Obesity and ectopic fat disposition are risk factors for metabolic disease. Recent data indicate that IGFBP2 expression in liver is epigenetically inhibited during hepatic steatosis. The aim of this study was to investigate if epigenetic de-regulation of hepatic Igfbp2 occurs already early in life and is associated with increased risk for diet-induced obesity (DIO) during adolescence. Male C57BL/6J mice received a high-fat diet. After 3 weeks on this diet (age of 6 weeks), DIO-susceptible (responder, Resp) and DIO-resistant (non-responder, nResp) mice were identified by early weight gain. At the age of 6 weeks, Resp mice exhibited elevated blood glucose (p < 0.05), plasma insulin (p < 0.01), HOMA-IR and leptin/adiponectin ratio, whereas liver triglycerides were identical but significantly increased (p < 0.01) in Resp mice at 20 weeks of age. Igfbp2 expression was reduced in young Resp compared with nResp mice (p < 0.01), an effect that correlated with elevated DNA methylation of intronic CpG2605 (p < 0.01). The epigenetic inhibition of Igfbp2 was stable over time and preceded DIO and hepatosteatosis in adult mice. In vitro studies demonstrated that selective methylation of CpG2605 significantly reduced reporter activity by ∼85%, indicating that Igfbp2 expression is modulated by methylation. In human whole blood cells, methylation of IGFBP2 at the homologous CpG site was increased in obese men with impaired glucose tolerance. In conclusion, our data show that increased methylation of hepatic Igfbp2 during infancy predicts the development of fatty liver later in life and is linked to deterioration of glucose metabolism.

Introduction

Obesity has become a major health threat, affecting more than half a billion people worldwide in 2014 (1). In particular, obesity increases the risk for insulin resistance, hypertension, coronary heart disease and type 2 diabetes (T2D). The risk of obesity is largely influenced by the family background, indicated by heritability rates of up to 40% for body mass index (BMI) (2). However, genetic variants identified so far only explain <5% of the observed heritability, a phenomenon referred to as “missing heritability” (3,4).

Epigenetics is suggested to explain parts of the missing heritability. Epigenetic alterations (e.g. DNA methylation at CpG dinucleotides or histone modifications) modify gene expression without altering genomic sequences and are therefore missed in conventional genome-wide association studies (5,6). Intriguingly, even in inbred mouse strains like C57BL/6J, individual mice largely differ in their response to diet-induced obesity (DIO), rendering these strains a suitable model to unravel non-genetic mechanisms in body weight regulation (7). In this context, Koza et al. (7) identified three genes (Sfrp5, Mest and Bmp3) in adipose tissue, which are differentially expressed in DIO-responders already at 7 weeks of age and associated with increased susceptibility to DIO.

In addition to adipose tissue, the liver plays a major role in the regulation of energy metabolism and hepatic insulin sensitivity, and consequently in the pathogenesis of obesity and T2D (8,9). Accordingly, hepatic genes involved in glucose and lipid metabolism have been shown to be altered in mice prone to DIO, including insulin-like growth factor 2 (Igf2) (10). The IGF-axis appears to play a crucial role in the regulation of body weight early in life. IGF-2 is involved in embryonic growth, and altered methylation of the human IGF2 gene is associated with childhood obesity (11–13). This is of particular interest, since plasma IGFBP-2 levels have been found to be decreased in obese males and females (14,15). Furthermore, reduced circulating IGFBP-2 levels in obese subjects are associated with increased bioavailability of IGF-1, and hyperinsulinemia (14). Furthermore, Igfbp2 transgenic mice show reduced body weight gain on a standard diet and are protected against the development of a DIO and insulin resistance (16,17). Recently, Ahrens et al. (18) reported that hepatic expression of IGFBP2, a modulator of IGF-1/2 bioavailability, is epigenetically inhibited in patients with different grades of hepatic lipid accumulation.

In this study, we investigated whether the variance in DIO and hepatic steatosis of genetically identical C57BL/6J mice is reflected by epigenetic modulation of hepatic Igfbp2 early in life. The establishment of weight prediction parameters allowed us to isolate liver tissues before the onset of fatty liver, and to study differences between mice prone to DIO and mice resistant against it. We found Igfbp2 to be hypermethylated and transcriptionally repressed in young mice prone to DIO, already before the onset of hepatic steatosis. These transcriptional and epigenetic differences of Igfbp2 that are present early in life and linked to obesity and insulin resistance in adolescence underline the significance of IGFBP-2 for body weight development and fatty liver. Interestingly, differential methylation occurs at the CpG site homologous to the human site in the IGFBP2 gene published by Ahrens et al. (18). In whole blood cells of human subjects exhibiting an impaired glucose tolerance the IGFBP2 methylation was significantly increased.

Results

Prediction of DIO by early weight parameters

In order to investigate early modulators of Igfbp2 expression, it was necessary to discriminate between mice prone and resistant to DIO at the earliest time point possible. Therefore, we established early predictors of obesity in a larger cohort of 324 male C57BL/6J mice fed a high-fat diet (HFD, 60 kcal% from fat) for 20 weeks. In this cohort, body weights varied in a range of 27.2–52.7 g (

, Fig. S1). DIO responders (Resp) were defined as individuals that reached body weights >38 g, while non-responders (nResp) reached <38 g at 20 weeks of age. Analysis of the individual weight courses indicated that weight gain during the first 3 weeks after weening (life week 3–6) in combination with the absolute body weight at week 6 is the strongest predictor for DIO susceptibility. In detail, nResp mice were characterized by an early weight gain of <0.9 g/day plus body weights <22 g in week 6, resulting in a probability for correct prediction of p = 0.81. Resp mice were characterized by an early weight gain of >0.7 g/day plus body weights >24 g in week 6 (p = 0.80). Next, the quality of prediction was confirmed in a second cohort of mice for which the prediction parameters were applied in week 6 (Fig. 1A). Low-fat diet (LFD)-fed mice served as a lean control. The early difference in body weight between Resp and nResp mice was reflected by both lean and fat mass, and increased over time (Fig. 1B, Table 1). Early body weight differences between both groups were not because of differences in body length, or food and water intake that were detected in single caged mice (Table 1).
Figure 1.

Impaired insulin sensitivity precedes liver fat accumulation. (A) Body weight development of DIO-responders (Resp, n = 9) and DIO-non-responders (nResp, n = 11) upon identification in week 6. Mice receiving an LFD served as control group. Body composition (B), fasted blood glucose (C), plasma insulin (D) and leptin (E) levels. (F) HOMA-IR calculated from fasted blood glucose and insulin levels. (G) LAR. (H) Liver triglyceride levels. All data represented as mean ± SEM of n = 10–15 animals per group. Differences between DIO-responder (Resp, black circle) and DIO-non-responder (nResp, white circle) mice were calculated by Student’s t-test. *p < 0.05, **p < 0.01, ***p < 0.001.

Figure 1.

Impaired insulin sensitivity precedes liver fat accumulation. (A) Body weight development of DIO-responders (Resp, n = 9) and DIO-non-responders (nResp, n = 11) upon identification in week 6. Mice receiving an LFD served as control group. Body composition (B), fasted blood glucose (C), plasma insulin (D) and leptin (E) levels. (F) HOMA-IR calculated from fasted blood glucose and insulin levels. (G) LAR. (H) Liver triglyceride levels. All data represented as mean ± SEM of n = 10–15 animals per group. Differences between DIO-responder (Resp, black circle) and DIO-non-responder (nResp, white circle) mice were calculated by Student’s t-test. *p < 0.05, **p < 0.01, ***p < 0.001.

Table 1.

Physiological parameters of nResp and Resp mice

 Week 6 Week 20 
nResp Resp p-value nResp Resp p-value 
Amount SEM Amount SEM Amount SEM Amount SEM 
Body weight (g) 21.3 0.2 24.9 0.3 <0.001 33.4 0.6 45.7 1.0 <0.001 
Lean mass (g) 17.1 0.2 19.1 0.3 <0.001 24.1 0.3 30.4 0.5 <0.001 
Fat mass (g) 4.8 0.1 6.2 0.3 <0.01 10.8 0.7 19.8 1.0 <0.001 
Body length (cm) 9.0 0.1 9.5 0.2 0.071 10.3 0.1 10.8 0.1 0.008 
Week 10 
Food intake (g/day/g BW) 0.087 0.002 0.082 0.004 0.205      
Water intake (g/day/g BW) 0.098 0.006 0.090 0.006 0.354      
 Week 6 Week 20 
nResp Resp p-value nResp Resp p-value 
Amount SEM Amount SEM Amount SEM Amount SEM 
Body weight (g) 21.3 0.2 24.9 0.3 <0.001 33.4 0.6 45.7 1.0 <0.001 
Lean mass (g) 17.1 0.2 19.1 0.3 <0.001 24.1 0.3 30.4 0.5 <0.001 
Fat mass (g) 4.8 0.1 6.2 0.3 <0.01 10.8 0.7 19.8 1.0 <0.001 
Body length (cm) 9.0 0.1 9.5 0.2 0.071 10.3 0.1 10.8 0.1 0.008 
Week 10 
Food intake (g/day/g BW) 0.087 0.002 0.082 0.004 0.205      
Water intake (g/day/g BW) 0.098 0.006 0.090 0.006 0.354      

Impaired glucose homeostasis precedes the onset of hepatosteatosis

Next, we investigated whether differences in body weight gain were associated with changes in blood glucose, insulin or leptin levels. Surprisingly, blood glucose levels in Resp mice were elevated compared with nResp mice already at 6 weeks of age (8.5 ± 0.5 versus 6.8 ± 0.4 mm, p < 0.05) (Fig. 1C). Furthermore, plasma insulin (0.88 ± 0.10 versus 0.44 ± 0.06 µg/l, p < 0.01) and leptin (0.54 ± 0.08 versus 1.62 ± 0.31 ng/ml, p < 0.01) levels were also increased in Resp mice at this early time point (Fig. 1D and E). The differences of these blood parameters were stable and further increased over time (Fig. 1C–E). Accordingly, HOMA-IR indicated that Resp mice already at 6 weeks of age were less insulin sensitive compared with nResp mice (8.3 ± 1.1 versus 3.4 ± 0.6, p < 0.001) (Fig. 1F). The leptin per adiponectin ratio (LAR) is a second measure of systemic insulin resistance (19) and showed the same differences between Resp and nResp mice (Fig. 1G). Therefore, the impaired insulin sensitivity observed at the age of 20 weeks in Resp mice (Fig. 1F and G) was already manifested at 6 weeks of age and preceded the obese phenotype. Importantly, liver triglyceride levels were not different at 6 weeks of age but increased in Resp mice at 20 weeks of age (0.82 ± 0.08 versus 1.21 ± 0.07 µg/µgprotein) (Fig. 1H). Absolute liver weight was not different between Resp and nResp mice at 6 (42.2 ± 1.3 versus 39.3 ± 1.2 mg/gbody weight) and 20 (33.3 ± 1.1 versus 31.8 ± 0.9 mg/gbody weight) weeks of age, respectively.

Reduced Igfbp2 expression in young DIO responders

Recently, it has been shown that human hepatic steatosis is associated with reduced expression of hepatic IGFBP2 (18). In Resp mice, expression of hepatic Igfbp2 was reduced by 35.55 ± 10.20% (p < 0.01) and 40.68 ± 8.11% (p < 0.001) at weeks 6 and 20, respectively, as compared with nResp mice (Fig. 2A and B). This decrease of gene expression was paralleled by reduced IGFBP-2 protein level in liver tissue (Fig. 2C). In order to investigate whether plasma levels of other IGFBP isoforms are also altered in Resp mice, we performed quantitative western ligand blots against IGF-2 (20). IGFBP-2 levels were decreased in Resp mice (p < 0.01) (Fig. 2D), whereas IGFBP-3 levels were increased at week 6 (p < 0.01) (Fig. 2E). Expression of IGFBP-4 was below the limit of quantification of the analytical system. In the circulation, IGFBP-2 binds to IGF-1 and decreases its bioavailability (21). In obese subjects, reduced expression of IGFBP2 results in increased serum levels of free IGF-1 (14). Accordingly, serum levels of total IGF-1 were increased in Resp mice compared with nResp mice (Fig. 2F).

Figure 2.

Modulation of the IGF axis. Hepatic Igfbp2 expression in 6 (A) and 20 (B) weeks old mice and the corresponding protein levels (C). Plasma IGFBP-2 (D) and IGFBP-3 (E) levels of the same animals determined by quantitative western ligand blot against IGF-2. (F) Plasma IGF-1 levels. All data represented as mean ± SEM of n = 8–12 animals per group. Differences between DIO-responder (Resp, black circle) and DIO-non-responder (nResp, white circle) mice were calculated by Student’s t-test. *p < 0.05, **p < 0.01, ***p < 0.001.

Figure 2.

Modulation of the IGF axis. Hepatic Igfbp2 expression in 6 (A) and 20 (B) weeks old mice and the corresponding protein levels (C). Plasma IGFBP-2 (D) and IGFBP-3 (E) levels of the same animals determined by quantitative western ligand blot against IGF-2. (F) Plasma IGF-1 levels. All data represented as mean ± SEM of n = 8–12 animals per group. Differences between DIO-responder (Resp, black circle) and DIO-non-responder (nResp, white circle) mice were calculated by Student’s t-test. *p < 0.05, **p < 0.01, ***p < 0.001.

Increased Igfbp2 methylation detected early in life of Resp mice

Reduced expression of IGFBP2 in NAFLD and NASH patients was associated with increased methylation of one CpG (cg11669516) in the first intron of the human IGFBP2 gene at position 2965 bp downstream the transcription start site (TSS) (18). Because the genomic region around cg11669516 is highly conserved between humans and mice (Fig. 3A), we investigated whether methylation at this site also occurs in young obesity-prone mice, and whether this methylation is paralleled by early differences in Igfbp2 expression. Indeed, methylation at the corresponding mouse CpG (CpG2605, 2605 bp downstream the TSS) was increased by 6.2 ± 1.2% (p < 0.001) in Resp mice at 6 weeks of age (Fig. 3B). In addition, we used direct bisulfite sequencing PCR (dBSP) to study CpG methylation of the proximal promoter region and identified three additional CpG sites (CpG842, CpG738 and CpG884) that were methylated differentially in nResp and Resp mice (

, Fig. S2A and B). However, as these sites could not be validated by pyrosequencing (, Fig. S2CE), we concentrated on CpG2605 for further analyses. Methylation of the intronic CpG2605 in the young animals was inversely correlated with expression of Igfbp2 (R2 = 0.551, p < 0.001) and positively correlated with body weight (R2 = 0.686, p < 0.001) (Fig. 3C and D). Neither blood glucose nor HOMA-IR was correlated with CpG2605 methylation; however, the LAR was positively correlated with CpG2605 methylation (, Fig. S3A–C). The methylation difference between both groups persisted, as Resp mice displayed increased CpG2605 methylation compared with nResp mice also at 20 weeks of age (6.0 ± 1.3%, p < 0.001) (Fig. 3E). Again, methylation of CpG2605 was negatively correlated with gene expression (R2 = 0.392, p < 0.01) and positively with body weight (R2 = 0.339, p < 0.01) (Fig. 3F and G). At this later time point, methylation of CpG2605 was also correlated with blood glucose and HOMA-IR but not with LAR (). In summary, these data indicate that susceptibility to DIO and insulin resistance is associated with increased Igfbp2 methylation and reduced expression early in life.
Figure 3.

Differential methylation of hepatic Igfbp2 at CpG2605. (A) Genomic localization of cg11669516 in the human IGFBP2 gene (region −500 bp and +3500 bp relative to the transcriptional start site). Methylation of the mouse homologue CpG2605 in week 6 (B) and in week 20 (E), respectively. Correlation of Igfbp2 expression and CpG2605 methylation at 6 (C) and 20 (F) weeks of age. Correlation of CpG2605 methylation and body weight at 6 (D) and 20 (G) weeks of age. All data represented as mean ± SEM of n = 10–14 animals per group. Differences between Resp (black circle) and nResp (white circle) mice were calculated by Student’s t-test. *p < 0.05, **p < 0.01. R2 = Pearson’s correlation coefficient.

Figure 3.

Differential methylation of hepatic Igfbp2 at CpG2605. (A) Genomic localization of cg11669516 in the human IGFBP2 gene (region −500 bp and +3500 bp relative to the transcriptional start site). Methylation of the mouse homologue CpG2605 in week 6 (B) and in week 20 (E), respectively. Correlation of Igfbp2 expression and CpG2605 methylation at 6 (C) and 20 (F) weeks of age. Correlation of CpG2605 methylation and body weight at 6 (D) and 20 (G) weeks of age. All data represented as mean ± SEM of n = 10–14 animals per group. Differences between Resp (black circle) and nResp (white circle) mice were calculated by Student’s t-test. *p < 0.05, **p < 0.01. R2 = Pearson’s correlation coefficient.

CpG2605 methylation modulates expression of the Igfbp2 gene

In order to investigate whether methylation of CpG2605 and expression of Igfbp2 are causally linked, we generated a luciferase reporter construct. It carried a 700-bp fragment of intron 1 in a CpG-free backbone vector (pCpGL) and was in vitro methylated by methyltransferase HpaII (CCGG) and M.SssI (CG), respectively (Fig. 4A). Of note, CpG2605 is embedded into an HpaII recognition site (Fig. 3A) that is unique in pCpGL-Igfbp2. Therefore, in vitro methylation of pCpGL-Igfbp2 by HpaII allowed us to selectively study the impact of CpG2605 methylation on Igfbp2 expression. After in vitro methylation, luciferase activity was assessed in AML12 hepatocytes. While full methylation by M.SssI almost completely abolished luciferase activity (−91.3 ± 3.9%, p < 0.001), selective methylation of CpG2605 with HpaII led to a reduction of luciferase activity by 84.8 ± 4.7% (p < 0.001) (Fig. 4B), indicating that Igfbp2 expression is indeed modulated by methylation of CpG2605. Because the differences in body weight are rather small within HFD-fed C57BL/6J mice, we also compared male C57BL/6J mice with male New Zealand Obese (NZO) mice fed a standard diet at the age of 6 weeks. NZO mice develop severe obesity and are, in contrast to C57BL/6J, susceptible to T2D (22,23). Already at the age of 6 weeks, NZO mice show increased body weight mainly because of increased fat mass compared with C57BL/6J mice which is also reflected by the significantly elevated plasma leptin concentrations. Accordingly, the young NZO mice are insulin resistant as fasted insulin levels and LAR are significantly higher than in C57BL/6 mice. However, both strains do not differ significantly in their fasted blood glucose levels (Table 2). Indeed, NZO mice displayed decreased hepatic Igfbp2 expression levels and increased methylation of CpG2605 compared with C57BL/6J mice (Fig. 4C and D). Again, the degree of CpG2605 methylation correlated with body weight (Fig. 4E) as well as with insulin concentration (Fig. 4F) and LAR (Fig. 4G).

Figure 4.

pCpGL–Igfbp2 luciferase reporter assays. (A) Intragenic region of murine Igfbp2 gene spanning from 2320 to 2997 bp relative to TSS. Recognition sites for methyltransferases HpaII and M.SssI, CpG2605 depicted in grey. (B) Luciferase activity upon selective methylation of pCpGL-Igfbp2. Igfbp2 expression (C) and CpG2605 methylation (D) in NZO mice. Correlation of CpG2605 methylation and body weight (E), insulin concentrations (F) and LAR (G) of 6 weeks old HFD-fed C57BL/6J and NZO mice. All data represented as mean ± SEM of n = 3–4 independent experiments and n = 7 animals per group, respectively. *p < 0.05, **p < 0.01 and ***p < 0.001 compared with untreated control group.

Figure 4.

pCpGL–Igfbp2 luciferase reporter assays. (A) Intragenic region of murine Igfbp2 gene spanning from 2320 to 2997 bp relative to TSS. Recognition sites for methyltransferases HpaII and M.SssI, CpG2605 depicted in grey. (B) Luciferase activity upon selective methylation of pCpGL-Igfbp2. Igfbp2 expression (C) and CpG2605 methylation (D) in NZO mice. Correlation of CpG2605 methylation and body weight (E), insulin concentrations (F) and LAR (G) of 6 weeks old HFD-fed C57BL/6J and NZO mice. All data represented as mean ± SEM of n = 3–4 independent experiments and n = 7 animals per group, respectively. *p < 0.05, **p < 0.01 and ***p < 0.001 compared with untreated control group.

Table 2.

Physiological parameters of male C57BL/6J and NZO mice

 C57BL/6J (n = 7) NZO (n = 7) p-value 
Amount SEM Amount SEM 
Body weight (g) 22.9 0.69 33.7 0.93 <0.001 
Lean mass (g) 20.5 0.57 27.2 0.46 <0.001 
Fat mass (g) 1.2 0.14 6.3 0.24 <0.001 
Fasted BG (mm6.3 0.21 7.5 0.78 0.175 
Fasted insulin (µg/L) 0.7 0.08 2.9 0.45 <0.001 
HOMA-IR (AU) 5.2 0.62 24.5 4.46 0.001 
Leptin (ng/ml) 1.1 0.23 16.67 1.24 <0.001 
Adiponectin (ng/ml) 6131 430 6227 458 0.881 
LAR (pg/ng) 0.157 0.028 2.800 0.328 <0.001 
 C57BL/6J (n = 7) NZO (n = 7) p-value 
Amount SEM Amount SEM 
Body weight (g) 22.9 0.69 33.7 0.93 <0.001 
Lean mass (g) 20.5 0.57 27.2 0.46 <0.001 
Fat mass (g) 1.2 0.14 6.3 0.24 <0.001 
Fasted BG (mm6.3 0.21 7.5 0.78 0.175 
Fasted insulin (µg/L) 0.7 0.08 2.9 0.45 <0.001 
HOMA-IR (AU) 5.2 0.62 24.5 4.46 0.001 
Leptin (ng/ml) 1.1 0.23 16.67 1.24 <0.001 
Adiponectin (ng/ml) 6131 430 6227 458 0.881 
LAR (pg/ng) 0.157 0.028 2.800 0.328 <0.001 

Plasticity of Igfbp2 methylation

In humans, IGFBP2 expression is increased by caloric restriction and exercise, two mechanisms known to reduce liver fat content (24,25). In order to investigate the plasticity of Igfbp2 methylation under conditions that modulate Igfbp2 gene expression, we performed a second set of experiments with obese mice that had received HFD for 18 weeks. The first set of mice was killed directly after the HFD feeding (HFD group); the second group of mice was subjected to 25% weight loss by restrictive feeding of an LFD (10 kcal% from fat, 16.1 kJ/g, D12450B, Research Diets, Inc.) for 3 weeks (HFD-CR group). The third group of mice was also caloric restricted for 3 weeks but received ad libitum HFD again afterwards for another 3 weeks (HFD–CR–HFD group). Caloric restriction (CR) led to the desired weight loss, and re-feeding with HFD led to a complete re-gain of body weight (Fig. 5A). Weight loss and re-gain were because of changes in both lean and fat mass (Fig. 5B). In parallel to the weight loss, expression of Igfbp2 increased in mice of the HFD–CR group (+52.2 ± 15.7%, p < 0.05) and decreased again upon HFD re-feeding (Fig. 5C). Accordingly, methylation of CpG2605 revealed the opposite trends, with a decrease during CR (−2.6 ± 2.2%, p = 0.25) and increase (+5.8 ± 2.7%, p = 0.04) during HFD re-feeding (Fig. 5D). To test if changes in Igfbp2 expression correlate with alterations in blood glucose or plasma insulin we measured these two parameters in the indicated groups. Blood glucose concentrations did not change in response to weight loss but increased significantly after the re-feeding period (Fig. 5E). As expected caloric restriction resulted in a decrease of insulin levels which rose after re-feeding, however, without reaching the initial concentration (Fig. 5F). These data indicate the changes in methylation of CpG2605 occur in a similar manner as that of insulin levels.

Figure 5.

Igfbp2 expression and methylation during weight loss and regain. (A) Body weight development of mice fed an HFD for 20 weeks (HFD) which were then caloric restricted for 3 weeks (HFD–CR) and again fed an HFD for 3 weeks (HFD–CR–HFD). (B) Modulation of lean mass (LM) and fat mass (FM) during weight cycling. Hepatic Igfbp2 expression (C) and methylation of CpG2605 (D) during weight cycling. Blood glucose (E) and plasma insulin concentrations (F) of the indicated groups. All data represented as mean ± SEM (n = 10 animals per group). Differences between HFD, HFD–CR and HFD–CR–HFD mice were calculated by Student’s t-test. *p < 0.05, ***p < 0.001. R2 = Pearson’s correlation coefficient.

Figure 5.

Igfbp2 expression and methylation during weight loss and regain. (A) Body weight development of mice fed an HFD for 20 weeks (HFD) which were then caloric restricted for 3 weeks (HFD–CR) and again fed an HFD for 3 weeks (HFD–CR–HFD). (B) Modulation of lean mass (LM) and fat mass (FM) during weight cycling. Hepatic Igfbp2 expression (C) and methylation of CpG2605 (D) during weight cycling. Blood glucose (E) and plasma insulin concentrations (F) of the indicated groups. All data represented as mean ± SEM (n = 10 animals per group). Differences between HFD, HFD–CR and HFD–CR–HFD mice were calculated by Student’s t-test. *p < 0.05, ***p < 0.001. R2 = Pearson’s correlation coefficient.

Increased methylation of human CpG2965 in whole blood cells of obese and glucose intolerant men

Our data indicate that methylation of CpG2605 is a marker for obesity and linked to changes of glucose tolerance or insulin sensitivity. Therefore, we next measured methylation of human CpG2965 in whole blood cells of men with different stages of metabolic disease. Subjects were grouped according to their glucose tolerance obtained during an oral glucose tolerance test (oGTT) as well as BMI into normal glucose tolerant subjects with a BMI <27 (NGT.BMI <27) or BMI >30 (NGT.BMI >30) as well as impaired glucose tolerant with a BMI <27 (IGT.BMI <27) or BMI >30 (IGT.BMI >30). Compared with the NGT.BMI <27 group, methylation of CpG2965 was not different in the NGT.BMI >30 and IGT.BMI <27 groups, but increased significantly (p < 0.05) by 5.3 ± 2.2% in the IGT.BMI >30 group (Fig. 6A). Furthermore, methylation of CpG2965 correlated with fasted blood glucose levels (R2 = 0.077, p < 0.01) (Fig. 6B), but neither with BMI (R2 = 0.00023, p = 0.888) nor age (R2 = 0.002, p = 0.667).

Figure 6.

Methylation of CpG2965 in human whole blood cells. (A) DNA methylation of CpG2965 in whole blood cells from human subjects classified into normal glucose tolerant with BMI <27 (NGT.BMI <27) or BMI >30 (NGT.BMI >30) or impaired glucose tolerant with BMI <27 (IGT.BMI <27) or BMI >30 (IGT.BMI >30). Correlation of CpG2965 methylation and fasted blood glucose (B). All data represented as mean ± SEM. Differences between groups were calculated by Student’s t-test. *p < 0.05. R2 = Pearson’s correlation coefficient.

Figure 6.

Methylation of CpG2965 in human whole blood cells. (A) DNA methylation of CpG2965 in whole blood cells from human subjects classified into normal glucose tolerant with BMI <27 (NGT.BMI <27) or BMI >30 (NGT.BMI >30) or impaired glucose tolerant with BMI <27 (IGT.BMI <27) or BMI >30 (IGT.BMI >30). Correlation of CpG2965 methylation and fasted blood glucose (B). All data represented as mean ± SEM. Differences between groups were calculated by Student’s t-test. *p < 0.05. R2 = Pearson’s correlation coefficient.

Discussion

In this study, we investigated early inter-individual differences in the epigenetic regulation of hepatic Igfbp2 that determine DIO susceptibility in inbred C57BL/6J mice. Already at a very young age hepatic Igfbp2 expression was decreased and methylation at CpG2605 was increased in DIO responders. This epigenetic repression of hepatic Igfbp2 was stable over time and, most importantly, preceded the development of fatty liver. Selective in vitro methylation of CpG2605 decreased luciferase reporter activity, indicating a direct influence of methylation at this site on Igfbp2 expression levels. Transcriptional regulation of Igfbp2 largely followed body weight dynamics, as caloric restriction and HFD feeding conversely affected Igfbp2 expression and methylation. Furthermore, methylation of Igfbp2 was associated with blood glucose, a finding that was confirmed in human whole blood cells of obese men. In young individuals, enhanced Igfbp2 methylation reflects early deteriorations of glucose metabolism, which in turn represents an enhanced risk for fatty liver and insulin resistance later in life.

Serum levels of insulin-like growth factor 1 (IGF-1) are tightly regulated to ensure proper body weight development and growth in the neonatal period (26,27). IGF-1 binding proteins (IGFBPs) control the bioavailability of circulating IGF-1 (28). For example, the repression of IGFBP1 and IGFBP2 during chronic hyperinsulinemia increases the bioavailability of IGF-1 (free IGF-1) and supports body weight gain (14,29). Our present finding of reduced hepatic Igfbp2 expression in mice with elevated plasma insulin levels is in line with these published observations. Recently, we presented data of mouse lines selected for either high or low body weight gain for 58 generations (20). We showed that the sub-population of high body weight gainers exhibited a reduced hepatic Igfbp2 expression (20), similar to the data presented here. Importantly, the differential Igfbp2 expression and methylation reported in this study occurred in a single generation of mice, and differences were observed even between littermates.

We could show that repression of Igfbp2 in livers of obesity-prone mice is linked with increased methylation of CpG2605 in the first intron of the gene, already at a very young age. However, whether increased CpG2605 methylation is a result of elevated plasma insulin and glucose levels or independently established during development could not be resolved by our experiments. Nevertheless, our data indicate that epigenetic alterations of Igfbp2 occur early in life and precede the development of fatty liver.

In line with our observation, low plasma IGFBP-2 levels have also been found in obese children (30,31). Increased IGF-1 bioavailability at least partially explains the accelerated growth observed in children when reaching the pre-pubertal period (30). At this age, decreased IGFBP-2 serum levels were associated with increased body weight, body fat mass, serum insulin and leptin levels (32,33). Again, the phenotypic characteristics of our obesity-prone mice are in line with the observations in young humans. In addition, our study in inbred mice indicates that epigenetic variability early in life might contribute to the observed differences in circulating IGFBP-2 levels. In contrast to IGFBP-2, IGFBP-3 levels were increased in young mice prone to weight gain. In humans, serum levels of IGFBP-3 are decreased in children born small for gestational age, and tend to be increased in obese children (34,35). Furthermore, mice transgenic for Igfbp3 display increased serum levels of both IGF-1 and IGFBP-3 as well as impaired glucose tolerance and insulin resistance (36). Therefore, the increased IGFBP-3 levels found in young Resp mice might also contribute to the metabolic phenotype.

There are several possible mediators of Igfbp2 methylation and expression in obesity-prone individuals. Besides glucose, we found circulating insulin and leptin to be increased in obesity-prone mice compared with obesity-resistant mice. As mentioned above, insulin is a potent inhibitor of Igfbp2 expression (37,38). Therefore, it can be speculated that low insulin concentration in obesity-resistant mice may cause an increase in Igfbp2 expression. In contrast to insulin, leptin increases Igfbp2 expression in a dose-dependent manner (39). In DIO, however, Igfbp2 expression is rather low despite high levels of circulating leptin (39). This could be a result of the concomitant hyperinsulinemia or of the leptin resistance. The correlation of CpG2605 methylation and plasma leptin levels in our young animals might therefore reflect early metabolic disturbances induced by the HFD (39). .As overnutrition is known to suppress Igfbp2 expression (14) it is possible that the observed alterations in Igfbp2 mRNA levels and DNA methylation are just an epiphenomenon of elevated dietary intake. However, as food intake that was measured in week 10 on the individual basis was not different between responder and non-responder mice, we believe that alterations in Igfbp2 expression and methylation are rather a consequence of intrauterine and/or early postnatal programming.

As mentioned earlier, lifestyle interventions that improve insulin sensitivity and liver fat content, e.g. caloric restriction and exercise, have been shown to increase IGFBP2 expression (24,25). Caloric restriction also resulted in increased expression of hepatic Igfbp2 in our obese mice. De-methylation of CpG2605 under these conditions is apparently slower and did not reach statistical significance in our experiment. However, induction of CpG2605 methylation by re-feeding the HFD significantly increased methylation again. Together, these data indicate that long-term caloric intake can alter the methylation of CpG2605. The differential Igfbp2 expression as well as methylation associated with changes in body weight before the onset of liver fat accumulation might already play an important role during fetal programming. Human studies found an association of high IGFBP-2 cord serum concentrations with inter-uterine growth retardation in newborns (40), indicating an effect of IGFBP-2 on fetal growth.

In conclusion, our data suggest that increased methylation of hepatic Igfbp2 during infancy indicates an increased risk to develop obesity and ectopic fat accumulation in adolescence. Furthermore, our data suggest that early deteriorations of glucose metabolism parallels epigenetic deregulation of Igfbp2.

Materials and Methods

Animals

C57BL/6J breeding pairs (Charles River, Germany) received a standard chow (ssniff) and were housed in temperature-controlled room (22 + 1°C) on a 12:12 h light dark cycle. After weaning male mice were fed an HFD (60 kcal% fat, 21.9 kJ/g, D12492, Research Diets, Inc., USA) ad libitum in groups of two to six animals per cage. The mice were housed and handled in accordance with the ‘Principles of laboratory animal care’ (41). The animal welfare committees of the DIfE as well as the local authorities (LUGV, Brandenburg, Germany) approved all animal experiments. Weekly body weight measurements were performed in the morning (8–10 am) from 3 to 20 weeks of age. Body composition was analyzed by nuclear magnetic resonance (NMR, Minispec LF50, Bruker Biospin). Blood glucose was measured from tail blood with a Glucometer Elite (Bayer, Germany). At 10 weeks of age, food intake was monitored with an automatic drinking and feeding monitor system (TSE Systems, Germany). Mice were housed in individual cages and adapted to food baskets connected to weight sensors for 2 days. During the measurement period of 3 days, the baskets contained HFD pellets to which mice had ad libitum access. At the age of 6 and 20 weeks, respectively, mice were killed after 6 h fasting. All tissues were directly frozen in liquid nitrogen and stored at −80°C until further processing.

Plasma analysis

Plasma insulin was measured using the Mouse Ultra-sensitive insulin ELISA kit from ALPCO (USA). Insulin-like growth factor binding protein 2 (IGFBP-2) as well as insulin-like growth factor 1 (IGF-1) plasma levels were quantified using an enzyme-linked immunosorbent assays from R&D Systems (mouse-IGFBP-2, DY797; mouse IGF-1, DY791; R&D Systems, USA). The assays were performed according to the manufacturers’ instructions.

Hepatic triglycerides

Liver triglycerides were quantified using a commercial kit (RandoxTR-210, UK). Livers were homogenized in 10 mmol/l sodium dihydrogen phosphate, 1 mmol/l EDTA and 1% (vol./vol.) polyoxyethylene-10-tridecyl ether and incubated for 5 min at 37°C. After centrifugation, the triacylglycerols in the supernatant fraction could be detected.

Quantitative real-time PCR (qPCR)

Total RNA from liver tissue of mice was extracted, and cDNA synthesis as well as TaqMan gene expression assays were performed as described previously (42).

Quantitative western ligand blot analysis of IGFBP

Insulin-like growth factor binding proteins were analyzed in plasma by quantitative western ligand blot analysis, as described previously (43,44). Plasma as well as recombinant human IGFBP standards (R&D Systems, USA) were diluted 1:20 in artificial serum matrix (Biopanda) and boiled in sample buffer (312.5 mm Tris (pH 6.8), 50% (wt/vol) glycerol, 5 mm EDTA (pH 8), 1% (wt/vol) SDS and 0.02% bromophenol blue) for 5 min. Proteins were separated by electrophoreses [12% polyacrylamide gel (Peqlab “Perfect Blue”)] and transferred to polyvinylidene fluoride membrane (Millipore Corp., USA), followed by immunoblotting. The blots were incubated with biotin-labeled human IGF-2 (1:500; BioIGF-2-10; Ibt GmbH, Germany). IGFBP were detected by enhanced chemiluminescence using the reagent Luminata Forte (Millipore Corp., USA) and bands were visualized on KODAK Image Station 4000 MM (Molecular Imaging Systems; Carstream Health Inc., USA) and afterwards semi-quantified by using ImageQuant 5.2 software (Molecular Dynamics, USA). For each IGFBP, curve fitting was done by a 7-parametric nonlinear regression (Hill equation).

Isolation of genomic DNA and bisulfite treatment

Genomic DNA (gDNA) from liver tissue was isolated using the Invisorb Genomic DNA Kit II (Stratec, Germany) according to manufacturer’s instructions. Bisulfite conversion of 500 ng gDNA was performed using the EpiTect Fast Bisulfite Kit (QIAGEN, Germany). Bisulfite-converted DNA (bsDNA) was purified as recommended and eluted in a total volume of 15 µl.

dBSP and pyrosequencing

Bisulfite-specific primers were designed using MethPrimer (45). PCRs (2.5 µl bsDNA) were performed using TaKaRa EpiTaq™ HS (Takara Bio Inc., Japan) and amplicons purified with the QIAquick PCR Purification Kit (QIAGEN, Germany). Sanger sequencing of the samples was carried out by GATC Biotech (Germany). Quantification of dBSP was done as described previously (46). In order to avoid over-representation of cytosine peaks after normalization, the raw sequencing data were used for quantification (CodonCode Aligner Software, version 5.1.5, CodonCode Corporation, USA). Primers for pyrosequencing were designed using the PyroMark Assay Design software (QIAGEN, Germany). Pyrosequencing was performed on a PyroMark Q96 Pyrosequencer according to the manufacturer’s instructions and quantified using PyroMark Q96 software 2.5.7 (QIAGEN, Germany).

Construction of CpG-free reporter gene plasmid

A fragment of the murine Igfbp2 gene, 2320–2997 bp downstream of the TSS, was PCR amplified using primers generating restriction sites for BamHI (forward primer 5′-GATC GGATCC GAGC CTG GTT TTG CCA GGT TG) and NcoI (reversed primer 5′-GATC CCATGG GCC TGG CCT GAA TTC TGA AAG G) with JumpStart™ REDAccuTaq® LA DNA Polymerase (Sigma-Aldrich, USA). The purified PCR amplificates were cloned into the promoter-less CpG-free pCpGL-Basic vector (kindly provided by Prof. Rehli, Regensburg, Germany) upstream of the luciferase reporter gene. For plasmid amplification, the construct (pCpGL-Igfbp2) was transformed into One Shot® PIR1 Competent Escherichia coli (Invitrogen, USA).

In vitro methylation of pCpGL-Igfbp2 was performed using either 4 U/µg DNA M.SssI or 2 U/µg DNA HpaII (New England Biolabs, USA) in the presence of 160 µmS-adenosylmethionine and incubated at 37°C for 1.5 h. After plasmid purification using QIAGEN QIAquick PCR Purification Kit (QIAGEN, Germany) concentration was measured with the NanoDrop Spectrophometer (ND-1000, PeqLab, Germany).

AML12 cell culture and luciferase activity assay

The murine AML12 cell line (ATCC® CRL-2254™) was cultured in a 1:1 mixture of Dulbecco’s modified Eagle’s medium and Ham’s F12 medium (PAN-Biotech, Germany) with 0.005 mg/ml insulin (Roche Life Science, USA), 0.005 mg/ml transferrin (Sigma-Aldrich, USA), 5 ng/ml selenium (Sigma-Aldrich, USA) and 40 ng/ml dexamethasone (Sigma-Aldrich, USA) supplemented with 10% fetal bovine serum (Biochrom, Germany). Cells were maintained in humidified air replenished with 5% CO2. In a 48-well plate, 4 × 104 AML12 cells/well were co-transfected using Lipofectamine®2000 (Invitrogen, USA) with 500 ng/well unmethylated or methylated pCpGL-Igfbp2 and 5 ng/well pRL-TK Renilla control vector (Promega, USA). After 48 h, cells were harvested and analyzed for firefly and renilla luciferase activity using the dual-luciferase reporter assay system (Promega, USA). Firefly luciferase activity was normalized to its renilla luciferase transfection control.

Human participants and whole blood collection

The study population consisted of 109 German individuals recruited from the ongoing Tübingen Family Study. The study currently comprises >2500 non-related individuals at increased risk for T2D, i.e. non-diabetic subjects with family history of type-2 diabetes, BMI ≥27 kg/m2, impaired fasting glycaemia, and/or previous gestational diabetes and healthy controls. All participants underwent assessment of medical history, smoking status and alcohol consumption habits, physical examination, routine blood tests and oral glucose tolerance tests (oGTTs). The study population consisted of individuals with complete oGTT and documented absence of medication known to influence glucose tolerance, insulin sensitivity or insulin secretion. The study adhered to the Declaration of Helsinki and each participant signed a written informed consent. The Ethical Committee of the Medical Faculty of the University of Tübingen approved the study protocol. Subjects are grouped into normal glucose tolerant with BMI <27 (NGT.BMI <27, n = 29) or BMI >30 (NGT.BMI >30, n = 18) and impaired glucose tolerant with BMI <27 (IGT.BMI <27, n = 17) or BMI >30 (IGT.BMI >30, n = 31). Participant characteristics are shown in Table 3. A standardized 75 g oGTT was performed following a 10-h overnight fast. For the determination of plasma glucose venous blood samples were drawn at baseline and at time-points 30, 60, 90 and 120 min of the oGTT. For DNA methylation analysis, DNA was isolated from whole blood using a commercial kit (NucleoSpin, Macherey & Nagel, Düren, Germany).

Table 3.

Characteristics of normal and impaired glucose tolerant men

 NGT.BMI <27 (n = 29) NGT.BMI >30 (n = 18) IGT.BMI <27 (n = 17) IGT.BMI >30 (n = 31) 
Amount SEM Amount SEM Amount SEM Amount SEM 
Patient age (years) 57.8 1.73 60.2 1.27 61.2 1.37 57.2 1.59 
BMI 24.5 0.33 33.5 0.73 25.43 0.21 34.3 0.65 
Fasted BG (mm5.4 0.08 5.6 0.11 5.7 0.14 6.0 0.11 
oGTT (mm5.9 0.14 6.0 0.27 9.3 0.20 9.2 0.19 
 NGT.BMI <27 (n = 29) NGT.BMI >30 (n = 18) IGT.BMI <27 (n = 17) IGT.BMI >30 (n = 31) 
Amount SEM Amount SEM Amount SEM Amount SEM 
Patient age (years) 57.8 1.73 60.2 1.27 61.2 1.37 57.2 1.59 
BMI 24.5 0.33 33.5 0.73 25.43 0.21 34.3 0.65 
Fasted BG (mm5.4 0.08 5.6 0.11 5.7 0.14 6.0 0.11 
oGTT (mm5.9 0.14 6.0 0.27 9.3 0.20 9.2 0.19 

Statistical analysis

Statistical analysis was performed by Student’s t-test, one-way ANOVA or linear regression analysis using the software Prism6 from Graph Pad Software (La Jolla, USA). Significance levels were set for p-values <0.05 (*), 0.01 (**) and 0.001 (***).

Supplementary Material

is available at HMG online.

Acknowledgements

The authors thank Christine Gumz, Andrea Teichmann and Kathrin Warnke of the German Institute of Human Nutrition Potsdam-Rehbruecke for their skillful technical assistance. Furthermore, we thank Prof. Rehli from Regensburg University (Germany) for providing the pCpGL-basic vector used in the luciferase assays and Dr. Christine Höflich (Ligandis GbR, Germany) for performing the western ligand blot analysis.

Conflict of Interest statement. None declared.

Funding

This work was supported by the German Ministry of Education and Research [BMBF, DZD, grant 01GI0922]. Funding to pay the Open Access publication charges for this article was provided by German Ministry of Education and Research.

References

1
WHO
. (
2014
)
Global status report on noncommunicable diseases 2014
 .
World Health Organization 2014
.
2
Phillips
C.M.
(
2013
)
Nutrigenetics and metabolic disease: current status and implications for personalised nutrition
.
Nutrients
 ,
5
,
32
57
.
3
Basile
K.J.
Johnson
M.E.
Xia
Q.
Grant
S.F.
(
2014
)
Genetic susceptibility to type 2 diabetes and obesity: follow-up of findings from genome-wide association studies
.
Int. J. Endocrinol
 .,
2014
,
769671
.
4
Locke
A.E.
Kahali
B.
Berndt
S.I.
Justice
A.E.
Pers
T.H.
Day
F.R.
Powell
C.
Vedantam
S.
Buchkovich
M.L.
Yang
J
., et al.  . (
2015
)
Genetic studies of body mass index yield new insights for obesity biology
.
Nature
 ,
518
,
197
206
.
5
Trerotola
M.
Relli
V.
Simeone
P.
Alberti
S.
(
2015
)
Epigenetic inheritance and the missing heritability
.
Hum. Genomics
 ,
9
,
17.
6
Feil
R.
Fraga
M.F.
(
2011
)
Epigenetics and the environment: emerging patterns and implications
.
Nat. Rev. Genet
 .,
13
,
97
109
.
7
Koza
R.A.
Nikonova
L.
Hogan
J.
Rim
J.S.
Mendoza
T.
Faulk
C.
Skaf
J.
Kozak
L.P.
(
2006
)
Changes in gene expression foreshadow diet-induced obesity in genetically identical mice
.
PLoS Genet
 .,
2
,
e81.
8
Stefan
N.
Kantartzis
K.
Haring
H.U.
(
2008
)
Causes and metabolic consequences of Fatty liver
.
Endocr. Rev
 .,
29
,
939
960
.
9
Ekstedt
M.
Franzen
L.E.
Mathiesen
U.L.
Thorelius
L.
Holmqvist
M.
Bodemar
G.
Kechagias
S.
(
2006
)
Long-term follow-up of patients with NAFLD and elevated liver enzymes
.
Hepatology
 ,
44
,
865
873
.
10
Attig
L.
Vige
A.
Gabory
A.
Karimi
M.
Beauger
A.
Gross
M.S.
Athias
A.
Gallou-Kabani
C.
Gambert
P.
Ekstrom
T.J
., et al.  . (
2013
)
Dietary alleviation of maternal obesity and diabetes: increased resistance to diet-induced obesity transcriptional and epigenetic signatures
.
PLoS One
 ,
8
,
e66816.
11
Waterland
R.A.
Lin
J.R.
Smith
C.A.
Jirtle
R.L.
(
2006
)
Post-weaning diet affects genomic imprinting at the insulin-like growth factor 2 (Igf2) locus
.
Hum. Mol. Genet
 .,
15
,
705
716
.
12
Deodati
A.
Inzaghi
E.
Liguori
A.
Puglianiello
A.
Germani
D.
Brufani
C.
Fintini
D.
Cappa
M.
Barbetti
F.
Cianfarani
S.
(
2013
)
IGF2 methylation is associated with lipid profile in obese children
.
Horm. Res. Paediatr
 .,
79
,
361
367
.
13
Perkins
E.
Murphy
S.K.
Murtha
A.P.
Schildkraut
J.
Jirtle
R.L.
Demark-Wahnefried
W.
Forman
M.R.
Kurtzberg
J.
Overcash
F.
Huang
Z
., et al.  . (
2012
)
Insulin-like growth factor 2/H19 methylation at birth and risk of overweight and obesity in children
.
J. Pediatr
 .,
161
,
31
39
.
14
Nam
S.Y.
Lee
E.J.
Kim
K.R.
Cha
B.S.
Song
Y.D.
Lim
S.K.
Lee
H.C.
Huh
K.B.
(
1997
)
Effect of obesity on total and free insulin-like growth factor (IGF)-1, and their relationship to IGF-binding protein (BP)-1, IGFBP-2, IGFBP-3, insulin, and growth hormone
.
Int. J. Obes. Relat. Metab. Disord
 .,
21
,
355
359
.
15
Allen
N.E.
Appleby
P.N.
Kaaks
R.
Rinaldi
S.
Davey
G.K.
Key
T.J.
(
2003
)
Lifestyle determinants of serum insulin-like growth-factor-I (IGF-I), C-peptide and hormone binding protein levels in British women
.
Cancer Causes Control
 ,
14
,
65
74
.
16
Hoeflich
A.
Wu
M.
Mohan
S.
Foll
J.
Wanke
R.
Froehlich
T.
Arnold
G.J.
Lahm
H.
Kolb
H.J.
Wolf
E.
(
1999
)
Overexpression of insulin-like growth factor-binding protein-2 in transgenic mice reduces postnatal body weight gain
.
Endocrinology
 ,
140
,
5488
5496
.
17
Wheatcroft
S.B.
Kearney
M.T.
Shah
A.M.
Ezzat
V.A.
Miell
J.R.
Modo
M.
Williams
S.C.
Cawthorn
W.P.
Medina-Gomez
G.
Vidal-Puig
A
., et al.  . (
2007
)
IGF-binding protein-2 protects against the development of obesity and insulin resistance
.
Diabetes
 ,
56
,
285
294
.
18
Ahrens
M.
Ammerpohl
O.
von Schonfels
W.
Kolarova
J.
Bens
S.
Itzel
T.
Teufel
A.
Herrmann
A.
Brosch
M.
Hinrichsen
H
., et al.  . (
2013
)
DNA methylation analysis in nonalcoholic fatty liver disease suggests distinct disease-specific and remodeling signatures after bariatric surgery
.
Cell Metab
 .,
18
,
296
302
.
19
Finucane
F.M.
Luan
J.
Wareham
N.J.
Sharp
S.J.
O’Rahilly
S.
Balkau
B.
Flyvbjerg
A.
Walker
M.
Hojlund
K.
Nolan
J.J
., et al.  . (
2009
)
Correlation of the leptin:adiponectin ratio with measures of insulin resistance in non-diabetic individuals
.
Diabetologia
 ,
52
,
2345
2349
.
20
Hoeflich
A.
Schmidt
P.
Foll
J.
Rottmann
O.
Weber
M.M.
Kolb
H.J.
Pirchner
F.
Wolf
E.
(
1998
)
Altered growth of mice divergently selected for body weight is associated with complex changes in the growth hormone/insulin-like growth factor system
.
Growth. Horm. IGF Res
 .,
8
,
113
123
.
21
Jones
J.I.
Clemmons
D.R.
(
1995
)
Insulin-like growth factors and their binding proteins: biological actions
.
Endocr. Rev
 .,
16
,
3
34
.
22
Jurgens
H.S.
Schurmann
A.
Kluge
R.
Ortmann
S.
Klaus
S.
Joost
H.G.
Tschop
M.H.
(
2006
)
Hyperphagia, lower body temperature, and reduced running wheel activity precede development of morbid obesity in New Zealand obese mice
.
Physiol. Genomics
 ,
25
,
234
241
.
23
Kluth
O.
Mirhashemi
F.
Scherneck
S.
Kaiser
D.
Kluge
R.
Neschen
S.
Joost
H.G.
Schurmann
A.
(
2011
)
Dissociation of lipotoxicity and glucotoxicity in a mouse model of obesity associated diabetes: role of forkhead box O1 (FOXO1) in glucose-induced beta cell failure
.
Diabetologia
 ,
54
,
605
616
.
24
Counts
D.R.
Gwirtsman
H.
Carlsson
L.M.
Lesem
M.
Cutler
G.B.
Jr.
(
1992
)
The effect of anorexia nervosa and refeeding on growth hormone-binding protein, the insulin-like growth factors (IGFs), and the IGF-binding proteins
.
J. Clin. Endocrinol. Metab
 .,
75
,
762
767
.
25
Berg
U.
Enqvist
J.K.
Mattsson
C.M.
Carlsson-Skwirut
C.
Sundberg
C.J.
Ekblom
B.
Bang
P.
(
2008
)
Lack of sex differences in the IGF-IGFBP response to ultra endurance exercise
.
Scand. J. Med. Sci. Sports
 ,
18
,
706
714
.
26
Liu
J.P.
Baker
J.
Perkins
A.S.
Robertson
E.J.
Efstratiadis
A.
(
1993
)
Mice carrying null mutations of the genes encoding insulin-like growth factor I (Igf-1) and type 1 IGF receptor (Igf1r)
.
Cell
 ,
75
,
59
72
.
27
Wu
Y.
Sun
H.
Yakar
S.
LeRoith
D.
(
2009
)
Elevated levels of insulin-like growth factor (IGF)-I in serum rescue the severe growth retardation of IGF-I null mice
.
Endocrinology
 ,
150
,
4395
4403
.
28
Russo
V.C.
Azar
W.J.
Yau
S.W.
Sabin
M.A.
Werther
G.A.
(
2015
)
IGFBP-2: the dark horse in metabolism and cancer
.
Cytokine Growth Factor Rev
 .,
26
,
329
346
.
29
Hoeflich
A.
Russo
V.C.
(
2015
)
Physiology and pathophysiology of IGFBP-1 and IGFBP-2—consensus and dissent on metabolic control and malignant potential
.
Best Pract. Res. Clin. Endocrinol. Metab
 .,
29
,
685
700
.
30
Ballerini
M.G.
Ropelato
M.G.
Domene
H.M.
Pennisi
P.
Heinrich
J.J.
Jasper
H.G.
(
2004
)
Differential impact of simple childhood obesity on the components of the growth hormone-insulin-like growth factor (IGF)-IGF binding proteins axis
.
J. Pediatr. Endocrinol. Metab
 .,
17
,
749
757
.
31
Street
M.E.
Smerieri
A.
Montanini
L.
Predieri
B.
Iughetti
L.
Valenzise
M.
De Luca
F.
Vigone
M.
Weber
G.
Maghnie
M
., et al.  . (
2013
)
Interactions among pro-inflammatory cytokines, IGF system and thyroid function in pre-pubertal obese subjects
.
J. Biol. Regul. Homeost. Agents
 ,
27
,
259
266
.
32
Ko
J.M.
Park
H.K.
Yang
S.
Hwang
I.T.
(
2012
)
Influence of catch-up growth on IGFBP-2 levels and association between IGFBP-2 and cardiovascular risk factors in Korean children born SGA
.
Endocr. J
 .,
59
,
725
733
.
33
de Kort
S.W.
van Doorn
J.
van de Sande
A.G.
Leunissen
R.W.
Hokken-Koelega
A.C.
(
2010
)
Serum insulin-like growth factor-binding protein-2 levels and metabolic and cardiovascular risk factors in young adults and children born small for gestational age
.
J. Clin. Endocrinol. Metab
 .,
95
,
864
871
.
34
Marzano
F.
Ventura
A.
Caratozzolo
M.F.
Aiello
I.
Mastropasqua
F.
Brunetti
G.
Cavallo
L.
Sbisa
E.
Faienza
M.F.
Tullo
A.
(
2015
)
The p53 family member p73 modulates the proproliferative role of IGFBP3 in short children born small for gestational age
.
Mol. Biol. Cell
 .,
26
,
2733
2741
.
35
Rubin
D.A.
Castner
D.M.
Pham
H.
Ng
J.
Adams
E.
Judelson
D.A.
(
2014
)
Hormonal and metabolic responses to a resistance exercise protocol in lean children, obese children and lean adults
.
Pediatr. Exerc. Sci
 .,
26
,
444
454
.
36
Silha
J.V.
Gui
Y.
Murphy
L.J.
(
2002
)
Impaired glucose homeostasis in insulin-like growth factor-binding protein-3-transgenic mice
.
Am. J. Physiol. Endocrinol. Metab
 .,
283
,
E937
E945
.
37
Suikkari
A.M.
Koivisto
V.A.
Rutanen
E.M.
Yki-Jarvinen
H.
Karonen
S.L.
Seppala
M.
(
1988
)
Insulin regulates the serum levels of low molecular weight insulin-like growth factor-binding protein
.
J. Clin. Endocrinol. Metab
 .,
66
,
266
272
.
38
Boni-Schnetzler
M.
Schmid
C.
Mary
J.L.
Zimmerli
B.
Meier
P.J.
Zapf
J.
Schwander
J.
Froesch
E.R.
(
1990
)
Insulin regulates the expression of the insulin-like growth factor binding protein 2 mRNA in rat hepatocytes
.
Mol. Endocrinol
 .,
4
,
1320
1326
.
39
Hedbacker
K.
Birsoy
K.
Wysocki
R.W.
Asilmaz
E.
Ahima
R.S.
Farooqi
I.S.
Friedman
J.M.
(
2010
)
Antidiabetic effects of IGFBP2, a leptin-regulated gene
.
Cell Metab
 .,
11
,
11
22
.
40
Smerieri
A.
Petraroli
M.
Ziveri
M.A.
Volta
C.
Bernasconi
S.
Street
M.E.
(
2011
)
Effects of cord serum insulin, IGF-II, IGFBP-2, IL-6 and cortisol concentrations on human birth weight and length: pilot study
.
PLoS One
 ,
6
,
e29562.
41
U.S. Office of Science and Technology Policy
(
1985
)
Laboratory animal welfare; U.S. government principles for the utilization and care of vertebrate animals used in testing, research and training; notice
.
Fed. Regist
 .,
50
,
20864
20865
.
42
Schwenk
R.W.
Jonas
W.
Ernst
S.B.
Kammel
A.
Jahnert
M.
Schurmann
A.
(
2013
)
Diet-dependent alterations of hepatic Scd1 expression are accompanied by differences in promoter methylation
.
Horm. Metab. Res
 .,
45
,
786
794
.
43
Hossenlopp
P.
Seurin
D.
Segovia-Quinson
B.
Binoux
M.
(
1986
)
Identification of an insulin-like growth factor-binding protein in human cerebrospinal fluid with a selective affinity for IGF-II
.
FEBS Lett
 .,
208
,
439
444
.
44
Wirthgen
E.
Hoflich
C.
Spitschak
M.
Helmer
C.
Brand
B.
Langbein
J.
Metzger
F.
Hoeflich
A.
(
2016
)
Quantitative western ligand blotting reveals common patterns and differential features of IGFBP-fingerprints in domestic ruminant breeds and species
.
Growth. Horm. IGF Res
 .,
26
,
42
49
.
45
Li
L.C.
Dahiya
R.
(
2002
)
MethPrimer: designing primers for methylation PCRs
.
Bioinformatics
 ,
18
,
1427
1431
.
46
Jiang
M.
Zhang
Y.
Fei
J.
Chang
X.
Fan
W.
Qian
X.
Zhang
T.
Lu
D.
(
2010
)
Rapid quantification of DNA methylation by measuring relative peak heights in direct bisulfite-PCR sequencing traces
.
Lab. Invest
 .,
90
,
282
290
.
This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (http://creativecommons.org/licenses/by-nc/4.0/), which permits non-commercial re-use, distribution, and reproduction in any medium, provided the original work is properly cited. For commercial re-use, please contact journals.permissions@oup.com

Supplementary data