Effects of Insulin Therapy on Weight Gain and Fat Distribution in the HF&Sol

Home , G Sat

ORIGINAL ARTICLE Effects of insulin therapy on weight gain and fat distribution in the HF/HS-STZ rat model of type 2 diabetes

S Skovsø1, J Damgaard2, JJ Fels2, GS Olsen2, XA Wolf2, B Rolin2 and JJ Holst1

BACKGROUND/OBJECTIVES: Insulin therapy is required for many patients with the obesity-related disorder type 2 diabetes, but is also associated with weight gain. The specific location of adipose tissue location matters to cardiovascular disease (CVD) risk. We investigated effects of exogenous insulin on fat distribution in the high-fat/high-sucrose fed rat treated with streptozotocin (HF/HS-STZ) rat model of type 2 diabetes. We also examined effects of insulin therapy on circulating CVD markers, including adiponectin, triglycerides (TGs), total cholesterol and high-density lipoprotein. SUBJECTS/METHODS: Male SD rats were HF/HS fed for 5 weeks followed by STZ treatment to mimic the hallmarks of human obesity-associated insulin resistance followed by hyperglycemia. Magnetic resonance imaging and computed tomography were used to determine total fat, abdominal fat distribution and liver fat before and after insulin therapy in HF/HS-STZ rats. HbA1c%, TGs, cholesterol, high-density lipoprotein and adiponectin were analyzed by conventional methods adapted for rats. RESULTS: Insulin therapy lowered HbA1c (Po0.001), increased body weight (Po0.001), increased lean mass (Po0.001) and led to a near doubling of total fat mass (Po0.001), with the highest increase in subcutaneous adipose tissue as compared with visceral adipose tissue (Po0.001). No changes in liver fat were observed after insulin therapy, whereas plasma TG and cholesterol levels were decreased (Po0.001, Po0.01), while high-density lipoprotein (HDL) and adiponectin levels were elevated (Po0.01, Po0.001). CONCLUSIONS: Using the HF/HS-STZ rat as an animal model for type 2 diabetes, we find that insulin therapy modulates fat distribution. Specifically, our data show that insulin has a relatively positive effect on CVD-associated parameters, including abdominal fat distribution, lean body mass, adiponectin, TGs and HDL in HF/HS-STZ rats, despite a modest gain in weight. International Journal of Obesity (2015) 39, 1531–1538; doi:10.1038/ijo.2015.92

INTRODUCTION visceral adipose tissue (VAT) rather than subcutaneous adipose 22–25 More than 380 million people worldwide have type 2 diabetes.1 tissue (SAT) is associated with a higher CVD risk, whereas The progression of type 2 diabetes is mainly characterized by the gluterofemoral subcutaneous fat seems to have a protective effect 26 initial presence of obesity and insulin resistance, which, in against CVD risk. However, the specific effects of insulin on fat combination with a probable genetic inability of β-cells to distribution remain poorly understood. To test the hypothesis that fi compensate with adequate insulin secretion, eventually leaves insulin might induce site-speci c weight gain, we employed the 2 high-fat/high-sucrose fed rat treated with streptozotocin (HF/HS- the subject with hyperglycemia and overt type 2 diabetes. Over 19 β STZ) as a pre-clinical model of type 2 diabetes. First, we time, a further decline in -cell function, often results in a need for fi external administration of insulin. This anabolic hormone, among con rmed the presence of hypertriglyceridemia, hyperinsulinemia, increased total fat mass and normoglycemia as an indication of many other physiological roles, possesses both anti-lipolytic as – insulin resistance before the introduction of STZ-induced hyper- well as lipogenic and protein synthesis effects.3 5 Thus, it is not glycemia in HF/HS-STZ rats. Second, we investigated if insulin surprising that many people with type 2 diabetes receiving insulin 6–11 therapy (Neutral protamine Hagedorn, NPH) led to reductions in therapy gain weight. However, obesity is already present in HbA1c and increases in body weight in HF/HS-STZ rats, which most type 2 diabetes individuals before initiation of insulin 12 would parallel the effects seen in humans. Third, we examined the therapy. Obesity is an independent risk factor of cardiovascular effect of insulin therapy of HF/HS-STZ rats with respect to fat disease (CVD), which is the most common cause of death in type distribution and other well-known markers of CVD risk, including 2 diabetes.13–15 Thus, the focus has been on developing type 16–18 circulating adiponectin, triglycerides (TGs), cholesterol and high- 2 diabetes drugs associated with little or no weight gain. To density lipoprotein (HDL). develop such treatments, pre-clinical models in which weight gain can be examined are employed.19 Weight gain, however, affects CVD risk differently depending on MATERIALS AND METHODS its location. For example, a gain in muscular tissue is associated Animals with a markedly lower CVD risk than a gain in adipose tissue. The study was carried out in accordance with the guidelines and Similarly, it has been known for more than 60 years that the risk of recommendations provided by the Federation of European Laboratory developing type 2 diabetes and CVD risk is dependent on the Animal Science Associations (FELASA)27 at Novo Nordisk A/S (Måløv, specific location of fat within the adipose tissue.20,21 Specifically, Denmark), and were approved by the animal unit at Novo Nordisk A/S.

1Department of Biomedical Sciences, Endocrinology Research Section, Panum Institute, Copenhagen, Denmark and 2Diabetes Research Unit, Novo Nordisk A/S, Måløv, Denmark. Correspondence: Dr S Skovsø, Cellular and Physiological Sciences, University of British Columbia, Point Grey Campus, 5358–2350 Health Sciences Mall, Vancouver, V6T 1Z3 British Columbia, Canada. E-mail: [email protected] Received 13 October 2014; revised 5 April 2015; accepted 6 May 2015; accepted article preview online 26 May 2015; advance online publication, 30 June 2015 Insulin therapy in a type 2 diabetes rat model S Skovsø et al 1532 Sixty 12-week-old male Spraque Dawley rats (Charles River, Sulzfeld, reference scan (pelvis start) were included in the determination of VAT and Germany) were acclimatized on a regular chow diet 1 week before study SAT. The inner layers of the abdominal wall were used to distinguish initiation. Throughout the study, 30 rats were fed ad libitum a HF/HS diet abdominal VAT and SAT from each other (Figure 2). Four trans-sectional (45 kcal % fat, 35% carbohydrate, 20% protein; D15451, Research Diets), scans were included for liver density analyses (Figure 2). A quantitative whereas 30 control rats were fed ad libitum a low-fat (LF) Purina 5008 diet magnetic resonance (QMR) EchoMRI-700 body composition analyzer (17 kcal % fat, 56% carbohydrate, 26% protein; International Product (EchoMRI, Houston TX, USA) was used for analysis of total body fat and Supplies Limited, London, UK). The LF diet was not intended to be a weight total lean body mass. Both CT and QMR measurements were performed loss diet for the purpose of this study. post the HF/HS vs LF feeding regimen (pre-STZ treatment), post-STZ Figure 1 illustrates our study design. Following 6 weeks of feeding, all 30 treatment (before insulin/vehicle therapy initiation) and again after 3 weeks HF/HS rats were anaesthetized with isoflurane and dosed (i.p.) with STZ of insulin/vehicle therapy (Figure 1). (20 mg per kg body weight, kg BW) (Sigma S-0130) in 0.1 M citrate buffer (pH 4.5). Rats were dosed with STZ for 3 consecutive days following the 6 weeks of HF/HS vs LF feeding. Based on our experience, STZ treatment Statistical methods does not lead to hyperglycemia in 100% cases, thus inclusion criteria were − Results are presented as means ± s.e.m. Body weight changes were established. Inclusion criteria were blood glucose (BG)47 mmol l 1 and HbA1c% (%) within the interval: 4.60%oHbA1c%o5.60% on the day analyzed by a two-way repeated-measures analysis of variance (not preceding initiation of insulin therapy ( = 21 days after the first STZ dose adjusted for multiple comparisons) with time and treatment (insulin vs was administered). Twenty HF/HS-STZ rats were assigned, based on HbA1c vehicle therapy) as factors. Other data were tested by paired and unpaired % stratification, to be treated with either NPH insulin (NN A/S, Denmark; two-tailed t-tests, respectively. Data appeared normal distributed. o fi given in a dose of 30 nmol per kg BW in NPH formulation and a volume of A P-value 0.05 was considered signi cant. The correlations are based 0.05 ml kg − 1 from a solution of 600 nmol ml − 1; n = 10) or vehicle (Veh; on data obtained from vehicle- and insulin-treated animals measured n = 10). The HbA1c% stratification was performed in the following way. All before STZ, and before and after insulin/vehicle therapy. Pearson’s animals were sorted from highest to lowest Hba1c% value. The first correlation coefficients are designated rp. The 95% confidence intervals animals were assigned the insulin-treated group, the second animal was are depicted as dashed lines in figures showing correlations between data assigned vehicle-treated group, the third animal was assigned to the sets. The main author was not blinded to group allocations, however, all insulin-treated group, the fourth animal was assigned vehicle group, and other investigators were. Fat distribution and liver densities analyses were this pattern was repeated. Both groups were injected subcutaneously performed in a semi-automatic way including both manual and automatic twice daily (0700 hours and 1900 hours) for 3 weeks. Body weights were parts (see Figure 2 for details). measured in non-fasted rats three times per week between 0700 hours and 0900 hours. No power calculation was performed before study initiation. The number was based on our previous experience with this RESULTS model including a pilot study where the effect of three different doses of STZ (10, 20, 30 mg kg − 1) was tested on BG levels. Indications of diet-induced insulin resistance followed by hyperglycemia in HF/HS-STZ rats Biochemical analyses HF/HS and LF feeding led to a BW of 604 ± 10 g, which was 6% Plasma (tongue blood) from conscious semi-fasted (3 h after lights on) rats higher in HF/HS rats than the BW of 568 ± 10 g in LF controls rats was analyzed for TGs, HDL, cholesterol, alanine amino transferase and (Po0.05). Total fat mass, as measured by QMR, was 83 ± 5 g in aspartate amino transferase on the Cobas 6000 analyzer series module HF/HS rats as compared with 45 ± 3 g in LF control rats (Po0.001, c501 (Hoffman La-Roche, Mannheim, Germany). The Cobas analyzer is 28,29 Figure 2c). Levels of semi-fasted plasma TGs, semi-fasted plasma known to be highly precise. Tail blood was analyzed for HbA1C and BG c-peptide and HbA1c% significantly increased to means of on the Cobas 6000 analyzer and EBIO plus glucose analyzer (Eppendorf AG, Hamburg, Germany), respectively. Blood samples analyzed following the 2.37 ± 0.24 mM, 1525 ± 108.7 pM, 3.77 ± 0.027% in HF/HS rats as 3 week insulin/vehicle intervention were taken exactly 2 h after the compared with 1.75 ± 0.18 mM, 1052 ± 101.3 pM and 3.63 ± 0.033%, morning NPH insulin/vehicle injection. Following euthanasia, the liver was respectively, in LF control rats (Po0.05, Po0.001 and Po0.001). immediately dissected, quickly weighed and freeze-clamped in liquid Similar levels of semi-fasted BG were present in HF/HS and LF nitrogen. Liver samples were homogenized in a 0.15 M sodium acetate control rats after the feeding regimen (6.05 ± 0.064 vs buffer containing Triton X-100 for liver TG analysis. The homogenate was 5.99 ± 0.069 mM, P = not significant). Following HF/HS feeding, boiled for 2 min at 100 °C, then cooled on ice and analyzed by the Cobas HF/HS rats were STZ treated, quickly leading to hyperglycemia 6000 analyzer. Plasma adiponectin was assayed by MilliPlex MAP Rat Single Plex Adiponectin, RADPK-81 K-ADPN (Millipore Corporation, Billerica, MA, 4 days after initiation of STZ treatment (21.2 ± 0.7 mM), while USA). Rat c-peptide was analyzed using an in-house luminescence oxygen maintaining some endogenous C-peptide production (297 ± 74.3 channeling immunoassay (also called AlphaLisa). pM), indicative of some residual β-cell mass in this model. Following 7 days of STZ treatment, the HF/HS-STZ rats that were Measurements of whole body fat, SAT, VAT and liver density later assigned to be treated with insulin had an 8% body weight For estimation of SAT, VAT and liver densities, anaesthetized (Isoflurane) reduction, whereas the vehicle-treated group had lost 9% rats were scanned in a Latheta computed tomography (CT) scanner (Figure 3a). These numbers had increased to 10% and 12%, (LCT-100 series, Aloka co., LTD, Tokyo, Japan). Subsequently, three images respectively, at day 21 following STZ treatment. (spaced 3 mm apart) upstream and two images downstream of the

Fig 2 Fig 3 Fig 5 Effect of insulin therapy on HbA1c% and bodyweight Twenty-one days after STZ-treatment, the HF/HS-STZ rats were STZ (n = 30) Treatment: insulin (n = 10) or divided into two groups, and treated with either insulin or vehicle week 7-9 vehicle (n =10) injections, designated HF/HS-STZ (Ins) and HF/HS-STZ (Veh), week 10-12 respectively. Three days after initiation of insulin therapy a HF/HS (n = 30 ) or signiﬁcant weight gain was observed in the HF/HS-STZ (Ins) as HF/HS Feeding (n = 20) LF Feeding compared with HF/HS-STZ (Veh) rats (Po0.001; Figure 3a). Three week 1-6 week 7-12 weeks of insulin therapy led to stable lowered semi-fasted glucose levels, as expected, but not hypoglycaemia or hyperphagia. No Figure 1. Study design. Illustrates the time line of the study design, ﬁ including feeding regimens (high fat/high sucrose (HF/HS); low fat signi cant difference was observed between HF/HS-STZ (Ins) and (LF)), streptozotocin (STZ) treatment (20 mg/kg body weight) and HF/HS-STZ (Veh) for any of the measured parameters just before insulin (Ins) vs vehicle (Veh) treatment. initiation of insulin therapy.

Figure 2. Quantification of SAT, VAT and liver density measured by micro-CT. Upper panel: A representative scout scan of an anaesthetized rat positioned in a micro-CT scanner. The two purple lines demarcate the area in which a transactional scans were performed for every third mm (~32 scans pr rat). The blue lines represent the location of the four transactional scans depicted in the middle panel. The solid red lines illustrate the location of the reference transactional scan marked in red in the lower panel. All the red lines (dashed and solid) show the location of the six transactional scans that were used for VAT and SAT determination. Middle panel: Representatives of four transactional scans used for liver density analysis. For each rat, the liver density was automatically calculated as an average of the liver density measured within the area marked by the manually drawn dashed green line. Lower panel: Three versions of the same transactional scan located in the abdominal region in a rat demonstrating the semi-automatic process of the separation of subcutaneous adipose tissue (SAT) vs visceral adipose tissue (VAT). Selection of the included six transsactional scans were based on a reference scan (that is, the scan at which the iliac crests first appeared). Three scans upstream and two scans downstream of the reference scan were included in the SAT and VAT determination analysis. For each of the six transactional scans, the inner layers of the abdominal wall were manually drawn as illustrated by the dashed white line drawn on top of the clearly visible inner layers of the abdominal wall. Fat located inside the dashed white line of the abdominal wall was automatically calculated by the LaTheta software being VAT, whereas fat outside the white dashed line was automatically calculated as being SAT. The final separation of VAT and SAT by the LaTheta software is demonstrated by the colours purple and yellow, respectively, on the far most right transactional scan seen. For both the VAT and SAT calculation, the final number was automatically calculated as an average based on the selected six transactional scans.

Effects of insulin on lean body mass and fat distribution in insulin-treated HF/HS-STZ rats, when compared with vehicle- in HF/HS-STZ rats treated HF/HS-STZ rats (36.2 ± 1.7% vs 17.3 ± 2.1%, Po0.001; Three weeks after insulin therapy initiation, changes in fat location Figure 3f). We observed significant associations between SAT% in the HF/HS-STZ (Ins) rats were examined. A significant increase in and the CVD risk marker HbA1c% (rp=− 0.58, Po0.001; total lean mass (387 ± 15 vs 344 ± 16 g, Po0.001; Figure 3b) and Figure 4b). total fat mass (99 ± 13 g vs 58 ± 8 g, Po0.001; Figure 3c) from before to after insulin therapy was observed in insulin-treated rats. Insulin effects on liver fat and liver functionality markers No change in total lean mass and a decrease in total fat mass in HF/HS-STZ rats (57 ± 7 vs 46 ± 5 g, Po0.05; Figures 3b and c) were observed in vehicle-treated rats. Insulin therapy significantly increased VAT The effect of insulin therapy on liver fat content was assessed in (12.1 ± 1.0 vs 8.6 ± 0.8 g, Po0.05; Figure 3d) and SAT (7.0 ± 0.8 vs two ways. Non-invasive liver density measurements were obtained 2.0 ± 0.3 g, Po0.001; Figure 3e). However, only SAT, not VAT, through CT scanning both before and after insulin therapy (5A). increased to a level significantly higher than the level observed Invasive measures, such as biochemical analysis of liver TG before STZ treatment (data not shown). The separation of VAT and content, were performed only after insulin therapy (Figure 5b). SAT by the inner layers of the abdominal wall can be seen on the No significant difference was observed between the liver density transactional scan shown in Figure 1. We observed a significant measured in HF/HS-STZ (Ins) rats upon termination of insulin association between total fat mass measured by QMR scanning therapy, when compared with HF/HS-STZ (Veh) rats (51.7 ± 3.8 and total abdominal fat (SAT+VAT) measured by CT scanning Hounsfield Units (HU) vs 46.2 ± 3.5 HU, P = 0.29; Figure 5a). (rp = 0.90, Po0.001; Figure 4a). Overall, insulin therapy resulted in Remarkably, liver density increased over time in insulin as well a shift towards SAT, as reflected by an increased SAT% (SAT/SAT+VAT) as vehicle-treated rats (Po0.01). The liver density is inversely

Body weight Total lean Body Mass Vehicle *** 700 Vehicle 500 Insulin Insulin 400 600 * 300

500 200 Body weight (g)

Lean Body mass (g) 100 400 STZ Insulin 0 -42 -35 -28 -21 -14 -7 0 7 14 21 28 35 Before Treatment After Treatment Time (days)

Total Fat Mass Visceral Adipose Tissue Vehicle *** Vehicle Insulin *** ** Insulin * 150 * 15 ***

100 10

50 VAT (g) 5 Total fat mass (g)

0 0 Before Treatment After Treatment Before Treatment After Treatment

Subcutaneous Adipose Tissue SAT % Vehicle Vehicle Insulin *** *** * Insulin *** 10 ** 40 *** 8 30

6 20

SAT (g) 4

10 2 SAT/(SAT+VAT) (%)

0 0 Before Treatment After Treatment Before Treatment After Treatment Figure 3. The effect of insulin on fat distribution in HF/HS-STZ rats. Body weight (g) depicted over time for HF/HS-STZ rats (Ins; closed squares) and HF/HS-STZ rats (Veh; open circles; n = 10 in each group; a). Total lean body mass (g; QMR measurement; b), total fat mass (g; QMR measurement; c), visceral adipose tissue (g; d), (e) subcutaneous adipose tissue (g), SAT% (SAT/(SAT+VAT) (%; f), Figures 2b and f: Treatment refers to either vehicle treatment (white bars) or insulin treatment (black bars) of HF/HS-STZ rats (n = 10 in each group), measured before and after 3 weeks of insulin/vehicle treatment. *Po0.05, **Po0.01, ***Po0.001. Data are presented as means ± s.e.m.

associated with the liver TG content (rp=− 0.80, Po0.001; (4.1 ± 0.6 vs 10.9 ± 0.9 mmol l − 1, Po0.001; Figure 6a) and − Figure 4d). Likewise, a tendency towards a reduced level of liver plasma cholesterol (1.9 ± 1.0 vs 3.1 ± 0.4 mmol l 1, Po0.05; TGs was present after ended insulin therapy (20.7 ± 2.8 vs Figure 6b) as well as increased plasma HDL levels (1.4 ± 0.1 vs − 28.1 ± 3.3 μmol g − 1, P = 0.11, Figure 5b) when comparing insulin- 1.1 ± 0.1 mmol l 1, Po0.01; Figure 6c), and the ratio between HDL and vehicle-treated groups. Insulin therapy reduced the level of and total cholesterol (1.1 ± 0.09 vs 0.4 ± 0.04, Po0.001), when alanine amino transferase (37.2 ± 1.9 vs 116.1 ± 13.4 U L − 1, comparing insulin-treated with vehicle-treated rats. Notably, there Po0.001; Figure 5c) and aspartate amino transferase (71.4 ± 2.3 was difference in total cholesterol when comparing the same HF/ vs 124.0 ± 23.34 U L − 1, Po0.05; Figure 5d), when comparing HS-STZ rats before and after insulin therapy (Figure 6b). Appar- insulin- and vehicle-treated HF/HS-STZ rats at the end of the study. ently, insulin therapy prevented the rise in total cholesterol with continued untreated hyperglycemia (LF-rats). The level of non- HDL (total cholesterol minus HDL) was significantly reduced in Insulin effects on circulating lipids and adiponectin in HF/HS-STZ insulin-treated rats, when comparing HF/HS-STZ rats before and rats after insulin therapy (1.3 ± 0.26 vs 0.55 ± 0.11 mmol l − 1, Po0.05). We also assessed the effects of insulin therapy on circulating CVD At the end of the study, the level of adiponectin was risk markers. Insulin therapy significantly decreased plasma TGs significantly higher in HF/HS-STZ (Ins) rats, when compared with

abFat (CT) vs. Fat (MRI) SAT% vs. HbA1c% 250 10

200 8

150 6

100 4 Hba1c% (%)

Total fat mass (g) 50 2

0 0 0 10203040 0 1020304050 SAT + VAT (g) SAT% (%)

cdSAT% vs. Adiponectin Liver triglycerides vs. Liver density 200 80

150 60 100 40 50

20 adiponctin (µg/ml)

0 Liver density (HU)

-50 0 0 1020304050 0204060 SAT% (%) Liver Triglycerides (µmol/g) Figure 4. Correlations between fat distribution and blood glucose parameters. Total fat mass measured by QMR scanning vs abdominal fat (subcutaneous adipose tissue (SAT)+visceral adipose tissue (VAT); rp = 0.90, Po0.001; a). HbA1c% vs SAT% (rp = − 0.58, Po0.001; b). Plasma adiponectin vs SAT% (rp = 0.57, Po0.001; c). Liver density measured by CT scanning vs liver triglycerides measured biochemically (rp = − 0.80, Po0.001; d). 95% conﬁdence intervals are shown with dashed lines.

Liver Density Liver Triglycerides Vehicle ** Insulin Vehicle * 40 60 Insulin

30 40 20

20 10 Liver Tg (umol/g) Liver Density (HU)

0 0 Before Treatment After Treatment After Treatment

ALAT ASAT Vehicle * Insulin *** Vehicle * 150 *** 200 Insulin

150 100

100

50 ASAT (U/L) ALAT (U/L) 50

0 0 Before Treatment After Treatment Before Treatment After Treatment Figure 5. The effect of insulin on liver fat and functionality markers in HF/HS-STZ rats. (a) Liver density (HU). (b) Liver triglycerides (μmol ml − 1). (c) Alanine aminotransferase (ALAT; U l − 1). (d) Aspartate aminotransferase (ASAT; U l − 1). Treatment refers to either vehicle treatment (white bars) or insulin treatment (black bars) of HF/HS-STZ rats (n = 10 in each group), measured before and/or after 3 weeks of insulin/vehicle treatment. *Po0.05, **Po0.01, ***Po 0.001. Data are presented as means ± s.e.m.

HF/HS-STZ (Veh) rats (106.8 ± 15.8 vs 30.1 ± 1.7 mmol l − 1, data obtained from the insulin-treated HF/HS-STZ rats after Po0.001; Figure 6d). We observed signiﬁcant associations 3 weeks of insulin treatment include clear outlier points on the between adiponectin and SAT% (rp = 0.57, Po0.001; Figure 4c), far right side of the correlation plot. These data have a tendency which seemed to be independent of insulin therapy. However, towards a steeper relationship between SAT% and adiponectin,

Triglycerides Total cholesterol Vehicle Vehicle Insulin Insulin * 15 * 4 * ***

3 10

2 5 Tg (mmol/L) 1 Total cholesterol (mmol/L) 0 0 Before Treatment After Treatment Before Treatment After Treatment

HDL Adiponectin Vehicle Vehicle 2.0 Insulin Insulin ** ** 150 * *** 1.5 100

1.0

50 HDL (mmol/L) 0.5 Adiponectin (ug/ml)

0.0 0 Before Treatment After Treatment Before Treatment After Treatment Figure 6. The effect of insulin on circulating cardiovascular disease risk markers in HF/HS-STZ rats. Circulating triglycerides (mmol l − 1; a), circulating total cholesterol (mmol l − 1; b), circulating HDL (mmol l − 1; c), and circulating adiponectin (μgml− 1; d) measured in vehicle-treated (white bars) or insulin-treated (black bars) HF/HS-STZ rats measured before and after 3 weeks of insulin/vehicle treatment (n = 10 in each group). *Po0.05, **Po0.01, ***Po0.001. Data are means ± s.e.m.

when compared with the same relationship obtained from the lean body mass have been shown after insulin therapy in type 2 same insulin receiving HF/HS-STZ rats before the beginning of diabetes subjects.31,32 Our observation that insulin therapy insulin therapy (data not shown). Clearly, future additional studies resulted in elevated levels of SAT in HF/HS-STZ rats also agrees are warranted to clarify the links between insulin and adiponectin with data from some human studies.32,33 However, three other in this model. clinical studies observed no change in SAT after insulin therapy.10,31,34 We observed an insulin-induced increase in VAT, which likewise has been reported in human studies.34 Other DISCUSSION human studies demonstrated no effect of insulin therapy on This study was conducted to determine the effects of insulin VAT.10,31,33 We favour the idea that insulin’s effects are direct on therapy on weight gain and fat distribution in the HF/HS-STZ rat adipose tissue. We have not observed hypoglycaemia or model of type 2 diabetes. We confirmed that the blood glucose hyperphagia in this model. lowering and weight increasing effect seen in humans after insulin Liver fat content was not affected by insulin therapy in our therapy was also apparent in HF/HS-STZ rats. The major finding of study, although we observed a tendency towards a decline, our study was that insulin has specific effects on fat distribution, indicated by both invasive and non-invasive measures of liver fat increasing subcutaneous rather than visceral fat. content. Insulin therapy has been reported to induce significant Our results suggest that the HF/HS-STZ rat is a useful pre-clinical decrease in liver fat accumulation in humans with type 2 model for testing the effects of anti-diabetic drugs, such as insulin, diabetes.31,35 The possible discrepancies between the human on obesity-related parameters including weight gain, fat distribu- studies are likely affected by the length of disease duration, the tion and the related CVD risk. In this study, we demonstrated that choice and intensity of insulin therapy and inclusion of insulin the HF/HS rat exhibited indicators of insulin resistance including therapy naïve or previously insulin therapy exposed subjects. hypertriglyceridemia, hyperinsulinemia, increased total fat mass Collectively, our data tend to agree with the sparse amount of and normoglycemia before STZ-induced hyperglycemia. Future human data available, showing a favouring of fat accumulation in studies should investigate β-cell mass and β-cell function in this the subcutaneous fat compartment over the visceral compartment HF/HS-STZ rat model. This animal model of type 2 diabetes has as a consequence of insulin therapy. Our data are also in recently been reviewed and found to be a reasonable model of agreement with in vitro data showing subcutaneous adipocytes late-stage type 2 diabetes, when insulin is prescribed.19,30 Our to be more sensitive towards insulin-induced fat accumulation current data are also consistent with the concept that the HF/HS- than visceral adipocytes.36–40 Of note, we showed a significant STZ model mimics the main pathological events in human type 2 association between liver density and liver TGs measured non- diabetes, insulin resistance followed by hyperglycaemia,2 as well invasively by CT scanning and invasively by a biochemical as the effects of insulin therapy. Indeed, insulin therapy lowered procedure, respectively. This observation calls for future long- – HbA1c% and increased body weight as seen in human studies.6 11 itudinal studies of hepatic lipid accumulation. Specifically, our data showed an insulin-induced gain in both total The ratio between SAT and VAT is suggested to be a better lean mass and total fat mass. Remarkably, however, the primary CVD risk marker, than both SAT and VAT alone.41 Our observed location of the newly stored fat was SAT, with only a modest significant associations between SAT% and Hba1c% as well increase in VAT. In accordance with our data, increased levels of between SAT% and adiponectin support SAT% to be a good

Insulin Weight Gain SAT VAT

Fat Storage Fat Storage

Glucose Adiponectin TG

HDL Liver fat Tot Chol ALAT ASAT

Figure 7. Summary. Insulin therapy led to a modest increase in body weight in high-fat/high-sucrose fed rats treated with streptozotocin. The weight gain primarily took place in the subcutaneous adipose tissue (SAT) rather than in the visceral adipose tissue (VAT). Insulin therapy also led to a decrease in circulating blood glucose, triglycerides (TG) and total cholesterol (Tot Chol) as well as an increase of high-density lipoprotein (HDL) and adiponectin levels. Insulin therapy further led to a decrease in alanine aminotransferase (ALAT) and aspartate aminotransferase (ASAT), and had no effect on liver fat measured by liver density assessed by computed tomography. Overall, we interpret these data as a relative improvement of the CVD risk in the HF/HS-STZ rat model treated with insulin therapy as compared with untreated diabetes, despite of a modest weight gain.

CVD risk marker. Besides having no effect on hepatic fat phase of this manuscript. Helle Nygaard and Helle Hebo, Novo Nordisk A/S (Måløv, accumulation along with primarily promoting subcutaneous fat Denmark), are thanked for their impressive technical support throughout the storage, insulin therapy lowered circulating TGs and cholesterol conduction of this study. Marianne Schiødt is thanked for her kind and technical as well as increased HDL and adiponectin. Decreased levels of support of the adiponectin analyses. Finally, Gitte Ronald Andersen and Stine Drent serum TGs and small augmented levels of serum HDL and Larsen are thanked for their unique caretaking of the rats in the animal facility at adiponectin have been reported in subjects with type 2 diabetes Novo Nordisk A/S (Måløv, Denmark). This work was supported by the Department of after insulin therapy interventions.42,43 However, the function- Insulin Pharmacology, Novo Nordisk A/S, Måløv, Denmark, the Danish In vivo ality rather than the level of HDL per se has recently been shown Pharmacology Graduate program and the LIFEPHARM Centre, Faculty of Health and to be important for the protective effects of HDL with respect to Medical Sciences, University of Copenhagen, Denmark. CVD risk.44 In contrast, the causative role of non-HDL in CVD development remains established.45 Our data indicate that the AUTHOR CONTRIBUTIONS HF/HS-STZ model is a more clinical relevant model than the SS and JD designed the study and performed all the experiments. SS analyzed leptin-compromised Zucker diabetic fatty rat with respect to fi insulin-induced changes in lipid parameters, which do not mimic as well as interpreted the data, made the gures and wrote the manuscript. the clinical situation.46 Of note, rats have an altered lipoprotein XAW, BR and JJH performed additional interpretation of the data and they fi profile, when compared with humans.47 However, we here contributed signi cantly to the writing of the manuscript. JJF and GSO demonstrate that relative changes in TGs, cholesterol and HDL contributed with data acquisition of C-peptide and adiponectin data, are detectable in rats as a consequence of insulin therapy. The respectively. data obtained in this study are limited to the effect of NPH insulin therapy for 3 weeks. REFERENCES Collectively, our data demonstrate (Figure 7) that the weight gain caused by insulin therapy is primarily due to expansion of the 1IDF.International Diabetes Federation, IDF Diabetes Atlas, 6th edn, 2013. Interna- tional Diabetes Federation: Brussels, Belgium, 2014. Available from: http://www. subcutaneous adipose tissue (and lean body mass) and that the fi idf.org/diabetesatlas (accessed April 19 2014). overall CVD risk pro le appears to be improved upon insulin 2 Prentki M, Nolan CJ. Islet beta cell failure in type 2 diabetes. J ClinInvest 2006; 116: therapy, when compared with untreated type 2 diabetes. The HF/ 1802–1812. HS-STZ rat model of type 2 diabetes showed insulin therapy- 3 Lindegaard B, Ditlevsen S, Plomgaard P, Mittendorfer B, Pedersen BK. induced changes in weight gain, fat distribution and other well- Acute reduction of lipolysis reduces adiponectin and IL-18: evidence from known CVD risk markers that mimicking those seen in humans. an intervention study with acipimox and insulin. Diabetologia 2013; 56: Therefore, the HF/HS-STZ rat model is likely to be valuable for 2034–2043. testing of new and better anti-diabetic drugs, such as new insulin 4 Mehran AE, Templeman NM, Brigidi GS, Lim GE, Chu KY, Hu X et al. Hyper- analogues. insulinemia drives diet-induced obesity independently of brain insulin production. Cell Metab 2012; 16:723–737. 5 Timmerman KL, Lee JL, Fujita S, Dhanani S, Dreyer HC, Fry CS et al. Pharmaco- CONFLICT OF INTEREST logical vasodilation improves insulin-stimulated muscle protein anabolism but not glucose utilization in older adults. Diabetes 2010; 59: 2764–2771. JD, JJF, GSO, XAW and BR are all employees at Novo Nordisk A/S and hold stocks in 6 Intensive blood-glucose control with sulphonylureas or insulin compared with the company. JJH is a consultant for Novo Nordisk A/S. NPH, used for insulin therapy conventional treatment and risk of complications in patients with type 2 diabetes in this study, is a product of Novo Nordisk A/S. (UKPDS 33). UK Prospective Diabetes Study (UKPDS) Group. Lancet 1998; 352: 837–853. 7 Best JD, Drury PL, Davis TM, Taskinen MR, Kesaniemi YA, Scott R et al. Glycemic ACKNOWLEDGEMENTS control over 5 years in 4,900 people with type 2 diabetes: real-world diabetes Professor Jim Johnson, University of British Columbia (Vancouver, Canada), is thanked therapy in a clinical trial cohort. Diabetes Care 2012; 35: 1165–1170. for discussing the data and for providing advice throughout the writing of this 8 Fonseca V, McDuffie R, Calles J, Cohen RM, Feeney P, Feinglos M et al. Determi- manuscript. Moreover, senior scientist Ulla Ribel, Novo Nordisk A/S (Måløv, Denmark), nants of weight gain in the action to control cardiovascular risk in diabetes trial. is thanked for her discussions and guidance given in the early phase of the writing Diabetes Care 2013; 36: 2162–2168.

© 2015 Macmillan Publishers Limited International Journal of Obesity (2015) 1531 – 1538 Insulin therapy in a type 2 diabetes rat model S Skovsø et al 1538 9 Pontiroli AE, Miele L, Morabito A. Increase of body weight during the first year of 29 Conde-Sanchez M, Roldan-Fontana E, Chueca-Porcuna N, Pardo S, Porras-Gracia J. intensive insulin treatment in type 2 diabetes: systematic review and meta-ana- Analytical performance evaluation of a particle-enhanced turbidimetric cystatin C lysis. Diabetes Obes Metab 2011; 13: 1008–1019. assay on the Roche COBAS 6000 analyzer. Clin Biochem 2010; 43: 921–925. 10 Shah PK, Mudaliar S, Chang AR, Aroda V, Andre M, Burke P et al. Effects of 30 Association AD. Standards of medical care in diabetes--2014. Diabetes Care 2014; intensive insulin therapy alone and in combination with pioglitazone on body 37 (Suppl 1): S14–S80. weight, composition, distribution and liver fat content in patients with type 2 31 Juurinen L, Tiikkainen M, Hakkinen AM, Hakkarainen A, Yki-Jarvinen H. Effects of diabetes. Diabetes Obes Metab 2011; 13:505–510. insulin therapy on liver fat content and hepatic insulin sensitivity in patients with 11 Yki-Jarvinen H, Ryysy L, Kauppila M, Kujansuu E, Lahti J, Marjanen T et al. Effect of type 2 diabetes. Am J Physiol Endocrinol Metab 2007; 292: E829–E835. obesity on the response to insulin therapy in noninsulin-dependent diabetes 32 Son JW, Jeong HK, Lee SS, Kim SR, Cha BY, Son HY et al. The effect of early mellitus. J Clin Endocrinol Metab 1997; 82: 4037–4043. intensive insulin therapy on body fat distribution and beta-cell function in newly 12 Field AE, Coakley EH, Must A, Spadano JL, Laird N, Dietz WH et al. Impact of diagnosed type 2 diabetes. Endocr Res 2013; 38:160–167. overweight on the risk of developing common chronic diseases during a 10- 33 Takei I, Takayama S, Yamauchi A, Nakamoto S, Kitamura Y, Katsukawa F et al. year period. Arch Inter Med 2001; 161: 1581–1586. Effect of insulin therapy on body fat distribution in NIDDM patients with sec- 13 Hubert HB, Feinleib M, McNamara PM, Castelli WP. Obesity as an independent risk ondary sulfonylurea failure: a preliminary report. Eur J Clin Nutr 1998; 52: 153–154. factor for cardiovascular disease: a 26-year follow-up of participants in the Fra- 34 Elsayed A, ElBebely S, Galal A. Insulin therapy induced adposity evaluated by mingham Heart Study. Circulation 1983; 67:968–977. computed tomography is not visceral. Pak J Med Sci 2007; 23:161–166. 14 Van Gaal LF, Mertens IL, De Block CE. Mechanisms linking obesity with cardio- 35 Tang A, Rabasa-Lhoret R, Castel H, Wartelle-Bladou C, Gilbert G, Massicotte-Tisluck vascular disease. Nature 2006; 444:875–880. K et al. Effects of Insulin Glargine and Liraglutide Therapy on Liver Fat As Mea- 15 Yusuf S, Hawken S, Ounpuu S, Bautista L, Franzosi MG, Commerford P et al. sured by Magnetic Resonance in Patients With Type 2 Diabetes: A Obesity and the risk of myocardial infarction in 27,000 participants from 52 Randomized Trial. Diabetes Care 2015; e-pub ahead of print 26 March 2015. countries: a case-control study. Lancet 2005; 366:1640–1649. 36 Bolinder J, Kager L, Ostman J, Arner P. Differences at the receptor and post- 16 Davies MJ, Derezinski T, Pedersen CB, Clauson P. Reduced weight gain with insulin receptor levels between human omental and subcutaneous adipose tissue in the detemir compared to NPH insulin is not explained by a reduction in hypoglyce- action of insulin on lipolysis. Diabetes 1983; 32:117–123. mia. Diabetes Technol Ther 2008; 10:273–277. 37 Kraunsoe R, Boushel R, Hansen CN, Schjerling P, Qvortrup K, Stockel M et al. 17 Fajardo Montanana C, Hernandez Herrero C, Rivas Fernandez M. Less weight gain Mitochondrial respiration in subcutaneous and visceral adipose tissue from and hypoglycaemia with once-daily insulin detemir than NPH insulin in intensi- patients with morbid obesity. J Physiol 2010; 588 (Pt 12): 2023–2032. fication of insulin therapy in overweight Type 2 diabetes patients: the PREDICTIVE 38 Livingston JN, Lerea KM, Bolinder J, Kager L, Backman L, Arner P. Binding and BMI clinical trial. Diabetic Med 2008; 25: 916–923. molecular weight properties of the insulin receptor from omental and sub- 18 Rosenstock J, Lorber DL, Gnudi L, Howard CP, Bilheimer DW, Chang PC et al. cutaneous adipocytes in human obesity. Diabetologia 1984; 27: 447–453. Prandial inhaled insulin plus basal insulin glargine versus twice daily biaspart insulin 39 Marin P, Andersson B, Ottosson M, Olbe L, Chowdhury B, Kvist H et al. The for type 2 diabetes: a multicentre randomised trial. Lancet 2010; 375: 2244–2253. morphology and metabolism of intraabdominal adipose tissue in men. Metabo- 19 Skovsø S. Modeling type 2 diabetes in rats using high fat diet and streptozotocin. lism 1992; 41: 1242–1248. J Diabetes Invest 2014; 5:349–358. 40 Zierath JR, Livingston JN, Thorne A, Bolinder J, Reynisdottir S, Lonnqvist F et al. 20 Vague J. The degree of masculine differentiation of obesities: a factor determining Regional difference in insulin inhibition of non-esterified fatty acid release from predisposition to diabetes, atherosclerosis, gout, and uric calculous disease. 1956. human adipocytes: relation to insulin receptor phosphorylation and intracellular Nutrition 1999; 15:89–90. signalling through the insulin receptor substrate-1 pathway. Diabetologia 1998; 21 Vague J. Sexual differentiation. A determinant factor of the forms of obesity. 1947. 41: 1343–1354. Obesity Res 1996; 4: 201–203. 41 Kaess BM, Pedley A, Massaro JM, Murabito J, Hoffmann U, Fox CS. The ratio of 22 Demerath EW, Reed D, Rogers N, Sun SS, Lee M, Choh AC et al. Visceral adiposity visceral to subcutaneous fat, a metric of body fat distribution, is a unique cor- and its anatomical distribution as predictors of the metabolic syndrome and relate of cardiometabolic risk. Diabetologia 2012; 55: 2622–2630. cardiometabolic risk factor levels. Am J Clin Nutr 2008; 88: 1263–1271. 42 Galland F, Duvillard L, Petit JM, Lagrost L, Vaillant G, Brun JM et al. Effect of insulin 23 Klein S, Fontana L, Young VL, Coggan AR, Kilo C, Patterson BW et al. Absence of an treatment on plasma oxidized LDL/LDL-cholesterol ratio in type 2 diabetic effect of liposuction on insulin action and risk factors for coronary heart disease. N patients. Diabetes Metab 2006; 32:625–631. Engl J Med 2004; 350: 2549–2557. 43 Li F, Zhao T, Wen X. Changes in serum adiponectin concentrations and endo- 24 Marques MD, Santos RD, Parga JR, Rocha-Filho JA, Quaglia LA, Miname MH et al. thelial function after intensive insulin treatment in people with newly diagnosed Relation between visceral fat and coronary artery disease evaluated by multi- type 2 diabetes: a pilot study. Diabetes Res Clin Pract 2011; 94:186–192. detector computed tomography. Atherosclerosis 2010; 209: 481–486. 44 Hafiane A, Genest J. HDL, atherosclerosis, and emerging therapies. Cholesterol 25 Newton Jr RL, Bouchard C, Bray G, Greenway F, Johnson WD, Ravussin E et al. 2013; 2013: 891403. Abdominal adiposity depots are correlates of adverse cardiometabolic risk factors 45 Varbo A, Benn M, Tybjaerg-Hansen A, Jorgensen AB, Frikke-Schmidt R, Nordest- in Caucasian and African-American adults. Nutr Diabetes 2011; 1: e2. gaard BG. Remnant cholesterol as a causal risk factor for ischemic heart disease. 26 Manolopoulos KN, Karpe F, Frayn KN. Gluteofemoral body fat as a determinant of J Am Coll Cardiol 2013; 61: 427–436. metabolic health. Int J Obes (Lond) 2010; 34:949–959. 46 Sparks JD, Shaw WN, Corsetti JP, Bolognino M, Pesek JF, Sparks CE. Insulin-treated 27 Guillen J. FELASA guidelines and recommendations. J Am Assoc Lab Anim Sci 2012; Zucker diabetic fatty rats retain the hypertriglyceridemia associated with obesity. 51: 311–321. Metabolism 2000; 49: 1424–1430. 28 van Gammeren AJ, van Gool N, de Groot MJ, Cobbaert CM. Analytical perfor- 47 Erion DM, Ignatova ID, Yonemitsu S, Nagai Y, Chatterjee P, Weismann D et al. mance evaluation of the Cobas 6000 analyzer—special emphasis on trueness Prevention of hepatic steatosis and hepatic insulin resistance by knockdown of verification. Clin Chem Lab Med 2008; 46:863–871. cAMP response element-binding protein. Cell Metab 2009; 10: 499–506.