Measurements of Insulin Secretory Capacity and Glucose Tolerance to Predict Pancreatic β-Cell Mass In Vivo in the Nicotinamide/Streptozotocin Göttingen Minipig, a Model of Moderate Insulin Deficiency and Diabetes Free

Adult male Göttingen minipigs, aged 11–16 months, were obtained from the barrier unit at Ellegaard Göttingen minipigs ApS (Dalmose, Denmark). Animals were housed in single pens under controlled conditions (temperature was kept between 18 and 22°C and relative air humidity was 30–70% with four air changes per hour) with a 12 h light:12 h dark cycle and fed 140 g SDS minipig diet (SDS, Essex, U.K.) twice daily and 240 g of a commercial swine fodder (Svinefoder 22; Østsjallands Andel, Karise, Denmark) and allowed free access to water. The pigs were studied at least 2 weeks postsurgery and trained carefully in all experimental procedures before the start of experiments. A total of 43 animals were used in the studies, weighing 26 ± 4 kg (range 18–33). The type of study was approved by the Animal Experiments Inspectorate, Ministry of Justice, Denmark.

All 43 animals included in the study underwent a mixed meal oral glucose tolerance test (OGTT) with measurement of glucose and insulin levels, and the pancreata from all animals were evaluated histologically. Furthermore, 30 of the included animals underwent an insulin secretion test (6 normal animals, 20 animals dosed with 67 mg/kg NIA and 125 mg/kg STZ, and 4 animals dosed with 20 ([n = 1]) or 0 mg/kg NIA ([n = 3]) and 125 mg/kg STZ).

Two central venous catheters (Certo 455; B. Braun Melsungen AG, Melsungen, Germany) were inserted surgically under general anesthesia as described previously (35). Postsurgical analgesia was maintained by injection of 0.03 mg/kg buprenorfine (0.3 mg/ml Anorfin; GEA, Frederiksberg, Denmark) and 4 mg/kg carprofen (50 mg/ml Rimadyl vet.; Pfizer, Ballerup, Denmark) intramuscularly before the end of anesthesia and for 3 days postsurgery by once-daily injection of 4 mg/kg carporfen intramuscularly. At the start of the study period, all animals had recovered fully from the surgical procedure, as evaluated by normal behavior and eating patterns.

β-Cell mass was reduced using a combination of NIA (Sigma N-3376) and STZ (Sigma S-0130). Six animals were kept as normal controls, while NIA was given in a range of doses (100 [n = 6], 67 [n = 25], 20 [n = 2], or 0 mg/kg [n = 4]) 15 min before STZ (125 mg/kg) to accomplish a partial protection of the β-cells against the diabetogenic action of STZ. Administration of NIA and STZ was performed after an 18-h overnight fast in conscious animals. Animals were offered SDS fodder 2 h after treatment and observed frequently during the first 48 h after administration of NIA and STZ, when blood glucose was monitored regularly to avoid episodes of hypoglycemia due to sudden hyperinsulinemia caused by necrosis of β-cells.

NIA was weighed out in individual portions and protected from light and dissolved in sterile saline (0.9%) (SAD, Copenhagen, Denmark) to a concentration of 300 mg/ml immediately before injection. STZ was weighed out in individual portions and dissolved in sodium citrate buffer (no. 929546; Bie & Bertsen, Roedovre, Denmark) (pH = 4.7) to a concentration of 62.5 mg/ml immediately before injection.

The OGTT was performed in nonrestrained freely moving animals in their usual pens to reduce the amount of stress experienced by the animals during testing. In animals dosed with NIA plus STZ or STZ alone, the OGTT was performed 7 ± 1 days (range 6–8) after dosing.

After an 18-h overnight fast, animals were offered an OGTT consisting of 25 g SDS minipig fodder and 2 g/kg glucose (500 g/l; SAD). The meal was eaten from a bowl under supervision.

Blood samples were obtained from the jugular vein catheters at −15, −5, 0, 15, 30, 45, 60, 90, 120, 150, and 180 min relative to the glucose load.

The insulin secretion test was performed in nonrestrained freely moving animals in their usual pens to reduce the amount of stress experienced by the animals during testing.

In animals dosed with NIA plus STZ or STZ alone, the insulin secretion test was performed 19 ± 4 days (range 12–26) after dosing. Since β-cell response to glucose alone has been shown to be more affected by progression to type 1 diabetes than the response to arginine (39), a bolus of arginine was included as an alternative insulin secretagogue to glucose.

After an 18-h overnight fast, animals were dosed with glucose (0.3 g/kg) (500 g/l; SAD) at time zero. At 60 min, another glucose bolus (0.6 g/kg) was given and an infusion of glucose (2 g · kg⁻¹ · h⁻¹) (200 g/l; SAD) was given from 60 to 100 min to maintain hyperglycemia at ∼20–30 mmol/l, since maximum insulin secretion in response to a bolus of arginine has been reported at ∼25 mmol/l in human subjects (23). (The rate of glucose infusion had to be reduced to 0.6 g · kg⁻¹ · h⁻¹ in severely diabetic animals [n = 4] to allow comparison with the rest of the animals studied.) At 90 min, a bolus of arginine (67 mg/kg) was given. All compounds were given intravenously through a central venous catheter. Blood samples were obtained from the jugular vein catheters at −15, −10, −5, 1, 3, 5, 7, 10, 45, 50, 55, 61, 63, 65, 67, 70, 80, 85, 91, 93, 95, 97, and 100 min relative to the first glucose bolus.

Blood samples (2 ml whole blood from the OGTT and 1.5 ml whole blood from the insulin secretion test) were immediately transferred to vials containing EDTA (1.6 mg/ml final concentration) and 500 KIU/ml full blood aprotinin (10,000 KIU/ml Trasylol; Bayer, Lyngby, Denmark) and kept on ice until centrifugation. Samples were centrifuged (4°C for 10 min at 3,500 rpm) and plasma separated and stored at −20°C until analysis. Plasma glucose was analyzed using the immobilized glucose oxidase method, 10 μl plasma in 0.5 ml buffer (EBIO plus autoanalyzer and solution; Eppendorf, Hamburg, Germany). Plasma insulin was analyzed in a two-site immunometric assay, with monoclonal antibodies as catching and detecting antibodies (catching antibody HUI-018 raised against the A-chain of human insulin, detecting antibody OXI-005 raised against the B-chain of bovine insulin) (40) and using purified porcine insulin for calibration of the assay. The minimal detectable concentration was 3.2 pmol/l, the upper limit was 1,200 pmol/l (no sample dilution), and the inter- and intra-assay variations (at three concentration levels) were 15.3 and 3.2% (at 342 pmol/l), 9.9 and 7.6% (at 235 pmol/l), and 14.6 and 4.4% (at 87 pmol/l). Recovery at high, medium, and low concentration levels was 97.1, 97.9, and 101%, respectively. Cross-reactivity against a number of peptides has been tested: IGF-1 (human) = 0.03%, growth hormone (porcine) = 0.001%, glucagon (porcine) = 0.4%, insulin (rat) = 0.03%, somatostatin = 0.2%, pancreatic polypeptide (porcine) = 0.2%, C-peptide (porcine) = 0.01%, intact proinsulin (human) = 0.3%, 32-33 split proinsulin (human) = 0.3%, des 31-32 split proinsulin (human) = 0.5%, 65-66 split proinsulin (human) = 30%, and des 64-65 split proinsulin (human) = 63% (40).

A commercial kit was used to measure C-peptide concentrations (Porcine C-peptide RIA kit, cat. no. PCP-22K; Linco, St. Charles, MO).

After euthanasia with pentobarbitone (20 ml per animal as an intravenous bolus) (200 mg/ml; Pharmacy of the Royal Veterinary and Agricultural University, Copenhagen, Denmark) at the end of the study period, the pancreata were isolated in toto for histological examination, as previously described (35). In animals dosed with NIA plus STZ or STZ alone, pancreata were obtained 35 ± 27 days (range 11–84) after the OGTT and 8 ± 7 days (range 0–27) after the insulin secretion test. In the normal control animals, pancreata were obtained 21 ± 4 days (range 17–26) after the OGTT and 13 ± 7 days (range 6–23) after the insulin secretion test.

In short, pancreata were fixed in paraformaldehyde (Bie & Berntsen, Copenhagen, Denmark), embedded in 3% agar solution (cat. no. 303289; Meco-Benzon, Copenhagen, Denmark), and sectioned as practiced in the smooth fractionator method (41,42). The deparaffinized sections were stained for insulin and a mixture of antibodies to glucagon, somatostatin, and pancreatic polypeptide to visualize β- and non-β endocrine cells. Furthermore, sections were counterstained with Mayer’s hematoxyline. β- and non-β endocrine cell mass was evaluated stereologically in two to three sections with the origin of the sections blinded to the observer. Mass of endocrine cells is expressed in milligrams per kilogram body weight.

The area under the curve (AUC) for glucose and insulin during OGTT was calculated using the trapezoidal method (baseline = 0 min). Acute insulin response (AIR) to glucose and arginine was calculated as the average increase in insulin levels during the first 10 min after dosing [AIR = (mean insulin 0–10 min) − (baseline before dosing)]. The effect of NIA plus STZ on parameters of glucose tolerance obtained during the OGTT (fasting plasma glucose [FPG], AUC glucose, and maximum insulin and C-peptide levels during the OGTT) was evaluated by comparison to similar parameters obtained in 51 normal animals.

All calculations and statistical evaluation of results were performed using linear and nonlinear regression and one-way ANOVA with Tukey’s multiple comparison test as posttest using Excel (2000) and GraphPad Prism version 3.00 for Windows (GraphPad Software, San Diego, CA). P values ≤0.05 were considered significant. Data are presented as means ± SD.

β-Cells accounted for 74 ± 7% of the total endocrine cell mass in the pancreas of the normal minipig, and the pancreatic content of endocrine tissue was 1.5 ± 0.6%. Pancreas weight relative to body weight was 1.6 ± 0.2 g/kg, and the β-cell mass relative to body weight in normal animals was 19.6 ± 6.0 mg/kg. β-Cell mass was reduced significantly by all combinations of NIA and STZ used (P < 0.0001) (Table 1).

FPG was increased significantly with falling doses of NIA (P < 0.0001), with the increase in FPG being significant for 67, 20, and 0 mg/kg NIA (Table 1). Similarly, AUC for glucose during the OGTT was increased significantly from the normal value when the NIA dose was reduced (P < 0.0001), with the increase being significant in the 67, 20, and 0 mg/kg NIA groups. Furthermore, maximum insulin and C-peptide levels during the OGTT were significantly reduced by a reduction in the NIA dose (P < 0.0001 for both parameters) (Table 1).

The AIR to intravenous glucose (0.3 g/kg) was reduced by administration of 67 mg/kg NIA in combination with STZ or 20 or 0 mg/kg NIA in combination with STZ (P < 0.001 for both) versus the normal value. Similarly, AIR to 0.6 g/kg glucose (P < 0.001 for both groups) and arginine (P < 0.01 for 67 mg/kg NIA plus STZ and P < 0.05 for 20 or 0 mg/kg NIA plus STZ), as well as maximum insulin levels during the insulin secretion test (P < 0.001 for 67 mg/kg NIA plus STZ and P < 0.01 for 20 or 0 mg/kg NIA plus STZ), were reduced compared with normal values (Table 2).

From the OGTT data, it appears that FPG (r² = 0.1744, P < 0.01), AUC glucose (r² = 0.2706, P < 0.001), and maximum insulin (r² = 0.2160, P < 0.01) and C-peptide (r² = 0.1992, P < 0.01) all correlate with β-cell mass determined postmortem (Fig. 1). Data on FPG and AUC glucose indicated a large compensatory capacity, so that no dramatic change in either of these parameters was seen before β-cell mass was reduced to ∼5 mg/kg. At this point, there was a sharp increase in both FPG and AUC glucose, with further reduction of β-cell mass. For practical purposes, this relation between β-cell mass and FPG or AUC glucose is illustrated by fitting the data to an exponential function in Fig. 1 (FPG, r² = 0.7848 and AUC glucose, r² = 0.7537).

From the data from the insulin secretion test, it was found that AIR to 0.3 (r² = 0.6155, P < 0.0001) and 0.6 g/kg glucose (r² = 0.7321, P < 0.0001), as well as AIR to arginine (r² = 0.7732, P < 0.0001) and maximum insulin secretion during the test (r² = 0.8192, P < 0.0001), all correlated significantly to β-cell mass. Furthermore, for all parameters derived from the insulin secretion test, except for AIR to 0.3 g/kg glucose, the regression line intercepted the axes close to zero (Fig. 2).

The β-cell content of the endocrine tissue in the normal minipig is in the same range as the 60–80% reported in humans (3,4,8), whereas β-cell mass relative to body weight is almost twice as high in the minipig as compared with that in humans (12.7 ± 5.1 mg/kg, P < 0.01) (1). The high β-cell mass relative to body mass in the minipig could indicate a greater insulin secretory reserve compared with the human.

As previously reported (35), the administration of a combination of NIA and STZ reduces glucose tolerance and β-cell function and mass in the Göttingen minipig. Both fasting and postprandial plasma glucose seem to be relatively resistant to quite considerable reductions in β-cell mass, and are only affected when β-cell mass is reduced to ∼5 mg/kg versus the normal value of ∼20 mg/kg. The same relationship has been described for β-cell function and FPG in humans with normal insulin sensitivity, where a reduction of β-cell function of ∼75% is seen before overt fasting hyperglycemia occurs (43). The reduction of β-cell mass (∼75%), resulting in fasting hyperglycemia in the present study, is slightly different from results from a previous study, showing that a β-cell mass reduction of ∼60% results in increased fasting hyperglycemia in minipigs (36). Similar studies in baboons have shown that a reduction of β-cell mass of 50–60% causes a dramatic decrease in β-cell function (13), whereas studies in both humans (2,3,6) and rats (15) have indicated that a greater than 80–90% reduction of β-cell mass is required before overt insulin-dependent diabetes develops. It cannot be concluded from the present study whether the ability to compensate for a reduction of β-cell mass differs between species. However, most likely, the different results obtained are, to some extent, explained by the relatively small number of observations in each study and the possibility that variability in β-cell function at very low values of β-cell mass seems to occur.

Transplantation studies in both humans (16) and pigs (17) have shown a strong relation between islet mass and metabolic control. Since the β-cell mass (milligrams per kilogram body weight) in pigs is higher than in humans, this could affect the way in which functional tests can be used as a predictor of β-cell mass in each species. However, since a reduction of β-cell mass of ∼80–90% is required in both species for development of overt diabetes (2,3,6,14), it seems that effects of a relative reduction of β-cell mass could be compared in the two species. Further studies would have to clarify this in more detail.

Minimal modeling of data obtained during OGTT could be a useful tool for evaluation of β-cell function in nondiabetic humans (44), whereas other methods using OGTT data were not efficient predictors of β-cell function in healthy human subjects (45). In the present study, parameters derived from the OGTT did not show the same strong correlation with actual measures of β-cell mass as those derived from the insulin secretion test. This could be due to the longer time span between OGTT and histology compared with the time between the insulin secretion test and histology. It could also be due to the more acute stimulation of β-cells by the intravenously administered glucose and arginine than by the orally ingested glucose, where several factors influence the rate of appearance of the glucose in the systemic circulation.

A number of the tests available for evaluation of β-cell function in humans, such as the hyperglycemic clamp (21), the minimal model technique (22), and induction of pulsatile insulin secretion (26), rely on intravenous administration of glucose, whereas other tests include administration of arginine (23,24) or GLP-1 (25), and it seems that the insulin secretion induced by intravenous glucose and/or arginine could be a possible tool for predicting β-cell mass in humans (16).

A better correlation between AIR glucose and β-cell mass compared with AIR arginine versus β-cell mass has been reported in β-cell-reduced baboons (13). However, in the present study, AIR to arginine correlated better to β-cell mass than AIR to the same dose of glucose (0.3 g/kg) as used in the baboons (13). This difference could possibly be explained by the higher levels of glucose during arginine stimulation in the present study compared with those in the baboons. In the present study, it seems that maximum insulin concentration during the insulin stimulation test has a slightly higher correlation to β-cell mass than any of the other investigated parameters. Therefore, this would indicate that a maximal stimulation of insulin secretion with a combination of glucose and arginine is a better predictor of β-cell mass in vivo compared with oral or intravenous glucose alone. However, even a simple method, such as the insulin secretory response to a glucose bolus of 0.3 g/kg, could be a significant predictor of β-cell mass.

It should be noted that even though correlations obtained in the present study are highly significant, it is likely that there is considerable biological variation inherent in the technique. Therefore, to further improve the utility of the method, it could be advantageous to study changes over time in individual animals, and this is exactly the strength of the study presented here—it allows for ongoing studies of β-cell function, shown to be a predictor of β-cell mass in live animals, thereby allowing for detection of dynamics of β-cell mass in vivo.

In conclusion, the present study supports the use of functional in vivo tests as a method of evaluating β-cell mass in vivo and has established a validated basis for developing a mathematical method for estimation of β-cell mass in vivo in the Göttingen minipig. This method could be useful for studies of dynamics of β-cell mass in diabetes, with special reference to the pathology and treatment of the disease.

We thank Helle Nygaard, Lene Sejersen Winther, Lotte Gotlieb Sørensen, Anne Grethe Juul, Nanna Kasmira Nowa Hansen, Annemette Petersen, Susanne Primdal, Steen Kryger, Karsten Larsen, Ejnar Eriksen, and Hans Rasmussen for excellent technical assistance. Carolyn F. Deacon is greatly acknowledged for helpful comments in preparation of the manuscript, and Sten Madsbad is greatly acknowledged for helpful advice in preparation of protocols.