Restriction of Advanced Glycation End Products Improves Insulin Resistance in Human Type 2 Diabetes: Potential role of AGER1 and SIRT1

The study enrolled 18 type 2 diabetic patients (14 women, 4 men; average age, 61 ± 4 years), without renal disease or overt cardiovascular disease, and a usual diet rich in AGE (dietary AGE intake ∼>20 AGE equivalents [Eq]/day; Table 1). All study subjects continued to receive standard medical care: 20% were treated only with diet, and 80% received oral antidiabetic medications, including 15 with metformin and 1 each with pioglitazone, glipizide, and glimepiride; 60% were receiving statins and ACE inhibitors or angiotensin-receptor blockers, and 40% were taking aspirin (81 mg daily).

They were randomly assigned to an AGE-restricted diet (n = 12) or their usual diet (n = 6) for 4 months. The AGE-restricted group received instructions on how to modify cooking time and temperature but not the quantity or nutrient composition of food. They were advised to boil, poach, stew, or steam food and to avoid frying, baking, or grilling, methods, which limit dietary AGE intake by ∼40–50% (9,19,20). The control group comprised 18 age-matched (age 67 ± 1.4 years), healthy volunteers selected from a larger well characterized cohort that was randomized to the same AGE-restricted diet for 4 months (9).

A dietitian monitored participants by telephone 1–2 times per week and at monthly clinic visits. The study diets were designed to maintain daily requirements of calories and nutrients in both arms but to restrict AGE intake only on the AGE-restricted arm. Nutritional intake was closely monitored in both groups.

The study and consent forms were approved by the Mount Sinai School of Medicine Institutional Review Board. All participants provided informed consent.

Participants received a physical examination and provided a fasting blood sample and a 24-h urine collection at baseline and at 4 months. Routine blood and urine tests were performed by the hospital clinical laboratory.

Daily dietary AGE content, from 3-day food records that emphasized cooking methods, was estimated from a database that listed AGE values of ∼560 foods and was expressed as AGE equivalents/day (AGE Eq = 1,000 kilounits) (20). The 3-day food record is based on established guidelines developed to allow estimation of food and beverage consumption at home and away from home. The dietitian reviewed the record with the subject, probing for details on portion sizes and preparation methods, to ensure accuracy and compliance. Nutrient calculations were estimated by using the Food Processor 10.1 nutrient software program (ESHA Research, Salem, OR).

Derivatives of N^ε-carboxymethyl-lysine (CML) and methylglyoxal (MG) in serum were quantified by established enzyme-linked immunosorbent assays (ELISAs) using monoclonal antibodies (4G9 and MG3D11) (9,21), validated against synthetic standards, CML-BSA (23 modified lysines/mol) and MG-BSA (20 MG-modified arginines/mol), based on high-performance liquid chromatography and gas chromatography-mass spectroscopy (21). Insulin was measured by an ELISA kit (ALPCO Diagnostics, Salem, NH). IR was estimated by the homeostasis model assessment (HOMA) index as [FI × (fasting glucose/22.5)], where FI is insulin in microunits per milliliterand fasting glucose is in millimoles per liter(9). ELISA kits were used for measuring plasma leptin, adiponectin (Millipore, Billerica, MA), and PMNC tumor necrosis factor (TNF)-α (Biosource International, Camarillo, CA).

PMNCs were separated from fasting, EDTA-anticoagulated blood by Ficoll-Hypaque Plus gradient (American Biosciences, Uppsala, Sweden) and used to isolate mRNA and protein (9,19). Total RNA was extracted by TRIzol (Invitrogen-Molecular Probes, Inc., Eugene, OR). The extracted RNA had an optical density of 280/260 ratio of between 1.8 and 2.0. Total RNA was reverse-transcribed using Superscript III RT (Invitrogen, Carlsbad, CA).

THP-1 human monocytes (1 × 10⁶/mL) (American Type Culture Collection, Manassas, VA) were cultured in RPMI-1640 medium (1% FBS). THP-1 cells were transfected with AGER1, SIRT1 or short-hairpin RNA AGER1 (0.5 µg of DNA) by using the human monocyte cell line nucleofector kit (Amaxa Biosystems, Cologne, Germany). Quiescent cells were exposed to CML-BSA, MG-BSA, or BSA, with or without the following inhibitors: sirtinol (10 µmol/L; Calbiochem, La Jolla, CA); N-acetylcysteine ([NAC], 5 mmol/L; Sigma-Aldrich, St. Louis, MO); apocynin (300 µmol/L; Sigma-Aldrich); and Mn superoxide dismutase (1,000 units/mL; Sigma-Aldrich).

Synthetic preparations of CML-BSA (23 CML-Lys/mol), MG-BSA (22 MG-Arg/mol) (21), and BSA (Fraction V; Sigma-Aldrich) were rendered LPS-negative before use by a Detoxigel column (Pierce, Rockford, IL), based on Limulus assay (BioWhittaker, Walkerville, MD).

AGER1, receptor for AGE (RAGE), the 66-kDa protein from the src homology and collagen homology domain (p66^shc), SIRT1, and NADPH oxidase 1 (NOX) p47^phox mRNA levels were assessed by quantitative SYBR Green real-time PCR (9). Primer sequences were as follows:

• AGER1, forward primer 5′-CTGGGGCTCTTCATCTTCAG-3′, reverse primer 5′-GTTGCATCTCCCACAGAGGT-3′;

• RAGE, forward primer 5′-AGGAGCGTGCAGAACTGAAT-3′, reverse primer 5′-TTGGCAAGGTGGGGTTATAC-3′;

• p66^shc, forward primer 5′-AGGAAGGGCAGCTGATGAT-3′, reverse primer 5′-GCGTGGGCTTATTGACAAAG-3′;

• NOXp47^phox, forward primer 5′-ACAGCGTCCGTTTTCTGC-3′, reverse primer 5′-AGCGGTTCAGGATGAGGTC-3′; and

• SIRT1, forward primer 5′-CGGAAACAATACCTCCACCT-3′, reverse primer 5′-CACCCCAGCTCCAGTTAGAA-3′.

Emission from each sample was recorded during thermal cycling, and the raw fluorescence data were processed using Sequence Detection System software (Applied Biosystems, Carlsbad, CA) to produce threshold cycle (Ct) values for each sample. β-Actin and glyceraldehyde-3-phosphate dehydrogenase (GAPDH) housekeeping genes were used for internal normalization. The transcript copy number of target genes was determined based on their Ct values (9).

Cell proteins were separated on 8% SDS-PAGE gels, transferred onto nitrocellulose (NT) membranes for probing with primary and secondary antibodies, and visualized by an enhanced chemiluminescence system (Roche, Mannheim, Germany) (7,18). For immunoprecipitation, the lysate (300 µg protein) was incubated overnight at 4°C with the appropriate antibody, followed by 60 μL protein A/G plus agarose beads (Santa Cruz Biotechnology, Santa Cruz, CA) for 2 h. Bound immune complexes in radioimmunoprecipitation assay lysis buffer were used for immunoblotting after SDS-PAGE and NT transfer (9).

Cells (PMNCs, THP-1) were disrupted in lysis buffer (Sigma-Aldrich, C2978), and 100 µg protein was immunoprecipitated with anti-NF-κB-p65 (Santa Cruz Biotechnology) at 4°C overnight. A total of 60 µL protein agarose (A/G) beads were added, and immunoprecipitates were immunoblotted for acetyl-lysine residues (4). For densitometric analysis, acetyl-p65 was normalized for total p65.

NAD and NADH were measured using an Amplite Fluorimetric NAD/NADH assay kit (AAT Bioquest, Inc., Sunnyvale, CA). Cells (wild-type or AGER1-transfected) were lysed after stimulation with MG-BSA (60 µg/mL) for 72 h in the presence or absence of antioxidants NAC (5 mmol/L) and apocynin (300 µmol/L).

Data in the tables and figures are presented as means ± SEM. The Kolmogorov-Smirnov goodness-of-fit test was used to test for normal distribution. Variables not normally distributed were logarithmically converted for analyses. Differences of means between groups were analyzed by the Student t test or ANOVA, followed by the Bonferroni correction for multiple comparisons, depending on the number of groups. Correlation analyses were also examined by the Pearson correlation coefficient.

Significance of changes during the interventional study was assessed by comparing 1) change of means between baseline and end of study within each group by paired t test, 2) percentage of change from baseline to the end of the study between the AGE-restricted and the regular AGE diet groups by the Mann-Whitney test, and 3) differences between the means of both groups at the end of the study by unpaired t test. Significant differences were defined as a value of P < 0.05 and are based on two-sided tests. Data analysis was performed in consultation with a statistician, using SPSS 17.0 software (SPSS, Chicago, IL).

The type 2 diabetic patients, age- and sex-matched with nondiabetic control subjects, had similar levels of creatinine clearance (Table 1). Compared with healthy subjects, type 2 diabetic participants had significantly higher fasting blood glucose, plasma insulin, HOMA, BMI, waist circumference, serum CML (sCML), serum MG (sMG), and leptin but lower adiponectin. They also had significantly higher levels of serum AGEs and plasma 8-isoprostanes (Table 1).

AGER1 and SIRT1 mRNA levels in PMNCs from diabetic subjects were significantly lower than in healthy subjects (Table 1). Concentrations of pro-OS p66^shc and NAPDH oxidase p47^phox mRNA and TNF-α protein, a proinflammatory cytokine, were elevated, as were intracellular AGEs (iCML; Table 1). These data were consistent with high baseline OS.

AGE restriction (by 50%), without altering nutrient intake, led to markedly lower levels of sCML and sMG, as well as 8-isoprostanes in type 2 diabetic subjects (Table 2, Fig. 1A). Plasma insulin, HOMA, and leptin levels were also decreased by ∼30% below baseline by AGE restriction (Table 2, Fig. 1A). However, adiponectin levels were doubled after AGE restriction, resulting in a markedly lower leptin/adiponectin ratio (Fig. 1A, inset).

Type 2 diabetic subjects on the regular diet showed no significant changes, except for higher sMG, iCML, leptin, and TNF-α values (Table 2), consistent with sustained high AGE intake and AGE accumulation. Neither the diabetic patients nor the healthy subjects had changes in weight (96.0 ± 7 to 97.3 ± 8 kg vs. 83.6 ± 6 to 82.0 ± 6 kg, respectively), BMI, plasma lipids, blood glucose, or HbA_1c after 4 months. Healthy subjects on the AGE-restricted diet had no significant changes, with the exception of decreased sAGEs and 8-isoprostanes (Table 2).

The suppressed basal mRNA levels of AGER1 and SIRT1 in type 2 diabetic patients were increased by ∼50%, almost to the normal range in the AGE-restricted cohort (Fig. 1A; Table 2) (9). The changes in mRNA of AGER1 and SIRT1 correlated with changes in protein expression (Fig. 1B). Intracellular AGE levels in the AGE-restricted group were decreased, iMG (P = 0.002) and iCML (P = 0.148; Table 2), consistent with increased AGER1 expression.

There was also a reduction in NF-κB acetyl-p65 (Fig. 1C), associated with trends of lower RAGE mRNA and TNF-α protein in PMNCs from AGE-restricted type 2 diabetic subjects (Table 2). In contrast, in diabetic subjects who ate the regular diet, PMNC SIRT1 remained depressed (Fig. 1B and C; Table 2), whereas acetyl-p65, TNF-α, and RAGE remained elevated in diabetic subjects, consistent with higher circulating oxidants, AGEs, and 8-isoprostane (Table 2).

Because SIRT1 (4,23) and AGER1 (6–8) are suppressed in patients with diabetes (high OS), monocyte-like THP-1 cells were used as inflammatory cells to investigate the observations in PMNCs. AGER1 and SIRT1 protein expression levels were coordinately suppressed in THP-1 cells after chronic exposure to CML or MG (but not unmodified BSA; Fig. 2A). In addition, AGER1 and SIRT1 expression was enhanced in AGER1-overexpressing (AGER1⁺) cells treated with MG but reduced in AGER1-silenced THP-1 cells (Fig. 2B and C). After a transient induction, a marked time–dependent reduction of the NAD⁺/NADH ratio was noted in association with suppressed SIRT1 levels, and this effect was prevented by antioxidants (NAC, apocynin; Fig. 2D) and by AGER1 overexpression (AGER1⁺; Fig. 2E). In addition, the AGE-induced NF-κB p65 hyperacetylation seen in wild-type cells was blocked in AGER1⁺ cells (Fig. 2F), consistent with preserved SIRT1 function when AGER1 is intact. These data were in line with the PMNC findings from type 2 diabetic patients.

IR remains an unresolved scientific and therapeutic problem. Increased OS and impaired anti-OS defenses are important factors in the development and persistence of IR (2–4). The current study shows that AGE restriction can improve IR in type 2 diabetes. Plasma insulin, markers of inflammation, and OS were substantially improved in diabetic subjects by an intervention that restricts oxidant (AGE) load without altering energy balance. These findings are consistent with evidence showing that excessive food AGEs promote inflammation and OS in humans (9,19) and IR in mice (15,16). That plasma insulin was nearly normalized after the AGE restriction indicates that tissue glucose uptake was improved. Therefore, a reduced demand on β-cells for insulin release appears to be directly related to the AGE restriction. While the findings must be confirmed in larger cohorts, AGE restriction may have therapeutic advantages in IR.

Although endogenous glucose or lipids contribute to elevated AGEs (22), exogenously AGE-modified nutrients are an important source in both human subjects and animals (9,14,16,19,20). The significant reduction in serum AGEs and lipid peroxides (8-isoprostanes) after AGE restriction in the current study confirms the hypothesis that elevated AGEs in type 2 diabetes result from increased dietary AGE intake, independently from glycemic control (19).

AGE restriction normalized AGER1, SIRT1, and SIRT1-dependent NF-κB p65 deacetylation levels, and there was a concomitant attenuation of proinflammatory mediators (i.e., RAGE and TNF-α) (9). The decreased basal AGER1 and SIRT1 (mRNA and protein) levels in type 2 diabetic PMNCs suggests a link between these two mechanisms. Since plasma insulin and leptin levels fell, whereas adiponectin levels increased with AGE restriction, the changes in AGER1 expression reflected adjustments in both the inflammatory and the metabolic states, effects previously attributed to SIRT1 (2–4,23). Thus, high AGE and low AGER1 levels in type 2 diabetes may partly account for the persistence of IR in these patients.

Given that downregulation of AGER1 (7–9) and SIRT1 is linked to high OS conditions (4,5), the current findings also indicate that normal SIRT1 function may depend on the ability of intact AGER1 to control AGEs and OS. Thus, the loss of AGER1 function may be one mechanism leading to SIRT1 downregulation in type 2 diabetes. This postulate is supported in the current study by the high basal levels of intracellular AGEs and low AGER1 levels in type 2 diabetic PMNC. The fact that these levels are nearly normalized after AGE restriction lends further support to this hypothesis. These data also suggest that low steady-state levels of intracellular AGEs may be crucial for maintaining AGER1 intact (7–9,16,24).

Because this was the first time that AGEs and AGER1 were linked to the deacetylases and to SIRT1, we probed these relationships in inflammatory monocyte–like THP-1 cells. After prolonged exposure to AGEs, AGER1 and SIRT1 were both suppressed in these cells or became unresponsive. AGEs acted via an OS-dependent decrease in NAD⁺, explaining the eventual downregulation of SIRT1, and consequently, the hyperacetylation of NF-κB p65 observed in diabetic PMNCs (2,4). Since the AGE-induced effects on SIRT1 and NF-κB were blocked in THP-1 cells by genetic modulation of AGER1 expression, AGER1 is likely to be involved in the regulation of inflammatory responses controlled by SIRT1. The data also suggest that controlling the pro-oxidant effects of AGEs could help preserve normal AGER1 and SIRT1 functions.

From a clinical perspective, AGE restriction markedly increased adiponectin levels and reduced the initially high leptin/adiponectin ratio, a marker of IR, indicating that AGEs also affect insulin-sensitive tissues (2–5). Adiponectin improves insulin sensitivity and is controlled by SIRT1 (25); therefore, the doubling in adiponectin levels after AGE-restriction provides another insight into the improved metabolic state in type 2 diabetic patients.

These data show that chronic exposure to excessive exogenous (food) oxidants fosters an impairment in both native antioxidant defenses as well as insulin action and that reducing oral AGEs can ameliorate these defects. This approach does not restrict energy or caloric intake. Since the effects of AGE restriction were additive to standard medical therapy, the intervention could be a valuable addition to the current management of patients with type 2 diabetes. Larger trials are warranted to validate the findings and establish the long-term effects of this intervention on diabetes.

This work was supported by MERIT grants AG-23188 and AG-09453 (to H.V.) and from the National Institutes of Health, National Institute of Research Resources (Grant M01-RR-00071) to the General Clinical Research Center at Mount Sinai School of Medicine for clinical and statistical support.

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

J.U. researched the data and reviewed and edited the manuscript. W.C. researched the data and wrote, reviewed, and edited the manuscript. M.R. researched the data. S.G. researched the data and contributed to the discussion. R.P., X.C., and L.Z. researched the data. G.E.S. reviewed and edited the manuscript. H.V. designed the study, obtained support, researched the data, contributed to the discussion, and wrote, reviewed, and edited the manuscript.

Part of this study was submitted in abstract form to the 71st Scientific Sessions of the American Diabetes Association, 24–28 June 2011, San Diego, California.