity of the CD related target autoantigens with different lectins

As analyzed by IIFT, the cleared supernatant of homogenized human pancreas was able to abolish PAb reactivity, thereby verifying the presence of PAb targets in the soluble fraction. The reactivity of the targets with a set of lectins is shown in (Fig. 1A). Strong reactions were observed with UEA I, DBA, SBA, WGA, and RCA, compared to missing reactions with PNA or ConA. UEA I was chosen for further preparations.

Figure 1

Binding of pancreatic autoantigens to lectins using a glycoprotein-capture ELISA (A): streptavidin-coated microplates were used to immobilize biotinylated lectins (UEA I, WGA, DBA, SBA, RCA, PNA, or ConA). The solid phases were then incubated, subsequently with cell-free pancreas lysate, human sera from patients with Crohn's disease, anti-human IgG HRP conjugate, and tetramethylbenzidine substrate. Photometry was carried out at 450 nm after the addition of 1 volume 1 mol/L H2SO4. For each serum sample, the OD values of a second incubation omitting the cell-free pancreas lysate were subtracted. Results of representative incubations. (B) Coomassie-stained SDS-PAGE of the UEA I-purified pancreas fraction. Lane 1: molecular mass markers; kDa indicated; lane 2: 4 μg UEA I-purified pancreas fraction.

3.2 Nature of the UEA I affinity chromatography purified pancreatic autoantigens

The fraction retained by UEA I agarose was able to neutralize PAb in all 31 positive sera (29 CD, 2 UC) used within this study. This was determined by IIFT as well as by an indirect ELISA using the glycoproteins immobilized to microplates via UEA I as solid phase.

When analyzed by SDS-PAGE, the UEA I eluate displayed only two broad protein bands of about 70 and 100 kDa (Fig. 1B). The bands were excised from the gel, treated with trypsin and applied to MALDI-TOF fingerprinting. At both positions, several signals corresponding to the membrane major glycoprotein GP2 of the pancreatic zymogen granules were detected. Additionally, mass signals corresponding to the CUB/zona pellucida-like domain-containing protein CUZD1 were found at the position of the larger protein. Four fragments of each protein were verified by de novo MALDI-QIT-TOF sequencing.

3.3 UEA I-purified proteins bind autoantibodies only in a non-denatured form

The UEA I-purified pancreas fraction was then used to coat the solid surface in an ELISA protocol for the determination of PAb (IgG only because of the limited additional value of IgA). All sera used within this study were analyzed for their reactivity. Of the CD sera, 24 (25.0%) produced positive signals, but only 1 (2.6%) of the UC sera and none of the samples from healthy blood donors (Fig. 2). ELISA and IIFT results agreed significantly (P < 0.0001; IgG detection) in the 96 CD and 39 UC and 50 HBD cases (Table 3). In contrast, after reducing or non-reducing SDS-PAGE of the UEA I-purified proteins, none of the sera was reactive in the Western blot. Thus, the patients' autoantibodies seemed to react only with conformational epitopes of both glycoproteins, CUZD1 and GP2.

Figure 2

ELISA reactivity of the UEA I-purified pancreas fraction. Microplates were used to immobilize the UEA I-purified pancreas fraction. The solid phase was then incubated, subsequently with human sera, anti-human IgG HRP conjugate, and tetramethylbenzidine substrate. Photometry was carried out at 450 nm after the addition of 1 volume 1 mol/L H2SO4. Sera from 96 Crohn's disease patients, 39 ulcerative colitis patients, and from 50 healthy blood donors were analyzed.

View this table:
Table 3

Correlation of IgG reactivities in indirect immunofluorescence using human pancreas sections, and in ELISA using UEA I-purified pancreas fraction. Fisher's exact test showed P < 0.000001 for the correlation of exocrine pancreas- and UEA I-purified pancreas fraction-associated antibodies.

n = 185 IgGIIFT on human pancreas
ELISA with UEA I-purified pancreas fractionPositive241

3.4 The two PAb patterns correlate with anti-CUZD1 and anti-GP2

Using frozen sections of the human pancreas, PAb were detectable in 29 (30.2%) CD and in 2 (5%) UC patients by IIFT. The reticulogranular pattern appeared in 15 patients, the droplet pattern in 10, and both patterns were observed in 6 of all positive patients (Table 4). Transfected HEK293 cells expressing CUZD1, used as substrates in IIFT, reacted with PAb of the reticulogranular pattern (Fig. 3A), whereas GP2-HEK reactions were significantly associated with the droplet pattern (Fig. 3B). Of the 39 sera from patients with ulcerative colitis, the two positive showed conclusive reactivity (one with the reticulogranular pattern: anti-CUZD1; the other with the droplet pattern: anti-GP2). None of the healthy blood donors' sera reacted with one or both recombinant substrates.

Figure 3

Immunofluorescence double staining of human pancreas and transfected HEK293: frozen sections of human pancreas and HEK293 transfected with either CUZD1 or GP2 were co-incubated with 1:100 diluted human serum and monoclonal antibodies. Bound antibodies were visualized with anti-human IgG Cy3 and anti-mouse IgG Cy2 conjugate. (A) PAb of the reticulogranular pattern versus 0.8 μg/mL monoclonal antibody against CUZD1; (B) PAb of the droplet pattern versus 0.8 μg/mL monoclonal antibody against GP2. (C) Frozen sections of human pancreas were co-incubated with 0.8 μg/mL Cy3-labeled anti-CUZD1 and 0.8 μg/mL Cy2-labeled anti-GP2 monoclonal antibodies. The scale bar represents a width of 20 μm.

View this table:
Table 4

Correlation of indirect immunofluorescence using human pancreas sections or HEK293 expressing either CUZD1 or GP2. Fisher's exact test showed highly significant correlations of reticulogranular pattern- and CUZD1-associated antibodies (P < 0.000001) and of droplet pattern- and GP2-associated antibodies (P < 0.000001).

n = 185 IgGHuman pancreas
Reticulo-granular patternDroplet patternBothNone
Transfected HEK293CUZD111001
Transfected HEK293CUZD17004

Five CD sera showed positive reactions with the CUZD1-, one serum with GP2-expressing cells and 1 serum with both, but the antibodies could not be observed with the frozen pancreas sections. Thus, by additional application of the recombinant cells to the sections, PAb prevalence in the Crohn's disease patients of this study could be increased from 30.2% to 35.4%. As tested using frozen sections and recombinant cells, in the positive cases, IgA alone was prevalent in 27%, IgG in 51%, both in 24%. In general, discrimination between positive and negative reactions was easier with the transfected cells, since the tissue sections occasionally showed some interfering unspecific fluorescence caused by antibodies against cytokeratin, mitochondria, ribosomes and other antigens.

Correspondingly, homogenates of CUZD1-expressing HEK293 were able to neutralize PAb in sera exhibiting the reticulptogranular, but not the droplet staining pattern and homogenates of GP2-HEK293 abolished PAb reactivity of the droplet, but not the reticulogranular pattern. Homogenized non-transfected cells did not show any neutralizing effect at all.

3.5 Features of murine monoclonal PAb

In 3 of 4 mice immunized with UEA I-purified pancreatic glycoproteins, it was possible to induce antibodies reacting with human exocrine pancreas as detected by IIFT. The spleens were used to generate hybridomas. Supernatants were selected which presented reticulogranular and droplet patterns. Finally, four clones were established exhibiting reticulogranular patterns and 5 clones exhibiting droplet patterns. In concordance with the CD-associated PAb, in all these monoclonal antibodies raised, the granular pattern was associated with anti-CUZD1 reactivity and the droplet pattern with anti-GP2 reactivity, as proved by IIFT. Furthermore, monoclonal anti-CUZD1 co-localized significantly with PAb of the reticulogranular pattern (Fig. 3A). In contrast, monoclonal anti-GP2 co-localized mainly with PAb of the droplet pattern (Fig. 3B). Double stainings with fluorophore-labeled monoclonal anti-CUZD1 and anti-GP2 showed that both glycoproteins were not exactly co-localized although both antibodies stained the secretory storage granules (Fig. 3C).

3.6 Discriminating properties of CUZD1- and GP2-capture ELISA systems

Clone HL1b (specific for CUZD1) and clone HL2a (specific for GP2) in combination with UEA I-purified pancreatic glycoproteins were used to form the solid phase in a two separate antigen capture ELISA. Applying these two different substrates in parallel, it was possible to clearly distinguish between both PAb specificities: anti-CUZD1 and anti-GP2. Positive reactions in the CUZD1 capture ELISA correlated significantly (P < 0.000001) with the reticulogranular pattern in IIFT (Fig. 4A), whereas the GP2 capture ELISA agreed (P < 0.000001) with the droplet pattern (Fig. 4B).

Figure 4

Antigen capture ELISA for the determination of human IgG antibodies. Microplates were used to immobilize monoclonal antibodies against either CUZD1 or GP2. The solid phase was then incubated, subsequently with the UEA I-purified pancreas fraction, human sera, anti-human IgG HRP conjugate, and tetramethylbenzidine substrate. The sera from 96 Crohn's disease patients, 39 ulcerative colitis patients, and from 50 healthy blood donors were analyzed. (A) Indirect ELISA as in Fig. 2 versus CUZD1 capture ELISA; (B) indirect ELISA as in Fig. 2 versus GP2 capture ELISA. None of the healthy blood donor sera exhibited reactions above the cut-off limits of each test system.

3.7 PAb reactive epitopes of CUZD1 and GP2

In further experiments, different mutants of CUZD1 and GP2 were tested for PAb reactivity. Silencing of all N-glycan sequons (Δglyco-CUZD1) by an exchange of asparagine against glutamine (N29Q, N53Q, N57Q, N67Q, N148Q, N271Q, N370Q, N394Q, and N419Q) resulted in a protein expressed by HEK293 which was still able to react in IIFT with all four monoclonal antibodies and all anti-CUZD1 positive human sera. In contrast, complete silencing of the N-glycan sequons in GP2 (N58Q, N81Q, N115Q, N127Q, N194Q, N206Q, N250Q, N281Q, and N352Q) led to a total loss of its reactivity with all five monoclonal antibodies and with all human anti-GP2 of each patient, although the recombinant protein was able to react with the polyclonal anti-GP2 rabbit serum. Partial silencing was associated with reduced PAb reactivity (data not shown).

Three of the 5 monoclonal antibodies, but none of the anti-GP2 positive human sera reacted with the secreted GP2-ΔGPI variant (lacking the C-terminal signal peptide, which is responsible for the attachment of the glycosyl phosphatidyl inositol anchor) and with GP2-ΔN505 (lacking its attachment site, asparagine residue 505).

4 Discussion

Autoantibodies against exocrine pancreas (PAb) are pathognomonic markers of Crohn's disease.2,4,8,29 At the 6th International Congress on Autoimmunity 2008 in Porto, Portugal, the corresponding antigens have already been reported by two independent groups: Conrad et al.20 and our group.18 Whereas Conrad et al. restricted the investigations to GP2, we have additionally identified the equally relevant antigen CUZD1 as a major target. Ignoring CUZD1, a considerable proportion of PAb kept unrevealed in the following studies using an ELISA system based on recombinant GP2.2224,30 Reactions with CUZD1 contribute to more than half of the PAb-based diagnosis of Crohn's disease!

Both, CUZD1 and GP2, are glycosylated membrane proteins residing in the acinar secretory storage granules of the pancreas. The pattern caused by the corresponding autoantibodies on pancreas tissue sections is the same as that obtained by labeling with different FITC-labeled lectins such as SBA and UEA I (referring to our own observations). Both autoantigens bind to the lectins SBA and UEA I, indicating the presence of fucose and N-acetyl-galactosamine moieties.2,8,31,32 Moreover, an SBA-based capture assay was able in an earlier work to demonstrate the binding of PAb to glycosylated antigens in pancreatic juice.33 Interestingly to note, CUZD1 and GP2, though characterized as membrane proteins, are constituents of pancreatic juice2 and can easily be released in vitro from pancreatic acinar membranes by treatment with a mild alkaline buffer.34

Conrad et al. identified GP2 by 2D gel electrophoresis: rat pancreatic fractions were first separated by isoelectric focussing and denaturing SDS-PAGE. Subsequently, the serum from a patient with a high-titer PAb was used to determine the position of the corresponding autoantigen in a Western blot.20,22 In our hands, not any PAb of the patients with Crohn's disease showed a reaction with the denatured forms of human CUZD1 and GP2, as are presented in the Western blot and in bacterially expressed recombinant GP2 that was purified under denaturing conditions. Antibodies against linear epitopes of GP2 could only be induced by active immunization of rabbits.

The competence of human GP2 to bind human PAb requires the presence of its GPI anchor, which affects further post-translational modifications. The deletion of only one asparagine residue near the C-terminus – the proposed attachment site for the GPI anchor35,36 – leads to the complete loss of its ability to bind human PAb. The importance of a correct glycosylation is also demonstrated by the gradually decreased competence of recombinant GP2 to recognize PAb in the antigen variants, in which N-glycans are stepwise reduced. In fact, the removal of all putative N-glycosylation sites results in the total elimination of PAb binding.

The few monoclonal antibodies against CUZD1 and GP2, generated in this study, recognized only conformational, but not linear epitopes. The corresponding epitopes were not identical to those binding human PAb, thus they could be used as a solid phase in two separate antigen capture ELISAs for the determination of PAb. Moreover, double staining experiments applying monoclonal antibodies against CUZD1 and GP2 in parallel demonstrated that both proteins reside in the pancreas secretory storage granules.34,37,38 It could, however, clearly be shown that both proteins are not exactly co-localized.

The autoantigens CUZD1 and GP2 could be expressed recombinantly in a human cell line. The fact that each of the recombinant cell substrates reacted with one of the PAb types – the CUZD1 cells with antibodies that produce the reticulogranular pattern and the GP2 cells with antibodies causing the droplet pattern – can be considered as a proof of correct identification. The use of both recombinant substrates in parallel allows a reliable detection of Crohn's disease-associated pancreatic autoantibodies by IIFT. This makes human tissue dispensible. Moreover, the number of positive PAb results increased from 30.2% to 35.4%, compared to pancreas tissue sections. Finally, it is much easier to distinguish between positive and negative results when recombinant cells are used.

The information on CUZD1 is sparse: it is highly expressed in exocrine pancreas, it has been reported as a marker of ovarian cancer39,40 and it belongs to the same protein family as DMBT1/gp34041 that is involved in innate immunity. The better studied GP2 has been considered as a defensive tool against microbial invasion because of its capability of binding to bacterial type I-fimbriae,42,43 similar to its kidney homologue, Tamm–Horsfall protein,35,44 and because of its presence at the outer membranes of M cells located in follicle-associated epithelium of healthy intestinal mucosae.45 Based on the available data, we hypothesize that – upon the release of considerable amounts of soluble CUZD1 and GP2 from pancreas in concert with its digestive enzymes – both proteins might serve as parts of the innate immunity in the intestinal lumen. Both proteins share a C-terminal zona pellucida domain that is found in proteins that polymerize under specific environmental conditions34,46 and might therefore aggregate bacteria and prevent their adhesion to mucosal cells. The generation of autoantibodies against them might either interfere with this process or be provoked as a result of specific microbial challenge, e.g. AIEC,47,48 to the intestinal barrier. This latter hypothesis is supported by the finding that autoantibodies against Tamm–Horsfall protein appear after urogenital infection49 and that their generation seems to be an epiphenomenon driven through a pathway of the innate immune system in mice.50

In conclusion, knowledge of the target antigens for PAb – CUZD1 and GP2 – both of which belong to protein families supposedly involved in innate immunity might help to unravel the mechanisms leading to autoimmunity in Crohn's disease.


We thank Beatrice Schneider, Stephanie Niemann and Antje Friedrich for their excellent technical assistance and Sandra Saschenbrecker and Stephan Tiede for their critical review of the manuscript.


View Abstract