Clinical and Experimental Pancreatic Islet Transplantation to Striated Muscle: Establishment of a Vascular System Similar to That in Native Islets

Male C57Bl/6 mice (25–30 g ([B&K Universal]) and C57Bl/6 nu/nu mice (25–30 g [Taconic M&B]) were used. Experiments were approved by the Uppsala Laboratory Animal Ethical Committee. An online appendix describes additional methods (http://diabetes.diabetesjournals.org/cgi/content/full/db10-0205/DC1).

Mouse islets of Langerhans were isolated and cultured over night as previously described (22). Human islets from five heart-beating female donors (age 57 ± 7 years) were isolated and cultured as previously described (23). Islets were fluorescently labeled immediately before transplantation with the intracellular probes Celltracker Blue CMAC or Celltrace Far Red DDAO (Invitrogen), depending on the type of imaging performed later. For transplantation into the cremaster muscle (surrounding the testis) of nondiabetic mice, 5–20 islets in suspension were, by the use of a butterfly needle (25G), repeatedly injected subfascially at different spots to allow for single islet engraftment. For transplantation to the liver of nondiabetic mice, 200 islets were infused via the portal vein.

Diabetic mice (injected with 75 mg/kg i.v. alloxan; Sigma-Aldrich) had plasma glucose concentrations of 25.8 ± 1.0 mmol/l on the day of transplantation. A suspension with 300 islets was injected superficially between abdominal external oblique muscle fibers (on the side of the abdomen) or infused into the portal vein to the liver. Plasma glucose levels <11.1 mmol/l were deemed normalized, and mice were killed if blood glucose levels were >20 mmol/l 7 days posttransplantation.

The cremaster muscles of isoflurane-anaestetized (Abbott) mice with implanted islets were prepared for intravital microscopy as previously described (24). A catheter in the femoral artery allowed retrograde infusion close intra-arterially to the muscle. An intravital microscope (Leica Microsystems DM5000B equipped with a Hamamatsu Orca-R2 CCD camera and HCX Apo L 20×/0.50W and 40×/0.80W objectives; Volocity software) was used to visualize the microcirculation of transplanted islets and surrounding muscle. Recordings were made for analysis of adherent (stationary for >30 s within 100 μm length of venule in 3 min) and emigrated leukocytes (cells in the extravascular space per field of view [0.05 mm²]).

Laser scanning confocal microscopy (Nikon C-1 with Plan Fluor ELWD 20×/0.45 and 40×/0.60 objectives; Nikon EZ-C1 software), spinning disk confocal microscopy (Olympus BX51/Quorum WaveFx with a Hamamatsu C9100-13 EMCCD camera and an XLUM Plan F1 20×/0.95W objective; Volocity software), or multiphoton microscopy (Olympus FV300/Chameleon ti:sapphire laser with a 40×/0.80W objective and Olympus FluoView and ImageJ software or Zeiss 710 NLO with a Plan-Apo 20×/1.0W objective; Zeiss Zen software) was performed after intra-arterial injection of 25 μg Alexa Fluor 555 or 594 anti–CD31 monoclonal antibody (mAb) (clone 390; eBioscience) to stain endothelium and 15 μg fluorescein isothiocyanate (FITC) anti–Gr-1 mAb (clone RB6–8C5; eBioscience) to stain neutrophils.

Paraffin sections of transplants were stained for insulin with anti-insulin antibody (Fitzgerald) and for endothelium with the lectin Bandeiraea simplicifolia-1 (BS-1) (Sigma-Aldrich) (supplemental Figs. 1 and 2). Vessel functionality within grafts was studied through intrajugular injection of 100 μg soybean agglutinin (SBA) lectin, which stains endothelium and allows for detection of perfused vessels. Endothelium was stained in cryosections with anti-CD31 mAb (conjugated to Alexa Fluor 555), and nuclei were stained with Hoechst 33342 (Invitrogen).

Mice were rendered neutropenic by intraperitoneal injections of anti–Gr-1 (clone RB6-8C5; eBioscience). Immunoglobulin G (IgG) isotype antibody (eBioscience) was injected in controls. One day prior to transplantation (day zero), 150 μg was injected; an additional 150 μg was injected on day three. Determination of treatment efficiency was made by blood analysis and differential count. Total leukocyte counts were reduced by 50% (8.2 ± 1.0 × 10⁹/l vs. 4.1 ± 0.6 × 10⁹/l), and neutrophil counts were reduced by 80% (19.2 ± 5.2 × 10⁸/l vs. 4.2 ± 1.2 × 10⁸/l). No Gr-1–postive cells were found nearby the grafts in Gr-1 antibody–treated animals.

The study was approved by the Regional Ethics Board, Uppsala, Sweden, and was performed in accordance with local institutional and Swedish national rules and regulations.

Three patients (two men and one woman) with intraductal papillary mucinous neoplasm underwent total pancreatectomy with a Whipple procedure and autologous intramuscular islet transplantation (Table 1). Ex vivo the pancreas was immediately perfused with cold (4°C) University of Wisconsin solution and shipped to the islet isolation laboratory of the Nordic Network for Clinical Islet Transplantation (cold ischemic time <2 hours). Islet isolation was performed as previously described (23), and islets were maintained in culture for 24 h before characterization (supplemental Table 1).

Absence of functioning endocrine pancreatic tissue before islet transplantation was confirmed in all patients by lack of C-peptide (Table 1) (25). Under brachial plexus anesthesia, islets in volumes of 50–100 μl were injected into the muscle fibers of the brachioradialis muscle (in the forearm) with the help of a central venous catheter (Secalon-T, 16G, 130 mm; BD Biosciences). Islet graft function after transplantation was assessed by circulating C-peptide (Table 1).

The subjects were scanned 3–6 months posttransplantation using a 1.5 T clinical magnetic resonance scanner (Gyroscan Intera; Philips Medical Systems). The body coil was used for radio frequency transmission and a 45-mm circular linear coil for radio frequency reception. High-resolution axial images were obtained using a T1-weighted three-dimensional gradient echo acquisition with 32 slices, (in-plane resolution of 200 × 200 μm² and a slice thickness of 1.0 mm; TR/TE/flip 16/5/10).

Data are means ± SEM. Paired and unpaired two-tailed Student t tests were used to compare between groups.

This study used a novel experimental model where isolated pancreatic islets are transplanted into the mouse cremaster muscle for ease of visualization. The reestablished circulation to the transplanted tissue was studied by transillumination, epifluorescence, spinning disk, laser scanning confocal and multiphoton microscopy, and the functionality of intramuscularly engrafted islets was tested in diabetic recipients. For translational purposes, ultra-high resolution magnetic resonance imaging was performed on autotransplanted islets engrafted in the brachioradialis muscle of patients in order to evaluate islet revascularization through measurements of plasma volume in transplanted islets versus surrounding muscle tissue.

Islets syngeneically transplanted to the cremaster muscle of nondiabetic mice rapidly became vascularized and showed functional intraislet blood vessels already at 3–5 days after transplantation, as confirmed by close intra-arterial injection of fluorescently labeled anti-CD31 mAb (Fig. 1,A), with a vessel density of 407 ± 94 vessels/mm² (Fig. 1,D). Following these early events, the vasculature continued to develop and, 2 weeks after transplantation, had the characteristic glomerular-like vascular network (Fig. 1,B) seen in native islets (Fig. 1,C). Two to four weeks posttransplantation, intraislet vascular densities had increased threefold (1,187 ± 187 and 1,162 ± 120 vessels/mm², respectively) and were similar to those observed in islets in intact pancreas (1,074 ± 174 vessels/mm²) (Fig. 1,D). Imaging of islet vasculature with deeply penetrating multiphoton microscopy 2 weeks posttransplantation revealed a complete vascular network throughout the islet (supplemental Video 1). When intraislet vessel densities were analyzed in islets transplanted intraportally to the liver of nondiabetic mice, vessels surrounding the whole islets were detected but no intraislet vessels could be found 2 or 4 weeks after transplantation (n = 7 mice) (supplemental Fig. 1). Transplanted islets at both sites stained positive for insulin (supplemental Fig. 1). The intraislet capillary diameter was enlarged 3–5 days posttransplantation to muscle but was normalized to diameters found in islets in intact pancreas after 2 weeks (Fig. 1 E) as previously reported (26).

To investigate whether the improved vascularization of mouse islets transplanted to muscle also applied to human islets, human islets were implanted into the cremaster muscle of immune incompetent mice (nu/nu). For logistic reasons, the human islets were cultured for an average of 4 days prior to transplantation and, for comparison, mouse islets were also cultured for an equal period of time. Four weeks posttransplantation, human islets were highly vascularized with an intraislet vascular density comparable with that of cultured mouse islets (Fig. 1,F). Intravital and confocal microscopy showed functional vessels in transplanted human islets, although the vessel structure was somewhat different than that observed for transplanted mouse islets: intraislet capillary diameters were more heterogeneous (9–20 μm) and had a differently organized, less glomerular-like vasculature (Fig. 1,G and supplemental Video 2) than that observed in mouse islets (Fig. 1,B and C). The number of remnant endothelial cells has been reported to decrease with time in culture (27). However, the vascular densities of mouse islets cultured for 4 days were not significantly lower compared with mouse islets cultured overnight (942 ± 50 and 1,162 ± 120 vessels/mm², respectively) (Fig. 1 D and F), suggesting that the decreased number of remnant donor endothelial cells does not affect the vascularization 4 weeks after transplantation.

The functionality of the newly formed vessels within transplanted islets was investigated by intravenous injections of fluorescent SBA lectin (Fig. 2,A and B), which ensures endothelial staining of perfused vessels only. After termination of experiments, muscle cryosections were stained for endothelial cells with anti-CD31 mAb and the numbers of lectin- and CD31-positive vessels were analyzed. In islets transplanted to striated muscle and in islets in the intact pancreas, close to 100% of the CD31-positive vessels were also positive for lectin (Fig. 2,C–E), indicating that nearly all vessels were indeed perfused. Contrary, the very few CD31-positive cells that could be found within islets transplanted to liver showed low staining overlap with intravenously administered lectin, suggesting mainly nonfunctional vessels, i.e., possibly remnant donor endothelial cells (Fig. 2 F).

In rodents, the β-cell core of the islet receives blood before the endocrine cell types in the islet mantle: the α-, δ- and PP-cells (i.e., β-α-δ blood flow pattern) (14,28). To investigate the blood flow pattern following transplantation to muscle, fluorescently labeled dextran was injected close intra-arterially and islet blood flow was recorded by epifluorescence microscopy. Indeed, 4 weeks after transplantation, mouse islets in muscle were perfused from the core outwards because the fluorescent signal was initially observed in the central part of the islet (supplemental Fig. 3A, C, and E). This indicates that transplantation and revascularization of islets at the intramuscular site do not change the inherent perfusion pattern.

Contrary to the mouse islet organization described above, human islets have a differently organized distribution of endocrine cells throughout the islet with clusterized subunits (29,,–32). Blood flow recordings of human islets transplanted to the cremaster muscle of nu/nu mice during fluorescent dextran injections revealed a differently organized perfusion pattern (4 weeks posttransplantation [supplemental Fig. 2B, D, and F]). While the blood perfusion of mouse islets was directed from the core and outward, human islets were observed to be perfused from one side to the other in a more polarized manner and no core-to-mantle perfusion could be detected. This implies that the blood flow of pancreatic islets differs between species and that the inherent organization of different endocrine cell types within the islet influences revascularization and thereby gives the perfusion order.

Syngeneic transplantation is not associated with acquired immune cell activation or concomitant graft rejection because donor and recipient share the same genetic background. However, intravital microscopy of newly formed vasculature in these grafts exposed interactions between circulating leukocytes and endothelial cells within the islet capillaries and in the venules draining the islets but not at sites receiving sham injections of islet culture medium. Significantly more leukocytes adhered in venules draining islets compared than in venules draining adjacent muscle tissue 3–5 days posttransplantation (Fig. 3,A), and emigrated leukocytes were found in the perivascular tissue in the proximity of recently transplanted islets (Fig. 3,B). In vivo fluorescent staining of leukocytes with an acute injection of low-dose anti–Gr-1 mAb intra-arterially revealed that the majority of recruited leukocytes were Gr-1–positive leukocytes (Fig. 3,C). The number of recruited leukocytes decreased with time after transplantation, indicating a role of leukocyte recruitment during the initial stages of reestablishment of circulation (Fig. 3,B). Gr-1–positive leukocytes have been shown to be involved in muscle healing and regeneration (20,21); we therefore investigated their involvement in the revascularization of transplanted islets. Transplantation of islets to neutropenic mice resulted in complete inhibition of revascularization of islets 3–5 days posttransplantation (Fig. 3 D and E), and no intraislet vessels could be detected.

To investigate the ability of islets transplanted to muscle to regulate blood glucose, recipient mice were rendered diabetic through alloxan treatment, resulting in plasma glucose levels of 25.8 ± 1.0 mmol/l. Thereafter, 300 islets were either transplanted into the abdominal external oblique muscle (n = 8) or into the liver (n = 6). There was not a difference in time to normalization of blood glucose levels between the two sites; both groups were below 11.1 mmol/l at just over 1 week posttransplantation (Fig. 4,A). Three mice with plasma glucose levels >20 mmol/l 1 week posttransplantation were killed because of ethical requirements (two with grafts in muscle and one with grafts in the liver). Plasma glucose levels in the successful recipients remained <11.1 mmol/l for at least 2 months. Islet functionality was tested by an intraperitoneal glucose tolerance test. Animals with islets transplanted to muscle did not significantly differ in glucose handling from nondiabetic control mice, but animals with islets transplanted to liver had a significant delay in lowering blood glucose after glucose load (Fig. 4,B), resulting in an increased area under the curve (Fig. 4,C). Islets transplanted to the abdominal external oblique muscle of diabetic mice had vascular densities similar to those of islets transplanted to the cremaster muscle of nondiabetic mice (967 ± 65 vs. 1,162 ± 120 vessels/mm²) (Fig. 4,D). Insulin staining of pancreata after termination of experiments confirmed maintained low densities of β-cells in all alloxan-treated mice, excluding the possibility of normalization of blood glucose levels due to β-cell regeneration. The intramuscular site was visualized 3 days after transplantation of fluorescently labeled islets through IVIS imaging (Fig. 4 E).

Three pancreatectomized patients receiving autotransplantation of isolated islets to the brachioradialis muscle were scanned using magnetic resonance imaging 3–6 months after surgery. The MRI technique used yields a very high in-plane resolution of 200 μm in vivo (33), which is required to accurately visualize the islet grafts. Islets could indeed be detected in muscle tissue at the site of injection in all three patients (Fig. 5,A, C, E, and F and supplemental Fig. 5). Therefore, fractional plasma volume, which corresponds to capillary density of the scanned area, was investigated through kinetic modeling. In two of the subjects, the capillary density was assessed through dynamic contrast enhancement of the equivalent parameter, the plasma volume, and was found to be 2.5 and 3.4 times higher in the implanted clusters of islets compared with the surrounding muscle tissue. The center of the larger islet grafts (Fig. 5,C and D) had a nonperfused area, which most likely corresponds to central fibrosis due to a too-large volume of islets implanted at the same site. Hepatic steatosis has been reported following islet transplantation to the liver (34,35), but no focal lipid accumulation surrounding the muscle-implanted grafts in the first two subjects could be detected (Fig. 5,A and C). Even when high-resolution morphological imaging on the third subject was used, no lipid deposits could be detected close to engrafted islets (Fig. 5 E and F).

Transplantation of isolated pancreatic islets to patients suffering from diabetes is limited by loss of function of the grafts, which appears to be related to the site of engraftment (36) including poor revascularization of the islets (17). In the present study we report that vasculature within islets engrafted in striated muscle are functionally and morphologically restored within 2 weeks of transplantation. The results were obtained using intravital and confocal microscopy and by immunohistochemical evaluation and curative transplantations of diabetic mice. Initiation of revascularization was dependent on Gr-1–positive leukocytes recruited to the implantation site; no revascularization of transplanted islets could be observed after depletion of this leukocyte subset. Ultimately, an increased capillary density was detected in islet grafts compared with adjacent tissue in pancreatectomized patients with isolated islets autotransplanted to muscle using an ultra-high resolution acquisition technique that was applied to the study of transplanted islets for the first time. This technique enables registration of plasma volumes proportional to capillary densities in small superficial, graft-containing tissue fractions corresponding to transplanted islets. In addition to verifying the experimental observations, this technique will be instrumental during our forthcoming development and evaluation of clinical islet transplantations to the promising muscle site.

Pancreatic islets are highly organized miniature organs with the primary function of regulating blood glucose levels through efficient sensing of present blood glucose concentration and concomitant release of the appropriate hormone (i.e., insulin and glucagon) from each specialized cell type. Justifiably, high blood perfusion and minimal diffusion barriers between the blood, the sensor, and the hormone-releasing cells are prerequisites for efficient blood glucose control (8,9). During isolation, all vascular connections to the islets are interrupted and, contrary to solid organ transplantations, are not reconnected during surgery. Transplanted islets therefore depend completely on revascularization to occur. When islets are grafted into the liver, blood vessels surround the islets but few penetrate the islets (2,17) (supplemental Fig. 1), which may explain the poor long-term function of the grafts. Experimental islet transplantation to other commonly used implantation sites such as the kidney and spleen also results in impaired revascularization (2). Indeed, the present study—where islets were engrafted into striated muscle—reports, for the first time, islet vascular density, vessel morphology, and blood perfusion equal to that of islets in the intact pancreas. These findings were not restricted to the cremaster muscle but could also be seen in islets transplanted to abdominal external oblique, gluteal, and psoas muscles (Fig. 4 D and supplemental Fig. 4). In addition, the experimental data were confirmed in autotransplanted patients where higher plasma volumes were measured in islets engrafted in forearm muscle compared with adjacent muscle tissue, indirectly demonstrating higher capillary densities in the grafts compared with muscle. Hepatic steatosis of tissue surrounding the grafts has been reported following islet transplantation to the liver (34,35) due to paracrine insulin effects on adjacent tissue. In the present study, no fat deposits could be detected after autotransplantation to muscle of pancreatectomized patients.

Striated muscle has successfully been used for autotransplantation of parathyroid glands, where these endocrine glands engraft and become vascularized (18). Muscle has also been investigated under experimental settings as a site for pancreatic islet transplantation but with varied success (37,–39). The variations in outcome may relate to the techniques used to implant the islets. Indeed, the negative influence of large clusters of islets were evident in the pancreatectomized patients receiving autotransplanted islets to striated muscle: fibrosis could be detected in the center of large number of islets grouped together. In the experimental part of this study, a small volume of 300 nonpacked islets was injected superficially into abdominal muscle in a pearls-on-a-string fashion (39). This resulted in a complete cure of diabetic mice within approximately 8 days following transplantation.

The increased revascularization of islets engrafted in muscle observed in the present study might be dependent on the specific plasticity of the endothelial cell population in muscle. Endothelial cells are heterogeneous, with specialized properties depending on the organ they reside in and their order in the vascular tree. For instance, angiogenesis is induced during a variety of pathologies, but during physiological conditions in adults it is only provoked in ovaries (during ovarian cycling), in the placenta (during placental development), and in muscles (during exercise) (40). Mechanical stretch and increased vessel wall tension during exercise is reported to result in increased capillary supply due to increased local levels of vascular endothelial growth factor (VEGF), hypoxia-inducible factor-1β, and matrix metalloproteinase-2 (41). Thus, a clear difference between striated muscle and other implantation sites investigated for islet vascular densities is that muscle endothelium is programmed to proliferate during muscle overload and concomitant hypoxia (40).

It has recently been reported that leukocytes contribute to angiogenesis occurring during remodeling and healing of muscle tissue (20,21,42). Hypoxic mediators are known to recruit leukocytes (43,44) and might account for the leukocyte–endothelial cell interactions observed early after islet implantation in this study but not after sham injections. The number of recruited leukocytes decreased with time after transplantation, which parallels islet revascularization and associated reduced hypoxia (45). No revascularization of islets 3–5 days after transplantation could be detected when Gr-1–positive leukocytes had been depleted in recipients, indicating a crucial role of these cells in the angiogenic switch. Gr-1 is reported to be expressed on neutrophils as well as inflammatory monocytes but not on the monocyte population involved in tissue repair (20,42,46). The Gr-1–positive leukocytes recruited to the site of engraftment in the current study that were responsible for initiation of revascularization are therefore most likely neutrophils, while the Gr-1–negative monocyte population that has been shown to be involved in tissue remodeling (21) and tumor angiogenesis (42) does not seem to contribute at these early time points. The mechanism by which neutrophils induce angiogenesis might be release of VEGF—either from intracellular storages or through secretion of matrix metalloproteinase-9, which has been shown to contribute to angiogenesis through digestion of extracellular matrix and release of immobilized VEGF (42).

This study presents a novel paradigm for islet transplantation whereby neutrophils recruited to the site for engraftment were crucial for the functionally restored intraislet blood perfusion following transplantation to striated muscle. In the clinical setting, this site proved to promote islet revascularization and also permitted longitudinal graft monitoring by high-resolution magnetic resonance imaging.

This study was supported by grants from the Swedish Medical Research Council, the Juvenile Diabetes Research Foundation, the Swedish Diabetes Foundation, the Wallenberg Foundation, The Royal Swedish Academy of Sciences, Vinnova SAMBIO 2006, the Magnus Bergvalls Foundation, the Swedish Juvenile Diabetes Fund, the Swedish Society for Medical Research, the Lars Hiertas Foundation, the Harald and Greta Jeanssons Foundation, the Gunvor and Josef Anérs Foundation, and the family Ernfors foundation.

No potential conflicts of interest relevant to this article were reported.

G.C. conceived, designed, and performed the experiments, analyzed data, and wrote the manuscript. J.H. performed the experiments. L.J. conceived, designed, and performed the experiments and analyzed data. C.R. conceived and designed the experiments. H.A. performed the experiments. J.C.-C. performed the experiments. R.S. performed the experiments. J.P. performed the experiments. O.K. conceived and designed the experiments and wrote the manuscript. P.-O.C. conceived and designed the experiments and wrote the manuscript. M.P. conceived, designed, and performed the experiments, analyzed data, and wrote the manuscript.

We thank Professor Michael Perry, University of New South Wales, Sydney, Australia, for revising the language of the manuscript. Eva Törnelius, Lisbeth Sagulin, Astrid Nordin, Johanna Svensson, Anders Ahlander and Marianne Ljungkvist, all at the Department of Medical Cell Biology, Uppsala University, Uppsala, Sweden, are acknowledged for their skilled technical assistance.