HDL Improves Cholesterol and Glucose Homeostasis and Reduces Atherosclerosis in Diabetes-Associated Atherosclerosis

Background and Aims Apolipoprotein A-I (ApoA-I), the main component of high-density lipoprotein (HDL), not only promotes reverse cholesterol transport (RCT) in atherosclerosis but also increases insulin secretion in pancreatic β-cells, suggesting that interventions which raise HDL levels may be beneficial in diabetes-associated cardiovascular disease (CVD). Previously, we showed that TNF-related apoptosis-inducing ligand (TRAIL) deletion in Apolipoprotein Eknockout (Apoe−/−) mice results in diabetes-accelerated atherosclerosis in response to a “Western” diet. Here, we sought to identify whether reconstituted HDL (rHDL) could improve features of diabetes-associated CVD in Trail−/−Apoe−/− mice. Methods and Results Trail−/−Apoe−/− and Apoe−/− mice on a “Western” diet for 12 weeks received 3 weekly infusions of either PBS (vehicle) or rHDL (containing ApoA-I (20 mg/kg) and 1-palmitoyl-2-linoleoyl phosphatidylcholine). Administration of rHDL reduced total plasma cholesterol, triglyceride, and glucose levels in Trail−/−Apoe−/− but not in Apoe−/− mice, with no change in weight gain observed. rHDL treatment also improved glucose clearance in response to insulin and glucose tolerance tests. Immunohistological analysis of pancreata revealed increased insulin expression/production and a reduction in macrophage infiltration in mice with TRAIL deletion. Furthermore, atherosclerotic plaque size in Trail−/−Apoe−/− mice was significantly reduced associating with increased expression of the M2 macrophage marker CD206, suggesting HDL's involvement in the polarization of macrophages. rHDL also increased vascular mRNA expression of RCT transporters, ABCA1 and ABCG1, in Trail−/−Apoe−/− but not in Apoe−/− mice. Conclusions. rHDL improves features of diabetes-associated atherosclerosis in mice. These findings support the therapeutic potential of rHDL in the treatment of atherosclerosis and associated diabetic complications. More studies are warranted to understand rHDL's mechanism of action.


Introduction
Diabetes is associated with an increased risk of cardiovascular diseases (CVD), including atherosclerosis. In diabetes, the pathophysiological mechanisms that promote atherosclerosis and CVD are numerous and in part still not completely clarified [1][2][3]. Diabetic patients have elevated plasma levels of low-density lipoprotein (LDL) and triglycerides and low levels of high-density lipoprotein (HDL) [4]. Early animal and human studies have shown HDL infusions to have favorable cardiovascular effects [5,6], even in diabetes. Infusing reconstituted HDL (rHDL) or increasing HDL levels with overexpression of apolipoprotein A-I (ApoA-I), the major HDL apolipoprotein, reduced plaque size by altering the histological composition of experimental lesions in animal models of atherosclerosis and restenosis [7][8][9][10][11][12]. The earliest proof-of-concept studies in humans demonstrated that infusing ApoA-I, rHDL, and its delipidated form could inhibit LDL oxidation, modulate inflammation, and elicit plaque regression [13][14][15]. Furthermore, ApoA-I protected against diabetes by reducing stress-induced pancreatic β-cell apoptosis and increasing insulin secretion [16], and HDL attenuated islet cell inflammation in type 2 diabetes via ATPbinding cassette transporter-A1 and ATP-binding cassette transporter-G1 (ABCA1 and ABCG1) [16,17]. In addition, ApoA-I improved insulin sensitivity with decreased systemic and hepatic inflammation in mice fed a high-fat diet [18]. Collectively, these suggest that interventions which raise HDL levels may be beneficial in lowering cholesterol, reducing atherosclerosis, and improving glucose homeostasis in diabetes.
TRAIL is a member of the TNF superfamily and was initially thought to selectively induce apoptosis in cancer cells. However, more recently, it has been discovered that TRAIL is implicated in coronary artery disease, acute coronary syndromes, and diabetes [19][20][21][22]. Studies have shown that TRAIL deficiency is associated with type-1 autoimmune diabetes in mice [23,24]. Injections of soluble TRAIL receptor (an antagonist of TRAIL signaling) into nonobese diabetic (NOD) mice or streptozotocin-treated Trail -/mice increased incidence of diabetes [23,24]. In addition, TRAIL deficiency in mice on a "Western" style high-fat diet for 12 w markedly accelerated atherosclerosis and promoted features of diet-induced diabetes including weight gain, hyperglycemia, hypoinsulinemia, and pancreatic β-cell dysfunction [25]. Using this mouse model of physiologically induced diabetes-associated atherosclerosis, we sought to determine the effects of reconstituted HDL (rHDL) on improving atherosclerosis and features of diabetes.

Animals.
Male Trail -/-Apoe -/or Apoe -/mice aged 6 weeks and weighing approximately 18 to 20 g were placed on a high-fat high-cholesterol "Western" diet (Semi-Pure Rodent Diet SF00-219; 22% fat, 0.15% cholesterol; Specialty Feeds, Glen Forrest, WA, Australia) as previously described [25]. After 10 weeks on a Western diet, mice were randomly assigned to receive either PBS vehicle or rHDL (20 mg/kg) for 2 weeks (3 times weekly). Mice were monitored daily, and body weights were recorded weekly. Mice were euthanised by cardiac exsanguination and indicated tissues isolated. All animal work was conducted according to the Animal Care and Ethics Committee guidelines, the University of New South Wales, or the Sydney Local Health District Animal Welfare Committee (Sydney, NSW, Australia).

Plasma and Liver
Analysis. Blood was collected by cardiac puncture at the time of euthanasia. Plasma samples were stored at −80°C in EDTA-Na 2 until required for analysis. Triacylglycerol and total cholesterol from plasma were measured using commercial kits (Wako Chemicals, Osaka, Japan). Fasting blood glucose was measured by a glucometer (Accu-check Performa; Roche, Mannheim, Germany). Insulin was measured using commercially available ELISA (Mercodia).

Insulin and Glucose Tolerance
Tests. At 10 and 12 weeks of high-fat feeding, either 1 g/kg body weight D-glucose (Sigma-Aldrich, Sydney, NSW, Australia) was injected into mice intraperitoneally following overnight fasting or 1 U/kg body weight human insulin was injected into non-fasted mice intraperitoneally. Blood was collected by pinprick from the tail vein and plasma glucose measured using a glucometer.
2.5. RNA Extraction and Real-Time Quantitative PCR. Liver and thoracic aortae were snap-frozen in liquid nitrogen and stored at −80°C. Tissue was homogenised (MP Biomedicals, Sydney, NSW, Australia) and total RNA extracted in TRI reagent (Sigma) [29]. RNA was then reverse transcribed to cDNA using iSCRIPT (Bio-Rad, Sydney, NSW Australia). Real-time PCR was performed in triplicate using iQSybr Green Master Mix (Bio-Rad, Sydney, NSW Australia) in the CFX96 thermocycler (Bio-Rad, Sydney, NSW Australia). Relative changes in mRNA levels between groups were determined using the 2 −ΔΔC t method [30]. Expression was normalised to β-actin, and changes were compared with Apoe −/− mice. Primer details can be found in Table 1. 2.6. Histology and Immunohistochemistry. Brachiocephalic arteries were processed as described [31]. Pancreata were removed and fixed in 10% formalin (wt/vol). Haematoxylin and eosin stain was used to assess tissue architecture; pancreata were stained for insulin (1 : 500; Cell Signalling) and macrophages (MAC3, 1 : 100; BD Pharmingen). All IgG controls were negative. Digital images of sections were acquired using an Olympus DP72 microscope (Olympus, Mount Waverley, Victoria, Australia) and a Zeiss Axio Imager Z2 microscope.

Morphometric
Analysis. Morphometric analysis of plaque area : total artery area, media area : total artery area, and necrotic core area : plaque area was performed on haematoxylin and eosin-stained sections using ImageJ (NIH, Bethesda). The percentage of positive staining in the plaque or islets was determined using cellSens Software (Olympus). Thresholds for positive staining for each antibody were determined, and sections were analysed by an investigator blinded to the mouse genotype.
2.8. Statistics. All results are expressed as the mean ± SEM. Statistical comparisons were made by Mann-Whitney t -tests and one-or two-way ANOVA, with Bonferroni's correction where appropriate. The statistics program in GraphPad Prism Version 6.0 was used (GraphPad Software, San Diego, CA, USA). A value of p < 0:05 was considered significant.

Journal of Diabetes Research
No changes to plasma glucose or insulin in response to rHDL in Apoe -/mice were observed (data not shown). We next performed glucose and insulin tolerance tests (GTT and ITT) in Trail -/-Apoe -/mice to assess glucose and insulin sensitivity. Vehicle-treated Trail -/-Apoe -/mice had significantly increased plasma glucose levels at 15 and 30 min following a glucose challenge (Figure 1(c)), which was cleared over the 2 h timepoint. In contrast, Trail -/-Apoe -/mice treated with rHDL demonstrated faster glucose clearance (Figure 1(c)). When the area under the curve was quantified, rHDL improved glucose tolerance in Trail -/-Apoe -/-> 2-fold compared with vehicle-treated mice (Figure 1(c)). rHDL-treated mice also showed improved insulin sensitivity with greater reductions in plasma glucose at 15 min following an insulin bolus and a~25% reduction overall when the area under the curve was quantified (Figure 1(d)). These findings were associated with a significant increase in endogenous insulin protein expression (Figure 1(e)) and reduced MAC3 staining indicating reduced macrophage content in pancreatic islets of Trail -/-Apoe -/mice treated with rHDL (Figure 1(f)). In contrast, rHDL had no effect on glucose or insulin tolerance in Apoe -/mice ( Supplementary Figure 2A-B). Together, these findings suggest that rHDL improves features of Western diet-induced diabetes in Trail -/-Apoe -/mice.

Discussion
Although epidemiological studies have shown that low HDL levels correlate with increased risk of diabetes mellitus and cardiovascular disease [40,41], preclinical and clinical studies aimed at raising HDL levels as a potential therapeutic to reduce diabetes and cardiovascular complications have been conflicting [42]. It is now known that HDL has pleiotropic functions, and HDL preparations can have multiple different effects on disease. Here, we sought to identify whether rHDL could improve features of diabetes-associated atherosclerosis using Western diet-fed Trail -/-Apoe -/mice. We identified 3 novel findings. First, rHDL reduced plasma glucose and insulin levels and improved glucose and insulin tolerance in these mice. Second, rHDL infusion reduced cholesterol and triglyceride levels observed, as well as restoring ABCA1 and ABCG1 mRNA in the vessel wall. Third, atherosclerosis was only reduced in Trail -/-Apoe -/mice with diabetic features, but not Apoe -/mice. Collectively, these suggest that rHDL therapy could benefit patients with diabetes- The beneficial effects of HDL on diabetes and changes to metabolic disease are well known. For example, mice deficient in ApoA-I-the main protein component on HDL-have impaired glucose tolerance compared to wildtype mice [43]. Furthermore, rHDL inhibited β-cell apoptosis and enhanced insulin secretion in vitro 16,17 , and ApoA-I treatment of insulin-resistant mice increased insulin secretion and glucose clearance [44]. Indeed, intravenous infusions of rHDL to patients with type 2 diabetes stimulated glucose uptake into skeletal muscle, reduced plasma glucose levels, and increased insulin secretion [45] confirming preclinical findings. rHDL treatment in Trail -/-Apoe -/mice also supports this view. We showed that improved glucose tolerance and insulin sensitivity occur in part, due to rHDL's anti-inflammatory effects on pancreata; rHDL reduced macrophage accumulation and increased insulin expression in β-cells in islets of Western diet-fed Trail -/-Apoe -/mice. In addition to type 2 diabetic features, Trail -/-Apoe -/mice have significantly increased total cholesterol and triglycerides as well as increased VLDL and LDL compared to Western diet-fed Apoe -/mice [25]. HDL's primary role is to mediate cholesterol efflux from atherosclerotic plaques, initially through interactions with ABCA1, then after remodelling through cholesterol exchange via SR-BI and ABCG1 [46]. Indeed, a deficiency or inhibition of Abca1 and Abcg1 enhanced cholesterol accumulation in atherosclerotic plaque via enhanced inflammation and monocyte/macrophage infiltration [47,48]. Furthermore, Trail-negative macrophages had reduced cholesterol efflux, in part due to reduced ABCA1 and ABCG1 mRNA expression when compared with Trailpositive macrophages [38]. In the current study, ABCA1 and ABCG1 mRNA expression was significantly reduced in the vascular tissues of Trail -/-Apoe -/mice, and importantly, rHDL administration restored this expression back to control levels. Because plaque from Trail -/-Apoe -/mice have increased monocyte/macrophage accumulation and inflammation and considering that ABCA1 and ABCG1 mRNA was unaltered in the liver of Trail -/-Apoe -/mice with rHDL infusion, rHDL may target lipid-laden macrophages in the vessel wall. This may also be relevant to lipid-laden macrophages in pancreata since downregulation of ABCA1 was associated with cholesterol accumulation in pancreatic βcells, greater inflammation, and reduced insulin secretion [41]. However, this requires further elucidation.
A large number of preclinical studies of atherosclerosis suggest that HDL intervention reduces plaque size and inflammation [11,49,50]. We also observed that rHDLtreated Trail -/-Apoe -/mice had significantly reduced atherosclerotic plaque size compared to mice that received PBS; however, no changes were observed in atherosclerotic Apoe -/mice. This is in contrast with early studies where infusing rHDL in cholesterol-fed rabbit models attenuated development of atherosclerosis [8,9] and in Apoe-deficient mice where single infusions of HDL containing ApoA-I (apolipoprotein A-I) Milano and dipalmitoyl phosphatidylcholine reduced plaque size, lipid, and macrophage content in atherosclerotic plaques [50]. As a caveat to these findings, other animal and HDLtargeted gene therapy studies demonstrated that increasing HDL with either infusions or overexpression of ApoA-I did not reduce the formation of atherosclerotic lesions but rather remodelled more advanced preexisting lesions similar to what is observed in human plaques as a more stable plaque phenotype [11,51,52]. Diabetic patients with obstructive coronary disease are known to have multiple metabolic features including elevated levels of metabolic stress markers [53], endothelial dysfunction [54], and increased inflammation in pericoronary adipose tissue [55], highlighting the complexity of the disease and its resulting cardiovascular complications. High fat dietfed mice lacking TRAIL have many of these contributing metabolic changes including adiposity [25], endothelial dysfunction [56], and alterations in metabolic function [25,57]. These are known to associate with atherosclerosis and CVD. Our findings highlight that rHDL treatment may be more useful in CVD patients with metabolic dysfunctions.
In summary, we have highlighted the importance of using a physiologically relevant diabetes-associated atherosclerosis model to investigate therapeutic strategies. In this model, HDL reduced features of diabetes associated with atherosclerosis. With much controversy surrounding the use of HDL in cardiovascular disease and the contradicting studies in preclinical models, this study identifies a continuing need for HDL research and a possible role for its use in targeted therapeutic strategies in people with diabetes-accelerated atherosclerosis.

Data Availability
Data are available on request.

Disclosure
This manuscript was presented in part as a poster for the Arteriosclerosis, Thrombosis, and Vascular Biology Scientific Meeting.

Conflicts of Interest
The authors declare no conflicts of interest.
Authors' Contributions B.A.D., S.P.C., and M.M.K. were responsible for experimentation, conception, design, analysis, and interpretation of data, manuscript preparation, and intellectual input. S.G. and P.M.C. were responsible for experimentation and analysis. K.A.R. was responsible for interpretation of data, intellectual input, and manuscript preparation. All authors gave final approval for publication.

Acknowledgments
Trail −/− mice were originally sourced from Amgen. BAD was supported by a National Health and Medical Research Council of Australia Postdoctoral Fellowship (GNT1037074). MMK was supported by a Heart Foundation of Australia Career Development Award (Award ID: CR 12S 6833).

Supplementary Materials
Supplementary Figure 1: Western diet-fed Trail -/-Apoe -/mice have impaired glucose tolerance compared to Apoe -/mice. Supplementary Figure 2: HDL has no effect on glucose and insulin tolerance in Apoe -/mice. Supplementary Figure  3: rHDL has no effect on plasma cholesterol, plasma triacylglycerol, or atherosclerosis in 12 w Western diet-fed Apoe -/mice. (Supplementary Materials)