Intraislet Endothelial Cells Contribute to Revascularization of Transplanted Pancreatic Islets

Flk-1 (kinase insert domain-containing receptor, vascular endothelial growth factor receptor 2) heterozygote mice with lacZ-tagged endothelial cells (Flk-1^wt/lacZ) (19), C57BL/6 mice, and NOD-SCID mice were obtained from The Jackson Laboratory (Bar Harbor, ME). To identify the lacZ insert in the Flk-1 gene, the offspring of Flk-1^wt/lacZ mice (stock no. 002938; background strain C57BL/6) were genotyped by PCR using the following primers to detect the LacZ insertion (5′ primer: ATC CTC TGC ATG GTC AGG TC; 3′ primer: CGT GGC CGT ATT CAT TTC) and the wild-type locus (5′ primer: CAA ATG TTG CTT GTC TGG TG; 3′ primer: GTC AGT CGA GTG CAC ATG TT). Mouse islets from Flk-1^wt/lacZ donors and human islets were transplanted into an immunodeficient NOD-SCID mouse model from The Jackson Laboratory (for additional information see http://jaxmice.jax.org/jaxmice-cgi/jaxmicedb.cgi?objtype=pricedetail&stock=001303).

Islets were isolated from Flk-1^wt/wt mice by dissection of the splenic portion of the pancreas followed by collagenase P digestion (Roche Molecular Biochemicals, Indianapolis, IN) as previously described (20). To increase the yield of islets isolated from Flk-1^wt/lacZ mice, 3 ml collagenase P in Hank’s buffered saline (0.6 mg/ml) was first directly infused into the pancreas through the bile duct. Groups of two pancreata were then digested in 6.7 ml collagenase P (0.6 mg/ml) for 4–5 min at 37°C using a wrist-action shaker. Islets were handpicked under microscopic guidance and washed three times with 10 mmol/l PBS containing 1% mouse serum. Finally, 200–240 islets were suspended in 30 μl of the same solution and transplanted into mouse recipients immediately following the isolation procedure.

Human islets were obtained through the Juvenile Diabetes Foundation Human Islet Distribution Program and from Dr. David M. Harlan at the Transplantation and Autoimmunity Branch of the National Institute of Diabetes and Digestive and Kidney Diseases of the National Institutes of Health. After isolation, human islets were shipped in CMRL media by overnight courier to Vanderbilt and cultured for additional 24 h in CMRL media, 95% CO₂/5% O₂ at 37°C. After culture, 500–2,000 islets were transplanted into NOD-SCID mice (∼48 h after islet isolation).

The mice were anesthetized by an intraperitoneal injection of sodium pentobarbital (50 mg/kg body wt; Abbott Laboratories, North Chicago, IL). After adequate anesthesia, the left flank was shaved, prepped, and draped in sterile fashion. With a left flank incision, the left kidney was identified, exposed, and irrigated with saline. The islet suspension (30 μl) was injected between the capsule and renal parenchyma of the left kidney using a 23-gauge butterfly needle. After withdrawal of the needle, the insertion point was cauterized and the wound was closed with subcutaneous sutures (Prolene, size 7-0 with cutting needle; Ethicon, Somerville, NJ) and skin staples (Autoclips, 9-mm size; Clay Adams, Parsippany, NJ).

Adult pancreata, as well as kidneys bearing islet transplants (3–5 weeks after transplantation), were dissected in ice-cold 10 mmol/l PBS and fixed in freshly prepared 4% paraformaldehyde (Electron Microscopy Sciences, Hatfield, PA)/100 mmol/l PBS for 1.5 h on ice. Following fixation, the tissues were washed 4–6 times with 100 mmol/l PBS over a period of 2 h and then equilibrated in 30% sucrose/10 mmol/l PBS overnight at 4°C. The tissues were cryopreserved in optimum cutting temperature compound (VWR Scientific Products, Willard, OH) at −80°C, and 10- or 60-μm sections were mounted on charged slides.

In the case of the whole pancreas, the fixed tissue was permeabilized twice for 30 min at room temperature in permeabilization solution (2 mmol/l MgCl₂, 0.01% sodium deoxycholate, and 0.02% Nonidet P-40 in 10 mmol/l PBS). β-Galactosidase activity was detected by incubating the tissue in staining solution (2 mmol/l MgCl₂, 5 mmol/l K ferricyanide, 5 mmol/l K ferrocyanide, 100 mmol/l Tris, pH 7.3, and 1 mg/ml X-gal [5-bromo-4-chloro-3-indolyl-β-d-galactopyranoside; Research Products International, Mt. Prospect, IL]) overnight in the dark at room temperature. Tissue was then rinsed with 10 mmol/l PBS, postfixed in fresh ice-cold 4% paraformaldehyde/10 mmol/l PBS for 1 h at 4°C, and rinsed three times with 10 mmol/l PBS. Whole-mount images of the pancreas were collected on an Olympus SZX9 microscope with an Olympus pm-C35 camera using Kodak Elite Chrome 160T film.

β-Galactosidase activity was also detected on 10-μm cryosections prepared as described above. The cryosections were postfixed with 0.2% glutaraldehyde/1% paraformaldehyde (Electron Microscopy Sciences) for 15 min at room temperature, washed three times for 5 min with 10 mmol/l PBS, and permeabilized with permeabilization solution for 10 min at room temperature. Sections were incubated with X-gal staining solution in a humidified chamber overnight at 37°C, rinsed three times with 10 mmol/l PBS, and mounted with AquaPoly/Mount (Polysciences, Warrington, PA).

Ten-micron cryosections were permeabilized in 0.2% Triton X-100 for 10 min at room temperature, blocked with 5% normal donkey serum (Jackson ImmunoResearch Laboratories, West Grove, PA) for 1.5 h, and then incubated with primary antibodies overnight at 4°C. Secondary antibodies were applied to the tissue sections for 1 h at room temperature. Both primary and secondary antibodies were diluted in 10 mmol/l PBS containing 1% BSA and 0.1% Triton X-100. Digital images of the 10-μm cryosections mounted with AquaPoly/Mount were acquired with a MagnaFire digital camera (Optronics, Goleta, CA) connected to an Olympus BX-41 fluorescence microscope.

Freshly isolated islets were attached to MatTek dishes (cat. No. P35G-0-14-C; MatTek, Ashland, MA) precoated with Cell-Tak adhesive (Becton Dickinson Labware, Bedford, MA) and fixed in 4% paraformaldehyde/10 mmol/l PBS for 25 min at room temperature. The fixation was followed by three 30-min washes in 10 mmol/l PBS and 3-h permeabilization with 0.3% Triton X-100/10 mmol/l PBS. The islets were blocked with 5% normal donkey serum/0.15% Triton X-100/10 mmol/l PBS overnight at 4°C and then equilibrated in antibody dilution buffer twice for 20 min at room temperature. The primary and secondary antibodies were diluted in 1% BSA/0.2% Triton X-100/10 mmol/l PBS, and the incubations were carried out for 24 h at 4°C. Sixty-micron cryosections of the islet transplants under the kidney capsule were permeabilized, blocked, and stained under the same conditions as isolated islets. The islets and tissue sections were mounted with AquaPoly/Mount (Polysciences). Samples were subjected to optical sectioning using a Zeiss LSM410 or LSM510 META confocal laser scanning microscope. Digital images were analyzed and three-dimensionally reconstructed using MetaMorph 5.0 software (Universal Imaging, Downington, PA).

Rat anti-mouse CD31 (platelet endothelial cell adhesion molecule [PECAM]-1) (1:200) and mouse anti-human CD31 (1:50) monoclonal antibodies were from BD Biosciences Pharmigen (San Diego, CA). Guinea pig anti-human insulin IgG (1:1,000) was from Linco Research (St. Charles, MO), sheep anti-somatostatin IgG (1:1,000) was from American Research Products (Belmont, MA), rabbit anti–Flk-1 IgG (1:500) was a gift from Rolf Brekken at The Hope Heart Institute, and rabbit anti–β-galactosidase IgG (1:5,000) was from ICN Pharmaceuticals (Costa Mesa, CA). The antigens were visualized using appropriate secondary antibodies conjugated with Cy2, Cy3, and Cy5 fluorophores from Jackson ImmunoResearch Laboratories. Secondary antibodies were used at concentrations recommended by the manufacturer.

At 3 to 5 weeks after transplantation, mice transplanted with Flk-1^wt/lacZ islets were anesthetized with sodium pentobarbital (80 mg/kg body wt; Abbott Laboratories). Fluorescein isothiocyanate–conjugated tomato lectin (Lycopersicon Esculentum, 1 mg/ml; Vector Laboratories) was injected into the jugular vein (0.1 ml/mouse) and allowed to circulate for 3 min, after which animals were killed. Kidneys bearing islet transplants were dissected and preserved for cryosectioning as described above. The permeabilization step was omitted because detergent destabilizes lectin binding. Ten-micron cryosections were blocked with 5% normal donkey serum/10 mmol/l PBS/1 mmol/l Ca²⁺ for 15 min. Incubations with primary and secondary antibodies were carried out at room temperature for 1 h and 30 min, respectively. Both primary and secondary antibodies were diluted in 10 mmol/l PBS containing 1% normal donkey serum and 1 mmol/l Ca²⁺. Washes were performed using the same buffer. Sections were mounted with Vectashield mounting medium (Vector Laboratories) and subjected to optical sectioning using a LSM510 META confocal laser scanning microscope.

Even though islet isolation severs arterial and venous connections, isolated islets retain their capillary network (Fig. 1A). Therefore, we asked whether these intraislet endothelial cells contribute to the revascularization of transplanted islets. To follow the fate of the intraislet endothelial cells, we used a model in which endothelial cells are tagged with lacZ (knock-in of lacZ to the Flk-1 locus termed Flk-1^wt/lacZ). LacZ encodes the β-galactosidase enzyme. Figure 1B shows prominent X-gal staining (reflecting β-galactosidase activity) of islets in the whole-mount Flk-1^wt/lacZ pancreas. Pancreatic sections in Figs. 1C and D demonstrate that lacZ expression recapitulates expression of Flk-1. Similar to Flk-1 expression, there was a greater density of lacZ⁺ capillary structures in the islets compared with exocrine tissue, reflecting the higher vascularity of islets (Figs. 1C and D).

Immediately after islet isolation, wild-type islets (Flk-1^wt/wt) were transplanted into wild-type animals (Flk-1^wt/wt, n = 3) (Figs. 1E–G) and mice that contain the endothelial lacZ tag (Flk-1^wt/lacZ, n = 4) (Figs. 1H–J). At the same time, murine islets containing tagged endothelial cells (Flk-1^wt/lacZ) were transplanted into mice that do not carry the endothelial cell marker (n = 10) (Figs. 1K–M). We used NOD-SCID mice as the recipient of Flk-1^wt/lacZ islets to avoid a possible immune reaction to endothelial cells expressing lacZ.

Kidneys bearing transplanted islets were retrieved 3–5 weeks after transplantation, a time in which the revascularization is completed, as shown by previous studies (9–16). In all three types of transplants, the islets were vascularized as detected by Flk-1 expression in vascular structures within and surrounding the islet graft (Fig. 1E, H, and K). As shown by detection of β-galactosidase activity (Figs. 1J and M), both donor and recipient endothelial cells were found within the islet graft area positive for insulin (Figs. 1I and L). Occasionally, there were a few lacZ⁺ intraislet endothelial cells migrating further away from the transplant into kidney cortex (Fig. 1M). These data indicate that intraislet endothelial cells survive and possibly contribute to the revascularization process.

To determine whether intraislet endothelial cells participate in revascularization and to access the structure and composition of blood vessels in the revascularized grafts, 60-μm sections of the islet grafts were labeled for the mouse endothelial marker PECAM-1, which is ubiquitously expressed on the surface of all (both donor and recipient) endothelial cells. The sections were colabeled for lacZ-encoded β-galactosidase, which is only expressed by the endothelium of donor Flk-1^wt/lacZ islets (Fig. 1K, L, and M) and unlike PECAM-1 has a more cytoplasmic localization. In both mouse and human native islets and islet grafts, mouse PECAM-1 or human CD31 and Flk-1 are coexpressed in islet microvasculature (data not shown). Mounted sections were then subjected to optical sectioning using a laser scanning confocal microscope. The three-dimensional reconstruction of optical sections through the islet grafts (Fig. 2) indicated the existence of two types of blood vessels in the revascularized islet graft: 1) capillaries formed predominantly of either donor or recipient endothelial cells directly connected to each other and 2) chimeric blood vessels formed from a mixture of donor and recipient endothelial cells (online data supplement 1C and D [available at http://diabetes.diabetesjournals.org]). By examining optical sections of the islet grafts in three dimensions, both donor and recipient endothelial cells were found to be components of tubular structures consistent with vessels that traversed throughout the islet graft (Figs. 2E and F and online data supplement 1C and D). To estimate the contribution of donor and recipient endothelial cells to the graft revascularization, we used MetaMorph software and calculated the volume of PECAM-1⁺ and β-galactosidase⁺ endothelial cells in the insulin⁺ graft area. This calculation was applied to four different grafts, and three to six fields were examined per each graft. These data suggested that as much as 40 ± 3% (n = 18, range 18–68%) of endothelial cells in the revascularized graft originated from the donor islets.

Human islets, transplanted under the kidney capsule of NOD-SCID mice (n = 4), were analyzed for the presence of the donor and recipient endothelial cells using species-specific antibodies to the endothelium-specific marker CD31. Sixty-micron sections of the human islet grafts were stained for mouse CD31 (PECAM-1) and human CD31. Mounted sections were then subjected to optical sectioning using a laser scanning confocal microscope. The three-dimensional reconstruction of optical sections through the islet grafts (Fig. 3) indicated that human intraislet endothelial cells, similar to the murine islets, survive after transplantation. In contrast to mouse intraislet endothelial cells, the human intraislet endothelial cells did not form chimeric blood vessels. Moreover, examination of human islet grafts in three dimensions revealed fewer possible connections of human endothelial cells with the host vasculature (Figs. 3D and E) than was seen in mouse islet grafts. The human intraislet endothelial cells also appeared to more readily migrate into the kidney cortex (Fig. 3F) than was observed in transplants of murine Flk-1^wt/lacZ islets (Fig. 1M).

Even though intraislet endothelial cells survived transplantation and were a part of vascular-like structures within the revascularized islet grafts, it was not known whether they integrated into a functional blood vessel network of the revascularized graft. To address this question, mice that received transplants of Flk-1^wt/lacZ islets were infused with fluorescein isothiocyanate–conjugated tomato lectin (n = 4), which binds with high affinity to the surface of mouse endothelial cells (21). Lectin infusion, a well-established technique, has been widely used in the area of angiogenesis (22). For example, many investigators (22) feel that it is the technique of choice for identifying functional blood vessels since the only way lectin can reach the endothelial cells is via the circulation. Furthermore, it has been shown previously that L. Esculentum lectin, which is utilized in our studies, binds uniformly to the luminal surface of murine endothelial cells in arteries, veins, and capillaries and is colocalized with CD31 in functional blood vessels (22). We confirmed the uniformity of L. Esculentum lectin binding in both pancreas and kidney. Optical sections of the lectin-perfused mouse tissue specimens acquired with a laser scanning confocal microscope show that the lectin signal is localized in the luminal lining of blood vessels (online data supplement 2A and B).

The lectin-perfused islet grafts were additionally analyzed for expression of insulin and β-galactosidase by immunocytochemistry and laser scanning confocal microscopy (Fig. 4) (online data supplement 3A–F). The series of optical sections was acquired at a 0.5-μm interval in the axial (z) dimension and an appropriate pinhole setting to alleviate any concerns about overlapping signals from the layers of the specimen above and below the given focal plane. Individual optical sections of the islet grafts demonstrate that in the same focal plane, β-galactosidase, which marks intraislet endothelial cells, is colocalized with lectin, thus proving unequivocally the functionality of these β-galactosidase–positive blood vessels. The presence of cells double positive for lectin and β-galactosidase indeed proves that blood flows through donor-derived blood vessels in revascularized mouse islet grafts (Figs. 4B and D). Immunocytochemistry of the lectin-perfused specimens required several modifications to avoid lectin leaching during the staining procedure (see research design and methods) and underestimated the number of β-galactosidase⁺ endothelial cells (especially in cells with lower β-galactosidase expression). This was based on a comparison of β-galactosidase⁺ cells in consecutive sections stained with either β-galactosidase antibody (modified immunocytochemistry procedure) or using an enzymatic reaction of β-galactosidase with X-gal substrate. Because there were fewer β-galactosidase⁺ cells detected by immunocytochemistry compared with enzymatic reaction, we did not feel it was appropriate to estimate the ratio of β-galactosidase⁺ cells to cells double positive for β-galactosidase⁺ and lectin⁺. We did find a few cells positive for β-galactosidase and negative for lectin, and, for example, one such cell appeared in Fig. 4B (arrow) (online data supplement 3C and F). These β-galactosidase⁺/lectin⁻ cells could be either intraislet endothelial cells that survived the transplantation but did not establish a lumen or proliferating donor-derived endothelial cells.

In the human islet grafts perfused with L. Esculentum lectin, we did not find human endothelial cells double positive for human CD31 and lectin. While binding of L. Esculentum lectin to mouse endothelium is well documented (22), it is unknown if this lectin binds to human endothelial cells. The human CD31⁺ cells remained negative for the lectin even when the sections of human islet grafts were subsequently stained with L. Esculentum lectin (a standard method to assess if a lectin binds to endothelial cells). These observations indicate that human intraislet endothelial cells may lack the N-acetyl-d-glucosamine olygomer moieties that are recognized by L. Esculentum lectin. The formation of tubular structures by human endothelial cells is evident from the three-dimensional projections in Fig. 3; however, it is still possible that they were incompletely connected to the recipient’s vasculature.

Transplanted pancreatic islets, unlike transplanted solid organs, must revascularize to survive and function, and this process of angiogenesis is likely an important factor that ultimately determines whether an islet transplant secretes sufficient insulin to reverse diabetes. The reasons for the current requirement of islets from two pancreata for islet transplantation to reverse diabetes are uncertain, but this limitation is a major obstacle to advancing islet transplantation. While continued focus on improving immunosuppression after islet transplantation is crucial, a better understanding of the molecular events of islet engraftment or revascularization could lead to a dramatic increase in islet survival and improve islet transplantation as a therapy for type 1 diabetes. In contrast to the current thinking that revascularization is solely a function of the transplant recipient (10,13,16–18), our results show that intraislet endothelial cells from the donor survive after transplantation and play an important role in the revascularization process. The survival of intraislet endothelial cells was recently reported by investigators (19) using the transgenic Tie2-lacZ mouse model, but this report did not examine whether these cells can contribute to functional vasculature.

Three-dimensional reconstructions of optical sections through islet grafts allowed a more detailed examination of blood vessels within the revascularized mouse and human islet grafts. In mouse Flk-1^wt/lacZ islet transplants, donor and recipient endothelial cells cooperated effectively during the process of angiogenesis in that some vessels were lined with donor or recipient endothelial cells, while other vessels were a chimera of donor and recipient endothelial cells. In contrast, there was less interaction between human intraislet endothelial cells and recipient mouse endothelial cells. Examination of optical sections in three dimensions did not demonstrate chimeric blood vessels in the human islet grafts and revealed fewer possible connections of mouse and human endothelial cells. The limited interaction of mouse and human endothelial cells may reflect species differences in endothelial cells. This may also explain the relatively extensive migration of human endothelial cells into the kidney cortex, which was only occasionally found in mouse Flk-1^wt/lacZ islet transplants. In addition, human islets (unlike those of mice) were procured from brain-dead donors and cultured longer, which may have influenced their intraislet endothelial cell biology.

Even though the presence of vascular connections between capillaries derived from donor and recipient endothelial cells seemed possible based on three-dimensional projections of islet grafts, it does not prove that these vessel-like structures are functional. Infusion of endothelium-binding lectin in the mouse islet–transplant model demonstrated that blood flows through donor-derived capillaries and, thus for the first time, showed that the intraislet endothelial cells have a capacity to integrate into the functional vasculature within the revascularized graft. These results have several implications for the handling of pancreatic islets destined for transplantation. Pancreas procurement and islet isolation procedures must not only strive to maximize the health of islet endocrine cells, but must now also consider the effects on endothelial cell health and survival. In contrast to the original Edmonton procedure, many human islet transplantation centers currently culture pancreatic islets for several days before transplantation; the effect of such culture on endothelial cell health and survival after transplantation is unknown and warrants further study. The model system in which intraislet endothelial cells are tagged with lacZ should allow one to address such questions. In addition, our transplantation studies with both mouse and human islets did not utilize immunosuppression; the effect of immunosuppressive agents on endothelial cell health and islet revascularization is not known and warrants further study.

Since islet revascularization may be a limiting factor in islet survival and since transplanted islets have reduced vessel density compared with islets in the pancreas (10,13,23), the capacity of intraislet endothelial cells to integrate into functional graft vasculature suggests a number of avenues of investigation to enhance the revascularization process. For example, can intraislet endothelial cells be activated or primed ex vivo (pretransplantation) to accelerate angiogenesis? Will inclusion of exogenous endothelial cells with transplanted islets promote islet engraftment and revascularization? Emerging evidence supports a direct interaction between pancreatic islet cells and endothelial cells. Melton et al. (24) recently suggested that the developing aorta was essential for the initiation of endocrine cell differentiation during pancreas development. These investigators also found that transgenic overexpression of the angiogenic factor, vascular endothelial growth factor (VEGF), under the control of a pancreas-specific promoter from the pancreatic duodenal homeobox-1 (PDX-1) gene increased islet size. When combined with the current results, this raises the possibility that intraislet endothelial cells and the newly formed islet vasculature may nurture transplanted islet cells in addition to the reestablishment of islet blood flow. Future studies are needed to address this hypothesis.

Because of the shortage of organ donors and the large number of patients with type 1 diabetes, investigators are developing alternative sources of insulin-producing cells such as stem cells and genetically engineered cells. Like transplanted pancreatic islets, such cell transplants will likely need to become vascularized. Our studies would suggest that inclusion of exogenous endothelial cells with such insulin-producing cells may facilitate the survival and function of these surrogate islet cells. A better understanding of how transplanted cells survive and become vascularized will also be relevant to the development of cell transplantation therapy for other diseases.

Our studies also give rise to a number of questions about the molecular events of islet engraftment. For example, we transplanted islets beneath the renal capsule, whereas in human transplantation, islets are infused into the portal vein and embolize in the liver. Whether the angiogenic and revascularization processes are identical in the liver and beneath the renal capsule is unknown. What signals initiate angiogenesis after islet transplantation? Are islet cells the source of such signals? What are the steps of islet engraftment and capillary sprouting, from breakdown of the basement membrane of preexisting vessels within and surrounding the islet graft to endothelial cell migration? The current findings and future research to better understand the molecular events of islet engraftment should enhance islet revascularization, improve islet survival after transplantation, reduce the number of pancreatic islets required to reverse diabetes, and improve the outcome of islet transplantation in type 1 diabetes.

This study was supported by a Merit Review Award from the VA Research Service, research grants from the National Institutes of Health and the Juvenile Diabetes Research Foundation International, and the Vanderbilt Diabetes Research and Training Center (National Institutes of Health DK20593).

The authors thank Drs. David M. Harlan and Boaz Hirshberg at the National Institutes of Health for providing some of the human pancreatic islets and for helpful discussions. We also thank Ninche Alston for technical assistance.