Impact of Perturbed Pancreatic β-Cell Cholesterol Homeostasis on Adipose Tissue and Skeletal Muscle Metabolism

The ATP-binding cassette transporters ABCA1 and ABCG1 regulate cell cholesterol homeostasis by exporting cellular cholesterol to extracellular acceptors (1–4). In the absence of ABCA1 and ABCG1, cell cholesterol levels increase (5,6) and cell function is impaired. In the case of pancreatic β-cells, increased cholesterol levels result in reduced insulin secretion (7). Loss-of-function mutations in the human ABCA1 gene are also associated with cellular cholesterol accumulation, impaired glucose tolerance, and decreased insulin secretion but do not affect insulin sensitivity (8).

Increased adiposity, insulin resistance, impaired glucose tolerance, and/or impaired fasting glucose are all associated with a compensatory increase in β-cell mass and increased insulin secretion. This can eventually lead to β-cell failure, decreased insulin secretion, and development of type 2 diabetes mellitus (T2DM) (9,10). Decreased β-cell insulin secretion can precede the development of insulin resistance in individuals who are genetically predisposed to T2DM (11–14), in obese adolescents (15–19), in people of Japanese descent (20–22), in Pima Indian populations (23) and in people with loss-of-function mutations in the ABCA1 gene (24). These studies point to an impaired insulin response as an independent predictor of diabetes (25,26).

Previous studies have shown that mice with β-cell–specific deletion of ABCA1, alone (27,28) or together with global deletion of ABCG1 (6), have elevated β-cell cholesterol levels, reduced insulin secretion, and impaired glucose tolerance. However, because ABCA1 deletion in isolation leads to a compensatory elevation in ABCG1 expression (29) and ABCG1 knockout mice have very low adipose tissue mass and do not become glucose intolerant or insulin resistant when challenged with a high-fat diet (30), these studies have not provided insights into how β-cell dysfunction caused by perturbations in cholesterol homeostasis affects insulin target tissues, including adipose tissue and skeletal muscle. To better understand how dysregulated β-cell cholesterol homeostasis affects adipose tissue and skeletal muscle metabolism, we have generated mice with deletion of ABCA1 and ABCG1 only in β-cells (β-DKO mice).

β-DKO mice were generated by crossing B6.Cg-Tg(Ins2-cre)25Mgn/J (The Jackson Laboratory) and Abca1fl/flAbcg1fl/fl mice (31). All experiments were done in 16-week-old male mice, unless specified otherwise. The animals were maintained on a standard chow diet in a specific pathogen-free laboratory with a 12-h light/dark cycle. Food intake of 16-week-old mice was measured manually every 24 h over 3 consecutive days. Blood from the lateral tail vein was used for determining glucose levels (Accu-Chek Performa glucometer; Roche). Plasma interleukin (IL)-6 and MCP-1 levels were determined by ELISA (R&D Systems).

Pancreatic islets were isolated from 16-week-old mice as previously described (32), washed with PBS, and lysed with water. Total cholesterol levels in the cell lysates were quantified by high-performance liquid chromatography apparatus (33).

For glucose tolerance tests, mice were fasted for 5 h before an intraperitoneal injection of d-glucose (2 g/kg). For pyruvate tolerance tests, mice were fasted for 18 h before an intraperitoneal injection of sodium pyruvate (2 g/kg). For insulin tolerance tests, randomly fed mice received an intraperitoneal injection of 1 unit of human insulin per kilogram. Blood glucose levels (tail vein) were monitored at the indicated times.

Mice were fasted for 5 h before an intraperitoneal injection of d-glucose (3 g/kg) or arginine (1 g/kg). Blood was sampled from the lateral tail vein at 0, 5, 10, and 15 min. Plasma insulin concentrations were determined by ELISA (Merck Millipore).

Pancreata and epididymal adipose tissue were fixed with formalin and embedded in paraffin. Islets were visualized with an anti-insulin antibody (Cell Signaling) and an EnVision+ System-HRP kit (Dako). Adipose tissue sections were immunostained with hematoxylin and eosin for cell area determination or with an anti-F4/80 antibody for macrophage quantification (Abcam). Immunostained sections were quantified using the EnVision+ System-HRP kit. β-Cell mass was determined as previously described (34). Adipocyte area was determined using ImageJ software (National Institutes of Health).

The percentage of body fat was determined using an EchoMRI-900 Body Composition Analyzer (EchoMRI, Biological Resources Imaging Laboratory, Mark Wainwright Analytical Centre, University of New South Wales, Sydney, Australia). The percentage body fat was calculated as ([body fat mass]/[body fat mass + lean mass]) × 100.

Fasted mice were injected with [³H]-2-deoxy-glucose (50 µCi/mouse) in 20% glucose (2 g/kg i.p.) and sacrificed after 2 h. Skeletal muscle and epididymal adipose tissue were harvested and frozen in liquid nitrogen until analysis. Tissues were homogenized in 0.5% (volume for volume [v/v]) perchloric acid and centrifuged at 2,000g for 20 min at 4°C. The supernatant was adjusted to pH 7.5 with KOH and split into two aliquots. Then, 2-deoxy-glucose phosphate was precipitated from one aliquot with 0.5 volume each of 0.3 mol/L BaOH and 0.3 mol/L ZnSO₄ and centrifuged at 16,000g for 5 min. Glucose uptake (as 2-deoxy-glucose phosphate) was quantified by liquid scintillation counting of the supernatant and the untreated aliquot and calculated as the difference between the two values (35).

Epididymal adipose tissue and liver samples (100 mg) were homogenized in ice-cold PBS containing 0.25 mol/L sucrose, 1 mmol/L dithiothreitol, 1 mmol/L EDTA-Na₂, and protease inhibitor (36). The samples were centrifuged at 20,000g for 10 min at 4°C, and then at 100,000g for 1 h at 4°C. The infranatant was mixed 1:1 (v/v) with 500 mmol/L phosphate buffer containing 0.5 mmol/L dithiothreitol (pH 7.4) and activated for 30 min at 37°C. The activated solution was added to two volumes of assay buffer (500 mmol/L potassium phosphate, 0.25 mmol/L NADPH, 0.1 mol/L EDTA-Na₂, and 1 mmol/L β-mercaptoethanol, pH 7), and incubated briefly. The reaction was started by adding substrate solution (10 µL of 5 mmol/L malonyl-CoA mixed 4:3 [v/v] with 5 mmol/L acetyl-CoA). The absorbance (340 nm) was measured continuously for 20 min. Fatty acid synthase activity was defined as 1 nmol NADPH consumed/min/mg protein using a molar (M) extinction coefficient of 6,220/M/cm.

Livers were dissected, weighed, and snap frozen. Before analysis, the samples were placed on ice and homogenized in water. Glycogen levels were determined using a glycogen assay kit (Sigma-Aldrich). Plasma lactate levels were determined using a lactate assay kit (Abcam).

Epididymal adipose tissue was isolated and homogenized with ice-cold radioimmune precipitation assay buffer. Cell lysates were electrophoresed on 4% homogeneous or 4–12% gradient SDS-PAGE gels for detection of fatty acid synthase and β-actin, respectively, and transferred to nitrocellulose membranes. The membranes were probed with monoclonal anti-rabbit fatty acid synthase (1:1,000; Cell Signaling) or anti-mouse β-actin (1:2,000; Abcam) primary antibodies and polyclonal sheep–anti-rabbit or sheep–anti-mouse horseradish peroxidase secondary antibodies (1:5,000; Abcam), developed with ECL Prime (GE Healthcare Life Sciences), and imaged using an ImageQuant LAS-4000 Mini (GE Healthcare Life Sciences). Band intensities were quantified using ImageJ.

Total RNA was isolated from adipocytes (miRNeasy Mini kit; Qiagen) and reverse transcribed using 0.1 µg of RNA. Real-time PCR was performed with the MyiQ single-color real-time PCR detection system, using iQ SYBR Green Supermix (Bio-Rad) for relative mRNA quantification of the acetyl-CoA carboxylase-α (Acaca) and malonyl-CoA decarboxylase (Mlycd) genes. Values were normalized to GAPDH and expressed as 2^{−(Ct[gene of interest]−Ct[GAPDH])}.

A circular polarized whole-body radiofrequency 50-mm volume quadrature coil was used for radiofrequency transmission and reception. For all MRI procedures, general anesthesia was induced with 4% isoflurane in oxygen (1 L/min) and maintained with 2–2.5% isoflurane. The animals were placed prone on the animal bed frame. The head was stabilized with a tooth-bar, and the liver region was centered within the scanner. A pressure-sensitive pad was used to monitor respiratory motion during imaging. The monitoring signal was used to trigger all imaging procedures to minimize image artifacts from abdominal motion. Animal body temperature was maintained at 36°C with a temperature-controlled circulating water warming blanket.

Images of the entire abdominal region were acquired in axial slice orientation using an optimized two-dimensional T1-weighted rapid acquisition with relaxation enhancement (TurboRARE) sequence. Parameters for the axial acquisitions were effective echo time = 8.5 ms, repetition time = 1,600 ms, field of view = 40 × 40 mm, acquisition matrix = 256 × 256 mm, in plane resolution = 156 × 156 µm, slice thickness = 1.0 mm, slice gap = 0.5 mm, number of slices = 40, resulting volume coverage = 59 mm, and 8 average, resulting in a total scan time of 10 min 20 s. All acquisitions were triggered “per phase step” on the breathing monitoring signal. To selectively highlight and identify fat localization, two identically positioned MRI volumes were acquired subsequently with and without fat presaturation but with otherwise identical imaging parameters. A fat suppression scheme using a Gaussian pulse with fat suppression bandwidth of 140 Hz, and 2 ms was used for the fat-saturated scan.

Mice were sacrificed by cervical dislocation, and gastrocnemius muscles were immediately dissected, weighed, snap frozen in liquid nitrogen, and stored at −80°C until processed. The samples (100 mg) were homogenized in ice-cold radioimmune precipitation assay buffer. Cell lysates were clarified by centrifugation at 10,000g for 10 min at 4°C. Proteins remaining in solution were precipitated by incubation with four volumes of ice-cold acetone and collected by centrifugation at 13,000g for 10 min at 4°C. The cell pellets were resuspended in 50 mmol/L ammonium bicarbonate (pH 8.0), and protein concentrations were quantified by bicinchoninic acid assay (Life Technologies).

Whole-cell protein extracts (100 µg) were adjusted to 20 µL with 50 mmol/L NH₄HCO₃ (pH 8). Trypsin (Promega) was added (final enzyme-to-protein ratio was 1:100 [w/w]) and incubated overnight at 37°C. The reaction was stopped by adjustment to ∼pH 3 with neat formic acid, and the samples were dried (vacuum centrifuge) and stored at −20°C until analyzed. The resulting peptides were purified using C18 StageTips (Thermo Fisher Scientific).

Peptides were reconstituted in 10 µL of 0.1% formic acid and separated by nano-LC using an Ultimate3000 high-performance liquid chromatography apparatus and autosampler (Dionex). Samples (0.2 µL, ∼2 µg in total) were loaded onto a micro C18 precolumn (500 µm–2 mm; Michrom Bioresources) with Buffer C (97.95% H₂O, 1.95% CH₃CN, 0.1% trifluoroacetic acid) at a flow rate of 10 µL/min. After a 4-min wash, the precolumn was switched (Valco 10-port valve; Dionex) into line with a fritless nano column (75 µm internal diameter, 12 cm) containing reverse-phase C18 media (3 µm, 200 Å Magic, Michrom Bioresources). The peptides were eluted using a linear gradient of Buffer C to Buffer D (97.95% CH₃CN, 1.95% H₂O, 0.1% formic acid) at 250 nL/min over 1 h. High voltage (2,000 V) was applied to a low volume tee (Upchurch Scientific) and the column tip positioned 0.5 cm from the heated capillary (T = 280°C) of an Orbitrap Velos (Thermo Electron, Thermo Fisher Scientific) mass spectrometer. Positive ions were generated by electrospray, and the Orbitrap operated in data-dependent acquisition mode. A survey scan mass (m)-to-charge (z) ratio of 350–1,750 was acquired (resolution = 30,000 at an m-to-z ratio of 400, with an accumulation target value of 1,000,000 ions) with lock mass enabled. Up to the 10 most abundant ions (45,000 counts) with charge states +2 to +4 were sequentially isolated and fragmented within the linear ion trap using collisionally induced dissociation with an activation q = 0.25 and activation time of 30 ms at a target value of 30,000 ions. The m-to-z ratio selected for tandem mass spectrometry (MS/MS) was dynamically excluded for 30 s. MS peak intensities were analyzed using Progenesis LC–MS 4 data analysis software (Nonlinear Dynamics). Ion intensity maps from each run were aligned to a reference sample. Ion feature matching was achieved by aligning consistent ion m-to-z ratio and retention times. The peptide intensities were normalized against total intensity (sample specific log-scale abundance ratio-scaling factor) and compared by ANOVA. Type I errors were controlled for by false discovery rate with q value for significance set at 0.01 (37). MS/MS spectra were searched against the National Center for Biotechnology Information nonredundant protein database using MASCOT (Matrix Science). Parent and fragment ions were searched with tolerances of 6 ppm and 0.6 Da, respectively, with mouse proteins selected and “no-enzyme” specified.

Mini osmotic pumps (Model 1004, Alzet) loaded with 33.34 units/mL Humulin R (corresponding to a dose of 0.1 units/day; Eli Lily) or an equivalent volume of PBS were inserted subcutaneously between the shoulder blades of anesthetized 12-week-old β-DKO mice (38). After sacrifice at 16 weeks of age, the percentage of body fat was determined by EchoMRI. Adipocyte area and adipose tissue macrophage content were quantified using ImageJ (39).

Results are presented as mean ± SD. Significant differences between experimental groups were determined using an unpaired, two-tailed Student t test or one-way ANOVA with Tukey posttest, as appropriate. P < 0.05 was considered as significant.

β-DKO mice were generated by crossing B6.Cg-Tg(Ins2-Cre)25Mgn/J (Ins2Cre) and Abca1fl/flAbcg1fl/fl mice (31). ABCA1 and ABCG1 protein levels in the β-DKO mice islets were decreased by 82.0 ± 5.1% and 84.4 ± 1.2%, respectively, relative to Abca1fl/flAbcg1fl/fl mice (Fig. 1A and B).

The body weight (Fig. 2A) and food intake (Fig. 2B) of the β-DKO mice were comparable to that of the Abca1fl/flAbcg1fl/fl mice and Ins2Cre mice. Although there were no differences in fed or fasting blood glucose levels in 6-, 12-, or 16-week-old β-DKO, Abca1fl/flAbcg1fl/fl, and Ins2Cre mice (Fig. 2C), values for the β-DKO mice were more variable, especially under fasting conditions. Fed and fasting plasma insulin levels were both significantly decreased in 6-, 12-, and 16-week-old β-DKO mice compared with Abca1fl/flAbcg1fl/fl littermates (P < 0.001 for all) (Fig. 2D).

Intraperitoneal glucose tolerance tests revealed that the β-DKO mice were glucose intolerant relative to Abca1fl/flAbcg1fl/fl and Ins2Cre mice (incremental area under the curve: 589.3 ± 203.8, 608.5 ± 187.3, and 1,657.0 ± 272.7 for Abca1fl/flAbcg1fl/fl, Ins2Cre, and β-DKO mice, respectively; P < 0.001 for β-DKO vs. Abca1fl/flAbcg1fl/fl mice) (Fig. 2E and F). Pyruvate tolerance tests indicated normal gluconeogenesis in β-DKO mice (Fig. 2G). Liver mass and glycogen levels were also comparable in the β-DKO and Abca1fl/flAbcg1fl/fl mice (Supplementary Table 1). β-DKO mice also had normal insulin sensitivity (Fig. 2H). Fasting insulin levels were significantly lower in β-DKO mice (0.67 ± 0.19 ng/mL) compared with Abca1fl/flAbcg1fl/fl mice (1.03 ± 0.18 ng/mL, P < 0.001). The first phase of insulin secretion in response to both a glucose (Fig. 2I) and arginine (Fig. 2J) challenge was lost in the β-DKO mice.

Islet morphology and β-cell mass were normal in the β-DKO mice (Fig. 2K and L) despite islet cholesterol levels being elevated to 19.2 ± 1.9 mg/mg protein compared with 8.7 ± 1.0 mg/mg protein in Abca1fl/flAbcg1fl/fl mice (P < 0.001).

The 16-week-old, chow-fed β-DKO mice had strikingly increased visceral and subcutaneous fat deposition relative to Abca1fl/flAbcg1fl/fl mice (Fig. 3A). This was confirmed by MRI (Fig. 3B), which showed that 16-week-old β-DKO mice had 16.2 ± 1.3% body fat compared with 10.5 ± 0.9% for Abca1fl/flAbcg1fl/fl mice (P < 0.001) (Fig. 3C). Because the body weights of the β-DKO and Abca1fl/flAbcg1fl/fl mice were comparable (Fig. 2A), this suggested that the adipose tissue expansion may have been at the expense of a reduction in skeletal muscle mass.

This was confirmed by establishing that hind limb gastrocnemius muscle mass was decreased to 87.0 ± 11.4 mg in the β-DKO mice compared with 123.8 ± 17.6 mg in Abca1fl/flAbcg1fl/fl mice (P < 0.005) (Fig. 3D and E). The increased percentage of body fat in the β-DKO mice was caused by adipocyte hypertrophy, with an average adipocyte area in the β-DKO mice of 4,445.7 ± 514.3 µm² compared with 3,386 ± 439.7 µm² for Abca1fl/flAbcg1fl/fl mice (P < 0.001) (Fig. 3F).

Macrophage infiltration was apparent in the β-DKO mouse adipose tissue, with 1.71 ± 1.056% of the cells staining positive for F4/80+ compared with 0.22 ± 0.38% in the Abca1fl/flAbcg1fl/fl mice (P < 0.01) (Fig. 4A and B). Plasma IL-6 (Fig. 4C) and MCP-1 (Fig. 4D) levels were also increased in the β-DKO mice (11.0 ± 3.2 and 57.1 ± 4.0 pg/mL, respectively) relative to Abca1fl/flAbcg1fl/fl mice (2.8 ± 0.2 and 11.4 ± 5.6 pg/mL, respectively; P < 0.005 for IL-6 and P < 0.001 for MCP-1).

We next asked whether the increased adiposity and decreased muscle mass in the β-DKO mice could be explained by the limited availability of insulin (Fig. 2D, I, and J) causing glucose disposal to be redirected from skeletal muscle to adipose tissue. This was found to be the case, with the uptake of [³H]-2-deoxyglucose by skeletal muscle reduced to 176.4 ± 23.1 nmol/g/min in the β-DKO mice compared with 241.1 ± 25.6 nmol/g/min in Abca1fl/flAbcg1fl/fl mice (P < 0.01) (Fig. 5A). [³H]-2-deoxyglucose uptake into epididymal adipose tissue was concomitantly increased from 24.5 ± 5.4 nmol/g/min in Abca1fl/flAbcg1fl/fl mice to 36.7 ± 8.9 nmol/g/min in β-DKO mice (P < 0.05) (Fig. 5B).

The underlying cause of the preferential disposal of glucose in adipose tissue in β-DKO mice was also investigated. Hepatic fatty acid synthase activity was comparable in the β-DKO and control mice (Fig. 5C). Fatty acid synthase activity was significantly elevated in β-DKO mouse adipose tissue (7.77 ± 0.50 nmol/min/mg compared with 5.10 ± 1.70 nmol/min/mg for Abca1fl/flAbcg1fl/fl mice and 3.84 ± 0.31 nmol/min/mg protein for Ins2Cre mice; P < 0.05 β-DKO vs. Abca1fl/flAbcg1fl/fl, and P < 0.01 β-DKO vs. Ins2Cre) (Fig. 5D) Despite the increased adipose tissue fatty acid synthase activity, adipocyte fatty acid synthase protein levels were decreased in the β-DKO mice relative to the Abca1fl/flAbcg1fl/fl mice (Fig. 5E and F).

Protein levels of the fatty acid synthase substrate, malonyl-CoA, and mRNA levels of acetyl-CoA carboxylase-α (ACACA), the enzyme that converts acetyl-CoA to malonyl-CoA and malonyl-CoA decarboxylase (MLYCD), which converts malonyl-CoA back into acetyl-CoA, were also determined. Adipocyte malonyl-CoA levels were decreased in β-DKO mice relative to Abca1fl/flAbcg1fl/fl mice (2.73 ± 0.943 vs 8.03 ± 2.74 ng/mg protein, respectively; P < 0.01) (Fig. 5G) Adipocyte MLYCD mRNA levels were, by contrast, increased in the β-DKO mice (Fig. 5H). There was a trend toward reduced ACACA mRNA levels in the β-DKO mice, but this did not reach statistical significance (Fig. 5I).

The effect of decreased glucose uptake on skeletal muscle metabolism was investigated by MS. Significant changes were observed in the expression of 53 proteins (Supplementary Table 2 and Supplementary Fig. 1). Ingenuity Pathway analysis of the differentially expressed proteins revealed significant changes in a number of skeletal muscle enzymes involved in glycolysis, gluconeogenesis, and amino acid metabolism in β-DKO mice relative to Abca1fl/flAbcg1fl/fl mice (Table 1).

To ascertain whether the increased adiposity in the 16-week-old β-DKO mice was caused by their suboptimal plasma insulin levels (Fig. 2D, I, and J), osmotic pumps were loaded with insulin (Humulin R, 0.1 units/day) or PBS (control) and implanted in two groups of 12-week-old β-DKO mice (38). The animals were sacrificed 4 weeks postpump implantation. No difference in body weight was observed between the insulin-supplemented and PBS-supplemented β-DKO mice (Fig. 6A). At 1 week postpump implantation, plasma insulin levels in the insulin-supplemented mice were 1.0 ± 0.2 ng/mL compared with 0.5 ± 0.1 ng/mL in the PBS-treated mice (P < 0.005) (Fig. 6B). Insulin levels in the insulin-supplemented mice remained constant and comparable to the insulin levels in the Abca1fl/flAbcg1fl/fl mice until sacrifice at 16 weeks of age. Insulin supplementation did not affect β-DKO mice blood glucose levels (Fig. 6C), but it completely ablated body fat accumulation. The insulin-supplemented β-DKO mice had 13.4 ± 2.3% body fat at 16 weeks of age compared with 18.0 ± 2.1% body fat for the PBS-treated β-DKO mice (P < 0.05) (Fig. 6D). Insulin supplementation also increased β-DKO mouse gastrocnemius muscle weight to 105.0 ± 8.9 mg compared with 85.8 ± 8.5 mg for PBS-treated β-DKO mice (P < 0.01) (Fig. 6E). Moreover, adipocyte area decreased from 4,380 ± 342 µm² in the PBS-treated β-DKO mice to 3,923 ± 213 µm² in the insulin-supplemented β-DKO mice (P < 0.01) (Fig. 6F). Insulin supplementation did not change β-cell mass (Fig. 6G), adipose tissue macrophage content (Fig. 6H), or plasma IL-6 (Fig. 6I) and MCP-1 levels (Fig. 6J). Although insulin tolerance was not altered in the insulin-supplemented mice (Supplementary Fig. 2A), glucose tolerance was significantly improved, most likely as a result of increased skeletal muscle mass (Supplementary Fig. 2B).

This study establishes that the increased β-cell cholesterol levels in β-DKO mice reduces insulin secretion in response to glucose and leads to glucose intolerance but does not cause insulin resistance. It also shifts glucose uptake from skeletal muscle to adipose tissue in a process that is accompanied by adipocyte hyperplasia, infiltration of macrophages into adipose tissue, and systemic inflammation.

Cholesterol is known to be toxic to β-cells (40), but the extent of β-cell cholesterol accumulation in β-DKO mice was clearly below the threshold that causes cell death. This is in contrast to what has been reported for mice transgenic for SREBP-1c, where increased β-cell cholesterol levels decreased islet size and number (41). Furthermore, islet hypertrophy was not evident in the β-DKO mice, which is consistent with the absence of insulin resistance (42).

Conditional β-cell deletion of ABCA1 and ABCG1 did not affect body weight or gross locomotion (42) but did increase body fat and decrease muscle mass relative to littermate controls. This is in contrast to reports of mice with β-cell–specific deletion of ABCA1, in isolation (28) or on a global ABCG1 knockout background (6), where body composition did not change. This discrepancy is most likely caused by a compensatory rise in ABCG1 expression in mice with β-cell–specific deletion of ABCA1 only (29), and ABCG1 deletion protecting animals with β-cell–specific deletion of ABCA1 from obesity (30).

The absence of hyperglycemia in β-DKO mice is consistent with other reports of normal blood glucose levels in mice with low plasma insulin levels (43–45). In the case of β-DKO mice, this phenotype is likely a result of normal insulin sensitivity combined with plasma insulin levels that are still sufficient to normalize blood glucose levels. However, increased variability in fasting blood glucose levels was observed in the β-DKO mice. This suggests that regulation of glucose homeostasis in these animals is impaired, which is consistent with the altered levels of skeletal muscle proteins involved in glycolysis and gluconeogenesis. Furthermore, because type 10,17 β-hydroxysteroid dehydrogenase (HSD17B10) and acetyl-CoA acetyltransferase (ACoAAT), which, respectively, catalyze the final steps in the conversion of isoleucine and methylated fatty acids to acetyl-CoA (46), were both upregulated in β-DKO mouse gastrocnemius muscle, it follows that these animals must use substrates other than glucose for energy production in skeletal muscle. In addition, the comparable glucose levels in β-DKO and control mice, despite significant differences in insulin levels, suggests that noninsulin-dependent glucose uptake is enhanced in β-DKO mouse skeletal muscle. Although no difference in AMPK activation between the β-DKO and control mice was observed (data not shown), it is possible that other pathways, such as Ca²⁺/calmodulin signaling, are involved.

A key finding from this study is that insulin insufficiency shifted glucose disposal from skeletal muscle to adipose tissue in β-DKO mice. This is probably related to the fact that approximately twice as much insulin is required to mediate the same amount of glucose uptake and metabolism in skeletal muscle than in adipose tissue (47). This suggests that when insulin availability is compromised, glucose uptake and metabolism in skeletal muscle are affected more severely than in adipose tissue. It is also consistent with altered skeletal muscle metabolism in the β-DKO mice and is in agreement with the observed increase in fatty acid synthase activity in adipose tissue. The increased fatty acid synthase activity in β-DKO mouse adipose tissue may be a reflection of enhanced enzyme dimerization in the hypertrophied adipocytes (48). The reduced adipocyte fatty acid synthase protein levels in the β-DKO mice, by contrast, are most likely a downstream consequence of insulin insufficiency, because insulin is required for the induction of fatty acid synthase gene transcription (49). Despite previous reports that malonyl-CoA levels rise during fatty acid synthesis (50), the level was decreased in adipocytes from β-DKO mice. This may be caused by increased incorporation of malonyl-CoA into fatty acids and increased MLYCD-mediated conversion of malonyl-CoA to acetyl-CoA. Interestingly, increased adiposity was not observed in insulin-insufficient, chow-fed Ins1^+/−Ins2^−/− mice (45), possibly as a result of increased energy expenditure and expression of uncoupling protein 1 in white adipose tissue.

Although inflammation is usually associated with increased circulating insulin levels and insulin resistance (51), the results in this study provide compelling evidence that inflammation can also be driven by adipose tissue expansion in the absence of insulin resistance. It is possible that insulin supplementation did not reduce adipose tissue inflammation in β-DKO mice because chronically elevating plasma insulin levels in these animals may enhance adipose tissue lipolysis (52), which increases macrophage recruitment (53). Moreover, the absence of insulin resistance in β-DKO mice with adipocyte hypertrophy and adipose tissue inflammation is in line with reports of insulin resistance being driven by lipid accumulation in nonadipose tissues (54,55).

In conclusion, this study shows that elevated β-cell cholesterol levels lead to β-cell dysfunction, increased adiposity, reduced skeletal muscle mass, and systemic inflammation. These findings indicate that β-cell dysfunction is a potential therapeutic target in people at risk for developing T2DM and possibly in individuals with maturity-onset diabetes of the young. Although insulin therapy increases body fat in people with established T2DM, this may not be the case for people with maturity-onset diabetes of the young who have β-cell dysfunction but are not insulin resistant. These individuals may derive benefit from insulin therapy as a means of averting weight gain and delaying disease progression.

Funding. M.W. was supported by VIDI grant 91715350 from the Netherlands Organisation for Scientific Research and a Rosalind Franklin Fellowship from the University of Groningen. This research was also supported by a National Health and Medical Research Council of Australia grant (APP1037903) to P.J.B. and K.-A.R. This work was partially funded by National Institutes of Health grant P01 HL087123 (to A.R.T.).

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. B.J.C. was responsible for the experimental design, performed the experiments and data analysis, and wrote the manuscript. L.H., A.P.C.M., B.M.M., F.T., A.S., L.C.T., S.T., S.S., P.S., and M.P. performed the experiments and data analysis. W.J.R., A.B., M.M., and V.C.W. were responsible for the experimental design. M.W., A.R.T., P.J.B., and K.-A.R. were responsible for the experimental design and wrote the manuscript. K.-A.R. is the guarantor of this work and, as such, has full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Prior Presentation. Parts of this study were presented in abstract form at the 83rd European Atherosclerosis Society Congress, Glasgow, Scotland, 22–25 March 2015; the 17th International Symposium on Atherosclerosis, Amsterdam, the Netherlands, 23–26 May 2015; and the Arteriosclerosis, Thrombosis and Vascular Biology 2014 Scientific Sessions, Toronto, Ontario, Canada, 1–3 May 2014.