Human Pancreatic Islets Produce and Secrete MCP-1/CCL2: Relevance in Human Islet Transplantation

Pancreata were obtained from heart-beating cadaveric multiorgan donors through the North Italian Transplant Organization. Islets were isolated according to a modification of the automated method (43,44). Samples of the pancreas and samples of isolated islets were taken for histological examination and for in vitro study. The purification was assayed by computerized morphometric method (Leica Imaging System LDD, Cambridge, U.K.).

The purified islets were cultured in a sterile flask containing 25 ml of M199 medium (Seromed Biochrom, Berlin, Germany) supplemented with 10% fetal calf serum, 1% l-glutamine, 100 units/ml penicillin, and 100 μg/ml streptomycin (complete medium). The islets were incubated at 30°C in 5% CO₂ and 95% humidified air. Islet viability was assessed by trypan blue and by the evaluation of the basal cytosolic calcium concentrations ([Ca²⁺]_i) in fura-2– loaded islets by digital microscopy. Fluorescence images were collected at a middle plane of the islets by an intensified charged-coupled device camera and fed into a digital image processor, where video frames were digitized and integrated (typically 10 consecutive frames) in real time (45). Measurements were done in at least 30 individual islets from each preparation. Images were then processed to convert fluorescence data in [Ca²⁺]_i images (340/380 nm excitation wavelength ratio method). In five islet preparations, the assessment of islet viability, with double staining for propidium iodide and acridine orange, also was applied. Criteria for islet viability were the following: trypan blue exclusion <10% of cells; [Ca²⁺]_i as 340/380 <0.7 that means [Ca²⁺]_i >350 nmol/l; propidium iodide <10%, acridine orange >90%. Islets were initially tested for sterility and endotoxin (<0.05 EU/ml in all tested preparations; Chromogenic LAL test; Bio Whittaker, Walkersville, MD) and mycoplasm (Mycoplasma detection kit; Boehringer Mannheim, Indianapolis, IN) content.

Aliquots of pancreatic islet preparations (500 islets/ml) were cultured in complete medium in a 24-well plate. Culture supernatants were harvested at different time points and stored at −80°C. To investigate regulation of MCP-1 secretion by inflammatory cytokines, we added IL-1β (Pepro Tech EC, London, U.K.; specific activity 1 × 10⁷ units/mg), TNF-α (specific activity >2 × 10⁷ units/mg; Pepro Tech EC), INF-γ, and LPS (Sigma, St. Louis, MO) at different concentrations and times. Human MCP-1, TNF-α, IL-1β, and IL-8 were detected using a sandwich enzyme-linked immunosorbent assay (ELISA) as described (46,47). ELISA for MCP-1 is specific for human MCP-1 and does not detect the mouse equivalent or the closely related human chemokines MCP-2 and MCP-3 (47).

Immunohistochemistry for MCP-1 was performed on frozen and paraffin-embedded 5-μm-thick sections obtained from surgical pancreatic specimens and from isolated pancreatic islets. Serial sections were immunostained for MCP-1 (clones 5D3, 2A1, 9G6, 3A3 1 μg/ml; the clones recognize different epitopes as shown by competitive binding with peptide aa 43–65 and aa 63–85 of MCP-1) and insulin (clone E2E3, from Signet, prediluted) to localize Langerhans islets better, and CD45 (clone T29/33 from Dako, dilution 1:200) to exclude potential synthesis of MCP-1 by lymphocytes by using an avidin-biotin-complex peroxidase method (Vector), using diaminobenzidine (DAB) as chromogen. In situ hybridization (ISH) detection of MCP-1-mRNA was performed on sections obtained from surgical pancreatic specimens previously embedded in OCT compound and snap-frozen in liquid nitrogen. Briefly, 5-μm cryostat sections were fixed for 20 min in 4% paraformaldehyde, washed in buffer, dehydrated in alcohol, and then hybridized with a cocktail of biotin-labeled MCP-1 probe (R&D Systems, Minneapolis, MN) at a concentration of 2 ng/μl in hybridization solution at 37°C for 3 h. After stringent washes, the sections were incubated in an avidin-biotin-peroxidase complex (Vector) for 30 min at room temperature. The labeling was revealed by DAB. Negative controls were performed by omission of the probe or by using an indifferent biotinylated probe.

Total cellular RNA was prepared from hand-picked human pancreatic cultured islets. Total RNA was extracted by the guanidinium thiocyanate method, blotted, and hybridized as previously described (48). MCP-1 and CCR2B cDNAs were obtained by PCR amplification of the reported sequences (49). RNAse protection assay (RPA) was performed using the RiboQuant Multi-Probe template sets (hCK-5), following the manufacturer’s instructions (PharMingen, San Diego, CA).

Cell migration was evaluated using a chemotaxis microchamber technique as previously described (50). Supernatants from different pancreatic islet preparations were used as such or mixed with 1 μg of anti–MCP-1 antibody (clone 24822.111; R&D Systems, Minneapolis, MN). Recombinant human MCP-1 (PeproTech, Rocky Hill, NJ) was used as reference chemoattractant. Results are expressed as the mean number of migrated cells counted in five microscope high-power fields (magnification ×1,000).

After 2 days of culture, the insulin released in response to glucose and to human MCP-1 (300 ng/ml) was assessed by static incubation with a Krebs-Ringer solution buffered with HEPES plus 0.1% bovine serum albumin. After 30 min of equilibration with 3.3 mmol/l glucose, human islets were stimulated for 30 min with 16.7 mmol/l glucose. MCP-1 (300 ng/ml) was added to both 3.3 mmol/l and 16.7 mmol/l glucose to assess the effect in both basal and stimulated conditions. All samples were frozen at −20°C, and insulin was assessed by RIA using a commercial kit (Incstar, Stillwater, MN). Data are expressed as the mean ± SD [Ca²⁺]_i changes after 300 ng/ml–1 μg/ml MCP-1 was assessed in islets loaded with fura-2 and examined by digital microscopy as described previously (45).

Preparations were considered adequate for transplantation according to the following criteria: 1) number of equivalent islets >5,000/kg body wt, 2) purity >20% (computerized morphometric determination of islet/total purified preparation; Leica Imaging System LDD), and 3) islet viability. Transplantation was performed between 24 and 72 h after isolation. The islet preparation was suspended in 100 ml of Hanks’ solution (Clinical Grade, SALF, Bergamo, Italy) containing 1,000 units of heparin and 2% human albumin. Percutaneous trans-hepatic injection (under local anesthesia) was performed according to the protocol approved by the local ethics committee. All patients were already under immunosuppression therapy with steroids and cyclosporine for a previous kidney transplant (44). After islet transplantation, the therapy was changed as follows: antilymphocyte globulin (125 mg/day for 10 days; IMTIX, Marseille, France), cyclosporine (7.5 mg · kg⁻¹ · day⁻¹), mycophenolate mofetil (2 g/day), and methylprednisone (500 mg immediately before surgery, 0.25 mg · kg⁻¹ · day⁻¹ for 2 months after surgery, then lowered to 5 mg/day). All recipients were negative for C-peptide (Dako, Cambridgeshire, U.K.), with the exception of patients 30 (C-peptide = 0.39 nmol/l, insulin requirement 34 units/day) and 32 (0.26 nmol/l, insulin requirement 27 units/day), in whom C-peptide values were not responsive to arginine test (data not shown). Before transplantation, patients 30, 32, and 35 had autoantibodies to glutamic acid decarboxylase (GAD) and tyrosine phosphatase-like protein (IA-2), measured by radiobinding assay (51). Patients 29, 34, 42 had autoantibody only to GAD; patient 41 had autoantibody only to IA-2. During the first 10 days posttransplantation, blood glucose was maintained between 4.4 and 7.0 mmol/l by continuous insulin infusion.

We investigated the capacity of human pancreatic islets to secrete MCP-1. Supernatants from 20 different primary cultures of human pancreatic islets were studied (Fig. 1A). Every islet preparation (210,000 ± 49,000 equivalent islets) remained sterile (anaerobic, aerobic, fungal, mycoplasm determination), with low endotoxin content (<0.05 EU/ml) and viable (trypan blue and fura-2 assays) for the culture period (7 days). The purification of the islet preparation was 35 ± 11% (islet/total purified preparation) with 67 ± 8% of β-cells (computerized morphometric analysis; Leica Imaging System LDD). In five different islet preparations, islet viability was confirmed also by double staining with propidium iodide (<10% positive) and with acridine orange (>90% positive). High levels of MCP-1 were detected in the culture medium. The secretion seemed to be time-dependent and with a wide range of concentration (Fig. 1A). Eight supernatants were also studied for the presence of TNF-α and IL-1. No detectable levels of TNF-α or IL-1 were present even after 7 days of culture (Fig. 1A). MCP-1 expression was analyzed by Northern blotting. Total cellular RNA was prepared from hand-picked human islets from three different preparations. High levels of MCP-1 mRNA were detected, whereas expression of mRNA for CCR2, the MCP-1 receptor, was not found (Fig. 1D). To investigate the biological activity of MCP-1 secreted by human pancreatic islets, we tested supernatants as chemoattractant for monocytes in vitro. Islet supernatants were able to attract human monocytes in a classical chemotaxis assay (Fig. 1B). Three different supernatants with different determination of MCP-1 (<0.08, 20.3, and 33.7 ng/ml) induced migration of 99 ± 8, 266 ± 4, and 351 ± 24 monocytes, respectively. The chemotactic activity of a recombinant MCP-1 used as standard was strikingly similar to that of islet-derived MCP-1. Migration was near totally inhibited by pretreatment with a blocking anti–MCP-1 antibody, indicating that MCP-1 is the major chemokine for monocytes in the supernatants of cultured pancreatic islets. We also tested the chemotactic activity of islet supernatants with polymorphonuclear cells and phytohemagglutinin (PHA)-activated T-cells and did not find significant migration. Modulation of MCP-1 production by cultured human pancreatic islets from six different preparations in the presence of inflammatory stimuli was investigated (Fig. 2A). IL-1β, TNF-α, and LPS enhanced the production of MCP-1 in a dose-dependent manner. INF-γ induced only a small decrease in the secretion, at the highest concentration used. The upregulatory effects of IL-1β and TNF-α on the production of MCP-1 were already demonstrable at 24 h and increased over time. To study whether other chemokines are secreted by human islets especially after stimulation with inflammatory stimuli, we performed an RPA from three different islet preparations (Fig. 2B). As expected, MCP-1 expression was increased after IL-1β or LPS stimulation. No mRNA for macrophage inflammatory protein (MIP)-1α, MIP-1β, interferon inducible protein 10 (IP-10), or RANTES was detected in resting condition or after exposure to IL-1β or LPS. In contrast, a positivity for IL-8 mRNA was present, which was increased after stimulation. By ELISA, low levels of IL-8 were detected in the culture supernatants of pancreatic islet (Fig. 1C). Mean values of IL-8 in 20 different preparations were 0.1 and 1.75 pg/islet at 24 h and 7 days of culture, respectively. It should be noted that levels of IL-8 were 10 times lower than MCP-1 levels. It is interesting that islet supernatants with highest MCP-1 content also showed the highest IL-8 production (Fig. 1C).

The observation that MCP-1 is produced from endocrine cells of human islets suggested a possible role for MCP-1 in modulating islet endocrine secretion. An early intracellular event in leukocytes activated by MCP-1 is a rise in [Ca²⁺]_i. Although no expression of CCR2 mRNA was detected in pancreatic islet, it was not possible to rule out the presence of another (previously unidentified) MCP-1 receptor. Therefore, it was of interest to evaluate [Ca²⁺]_i in human islet after exposure to MCP-1. In a range of concentrations from 100 ng/ml to 1 μg/ml, MCP-1 did not induce an increase in [Ca²⁺]_i in fura-2–loaded human islets (n = 5 from three different preparations; Fig. 3A). Next, we evaluated the possibility that MCP-1 might influence insulin secretion in a system of static incubation. Human islets (n = 6 from six different islet preparations) exposed for 30 min to 100 ng/ml MCP-1 did not change their insulin release at either 3 (basal condition) or 16 mmol/l (stimulated condition) of glucose (Fig. 3B). Conversely, to assess the capacity of glucose to modulate MCP-1 secretion, we studied MCP-1 release after exposure to 16.7 mmol/l glucose. As expected, insulin secretion was increased whereas MCP-1 release was not affected (Fig. 3C). These results suggest that MCP-1 production by pancreatic islets is not associated and is not modulated by the endocrine activity of islets.

Standard methodology of immunohistochemistry was used to identify MCP-1 in human islets (Fig. 4). To evaluate MCP-1 expression before enzymatic digestion, we studied whole pancreas from surgery specimen (tumor, n = 3; chronic pancreatitis, n = 3; pancreas from explanted organs, n = 8). MCP-1 was visualized within the cytoplasm of islet cells that were also positive for insulin (Figs. 4A and B). MCP-1–positive cells did not react with a specific anti-CD45 antibody (Fig. 4C), indicating that MCP-1 staining was not due to infiltrating leukocytes. A less intense positivity was also seen in ductal cells (in particular, small ducts), whereas exocrine tissue seemed to be negative. To confirm these results, we used three other anti–MCP-1 monoclonal antibodies, recognizing different epitopes of MCP-1, with identical pattern of positivity (not shown). Immunoelectron microscopy confirmed the cytoplasmic positivity for MCP-1 of endocrine cells in human islet (not shown). By ISH, MCP-1 mRNA was detected in endocrine cells of human islets (Fig. 4D). Insulinoma and glucagonoma were also analyzed, and MCP-1 immunostaining was detected in the endocrine cells, confirming their capacity to secrete MCP-1 (data not shown). Whole pancreas from surgery specimen was also studied for IL-8 expression. IL-8 was not detected by immunohistochemical analysis within the cytoplasm of islet cells or other pancreatic components (not shown).

Twenty islet preparations were transplanted into 14 patients with type 1 diabetes, who were already immunosuppressed for a previous kidney graft. Patients’ characteristics are detailed in Tables 1 and 2. To reach the minimum number of 5,000 IE/kg body wt, 6 of the 14 patients received two islet preparations, either within 10 days from the first preparation (30, 33, 34, 38, 40) or 4 months later (41). Starting from the second month after transplant, insulin requirement reduction correlates inversely with the MCP-1 secretion capacity of islets (Fig. 5A). Patients were divided into two groups according to values of MCP-1 released by the islet preparations during a 24-h culture. Cutoff value was the 50th percentile, 4.56 pg · islet⁻¹ · 24 h⁻¹. Group 1 (low MCP-1) received islet preparations below the 50th percentile, (MCP-1 mean value, 2.04 ± 0.51 pg · islet⁻¹ · 24 h⁻¹); group 2 (high MCP-1) received islet preparations above the 50th percentile (MCP-1 mean value, 34.03 ± 12.47 pg · islet⁻¹ · 24 h⁻¹). Islet preparations in the two groups of patients did not differ in viability, purity, number, and timing of culture before transplant (Table 1). Patients did not differ in terms of the following parameters: age, weight, time of diabetes, pretransplant insulin requirement, and metabolic control (Table 2). If we analyze the mean insulin requirement reduction in the two groups during the follow-up, patients in the high MCP-1 group showed a decrease in their insulin requirement that was significantly lower than that of patients in low MCP-1 group starting from the second month after the transplant (Fig. 5B). Glycosylated hemoglobin was significantly lower in the low MCP-1 group already 6 months from the transplant (5.7 ± 0.3% vs. 7.7 ± 0.8%; P < 0.05).

We evaluated islet transplants according the Islet Transplant Registry parameters (Fig. 5C) (34). Six patients in the low MCP-1 group became insulin-independent with a mean follow-up of insulin independence of 7 ± 3 months. All patients but one (31) are still insulin-free. In contrast, only two patients in the high MCP-1 group became insulin-independent for 2 ± 1 months; both patients are not still insulin-independent. One patient in the low MCP-1 group and three patients in the high MCP-1 group reduced their insulin requirement by more than 50% (partial function). Two patients in the high MCP-1 group and no patients in the low MCP-1 group showed a reduction of less than 50% (graft failure). No primary nonfunction was observed in our patients.

This is the first report demonstrating that human pancreatic islets prepared for a clinical transplant constitutively produce MCP-1 in the absence of detectable infections or endotoxin contamination. The produced MCP-1 is biologically active because it chemoattracts monocytes in chemotaxis assay and its activity (referred as ng/ml measured in ELISA) correlated well with that of a recombinant MCP-1 used as standard. Moreover, the chemotactic activity for monocytes was almost abrogated by a neutralizing anti–MCP-1 monoclonal antibody, indicating that no other chemotactic factors were present in the supernatants. In fact, RPA did not reveal expression of MIP-1α, MIP-1β, RANTES, or IP-10 even after LPS/IL-1β stimulation. We did find mRNA for IL-8, and protein levels were measured in pancreatic islet supernatants even if it is almost 1 log less than the release of MCP-1. In chemotaxis assay, pancreatic islet supernatants were unable to induce migration of human polymorphonuclear leukocytes (PMNs) (data not shown). A major goal of this study was to correlate the clinical outcome of patients who receive allogeneic transplants of pancreatic islets with biological parameters. Primary cultures of pancreatic islets displayed similar characteristics but differed in the production of MCP-1. Follow-up demonstrated that patients who received islets that produced high levels of MCP-1 (>4.56 pg · islet⁻¹ · 24 h⁻¹) were not able to induce long-term insulin independence, as two of seven patients reached insulin independence and only for a short period, whereas five patients had graft failure or partial graft function. In contrast, six of seven patients who received islets that produced low levels of MCP-1 reached long-lasting insulin independence. A possible interpretation is that high titers of MCP-1 at the graft site (liver parenchyma) attracts monocyte-macrophages locally. The recruited macrophages, activated by the inflammatory reaction, eventually lead to the destruction of the grafted organ. In support of this is the demonstration that elimination of macrophages improves graft survival in a rodent model of islet transplantation (42). The demonstration that islet MCP-1 release interferes with the engraftment may open new approaches in human islet transplantation. First, the identification of factors and/or drugs that are able to regulate the in vitro MCP-1 secretion could suggest strategies to identify new immunosuppressive approaches. Second, a screening of islet preparations based on their MCP-1 release before transplant may increase the success rate of transplantation.

Another major and intriguing finding of this study is the immunoreactivity for MCP-1 and also specific RNA by ISH in pancreatic sections from surgical specimen (pancreatic adenocarcinomas, chronic pancreatitis, pancreas from multiorgan donors). Although none of these clinical settings is perfectly normal, sections were carefully selected for lack of evident necrosis or hypoxia or hemorrhages. In all pancreata studied, islets stained positive for MCP-1 with four different anti–MCP-1 monoclonal antibodies. MCP-1 was produced by β-islet cells because both mRNA and protein stained insulin-producing cells. In addition, no CD45+ leukocytes were found within the islets. The presence of MCP-1 in pancreatic islets raises the question of which role MCP-1 might play, different from leukocyte recruitment, in an endocrine organ. Our data seem to exclude a role in the endocrine metabolism. In fact, MCP-1 does not affect the release of insulin or vice versa, and high glucose concentrations do not modulate MCP-1 release. MCP-1 does not induce elevation of free [Ca²⁺]_i (like glucose). It has long been known that chemokines also have no chemotactic functions, including control of angiogenesis, collagen production, and regulation of hematopoietic precursor proliferation. MCP-1, in particular, increases deposition of collagen type 1 by fibroblasts and upregulates endogenous expression of TGF-β, an important fibrogenic cytokine (52). The role of MCP-1 could facilitate the formation of the peri-islet capsule, which is severely damaged during in vitro isolation procedure (53) and is of crucial importance in vivo to separate the endocrine cells from the exocrine pancreas producing highly reactive enzymes. MCP-1 null mice did not show any gross alteration of organ structures, but organization of pancreatic islet architecture in these mice was not specifically addressed. The real function of MCP-1 in pancreatic islets, therefore, remains elusive. Other tissues produce MCP-1, among them are many tumors, from which MCP-1 was first identified. In human ovarian carcinomas, levels of MCP-1 correlate with infiltrating macrophages (54). In contrast, MCP-1 expression in pancreatic islets is not associated with an inflammatory infiltrate. This is reminiscent of regulatory mechanisms that may shut off the recruitment of potentially dangerous leukocytes. β-cells in pancreatic islets produce migration inhibitory factor (MIF), an old cytokine with inhibitory effects on monocyte migration (55). It might be that its presence with β-cells encompasses also the limitation of infiltrating monocytes. It is worth noting that some studies recently demonstrated that MIF also inhibited the MCP-1–induced migration of monocytes (56,57). In addition, we do not know which is the amount of MCP-1 released by islet in vivo and if in physiological condition is sufficient to induce monocyte recruitment. Finally, the monocyte recruitment is a multistep process that occurs through complex interactions of adhesion molecules and chemoattractants and their receptors and that involves endothelial cells. Some of our preliminary data (not shown) showed that islets are able to produce molecules that inhibit endothelia permeability and activation from inflammatory stimuli. A regulatory role to inducing a resting state on endothelial cells may explain the absence of a strong monocyte infiltration despite a low secretion of MCP-1.

Another major question is whether MCP-1 plays any role in diabetes onset. Pathological examinations in NOD mice have demonstrated that pancreatic islets are infiltrated early with cells of the monocyte-macrophage lineage and that this event precedes an inflammatory reaction (insulitis) and subsequent organ destruction. It was shown recently that the NOD genetic background is characterized by the presence of MCP-1 expression by pancreas-resident cells even in the absence of infiltrated mononuclear cells (58). If the constitutive production of MCP-1 in pancreatic islets does not result in insulitis, then it is reasonable to speculate that in the presence of primary inflammatory cytokines and increased levels of MCP-1, increased leukocyte recruitment may occur. It should be noted that MCP-1 transgenic mice, in which the transgene was under the control of the insulin promoter, have normal glycemia (59). These mice developed a monocytic infiltrate but never became diabetic, and there was no evidence for any local tissue destruction. This evidence strengthens the concept that, in vivo, MCP-1 functions mainly as a chemotactic factor without inducing cell activation, indicating that locally produced MCP-1 could recruit but not activate monocytes. It is very likely that other inflammatory signals (e.g., IL-1, TNF, LPS) are required to obtain the full activation of leukocytes. Of course, a possible involvement of MCP-1 in the pathogenesis of diabetes needs to be further analyzed. Other works recently reported the expression of MCP-1 mRNA and protein in rat and human islet cells exposed to cytokines (60,61). Similar to our results, they reported in human islet a great increase of MCP-1 release after IL-1β exposure and only a slight increase after TNF-α (not significant in the study of Chen et al. [60] but significant in our study). Furthermore, different from our data, the authors reported MCP-1 mRNA in human islets only after exposure to IL-1β and not in a constitutive situation. It must be noted that in the work of Chen et al., even when mRNA for MCP-1 is not present without IL-1β stimulation, protein production analyzed by ELISA showed that MCP-1 is also present in control unstimulated human islets. So the absence of mRNA in unstimulated islets was probably due to the sensitivity of the method.

In conclusion, this study is the first to report that unstimulated human pancreatic islets produce a high level of MCP-1, which plays a relevant role in the clinical outcome of islet transplantation in patients with type 1 diabetes. In fact, low MCP-1 secretion resulted as the more relevant factor for successful engraftment and long-lasting insulin independence. This finding opens new approaches in the management of human islet transplantation. MCP-1 production by pancreatic islets is strongly increased by inflammatory cytokines, suggesting a possible involvement of this chemokine in the induction of insulitis and pathogenesis of pancreatic endocrine diseases. Finally, the finding that MCP-1 is constitutively present in normal human islet cells, in the absence of an inflammatory infiltrate, suggests that this chemokine could have functions other than monocyte recruitment, and it opens a new link between the endocrine and immune systems.

This work was supported by grants of CNR (Finalized Project Biotechnology No. 97.01301. PF 49), Telethon E.443, and by a JDF Innovative Grant (5-2001+172) and a JDF Award (1-2000-780). We thank Phan Van Choung, Alessandra Caputo, Maria Santopinto, and Elena Dal Cin for excellent technical assistance.