Improved Vascular Engraftment and Graft Function After Inhibition of the Angiostatic Factor Thrombospondin-1 in Mouse Pancreatic Islets

Pancreatic islets from wild-type (TSP-1^+/+), heterozygous TSP-1^+/−, and TSP-1^−/− C57BL/6 mice of the F2-F3 generations were used for transplantation. The TSP-1^−/− mice were generated by homologous recombination in 129/Sv-derived ES cells implanted in C57BL/6 blastocysts (24). A breeding program of such mice was established at Uppsala University, and male mice 10–12 weeks of age were allocated to the studies. Age-matched wild-type male C57BL/6 mice were used as controls. Recipient C57BL/6 (nu/nu) mice weighing ∼30 g were purchased from M&B Research and Breeding Center (Ry, Denmark). For experiments with siRNA, adult, inbred C57BL/6 mice (M&B) were used both as islet donors and recipients. All animals had free access to water and food throughout the course of the study. The experiments were approved by the animal ethics committee for Uppsala University.

Islets from wild-type, TSP-1^+/−, and TSP-1^−/− C57BL/6 mice were prepared by collagenase digestion (25) and cultured at 37°C free-floating in 5 ml culture medium composed of RPMI 1640 (Sigma-Aldrich, Irvine, U.K.), to which we added 11 mmol/l glucose, 10% (vol/vol) FCS (Sigma-Aldrich), 0.17 mmol/l sodium benzylpenicillate, and 0.17 mmol/l streptomycin.

siRNA transfection was performed as previously described (26). Freshly isolated islets were dispersed at 37°C into single cells by addition of 5 mg/ml trypsin (Sigma-Aldrich) for <5 min. The trypsination was terminated with Ca²⁺-containing culture medium (RPMI 1640), followed by DNase I treatment (33 mU/μl; Amersham Life Sciences, Piscataway, NJ) for 1–2 min. The dispersed cells were washed in RPMI 1640 (Sigma-Aldrich) and divided into two equal fractions. Nonadhesive culture dishes were pretreated with FCS for 30 min at room temperature and then washed with RPMI 1640. Each fraction was placed into a nonadhesive culture dish followed by transfection with 100 nmol/l scramble or an equal mixture of four different TSP-1 siRNA sequences (Table 1) using 10 μg/600 μl Lipofectamine (Invitrogen, Lidingö, Sweden) in 600 μl RPMI 1640. Using this Lipofectamine protocol, we have previously observed uptake of siRNA in >90% of all islet cells (26). After 3 h, the transfection medium was replaced with culture medium. The dispersed islets were re-aggregated with a shaker overnight at standard culture conditions before transplantation.

Dispersed islet cells transfected with scramble or TSP-1 siRNA, or freshly isolated islets, were washed with PBS followed by total RNA isolation using Ultraspec (Biotecx Laboratories, Houston, TX). cDNA synthesis was performed with random nonamers (Sigma-Aldrich) and reverse transcriptase M-MuLV (Moloney murine leukemia virus) H⁻ (Finnzymes, Espoo, Finland). Amplification was obtained with a Lightcycler system (Roche-Diagnostic, Lewes, U.K.) using DyNAmo Capillary SYBR Green qPCR kit (Finnzymes). β-Actin was used as housekeeping gene. For primer sequences used, see Table 1.

Whole islets, kidney and spleen homogenates, and dispersed islet cells transfected with scramble, or TSP-1 siRNA and incubated 1–7 days were washed in cold PBS after lyses in SDS sample buffer (2% SDS, 0.15 mol/l Tris, pH 8.8, 10% glycerol, 5% β-mercaptoethanol, bromphenol blue, and 2 mmol/l phenylmethylsulfonyl fluoride), boiled for 2 min, and separated on a 6% SDS-PAGE gel. Proteins were transferred to Hybond-P membrane (GE Healthcare, Uppsala, Sweden). The membranes were blocked in 2.5% BSA at 4°C overnight followed by incubation with TSP-1 (Ab-2; Ab-4 Labvision, Fremont, CA) (TX-17.10; Santa Cruz Biotechnology, Santa Cruz, CA) and HSP60 antibodies (Stressgen, Ann Arbor, MI) and probed with horseradish peroxidase antibody. The bound antibodies were visualized with Kodak image station 4000MM (Kodak, New Haven, CT) using ECL⁺ (GE Healthcare). The band intensities were calculated using Kodak Molecular imaging software 4.5.1 SE.

Dispersed islet cells transfected with scramble or TSP-1 siRNA were re-aggregated overnight, cultured for 72 h, and investigated for glucose-stimulated insulin release and insulin content. Groups of islet cells from 10 islets were transferred in triplicate to glass vials containing 250 μl Krebs-Ringer bicarbonate HEPES (KRBH) buffer supplemented with 10 mmol/l HEPES (Sigma-Aldrich) and 2 mg/ml BSA (ICN Biomedicals, Aurora, OH); the buffer is hereafter referred to as KRBH buffer. The KRBH buffer contained 1.67 mmol/l d-glucose during the 1st h of incubation at 37°C (O₂:CO₂ = 95:5). The medium was removed and replaced by 250 μl KRBH buffer supplemented with 16.7 mmol/l glucose and incubated for a 2nd h. The medium was again removed, and the islet cells were harvested and homogenized in 200 μl redistilled water. The aqueous homogenate was then used for DNA measurements by fluorophotometry (PicoGreen dsDNA Quantitation kit; Molecular Probes, Eugene, OR). A fraction of the homogenate was mixed with acid-ethanol (0.18 mol/l HCl in 95% [vol/vol] ethanol), from which insulin was extracted overnight at 4°C. Insulin contents in incubation medium and homogenates were determined by ELISA (Mercodia, Uppsala, Sweden).

Groups of 250 wild-type, TSP-1^+/−, or TSP-1^−/− C57BL/6 islets cultured for 3–4 days or re-aggregated scramble or TSP-1 siRNA transfected C57BL/6 islet cells were packed in a braking pipette and syngeneically implanted beneath the capsule of the left kidney in nondiabetic C57BL/6 nu/nu mice anesthetized with avertin (2.5% [vol/vol] solution of 10 g 97% [vol/vol] 2,2,2-tribromoethanol [Sigma-Aldrich] in 10 ml 2-methyl-2-butanol) (Kemila, Stockholm, Sweden). For the transfected material, all re-aggregated islet cells, irrespective of the size of the organoids, were transplanted.

One month posttransplantation, the animals were anesthetized with avertin (compare with above), placed on a heated operating table (38°C), and tracheotomized. Polyethylene catheters were inserted into the right carotid artery and left jugular vein. The former catheter was connected to a Statham P23dB pressure transducer (Statham Laboratories, Los Angeles, CA) to monitor arterial blood pressure, whereas the latter catheter was used for continuous infusion of Ringer solution (5 ml · kg⁻¹ · h⁻¹) to substitute for loss of body fluid.

After a left subcostal flank incision, the graft-bearing left kidney was immobilized in a plastic cup. The kidney was embedded in cotton wool soaked in Ringer solution and covered with mineral oil (Apoteket, Gothenburg, Sweden) to prevent evaporation and keep the tissue moist at body temperature. The blood perfusion of the islet graft and the adjacent renal cortex was measured by laser-Doppler flowmetry (PF 4001-2; Perimed, Stockholm, Sweden). The laser-Doppler probe (outer tip diameter 0.45 mm) was positioned perpendicular to the immobilized tissue surface by the use of a micromanipulator, and care was taken not to cause any compression of the tissue. At least three blood flow measurements were performed in the transplanted islets and renal cortex in each animal. The mean of these measurements from each animal was calculated and considered to be one experiment. Because it is difficult to calibrate the instrument in physical units of blood flow, all blood flow values are given as arbitrary tissue perfusion units (TPU).

Oxygen tension was measured in the islet graft and the adjacent renal parenchyma with Clark microelectrodes (Unisense, Arhus, Denmark), as described in detail previously (27,28). The electrodes (outer tip diameter 2–6 μm) were inserted into the tissues by the use of a micromanipulator under a stereomicroscope. At least 10 measurements were performed in both the islet graft and the renal cortex. The mean of all measurements, in each tissue and animal, was calculated and considered to be one experiment. During the blood flow and oxygen tension measurements, blood pressure, body temperature, and tissue temperature were continuously monitored with a MacLab Instrument (AD Instruments, Hastings, U.K.). A mean arterial blood pressure <75 mmHg was used as a preset exclusion criteria from the study.

Blood glucose concentrations were determined with test reagent strips (Medisense; Baxter Travenol, Deerfield, IL) from samples obtained from the cut tip of the tail. At the end of blood flow and oxygen tension measurements, a blood sample was collected for analysis of hematocrit and blood gases. Animals with pH <7.30, pO₂ <10 kPa, pCO₂ >6.8 kPa, or hematocrit <40 were excluded from the study.

The graft-bearing kidneys were removed after the oxygen tension and blood flow measurements, fixed in 10% (vol/vol) formaldehyde, and embedded in paraffin. Consecutive sections (5 μm thick) of the islet grafts were prepared and stained with the lectin Bandeiraea simplicifolia (BS-1) or a monoclonal guinea pig anti-insulin antibody (ICN Biomedicals), as previously described (5), and counterstained with hematoxylin. In each case, ≥10 tissue sections stained with BS-1 or for insulin from all parts of the islet transplants were randomly chosen and evaluated. The blood vessel density and β-cell percentage in pancreatic islets was determined by a direct point-counting method (29). For this purpose, a grid with 121 intersections was placed onto each tissue section under a light microscope (BS-1, ×400; insulin, ×1,000) (19). The number of intersections overlapping endothelial cells (stained with BS-1) or insulin-positive cells was then counted by an examiner unaware of the origin of the samples.

Grafts from some of the animals transplanted with scramble or TSP-1 siRNA-transfected wild-type islet cells were investigated 1 month posttransplantation for glucose-stimulated insulin secretion as previously described (30). Briefly, the graft-bearing left kidney was removed together with a part of the aorta and inferior vena cava. The ureter and the renal vein were cut, while the aorta was cannulated and infused with a continuously gassed (O₂:CO₂ = 95:5) Krebs-Ringer bicarbonate buffer supplemented with 2.0% (wt/vol) each of BSA (fraction V; Miles Laboratories, Slough, U.K.) and dextran T70 (Pharmacia, Uppsala, Sweden). At different times during the perfusion, the medium contained either 2.8 or 16.7 mmol/l d-glucose. The medium was administered at a rate of 1 ml/min without recycling for 60 min with a perfusion pressure of ∼40 mmHg. The perfusion experiments started with a 15-min period using medium containing 2.8 mmol/l glucose, which was followed by 30 min using 16.7 mmol/l glucose. The perfusions were concluded by a 15-min perfusion with medium containing 2.8 mmol/l glucose. A 1.0-ml sample was collected at 14, 15, 16, 17, 18, 19, 20, 22, 25, 30, 35, 40, 45, 50, 55, and 60 min. The insulin concentrations of the effluent samples were measured by ELISA (Mercodia). The rate of insulin secretion was calculated by multiplying the insulin concentration in the sample by the flow rate, giving values of insulin expressed as nanograms per minute. The area under the curve (AUC) was then determined from these values.

Values are expressed as the means ± SE. For comparisons of nonparametric data, Wilcoxon's signed rank test was used for comparisons between two groups, and Kruskal-Wallis was used for multiple comparisons. When two groups of parametric data were compared, Student's paired or unpaired two-tailed t test was used. For all comparisons, P values <0.05 were considered statistically significant.

All recipient mice allocated to the study weighed ∼30 g at transplantation and had increased 5–10% in body weight when investigated 1 month later. Both donors of islets, including the TSP-1^−/− mice, and recipients were normoglycemic with nonfasting blood glucose concentrations of ∼7 mmol/l. All animals allocated to blood flow and oxygen tension measurements had a mean arterial blood pressure of 90–100 mmHg. No animals had to be excluded based on the preset exclusion criteria for blood gases and hematocrit. However, two animals were excluded from studies of islet graft blood flow and oxygen tension because of hypotension (mean arterial blood pressure <75 mmHg) following perioperative bleeding.

TSP-1 immunoreactivity was observed as a double band. When compared with spleen, pancreatic islets and kidneys contained low amounts of the protein TSP-1 (Fig. 1A).

Dispersed islet cells transfected with TSP-1 siRNA showed an ∼40% reduction in TSP-1 mRNA expression compared with dispersed islet cells transfected with scramble siRNA when evaluated 24 h posttransfection (Fig. 1B). At the protein expression level, the effects of TSP-1 siRNA transfection caused a delayed effect, with a ∼40% decrease first after 1 week (Fig. 1C and D).

There were no differences in insulin release at low glucose concentrations (50.2 ± 5.5 vs. 47.2 ± 6.7 ng · μg DNA⁻¹ · h⁻¹), in insulin release at high glucose concentrations (203.0 ± 18.8 vs. 187.2 ± 19.3 ng · μg DNA⁻¹ · h⁻¹), or in islet insulin content (8.2 ± 0.1 vs. 7.5 ± 0.3 ng/ng DNA) between islet cells transfected with scramble siRNA or TSP-1 siRNA 72 h before measurements.

One month posttransplantation, the blood perfusion of grafts composed of wild-type control islets or scramble siRNA-transfected islet cells were similar and 40–50% of that in the adjacent renal cortex (Fig. 2A and B, respectively). In comparison, grafts composed of islet tissue either genetically decreased or deficient in TSP-1 (Fig. 2A) or transfected with siRNA for TSP-1 (Fig. 2B) had a markedly higher blood perfusion when compared with their corresponding control.

The pO₂ levels in islet grafts composed of wild-type control islets (Fig. 3A) or in islet cells transfected with scramble siRNA before transplantation (Fig. 3B) were 7–8 mmHg 1 month posttransplantation. Improved oxygenization compared with this was observed in islet grafts composed of TSP-1^+/− islets, TSP-1^−/− islets, or islet cells transfected with TSP-1 siRNA before transplantation (Fig. 3A and B, respectively).

Both wild-type control islets (Fig. 4A and B) and scramble siRNA-transfected islets tissue (Fig. 4D and E) had a blood vessel density of ∼7% 1 month posttransplantation. By contrast, both TSP-1^−/− islets (Fig. 4A and C) and islet tissue transfected with siRNA for TSP-1 (Fig. 4D and F) had an ∼40% increase in vascular density. Islet grafts composed of TSP-1^+/− islets had a vascular density in-between wild-type islet grafts and TSP-1^−/− islet grafts (Fig. 4A). The increase in blood vessels numbers seemed to occur both in the center and peripheral regions of the TSP-1^−/− islets and TSP-1 siRNA-transfected islet tissue.

There were similar percentages of β-cells in grafts composed of wild-type islets or TSP-1^−/− islets (84.7 ± 1.3% [n = 4] vs. 86.2 ± 0.9% [n = 5], respectively). Likewise, there was no difference in β-cell percentage in the islet tissue composed of scramble siRNA-transfected and TSP-1 siRNA-transfected islet cells at 1 month follow-up (85.1 ± 0.5 vs. 86.9 ± 1.3%, respectively).

Perfusions of islet graft–bearing kidneys 1 month posttransplantation showed that islet tissue transfected with siRNA for TSP-1 had an improved first and second peak of insulin release when compared with scramble siRNA-transfected islet cells (Fig. 5A–C). Total AUC for glucose-stimulated insulin release in the TSP-1 siRNA-transfected islet tissue was thus increased by ∼40% (Fig. 5D).

Adult pancreatic islets have a dense glomerular-like angioarchitecture that requires VEGF-signaling from the β-cells for its formation and maintenance (31–34). However, despite chronic exposure to VEGF, vascular expansion does not normally occur in islets after the 1st postnatal week (35). Likewise, although increased amounts of not only VEGF but also other pro-angiogenic factors such as hepatocyte growth factor are secreted from newly transplanted islets (16,36), new blood vessel formation within the islets also during these conditions is sparse (5–7). Instead, a rich vascular network is formed in the stroma around each single transplanted islet (5). This suggests the presence of matrix- or tissue-bound angiostatic factors within the implanted islets that prevent intra-islet vascular formation. Marked overexpression of VEGF in the donor tissue has previously been shown in experimental studies to overcome this and to improve islet graft revascularization and function (9,10). However, silencing some of the angiostatic factors may be an even simpler strategy.

In this study, we tested the hypothesis that unopposing the action of already-present pro-angiogenic factors through silencing TSP-1 in islets would improve islet graft revascularization and function. The glycoprotein TSP-1 binds to the extracellular matrix (37,38) and exerts its main effects through inducing apoptosis selectively in activated endothelial cells, i.e., those that are forming new blood vessels, but not quiescent endothelium (39). It also inhibits angiogenesis by blocking the mobilization of pro-angiogenic factors, such as matrix metalloproteinase-9 and VEGF, and by inhibiting their access to co-receptors on the endothelial cell surface (40).

We investigated the influence of TSP-1 on islet graft revascularization using two different models: by transplanting genetically TSP-1 deficient islets and by transplanting islet cells transfected with siRNA for TSP-1 in vitro before transplantation. The first model thereby provides a persistent partial (TSP-1^+/−) or complete loss (TSP-1^−/−) of TSP-1 expression in the islets, whereas the latter causes only a transient decrease in TSP-1 levels. Using both approaches, we observed a markedly improved vascular engraftment of the transplanted islets with improved intra-islet vascular density at 1 month follow-up. In both models, the newly formed blood vessels also seemed to be functional, because the improved vascular density was reflected in an increased blood perfusion and pO₂ in the islet grafts. Notably, the pO₂ increase was especially prominent in the TSP-1^−/− mice, where TSP-1 expression is chronically deficient and of a greater magnitude than the increase in blood vessel numbers. It is possible that this reflects higher blood perfusion in individual blood vessels and thereby better oxygen transport capacities of islet blood vessels in the absence of TSP-1, e.g., due to loss of the previously described counteracting effect of TSP-1 on nitric oxide–mediated vasodilation (41). Nitric oxide is the main mediator of high blood perfusion in endogenous and transplanted islets (42,43). It should be noted that an intra-islet vasodilation is not necessarily reflected in increased total graft blood flow, which is measured by laser-Doppler flowmetry, because total graft blood flow represents not only the nutritive blood flow to the endocrine cells but also blood flow in the vast number of stroma capillaries. In contrast, it could be expected that the blood flow in capillaries in the endocrine parts mainly contributes to the delivery of oxygen to the endocrine cells due to limitations of oxygen diffusion (44).

In our characterization of TSP-1^−/− mice, we observed that these mice had fasting and nonfasting blood glucose levels that were normal and did not differ from those of wild-type animals. However, despite that TSP-1^−/− mice have been described to have an increased islet mass (23), closer investigations revealed that these animals were slightly glucose intolerant and had a decreased islet function (45), which probably reflects the role of TSP-1 in transforming growth factor-β₁ activation (23). Nevertheless, we hypothesized that a transient decrease in islet TSP-1 levels during the acute revascularization phase of transplanted islets may still be beneficial for long-term islet function, considering that such a decrease also can be seen physiologically during the islet vascular expansion of pregnant rats. In our protocol for TSP-1 siRNA transfection of islet cells using Lipofectamine as vector, there was a transient decrease in islet TSP-1 levels at the mRNA and protein levels of a similar magnitude as seen in islets during pregnancy, i.e., ∼40%. In rapidly dividing cells, the silencing effect of siRNA usually lasts for <1 week, but it lasts for up to 3–4 weeks in nondividing cells both in vitro and in vivo (46). Probably because of the low proliferation rate of adult endocrine cells, including β-cells, and turnover rate of the extracellular matrix binding of the TSP-1 protein, silencing of TSP-1 in these cells was not manifested until after 1 week at the protein level. A partial effect of TSP-1 siRNA may also be on the remnant donor islet endothelial cells, because these cells also contain TSP-1 (22). However, in this case, the effect is probably much shorter because of cell division as part of angiogenesis.

When graft function was evaluated by perfusion of graft-bearing kidneys 1 month posttransplantation, islet cells exposed to siRNA for TSP-1 before transplantation performed better than control islet cells exposed to scramble siRNA. The chosen model to evaluate graft function also enabled us to separately evaluate the effects of improved vascular engraftment on the insulin release kinetics from transplanted islets. It is noteworthy that the peak of the first phase of glucose-stimulated insulin secretion was substantially higher in the better revascularized islet grafts, which is similar to what we have previously observed when improving blood perfusion, but not revascularization, by angiotensin II receptor inhibition (47). However, in contrast to when only the blood perfusion was improved, we observed in the present study also an augmented second phase of glucose-stimulated insulin secretion. This suggests that the present findings do not merely reflect a “wash-out” effect of insulin retained in the islet grafts due to low blood perfusion. The general improvement of glucose-stimulated insulin release may instead reflect an improved survival, a better oxygenation, and/or a better endothelial paracrine support of the islet β-cells. The normally low oxygenation of transplanted islets may affect islet survival and function, because it means that their metabolism tends to be much more anaerobic than that of endogenous islets and results in lactate formation and tissue acidosis (48). Several recent reports also indicate that endothelial cell signaling may be important for β-cell function and growth (19,49,50).

We conclude that transient inhibition of TSP-1 early posttransplantation may be used to improve islet graft revascularization and function. Such a strategy, e.g., by development of pharmacological inhibitors, may be more feasible in the clinical setting than trying to increase the expression of pro-angiogenic factors in the transplanted tissue even further.

This work was supported by grants from The Swedish Research Council (55X-15043), The Juvenile Diabetes Research Foundation, the European Foundation for the Study of Diabetes/Juvenile Diabetes Research Foundation/Novo Nordisk Research Programme 2006, The Swedish Diabetes Association, The Swedish Juvenile Diabetes Fund, The Anér Foundation, and The Family Ernfors Fund.

The skilled technical assistance of Astrid Nordin, Birgitta Bodin, and Eva Törnelius is gratefully acknowledged.