DNA Methylation Variability at Growth-Related Imprints Does not Contribute to Overweight in Monozygotic Twins Discordant for BMI
Defective genomic imprinting is often associated with syndromes that include abnormal growth as a clinical phenotype. However, whether differential methylation at imprinted loci also contributes to nonsyndromic abnormal body weight regulation is yet unknown. In this study, we investigated a potential contribution of aberrant DNA methylation at nine differentially methylated regions (DMRs) to the development of nonsyndromic overweight. Sixteen monozygotic (MZ) twins discordant for BMI (BMI difference ranging from 2.9–9.5 kg/m2) were recruited from the East Flanders Prospective Twin Survey. DNA extracted from saliva samples was bisulfite-treated followed by PCR amplification of target regions in DMRs most representative for abnormal growth syndromes: KvDMR1, H19 CTCF4, H19 CTCF6, IGF2 DMR0, IGF2 DMR2, GRB10, MEST, SNRPN, GNAS XL-α-s and GNAS Exon1A. At the DMRs analyzed, methylation-dependent primer extension experiments revealed only small intrapair differences in methylation indexes (MI) between the heavy and lean co-twins. In addition, no significant correlations between intrapair BMI differences and intrapair differences in MI were observed. In conclusion, DNA methylation variability at the nine DMRs analyzed does not seem to contribute to the discordancy in BMI observed in these MZ twins.
High calorie diets combined with a sedentary lifestyle has resulted in a dramatic rise of the worldwide incidence of obesity (1). Besides the key role of these lifestyle factors, some individuals appear more susceptible to become obese than others, and family and twin studies have shown that genetic predisposition is important (2,3). Although much research is currently focused on the genetics of obesity (4), the awareness that epigenetic regulation of gene expression might also contribute to an individual's predisposition to obesity is currently growing (5,6).
An important epigenetic phenomenon is genomic imprinting, where genes are marked by DNA methylation during gametogenesis, resulting in monoallelic, parent-of-origin-specific expression in the offspring (7). Since several imprinted genes are critical for growth and development, genomic imprinting disorders often involve abnormal growth (6,8,9,10).
Human chromosome 11p15 comprises two imprinted control regions important for fetal and postnatal growth: the KvDMR1 and the insulin-like growth factor-2 (IGF2)/H19 differentially methylated region (DMR) (8). The maternally methylated KvDMR1 in the promoter of the paternally expressed noncoding RNA KCNQ1OT1, is often hypomethylated in patients with the overgrowth-associated Beckwith–Wiedemann syndrome. Furthermore, Beckwith–Wiedemann syndrome can also result from hypermethylation of the IGF2/H19 imprinted control region causing overexpressing of IGF2, while hypomethylation is associated with the Silver–Russell syndrome, characterized by growth retardation (8). Other potent candidates for Silver–Russell syndrome are the maternally methylated growth factor receptor-bound protein-10 (GRB10) gene, that inhibits growth by interacting with the insulin and IGF1 receptor (11,12), and the maternally imprinted mesoderm-specific transcript (MEST) gene, believed to be involved in adipose tissue expansion (11,13,14).
Besides abnormal growth, some imprinting syndromes are characterized by severe obesity. For instance, imprinting of the maternal and paternal copies of the normally maternally imprinted small nuclear ribonucleoprotein polypeptide N (SNRPN) gene has been observed in Prader–Willi syndrome patients, characterized by insatiable appetite causing morbid obesity (9). Furthermore, (epi)genetic defects at the GNAS complex locus (GNAS), that generates multiple imprinted and nonimprinted transcripts with opposite metabolic effects, can cause the obesity-associated syndrome pseudohypoparathyroidism type-1a (10,15).
It is yet unknown whether differential methylation at these imprinted loci also contributes to nonsyndromic abnormal body weight regulation. To study the contribution of epigenetic modifications to the establishment of complex phenotypes in human, monozygotic (MZ) twins provide a unique opportunity: they are considered genetically identical, thus changes in gene expression should be due to epigenetic phenomena. Moreover, MZ twins are matched for gender, age, and a range of environmental factors. To study the contribution of aberrant methylation at imprinted loci to nonsyndromic overweight, we analyzed DNA methylation in MZ twins discordant for BMI at DMRs most representative for abnormal growth syndromes, including KvDMR1, H19 CTCF4, H19 CTCF6, IGF2 DMR0, IGF2 DMR2, GRB10, MEST, SNRPN, GNAS XL-α-s, and GNAS Exon1A.
Methods and Procedures
Discordant twins (ΔBMI ≥3) were recruited from the East Flanders Prospective Twin Survey (EFPTS) (16). The Ethics Committee of the Faculty of Medicine of the Katholieke Universiteit Leuven approved the project and all participants gave informed consent.
BMI was calculated based on the subject's self-reported body weight and height, except for eight pairs of whom stadiometer height measurements were available. Physical activity levels were assessed using the Baecke questionnaire (17). In addition, twins were asked to compare their physical activity level and lifestyle with their twin brother or sister. Birth weight and gestational age were obtained from obstetric records. Zygosity was confirmed by genotyping 10 highly polymorphic microsatellite markers (accuracy rate >99%).
DNA methylation analysis is described in detail in the Supplementary Data online. Briefly, bisulfite-treated DNA, extracted from saliva, was amplified by PCR, followed by a single-nucleotide primer extension reaction, where a primer annealing next to a CpG is extended by a ddCTP or ddTTP depending on whether the site is methylated or not. Subsequently, single-nucleotide primer extension products were separated on a denaturing high-performance liquid chromatography column and methylation indexes (MIs) were obtained by measuring the peak heights (h) and calculating the ratio MI = h(C)/(h(C) + h(T)).
Statistical analyses were performed using SAS (v9.1, SAS Institute). Paired t-test and Wilcoxon's signed rank test were carried out to determine whether the phenotypic characteristics and MIs differed between the discordant twins. To assess whether intrapair BMI differences were correlated with intrapair differences in MIs, both Pearson's and Spearman's correlations coefficients were calculated. P values <0.05 were considered significant.
Power analysis showed that with 16 twins, 80% power is achieved to detect a ΔMI of 0.03 assuming a standard deviation of 0.04 and significance threshold of 0.05.
Results and Discussion
Since BMI is strongly correlated within MZ twins (3), MZ twins discordant for BMI are scarce. Nevertheless, we had DNA available of 11 female and 5 male discordant MZ twins, with a mean age of 30.9 years (19–44 years) (Table 1). Compared to the lean co-twins, BMI and body weight of the heavy co-twins were on average 4.5 kg/m2 (20%) and 13.6 kg (20%) higher (P < 0.0001), with intrapair differences in BMI and body weight ranging from 2.9–9.5 kg/m2 and 6.5–31.0 kg, respectively. Also minimum and maximum body weight ever was significantly higher in the heavy co-twins (P < 0.0001), but birth weight and adult body height did not differ between the discordant twins (P > 0.05).
Table 1. Phenotypic characteristics of the monozygotic (MZ) twins discordant for BMI
During leisure-time, the heavy co-twins were physically less active, however, this effect was only borderline significant (P = 0.02), and the working, sport, and overall activity indexes did not differ significantly between the discordant twins (P > 0.05) (Table 1). Furthermore, of six pairs both members reported that they have a similar lifestyle and physical activity level. This indicates that the BMI differences in these twins cannot be completely explained by differences in physical activity and lifestyle, suggesting that epigenetic factors might be involved.
The results of the methylation analysis are presented in Table 2. For each DMR analyzed, mean MIs were very similar among the heavy and lean co-twins. Also the mean intrapair differences in MIs between the discordant twins were small (ranging from −0.0006 for H19 CTCF6 to −0.0104 for GNAS Exon1A (CpG1)) and did not differ significantly from zero (P > 0.05) (Table 2). This indicates that methylation levels were not associated with overweight. The analysis was repeated by only including discordant MZ twins of whom both members reported to have a similar physical activity level and lifestyle; again no significant associations were observed (See Table S3 in Supplementary Data online). Also analyzing the data for men and women separately revealed no significant associations (data not shown). In addition, we tested whether intrapair BMI differences were correlated with intrapair differences in MIs, but correlations were insignificant (P > 0.05) (See Table S4 in Supplementary Data online).
Table 2. Results of the methylation analysis at nine differentially methylated regions (DMRs) involved in growth and metabolism in 16 monozygotic (MZ) twins discordant for BMI
Although methylation levels were generally not associated with BMI, a severe hypo- or hypermethylation at a particular DMR could be responsible for the BMI difference within a single pair. The largest ΔMI of −0.096 was observed for IGF2 DMR2 (CpG1) and although this difference is substantial, the ΔMI at the second CpG analyzed within this amplicon in the same twin pair was only 0.0055. In human fibroblast DNA, we also observed considerable differences in MIs of these CpGs and extensive sequencing of cloned bisulfite-PCR products revealed that IGF2 DMR2 displays a mosaic methylation pattern in fibroblasts (18). The current results suggest that IGF2 DMR2 also displays a mosaic methylation pattern in saliva and therefore the methylation difference at IGF2 DMR2 (CpG1) does probably not contribute to the difference in BMI within this twin pair. For the other DMRs, ΔMIs ranged only from −0.055 to 0.051 (Table 2) and it is unlikely that such modest differences explain the discordancy in BMI within a single pair.
Recently, Tobi et al. (19) determined methylation at imprinted and nonimprinted genes implicated in growth and metabolism (including KvDMR1, GRB10, and GNAS Exon1A) in individuals prenatally exposed to famine during the Dutch Hunger Winter, and they observed no significant associations between methylation levels and adult BMI, which is in agreement with our results. However, Netchine et al. (20) observed among Silver–Russell syndrome patients, that patients with H19 DMR hypomethylation had a lower birth weight, length, and postnatal BMI compared to patients with normal H19 methylation (20). Although this supports a role of the H19 DMR in weight regulation, it does not necessarily imply a contribution to the development of nonsyndromic overweight.
To our knowledge, the studies of Tobi et al. (19) and Netchine et al. (20) are the only studies that examined the relation between BMI and methylation at DMRs analyzed in the present study. Hence, thus far, no evidence exists that aberrant methylation at these DMRs contributes to nonsyndromic abnormal body weight regulation in human. However, we want to emphasize that our results should be interpreted cautiously. Although MZ twins are presumed to be genetically identical, we can, however, not exclude somatic mosaicism for e.g., mutations in obesity-related genes, chromosomal aberrations, or copy-number variations that might lead to aberrant growth. Moreover, these imprinted loci are particularly important during early human development and small alterations potentially present in childhood might be balanced during later age. Although we previously showed that DNA methylation levels in saliva are very similar to the levels observed in peripheral blood, skin fibroblasts, and buccal swab DNA (18), saliva DNA might not necessarily reflect the methylation status in e.g., adipose tissue. Taken together, our results indicate that in saliva DNA methylation variability at the nine DMRs analyzed does not seem to contribute to the discordancy in BMI within these MZ twins.
Supplementary material is linked to the online version of the paper at http:www.nature.comoby
The EFPTS has been partly supported by grants from the Fund for Scientific Research Flanders and by Twins, Association for Scientific Research in Multiple Births Belgium. We are grateful to all the twins participating in this study and thank Lut De Zeure for excellent fieldwork.
The authors declared no conflict of interest.