Vascular Reactivity in Arterioles From Normal and Alloxan-Diabetic Mice: Studies on Single Perfused Islets

Male C57BL/6 mice with a body mass of ∼25 g were purchased from Scanbur (Sollentuna, Sweden). Some of the animals were injected intravenously with alloxan (160 mg/kg body wt; Sigma-Aldrich, Irvine, U.K.) 5–7 days before the experiments, and only mice with blood glucose concentrations exceeding 15 mmol/l were selected for further experiments. All animals had access to tap water and pelleted food throughout the experiments. All experiments were approved by the local animal ethics committee at Uppsala University.

Mice were killed by cervical dislocation, and the pancreas was quickly removed and placed in cold (4°C) albumin-enriched (1%) Dulbecco’s minimal enriched medium (Sigma-Aldrich, Stockholm, Sweden). Islets were dissected with their arterioles intact (21) using a modification of a previously described technique for renal glomeruli (18,19). The time for dissection was limited to 60 min, and most of the obtained islets were large (diameters 400–600 μm). Occasionally, we were also able to isolate medium-sized islets (diameter 200–300 μm), especially in alloxan-diabetic mice, where islets were few and small. To evaluate possible effects of islet size on vascular reactivity, we compared the effects of d-glucose (see protocol 1) in large islets (diameter 500–600 μm) with that of medium-sized islets (diameter ∼200–300 μm). Isolation of the medium-sized islets is difficult to achieve, and since we detected no differences in the reactivity, we subsequently used larger islets. The dissection was performed under a stereo microscope at a magnification of 250× by means of sharpened forceps (no. 5; Dumont, LA-Chaux-de-Fonds, Switzerland). The islets, with their attached arterioles, were cut with miniblades and transferred into a chamber on a stage of an inverted microscope. This experimental set-up allows movement and adjustment of concentric holding and perfusion pipettes (Luigs & Neumann, Ratingen, Germany). The perfusion system (Vestavia Scientific, Vestavia Hills, AL) used manually produced pipettes from custom glass tubes (Drummond Scientific, Broomall, PA). A holding pipette was used to keep the islet in place while another holding pipette, into which the ends of the arterioles were aspirated, was put in place. The latter had an aperture of ∼30 μm, whereas the inner perfusion pipette, with an aperture of 5 μm, was advanced into the lumen of the blood vessel (Fig. 1A and B).

Isolation and preparation of renal glomeruli was performed as previously described (18,19). To ascertain the vascular reactivity of the preparations, the experiments ended with administration of KCl.

The technique used for perfusion of single islets was that used for renal glomeruli, with some modifications (18,19). The perfusion pipette was connected to a manometer and a reservoir containing the perfusion solution (Krebs-Ringer bicarbonate buffer with 10% HEPES and 1% BSA [hereafter referred to as KRBH], various concentrations of d-glucose [5.5 mmol/l, unless otherwise stated], and other substances as described below). The flow was adjusted by pressure measurements, aiming at 40 mmHg throughout the perfusion period. This corresponds to an approximate flow of 40 nl/min. The end magnification (∼300×) results from a Nikon X60/1.2 water immersion objective and a projection (×1) on a 0.3-inch chip digital camera (CB-3803S; GKB, Tai-chung, Taiwan). Data were stored on super video home systems video tapes on a video recorder (Panasonic NV-HS830; Matsushita Audio Video GmbHm, Lüneburg, Germany). These video sequences were digitalized using a frame-grabber card (pciGrabber-4plus; Phytec Technologie Holding, Mainz, Germany), and blood vessel diameters were later analyzed using customized software. This experimental set-up allowed us to continuously measure the diameter of the blood vessels and to record changes at a resolution of <0.2 μm.

KRBH, with pH adjusted to 7.4, was also used for the chamber in which the islets were located. However, the concentration of BSA was only 0.1%. All buffers were exposed to air throughout the experiments. Criteria for using an islet arteriole were remaining basal tone, no pronounced vasodilation, and a fast and complete constriction in response to administration of KCl solution (100 mmol/l) or angiotensin II (Ang II; 10⁻⁷ mol/l). Arterioles were allowed to recover for 10 min after the test.

In all series of experiments, the images from the last 10 s of each control or treatment period were used for statistical analyses. Only one concentration-response curve, with or without pretreatment with one other drug, was obtained in each of the perfused islets.

d-glucose, 3-O-methyl glucose, Ang II, adenosine, and l-nitro-arginine methyl ester (l-NAME) were purchased from Sigma-Aldrich, and spermine NONO-ate was obtained from VWR International (Stockholm, Sweden).

Each experiment began with a 15-min equilibrium period with buffer containing 5.5 mmol/l glucose in both the bath and perfusion solution. Thereafter, one of the different test substances was added to the perfusion chamber (all protocols) and/or the perfusion solution (protocols 1 and 2) for 15–30 min.

Islets were subjected to one of the following protocols after the equilibrium period, whereas renal glomeruli were exposed to one of the protocols (1, 2, or 4): 1) 15 min with 17 mmol/l d-glucose (performed separately on large [diameter 500 μm] and medium-sized [diameter 200–300 μm] islets), 2) 15 min with 5.5 mmol/l d-glucose plus 11.5 mmol/l 3-O-methyl glucose, 3) 15 min with adenosine (10⁻⁵ mol/l), 4) 14 min with Ang II (2 min with each concentration [10⁻⁶ to 10⁻¹² mol/l], beginning with the lowest), and 5) 15 min with l-NAME (10⁻⁴ mol/l) or l-NAME plus spermine NONO-ate (5 × 10⁻⁴ mol/l).

Each islet/glomerulus was exposed to one of the test substances, and the experiment was then terminated by administration of KCl (100 mmol/l) or Ang II (10⁻⁷ mol/l) to ascertain that the arterioles were able to contract. The islet was then fixed in 4% (wt/vol) formaldehyde for 4 h, dehydrated, and embedded in paraffin. Sections 7-μm thick were mounted and stained with hematoxylin and eosin or antibodies against insulin (22).

All values are given as means ± SE. Numbers of observations (n) represent numbers of observed islets, and usually only one islet was isolated per animal. Probabilities (P) of chance differences were calculated with Student’s unpaired t test or ANOVA with Sigmastat (SSSPD, Erkrath, Germany).

We were able to consistently microdissect islets from both normoglycemic and alloxan-diabetic mice. For obvious technical reasons, it was much easier to locate and prepare large islets (Fig. 1A). In each pancreas of normal mice, there were a few very large islets with diameters between 400 and 600 μm, and the diameters of the arterioles supplying these islets were 30–50 μm. We were occasionally able to isolate medium-sized islets, with a diameter of 200–300 μm, but this was tedious and the success rate low. The identity of these structures as islets was verified by insulin staining of paraffin sections. The islets chosen for the experiments contained one arteriole, and it was rare among these large islets to encounter one with multiple arterioles. The veins were difficult to identify, but we sometimes saw accumulations of erythrocytes at the periphery of the islets immediately after isolation (Fig. 2A). After initiating the perfusion, we could see erythrocytes escaping from some locations along the islet periphery (Fig. 2B), presumably representing cut venules.

The identity of the cannulated microvessels as arterioles was verified by the presence of a single layer of vascular smooth muscle in the media (Fig. 3), which reacted with contraction to administration of Ang II (∼20% decrease in diameter) (Fig. 3B) and KCl (almost 90% constriction) (Figs. 3C–D).

In most islets studied, the arteriole entering the islets branched into smaller tributaries almost immediately after entry into the islet (Fig. 2B). If this occurred in a discontinuity, the non–β-cell mantle of the islets could not be determined with the present technique. We were unable to determine whether some of the arterioles penetrated into the center of an islet before branching, but we cannot exclude that this may have occurred. We also found that smaller microvessels advanced from the branching arteriole into the central parts or the periphery of the islets. In alloxan-diabetic animals (duration of hyperglycemia 4–7 days), the number of visible islets was much smaller, but occasional islets with a diameter of 200–300 μm were found. Morphologically, these contained some β-cells, but, as expected, other endocrine cells predominated. Isolated renal glomeruli were smaller than islets and had a diameter of ∼100 μm, with afferent arterioles with a size of 8–12 μm.

The diameter of the renal afferent arterioles did not change when the glucose concentration of the perfusion medium was increased from 5.5 to 17 mmol/l (Figs. 4 and 5). Islet arterioles, on the other hand, displayed a 5% vasodilation in response to an increase in glucose concentration. This response was similar in islets with a diameter of 400–600 μm to that of islets with a diameter of 200–300 μm. There was an even more pronounced dilation of ∼10% in islets isolated from diabetic animals. A combination of d-glucose (5.5 mmol/l) and 3-O-methyl glucose (11.5 mmol/l), given the same osmolarity, did not influence either renal or islet arterioles derived from normo- or hyperglycemic mice.

Administration of adenosine caused a 5% dilation of the arterioles in all examined animals (Fig. 6). The response was statistically significant (P < 0.05) at 12 and 15 min after initiation of stimulation.

As expected, Ang II caused a pronounced dose-dependent constriction of glomerular afferent arterioles, with a maximum constriction of 40% at the highest dose (i.e., 10⁻⁶ mol/l) (Fig. 7). Islet arterioles also exhibited dose-dependent constriction but only responded to higher concentrations of Ang II. Furthermore, the maximal vasoconstriction attained was 20%. Islet arterioles prepared from normoglycemic and diabetic animals did not differ in this respect.

l-NAME alone caused a significant constriction of islet arterioles from normoglycemic mice, which occurred between 6 and 15 min after the start of administration (Fig. 8). The maximal constriction was seen at 15 min and amounted to ∼8%.

Combined administration of l-NAME and spermine NONO-ate caused a progressive vasodilation that began immediately after initiation of the drug administration and reached a plateau after 10–15 min (Fig. 8).

This study presents, for the first time, a technique that enables isolation of single pancreatic islets and their perfusion through the endogenous vasculature of the islet. This means that administered substances reach the islets through arterioles and may affect not only the endocrine cells within the islets but also endothelial and vascular smooth muscle cells (VSMs) in the microvessels. As shown, this allowed us to directly study vascular reactivity in islet arterioles without confounding the effects imposed by surrounding exocrine tissues or simultaneous vascular effects induced in exocrine blood vessels. Previous studies (23) have suggested that the control of islet blood flow is mainly precapillary, and this was supported by the vascular response to vasoactive drugs seen in the present study. Our major finding is that d-glucose, the main insulin secretagogue, had a selective dilating effect on VSM in islet arterioles but not in glomerular afferent arterioles. Furthermore, the response to glucose was amplified in islet arterioles in diabetic animals. This is in line with the previous observation of enhanced islet blood perfusion in animals with impaired glucose tolerance or overt type 2 diabetes (9), and the effect may contribute to impaired islet function as outlined further below.

One possible disadvantage of the present technique is that it does not enable us to study all islets (i.e., the whole islet organ) simultaneously as is done with whole-pancreas perfusions. It may be argued that islets are heterogeneous (24) and that the large islets we study are not representative of the average pancreatic islet. In previous studies, it has been demonstrated that small islets, which contribute little to the total pancreatic islet volume (25), do not possess an autonomous vasculature but are incorporated into the exocrine capillary network (1). It seems as if larger islets, which contain most of the endocrine cells (25), receive a major part of the islet blood perfusion (26). Furthermore, there appears to be a subset of islets that, during short-term (30 min) studies, are preferentially perfused (27); however, whether this is also reflected in long-term experiments is not known. It was recently suggested (28) that small rat islets (diameter <125 μm) were functionally superior to larger islets both in vitro and in transplantation outcomes, probably at least partially due to the fact that they are subjected to less ischemia in vitro due to their small size. In view of this, we performed separate studies on islets of two sizes (diameter of 400–600 or 200–300 μm). No differences in the response to glucose was noted, which argues in favor of the notion that the larger islets are representative of the islet organ as a whole. However, the islets referred to as small islets, i.e., a diameter <125 μm by MacGregor et al. (28), are impossible to isolate with the present technique, since they cannot with certainty be distinguished from small groups of exocrine acini. In the present study, we usually chose one islet per pancreas, and this islet was usually very large with a diameter of 400–600 μm. The number of such islets per pancreas was small, usually around three to five. In some normoglycemic animals and in most alloxan-diabetic mice, we could also isolate medium-sized islets, with a diameter of 200–300 μm. All isolated islets reacted similarly to the applied vasoactive drugs, irrespective of size. We therefore consider that the islets we have studied are representative of the islet endocrine organ as a whole and that we can draw conclusions about islets in general from our findings.

Of major interest, as mentioned above, was our finding that d-glucose dilated islet arterioles but not glomerular afferent arterioles. That this effect was specific for d-glucose was verified by the finding that the nonmetabolizable glucose analog 3-O-methyl glucose, which served as an osmotic control, had no effects. The time schedule for the effect of d-glucose was similar to that of adenosine, i.e., it was seen after ∼10 min. This is of considerable interest, since we know from previous studies (29) that the initial islet hyperemia in response to increased glucose concentrations is mainly mediated through the vagus nerve. However, starting ∼10 min after d-glucose administration, and continuing for an additional 15 min, the increased glucose metabolism within the islets produces and releases adenosine (30), which has been shown to be responsible for this later hyperemia (31). Thus, the present findings verify the in vivo observations and suggest that the released adenosine has a direct effect on the islet microcirculation. This may be caused by adenosine affecting arteriolar VSMs where the vessel penetrates into the islet and subsequent propagation of this signal through gap junctions. This emphasizes the complexity of the control mechanisms for islet blood perfusion and suggests the importance of these mechanisms for normal homeostasis.

Of even more interest in this context was the finding that arteriolar dilation in response to d-glucose was enhanced in islets derived from diabetic animals. This points to another mechanism, besides the already suggested central nervous effects of d-glucose (29,32), to explain the islet hyperperfusion of blood seen in animal models of type 2 diabetes and impaired glucose tolerance (9,14), namely that hyperglycemia may cause direct local dilation of islet arterioles. In view of the possible deleterious effects of increased blood flow caused by increased shear stress and the release of potentially harmful endothelial mediators (15–17), this is of interest. Whether the flow increase is due to increased glucose metabolism (i.e., adenosine), increased glucagon secretion from islets in diabetic animals (glucagon is a vasodilator [33]), increased production of local NO (11), or any other local factor remains to be determined. It should also be noted that both type 1 (34) and type 2 diabetes can be associated with local inflammatory reactions (35,36), which may also lead to an increased propensity for vascular dilation. It should also be noted that there are other cells present at the vascular pole of the islet, such as macrophages and adipocytes, the products of which may also influence arteriolar reactivity.

The finding that administration of l-NAME constricted islet arterioles, whereas spermine NONO-ate dilated them, confirms previous observations in vivo (11,12). Of interest in this context is that NO dilated the islet arteriole under normo- and hyperglycemic conditions. The present results, together with previous observations (37,38), suggest that NO is indeed involved in the regulation of the basal tone in islet arterioles. We have previously proposed that an intact NO production is a prerequisite for high basal islet blood perfusion, but the present findings suggest that not all of the NO effects are elicited through arteriolar dilation by a direct action on VSM. Thus, the decrease in diameter by only 8% on administration of l-NAME can theoretically— according to Poiseuille’s law, which states that the flow is proportional to the fourth power of the radius—diminish flow by 30%; however, findings in vivo suggest that the blood flow is decreased by several hundred percent (12). There is an increase of similar magnitude when NO is administered exogenously. This suggests that other, as yet unknown, effects of NO, besides its effects on VSM, are also involved in the NO dependency of the high basal islet blood perfusion. One possibility could be dilating effects on pericytes (39), but such cells are very sparse in normal islets (40), although they can be seen in islet tumorigenesis (41).

The constrictive effects of Ang II on islet arterioles were concentration dependent and very pronounced at higher concentrations. This is in accordance with the previous observation in vivo of a pronounced sensitivity of islet arterioles to this substance (13). Furthermore, we previously noted that inhibition of ACE increased islet blood flow in healthy rats (13).

The present technique also confirms previous observations on islet vascular anatomy (1,6). It should be kept in mind that we have mainly studied very large islets, and we have consistently noted that the arteriole branches at the periphery of the islet. This is in contrast to other reports suggesting that the distribution of the different cell types, with regard to the blood capillaries, is organized so that the arterial blood first reaches the β-cells, then the α-cell, and then the δ-cells (42). In the latter model, it is presumed that the arterioles penetrates into the center of the islets, i.e., to the β-cell–rich core, before branching into capillaries that then radiate toward the periphery of the islets. This direction of the blood flow has been claimed to be essential for normal islet function, since the high local hormone concentrations can either stimulate or suppress the secretion of the downstream endocrine cells (42). Other theories have also been proposed, and the issue is at the moment under debate (2). However, the present study shows that this is not the case for very large islets, but we are unable to determine what happens in smaller islets.

The study was supported by a generous Innovative Grant from the Juvenile Diabetes Research Foundation International and partially by grants from the Swedish Research Council (72X-109 and 04-03522-32), the Swedish Heart and Lung Foundation (20040645), the Wallenberg Foundation, the NOVO Nordic Fund, and the Ingabritt and Arne Lundberg Foundation.