Human Pancreatic Duct Cells Exert Tissue Factor-Dependent Procoagulant Activity

RESEARCH DESIGN AND METHODS

Cell preparations.

Human pancreatic duct cells were isolated from pancreata obtained from adult heart-beating organ donors as previously described (9). The organs were procured by European hospitals affiliated with the Eurotransplant Foundation (Leiden, the Netherlands) and sent to the Human β-Cell Bank in Brussels (Belgium) for the preparation of isolated fractions. Collagenase digests were separated by Ficoll gradient purification into an islet fraction and an exocrine fraction. The exocrine fraction was cultured in serum-free medium for at least 4 days, leading to a preparation enriched in duct cells. As recently reported (10), the purity of these duct cell preparations was routinely >90–95% as assessed by cytokeratin 19 (CK19) staining. The culture medium consisted of HAM’s F10 with 6 mmol/l glucose, supplemented with 0.5% BSA, 0.1 mg/ml streptomycin, 125 units/ml penicillin, and 2 mmol/l l-glutamine. Thereafter, cells were either maintained in a serum-free medium (cultures in suspension) or dissociated in a calcium-free medium and cultured with 10% FCS to obtain monolayers. Human adenocarcinoma pancreatic duct cell line (CAPAN-2) was cultured as previously described (11). In one set of experiments, β-cells and duct cells were directly isolated using fluorescence-activated cell sorting, as previously described (9). As control, peripheral blood mononuclear cells (PBMCs) from healthy volunteers were activated during 6 h with 1 μg/ml lipopolysaccharide from E. coli (serotype 0128:B12; Sigma Chemicals, Bornem, Belgium).

Procoagulant activity assay.

Procoagulant activity (PCA) was determined in duplicates by a single-stage clotting assay on culture supernatants or on cell extracts realized in PBS by repetitive freezing. Each sample (100 μl) was incubated at 37°C for 1 min with 100 μl of normal citrated plasma before the initiation of clotting by the addition of 100 μl of 25 mmol/l CaCl₂. Clotting time was recorded with a KC10 apparatus (Amelung, Lemgo, Germany), and PCA (in milliunits per milliliter) was determined by reference to a standard curve generated with a commercial rabbit thromboplastin (BioMérieux, Marcy l’Etoile, France). The amount of thromboplastin that yielded a clotting time of 12.4 s was assigned a value of 1 unit. To determine the role of the TF/factor VII pathway in the PCA, experiments were performed with factor VII-deficient plasma (Dade Behring, Anderlecht, Belgium). The number of cells in the extracts was determined by counting or DNA measurement.

Flow cytometry analysis.

Cells were washed in PBS supplemented with 1% BSA and 10% pooled human serum and incubated for 20 min at 4°C with the fluorescein isothiocyanate-conjugated IgG1 monoclonal antibody (mAb) against TF no. 4508CJ (American Diagnostica, Andresy, France) or the corresponding isotype-matched control mAb (Becton Dickinson, Erembodegem, Belgium). Cell fluorescence was analyzed using a FACScan flow cytometer (Becton Dickinson).

Immunohistochemistry studies.

Immunohistochemical analysis was performed on formalin-fixed, paraffin-embedded human pancreatic specimens. Five-micron-thick sections were cut onto coated slides and deparaffined by routine techniques. After antigen retrieval in EDTA and microwave pretreatment, the sections were incubated with a mouse anti-TF mAb (IgG1 n4509; American Diagnostica) or a mouse anti-CK19 mAb (clone RCK108; Dako, Merelbeke, Belgium) at 1/100 dilution. Negative controls were obtained by omitting the primary antibody. Labeling was revealed by the streptavidin-biotin-peroxidase complex (Dako).

RT-PCR.

mRNA was extracted from 0.2 × 10⁶ duct cells or 0.5 × 10⁶ PBMCs using the MagNA Pure mRNA extraction kit on the MagNA Pure instrument (Roche Applied Science, Brussels, Belgium) following the manufacturer’s instructions. One-step RT-PCR was performed on a Lightcycler instrument (Roche Applied Science) with primers synthesized at Biosource Europe (Nivelles, Belgium). RT-PCR for TF or β-actin was realized with the following primers: TF: sense primer, 5′-TGAATGTGACCGTAGAAGATGA-3′; antisense primer, 5′-GGAGTTCTCCTTCCAGCTCT-3′; and β-actin: sense primer, 5′-GGTCAGAAGGATTCCTATG-3′; antisense primer, 5′-GGTCTCAAACATGATCTGGG-3′. Products were separated by electrophoresis on 2% agarose and visualized with ethidium bromide. We also developed a real-time RT-PCR method using the following primers and probes: TF sense primer, 5′-GGGAATTCAGAGAAATATTCTACATCA-3′, TF antisense primer, 5′-TAGCCAGGATGATGACAAGGA-3′; alternatively spliced TF (asTF) sense primer, 5′-TCTTCAAGTTCAGGAAAGAAATATTCT-3′; asTF antisense primer, 5′-CCAGGATGATGACAAGGATGA-3′; and probe, 5′-TGGAGCTGTGGTATTTGTGGTCA-3′.

Tubing loop model.

A whole-blood experiment protocol was adapted from a model previously described (5). Loops made of polyvinylchloride tubing (inner diameter 6.3 mm and length 390 mm) and treated with a Corline heparin surface were purchased from Corline (Uppsala, Sweden). Loops were supplemented with cell samples in 150 μl PBS or with PBS alone before blood addition. When indicated, samples were preincubated at room temperature for 10 min with a neutralizing mAb against TF (IgG1 no. 4509) or PBS and added to the loops after two washing steps. Nonanticoagulated blood (5 ml) from healthy volunteers was added to each loop. To generate a blood flow of about 45 ml/min, loop devices were placed on a platform rocker inside a 37°C incubator. Blood samples were collected into EDTA (4.1 mmol/l final concentration) and citrate (12.9 mmol/l final concentration) tubes before and 30 min after perfusion began. Platelets were counted using a CellDyn 4000 automate (Abbott Laboratories, Abbott Park, IL), and fibrinogen was determined using a Behring coagulation system (Dade Behring).

RESULTS

PCA of pancreatic cell preparations in human plasma.

In a preliminary set of experiments, we compared the PCA in human plasma of several cell preparations obtained from human pancreata. As shown in Table 1, we first found that both the endocrine and exocrine fractions induced significant PCA. When the exocrine fraction was cultured in conditions resulting in duct cell enrichment (>95%), PCA increased to reach values three- to sixfold above that of the endocrine fraction. The PCA activity of the enriched duct cell preparations was factor VII-dependent since it was not observed in factor VII-depleted plasma. In accordance with these findings obtained on primary duct cells, we observed that the CAPAN-2 pancreatic duct cell line exerted a factor VII-dependent PCA that was also present in the culture supernatants of this cell line.

TF expression by duct cells.

The factor VII dependence of the duct cell-associated PCA strongly suggested the involvement of TF. TF expression by primary duct cells was first documented by flow cytometry (Fig. 1A). Additional immunohistochemical studies were conducted to confirm the constitutive expression of TF on duct cells in vivo. First, we confirmed a positive staining of islets for TF (data not shown). As shown in Fig. 1B, a clear staining for TF was observed on cells that express CK19, a marker specific for epithelial cells. The staining for TF was cytoplasmic and concentrated at the apical side.

Using RT-PCR, we confirmed that both the CAPAN-2 duct cell line and primary duct cells express TF mRNA (Fig. 2A). Interestingly, both the classical form and the recently described alternatively spliced variant (12) of TF mRNA were detected in the duct cell extracts. For control, we documented the upregulation of TF mRNA expression in PBMCs upon activation by bacterial lipopolysaccharide. In this cell type, the classical form of TF mRNA was clearly induced, whereas asTF was barely detected. In additional experiments, we used a real-time RT-PCR method to quantify TF and asTF mRNA expression in β-cells and duct cells directly purified by using flow-activated cell sorting. As shown in Fig. 2B (which is representative of three independent experiments), the expression of both TF and asTF mRNA was higher in duct cells (87% pure) than in β-cells (74% pure) obtained from the same donor.

PCA of primary duct cells in whole blood.

To investigate the PCA of duct cells in a model closer to the in vivo situation, we adapted the tubing loop system that was used to demonstrate the thrombotic reaction induced by islet preparations (5). This model is based on the injection of nonanticoagulated blood in plastic loops in which the inner surface is coated with heparin to prevent contact-dependent blood coagulation, thus allowing the investigation of coagulation in the presence of platelets. Blood coagulation was assessed by macroscopic examination for the presence of clots and by monitoring platelet counts and fibrinogen levels, which drop as a consequence of coagulation activation. As shown in Table 2, we found that as few as 10³ duct cells added to 5 ml of blood were sufficient to induce clot formation within 30 min. This was associated with a dramatic drop in platelet counts and a complete consumption of fibrinogen. In three independent experiments, pretreatment of duct cells with an anti-TF mAb prevented the formation of visible clots and reduced the drop in platelet counts and fibrinogen levels.

DISCUSSION

The demonstration that the CAPAN-2 duct cell line expresses TF is in line with previous studies (13) showing the presence of TF in pancreatic ductal adenocarcinoma. As far as the normal pancreas is concerned, TF expression was previously documented (14) during embryogenesis, but not in adult tissue. Development of more sensitive immunostaining techniques using different anti-TF antibodies probably explains the recent demonstration (5) of TF expression in human islets. In our study, we confirmed this observation and also provided evidence for TF expression by duct cells in normal pancreas (Fig. 2). The latter finding is consistent with the expression of TF by normal epithelial cells in breast and kidney (15,16).

Interestingly, duct cells were shown to express the classical TF mRNA, encoding the membrane form of TF as well as asTF, which encodes the molecule-soluble form (12). It has been suggested that asTF could be involved in the propagation of the coagulation in blood and lead to the formation of macroscopic thrombi (12). Expression of asTF by duct cells might thus contribute, possibly along with the release of TF in microparticles (17), to the PCA present in the supernatants of duct cell cultures.

The relevance of our observations to islet transplantation is suggested by experimental observations (18) in rats showing the formation of thrombi shortly after transplantation, particularly around the nonendocrine tissue that contaminates islet preparations. Furthermore, there is also clinical evidence (6) that insufficient islet purification can promote development of portal vein thrombosis. In allogeneic as well as autologous islet transplants, the regular use of heparin might contribute to limit these thrombotic complications. Activation of coagulation by TF could not only result in thrombotic events but also elicit an inflammatory reaction involving upregulation of adhesion molecule expression and chemokine production (19,20). In the context of islet transplantation, the so-called instant blood-mediated inflammatory reaction likely contributes to early β-cell loss (5). It could also enhance graft immunogenicity, thereby promoting transplant rejection (21). Our findings demonstrate that contaminating duct cells can contribute to this post-islet transplant blood-mediated inflammatory reaction originally described (5) as a consequence of TF expression by endocrine cells.