Islet Surface Heparinization Prevents the Instant Blood-Mediated Inflammatory Reaction in Islet Transplantation

Protocols for the isolation of human islets from cadaver donors and for all animal experiments were approved by the regional research ethics committees of Uppsala University, and experiments were performed in accordance with local institutional and Swedish national rules and regulations.

Human pancreata were obtained from normoglycemic cadaveric donors after appropriate consent for multiorgan donation. Islets of Langerhans were isolated using a modification of the previously described semiautomated digestion-filtration method (15), followed by purification on a continuous density gradient in a refrigerated cell processor (COBE 2991; COBE Blood Component Technology, Lakewood, CO).

The islet preparations were of good quality and were designated for experimental use because the total islet volume was too low for use in clinical transplantation. Islet purity was determined by staining with diphenylthiocarbazone. Islet preparations were 70–90% pure.

After the isolation and purification, the islet preparations were placed in untreated single transfer packs for platelets 1300, the maximum volume of the bag (Baxter Medical, Sweden), and kept at 37°C in an atmosphere of 5% CO₂ in humidified air in CMRL 1066 culture medium (Mediatech, Herndon, VA) supplemented with 10 nmol/l nicotinamide (Sigma Aldrich, Schnelldorf, Germany), 10 mmol/l HEPES buffer (Invitrogen, Paisley, Scotland), 0.25 μg/ml fungizone (Invitrogen), 50 μg/ml gentamicin (Invitrogen), 2 mmol/l l-glutamine (Invitrogen), 10 μg/ml Ciprofloxacin (Bayer, Leverkusen, Germany), and 10% (vol/vol) heat-inactivated human serum.

Porcine islets were isolated and purified at the Third Medical Department, Justus-Liebig-University, Giessen, according to published protocols (16). Isolation of mouse islets was performed using a collagenase digestion method as previously described (17).

Human, porcine, and mouse islets were biotinylated by incubating the islets for 1 h at 37°C in culture medium containing SNL biotin (EZ-Link Sulfo-NHS-LC-Biotin), tissue factor-P (EZ-Link tetraphenyl ester), or tissue factor-P biotin (EZ-Link tetrephenyl ester-PEO-Biotin) (Pierce Biotechnology, Rockford, IL) at 1 mg/ml. The islets were then rinsed by changing the culture medium three times. In the next step, the islets were incubated for 30 min at 37°C in culture medium supplemented with 1 mg/ml avidin (Pierce Biotechnology) and again rinsed by changing the culture medium three times. Finally, 1 mg/ml macromolecular conjugates of heparin, ∼70 heparin molecules covalently linked to an inert carrier chain (Corline Systems, Uppsala, Sweden), in culture medium were allowed to bind to the biotin/avidin coating for 60 min at 37°C. The islets were finally rinsed by changing the culture medium three times.

Although SNL biotin and tissue factor-P biotin gave similar results, a direct comparison showed slightly better results with SNL biotin, and, hence, this substance was chosen for experiments in the allogeneic porcine and syngeneic mice islet transplantation model.

The endothelial cell surfaces of human pancreatic arteries were modified by the same method (using SNL biotin). Each artery segment was cut in half; one-half was used for heparinization and one-half as a control. The artery segments, 3–4 cm in length, were connected at both ends to 20-cm–long pieces of polyethylene tubing (4 mm inner diameter). Ringer's acetate buffer (5 ml) was transferred into the segments and rocked back and forth over the endothelial lining several times for 1 min. The artery segments were then coated as described above, using the same regents, washes, and incubation times as described for the islets. The control artery was filled with 5 ml Ringer's solution and submerged in a Petri dish in the same solution during the heparinization process. Immediately after the immobilization process, both samples were snap frozen in liquid nitrogen and stored for later analysis.

The degree of heparinization of the islets and of cryosections of heparinized pancreatic arteries was visualized by confocal microscopy after incubating the tissues at room temperature for 15 min with antithrombin (1 unit/ml) labeled with Alexa Fluor 488. Propidium iodide (Sigma, St. Louis, MO) was used to stain nonviable cells, and Hoechst 33342 (Sigma) was used to detect cell nuclei in heparinized islets. Images were acquired with a confocal microscope (Zeiss LSM 510 Meta; Carl Zeiss, Jena, Gemany) equipped with an Axiovert 200 microscope stand. Z-stacks of the islet surfaces were acquired using the 488-nm laser line, a 20× objective, and a 505–550 bit plane filter. Three-dimensional projections of the acquired z-stacks were analyzed using Imaris software (Bitplane, Zurich, Switzerland). Images were acquired at 20× magnification. Because of the high level of autofluorescence emission within the spectrum of Alexa Fluor 488, autofluorescence was determined in lambda mode, and images were then acquired in fingerprinting mode using the appropriate spectra.

Insulin secretion in response to glucose stimulation in a dynamic perfusion system was assessed on day 1 (n = 3). Islets were initially perfused with 1.67 mmol/l glucose, then with 16.7 mmol/l, and again with 1.67 mmol/l. During the 120-min perfusion, fractions were collected at 6-min intervals. The concentration of insulin was analyzed using a commercial enzyme-linked immunosorbent assay kit (Mercodia, Uppsala, Sweden). Values are given as picomoles per liter.

All mice (25–30 g) were made diabetic by intravenous administration of alloxan (75 mg/kg body wt) as elsewhere described (18). Diabetes was confirmed by the presence of hyperglycemia (>25 mmol/l blood glucose for 2 consecutive days). Four to six days after the induction of diabetes, either 300 heparinized (using SNL biotin; n = 7) or 300 untreated control (n = 7) islets (in this model, 300 islets is the islet mass required for reversal of diabetes in mice receiving control islets) were transplanted under the left kidney capsule into B6 mice (Bomholdtgaard, Ry, Denmark) on a C57BL/6J background, as previously described (19). After transplantation, blood glucose was measured every third day for 4 weeks. At 5 weeks after the transplantation, all mice underwent an intraperitoneal glucose tolerance test by injection of 250 μl of a 30% glucose solution. Blood glucose concentrations were determined before the glucose administration (time point 0) and 15, 30, 60, and 120 min after the injection. At the end of the experiment, to ensure persistent absence of endogenous pancreatic insulin production, the graft-bearing kidney was surgically removed from all animals. Reversal of diabetes was defined as blood glucose >25 mmol/l in two consecutive measurements. After the nephrectomy, all grafts were retrieved, paraffin embedded, and evaluated by immunohistochemical staining for morphology and the presence of insulin.

The capacity of surface heparinized and control islets to induce IBMIR was evaluated using a modification of an in vitro loop model as previously described (4,13): 2 μl either heparinized (using tissue factor P or SNL biotin) or untreated control islets were resuspended in 100 μl CMRL 1066. Fresh ABO-compatible human blood (7 ml) was added to each loop before perfusion of the islets. In similar experiments, blood loops with heparinized islets (using SNL biotin) were compared with untreated islets, with and without the addition of the specific thrombin inhibitor Melagatran (4 μmol/l), which has previously been shown to be an efficient inhibitor of the IBMIR (10). In both experiments, control loops with 100 μl CMRL 1006 alone were included. To generate blood flow, the loops were placed on a rocking device inside a 37°C incubator for 60 min. To monitor the IBMIR, blood samples of 1 ml were collected in EDTA (4.2 nmol/l, final concentration) before perfusion and at 5, 15, 30, and 60 min after the start of the perfusion.

Platelet counts were assessed using a Coulter AcT diff analyzer (Beckman Coulter, Miami, FL). Plasma concentrations of TAT complexes were quantified using a commercially available enzyme-linked immunosorbent assay kit (Enzygnost TAT; Dade Behring, Marburg, Germany). The complement activation product C3a was quantified as previously described (20).

Five pairs of piglets of Swedish mixed-country breed, weighing 11–16 kg, were anesthetized using xylazine (2.2 mg/kg), tiletamin/zolazepam (6 mg/kg), and atropine (0.04 mg/kg) as a premedication given intramuscularly in the neck and then were given fetanyl (15–20 μg/kg) at the onset of anesthesia. They were orally intubated using a cuffed endotracheal tube and ventilated with mechanical ventilation in volume-controlled mode. A continuous infusion of ketamine (30 mg · kg⁻¹ · h⁻¹), fetanyl (5 μg · kg⁻¹ · h⁻¹), and midazolam (0.5 mg · kg⁻¹ · h⁻¹) was given. Pancuronium (0.3 mg/kg) was given intravenously. No systemic anticoagulant was administrated. A catheter was placed in the superior mesenteric vein. Heparinized (using SNL biotin) or untreated control adult porcine islets (7,500 islet equilvalent/kg body wt) isolated from the same donor animal were infused intraportally, using a bag system, over 5 min into each pair of pigs. Blood samples were collected into EDTA tubes before and 15, 30, and 60 min after islet transplantation. After 60 min, the pigs were killed with an overdose of KCl, and the livers were excised and the portal system exposed. Blood clots and liver biopsies were taken and fixed in paraformaldehyde or snap frozen in liquid nitrogen.

Transplanted grafts, liver biopsies, and clots retrieved from portal vessels were fixed in paraformaldehyde (Histofix; Histolab Products, Västra Frölunda, Sweden), embedded in paraffin, sectioned on a Leica Jung RM 2055 microtome at 4 μm, and air-dried at 37°C overnight. The sections were kept at room temperature until dewaxing. They were then fixed and stained for insulin using Rabbit Envision (DAKO, Copenhagen, Denmark).

All results are presented as means ± SEM. Prism, version 4.0a, for Macintosh was used for statistical calculations. The in vitro loop studies were evaluated using Friedman's ANOVA. Significance was determined at P < 0.05. For evaluation of the in vivo allogeneic porcine islet transplantation experiments, the Mann-Whitney nonparametric test (double tailed) was used.

Confocal microscopy of the heparinized islets consistently revealed evenly distributed fluorescence, demonstrating that the heparin coat covered the complete islet surface (Fig. 1B). Cross-sections of the islets showed that the heparin bound strictly to the surface of the islets, as reflected by the binding of Alexa Fluor 488–labeled antithrombin. This analysis confirmed that the heparinized islets were surrounded by a coherent and smooth coating of heparin, which was strictly confined to the surface of the islet. The coating was present on cultured human islets for at least 72 h but was not detectable on transplanted mouse islets after 4–5 weeks under the kidney capsule.

Islets treated with heparin conjugates alone, with heparin conjugates after biotinylation (excluding the avidin step) or with heparin conjugates after treatment of the islets with avidin (excluding the biotin step), demonstrated no binding of Alexa Fluor–labeled antithrombin (data not shown).

Sections of heparinized vessel walls treated with Alexa Fluor 488–labeled antithrombin demonstrated a distinct fluorescence following the endothelial cell lining, with sparse distribution into the subendothelium (Fig. 1C, left panel). No fluorescence was detected below the external elastic membrane or in the muscularis. Untreated control vessel wall showed no binding of Alexa Fluor 488–labeled antithrombin (Fig. 1C, right panel).

The macroscopic appearance of the islets after heparinization did not differ from that of untreated control islets, and there was no sign of islet loss, even after several days of culture. However, the heparinized islets preserved their integrity over time, with no tendency to adhere to plastic surfaces or to other islets.

Human and porcine islets were stained with propidium iodide and Hoechst 33342 and then analyzed by confocal microscopy. These analyses indicated that >90% of the cells were viable, with no difference seen between heparinized and control islets (Fig. 1B). Heparinized and untreated control islets from three islet preparations responded similarly to glucose stimulation, demonstrating that the normal biphasic insulin secretion was not affected by the heparinization procedure (Fig. 2A).

In a syngeneic transplantation model, all mice that received either heparinized (n = 7) or untreated control (n = 7) islets had normalized blood glucose levels within 10 days after transplantation and subsequently responded equally well to an intraperitoneal glucose tolerance test (Fig. 2B and C, respectively). After removal of the graft-bearing kidney, all 14 mice immediately became hyperglycemic and remained so for the 4 days until they were killed, demonstrating that regeneration of islets in the mouse pancreas had not occurred. Positive insulin staining and intact morphology indicated successful engraftment and survival of both heparin and untreated control islets (Fig. 2D).

Long-term graft survival was tested by transplanting heparinized (using SNL biotin) and untreated human islets under the kidney capsule in B6 athymic nude (nu/nu) mice. After 4 weeks, the grafts were retrieved and paraffin sections evaluated for morphology and for the presence of insulin by immunohistochemical staining. The results demonstrated successful engraftment and survival of both heparin and untreated control islets (data not shown).

Clotting was visible after a 60-min incubation of untreated islets, but not heparinized islets, in fresh nonanticoagulated human blood (Fig. 3A).

In the control tubing loops containing culture medium alone, only a slight drop in platelet count was observed, while nearly all of the platelets were consumed in the tubing loops containing untreated islets. The untreated islets also triggered a rise in the coagulation parameter thrombin antithrombin and an increase in C3a (Fig. 3B–D). In loops containing biotin (SNL)-heparinized islets, there was no macroscopic clotting, and the drop in platelet count and generation of TAT and C3a were markedly attenuated. Similar results were obtained with loops containing biotin (tissue factor P)-heparinized islets. In control experiments, islets treated with biotin and heparin conjugate alone (excluding the avidin step) were evaluated in the in vitro loop model. The results showed no difference between untreated control islets and treated islets, indicating that heparin binding is essential for the protective effect (data not shown). The heparin coating was found to be as efficient as 4 μmol/l melagatran (Fig. 3E).

Five pairs of pigs were transplanted intraportally with heparinized or untreated adult porcine islets (Fig. 4). After 60 min, visible intraluminal macroscopic clotting was found in the portal vein branches of the pigs receiving untreated control islets (n = 5) (Fig. 4A), whereas thrombi were scarce in those receiving heparinized islets (n = 5) (Fig. 4B). We also observed a reduction in the degree of macroscopically visible infarcted areas.

In all pigs (n = 5) that received untreated control islets, there was a rise in TAT during the first 15-min period after infusion of the islets. In contrast, the increase in TAT was significantly attenuated in pigs that were given heparinized islets (P < 0.05), indicating that the IBMIR was reduced (Fig. 4C). Significantly increased insulin release (an indicator of cell damage) and generation of C3a were observed in a few individual animals that received untreated control islets. In the corresponding control animals that received heparinized islets, the levels of insulin and C3a was considerably lower (Fig. 4D and E).

The consistent appearance of the IBMIR in clinical islet transplantation (3,4) calls for strategies to prevent or decrease the IBMIR to minimize islet loss in the immediate posttransplantation period. We have previously shown that soluble heparin, which is regularly used during clinical islet transplantation, only marginally counteracts the IBMIR in our in vitro loop model (13). Likewise, in clinical transplantation performed with systemic heparinization, a swift increase in coagulation activation followed by a marked release of C-peptide is observed (3,4) Surface heparinization is an attractive alternative to soluble heparin. It provides a means to render the islet surface biocompatible when exposed to blood, thereby mimicking the protective characteristics conferred by heparan sulfate on the endothelial cell lining of the vascular wall. In addition to its effects on the cascade systems and the cells of the blood, the heparin coating reduces exposure of collagen and other extracellular matrix proteins on the islets that may be prothrombotic and trigger inflammation.

In the present study, we have demonstrated both in vitro and in vivo that modification of pancreatic islets with surface-attached heparin can reduce the deleterious IBMIR associated with islet transplantation. Using a pig model to study the immediate time period after allogeneic intraportal islet transplantation, we were able to confirm the reduced IBMIR that we had previously demonstrated in the in vitro loop model. It is reasonable to assume that improved long-term islet graft function will be obtained if the IBMIR is attenuated and more islets survive (9). To be able to show an analogous effect in immunosuppressed and diabetic large animals is, however, a major undertaking that is beyond the scope of the present study.

We have used a water-soluble macromolecular conjugate in which ∼70 heparin molecules are covalently attached to a carrier carbon backbone (21) that is bound to the islet surface via biotin and avidin. Unlike previously published heparinization methods that involve multiple steps, the use of biotin and prefabricated heparin complexes reduce the handling time and the risk of damage to the islets. The present heparinization protocol is based on the fact that avidin expresses binding sites for both biotin and heparin (22), thus bolstering the multiattachment binding effect achieved by using a conjugate composed of multiple heparin molecules.

Biotin and streptavidin have previously been used to conjugate biologically active molecules to cells. In a heart transplantation model in mice, Askenasy et al. (23) successfully displayed exogenous streptavidin-conjugated proteins on the vascular endothelium of solid organ grafts and splenocytes, without any notable signs of toxicity being induced by the procedure. Similarly, no toxicity was associated with the heparin coating procedure in the present study: i.e., surface heparinization did not affect insulin release in response to a glucose challenge, and the long-term survival and function of the heparinized islets were unaffected when compared with the untreated islets transplanted into diabetic mice.

The heparinization approach is distinctly different from coatings such as microencapsulation that create a barrier for both molecules and cells. The conjugated heparin does not hinder cell migration, and therefore does not prevent revascularization, nor does it create a dead space that alters the diffusion properties and impairs the dynamics of glucose-stimulated insulin release.

Recently, our group has successfully induced a marked expression and secretion of the leech thrombin inhibitor hirudin using an adenoviral vector, indicating that human islets can successfully be modified by means of gene therapy to express molecules that can prevent the IBMIR (24). A number of proteins, such as APC (25) and CD39 (26), are also possible candidates for use. APC significantly reduced loss of functional islet mass after intraportal transplantation in diabetic mice, and islets from transgenic mice that express CD39 fail to induce coagulation in roughly one-half of the experiments when exposed to human blood. Gene therapy approaches are in many ways appealing, but they suffer from the fact that all existing procedures that introduce new DNA into the islets are associated with a risk of inducing inflammatory or even adaptive immune responses (27,28). To what extent the additional protein synthesis may affect the islets’ ability to produce insulin is currently under debate and may vary depending on the transduction protocol (24,29).

Tissue factor expression is accompanied by an array of proinflammatory events that affect the islets, such as the expression of MCP-1 (6,30), interluekin-8, and macrophage migration inhibitory factor (7,31). These events are linked to ischemia after organ procurement and upregulation of inflammatory genes occurring as a result of brain death of the donor (25,32). The heparin coating protects the affected cells from the innate immune response of the blood. Our results with successful heparin coating of artery vessel walls suggest that this heparin coating strategy may prove effective not only in reducing the adverse effects of the IBMIR in clinical islet transplantation but also in other types of cell, tissue, or organ transplantation and ischemia-reperfusion injury, without exposing the recipient to the side effects associated with systemic anticoagulant treatment (33).

In current protocols for clinical islet transplantation, a single dose of ∼3,000–5,000 IU is used. Systemic administration of anticoagulants will always be associated with an increased risk of bleeding, a consideration of particular importance in islet transplantation, since the procedure often involves transhepatic puncture of a large portal vein. Despite the use of heparin, portal thrombosis is a feared complication that occurs in ∼10% of transplantations (34). Covering the islet surface with heparin offers an attractive alternative with the capability of abrogating intraportal thrombosis and the IBMIR. This protection is achieved without systemic side effects, since the total amount of heparin attached to a full-size clinical islet transplant is as low as 40 IU. The technique is readily transferable to the clinical situation and holds promise for improving both the safety and outcome of clinical islet transplantation. The heparin coating may also be beneficial during the subsequent engraftment. Heparin has an affinity for a number of plasma proteins and growth factors (35), such as vascular endothelial growth factor (36) and fibroblast growth factor (37), which may facilitate revascularization and reinnervation.

This study was supported by grants from the National Institutes of Health, the Juvenile Diabetes Foundation International, the Swedish Research Council (16P-13568, 16X-12219, 16X-5674, and 72X-06817–19), the Åke Wiberg Foundation, the Nordic Insulin Fund, the Novo Nordisk Foundation, the Torsten and Ragnar Söderbergs Foundation, the Ernfors Family Fund, Barn Diabetes Fonden, the Swedish Diabetes Association, the Clas Groschinsky Foundation, the Marianne and Marcus Wallenberg Foundation, the Knut and Alice Wallenberg Foundation, and the Crafoord Foundation.

We thank Margareta Engkvist and Graciela Elgue for their excellent technical assistance and Dr. Deborah McClellan for editing the text.