Introduction
Macrovascular complications of diabetes mellitus (DM), primarily atherosclerosis are the primary cause of morbidity and mortality within affected patients. Approximately 68% of DM patients over the age of 65 die from a cardiovascular related-illness [
1]. Framingham data with 20 year follow-up revealed that patients with DM had a 2–3-fold increase in atherosclerotic disease [
2]. Similarly, in the Multiple Risk Factor Intervention Trial, males with type 2 DM had a threefold risk of coronary artery disease-related death, compared to non-diabetic controls [
3]. Most of the molecular and cellular mediators involved in atherosclerosis have been elucidated [
4] and an increasing body of evidence has identified the pivotal role of macrophages in atherosclerotic plaque formation [
5]. In addition, diabetes-associated hyperglycemia and oxidative stress contribute to accelerated atherosclerosis by directly effecting monocytes and macrophages [
6].
Atherosclerosis is a chronic progressive inflammatory disease and cholesterol metabolism disorder. Monocytes are recruited to, and migrate through the damaged endothelium where they differentiate to macrophages in the intimal layer [
7]. “Classically activated” MΦ1 macrophages are pro-atherogenic and differentiate upon exposure to T-helper 1 cytokines, IFN-gamma and IL-1beta; and sustain the ongoing inflammatory response via generation of TNF-alpha and IL-1beta. “Alternatively activated” MΦ2 macrophages are atheroprotective and arise from exposure to T-helper 2 cytokines, IL-4 and IL-13 [
8] and promote tissue repair and healing [
9].
Interestingly, MΦ1/MΦ2 macrophages show plasticity in response to cues from the microenvironment. MΦ1 and MΦ2 macrophages have a functional role in both obesity and atherosclerosis [
10,
11]. Macrophages within atherosclerotic plaques are typically MΦ1 promoting chronic inflammatory responses [
8]. However, it has been proposed that atherosclerosis-associated inflammation may not primarily be due to a skewing of the MΦ1:MΦ2 ratio, but also loss of anti-inflammatory MΦ2 macrophages resulting in impaired resolution of inflammation [
8,
12]. In the context of atherosclerosis, we have shown that there is an MΦ2 to MΦ1 switch during human plaque progression [
13]. In addition we showed that MΦ2 macrophages are localized to stable regions of atherosclerotic plaques and that expression of MΦ2 macrophage cell markers is inversely related to disease progression [
13]. Type 2 DM is characterized by a marked reduction in MΦ2 populations and a shift in the ratio between MΦ1 and MΦ2 macrophages is directly related to the development of insulin resistance in adipose tissue [
14]. This suggests that skewing macrophage phenotypes is an important mechanism through which DM induces cardiovascular disease (CVD) risk. We propose that therapeutic interventions which induce a MΦ2 switch may protect against diabetes-induced development of CVD.
Since the publication of the “Liraglutide effect and action in diabetes: evaluation of cardiovascular outcome results—a long term evaluation (LEADER) trial”, more emphasis has been placed on the potential protective effects of incretin-based therapies [
15]. Liraglutide, a glucagon-like peptide-1 receptor (GLP-1R) agonist currently used for the treatment of obesity-associated type 2 DM, confers minimum risk of hypoglycemia and promotes weight loss [
16]. The results of the LEADER trial showed that diabetic patients had a 13% reduction in risk of CV deaths following liraglutide administration [
15]. Furthermore, the Evaluation of Lixisenatide in Acute Coronary Syndrome (ELIXA) trial showed that lixisenatide confers a CV benefit rather than an absence of adverse effects [
17]. Since increased benefit in CV outcomes with GLP-1R agonists have now been established, investigating the effects of GLP-1R agonists on atherosclerosis-associated inflammation and determining their atheroprotective mechanism is important.
Liraglutide promotes anti-inflammatory responses [
18‐
20], reduces foam cell formation [
21,
22], inhibits expression of inflammatory markers and monocyte adhesion and attenuates atherosclerosis in vivo [
23,
24]. The aim of our study was to investigate if liraglutide induces an MΦ2 macrophage phenotype and promotes resolution in an in vivo model of early atherosclerosis progression.
Methods
Animals and diets
Homozygous ApoE
−/− mice (002052; C57BL/6J-ApoE
tm1Unc; RRID:IMSR_TAC:apoe) were purchased from Charles River Laboratories (Margate, UK) and housed in specific pathogen free conditions in 12 h light and dark cycle. All animal experiments were conducted according to Institutional guidelines and in compliance with the Health Regulatory Products Agency Ireland and Directive 2010/63/EU. Number of experimental animals was based on the sample size calculation (see Additional file
1: Methods). Diets were composed of 10% fat, 0% no added cholesterol (low-fat diet, LFD) or 60% fat with 1% cholesterol (high-fat, high-cholesterol diet, HFHCD). Both diets were sucrose matched (10%) and were supplied by Research Diets (New Brunswick, NJ, USA). Male ApoE
−/− mice were randomized at 8 weeks of age to receive LFD or HFHCD for 8 weeks. For the final 6 weeks mice received daily subcutaneous injections of PBS or liraglutide (Victosa, Novo Nordisk, Dublin, Ireland). Liraglutide dose was titrated upwards for the first 10 days (1, 3, 10, 30, 50, 100, 150, 200, 250 and 300 μg/kg) and maintained at 300 μg/kg. Mice were scored, weighed and water intake measured once weekly for the first 2 weeks and daily for the remaining 6 weeks. Food intake was measured weekly. Mice were euthanized by retro-orbital bleed and cervical dislocation and aortae and bones harvested. No adverse events were reported. The unit of analysis was a single animal.
Peripheral blood mononuclear cell isolation
Peripheral venous blood was collected from healthy volunteers at University College Dublin and Irish Blood Transfusion Service, National Blood Bank, St. James’s Hospital, James’s Street, Dublin 8., post institutional review board ethical approval. Written informed consent was obtained from all volunteers. Peripheral blood mononuclear cells (PBMCs) were isolated from blood as previously described [
25]. Briefly, 20 ml of blood was collected, layered onto polymorphoprep solution (Axis-Shield, Dundee, UK) (1:1) and centrifuged for 35 min at 500×
g and 20 °C. The mononuclear layer was removed, mixed with an equal volume of 0.45% NaCl (Sigma-aldrich, Dorset, UK) and centrifuged for 10 min at 400×
g and 20 °C. The pellet was re-suspended in 12 ml ice-cold water and 12 ml 1.8% NaCl and centrifuged for 5 min at 300×
g at 20 °C. Cells were re-suspended in complete medium [RPMI + Glutamax supplemented with 10% FBS and 100U penicillin–streptomycin (Gibco, Thermo Fisher Scientific, Waltham, MA, USA)], counted and seeded at a density of 2.5 × 10
5 cells/ml.
Cell line
THP-1 monocytes were seeded at a density of 1 × 10
6 cells/2 ml medium (ATCC
® TIB-202™, Teddington, Middlesex, UK, CLS Cat# 300356/p804_THP-1, RRID:CVCL_0006) and differentiated into MΦ0 cells using 320 nmol/l phorbol 12-myristate 13-acetate (PMA) (Sigma-aldrich, Dorset, UK) for 72 h, rested in complete medium for 24 h and polarized with 100 ng/ml lipopolysaccharide (LPS) (InvivoGen, Toulouse, France) and 20 ng/ml IFN-gamma (R&D Systems, Abingdon, UK) to induce a MΦ1, or 20 ng/ml IL-4 and IL-13 (R&D Systems, Abingdon, UK) for a MΦ2 macrophage phenotype for 48 h. THP-1 MΦ0, MΦ1 and MΦ2 macrophages were treated post-polarization with 250 nmol/l (~ 1 μg/ml) liraglutide for 6 h. mRNA and protein were isolated from cells for RT-qPCR and Western blotting (Additional file
1: Methods) analysis for mannose receptor (MR) expression. Cells tested negative for mycoplasma contamination.
Bone marrow-derived macrophage (BMDM) culture
Murine BMDMs were taken from C57BL/6 (RRID:MGI:3038854) and ApoE−/− femurs and tibiae and lavaged with complete medium. Cells were cultured in 75% RPMI + Glutamax supplemented with 10% FBS and 1% l-glutamine (Gibco, Thermo Fisher Scientific, Waltham, MA USA) and 25% L929-conditioned medium for 7 days. On day 3 the medium was replaced. On day 7 cells were treated with liraglutide (250 nmol/l for 6 h) or polarized into MΦ1 and MΦ2 macrophages using 100 ng/ml LPS and 20 ng/ml IFN-gamma or 20 ng/ml IL-4 and IL-13 (R&D Systems, Abingdon, UK), respectively for 18 h.
ELISA
Supernatants were collected from THP-1 and BMDM cells and analyzed by ELISA for human IL-1beta, IL-10, monocyte chemoattractant protein (MCP)-1 and TNF-alpha or murine IL-1beta, IL-10 and IL-6 (eBioscience, Thermo Fisher Scientific, and Biolegend, San Diego, CA, USA) as per the Manufacturer’s instructions.
Western blotting
THP-1 cells were harvested for protein. 20 μg of protein was quantified by a Bradford assay (Bio-rad, Fannin Ltd., Dublin, Ireland) as per the manufacturers instructions. The protein was run using 10% sodium dodecyl sulfate–polyacrylamide gel electrophoresis (Additional file
1: Table S1) at 90 V for 10 min and 120 V for a further 95 min. The proteins were then transferred onto nitrocellulose membrane (VWR, Dublin, Ireland) at 115 V for 1.5 h. The membrane was blocked in 5% non-fat skimmed milk for 1 h at room temperature and probed at 4 °C overnight for MR, 1:500, Signal transducer and activator of transcription (STAT)3 1:1000, STAT1 1:1000, phospho (p)-STAT1 1:1000 (Cell Signalling Technology, Dublin, Ireland), GLP-1R 1:1000 or beta-actin (Santa Cruz Biotechnology, Heidelberg, Germany) was used as the loading control at 1:1000. Membranes were then probed with anti-rabbit 1:1000 (MR, STAT3, STAT1, p-STAT1 and GLP-1R) or anti-mouse 1:2000 (beta-actin only) as a secondary antibody for 1 h at room temperature and developed in Pierce™ ECL Western blotting substrate (Thermo Fisher Scientific, Dublin, Ireland).
Quantitative RT-PCR
RNA from cells was extracted using the Qiagen RNeasy kit (Qiagen, Manchester, UK) as per Manufacturer’s instructions. RNA quantity and quality was determined using the NanoDrop™ 2000 (Life Technologies Ltd., Paisley, UK). mRNA was converted to cDNA and analyzed by RT-qPCR on ABI PRISM 7900HT (Applied Biosystems, Life Technologies, Ireland) using a TaqMan hydrolysis probe for mannose receptor C-type 1 (
Mrc1, HGNC ID:7228),
Il-
10 (HGNC ID: 601), arginase-1 (
Arg1, HGNC ID: 663) and 18S ribosomal RNA (HGNC ID: 1383) as a reference gene (Applied Biosystems, Life Technologies, Dublin, Ireland) and SYBR
® Green primers listed in Additional file
1: Table S2 (Eurofins, MWG Operon, Ebersberg, Germany).
En face analysis
Aortae were harvested from ApoE−/− mice and fixed in 10% formalin (DiaPath, Martinengo BG, Italy). Adventitial fat tissue was removed and the aortae opened longitudinally, rinsed in 70% ethanol for 1 min, stained in Sudan IV (35% ethanol/50% acetone) (Sigma-Aldrich, Dorset, UK) for 15 min with continuous agitation, destained in 80% ethanol for 2 min with agitation and rinsed until solution was clear in deionized water. Aortae were pinned onto black wax and imaged using a 4× magnification. Photoshop (Adobe Systems Inc., San Jose, CA, USA) was used to stitch images together and lesions were quantified using ImageJ (NIH, Bethesda, MD, USA) software. Percentage atherosclerotic lesion area was calculated as a ratio of the total aorta area to the lesion area.
Flow cytometry
BMDMs (2 × 10
6) from mice fed LFD or HFHCD (with/without liraglutide treatment) were labelled after 7 days in culture with anti-mouse monocyte and macrophage specific markers (BD Biosciences, Oxford, UK) (Table
1). Antibodies were combined to make a master mix using brilliant buffer (BD Biosciences, Oxford, UK) and PBS. Cells were stained with the antibody mix for 30 min at room temperature in the dark, washed and re-suspended in PBS before running on the Beckman Coulter CyAn Advanced Digital Processing (Beckman Coulter, Brea, CA, USA). Cells were also stained with a combination of all antibodies except one for fluorescence minus one controls for gating controls and amine C reactive positive and negative beads were stained with one antibody for single stained controls required for compensation. Flow cytometry standard (FCS) files were analyzed using FlowLogic software (Miltenyi Biotec Ltd., Surrey, UK).
Table 1
Antibodies for flow cytometry
CD45 | Brilliant ™ Blue (BB)515 | rat IgG2b, κ | 30-F11 |
CD11c | Phycoerythrin–cyanine 7 (PE-Cy7) | hamster IgG1, λ1 | HL3 |
Ly-6C | Allophycocyanin (APC) | rat IgM, κ | AL-21 |
F4/80 | Brilliant violet™ (BV)421 | rat IgG2a, κ | T45-2342 |
CD11b | BV510 | rat IgG2b, κ | M1/70 |
Statistical analysis
Results were analysed using GraphPad Prism 5.0c (GraphPad Software Inc., La Jolla, CA, USA) and are expressed as mean (SEM). Statistical comparisons for all THP-1 and BMDM experiments were made by paired (Friedman) and ApoE−/− in vivo experiments were analyzed by unpaired (Kruskal–Wallis) non-parametric ANOVAs followed by Dunn’s multiple comparison post-tests comparing all columns. Student’s paired (Wilcoxin-matched pairs signed rank) and unpaired (Mann–Whitney) non-parametric t tests were used to compare all columns of interest for in vitro or in vivo work involving RT-qPCR, flow cytometry and en face lesion quantification data. Statistical significance was considered when *p < 0.05, **p < 0.01 and ***p < 0.001.
Discussion
The developing atherosclerotic lesion induces a unique microenvironment consisting of immune cells that shift the normal balance towards a pro-inflammatory state. Determining if pharmacological interventions can limit progression or promote resolution of the disease by impacting on macrophage phenotype is important for improved treatment of DM macrovascular complications [
29]. The work described here shows that the GLP-1R agonist, liraglutide, has a direct effect on macrophage phenotype in early atherosclerosis resulting in inhibition of lesion formation in vivo.
To date in vivo studies which have examined the atheroprotective effect of GLP-1 receptor agonists have yielded conflicting results. Previous studies have examined the effect of GLP-1R agonists exendin-4 [
30] and taspoglutide [
31] in vivo. Whilst high doses of exendin-4 (24 nmol kg
−1day
−1) for 4 weeks suppressed atherosclerosis in ApoE
−/− mice fed a standard diet [
30] monthly doses of 0.4 mg taspoglutide for 12 weeks did not reduce plaque area in hyperglycemic ApoE
−/− mice fed a 45% fat diet [
31]. These results highlight the differences between exendin-based therapies and human GLP-1 analogues (taspoglutide) and the effect hyperglycemia and high fat diets have on responses to incretins. Previous studies have examined the effect of liraglutide on progression of atherosclerotic disease [
24,
32]. ApoE
−/− mice given 300 μg/kg liraglutide twice daily for 4 weeks, receiving a 22% fat and 0.15% cholesterol diet, had increased plaque stability although there was little effect on lesion area [
32]. A more recent study in LDL receptor null mice fed a 4.25% fat diet containing 0.3% cholesterol showed liraglutide at 1000 μg/kg daily for 13 weeks following a nephrectomy had attenuated atherosclerosis [
23]. Other in vivo studies have shown liraglutide is cardioprotective in an ischemic/reperfusion injured hearts model by preserving physiological levels of calcium [
33] and in a doxorubicin-induced cardiotoxicity model [
34]. In our study ApoE
−/− mice received a single daily dose of 300 μg/kg liraglutide for 6 weeks resulting in significant reduction of lesion progression at the iliac bifurcation of the aorta. Previous in vivo studies administered liraglutide for 4 weeks [
30‐
32,
35]. 300 μg/kg liraglutide has previously shown to have little effect on weight loss [
24,
36] meaning the atheroprotective effects we report are weight-independent. The equivalent dose of liraglutide given subcutaneously in a clinical setting is 1.8 mg/kg, as opposed to 3 mg/kg used to treat obesity [
37]. A unique aspect of our model is the high-fat high-cholesterol diet used. It has been extensively documented that ApoE
−/− mice on a high-cholesterol diet are more susceptible to rapidly developing atherosclerotic lesions compared mice on a high-fat diet alone [
38]. An obesogenic high-fat diet was used in this study to better mirror the current target population of liraglutide. Faber et al. investigated the effect of liraglutide treatment in patients and found no significant results in microvascular function after 10 weeks [
39] compared to Rizzo et al. who found significant reduction in waist circumference and intima-media thickness in carotid arteries in patients after an 18 month follow up [
40]. The main differences in these studies where the combination of drugs administered to patients and the length of treatment. Rizzo et al. used liraglutide in combination with metformin in contrast to Faber et al. who only gave liraglutide. Both studies used a slightly lower dose of 1.2 mg/kg compared to the standard dose of 1.8 mg/kg. Longer administration had the most profound effect. Although liraglutide had significant effects on intima-media thickness in carotid arteries in patients [
40], it had no significant effect in systolic function or exercise capacity [
41]. Therefore our data suggests that liraglutide may elicit a similar effect in patients with early stage atherosclerosis or at high risk of atherosclerosis development. Although results were variable in patients with established disease, our work warrants further investigation into translational research.
Previous studies have examined the potential mechanism through which GLP-1R agonists confer atheroprotection. It has been shown that liraglutide reduces monocyte adherence to human aortic endothelial cells in response to TNF-alpha and LPS stimulation [
42]. Of relevance to our study, is that exenatide induces MΦ2 macrophage polarization via signal transducer and activator of transcription three in human monocyte derived macrophages [
43]. However, few studies have addressed the effect of liraglutide on the inflammatory status of monocytes and macrophages in vivo in the context of atherosclerosis. Most recently Vinué et al. reported that lixisenatide decreases atherosclerosis in insulin resistant mice by reprogramming macrophages towards an MΦ2 phenotype [
35]. Indeed, our initial approach using THP-1 monocytes suggested that liraglutide influences monocyte/macrophage cell fate. Liraglutide attenuated IL-1beta in MΦ0 macrophages and showed a trend towards elevated secretion of IL-10 from both MΦ0 and MΦ2 macrophages. Indeed, similar results have been reported by others which showed decreased IL-1beta in MΦ1 THP-1 macrophages and increased IL-10 secretion from MΦ2 THP-1 macrophages [
28]. To date there is limited data on the effect of liraglutide on BMDM phenotypes or cell fate in vivo. This is important as we and others have shown that inhibition and regression of atherosclerosis is associated with a phenotypic MΦ1 to MΦ2 macrophage switch [
13,
44,
45]. To address this we examined the effect of ex vivo liraglutide treatment on pro-and anti-inflammatory cytokine generation and on expression of macrophage phenotypic markers TNF-alpha (MΦ1) and MR (MΦ2) in BMDMs from WT mice. Similar to what we observed in THP-1 cells, we found a trend towards reduced IL-1beta and enhanced IL-10 secretion in MΦ0, MΦ1 and MΦ2 WT BMDMs following ex vivo liraglutide treatment. This is in keeping with what has been observed with exenatide treatment in human monocyte derived macrophages [
43]. This suggests liraglutide acts as an immune modulator that can alter macrophage phenotype. EAT is the gonadal white fat found in male rodents and expands significantly when these mice are fed a high-fat diet [
46]. To confirm a role for liraglutide in inducing a pro-resolving macrophage phenotype we examined macrophage infiltration in EAT from ApoE
−/− mice fed a HFHCD. We showed reduced macrophage infiltration and a trend towards increased MΦ2 macrophage polarization in EAT (Additional file
11: Figure S10).
To address if liraglutide impacts on macrophage phenotype in disease we examined the effect of in vivo liraglutide treatment on monocyte/macrophage phenotype in the ApoE
−/− mice. Animals were administered a HFHCD for 2 weeks to induce a state of metabolic dysfunction which was confirmed with weight gain (Additional file
5: Figure S4), following administration of liraglutide in the presence of the HFHCD for a further 6 weeks. After a total of 8 weeks of diet challenge both the monocyte and macrophage populations were quantified. This is the first study to directly examine alteration in both monocyte and macrophage populations in vivo in response to liraglutide in the context of atherosclerosis. Pro-inflammatory and anti-inflammatory monocyte populations in mice are characterised by CD45
+F4/80
loLy-6C
hiCD11c
− and CD45
+F4/80
loLy-6C
loCD11c
+ respectively [
47]. Inflammatory monocytes are abundant in ApoE
−/− mice where they adhere to the endothelium, infiltrate, differentiate to macrophages and contribute to disease progression [
48]. However, the specific role of pro-resolving monocytes in atherosclerosis is less clear [
48,
49]. We found a significant change in MΦ1-like characterized monocytes, where these monocytes presented as less inflammatory following liraglutide treatment in vivo. This is in keeping with a recent study which showed that lixisenatide decreases proinflammatory monocytes in insulin resistant ApoE
−/− mice [
35]. Importantly we also showed a significant increase in alternative MΦ2 macrophage populations. This novel atheroprotective effect was further confirmed by analyzing MΦ2 gene expression in BMDMs from liraglutide-treated ApoE
−/− mice which showed a statistically significant increase in Arg-1, IL-10, MR and CD163. We have previously shown that induction of IL-10 by MΦ2 macrophages is required for systemic alteration of monocyte populations [
50]. This is important in the context of our other work which shows that MΦ2 macrophage accumulation in human atherosclerotic plaque is inversely related to disease progression [
13]. Here, we show that as well as enrichment of the MΦ2 phenotype there was a significant reduction in MΦ1 phenotypic markers specifically CCR7 and TNF-alpha.
Authors’ contributions
RB, SC, DC, GGC, FCMcG and OB designed the experimental work carried out. RB and DC carried out all in vitro experiments. RB, SC, SK and MEOR carried out all in vivo experiments. MBL was consulted for aorta lesion quantification training. All figures were prepared by RB. Analysis was carried out by RB, SC, DC and SK. RB and OB wrote and revised the manuscript. All authors read and approved the final manuscript.