Mitochondrial Epigenetic Changes Link to Increased Diabetes Risk and Early-Stage Prediabetes Indicator

Type 2 diabetes (T2D) is characterized by mitochondrial derangement and oxidative stress. With no known cure for T2D, it is critical to identify mitochondrial biomarkers for early diagnosis of prediabetes and disease prevention. Here we examined 87 participants on the diagnosis power of fasting glucose (FG) and hemoglobin A1c levels and investigated their interactions with mitochondrial DNA methylation. FG and A1c led to discordant diagnostic results irrespective of increased body mass index (BMI), underscoring the need of new biomarkers for prediabetes diagnosis. Mitochondrial DNA methylation levels were not correlated with late-stage (impaired FG or A1c) but significantly with early-stage (impaired insulin sensitivity) events. Quartiles of BMI suggested that mitochondrial DNA methylation increased drastically from Q1 (20 < BMI < 24.9, lean) to Q2 (30 < BMI < 34.9, obese), but marginally from Q2 to Q3 (35 < BMI < 39.9, severely obese) and from Q3 to Q4 (BMI > 40, morbidly obese). A significant change was also observed from Q1 to Q2 in HOMA insulin sensitivity but not in A1c or FG. Thus, mitochondrial epigenetic changes link to increased diabetes risk and the indicator of early-stage prediabetes. Further larger-scale studies to examine the potential of mitochondrial epigenetic marker in prediabetes diagnosis will be of critical importance for T2D prevention.


Introduction
Lifestyle contributes significantly to the epidemic of type 2 diabetes (T2D) [1][2][3][4]. The interactions between lifestyle factors and mitochondria (the primary metabolic platform) have been implicated in the pathogenesis of diabetes and obesity [5]. In particular, mitochondria undergo aberrant alteration and cause oxidative stress in diabetic subjects, and redox active compounds have shown promising potential to treat T2D [5][6][7][8]. Nevertheless, no known cure has been developed for diabetes to date [9,10], and early diagnosis of prediabetes for intervention has been an important strategy to prevent T2D [3]. To this end, biomarkers that can signify progression or early stage of prediabetes will be of clinical significance.
The risk of T2D increases with weight gain. For instance, the lifetime risk of developing diabetes at an age of 18 years increases by 9.9%, 37.2%, and 50.5% in overweight (25 < BMI < 30), obese (30 < BMI < 35), and severely obese (>35) males, respectively, compared with normal weight individuals; for female subjects, it increases by 18.3%, 37.5%, and 57.3%, respectively [11]. As such, the American Diabetes Association (ADA) recommends BMI as a primary index to screen for high-risk population and recommend of further testing to 2 Oxidative Medicine and Cellular Longevity assess risk of future diabetes development in asymptomatic people [12]. Following an initial weight screening, individuals having a fasting glucose (FG) ranging from 100 mg/dL (5.6 mmol/L) to 125 mg/dL (6.9 mmol/L) or an A1c of 5.7-6.4% are diagnosed as prediabetic [12]. Of note, the existing discordance between FG and A1c may lead to conflicting results, underscoring the utmost importance of identifying new markers that can consistently identify prediabetic individuals as early as possible for T2D prevention [13][14][15].
The effectiveness of diabetes prevention via lifestyle modification is strongly associated with improved insulin sensitivity [16] and epigenetic adaptation, including DNA methylation on the nuclear genome and mitochondrial genome [5,17]. Recently, we identified an insulin signalingepigenetic-genetic axis in the regulation of mitochondria, the major cellular metabolic platform that responds robustly to interventions [6,[18][19][20][21][22][23]. However, it is largely unknown whether mitochondrial epigenetic signature reflects prediabetes progression and how it interacts with the current diagnostic indices (A1c and FG). In the present work, we studied the interactions of mitochondrial DNA methylation with FG, A1c, and insulin sensitivity, the parameters that change stage-dependently during prediabetes progression [24][25][26]. DNA methylation in mitochondrial NADH dehydrogenase 6 (ND6) and displacement loop (D-loop) region was significantly correlated with changes in insulin sensitivity (the event starting in an earlier stage [24][25][26]) but had marginal interaction with A1c and FG (the later-stage markers in prediabetes progression [24][25][26]). Upon quartilizing BMI (the primary risk index for prediabetics screening [11,12]), we found significant changes in mitochondrial epigenetics during the progression from Q1 (20 < BMI < 24.9, lean) to Q2 (30 < BMI < 34.9, obese), but not from Q2 to Q3 (35 < BMI < 39.9, severely obese) or Q3 to Q4 (BMI > 40, morbidly obese); by contrast, FG showed significant changes only from Q3 to Q4, and the levels of A1c were unchanged in all cases. Discordance was observed when FG or A1c was used to diagnose prediabetes, presumably because they indicate late but different stages of prediabetes. Together, our data suggest that mitochondrial epigenetic changes link to diabetes risk and signify early-stage prediabetes, targeting which may have the potential to achieve early diagnosis of prediabetes for T2D intervention.

Ethics Statement.
A written informed consent was obtained from each participant. All procedures were conducted in accordance with NIH Guidelines, and the research protocol was approved by Institutional Review Boards (IRB) at Carillion Clinic and at Virginia Tech.

Statistical Analysis.
Pearson's correlation and regression analysis were applied to examine the relationships among obesity-related risk factor, metabolic indexes, and mitochondrial DNA methylation levels. In some cases, logarithmtransformed data were used for skewed variables (e.g., HOMA-IR). The data were expressed as the mean ± SE unless otherwise specified, and analysis of variance (ANOVA) was conducted to determine values; < 0.05 was considered statistically significant.  (Table 1). In line with the notion that an increase in fasting insulin (FI) indicates impairment of insulin signaling [35,36], we found that the homeostasis model assessment-estimated insulin resistance (HOMA-IR) index increased from 2.1 in the lean group to 6.0 in the overweight and obese group ( < 0.05) ( Table 1). In addition, lipid profile was dysregulated in the overweight and obese individuals, showing significant elevation in LDL, VLDL, total cholesterol, and triglyceride (Table 1). However, A1c showed no difference between the two groups (5.5 versus 5.7, > 0.05). These data suggest that BMI can reflect the changes in glucose and lipid metabolism, as well as the impairment in insulin signaling, supporting the notion that BMI represents an important risk index for both prediabetes and T2D development [11,37,38].

Insulin Resistance Reflected Diabetes Risk Better than
A1c and FG. Prediabetes is characterized by impaired fasting glucose and insulin sensitivity, which progress differently from A1c before the onset of T2D (Supplemental Figure 1 in Supplementary Material available online at http://dx.doi.org/10.1155/2016/5290638) [24][25][26]. Specifically, drastic changes were observed in insulin sensitivity 5 years prior to the diagnosis of T2D, which is 3 years earlier than FG and 4 years earlier than A1c (Supplemental Figure 1) [24][25][26] Figure 2). These data suggest that insulin resistance indices can better reflect diabetes risk. Given that impaired insulin sensitivity is observed prior to impaired FG and A1c in prediabetes [24][25][26], use of insulin resistance index may be superior to A1c or FG for early detection of prediabetes.

Mitochondrial Epigenetic Changes Were Associated with
Insulin Resistance. The onset of T2D in prediabetic individuals can be effectively delayed or prevented by lifestyle intervention, in response to which mitochondria undergo epigenetic adaptation [1-3, 5, 17, 18, 39]. To examine how mitochondrial epigenetics interacts with A1c, FG, and insulin resistance, we collected additional blood samples from a subset (40 out of the 87, including 8 lean/healthy, and 32 obese) of participants and analyzed DNA methylation in mitochondrial ND6 and D-loop, the regions critical for mitochondrial function and DNA replication [18,32]. The 40 individuals represented well the 87 participants, as demonstrated by the relations of BMI with A1c, FG, FI, and HOMA-IR (Supplemental Figure 3). More importantly, ND6 methylation level showed significant correlation with insulin resistance indices ( = 0.0054 for HOMA-IR and 0.0110 for FI; Figures 3(a) and 3(b)), but it had marginal interaction with FG ( = 0.1581; Figure 3(c)) and A1c ( = 0.3538; Figure 3(d)). Further regression analyses suggested that ND6 methylation had no significant correlation with lipid profile or age (Supplemental Figure 4, A-D, or data not shown), consistent with previous studies of global DNA methylation [40,41]. This phenotype was recapitulated by D-loop methylation, which was significantly correlated with insulin resistance (HOMA-IR, = 0.0337; Figure 3  These data suggest that mitochondrial epigenetic signature may represent an early-stage marker of prediabetes and its alteration parallels with the development of insulin resistance.
In comparison with Q1, the other three quartiles all showed significant increase in ND6 and D-loop methylation (Figures 4(c) and 4(d)). However, no further significant changes in DNA methylation were observed from Q2 to Q3 or to Q4. By contrast, insulin resistance indices (HOMA-IR and FI) underwent significant elevation from Q1 to Q2 and additionally from Q3 to Q4 (Figures 4(e) and 4(f)). The continuous changes in insulin resistance indices during high-risk stages (Q3 and Q4) are consistent with the trajectory showing that HOMA insulin sensitivity decreased steeply until Oxidative Medicine and Cellular Longevity the diagnosis of T2D (Supplemental Figure 1) [24,25]. Of note, A1c showed marginal changes in all stages (from Q1 to Q4), and a significant increase in FG was observed only during advanced progression (Q3 and Q4) stages (Supplemental Figure 5). These results support the notion that FG is a late-stage marker of prediabetes progression, and change of A1c is an even later-stage event (Supplemental Figure 1) [24][25][26]. Importantly, we found that mitochondrial epigenetic changes were most sensitive when prediabetes progressed from low-risk (Q1) to high-risk (Q2) stages but less responsive to further progression in higher risk (Q3 and Q4) stages. Therefore, targeting mitochondrial epigenetic markers has the potential to identify early-stage prediabetes for as-earlyas-possible lifestyle intervention.

Oxidative Medicine and Cellular Longevity
These studies have elegantly unraveled an epigenetic link to metabolic diseases, but the relationship between global DNA methylation and environmental and metabolic markers is more complex than expected and warrants further investigation.
Mitochondria underpin metabolic homeostasis [48][49][50]. We and others have shown that mitochondria undergo genetic, epigenetic, and functional changes in subjects with obesity, insulin resistance, or T2D [5,18,[50][51][52]. In the present study, we investigated the relationships between mitochondrial DNA methylation and stage-dependent metabolic markers and diabetes risk index (BMI). It is known that the progression of prediabetes to T2D shows stagedependent features: (1) HOMA insulin sensitivity decreases steeply 5 years prior to T2D occurrence; (2) FG undergoes abrupt increase 2-3 years before T2D onset; (3) significant elevation of A1c is not observed until 1 year before the diagnosis of T2D (Supplemental Figure 1) [24][25][26]. The different trajectories rank HOMA insulin sensitivity (or insulin resistance, assessed by HOMA-IR) higher than FG and A1c as a potential marker for identifying earlier stage of prediabetes. Indeed, HOMA-IR and FI showed significant correlation with the risk index BMI ( Figure 2). Moreover, insulin resistance was significantly associated with the changes in mitochondrial ND6 and D-loop DNA methylation, while the latter showed no correlation with A1c or FG (Figure 3). Based on quartiles of BMI, mitochondrial epigenetic traits underwent significant changes from healthy (20 < BMI < 24.9, Q1) to obese (30 < BMI < 34.9, Q2) status, which corresponds to the development of insulin resistance (Figure 4, Supplemental Figure 1). However, no significant change was observed in FG or A1c during the Q1 → Q2 transition (Supplemental Figure 5). These findings suggest that mitochondrial epigenetic changes represent an early-stage event in prediabetes progression and is associated with the development of insulin resistance.
The Prevent Diabetes STAT campaign calls for early identification and intervention to prevent the future development of T2D [53]. Thus, our ability to consistently identify prediabetic individuals as early as possible in the disease progression stages will be of utmost importance to address these challenges. In this study we examined the diagnosing power of FG and A1c by following the standards recommended by ADA [12]. FG indicated that 27.6% of individuals were prediabetic, 1.1% diabetic, and 71.3% healthy or normal. By A1c, however, 49.4% of the tested participants were considered prediabetic and 3.5% diabetic, leading to a diagnostic rate 1.8-fold and 3.2-fold higher than FG, respectively ( Figure 1). In addition, neither FG nor A1c showed significant correlation with BMI, the primary risk index used for screening (Figure 2), which may account for the wide variation in diagnosis thus highlighting the urgent need for additional diagnosis markers [14,15]. More importantly, mitochondrial epigenetic signature was significantly correlated with BMI and showed robust changes in early stage of prediabetes progression that corresponds to the development of insulin resistance (Figures 3 and 4). These results suggest that mitochondrial epigenetic marker may have greater potential than A1c or FG to achieve early diagnosis of prediabetes. Future studies seeking to determine the value of mitochondrial epigenetic signatures in accurately diagnosing prediabetes and assessing intervention effectiveness will be of critical importance to fulfill the mission of Prevent Diabetes STAT.
Taken together, effective prevention of T2D requires as-early-as-possible diagnosis of prediabetes for lifestyle intervention. We showed that the current diagnosis indices A1c and FG led to significant discordance, underscoring the urgent need of additional markers to achieve reliable diagnosis. Importantly, we found that the mitochondrial epigenetics signature (DNA methylation in ND6 and D-loop) characteristically signified the early stage of prediabetes when insulin resistance develops. By contrast, advanced stages of prediabetes progression lead to no significant changes in mitochondrial epigenetic signature. Thus, mitochondrial epigenetic changes represent an early-stage event in prediabetes progression, and further studies to systematically examining mitochondrial epigenetic signature in prediabetes diagnosis, particularly in larger-scale cohorts, will be of critical importance.