Sirtuin 6 Expression and Inflammatory Activity in Diabetic

Diabetes Volume 64, April 2015 1395

Maria Luisa Balestrieri,1 Maria Rosaria Rizzo,2 Michelangela Barbieri,2 Pasquale Paolisso,2 Nunzia D’Onofrio,1 Alfonso Giovane,1 Mario Siniscalchi,3 Fabio Minicucci,3 Celestino Sardu,2 Davide D’Andrea,3 Ciro Mauro,3 Franca Ferraraccio,4 Luigi Servillo,1 Fabio Chirico,5 Pasquale Caiazzo,5 Giuseppe Paolisso,2 and Raffaele Marfella2

Sirtuin 6 Expression and Inﬂammatory Activity in Diabetic Atherosclerotic Plaques: Effects of Incretin Treatment

Diabetes 2015;64:1395–1406 | DOI: 10.2337/db14-1149

The role of sirtuin 6 (SIRT6) in atherosclerotic pro- Cardiovascular disease represents the leading cause of gression of diabetic patients is unknown. We evaluated death in patients with type 2 diabetes (1). Diabetes leads SIRT6 expression and the effect of incretin-based to increased vulnerability for plaque disruption, and therapies in carotid plaques of asymptomatic diabetic mediates increased incidence and severity of clinical and nondiabetic patients. Plaques were obtained from fl events (2). In ammation, particularly in diabetes, plays COMPLICATIONS 52 type 2 diabetic and 30 nondiabetic patients undergo- a central role in the cascade of events that result in plaque ing carotid endarterectomy. Twenty-two diabetic patients erosion and fissuring (2). There is now increasing evi- were treated with drugs that work on the incretin system, dence that a number of transcription factors, including GLP-1 receptor agonists, and dipeptidyl peptidase-4 the Sir2 family of enzymes, namely sirtuins (SIRTs), reg- 6 inhibitors for 26 8 months before undergoing the end- ulate multiple genes whose products are putatively in- arterectomy. Compared with nondiabetic plaques, dia- volved in the regulation of inflammation and endothelial betic plaques had more inflammation and oxidative cell (EC) function (3). The Sir2 family consists of seven stress, along with a lesser SIRT6 expression and collagen enzymes (SIRT1 to SIRT7) that share a conserved core cat- content. Compared with non-GLP-1 therapy–treated alytic domain, but differ in their cellular localization and plaques, GLP-1 therapy–treated plaques presented greater tissue distribution (4). Among the SIRTs, SIRT6, a chromatin- SIRT6 expression and collagen content, and less in- associated deacetylase, is considered to have a leading flammation and oxidative stress, indicating a more stable plaque phenotype. These results were supported by in role in regulating genomic stability, cellular metabolism, – vitro observations on endothelial progenitor cells (EPCs) stress response, and aging (5 8). A recent study (9) in fl and endothelial cells (ECs). Indeed, both EPCs and ECs mice suggested a role for SIRT6 in in ammation. Moreover, treated with high glucose (25 mmol/L) in the presence of the knockdown of SIRT6 resulted in the increased expres- fl GLP-1 (100 nmol/L liraglutide) presented a greater SIRT6 sion of proin ammatory cytokines (interleukin [IL]-1b, and lower nuclear factor-kBexpressioncomparedwith IL-6, and IL-8), extracellular matrix remodeling enzymes cells treated only with high glucose. These findings es- (matrix metalloproteinase [MMP]-2, MMP-9, and plasmin- tablish the involvement of SIRT6 in the inflammatory ogen activator inhibitor 1), and intracellular adhesion pathways of diabetic atherosclerotic lesions and sug- molecule-1 (4). In ECs, the loss of SIRT6 was associated gest its possible positive modulation by incretin, the with an increased expression of nuclear factor-kB(NF-kB), effect of which is associated with morphological and whereas overexpression of SIRT6 was associated with de- compositional characteristics of a potential stable pla- creased NF-kB transcriptional activity (4), indicating that que phenotype. SIRT6 may be associated with the upregulation of genes

1Department of Biochemistry, Biophysics and General Pathology, Second Univer- Corresponding author: Raffaele Marfella, [email protected]. sity of Naples, Naples, Italy Received 25 July 2014 and accepted 2 October 2014. 2Department of Medical, Surgical, Neurological, Aging and Metabolic Sciences, This article contains Supplementary Data online at http://diabetes Second University of Naples, Naples, Italy .diabetesjournals.org/lookup/suppl/doi:10.2337/db14-1149/-/DC1. 3Department of Cardiology, Cardarelli Hospital, Naples, Italy 4Department of Clinical, Public and Preventive Medicine, Second University of © 2015 by the American Diabetes Association. Readers may use this article as Naples, Naples, Italy long as the work is properly cited, the use is educational and not for profit, and 5Department of Neurosurgery, Cardarelli Hospital, Naples, Italy the work is not altered. 1396 Atherosclerosis in Patients With Diabetes and Sirtuin 6 Diabetes Volume 64, April 2015 involved in inflammation, vascular remodeling, and an- peptidase-4 (DPP-4) inhibitors, were classified as “never giogenesis. However, the role of SIRT6 in human athero- incretin users.” The patients with diabetes who had al- sclerotic plaques has not yet been described. Although ready used GLP-1 agonists or DPP-4 inhibitors were clas- it has been demonstrated that diabetes may be impli- sified as “current incretin users.” Among the 52 diabetic cated in the regulation of SIRT6 expression in diabetes- patients enrolled in the study, 24 were current incretin induced neurodegeneration (10), still no evidence exists users, and 28 were never incretin users. The current about the potential role of SIRT6 in the evolution of incretin users were patients who had been treated with atherosclerotic plaques in diabetic patients. We hypothe- incretin for at least 6 months. Patients treated with sized that, by acting on SIRT6, diabetes may enhance the incretin for a period of ,6monthswereexcludedfrom inflammatory potential of atherosclerotic plaques, favoring the study. Information on the duration of treatment was their instability. Thus, this study was designed to identify available for all current users. The mean (6SD) duration differences in SIRT6 expression, as well as in inflamma- of incretin treatment was 26 6 8 months. No patients tory infiltration, between carotid plaques of asymptom- had clinical or laboratory evidence of heart failure, pre- atic diabetic and nondiabetic patients. Experimental vious stroke, valvular defects, malignant neoplasms, or studies suggest that in obese mice, GLP-1–based therapies secondary causes of hypertension. Carotid sonography may reduce inflammation (11–13) and enhance the pro- was performed on a single ultrasound machine (Aloka tein expression of SIRT1 (14). Moreover, human studies 5500). The study was approved by the local ethics com- showed that sitagliptin (15) and exenatide (16), even at mittee, and informed written consent was obtained for a single dose, exert a potent anti-inflammatory effect, and each patient. that many of these effects were persistent over a period of 12 weeks, thus suggesting that the anti-inflammatory Laboratory Analysis effects of GLP-1–based therapies could help to reduce After an overnight fast, plasma glucose, HbA1c, and serum atherogenesis. lipid levels were measured by enzymatic assays in the Here, we evaluated the effect of incretin therapy in hospital chemistry laboratory. GLP-1 levels (Active GLP-1 fi diabetic patients on SIRT6 expression in carotid plaques [7-36] Speci c ELISA Kit; Epitope Diagnostics) were mea- and early outgrown circulating endothelial progenitor sured after an overnight fast (at 8:00 A.M.) and after cells (EPCs). Furthermore, a set of in vitro experiments breakfast. A standardized breakfast contained 419 kcal in ECs and EPCs was designed to evaluate the effect (57% carbohydrate, 17% protein, and 26% fat). After of incretin on SIRT6 and NF-kB during high-glucose breakfast, blood samples for the measurement of GLP-1 treatment. were obtained every 30 min over a 2-h period. The mean of the four GLP-1 evaluations was defined as the post- RESEARCH DESIGN AND METHODS prandial GLP-1 value. The standardized meal tolerance Patients were recruited from the outpatient Department test and baseline evaluations were performed 7 days be- of Cardiology and Cardiovascular Surgery of the Cardarelli fore surgery. Thereafter, the patients were asked to self- Hospital, Naples, Italy, from January 2009 to June 2013. monitor their blood glucose level (fasting, postbreakfast, From among these patients, we selected 52 type 2 diabetic postmeal, and postdinner glucose levels) until the day and 30 nondiabetic patients (nondiabetic group) with of surgery. Levels of fasting blood glucose were eval- asymptomatic carotid stenosis (according to North Amer- uated before surgery. Fasting and postprandial plasma ican Symptomatic Carotid Endarterectomy Trial classifi- glucose data were obtained from the average of each cation), enlisted to undergo carotid endarterectomy for assessment. . extracranial high-grade ( 70%) internal carotid artery Atherectomy Specimens stenosis (17). Asymptomatic patients underwent a base- After surgery, the specimens were cut perpendicular to line clinical examination, gave a medical history, and had the long axis into two halves. The first half was frozen in never developed neurologic symptoms or cerebral lesions liquid nitrogen for the following ELISA analysis. A assessed by computed tomography. All patients under- portion of the other half of the specimen was immedi- went computed tomography scanning or MRI. Diabetes ately immersion fixedin10%bufferedformalin.Sections was categorized according to the criteria of the American were serially cut at 5 mm, mounted on lysine-coated Association of Clinical Endocrinologists and the American slides, and stained with hematoxylin-eosin and tri- Diabetes Association (18). Furthermore, the diabetic chrome stain. Carotid artery specimens were analyzed fi patients answered a speci c questionnaire about medi- by light microscopy. cines used for diabetes treatment before the beginning of the study, the dates of the beginning and the end of Immunohistochemistry treatment, the route of administration, and the duration After the surgical procedure, atherectomy specimens of use. Information from the medicine inventory during were immediately frozen in isopentane and cooled in the study and this specific questionnaire was used to liquid nitrogen. Similar regions of the plaque were classify the subjects. The patients with diabetes who never analyzed (Fig. 1). Serial sections were incubated with used incretin, such as GLP-1 agonists and dipeptidyl the following specific antibodies: anti-SIRT6 (Millipore); diabetes.diabetesjournals.org Balestrieri and Associates 1397 anti–HLA-DR; anti-CD68 (cluster of differentiation 68 gly- coprotein) and anti-CD3 (cluster of differentiation 3 T-cell coreceptor protein) (Dako), markers of T cells and macrophages, respectively; anti–MMP-9 (Santa Cruz Biotechnol- ogy), proteases involved in the degradation of the collagen content in the plaque; and anti–tumor necrosis factor-a (TNF-a) (R&D Systems). Analysis of immunohistochemistry was performed with a personal computer–based, quan- titative, 24-bit color image analysis system (IM500; Leica Microsystems AG). Sirius Red Staining for Collagen Content Sections were stained as previously described (19). After dehydration, the sections were observed under polarized light after being placed on coverslips. The sections were photographed with identical exposure settings for each section. Biochemical Assays Plaques were lysed and centrifuged for 10 min at 10,000g — m Figure 1 Representative example of the plaques studied. The in- at 4°C. After centrifugation, 20 g of each sample was set is representative of the plaque region analyzed. The section was loaded, electrophoresed in polyacrylamide gel, and electro- stained with hematoxylin-eosin. The main plaque section is shown blotted onto a nitrocellulose membrane. Each determina- at low magnification (35), and the section analyzed is shown at high tion was repeated at least three times. magnification (inset 3100). MMP-9, TNF-a, and nitrotyrosine levels were quantified in plaques using specificELISAkits(from Santa Cruz Biotechnology, R&D Systems, and Imge- nex). SIRT6 activity was measured using a fluorimetric Confocal Laser-Scanning Microscopy SIRT6 assay kit (Abcam, Cambridge, U.K.). NF-kBbinding Immunofluorescence detection of SIRT6 in deparaffinized to kB sites was assessed on nuclear extracts from plaque atherosclerotic plaque sections from diabetic and non- specimens (19) by the TransAM NF-kB p65 Transcrip- diabetic patients and in in vitro–cultured EPCs and ECs tion Factor Assay Kit (Active Motif Europe, Rixensart, was performed by confocal laser-scanning microscope anal- Belgium). ysis (LSM 510; Zeiss) by using specific antibodies against SIRT6 (1:500) (Cell Signaling Technology, Danvers, MA), Isolation and Culture of EPCs vimentin (1:1,000) (Sigma-Aldrich, St. Louis, MO), or anti- Early outgrown EPCs were isolated from the leukocyte- bodies against von Willebrand factor (1:500) (Abcam), as rich buffy coat of human healthy volunteers, and previously described (20,21). Secondary antibodies were con- peripheral blood from diabetic patients (n = 5 current jugated to Alexa Fluor 633 (1:1,000) or Alexa Fluor 488 incretin users and n = 5 never incretin users) and non- (1:1,000) (Life Technologies Italia, Monza, Italy) (20,21). diabetic patients (n = 10) was isolated and characterized as previously described (20,21). Western Blot Analysis Western blot analysis of atherosclerotic plaque sections, In Vitro Cell Treatments EPCs, or EC total protein extracts was performed, as ECs, human aortic ECs, were from Lonza (Cologne, previously described (21), using antibodies against SIRT6 Germany and Walkersville, MD). Cells were cultured (Cell Signaling Technology), NF-kB (Cell Signaling Tech- with an EGM-2MV BulletKit (Lonza) supplemented with nology), and g-tubulin protein (GTU-88) (Sigma-Aldrich). 5% FBS, 0.6% HEPES, maintained at 37°C in a 5% CO 2 As for plaque section homogenization, 800 mL of 2D lysis humidified atmosphere, and used in experiments be- buffer (7 mol/L urea, 2 mol/L thiourea, 4% CHAPS [3-([3- tween passage 4 and 7. EPCs (5 3 106 cells/mL medium) cholamidopropyl]dimethylammonio)-1-propane sulfonate] andECs(culturedat70–80% confluence in six-well buffer, 30 mmol/L Tris-HCl, pH 8.8), were added to tissues plates [Costar, Corning, NY]) were subjected to short- (400 mg) cut into small pieces. Tissue homogenized with term exposure to high glucose (25 mmol/L) in the pres- a Precellys 24 system (Bertin Technologies, Montigny- ence or absence of a GLP-1 receptor agonist (exenatide le-Bretonneux, France) was centrifuged at 800g for 10 min [Byetta]) (100 nmol/L) in complete culture media for at 4°C to collect the supernatant. Proteins were then 3 days. GLP-1 receptor agonists were added 30 min be- precipitated by adding 100% cold methanol. fore starting the high-glucose treatment and were left in the culture media throughout the high-glucose treat- Statistical Analysis ment. Control cells were cultured for 3 days under basal Data are presented as mean 6 SD. Continuous variables conditions. were compared among the groups of patients with one-way 1398 Atherosclerosis in Patients With Diabetes and Sirtuin 6 Diabetes Volume 64, April 2015

ANOVA for normally distributed data and Kruskal-Wallis never incretin users and current incretin users (Table 1 test for non–normally distributed data. When differences and Supplementary Table 1). However, basal and post- among the groups were found, Bonferroni correction to prandial GLP-1 levels were higher in current incretin make pairwise comparisons was used. P , 0.05 was con- users compared with never incretin users (P , 0.01) sidered to be statistically significant. All calculations were (Table 1). performed using SPSS 12. Plaque Composition Compared with nondiabetic patients, diabetic patients RESULTS (n = 52) had a significantly greater portion of plaque area Demographic data for the study population are presented occupied by macrophages and T cells, as well as greater in Table 1 and Supplementary Table 1. The percentage of expression of HLA-DR antigen (Table 1 and Fig. 2). Com- carotid diameter reduction, risk factors, and concomitant pared with the never incretin users group, the current nonhypoglycemic therapy did not differ among the incretin users group presented with a significantly smaller groups (Table 1 and Supplementary Table 1). In diabetic portion of plaque area occupied by macrophages (P , patients, mean, fasting, and postprandial plasma glucose 0.01) and T cells (P , 0.01), as well as lower expression levels at baseline as well as during the week before the of HLA-DR (P , 0.01) (Table 1 and Fig. 2). Both immu- surgery, HOMA-IR, and HbA1c levels did not differ among nohistochemistry and ELISA revealed markedly higher

Table 1—Characteristics of study patients Control subjects Never incretin users Current incretin users (n = 30) (n = 28) (n = 24) Age (years) 71 6 4696 7706 5 Sex Female 17 17 15 Male 13 11 9 Patient characteristics Family history of IHD 14 (47) 17 (60) 13 (54) Hypertension 12 (40) 13 (46) 12 (50) Hypercholesterolemia 12 (40) 13 (46) 10 (41.6) Coronary artery disease 16 (53) 17 (61) 15 (63) BMI (kg/m2) 27.3 6 3 29.8 6 4 28.7 6 2

HbA1c (%) 4.8 6 1.0*† 8.0 6 1.4 7.9 6 1.3 Blood glucose (mmol/L) 6 6 0.6*† 9.5 6 1.4 9.4 6 1.6 Insulin (mU/mL) 8.09 6 3.6*† 10.9 6 2.9 10.4 6 2.7 Basal GLP-1 (pmol/L) 7.9 6 1.1* 5.1 6 1.1 6.6 6 1.4* Postprandial GLP-1 (pmol/L) 22.6 6 3.2*† 11.4 6 2.8 20.1 6 2.8* HOMA-IR 2.14 6 0.8*† 4.60 6 1.5 4.59 6 1.4 CRP (mg/dL) 0.84 6 0.07*† 1.20 6 0.07 1.24 6 0.09 Total cholesterol (mmol/L) 5.66 6 0.08 5.68 6 0.10 5.62 6 0.05 HDL cholesterol (mmol/L) 1.24 6 0.11 1.20 6 0.09 1.21 6 0.09 LDL cholesterol (mmol/L) 3.58 6 0.10 3.59 6 0.11 3.49 6 0.09 Triglycerides (mmol/L) 1.84 6 0.36 1.98 6 0.31 1.97 6 0.39 Plaque characteristics Stenosis severity (%) 77.1 6 6.6 79.2 6 5.9 78.1 6 6.4 Macrophage-rich areas (%) 6 6 2*† 24 6 4176 3* HLA-DR–rich areas (%) 8 6 2*† 28 6 9146 8* T cells/mm2 section area 19.7 6 9.3*† 79.1 6 16.2 28.9 6 11.7* Collagen content (%) 28.2 6 3.8*† 9.7 6 3.9 19.1 6 3.2* p50 (ng/mg) 20.6 6 11.2*† 34.8 6 15.4* 71.1 6 16.4 p65 (ng/mg) 13.5 6 7.7*† 31.3 6 15.1* 59.9 6 16.3 MMP-9 (mg/mg) 4.3 6 2.5*† 17.6 6 2.8 8.8 6 2.1* Nitrotyrosine (nmol/pg) 1.3 6 0.7*† 6.1 6 0.9 3.5 6 1.2* TNF-a (pg/mg) 23.9 6 4.5*† 94.6 6 6.5 61.4 6 7.3* Incretin therapy GLP-1 agonists 4 (17) Exenatide 3 (12) Liraglutide 1 (4) DPP-4 inhibitors 20 (83) Data are presented as mean 6 SD or n (%). CRP, C-reactive protein; IHD, ischemic heart disease; IR, insulin resistance. *P , 0.05 compared with never users group. †P , 0.05 compared with current users group. diabetes.diabetesjournals.org Balestrieri and Associates 1399

Figure 2—Inﬂammation in atherosclerotic plaques. Immunochemistry for macrophages (CD68) (3400), lymphocytes (CD3) (3400), in- ﬂammatory cells (HLA-DR) (3400), and TNF-a (3400) in asymptomatic plaques of a control subject, a never incretin user, and a current incretin user. Similar regions of plaque are shown. These results are typical of the asymptomatic plaques of control subjects, never incretin users, and current incretin users.

staining and levels of TNF-a in all diabetic versus non- B) revealed that levels of SIRT6 are consistently lower in diabetic lesions (P , 0.001). In diabetic patients, staining plaques from diabetic patients compared with plaques and levels of TNF-a were significantly more abundant in from nondiabetic patients and, specifically, in plaques lesions from never incretin users than in lesions from from diabetic never incretin users (P , 0.01). Indeed, current incretin users (P , 0.001) (Table 1 and Fig. 2). levels of SIRT6 in plaques from diabetic current incretin Moreover, TNF-a levels were inversely correlated with users were significantly higher than those observed in GLP-1 levels (r = 20.67, P , 0.001). MMP-9 levels plaques from never incretin users (P , 0.05), and were were more abundant in diabetic than in nondiabetic near the values observed in nondiabetic patients (Figs. 3 lesions (P , 0.001). Specifically, in diabetic patients, and 4A and B). In addition, in order to define the cellular MMP-9 levels were more abundant in lesions from never type expressing SIRT6 within diabetic and nondiabetic incretin users than in lesions from current incretin users plaques, sections were incubated with antibodies against (P , 0.001). As for the content of interstitial collagen, SIRT6 and von Willebrand factor, a specificECmarker. a lower amount of interstitial collagen was found in Results showed the coexpression of SIRT6 and von Wille- plaques of all diabetic patients compared with nondiabetic brand factor, thus suggesting that SIRT6 is expressed by patients (P , 0.001). The amount of interstitial collagen ECs (Fig. 4B). Consistent with immunofluorescence find- in plaques from never incretin users was lower than in ings, Western blot analysis of SIRT6 protein levels in ath- lesions from current incretin users (P , 0.001) (Table 1 erosclerotic plaques from nondiabetic and diabetic patients and Fig. 3). Higher nitrotyrosine levels were found in di- (current and never incretin users) showed a similar trend abetic plaques than in nondiabetic plaques (P , 0.001). (Fig. 4C). Moreover, SIRT6 expression levels in plaques Among diabetic plaques, nitrotyrosine levels were signif- were inversely correlated with GLP-1 levels (r = 20.58, icantly higher in never incretin users than in current P , 0.001). NF-kB activation, as reflected by the selective incretin users (P , 0.01) (Table 1 and Fig. 3). analysis of the activated form of both p50 and p65, was Interestingly, both immunohistochemistry (Fig. 3) and significantly higher in plaques from both current incretin confocal laser-scanning microscopy analyses (Fig. 4A and users (P , 0.01) and never incretin users (P , 0.01) than 1400 Atherosclerosis in Patients With Diabetes and Sirtuin 6 Diabetes Volume 64, April 2015

Figure 3—Atherosclerotic plaque phenotypes. Immunochemistry for MMP-9 (3400), and sirius red staining for collagen (3400), nitrotyrosine (3400), and SIRT6 (3400) content in control subjects, never incretin users, and current incretin users with asymptomatic plaques. Similar regions of plaque are shown. These results are typical of control subjects, never incretin users, and current incretin users with asymptomatic plaques. in those of nondiabetic patients. In plaques from diabetic incretin therapies seem to prevent SIRT6 downregulation patients, the NF-kB level was significantly higher in never (Figs. 3, 4, and 5), we next evaluated whether the detri- incretin users than in current incretin users (Table 1). mental effect of high-glucose concentration on EPCs and ECs (20) is exerted via a SIRT6/NF-kB pathway. Western SIRT6 Protein Levels in EPCs From Diabetic and blot and confocal laser-scanning microscopy analysis Nondiabetic Patients revealed that short-term exposure of early EPCs (Fig. 6) It is already known from experimental and clinical studies and ECs (Fig. 7) to high-glucose concentrations induces (22) that atherosclerosis is associated with a reduced the downregulation of SIRT6 protein (P , 0.01) with number and dysfunction of EPCs. Here, for a better under- a concomitant upregulation of NF-kB protein expression standing of the molecular mechanisms underlying EPC im- (P , 0.01) (Figs. 6 and 7). pairment in atherosclerosis, we looked at the possible The dose-dependent response (from 1 to 1,000 nmol/L) involvement of SIRT6 by evaluating its expression levels in effect of GLP-1 on SIRT6 activity during short-term EPCs isolated from the peripheral blood of 10 nondiabetic exposure to high-glucose levels showed a significant effect patients and 10 diabetic patients (5 current incretin users starting at a concentration of 100 nmol/L (810 6 51 vs. and 5 never incretin users). Western blot analysis revealed 398 6 25 relative fluorescence units in high glucose– that EPCs from diabetic never incretin users had lower values treated cells) (P , 0.05) and with no further increase at of SIRT6 protein arbitrary units and that diabetic current 1,000 nmol/L. GLP-1 alone (from 1 to 1,000 nmol/L) incretin users had levels of SIRT6 higher than those observed showed no significant effect on SIRT6 activity. Thus, a in EPCs from never incretin users (P , 0.05) (Fig. 5). 100 nmol/L concentration was used to test the in vitro In Vitro Effect of High Glucose and GLP-1 Receptor effect of GLP-1 on SIRT6 and NF-kB protein levels in EPCs Agonist on SIRT6 and NF-kB Expression in EPCs and and ECs exposed to high glucose. Notably, in EPCs, ECs GLP-1 receptor agonist (100 nmol/L) significantly coun- In light of the observational data that diabetic never teracted the effect of high-glucose concentrations on incretin users showed decreased levels of SIRT6 protein in both SIRT6 and NF-kB protein levels (P , 0.05) (Fig. either atherosclerotic plaque sections or EPCs, and that 6A–C). diabetes.diabetesjournals.org Balestrieri and Associates 1401

Figure 4—Detection of SIRT6 expression in atherosclerotic plaques. Immunofluorescence detection of SIRT6 was performed on deparaffinized atherosclerotic plaque sections from diabetic and nondiabetic patients. A: Bar graph of SIRT6 arbitrary fluorescence units (AFU) determined by confocal laser-scanning microscopy. B: Representative confocal images of SIRT6 (red) and von Willebrand factor (green). C: Representative bar graph of SIRT6 protein levels determined by Western blot analysis of atherosclerotic plaque homogenates from nondiabetic patients (P1 and P2), diabetic never incretin user patients (P3 and P4), and diabetic current incretin user patients (P5 and P6). Inset, representative image of Western blot analysis. Lanes 1 and 2, nondiabetic patients (P1 and P2). Lanes 3 and 4, diabetic never incretin user patients (P3 and P4). Lanes 5 and 6, diabetic current incretin user patients (P5 and P6). Data are mean 6 SD. *P < 0.01 vs. nondiabetic patients; §P < 0.05 vs. never incretin users.

Similar results were obtained when ECs were subjected unstable phenotype and the downregulation of SIRT6 to high-glucose treatment in the presence or absence of expression. GLP-1 (100 nmol/L) (Fig. 7A–C). Indeed, the presence of A previous study (9) in mice has suggested a role for GLP-1 during high-glucose treatment prevented SIRT6 SIRT6 in inflammation, and a recent experimental study downregulation and NF-kB upregulation (P , 0.05) (4) on ECs has shown that the loss of SIRT6 is associated (Fig. 7A–C). According to Western blot analysis, measure- with the upregulation of genes involved in inflammation, ments of the SIRT6 fluorescence intensity units (Figs. 6D vascular remodeling, and angiogenesis. To date, there has and E, and 7D and E) showed similar results. been no evidence concerning the possible role of SIRT6 in subgroups of high-risk plaques, such as those found in DISCUSSION diabetic patients, and about the specific pathways trans- This study provides novel insights into the relationship ducing environmental stimuli in the modulation of SIRT6 between the SIRT6 pathway and the inflammatory pro- levels in atherosclerotic plaques. In our study, the novelty cess of atherosclerotic plaques of type 2 diabetic patients. is represented by the evidence that EC-associated SIRT6 In particular, SIRT6 protein expression was downregu- expression is markedly downregulated, and is associated lated in diabetic atherosclerotic lesions, compared with with more macrophages, T cells, and HLA-DR+ inflamma- nondiabetic lesions, and the impaired SIRT6 expression tory cells in the diabetic plaques compared with nondia- was associated with higher oxidative stress and higher betic plaques. These results suggest that the presence of NF-kB, proinflammatory cytokine, and MMP-9 levels along an active inflammatory reaction in diabetic plaques may with less interstitial collagen content. On the whole, all be associated with a lower production of SIRT6 in ECs. these factors might increase the risk of future acute is- In line with such evidence, lower expression of SIRT6 and chemic events precipitated by inflammation-dependent concomitantly higher levels of oxidative stress (nitro- rupture of atherosclerotic plaques. Moreover, we provide tyrosine levels) and inflammatory cytokines were found evidence that in diabetic patients the drugs that work on in plaques obtained from the asymptomatic patients with the incretin system, such as GLP-1 receptor agonists and type 2 diabetes compared with nondiabetic patients. In DPP-4 inhibitors, may prevent plaque progression to an agreement with the difference in the SIRT6 staining pattern, 1402 Atherosclerosis in Patients With Diabetes and Sirtuin 6 Diabetes Volume 64, April 2015

In this context, previous reports (14) evidenced the involvement of the incretin system in the regulation of the SIRTs, particularly under conditions of metabolic disease such as in diabetic and high-fat diet–induced obese mice. Incretin system deregulation plays a pivotal role in the pathogenesis of type 2 diabetes, and patients with type 2 diabetes show a significant reduction in meal-stimulated levels of GLP-1 (25). However, no evidence exists on the potential role of GLP-1 in the regulation of SIRT6 in atherosclerotic plaques of diabetic patients. In this study, we observed that plaque SIRT6 expression levels were inversely correlated with GLP-1 levels. SIRT6 antagonizes – Figure 5—SIRT6 expression in EPCs from diabetic and nondiabetic NF-kB induced gene expression programs by associating patients. Western blot analysis of SIRT6 protein levels on EPCs with chromatin-bound NF-kB, directing deacetylation of from nondiabetic and diabetic patients (never incretin users and histone H3 lysine 9 (H3K9), and destabilizing the binding current incretin users) was performed as described in the RESEARCH of NF-kB to chromatin (26). Thus, we hypothesized that DESIGN AND METHODS section. Representative bar graph and image of the Western blot analysis of SIRT6 expression in EPCs from non- the decreased expression of SIRT6 in plaques, as a conse- diabetic patients (P1, lane 1), never incretin users (P2, lane 2), and quence of GLP-1 reduction, may enhance NF-kBactivity, current incretin users (P3, lane 3). Data are expressed as mean 6 and that this could represent a crucial step in the patho- P < P < SD arbitrary units. * 0.05 vs. nondiabetic patients; § 0.05 physiology of diabetic plaque instability. In this context, vs. never incretin users. our data suggest the possibility of a novel pathway to be unveiled through which the incretin system impairment, by reducing SIRT6 expression, could mediate inflammatory the histological milieu of the lesions appears different activity in diabetic atherosclerotic plaques. Thus, the mod- with regard to cellularity, but not in the degree of vessel ulation of SIRT6 through the incretin system could be stenosis, thus suggesting that diabetic and nondiabetic beneficial for many inflammatory diseases associated with lesions are different only with regard to SIRT6 protein endothelial dysfunction (27) and could play a pivotal role in expression and inflammatory burden. These data are con- the stabilization of diabetic atherosclerotic plaques. In this sistent with our previous findings that the inflammatory context, the present findings also show a stimulatory effect response, as well as the oxidative stress level, were higher of the drugs that work on the incretin system, such as GLP-1 in diabetic than in nondiabetic plaques (23). Therefore, in receptor agonists and DPP-4 inhibitors, in the SIRT6 this study, the observed decreased expression of SIRT6 in pathway in diabetic lesions. Indeed, at the same blood carotid plaques of diabetic patients might be related to glucose levels, diabetic patients treated with both GLP-1 the expansion of oxidative and inflammatory processes, receptor agonists and DPP-4 inhibitors had the lowest thinning of the fibrous cap, and plaque instability. These levels of plaque inflammatory cells, cytokines, oxidative results are in agreement with studies showing that in ECs stress, and MMP-9 associated with the highest expression the loss of SIRT6 is paralleled by the increased expression of SIRT6 and amount of plaque interstitial collagen. Thus, of the proinflammatory transcription factor NF-kB, and, patients assigned to incretin-based therapy had less conversely, the overexpression of SIRT6 is associated with plaque progression to an unstable phenotype than patients a downregulation of NF-kB activity as well as the expres- treated without incretin-based therapy. In particular, the sion of its target genes (4). These findings suggest that increased SIRT6 expression observed in diabetic plaques SIRT6, a critical enzyme in the maintenance of genomic from the current incretin users suggests a low inflamma- stability, could also play a key role in reducing oxidative tory activity linked to decreased NF-kB activation. This stress and cellular damage associated with plaque insta- hypothesis is supported by the results of in vitro experi- bility by acting in the pathways controlling for vascular ments on EPCs and ECs showing that the loss of SIRT6 oxidative stress and inflammation (24). Collectively, this during short-term exposure with high glucose is paralleled suggests that SIRT6 is a negative regulator of vascular by the increased expression of NF-kB, whereas overex- oxidative stress and inflammation. In addition, our obser- pression of SIRT6 in the presence of cotreatment with vational results on the negative modulation of SIRT6 lev- GLP-1 relates to a decreased NF-kB expression. In the els in EPCs from diabetic patients open a new scenario in process of vascular inflammation, EPCs and activated the signaling pathways underlying the EPC functional im- ECs are critically involved in the formation of new blood pairment during atherosclerosis and diabetes (20,22). vessels, which plays an important role in several pathol- However, although our findings suggest that diabetes ogies, including atherosclerosis and diabetes (22). A determines an increase of oxidative stress, NF-kB activa- strong correlation between the number of circulating EPCs tion, and SIRT6 expression, the mechanism by which and the combined Framingham risk factor score for ath- SIRT6 acts in the regulation of the type 2 diabetic plaque erosclerosis exists, suggesting that EPCs can be used as phenotype is unknown. a predictive biomarker for cardiovascular risk and vascular diabetes.diabetesjournals.org Balestrieri and Associates 1403

Figure 6—Effect of short-term exposure to high glucose and GLP-1 on SIRT6 and NF-kB protein levels in EPCs. Peripheral blood mononuclear cells (5 3 106 cells/mL in medium), isolated from the leukocyte-rich buffy coat of healthy donor subjects, were subjected to short-term exposure to high glucose (25 mmol/L) in the presence or absence of the GLP-1 receptor agonist exenatide (100 nmol/L) for 3 days. Representative bar graph and image of Western blot analysis of SIRT6 (A and B) and NF-kB(A and C) protein levels in control (Ctr), high glucose–treated cells (hGluc), GLP-1–treated cells (GLP-1), and cells treated with high glucose in the presence of GLP-1 (hGluc + GLP-1). D: Representative confocal images of SIRT6 (red) and vimentin (green). E: Bar graph of SIRT6 ﬂuorescence intensity (in arbitrary ﬂuorescence units [AFU]) determined by confocal laser-scanning microscopy. Data are mean 6 SD (n = 6). *P < 0.01 vs. Ctr; §P < 0.05 vs. hGluc; ‡P < 0.01 vs. hGluc.

function (22). Our data, according to reports showing that exenatide (16), a GLP-1 agonist, have recently been shown the loss of SIRT6 in ECs is associated with increased ex- to exert a rapid and significant anti-inflammatory effect pression of intracellular adhesion molecule-1, NF-kB level, that is evident within 2 h of administration of the first and senescence (4), shed light on the role of SIRT6 as the dose of each of these drugs. Such results support the hy- regulator of endothelial function during altered glucose pothesis that the anti-atherosclerotic effect we observed homeostasis within a pathway that links inflammation, may also be due to a direct anti-inflammatory effect that metabolic diseases, and atherosclerosis. is independent of the effect on glycemic control. However, As a whole, our data might have strong clinical the full mechanism by which GLP-1 acts on SIRT6, which implications because in a large series of carotid endarter- was not uncovered by this study, requires further investi- ectomy specimens it has been shown that plaque in- gation. Indeed, we did not identify the complete mecha- flammation is one of the major determinants of ischemic nism by which GLP-1 can regulate the inflammatory status events in patients affected by carotid atherosclerotic of atherosclerotic plaque through the modulation of SIRT6 disease (28). Therefore, the direct anti-inflammatory levels. This study highlights the relevance of modulating effects of GLP-1 analogs and DPP-4 inhibitors, which in- SIRT6/NF-kB signaling with incretin-based therapy to re- crease plasma concentrations of GLP-1, and go above and duce the inflammatory burden of diabetic plaques. At the beyond glycemic control, should also be considered. In moment, we do not exclude the possibility that GLP-1 could addition to their metabolic actions, various beneficial car- inhibit NF-kB activation through additional mechanisms diovascular effects have been described for GLP-1–based independently of SIRT6. Indeed, it has previously been therapies (29). It has previously been reported (11) that reported (32) that exendin-4, a GLP-1 receptor agonist, GLP-1 reduces in vitro monocyte migration, while the attenuates atherosclerosis through cAMP-dependent pro- DPP-4 inhibitor sitagliptin reduced plaque inflammation tein kinase-phosphatidylinositol 3-kinase/Akt-endothelial 2 2 and enhanced plaque stability in ApoE / mice. Similar nitric oxide synthase-p38 mitogen-activated protein effects of DPP-4 inhibitors on plaque inflammation were kinase-Jun NH2-terminal kinase–dependent pathways via observed (30). Finally, sitagliptin, vildagliptin (15,31), and aGLP-1receptor–dependent mechanism, without affecting 1404 Atherosclerosis in Patients With Diabetes and Sirtuin 6 Diabetes Volume 64, April 2015

Figure 7—Effect of short-term exposure to high glucose and GLP-1 on SIRT6 and NF-kB protein levels in ECs. ECs were seeded in six-well plates at 70–80% confluence 12 h before treatments with high glucose (25 mmol/L) in the presence or absence of the GLP-1 receptor agonist exenatide (100 nmol/L). ECs were cultured for 3 days in complete media alone (control [Ctr]), with high glucose (hGluc), GLP-1, or high glucose in the presence of GLP-1 (hGluc + GLP-1). Representative bar graph and image of Western blot analysis of SIRT6 (A and B) and NF-kB(A and C) protein levels. D: Representative confocal images of SIRT6 (red) and vimentin (green). E: Bar graph of SIRT6 fluorescence intensity (in arbitrary fluorescence units [AFU]). Data are mean 6 SD (n = 6). *P < 0.05 vs. Ctr; **P < 0.01 vs. Ctr; §P < 0.05 vs. hGluc; ‡P < 0.01 vs. hGluc. metabolic parameters. Furthermore, recent evidence effects of incretin-mimetic drugs on cardiovascular events has shown a promising role for DPP-4 inhibitors in the and mortality cannot be considered to be established un- attenuation of atherosclerosis involving vascular dysfunc- less they are evaluated in long-term cardiovascular out- tion and endothelial inflammation by upregulating the comes trials in patients without cardiovascular diseases. Akt/endothelial nitric oxide synthase signaling pathway In conclusion, we demonstrated that GLP-1 reduced and suppressing the activation of extracellular signal– plaque macrophage infiltration and MMP-9 expression, related kinase 1/2 in vascular tissues (33). However, de- which resulted in increased plaque collagen content and spite the fact that incretin-mimetic drugs have shown a thickened fibrous cap in diabetic plaques from humans. favorable effects on pathogenic mechanisms of athero- As these plaque characteristics are features of plaque sclerosis, the recently published large outcome trials stability, the results of this study suggest that GLP-1 (34,35) have not shown any superiority of treatment reduces plaque vulnerability. These findings may be of with a DPP-4 inhibitor (alogliptin or saxagliptin) for the clinical importance for patients with type 2 diabetes, who outcomes of a composite of atherosclerosis-related events. are known to exhibit a higher burden of inflamed, rupture- However, it should be pointed out that these trials were prone atherosclerotic plaques in comparison with non- mandated by the U.S. Food and Drug Administration diabetic subjects (37). Overall, it is tempting to speculate simply to rule out cardiovascular harm. Furthermore, it that the use of activators of SIRT6 might be an effective should be noted that in both studies patients were ran- strategy for treating inflammatory vascular diseases such domized to receive DPP-4 inhibitors or placebo on top of as diabetes and atherosclerosis. Future studies are needed standard therapy, usually metformin (alone or in combi- to determine whether the beneficial effects of GLP-1 re- nation with other agents), which was taken by about two- ceptor agonists and DPP-4 inhibitors on features of plaque thirds of the patients. Moreover, almost half of the vulnerability through the SIRT6 pathway translate into a re- patients were taking sulfonylureas, which may be associ- duction of cardiovascular events in patients with type 2 ated with increased mortality (36). Finally, both studies diabetes treated with GLP-1–based regimens. Moreover, were conducted in patients with cardiovascular disease, this study also raises an important question about whether and, therefore, the assessment of the prevention of events asymptomatic diabetic patients with .70% carotid artery with incretin therapy may not be accurate. However, the stenosis who are receiving GLP-1–based therapies really diabetes.diabetesjournals.org Balestrieri and Associates 1405 need to undergo carotid endarterectomy, because they are 11. Vittone F, Liberman A, Vasic D, et al. Sitagliptin reduces plaque macrophage likely to have stable plaques. Prospectively randomized, content and stabilises arteriosclerotic lesions in Apoe (-/-) mice. Diabetologia large-scale clinical trials are required to further clarify 2012;55:2267–2275 this relationship. Finally, although the numbers of incretin 12. Arakawa M, Mita T, Azuma K, et al. Inhibition of monocyte adhesion to endothelial cells and attenuation of atherosclerotic lesion by a glucagon-like users are small, it would be of interest to know whether peptide-1 receptor agonist, exendin-4. Diabetes 2010;59:1030–1037 administration of GLP-1 agonists showed more SIRT6 ex- 13. Shah Z, Kampfrath T, Deiuliis JA, et al. Long-term dipeptidyl-peptidase 4 pression compared with administration of DPP-4 inhibi- inhibition reduces atherosclerosis and inflammation via effects on monocyte tors. Moreover, further investigations are required to recruitment and chemotaxis. Circulation 2011;124:2338–2349 determine whether GLP-1 agonists are more effective 14. Lee J, Hong SW, Chae SW, et al. Exendin-4 improves steatohepatitis by than DPP-4 inhibitors in modulating SIRT6 expression increasing Sirt1 expression in high-fat diet-induced obese C57BL/6J mice. PLoS and activity. Also, within the class of GLP-1 agonists, an One 2012;7:e31394 evaluation of the effect of the rules for GLP-1 agonist 15. Makdissi A, Ghanim H, Vora M, et al. Sitagliptin exerts an antinflammatory administration (once or twice daily and weekly) on SIRT6 action. J Clin Endocrinol Metab 2012;97:3333–3341 expression needs to be undertaken. 16. Chaudhuri A, Ghanim H, Vora M, et al. Exenatide exerts a potent antiin- flammatory effect. J Clin Endocrinol Metab 2012;97:198–207 17. Young B, Moore WS, Robertson JT, et al. An analysis of perioperative surgical mortality and morbility in the asymptomatic carotid atherosclerosis study. ACAS In- Funding. This research was supported by Ricerca Ateneo 2006–2010 and the vestigators. Asymptomatic Carotid Atherosclerosis Study. Stroke 1996;27:2216–2224 Second University of Naples. 18. Moghissi ES, Korytkowski MT, DiNardo M, et al.; American Association of Duality of Interest. No potential conflicts of interest relevant to this article Clinical Endocrinologists; American Diabetes Association. American Association were reported. of Clinical Endocrinologists and American Diabetes Association consensus Author Contributions. M.L.B. and R.M. contributed to the conception, statement on inpatient glycemic control. Diabetes Care 2009;32:1119–1131 design, analysis, and interpretation of the data, and the drafting and revision of 19. Marfella R, D’Amico M, Di Filippo C, et al. Increased activity of the ubiquitin- the manuscript; and gave final approval of the submitted manuscript. M.R.R., proteasome system in patients with symptomatic carotid disease is associated with M.B., N.D.O., A.G., M.S., F.F., and L.S. contributed to the analysis and inter- enhanced inflammation and may destabilize the atherosclerotic plaque: effects of pretation of the data, and approved the submitted manuscript. P.P., F.M., D.D.A., rosiglitazone treatment. J Am Coll Cardiol 2006;47:2444–2455 C.M., and F.C. contributed to the drafting and revision of the manuscript, and 20. Balestrieri ML, Servillo L, Esposito A, et al. Poor glycaemic control in type 2 approved the submitted manuscript. C.S. contributed to the analysis of the data. diabetes patients reduces endothelial progenitor cell number by influencing P.C. and G.P. contributed to the conception and design of the study, and ap- SIRT1 signalling via platelet-activating factor receptor activation. Diabetologia proved the submitted manuscript. R.M. is the guarantor of this work and, as 2013;56:162–172 such, had full access to all the data in the study and takes responsibility for the 21. Balestrieri ML, Giovane A, Milone L, Servillo L. Endothelial progenitor cells integrity of the data and the accuracy of the data analysis. express PAF receptor and respond to PAF via Ca(2+)-dependent signaling. Bio- chim Biophys Acta 2010;1801:1123–1132 References 22. Du F, Zhou J, Gong R, et al. Endothelial progenitor cells in atherosclerosis. 1. Grundy SM, Benjamin IJ, Burke GL, et al. Diabetes and cardiovascular Front Biosci (Landmark Ed) 2012;17:2327–2349 disease: a statement for healthcare professionals from the American Heart As- 23. Marfella R, D’Amico M, Esposito K, et al. The ubiquitin-proteasome system sociation. Circulation 1999;100:1134–1146 and inflammatory activity in diabetic atherosclerotic plaques: effects of rosiglita- 2. Creager MA, Lüscher TF, Cosentino F, Beckman JA. Diabetes and vascular zone treatment. Diabetes 2006;55:622–632 disease: pathophysiology, clinical consequences, and medical therapy: part I. 24. Etchegaray JP, Zhong L, Mostoslavsky R. The histone deacetylase SIRT6: at Circulation 2003;108:1527–1532 the crossroads between epigenetics, metabolism and disease. Curr Top Med 3. Michishita E, Park JY, Burneskis JM, Barrett JC, Horikawa I. Evolutionarily Chem 2013;13:2991–3000 conserved and nonconserved cellular localizations and functions of human SIRT 25. Drucker DJ, Nauck MA. The incretin system: glucagon-like peptide-1 re- proteins. Mol Biol Cell 2005;16:4623–4635 ceptor agonists and dipeptidyl peptidase-4 inhibitors in type 2 diabetes. Lancet 4. Lappas M. Anti-inflammatory properties of sirtuin 6 in human umbilical vein 2006;368:1696–1705 endothelial cells. Mediators Inflamm 2012;2012:597514 26. Beauharnois JM, Bolívar BE, Welch JT. Sirtuin 6: a review of biological 5. Kanfi Y, Peshti V, Gil R, et al. SIRT6 protects against pathological damage effects and potential therapeutic properties. Mol Biosyst 2013;9:1789–1806 caused by diet-induced obesity. Aging Cell 2010;9:162–173 27. Sundaresan NR, Vasudevan P, Zhong L, et al. The sirtuin SIRT6 blocks IGF- 6. Grimley R, Polyakova O, Vamathevan J, et al. Over expression of wild type Akt signaling and development of cardiac hypertrophy by targeting c-Jun. Nat or a catalytically dead mutant of Sirtuin 6 does not influence NFkB responses. Med 2012;18:1643–1650 PLoS One 2012;7:e39847 28. Spagnoli LG, Mauriello A, Sangiorgi G, et al. Extracranial thrombotically active 7. Kim HS, Xiao C, Wang RH, et al. Hepatic-specific disruption of SIRT6 in mice carotid plaque as a risk factor for ischemic stroke. JAMA 2004;292:1845–1852 results in fatty liver formation due to enhanced glycolysis and triglyceride syn- 29. Burgmaier M, Heinrich C, Marx N. Cardiovascular effects of GLP-1 and GLP- thesis. Cell Metab 2010;12:224–236 1-based therapies: implications for the cardiovascular continuum in diabetes? 8. Michishita E, McCord RA, Berber E, et al. SIRT6 is a histone H3 lysine 9 Diabet Med 2013;30:289–299 deacetylase that modulates telomeric chromatin. Nature 2008;452:492–496 30. Matsubara J, Sugiyama S, Sugamura K, et al. A dipeptidyl peptidase-4 9. Kawahara TLA, Michishita E, Adler AS, et al. SIRT6 links histone H3 lysine 9 inhibitor, des-fluoro-sitagliptin, improves endothelial function and reduces ath- deacetylation to NF-kappaB-dependent gene expression and organismal life erosclerotic lesion formation in apolipoprotein E-deficient mice. J Am Coll Cardiol span. Cell 2009;136:62–74 2012;59:265–276 10. Favero G, Rezzani R, Rodella LF. Sirtuin 6 nuclear localization at cortical 31. Barbieri M, Rizzo MR, Marfella R, et al. Decreased carotid atherosclerotic brain level of young diabetic mice: an immunohistochemical study. Acta Histo- process by control of daily acute glucose fluctuations in diabetic patients treated chem 2014;116:272–277 by DPP-IV inhibitors. Atherosclerosis 2013;227:349–354 1406 Atherosclerosis in Patients With Diabetes and Sirtuin 6 Diabetes Volume 64, April 2015

32. Erdogdu O, Eriksson L, Xu H, Sjöholm A, Zhang Q, Nyström T. Exendin-4 35. Scirica BM, Bhatt DL, Braunwald E, et al.; SAVOR-TIMI 53 Steering Com- protects endothelial cells from lipoapoptosis by PKA, PI3K, eNOS, p38 MAPK, and mittee and Investigators. Saxagliptin and cardiovascular outcomes in patients JNK pathways. J Mol Endocrinol 2013;50:229–241 with type 2 diabetes mellitus. N Engl J Med 2013;369:1317–1326 33. Zeng Y, Li C, Guan M, et al. The DPP-4 inhibitor sitagliptin attenuates the 36. Monami M, Genovese S, Mannucci E. Cardiovascular safety of sulfonyl- progress of atherosclerosis in apolipoprotein-E-knockout mice via AMPK- and ureas: a meta-analysis of randomized clinical trials. Diabetes Obes Metab 2013; MAPK-dependent mechanisms. Cardiovasc Diabetol 2014;13:32–36 15:938–953 34. White WB, Cannon CP, Heller SR, et al.; EXAMINE Investigators. Alogliptin 37. Moreno PR, Murcia AM, Palacios IF, et al. Coronary composition and after acute coronary syndrome in patients with type 2 diabetes. N Engl J Med macrophage inﬁltration in atherectomy specimens from patients with diabetes 2013;369:1327–1335 mellitus. Circulation 2000;102:2180–2184