The pancreas is an endodermally derived organ, consisting of two morphologically distinct tissues with disparate functions, the exocrine pancreas and the endocrine islets. The islets consist of five endocrine cell subtypes, α, β, δ, ε, and PP cells secreting glucagon, insulin, somatostatin, ghrelin, and pancreatic polypeptide hormones, respectively. The highly vascularized nature of the islets is necessary for proper blood glucose homeostasis by accurately sensing changes in blood chemistries. In addition to acting as conduits for blood, mounting evidence suggests that blood vessels also influence cell differentiation and organogenesis (Gittes, 2009; Lazarus et al., 2011; Shah et al., 2011).
During development, the embryonic pancreas is known to pass through three stages. The early stage is when the endoderm evaginates to initiate pancreatic morphogenesis (Gittes and Rutter, 1992). The second stage involves epithelial branching morphogenesis, including the separation of islet progenitors that lose their attachments to the basement membrane (Argent et al., 1992). The third and final stage begins with the formation of acinar cells at the tips of the ductal structures, with the later development of zymogen granules containing digestive enzymes(Kolacek et al., 1990). As the embryonic pancreas progresses through these stages, it receives signals from different tissues, including the nearby dorsal aortae. When the dorsal aortae were removed from developing Xenopus embryos, there was a resulting absence of pancreatic endocrine development (Lammert et al., 2001). Later, as the embryonic pancreas develops in utero and as β-cells start to aggregate to form islets, β-cells begin to express high levels of vascular endothelial growth factor (VEGF) to attract endothelial cells (Lammert et al., 2003). β-cells deficient in VEGF-A form islets with less capillaries, and experimental overexpression of VEGF-A has improved islet graft vascularization (Lammert et al., 2003; Zhang et al., 2004; Lai et al., 2005; Brissova et al., 2006; Kamba et al., 2006). These data underscore the critical role that the intraislet capillary network plays in the overall function of the islets (Akirav et al., 2011).
The vascular network within the mature islet is known to be tortuous and 5–10 times denser than the surrounding exocrine tissue (Jansson and Carlsson, 2002). Early attempts at visualizing the pancreatic vasculature entailed corrosion casting, followed by scanning electron microscopy (Gorczyca et al., 2010). Also in vivo light microscopy (McCuskey and Chapman, 1969; McCuskey, 1997) and using a variety of dye injections including Berlin Blue, Evans blue, and Indian Ink (Henderson and Daniel, 1979; Fraser and Henderson, 1980) have been used. More recently, live in vivo imaging of pancreatic islet blood-flow within the pancreas has also been achieved with great success (Nyman et al., 2008). However, that elegant study was focused more on blood flow, with a less detailed analysis of the three-dimensional (3D) vascular architecture (Figs. 2, 3 of Nyman et al., 2008). Here we describe a relatively simple method to visualize the static 3D architecture of islets in the adult pancreas. We studied the relationship between islet morphology and capillary network anatomy. We also used morphometric analysis to determine the total length of the intraislet vasculature, the number of branch points, and the number of islet entry and exit points. Nonobese diabetic (NOD) mouse pancreas, a murine model that develops spontaneous autoimmune diabetes mellitus resembling that of human Type 1 diabetes (Pozzilli et al., 1993), were compared to wild-type CD1 pancreas.
The University of Pittsburgh Institutional Animal Care and Use Committee approved all animal experiments described herein.
Ten-week-old female CD1 wild-type and 8-, 10-, and 12-week-old female NOD pancreases were harvested, fixed in 4% paraformaldehyde for 2 hr at 4°C. Pancreas was then minced manually into small pieces, ranging from 400 to 2,000 μm thickness, and then 100 pancreas pieces placed per well (three to four wells) in a 96-well plate. Nearly 1,000 μL of PBS was added per well with pancreas pieces, washed on a rotator in PBS for 10 min three times at room temperature (RT). Washed in methanol for 15 min at RT; serially five times. Blocked with 1,000 μL of 10% Normal Donkey serum (NDS) in 0.1% PBST for 45 min at RT. Incubated with primary antibody (insulin guinea pig anti-swine 1:250 or rabbit anti-glucagon 1:500 and CD-31 rat monoclonal anti-mouse 1:100, see Whole-mount Immunohistochemistry) in 1% NDS in 0.1% PBST at 4°C on a rotator overnight. Washed with 1,000 μL 0.1% PBST on a rotator at RT once an hour for 5 hr. Incubated with secondary antibody in 1% NDS in PBST in 4 degree on rotator overnight. Washed six times with PBST on rotator at RT for 10 min. Mounted on chamber slides with whole-mount media and coverslip (El-Gohary et al., 2012).
About 21 g of polyvinyl acid in 42 mL of glycerol and 52 mL of distilled water along with a few crystals of sodium azide. Then added 106 mL of Tris (0.2M, pH = 8.5) and stirred with low heat for a few hours or until reagents dissolved. Clarified mixture by centrifugation at 5,000g for 15 min and aliquoted the solutions and stored at 4°C.
An upright Axio Imager Z1 microscope with AxioCam MRc5 was used to take whole-mount images of the minced adult mouse pancreatic tissues. Capturing the images with AxioVs40 V22.214.171.124 software provided the 3D images of the pancreatic islets. The pictures were generated by overlay of the colors followed by merging of all color channels into one.
To create a 3D reconstructed image, an inverted Olympus Fluoview 1000 confocal microscope was utilized to confocally image the whole-mounted pancreas tissues, collecting multiple Z-sections (each layer 3-μm apart) of pancreas tissues immunostained with primary antibodies against insulin and CD31 (an endothelial cell marker, also known as antiplatelet/endothelial cell adhesion molecule-1 [PECAM]), followed by secondary antibodies conjugated to fluorescent dyes. The multiple image layers obtained were then linked to an image analysis computer that was configured with image tracing software (Stereoinvestigator, Microbrightfield). Using the Stereoinvestigator program, each individual image pancreas layer was traced, with the first layer representing the bottom/inferior region of the pancreas structure. All the Z-sections were traced serially through the entire pancreas structure, where different colors were used to trace individual structures such as the islet (including insulin and CD31). Once all the Z-sections were traced, a crude 3D stack was generated and subsequently converted to a solid 3D structure (Sims-Lucas et al., 2009).
Skeletonizing the Vasculature of the Adult Pancreas
Adult female CD1 wild-type and NOD mice were injected with fluorescein Lycopersicon Esculentum (tomato) lectin intravenously through tail vein (to visualize blood flow), while the mice were under anesthesia. The animal was then sacrificed 5 min after injection and the pancreas removed. Following fixation the pancreas was labeled with insulin using anti-insulin antibody conjugated to fluorescein-tagged secondary antibody. Individual islets were then imaged using a confocal microscope at 2-mm intervals. These stacks of images were then analyzed with the imaging software Imaris (Bitplane, South Windsor, CT). Using this software, the islet was then skeletonized using the filament tracer prompt of Imaris as previously described (Sims-Lucas et al., 2011). Intraislet capillary length refers to the total distance of intraislet vasculature that traverses an islet (insulin positive area). Segment refers to the intraislet capillary distance between each capillary branch point. Branch point refers to the intraislet vascular point where the capillary divides into different branches or segments.
Female NOD mice were obtained from the Jackson Laboratory (Bar Harbor, ME) and maintained under pathogen-free conditions. NOD mice were screened for hyperglycemia at 12 weeks of age and diagnosed with diabetes when two consecutive glucose levels were >200 mg dL−1. Glucose levels were measured in whole blood from the tail vein using a Glucometer.
Primary antibodies used insulin guinea pig anti-swine 1:250 (Dako, Carpinteria, CA), CD-31 rat monoclonal anti-mouse 1:100 (BD Pharmingen, CA), and rabbit anti-glucagon 1:500 (Linco). The fluorophores for secondary antibody used were: Cy3 ant-rat (Jackson) 1:200, FITC anti-rat (Jackson) 1:200, Dylight Cy5 anti-guinea pig (Jackson) 1:500, and Fluorescein Lycopersicon Esculentum (tomato) Lectin (Vector Laboratories, CA).
Initial immunohistochemical imaging of 6-μm sections showed the pancreatic and vascular architecture for comparison with the 3D images (Fig. 1). Whole-mount 3D double immunohistochemistry of the adult pancreas for the endothelial marker CD31 and insulin demonstrate a dense vascular network within the islets (Figs. 1, 4). The thickness of the imaged pancreatic fragments ranged from 400 to 2,000 μm. We thus used serial analysis of these fragments to scan through the entire mouse pancreas and quantified the number of islets detected in the whole pancreas (Fig. 2). We compared the total number of islets in NOD mouse pancreases at different ages, 8 weeks (1,845 islets ± 189.4), 10 weeks (1,158 islets ± 92.9), and 12 weeks of age (1260 islets ± 71.8), demonstrating a significant decrease in total number after 8 weeks of age (Fig. 2, n = 3). The adult mouse pancreas was also divided into different regions: gastric lobe, duodenal head, and tail and the number of islets quantified. The gastric lobe of the pancreas in a 10-week-old wild-type CD1 pancreas averaged 110 islets (SEM ± 15.4), duodenal head 453.8 islets (SEM ± 45.7), and the tail of the pancreas 686 islets (SEM ± 54.3) (Fig. 2, n = 6). More islets were documented in the pancreatic duodenal head of the NOD mice at different ages, NOD 8 weeks (869 islets SEM ± 270.6), NOD 10 weeks (549 islets SEM ± 139.6), NOD 12 weeks (634 islets SEM ± 60.4) compared to the pancreas tail, NOD 8 weeks (793 islets SEM ± 56.5), NOD 10 weeks (489 islets SEM ± 76), NOD 12 weeks (490 islets SEM ± 98.6). However, the reverse was seen in 10-week wild-type CD1 pancreas where more islets were documented in the pancreas tail (686 islets SEM ± 94.1) compared to the pancreatic duodenal head (453 islets SEM ± 112). In the context of the NOD mouse pancreas, as insulitis progresses, the total number of islets in the whole pancreas, as well as in the pancreatic gastric lobe, decreased from 182 islets (SEM ± 73.3) in an 8-week old to 120 islets (SEM ± 53.3) and 135 islets (SEM ± 29.1) in 10- and 12-week old, respectively (Fig. 2).
We used this whole-mount technique to analyze the detailed anatomical vascular variations that exist between different islets. We determined the total intraislet capillary length, the total number of branch points, number of segments and number of entrance/exit points in and out of the islet for the islet vasculature of each islet (Fig. 3). This morphometric analysis was performed in wild-type and diabetic NOD islets, comparing these different parameters (Fig. 3). There was a trend toward a decrease in the average intraislet vasculature length in diabetic NOD islets, 2.41 ± 1.1 mm, compared to wild-type islets, 2.83 ± 1.0 mm. Similarly, the intra-islet capillary volume trended to be lower in the NOD diabetic islets. Furthermore, we observed significantly fewer intraislet vascular segments and branch points in the NOD diabetic islets, compared to wild type islets (P = 0.016 and P = 0.016, respectively).
With no reliable immunohistochemical marker available to distinguish between vessels carrying blood flow toward (feeding/arteriole) the islet or away (exiting/venule) from the islet, the vessel caliber was used as a “best estimate” of whether the vessel was arterial or venous. A large caliber vessel indicated a venule, or blood flowing away from the islet (Fig. 4, arrowheads), and a thin caliber vessel indicated an arteriole or blood flowing towards the islet (Fig. 4, arrows). To better visualize the complex microcapillary network of the islets in the adult pancreas, we used confocal Z-stack imaging (Supporting Information Movie 1). Most islets had two to four vascular entrance/exit points, with the size of an islet loosely correlating with the number of vascular entry and exit points (Fig. 3).
We were able to survey the vascular anatomy of the whole pancreas, with morphology of the intraislet capillary varying from islet to islet. We found that the vascular anatomy of the islets varied depending on the size of the islet and the location within the pancreas in relation to large vessels (Fig. 4).
The morphological development of the pancreas is dictated by its two major functions, which are the production of digestive enzymes by the exocrine tissue, and the production of metabolically active hormones by the endocrine tissue. These two tissues exist together within the pancreas despite their contrasting morphology and function. Some have even described it as “two organs in one” because of the distinct function and organization of these two tissues (exocrine and endocrine) within the pancreas. These two tissues are functionally related and connected through the vascular anatomy of the pancreas (Figs. 1, 2A), (Henderson and Daniel, 1979; Fraser and Henderson, 1980; Henderson et al., 1981; von Schonfeld et al., 1994). The endocrine pancreas, which comprises of only 1–2% of the adult pancreatic mass, was first noticed in 1869 by Langerhans as histologically distinct regions within the rabbit pancreas, and later described in 1882 by Kühne and Lea as discrete vascular regions in a living rabbit pancreas, closely resembling the renal glomeruli (Henderson et al., 1981). These were later known as the islets of Langerhans, in memory of their original discoverer (Henderson and Daniel, 1979).
Mounting evidence implies a key role for endothelial cells and blood flow in pancreatic islet development and function (Lammert et al., 2001; Yoshitomi and Zaret, 2004; Shah et al., 2011). Branching morphogenetic studies in other organ systems, such as the lungs, have highlighted the role that the vasculature plays in 3D patterning of branching (Lazarus et al., 2011).
The basic morphology of the pancreatic vasculature, especially of the pancreatic islets, was studied over the past century either through in vivo light microscopy (McCuskey and Chapman, 1969; McCuskey, 1997) or by using a variety of dye injections including Berlin Blue, Evans blue, and Indian Ink (Henderson and Daniel, 1979; Fraser and Henderson, 1980) or by vascular corrosion casting with scanning electron microscopy (Weaver and Sorenson, 1989; Jansson and Carlsson, 2002; Nyman et al., 2008; Gorczyca et al., 2010). These techniques were cumbersome, but the latter offered a 3D high resolution view of the vascular network (Weaver and Sorenson, 1989; Lametschwandtner et al., 1990; Gorczyca et al., 2010). However, none of the previously described islet capillary imaging studies were coimaged with insulin or glucagon. Furthermore, due to the static and difficult nature of the casting preparation, it does not lend itself well to further quantitative analysis. The elegant blood flow and anatomic analysis of living pancreatic islets by Nyman et al., 2008 was quite dramatic, but such an analysis may not be easily applicable to a large-scale, detailed quantitative analysis of numerous islets.
To better understand the detailed anatomy of islet vasculature, we developed a whole-mount immunohistochemical protocol to image the entire intact vasculature in an adult pancreas. Here, we were able to image through the whole mouse pancreas, and quantify islet vasculature. Capturing these images using a standard upright Axio Imager Zeiss microscope, and staining with ant-insulin, anti-glucagon and anti-CD31 antibodies, followed by secondary antibodies conjugated with fluorescent dyes, similar to staining a standard pancreatic histologic section. This simple method does not require the need for sophisticated imaging software such as optical coherence tomography or optical projection tomography, or the use of labor-intensive electron microscopy (El-Gohary et al., 2012). In addition, one can scan through the whole pancreas with relative ease and speed (Fig. 2A). Previous whole-mount immunohistochemical imaging of the vascular system has been described in embryonic tissues, but not in the adult pancreas (Li and Mukouyama, 2011; Shah et al., 2011).
Islet microcirculatory architecture and blood flow analysis has received a lot of interest over the past few decades, especially in islet transplantation. Less than 30% of the transplanted islet mass undergoes successful revascularization (He et al., 2008). The loss of donor islet endothelial cells may significantly impair functional islet mass post-transplantation (Brissova et al., 2004; Nyqvist et al., 2005). Our whole-mount imaging modality may help to shed some light on the vascularization process in transplanted islets.
There is conflicting evidence regarding blood flow patterns in islets (Fujita, 1973; Henderson and Daniel, 1978; Fraser and Henderson, 1979; Weaver and Sorenson, 1989). Nyman et al. used in vivo imaging to elegantly demonstrate two predominant blood flow patterns. The majority (65%) of islets displayed blood perfusing the core first, and then the islet perimeter (inner-to-outer), but then 35% of islets had blood perfusing from one side of the islet to the other, regardless of cell type. However, their analysis of islet blood flow was focused on β-cells. With our whole-mount imaging, one can also analyze the α-cell distribution within an islet and its relationship to blood flow as seen in Fig. 4A′. Although endocrine cell distribution in human islets has been argued to be different from rodent islets, recent studies support that there are significant similarities (Bosco et al., 2010; El-Gohary et al., 2012).
Previous studies in NOD mice have documented the linear decline in β-cell mass along with a reduction in islet vascular area as a result of autoimmune destruction (Savinov et al., 2003; Akirav et al., 2011). Similarly, Nyman et al. used their in vivo system to document increased islet blood flow during hyperglycemia (Nyman et al., 2010). Quantitative analysis of islet vasculature has been achieved through exhaustive sectioning of the whole pancreas (Bock et al., 2003) or through vascular corrosion casting(Weaver and Sorenson, 1989). These approaches yielded high resolution images, but were labor-intensive. Others have used complex and expensive imaging modalities such as optical projection tomography (Hornblad et al., 2011). Here, we describe a relatively simple whole-mount imaging technique to confirm the previous observation of a reduction in islet vascular volume in diabetic NOD mice compared to wild-type islets, as well as to document total islet number with progression of insulitis. Insulitis in NOD mice begins at 3–5 weeks of age and gradually increases until 16–20 weeks (Signore et al., 1994). We found a significant decrease in the total number of islets in 12-week-old NOD diabetic mice compared to 8-week-old NOD mice, in keeping with insulitis progression. We also documented a greater number of islets in the duodenal head of the NOD mouse pancreas compared to the tail, but found the opposite with wild-type CD1 pancreas, indicating that different mouse strains might have differing islet distribution. As described in previous studies(Weaver and Sorenson, 1989), the vast majority of islets were small (30um to 150um in diameter). In addition, we also found a trend toward reduction in the overall intraislet vascular length, and a significant decrease in the number of branch points and the number of segments. These observations are also in keeping with the continuing reduction in β-cell mass in NOD diabetic islets. The complex vascular branching network that exists within an islet may be necessary for maintaining proper islet function by optimizing blood sampling, exchange, transportation of hormones, and so forth. Thus, a reduction in the overall vascular branching pattern of the islet as a result of autoimmune destruction may worsen islet function.
In addition to the morphometric data shown, we found that the vascular anatomy of the islets varied depending on the size of the islet and the location of the islet in relation to the large vessels within the pancreas. Islets may receive capillary branches in a polarized manner (Figs. 1B, 4B,B′) if they are located in the periphery of the pancreas, or receive branches directly from large vessels if the islet is more central (Figs. 4C,C′,E,E′). We were able to quantify the number of entrance/exit points per islet. Smaller islets (≤100 μm) usually had two to three vascular penetration points (Fig. 4E,E′) whereas larger islets typically displayed more penetration points (Fig. 4B,B′). Our technique allows us to analyze the relationship of α-cells to arterioles and venules as seen in Fig. 4A′. It appears that α-cells tend to locate more on the side of the islet near where arterial inflow is, and less often near efferent venules.
Here we describe a simple 3D technique to image the islet vasculature that confirmed the dense nature of the islet vasculature, but with a vast array of anatomical variations. Changes in islet vasculature seen in diabetic mice may exacerbate the existing β-cell loss.