Impact of Glycemic Variability on Chromatin Remodeling, Oxidative Stress, and Endothelial Dysfunction in Patients With Type 2 Diabetes and With Target HbA_1c Levels

The prevalence of type 2 diabetes (T2D) is extremely high, with ∼415 million people affected worldwide. Most importantly, this number is expected to rise to 642 million by the year 2040 (1,2). In the constellation of diabetes-related comorbidities, cardiovascular disease (CVD) carries the largest burden (3,4). Although a consistent body of evidence has unmasked a major pathophysiological role of hyperglycemia in the development and progression of vascular complications (5), the attempt to control CVD progression in T2D with intensive glycemic control (IGC) has been a disappointment for a long time (6). Indeed, ACCORD (Action to Control Cardiovascular Risk in Diabetes), ADVANCE (Action in Diabetes and Vascular Disease: Preterax and Diamicron MR Controlled Evaluation), and VADT (Veterans Affairs Diabetes Trial) have almost unanimously reported that IGC with a significant reduction of HbA_1c is not able to improve cardiovascular outcomes in patients with long-standing hyperglycemia (7,8). Recently, EMPA-REG OUTCOME (BI 10773 [Empagliflozin] Cardiovascular Outcome Event Trial in Type 2 Diabetes Mellitus Patients), LEADER (Liraglutide Effect and Action in Diabetes: Evaluation of Cardiovascular Outcome Results—A Long Term Evaluation), and SUSTAIN-6 (Trial to Evaluate Cardiovascular and Other Long-term Outcomes With Semaglutide in Subjects With Type 2 Diabetes 6) have shown a remarkable benefit of empagliflozin, liraglutide, and semaglutide on cardiovascular outcomes. However, cardiovascular benefit of these emerging antidiabetic agents was also largely explained by nonglycemic factors, namely blood pressure reduction, osmotic diuresis, and anti-inflammatory effects (9). Furthermore, a growing body of experimental evidence suggests that the hyperglycemic environment is somehow remembered in the vascular system (6,10). Whether this phenomenon occurs in patients with T2D remains to be elucidated. Epigenetic modifications are emerging as important players in CVD (11,12). Indeed, alterations of the epigenome may significantly affect the expression of oxidant and inflammatory genes (13). Major mechanisms of epigenetic regulation are DNA methylation of cytosine-phosphate-guanine (CpG) dinucleotide sequences and acetylation of histone proteins. DNA methylation is an important repressor of gene expression, whereas acetylation of histone tails favors an open chromatin, leading to active transcription (11). Reduced DNA methylation and increased histone acetylation represent an adverse epigenetic pattern that leads to dysregulation of genes with detrimental effects for cellular homeostasis (14). The adaptor protein p66^Shc is a key driver of mitochondrial oxidative stress and vascular damage in experimental diabetes (15–17). Indeed, diabetic p66^Shc−/− mice are protected against hyperglycemia-induced endothelial dysfunction and vascular redox changes (16). We have shown that epigenetic remodeling of p66^Shc is responsible for the persistence of endothelial dysfunction in diabetic mice despite glycemic control with insulin (18). Consistent with these findings, other investigators reported that transient spikes of hyperglycemia trigger inflammation through chromatin changes that persist even after restoration of normoglycemia (19,20). Additional evidence indicates that glucose fluctuations may also exert detrimental effects on the vasculature as a result of their ability to derail pathways implicated in cardiovascular homeostasis. In patients with T2D, glycemic variability during postprandial hyperglycemic swings have been reported to exert a triggering effect on oxidative stress compared with chronic sustained hyperglycemia (21). Whether glucose fluctuations affect chromatin structure and activity in humans is largely unknown. In the current study, we postulate that glycemic variability causes persistent epigenetic remodeling of p66^Shc and vascular dysfunction in patients with T2D. Understanding the relationship among glucose variability, p66^Shc-related epigenetic changes, and vascular disease in the clinical setting may have major implications for the development of mechanism-based therapeutic strategies in patients with T2D.

Between January and December 2012, we recruited 39 consecutive patients with uncontrolled T2D (HbA_1c >7.5%) in the outpatient service of Sant’Andrea and Agostino Gemelli University Hospitals in Rome, Italy. Exclusion criteria were overt atherosclerotic vascular disease as well as other relevant comorbidities. Individuals with an estimated glomerular filtration rate of <60 mL/min/1.73 m² were also excluded. Twenty-four healthy subjects of similar age and sex were recruited during the same period. Control subjects were not taking medications, and their blood pressure was <130/80 mmHg, LDL cholesterol <160 mg/dL, and fasting plasma glucose <100 mg/dL. A medical history was taken from all participants followed by anthropometric measurements and blood and urine sampling. The study was carried out according to the ethical principles stated in the Declaration of Helsinki. The protocol was approved by local ethics committee (Sant'Andrea Hospital and Agostino Gemelli University Hospital, Rome, Italy) and in accordance with institutional guidelines, and all participants were aware of the investigational nature of the study and gave written consent.

After enrollment, an IGC program was implemented in all patients with diabetes aiming to achieve an HbA_1c level ≤7% in accordance with current recommendations (22,23). All participants were subjected to follow-up visits at months 1, 3, and 6. Glycemic control was assessed by serial HbA_1c determinations during the study. Patients with T2D received diet and lifestyle counseling, glucose monitoring equipment, and antidiabetic medications. In addition to lifestyle modifications, the antidiabetic treatment at baseline was uptitrated and/or new glucose-lowering agents added if HbA_1c levels remained >7% or if >50% of premeal and postmeal capillary glucose readings were >100 mg/dL or 140 mg/dL, respectively. All drug combinations were allowed, and medications were reduced only in the presence of adverse effects or major hypoglycemic events (requiring intervention of a third party). The following analyses were performed at baseline in both study groups: 1) epigenetic changes of p66^Shc promoter in isolated peripheral blood monocytes (DNA methylation and histone 3 [H3] acetylation), 2) 24-h urinary excretion rates of 8-isoprostaglandin F_2α (8-isoPGF_2α), and 3) brachial artery flow-mediated dilation (FMD). After 6-months of the IGC program, patients with T2D repeated the same tests and performed continuous glucose monitoring (CGM) to assess markers of glycemic instability, such as the mean amplitude of glycemic excursion (MAGE) and postprandial incremental area under the curve (AUCpp) of blood glucose levels.

Blood was collected in a BD Vacutainer CPT Mononuclear Cell Preparation Tube–Sodium Heparin (BD Biosciences, Franklin Lakes, NJ) and centrifuged for 20 min at 1,800g at room temperature. The turbid white layer above the Ficoll-paque density gradient containing the mononuclear blood cells was transferred to a clean tube and washed twice with PBS. Subsequently, monocytes were isolated by using magnetic CD14-coated beads and magnetic activated cell sorting (Miltenyi Biotec, Bergisch Gladbach, Germany) (24).

PCR experiments for p66^Shc gene were performed with a TaqMan Gene Expression Assay kit and TaqMan Gene Expression Master Mix (Applied Biosystems, Foster City, CA). A p66^Shc predesigned primer (Hs01050695_g1; Applied Biosystems) and TATA-box binding protein (TBP) (Hs00427620_m1; Applied Biosystems) as endogenous control for normalizing RNA concentration were used.

Real-time PCR for SIRT1 (forward: 5′-GCCGGAAACAATACCTCCAC-3′; reverse: 5′-ACCCCAGCTCCAGTTAGAAC-3′) and DNMT3b (forward: 5′-AGTGACACGGGGCTTGAATA-3′; reverse: 5′-CTTCACGGTTCCAACAGCAA-3′) were performed in an Mx3000P PCR cycler (Stratagene) with SYBR Green JumpStart Taq ReadyMix (Sigma Aldrich, St. Louis, MO). TBP (forward: 5′-CGTGGCTCTCTTATCCTCATG-3′; reverse: 5′-GCCCGAACCGCCGAATATA-3′) were used as endogenous controls for normalizing RNA concentration. Differences in cycle threshold (Ct) values between test gene and endogenous controls (TBP, ΔCt) were calculated and used for statistical analysis.

Genomic DNA (gDNA) from peripheral blood monocytes was obtained by using the DNeasy Blood & Tissue Kit (QIAGEN, Hilden, Germany) according to the manufacturer’s instructions. One microgram of gDNA was denatured, and unmethylated cytosines were converted to uracil in the denatured samples (Cells-to-CpG Bisulfite Conversion Kit; Applied Biosystems). CpG methylation was quantified by methylation-specific real-time PCR by using 100 ng bisulfite-converted gDNA as the template and methylation-specific primers for CpG in the human p66^Shc promoter. Because the p66^Shc promoter region from −100 to +40 base pairs is rich in CpG island, a specific primer for methylated (forward: 5′-GTTTAGGTTTATTGTATGGGGTAGC-3′; reverse: 5′-CCTTCCTATCCTAATTAAACACTCG-3′) and unmethylated DNA (forward: 5′- TTAGGTTTATTGTATGGGGTAGTGG-3′; reverse: 5′-TTCCTATCCTAATTAAACACTCAAA-3′) located in this specific region (−45 to +45 base pairs) was predesigned. The methylation level of the p66^Shc promoter was calculated by using the methylation index as previously reported (25).

Chromatin immunoprecipitation (ChIP) assay was performed by using the Magna ChIP Assay Kit (Millipore, Billerica, MA) according to the manufacturer’s instructions. ChIP was performed with 10 μg anti-acetylated H3 antibody (06-599; Millipore) and an equivalent amount of mouse IgG as negative control. Washes and elution of the immunoprecipitant DNA were performed according to the Magna ChIP protocol (Millipore). ChIP quantifications to p66^Shc promoter were performed by real-time PCR (forward 1: 5′- ATTGCCTCATTCTCACCCTTG-3′; reverse 1: 5′-GCCAAGAGGAAGAGCAAAGC-3′; forward 2: 5′-GCAGATGTGTCTTCTGATCTCTCTGT-3′; reverse 2: 5′-TGAGAATGAGGCAATCAGGGTCC-3′; forward 3: 5′-CACCAGCTTTGCTCTTCCTCTTG-3′; reverse 3: 5′-ACAGTAAGCCTGGGCCATTAGC-3′).

Both at baseline and at follow-up, 24-h urinary samples were collected and incubated with the antioxidant 4-hydroxy-TEMPO (1 mmol/L) and immediately stored at −80°C until analyses for 8-isoPGF_2α. Levels of isoPGF_2α were assessed by using a commercially available kit (Cell Biolabs, San Diego, CA) according to the manufacturer’s instructions.

Endothelial-dependent vasodilation was assessed as dilation of the brachial artery in response to increased blood flow. Two expert sonographers (L.C. and G.R.) carried out the examinations with Doppler echocardiography (26,27). Endothelium-independent vasodilation was elicited by the administration of a low dose (25 μg) of sublingual glyceryl trinitrate (GTN). Recording time frames were 10 min for FMD studies (1 min for baseline, 5 min of ischemic period, 4 min for assessing changes in diameter after reactive hyperemia) and 6 min for GTN-mediated dilation (1 min for baseline, 5 min for assessing changes in diameter after GTN administration) (26,27).

Subcutaneous interstitial glucose levels were monitored on an ambulatory basis over 3 consecutive days by using a second-generation CGM system (iPro2; Medtronic). The sensor was inserted on day 0 and removed on day 3 at midmorning. The data were downloaded to a computer for evaluation of glucose variations, but these calculations were limited to data obtained on days 1 and 2 to avoid bias as a result of both insertion and removal of the sensor (insufficient stabilization of the monitoring system). The characteristic glucose pattern of each patient was calculated by averaging the profiles obtained on study days 1 and 2. MAGE and AUCpp were calculated as previously reported (21).

The normality of continuous variables was assessed by the Kolmogorov-Smirnov test. All normally distributed variables are expressed as mean (SD), unless otherwise stated. Data not normally distributed are shown as median (interquartile range [IQR]). Comparisons of continuous variables between control subjects and patients with T2D were performed by using unpaired two-sample t and Mann-Whitney U tests, as appropriate, whereas comparisons of variables at enrollment and 6-month follow-up in the T2D group were done by paired t or Wilcoxon test, as indicated. Multiple comparisons between normally distributed variables were performed by one-way ANOVA followed by Bonferroni correction. Non-Gaussian variables were compared by the Kruskal-Wallis test followed by Dunn’s post hoc test. Between-variable correlations were assessed by Spearman test. Multiple linear regression models were built to explore the independent link between glycemic markers (AUCpp, MAGE, and HbA_1c) and epigenetic changes of the p66^Shc promoter. Regressions were adjusted for potential confounders, namely age, sex, BMI, and antidiabetic treatment. P < 0.05 was considered statistically significant. We calculated the number of patients with T2D patients required for the study to reject the null hypothesis 99% of the time (one-tailed type II error rate of 0.01) when r ≥0.80 with a two-tailed type I error at the 0.05 level of significance (28). All statistical analyses were performed by using GraphPad Prism version 5.0 and SPSS version 20 software.

Clinical and laboratory characteristics of patients with T2D and control subjects are shown in Table 1. No significant age and sex differences existed between groups, whereas blood glucose, HbA_1c levels, BMI, waist circumference, and triglyceride levels were increased in the T2D group. In the T2D group, the mean disease duration was 6.9 (5.6) years. At baseline, 38 patients with T2D (97.5%) were receiving glucose-lowering treatment with hypoglycemic agents, insulin, or their combination, and 1 (2.6%) patient was not taking any glucose-lowering drug. After enrollment, an IGC program was implemented in all patients with T2D aiming to achieve an HbA_1c ≤7% (Fig. 1A). Comparison of clinical and laboratory data in patients with T2D at baseline and follow-up are shown in Table 2. Median HbA_1c decreased from 7.8% (IQR 7.5–8.5%) to 6.6% (6.3–7.0%; P < 0.001). No major hypoglycemic events (requiring intervention) were observed during follow-up. Anthropometric parameters (body weight/BMI), and other cardiovascular risk factors (blood pressure, lipids) did not change throughout the study.

Baseline endothelial function, as assessed by FMD of the brachial artery, was significantly impaired in patients with T2D compared with control subjects (median 4.9% [IQR 3.7–5.8%] vs. 8.5% [7.3–9.8%]; P < 0.001) (Fig. 1B). Endothelium-independent dilatation to nitroglycerine was comparable in the two groups (11.85% [10.9–13.6%] vs. 11.80% [8.5–14.8%]; P = 0.69), and no differences were observed in arterial diameter as well as in resting or hyperemic flow (data not shown). Patients with T2D also showed higher 24-h urinary excretion rates of 8-isoPGF_2α, an in vivo marker of oxidative stress (360.9 [351.2–411.5] vs. 155.8 [95.2–206.3] pg/mg of creatinine; P < 0.001) (Fig. 1C). Of note, 6-month IGC did not improve FMD (4.8% [4.1–6.2%]; P = 0.16) and 8-isoPGF_2α (357.5 [342.9–373.7] pg/mg of creatinine; P = 0.10) compared with baseline values, respectively (Fig. 1B and C ).

The p66^Shc gene expression was significantly higher in peripheral blood monocytes isolated from patients with T2D than in control subjects (4.87 [2.91] vs. 1.00 [0.55] arbitrary units [AU]; P < 0.001) (Fig. 1D). Linear regression models showed that p66^Shc mRNA levels were independently associated with 8-isoPGF_2α urinary excretion and FMD, regardless of potential confounders (Supplementary Table 1). We found that upregulation of p66^Shc was not reverted by IGC (4.77 [2.88] AU; P = 0.39 vs. baseline), suggesting that persistent p66^Shc overexpression may contribute to ongoing oxidative stress and vascular dysfunction in this setting (Fig. 1D).

Bisulfite analysis of transcriptionally active regions of p66^Shc promoter revealed that DNA methylation was significantly reduced in peripheral blood monocytes isolated from patients with T2D compared with control subjects (43.5% [19.3%] vs. 100% [0.37%]; P < 0.001), and IGC was not able to reverse such a detrimental signature (49.2% [22.7%]; P = 0.53 vs. baseline) (Fig. 2A). To understand the mechanisms of reduced CpG methylation, we assessed the expression of the methyltransferase DNMT3b, an important methyl-writing enzyme involved in the maintenance of DNA methylation (12). DNMT3b expression was inhibited in patients with T2D compared with control subjects, and such downregulation was not affected by IGC (Fig. 2B). Furthermore, ChIP experiments revealed that the interaction between DNMT3b and p66^Shc promoter in control subjects was strongly reduced in patients with T2D, and IGC could not rescue such an interaction (Fig. 2C). This latter finding confirms that DNMT3b-dependent methylation of p66^Shc promoter is suppressed in patients with T2D and remains unchanged despite reduction of HbA_1c levels.

Because DNA hypomethylation triggers gene expression by clustering with histone acetylation (12), we determined the acetylation status of H3 on p66^Shc promoter. H3 acetylation bound to p66^Shc promoter was increased in patients with T2D compared with control subjects (345.4 [185.4] vs. 100 [37.0] AU; P < 0.001) (Fig. 3A). Such a posttranslational mechanism of active transcription was not erased by IGC (354.4 [206.2] AU; P = 0.90 vs. baseline) (Fig. 3A). Accordingly, we found that the chromatin-modifying enzyme SIRT1 involved in H3 deacetylation (29) was downregulated in patients with T2D regardless of glycemic control (Fig. 3B). ChIP analysis showed that SIRT1-dependent deacetylation of p66^Shc promoter was markedly reduced in patients with T2D and not affected by IGC (Fig. 3C). To further strengthen these findings, we investigated the correlation between these epigenetic changes and p66^Shc gene expression. Both DNA methylation (r = −0.51; P < 0.009) and H3 acetylation (r = 0.48; P < 0.01) significantly correlated with p66^Shc gene transcription within the T2D cohort.

The results obtained so far suggest that the reduction of HbA_1c levels is unable to reprogram the adverse chromatin pattern underlying persistent p66^Shc transcription. Hence, we investigated whether glycemic excursions rather than HbA_1c explain the epigenetic pattern of p66^Shc promoter. At follow-up, CGM was performed in patients with T2D to assess MAGEs and AUCpp. After 3-day monitoring, mean values of MAGE and AUCpp were 84.9 (29.8) mg/dL and 457.6 (842) mg/dL · h, respectively. Only subjects with values >50th percentile of MAGE and AUCpp showed an adverse epigenetic remodeling of p66^Shc promoter (Fig. 4A and B). By contrast, the epigenetic profile did not differ between patients with HbA_1c above and below the median value (Fig. 4C). Linear regression models adjusted for age, sex, BMI, and glucose-lowering treatment confirmed that MAGE and AUCpp were independently associated with adverse epigenetic signatures on p66^Shc promoter (Supplementary Table 2). Taken together, these findings indicate that glycemic variability promotes chromatin changes, leading to persistent vascular dysfunction despite IGC (Fig. 5).

This study demonstrates that the reduction of HbA_1c levels cannot inhibit the overexpression of mitochondrial adaptor p66^Shc, resulting in persistent oxidative stress and endothelial dysfunction. Several lines of evidence support these conclusions. First, epigenetic changes of p66^Shc promoter, namely DNA hypomethylation and H3 acetylation, promote gene transcription in patients with T2D, and IGC does not reverse them. Second, persistent downregulation of methyl-writing DNMT3b and acetyl-erasing SIRT1 enzymes favors the adverse chromatin remodeling responsible for continuous p66^Shc upregulation. Finally, AUCpp and MAGE but not HbA_1c are independently associated with these epigenetic signatures.

Experimental studies in human endothelial cells have suggested that transient hyperglycemia elicits long-lasting epigenetic changes of oxidant and inflammatory genes, which may account for sustained cellular damage despite restoration of normoglycemic conditions (19,30). Consistently, we showed that adaptor protein p66^Shc is upregulated in diabetic mice and that such overexpression is not affected by restoration of normoglycemia. This phenomenon was associated with sustained generation of mitochondrial reactive oxygen species (ROS), reduced nitric oxide availability, and persistent endothelial dysfunction (18). In the current experimental model, we found that p66^Shc upregulation was triggered by hypomethylation of CpG dinucleotides and H3 acetylation of the promoter. Whether these mechanisms are active in patients with T2D is unknown.

The current study investigated whether p66^Shc is involved in persistent ROS generation and endothelial dysfunction in patients with T2D despite IGC. We show that targeting HbA_1c ≤7.0% did not revert diabetes-induced p66^Shc overexpression, oxidative stress, and endothelial dysfunction. Epigenetic changes of DNA/histone complexes may explain persistent upregulation of p66^Shc. Indeed, DNA hypomethylation and H3 acetylation were found on the p66^Shc promoter of patients with T2D, and IGC did not affect such detrimental chromatin modifications. That these epigenetic signatures are functionally linked is well established. Demethylation of CpG dinucleotides favors histone acetylation and chromatin accessibility to transcription factors (11). Persistent downregulation of chromatin-modifying enzymes methyltransferase DNMT3b and deacetylase SIRT1 may contribute to the epigenetic pattern observed on p66^Shc promoter of patients with T2D, even after IGC. In agreement with other in vitro studies (17,24,30), reprogramming of DNMT3b and SIRT1 blunts p66^Shc expression by resetting a condensed chromatin (25,31). Although DNMT3b and SIRT1 are well-established regulators of p66^Shc transcription (25,31), we cannot exclude that other chromatin modifiers may co-occur to modulate p66^Shc expression under hyperglycemic conditions. Future studies that use unbiased approaches may help with reaching a definite conclusion on this important aspect.

Of note, we found that restoration of HbA_1c target levels does not suppress the epigenetic changes of the p66^Shc gene. Although HbA_1c is a reliable marker of glycemic control, it may explain only <25% of the risk of developing diabetic complications (32). Indeed, HbA_1c does not correlate with glycemic variability when adjusted for mean blood glucose (33). Experimental evidence suggests that transient spikes of hyperglycemia may be considered an independent risk factor (19). In this regard, significant emphasis has been given to the relationship between postprandial hyperglycemia and cardiovascular complications (34). Glycemic variability can be quantified by MAGE and AUCpp (35). MAGE has been conceived to measure the mean of the difference between consecutive peaks and nadirs and to provide a reliable estimate of glycemic instability (21). On the other hand, AUCpp is a well-established marker that reflects meal-related hyperglycemic swings. Previous work has shown that postprandial glucose levels are significantly prolonged in patients with T2D compared with healthy control subjects and may trigger ROS generation, reduced nitric oxide bioavailability, and endothelial damage (21,36–39). Of note, postprandial hyperglycemia is an independent risk factor for micro- and macrovascular complications in patients with T2D (40–42). In this regard, the Study to Prevent NIDDM has shown that decreasing postprandial hyperglycemia is associated with a 49% relative risk reduction in the development of cardiovascular events (hazard ratio 0.51 [95% CI 0.28–0.95]; P = 0.03) in high-risk subjects with impaired glucose tolerance (43). Although a growing body of evidence supports the association between glycemic excursions and vascular damage, no previous studies have investigated the possible mechanisms that underlie this relation in patients with T2D. Moreover, the current study is the first to our knowledge to link glucose fluctuations with modifications of the epigenetic repertoire in humans. Previous work in human endothelial cells demonstrated that transient hyperglycemic spikes increase chromatin accessibility through epigenetic changes that favor a proatherosclerotic phenotype (19). We were prompted by this background to explore the possible interconnections among glycemic fluctuations, chromatin remodeling, oxidative stress, and endothelial dysfunction. We observed that epigenetic signatures of p66^Shc promoter are strongly associated with markers of glycemic instability (AUCpp and MAGE), regardless of relevant confounders such as age, sex, BMI, and diabetic treatment. By contrast, HbA_1c is unable to discriminate subjects with and without p66^Shc epigenetic remodeling. This observation suggests that a well-established marker of chronic sustained hyperglycemia may not detect gene-activating events elicited by glucose fluctuations. On the basis of these findings, tailoring glucose-lowering strategies only on the level of HbA_1c may leave patients with T2D exposed to a substantial burden of glycemic peaks and nadirs, which perpetuate the epigenetic changes responsible for dysregulation of vascular oxidative pathways. The use of glucose-lowering drugs that specifically suppress glycemic variability, namely α-glucosidase inhibitors, GLP-1 receptor agonists, and SGLT2 inhibitors, might contribute to erasing detrimental epigenetic modifications and restoring vascular homeostasis. The reduction of glucose fluctuations may have contributed to the beneficial cardiovascular effects of liraglutide, semaglutide, and empagliflozin in the LEADER, SUSTAIN-6, and EMPA-REG OUTCOME trials, respectively (44).

The current study has some limitations. The epigenetic changes of p66^Shc were assessed in peripheral blood monocytes from patients with T2D. However, an increasing body of evidence supports the concept that molecular changes in circulating mononuclear cells mirror early alterations in endothelial vasomotor function. In this regard, a strong correlation between oxidative stress in mononuclear cells and endothelium-dependent vasorelaxation in patients with T2D has been reported (45). These findings suggest that epigenetic changes observed in circulating cells may represent a reliable indicator of endothelial dysfunction and inflammation. Although the current analysis was conducted in a relatively small cohort and further studies are needed to confirm these results, we have unmasked a deleterious link between glucose fluctuations and chromatin remodeling and vascular dysfunction. These results provide insights into why targeting HbA_1c failed to improve cardiovascular outcomes in T2D and may set the stage for further investigations to exploit the impact of glycemic variability on the entire epigenetic landscape and its implications for CVD phenotypes (46). Although we could not directly prove a casual relationship between glycemic variability and epigenetic changes of p66^Shc promoter, previous experimental work demonstrated that glucose oscillations induce epigenetic signatures in human endothelial cells as well as in diabetic mice (19,47). Moreover, Quagliaro et al. (30) showed that glucose excursions compared with chronic sustained hyperglycemia induce the activation of protein kinase C, a master regulator of p66^Shc activity. Future clinical investigations specifically targeting glycemic variability will be invaluable to support the biological link between glucose fluctuations and epigenetic changes. Therefore, in the midst of a global diabetes epidemic, this work encourages efforts to assess and minimize glycemic variability as an HbA_1c-independent trigger of adverse chromatin alterations and may offer an attractive perspective to reducing the staggering cardiovascular burden of diabetes.

Funding. S.Co. was supported by Center for Gender Medicine, Karolinska Institute. F.P. is the recipient of a Sheikh Khalifa's Foundation Assistant Professorship at the Faculty of Medicine, University of Zurich. This study was supported by the Swiss Research Council Foundation (310030-135815 to T.F.L.) and the Swedish Research Council (VR, 2016-02706), Swedish Heart & Lung Foundation (20140360), Konung Gustaf:Vs och Drottning Victorias Frimurarestiftelse, European Foundation for the Study of Diabetes, Karolinska Institute, and Italian Ministry of Education PRIN (to F.C.).

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

Author Contributions. S.Co. and F.P. contributed to the study conception, data analysis and interpretation, and drafting of the manuscript. R.B., L.C., G.C., S.Ch., L.T., and G.R. contributed to the data analysis and interpretation. D.P., G.A.L., M.V., and T.F.L. contributed to the revision of the manuscript for important intellectual content. F.C. contributed to the study design and drafting and final approval of the manuscript. F.C. is the guarantor of this work and, as such, had 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.