In vivo PET imaging with [18F]FDG to explain improved glucose uptake in an apolipoprotein A-I treated mouse model of diabetes
- 1.4k Downloads
Type 2 diabetes is characterised by decreased HDL levels, as well as the level of apolipoprotein A-I (apoA-I), the main apolipoprotein of HDLs. Pharmacological elevation of HDL and apoA-I levels is associated with improved glycaemic control in patients with type 2 diabetes. This is partly due to improved glucose uptake in skeletal muscle.
This study used kinetic modelling to investigate the impact of increasing plasma apoA-I levels on the metabolism of glucose in the db/db mouse model.
Treatment of db/db mice with apoA-I for 2 h significantly improved both glucose tolerance (AUC 2574 ± 70 mmol/l × min vs 2927 ± 137 mmol/l × min, for apoA-I and PBS, respectively; p < 0.05) and insulin sensitivity (AUC 388.8 ± 23.8 mmol/l × min vs 194.1 ± 19.6 mmol/l × min, for apoA-I and PBS, respectively; p < 0.001). ApoA-I treatment also increased glucose uptake by skeletal muscle in both an insulin-dependent and insulin-independent manner as evidenced by increased uptake of fludeoxyglucose ([18F]FDG) from plasma into gastrocnemius muscle in apoA-I treated mice, both in the absence and presence of insulin. Kinetic modelling revealed an enhanced rate of insulin-mediated glucose phosphorylation (k 3) in apoA-I treated mice (3.5 ± 1.1 × 10−2 min−1 vs 2.3 ± 0.7 × 10−2 min−1, for apoA-I and PBS, respectively; p < 0.05) and an increased influx constant (3.7 ± 0.6 × 10−3 ml min−1 g−1 vs 2.0 ± 0.3 × 10−3 ml min−1 g−1, for apoA-I and PBS, respectively; p < 0.05). Treatment of L6 rat skeletal muscle cells with apoA-I for 2 h indicated that increased hexokinase activity mediated the increased rate of glucose phosphorylation.
These findings indicate that apoA-I improves glucose disposal in db/db mice by improving insulin sensitivity and enhancing glucose phosphorylation.
KeywordsDiabetes FDG Glucose uptake Insulin sensitivity Kinetic modelling
Positron emission tomography
Protein kinase A
Point spread function
Region of interest
Scavenger receptor class B type I
Type 2 diabetes is characterised by decreased HDL-cholesterol levels, as well as a decreased level of apolipoprotein A-I (apoA-I), the main HDL apolipoprotein . Therapeutic interventions that increase plasma HDL levels improve glycaemic control in patients with type 2 diabetes [2, 3], and treatment with HDL or apoA-I has been shown to increase glucose uptake into skeletal muscle [4, 5] and adipose tissue . Furthermore, treatment with HDL or apoA-I increases insulin secretion and production in isolated pancreatic islets and insulinoma beta cell lines [7, 8]. Recent studies have demonstrated that both single and repeated doses of apoA-I can also improve insulin resistance in fat-fed C57Bl6 mice [9, 10], as can overexpression of apoA-I .
Here, we have established a methodology using [18F]FDG and positron emission tomography/computed tomography (PET/CT) imaging, together with kinetic modelling, to quantify the in vivo, real-time impact of apoA-I treatment on glucose uptake into skeletal muscle. The db/db mouse is a well-established animal model of type 2 diabetes . While previous studies have been conducted using [18F]FDG to examine glucose uptake in db/db mice as an insulin resistant animal model, these studies only quantified endpoint uptake in the heart  and cardiac and skeletal muscle  and did not extend to determination of kinetics. Kinetic modelling of PET dynamic studies permits analysis beyond simple characterisation of global glucose uptake and determines the glucose influx and phosphorylation rates in specific tissues.
B6/BKS(D)-Lepr db /J (db/db) mice were obtained from The Jackson Laboratory (Bar Harbor, ME, USA). The experiments were conducted on 6-week-old male homozygous mice maintained on standard chow in pathogen-free conditions. All procedures were approved by the Sydney Local Health District and University of Sydney Animal Ethics Committees and followed the NIH Guide for the care and use of laboratory animals, Eighth edition (2011). Mice were randomly assigned to treatment groups by coin flip.
Mice received a single i.p. injection of human apoA-I (40 mg/kg, CSL Pty, Parkville, Australia) or an equivalent volume of PBS (n = 5/group). Blood was sampled from the lateral tail vein over 24 h and plasma apoA-I concentrations were determined by an ELISA that was specific for human apoA-I (Cell Biolabs, San Diego, CA, USA).
Glucose tolerance test
Mice were fasted for 3 h prior to receiving human apoA-I (40 mg/kg i.p.) or an equivalent volume PBS (n = 7/group). After a further 2 h, glucose (1 g d-glucose/kg; Sigma-Aldrich, St Louis, MO, USA) was injected i.p. Blood glucose levels were determined using an Accu-Chek Performa glucometer (Roche, Basel, Switzerland).
Insulin tolerance test
Random fed mice received a single i.p. injection of human apoA-I (40 mg/kg) or PBS (n = 6/group). After a further 2 h insulin (3 U human insulin/kg; Sigma-Aldrich) was injected i.p. and blood glucose levels measured as above.
Ex vivo skeletal muscle glucose uptake
Fasted mice were treated with apoA-I (40 mg/kg i.p.) or PBS (n = 6/group) and euthanised 2 h later. The gastrocnemius muscle was removed and 2-deoxy[3H]glucose (PerkinElmer, Waltham, MA, USA) uptake was quantified . The muscle was incubated for 1 h at 37°C with or without insulin (0.3 μmol/l), transferred to fresh buffer containing 1 mmol/l 2-deoxy[3H]glucose (37 kBq/ml) and 7 mmol/l d-[14C]-mannitol (14.8 kBq/ml; PerkinElmer) with or without insulin (0.3 μmol/l) for 10 min. Glucose uptake was terminated by immersion in ice-cold buffer containing 80 μmol/l cytochalasin B (Sigma-Aldrich). White and red muscle was isolated, snap frozen in liquid nitrogen and stored at −80°C for further use. Muscle samples were homogenised, and the 3H and 14C content determined by liquid scintillation counting.
Small-animal PET study
Fasted mice received an i.p. injection of apoA-I (40 mg/kg) or PBS (n = 4/group). After 2 h, the mice were anaesthetised with isoflurane administered via precision vaporiser (induction, 5%; maintenance, 1.5–2%), and maintained in a surgical plane of anaesthesia for the duration of the procedure. For mice treated with insulin, a dose of 3 U human insulin/kg was administered s.c. 30 min before the start of the PET/CT scan. After positioning the mice in the gantry of an Inveon PET/CT Scanner (Siemens, Munich, Germany), a 60 min PET scan commenced immediately after injection of [18F]FDG (10–18 MBq, pH 6.6–7.0, 0.1 ml; PETNET Solutions, Lucas Heights, Australia) via the lateral tail vein. After completing the PET scan, a CT scan was performed to allow the co-registration of radiotracer uptake with tissues.
Image reconstruction was performed using IAW 2.02 (Siemens). The listmode data was histogrammed into 16 frames (6 × 10 s, 4 × 60 s, 1 × 300 s, 5 × 600 s) for the period 0–60 min after tracer injection. Emission sinograms were reconstructed using 2D-filter back projection with a zoom of 1.5. The reconstructed images consisted of 16 frames of a 128 × 128 × 159 matrix with a voxel size of 0.52 × 0.52 × 0.796 mm3. They were corrected for attenuation (CT-based), scatter, randoms, normalisation, isotope decay, branching ratio, deadtime and were calibrated to Bq/ml.
[18F]FDG region of interest
Regions of interest (ROIs) (gastrocnemius muscle and inferior vena cava) were drawn on the image volumes using Brain-Visa/Anatomist version 3.1 (http://brainvisa.info/web/index.html) and time–activity curves were extracted from the dynamic frames. Radioactivity uptake values were quantified in Bq per cubic cm of tissue, and converted into the percentage of injected dose per cubic cm.
The FDG two-tissue compartment method with k 4 = 0 was used for modelling. The blood, plasma and gastrocnemius total activity counts were loaded into PMOD, version 3.17 (PMOD Technologies, Zürich, Switzerland) and the FDG model fitted. During fitting of the parameters, the V b (blood volume fraction) was initially fixed at 2%, and dispersion effect  was applied to the input functions as proposed by Oikonen . The optimum value for dispersion was found by minimising the χ2 value for the FDG model. A final fit was made using a floating V b and the optimised dispersion value, this yielded the rate constants (k 1–k 3) and influx constant was calculated as K i = (k 1 × k 3) / (k 2 + k 3). The researcher conducting the kinetic modelling was blinded to treatment groups.
L6 rat skeletal muscle cells were cultured at 37°C in a humidified, 5% CO2 atmosphere with α-MEM medium supplemented with 10% (vol./vol.) FCS. Cells were differentiated as previously described .
Hexokinase activity assay
Cells were serum starved for 6 h prior to treatment with apoA-I (final concentration 1 mg/ml) or equivalent volume PBS. After 1.5 h, cells were treated with insulin (final concentration 0.1 μmol/l) or equivalent volume PBS for a further 30 min. Hexokinase activity was determined by measuring NADH production in lysates after 30 min incubation period with a hexokinase assay kit (Abcam, Cambridge, MA, USA).
GLUT4 translocation was measured as described previously . L6 cells were incubated at 37°C for 2 h with or without apoA-I (1 mg/ml). Primary anti-GLUT4, specific for the extracellular domain (1:100, Santa Cruz Biotechnology, Dallas, TX, USA), and secondary DyLight488 donkey anti-goat (1:500, Abcam) antibodies were incubated together for 10 min at room temperature. Cells were washed with PBS and incubated with the antibody complex in the presence and absence of insulin (0.1 μmol/l) for 30 min. Cells were analysed using a FACSverse flow cytometer (BD Biosciences, San Jose, CA, USA).
All data presented are mean ± SD. Data was analysed using Prism 6 (GraphPad, La Jolla, CA, USA). A p value <0.05 was considered statistically significant. Glucose and insulin tolerance tests were compared using unpaired, two-tailed t tests. Mann–Whitney tests were used to compare the kinetic parameters between groups. Hexokinase activity and GLUT4 translocation experiments were compared by ANOVA and a Tukey’s multiple comparison test.
ApoA-I increases glucose tolerance and improves insulin sensitivity in db/db mice
Treatment with apoA-I improved glucose tolerance in the db/db mice (white circles) compared with saline (154 mmol/l NaCl)-treated animals (black squares) (AUC 2574 ± 70 mmol/l × min vs 2927 ± 137 mmol/l × min for apoA-I and PBS treatment, respectively; p < 0.05) (Fig. 1b). When subjected to an insulin tolerance test, blood glucose levels (black squares) decreased by 25%, from 10.83 ± 1.98 mmol/l to 8.13 ± 1.97 mmol/l in the PBS-treated mice and by 43%, from 9.62 ± 0.64 mmol/l to 5.52 ± 1.09 mmol/l, in the apoA-I-treated mice (white circles) (AUC 388.8 ± 23.8 mmol/l × min vs 194.1 ± 19.6 mmol/l × min for apoA-I and PBS treatment, respectively; p < 0.001) (Fig. 1c).
ApoA-I increases ex vivo glucose uptake by skeletal muscle in db/db mice
In the presence of insulin, glucose uptake into red muscle from the control animals was 3.02 ± 0.74 nmol mg−1 h−1, compared with 7.63 ± 0.65 nmol mg−1 h−1 in the apoA-I-treated animals (p < 0.001). Treatment with apoA-I also increased insulin-dependent glucose uptake into white muscle from 4.41 ± 0.65 nmol mg−1 h−1 in the PBS-treated animals to 8.04 ± 1.12 nmol mg−1 h−1 (p < 0.01). This indicates that treatment with apoA-I markedly improved insulin-dependent and insulin-independent glucose uptake by skeletal muscle in insulin resistant db/db mice.
In vivo [18F]FDG imaging of apoA-I-mediated increased glucose uptake
ApoA-I treatment increases hexokinase activity in vitro
In this study we have demonstrated that a single infusion of apoA-I significantly increases the rate of glucose uptake by skeletal muscle in db/db mice. Further, we have used real-time PET imaging to characterise the impact of apoA-I and insulin treatment on glucose uptake from the plasma into the gastrocnemius muscle in db/db mice. This proof of principle study has considerable flexibility and can be extended into studies of multiple organs and used longitudinally.
No significant increase in glucose uptake was observed when muscle samples were treated with insulin alone, which is a reflection of the insulin resistant state of db/db mice. Treatment of the mice with apoA-I alone significantly increased glucose uptake. When the mice were treated with apoA-I and the skeletal muscle was subsequently treated with insulin, glucose uptake was increased over and above that of apoA-I alone. This is indicative of apoA-I not only increasing the glucose uptake, but also improving insulin sensitivity. This is consistent with previous reports of apoA-I increasing skeletal muscle glucose uptake in both the presence and absence of insulin [2, 4].
HDL and apoA-I can activate a number of cell signalling pathways that have been linked to increased glucose uptake and insulin sensitivity. These include the AMP kinase pathway [21, 22, 23, 24, 25, 26, 27], which is known to be activated by apoA-I . HDL can also activate phosphatidylinositol-3 kinase (PI3K)/Akt via scavenger receptor class B type I (SR-BI) , which increases the trafficking of GLUT4 from intracellular stores to the cell surface , thereby increasing cellular glucose uptake. We have also previously demonstrated that apoA-I can activate protein kinase A (PKA) , and PKA-mediated phosphorylation of peroxisome proliferator-activated receptor α has been shown to improve insulin sensitivity in rats . The findings from the ex vivo skeletal muscle glucose uptake experiment suggested that apoA-I increased glucose uptake into skeletal muscle in an insulin-dependent and insulin-independent manner. To gain further insight into the impact of apoA-I treatment on glucose uptake and metabolism in skeletal muscle, we used kinetic analysis to model the uptake of [18F]FDG from the plasma into the muscle. Rather than collecting numerous arterial blood samples during the PET study, which is impractical due to the small blood volume of mice, we derived the model input function from the reconstructed image data using an ROI over the vena cava . To ensure the robustness of our variable estimates, PET images were deconvolved to limit partial volume effects  and the dispersion of the input function was included as a fit variable during kinetic modelling. Mouse input functions have previously been determined using the whole heart  as well as heart, liver and a blood samples .
A limitation of our [18F]FDG PET study is that it cannot determine whether small changes in k 1 are due to changes in blood flow in gastrocnemius tissue or altered glucose transport across the cell membrane. Multi-tracer approaches using [15O]H2O and 3-O-[11C]methylglucose together with [18F]FDG have previously been used by other investigators to assess glucose delivery, transport and phosphorylation in humans [33, 34, 35]. However, due to the 2-min half-life of 15O, this approach requires an on-site cyclotron and/or [15O]H2O generator to enable immediate injection which was not feasible for the current study. Without the ability to independently measure blood flow using [15O]H2O, a 3-O-[11C]methylglucose PET scan would not add any additional useful information since, like [18F]FDG, it cannot distinguish between flow dependent changes in delivery of tracer to the tissue and increased glucose transport across the cell membrane.
The time–activity curves for the gastrocnemius muscle of db/db mice treated with apoA-I showed increased uptake relative to control treatment in both the absence and presence of insulin. Furthermore, apoA-I and insulin treatment led to an increased rate of exit of [18F]FDG from the plasma compartment into tissue, relative to mice treated with insulin alone. As the rate of [18F]FDG uptake into the gastrocnemius (k 1) cannot account for this alone, it is likely to be due to increased cumulative whole body uptake, potentially via an HDL-mediated increase in production of the vasodilator endothelial nitric oxide synthase .
The rate of transport of [18F]FDG into the intracellular space (k 1) did not increase significantly in mice treated with apoA-I, either in the absence or presence of insulin. However, a trend towards increased k 1 values in apoA-I-treated animals was observed and could warrant further investigation in a PET-water study using [15O]H2O to specifically measure any transport effect mediated by apoA-I. However, as noted above, this requires an on-site [15O]H2O generation capability in the imaging facility. While the sample size used in this study is small (n = 4), it was sufficiently powered to clearly identify the effect of apoA-I treatment on both K i and k 3 values.
While previous studies have suggested HDL can improve glucose transport via increasing cell surface GLUT4 levels [5, 6], these studies were conducted in vitro and did not observe increased glucose transport with apoA-I treatment. HDL has previously been reported to increase GLUT4 translocation to the cell surface via SR-BI and Akt . However, apoA-I does not activate the Akt pathway  and this may explain the lack of a significant increase in k 1 in vivo and in L6 rat skeletal muscle cells treated with apoA-I.
While apoA-I treatment alone had no impact on the rate of [18F]FDG phosphorylation (k 3), treatment with apoA-I significantly increased the rate of insulin-dependent glucose phosphorylation. This appears to be somewhat contradictory to previous findings that glucose transport is rate limiting for glucose disposal in skeletal muscle . Hexokinase II (HK-II), the primary enzyme responsible for glucose phosphorylation in skeletal muscle, is phosphorylated by insulin-dependent activation of the Akt/PI3K signalling pathway, resulting in increased HK-II–mitochondrial association , which in turn increases HK-II activity [38, 39, 40]. Using the L6 rat skeletal muscle cell line, we confirmed that treatment of cells with apoA-I increased insulin-mediated HK-II activity, providing further evidence that while not able to activate the Akt/PI3K pathway directly, apoA-I potentiates insulin-mediated activation of HK-II activity. As such, this study suggests that the primary mechanism responsible for increased apoA-I-mediated insulin sensitivity is increased phosphorylation of glucose via increased potentiation of the insulin signalling pathway.
This work was supported by a National Imaging Facility Subsidised Access Grant to BJC.
Duality of interest
The authors declare that there is no duality of interest associated with this manuscript.
BJC was responsible for the experimental design, carried out the experiments, performed data analysis and wrote the manuscript. WJR was responsible for the experimental design, performed data analysis, and wrote the manuscript. AP, ST, AR, AA, AC and HH carried out the experiments, performed data analysis and revised the manuscript. PJB, LK, SRM, MCG and KAR were responsible for the experimental design and wrote the manuscript. All authors approved the final version of the manuscript for publication. BJC 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.
- 18.Iida H, Kanno I, Miura S, Murakami M, Takahashi K, Uemura K (1986) Error analysis of a quantitative cerebral blood flow measurement using H2(15)O autoradiography and positron emission tomography, with respect to the dispersion of the input function. J Cereb Blood Flow Metab 6:536–545CrossRefPubMedGoogle Scholar
- 19.Oikonen V (2002) Model equations for the dispersion of the input function in bolus infusion PET studies. Available from www.turkupetcentre.net/reports/tpcmod0003.pdf. Accessed 20 Mar 2015