WO2015067952A1

WO2015067952A1 - Phenotype prediction by determining the methylation status of genes

Info

Publication number: WO2015067952A1
Application number: PCT/GB2014/053308
Authority: WO
Inventors: Keith Malcolm GODFREY; Karen Ann LILLYCROP; Joanna Dawn HOLBROOK; Mark Adrian HANSON; Cyrus Cooper
Original assignee: University of Southampton; Agency for Science Technology and Research Singapore
Current assignee: University of Southampton; Agency for Science Technology and Research Singapore
Priority date: 2013-11-06
Filing date: 2014-11-06
Publication date: 2015-05-14
Anticipated expiration: 2016-05-06
Also published as: GB201319576D0

Abstract

Phenotypic characteristics selected from obesity and/or adiposity, bone health, and cardiovascular disease, may be predicted by determining the methylation status of one or more CpG dinucleotides of the, SLC6A4, CDKN2A, FEN1 and BACH1 genes.

Description

PHENOTYPE PREDICTION BY DETERMINING

THE METHYLATION STATUS OF GENES

Field of the Invention

The present invention relates to the use of epigenetic markers as a means for predicting the future development of one or more phenotypic characteristics in an individual. In particular, the present invention relates to a method of predicting the future development of one or more of obesity and adiposity, bone health and cardiovascular disease in an individual.

Background of the Invention

Fixed genomic variations, such as single nucleotide polymorphisms and copy number variations, explain only a fraction of the risk of developing obesity and/or metabolic disease in humans (1-4). There is now substantial evidence from both human and animal studies that the early life environment of an individual, through an epigenetic mechanism, can affect susceptibility to adiposity and metabolic disease in later life. For instance, in humans, an adverse intrauterine environment has been associated with the development of a range of chronic diseases in later life, including obesity, type-2 diabetes, cardiovascular disease, asthma, osteoarthritis and some forms of cancer (5). These findings have also been replicated in a variety of animal models where impaired nutrition, such as a low protein diet, caloric restriction, or even high fat diet during pregnancy, has been shown to induce long term changes in gene expression and metabolism in the offspring, and which persist throughout the life course (6,7).

The mechanism by which the early life environment can induce persistent phenotypic changes has been suggested to involve the epigenetic regulation of genes. Epigenetics is defined as processes that induce heritable changes in gene expression without a change in the DNA sequence. The major epigenetic mechanisms are DNA methylation, histone modification and non-coding RNAs. These processes control genomic imprinting and X chromosomal inactivation, and have also been implicated in the process of developmental plasticity, the mechanism by which an organism adjusts its developmental programme in response to environmental cues in early life, with long term effects on gene expression and phenotype.

Of particular relevance is the methylation of genomic regions called CpG dinucleotides. CpG dinucleotides, or CpG sites, are regions of DNA where a cytosine nucleotide occurs next to a guanine nucleotide. Cytosines in the CpG dinucleotide can be methylated to form 5- methylcytosine. This increase in epigenetic methylation can cause genes influenced by the CpG dinucleotide to become less actively expressed or even silenced, or "turned off". In contrast, a decrease in epigenetic methylation, i.e. hypomethylation, has been shown to be associated with over-expression of certain genes. Some genomic regions have a higher concentration of CpG dinucleotides than others. These regions are known as CpG Islands and are often located within the promoter region of a gene.

Feeding rats a modest reduction in protein during pregnancy induced the hypomethylation of the promoter regions of the nuclear receptors GR and PPARa in the liver of the offspring, which was accompanied by increased GR and PPARa expression and alterations in the metabolic processes that they control (6). In humans, differences in the methylation of a number of imprinted and non-imprinted genes have been found in the peripheral blood of individuals whose mothers were exposed to famine during pregnancy (8), suggesting that, as in the animal studies, DNA methylation changes induced by environmental cues in early life are largely maintained throughout the life-course.

Although DNA methylation patterns are often tissue-specific, recent studies have indicated that epigenetic traits in tissues such as cord, blood, and buccal tissues, may provide useful proxy markers of future disease risk in more disease-relevant cell types. For instance Godfrey et al., have shown that methylation of a CpG within the promoter of the RXRA gene in umbilical cord at birth is associated with childhood adiposity at 6 and 9 years in two independent cohorts (9).

The CDKN2A gene locus encodes two potent inhibitors of cell growth, p16^INK4a, which is an inhibitor of cyclin-dependent kinase 4, and p14^ARF (alternative reading frame relative to p16). Both inhibitors are involved in cell cycle regulation. p14^ARF induces cell cycle arrest, by interacting with MDM2 and promoting its degradation, thereby preventing MDM2 from carrying out its usual function of destabilising p53. The resultant increase in p53 levels triggers p53-mediated cell cycle arrest at both G1 and G2/M phases. CDKN2A is located in a region adjacent to the CDKN2B gene, encoding a cyclin-dependent kinase inhibitor known as p15^INK4b. The encoded protein functions as a cell growth regulator. Both p16^INK4a and p15^INK4b are well established tumour suppressors that play key roles in cell proliferation and differentiation by binding specifically to CDK4 and CDK6, blocking cell proliferation. This shared region is frequently mutated in a wide variety of tumours.

The CDKN2A gene locus also shares a divergent promoter region with CDKN2B-AS, also known as ANRIL (antisense non-coding RNA in the INK4 locus) (10). ANRIL is a 3.8k transcript which is thought to be involved in transcriptional repression. ANRIL negatively regulates both p16^INK4a and p15^INK4b. Single nucleotide polymorphisms within this region are associated with many types of cancers. The genes within the CDKN2A locus have also been shown to be important for a number of other cellular processes including cellular differentiation, ageing and in driving cellular senescence through irreversible cell cycle arrest (1 1-22).

The SLC6A4 (solute carrier family 6) gene encodes an integral membrane protein that functions as a serotonin transporter, transporting serotonin from the synaptic spaces into presynaptic neurons. This gene has previously been implicated in conditions such as Alzheimer's disease and depression.

The BACH1 gene encodes a transcription factor that belongs to the cap'n'collar type of basic region leucine zipper factor family. The encoded protein is thought to play a role in transcription activation and repression.

The FEN1 gene is located between the fatty acid desaturase 1 and 2 (FADS1 and FADS2) genes on chromosome 1 1 and encodes the enzyme Flap endonuclease 1. This enzyme is involved in DNA synthesis and repair.

With the escalating prevalence of lifestyle related disorders worldwide, such as obesity, particularly in early childhood and cardiovascular disease, the identification of an improved marker for determining whether or not an individual will develop such undesirable characteristics would be extremely valuable in order to assess individuals at risk from non- communicable disease, and would provide evidence for the role of early life environment within the normal range rather than extremes of environment in the development of obesity.

The current invention helps to solve this problem by providing a method for predicting development of an undesirable phenotypic characteristic in an individual, in particular, one or more of obesity, bone health and/or cardiovascular disease, and may be of assistance in identifying external factors relevant to the incidence of a particular characteristic in an individual, or in a population in general. In this way, information may be obtained which could be used to reduce the incidence of the condition and any associated metabolic disorders. Once the susceptibility for a particular characteristic has been determined in an individual, the lifestyle of that individual may be managed accordingly, in order to reduce the risk of the actual occurrence of the characteristic. Summary of the Invention

Thus, in a first aspect, the invention provides a method of predicting the development of one or more phenotypic characteristics in an individual, wherein the phenotypic characteristic is selected from one or more of:

(i) obesity and/or adiposity;

(ii) bone mineral content and/or bone health, and

(iii) cardiovascular disease,

wherein the method comprises determining the methylation status of one or more CpG dinucleotides of a gene in a sample, wherein said gene is selected from the group comprising SLC6A4, CDKN2A, FEN1 and BACH1 or a combination thereof, and wherein the methylation status of the one or more CpG dinucleotides provides an indication of the development of said phenotypic characteristic, or characteristics, in said individual.

In the alternative, or in addition, said method comprises detecting ANRIL, 14^ARF or p16^INK4a transcript, or a combination thereof, in a sample, wherein the presence/absence of ANRIL transcript is indicative of the development of said phenotypic characteristic.

Preferably, the level of gene expression of ANRIL is detected.

Preferably, the level of gene expression of 14^ARF and/or p16^INK4a is detected.

Advantageously, the methylation status is the level, or amount, of methylation, or the presence or absence of methylation. It may be determined by pyrosequencing, for example, as described hereinbelow, or by any suitable means known in the art. For example, the degree of DNA methylation may be also detected by, but not limited to, bisulfite sequencing, single molecular real time sequencing, methylation-sensitive PCR amplification, sequenom, bisulfite conversion followed by primer extension, or a combination thereof.

Preferably, the one or more phenotypic characteristic is one or more of obesity, adiposity, total body fat mass, percentage body fat mass and/or body fat distribution, and the one or more CpG dinucleotide is selected from the group comprising CpG5, CpG1 and CpG2 of the SLC6A4 gene, CpG1 , CpG2, CpG3, CpG4, CpG5, CpG6, CpG7, CpG8, and CpG9 of the CDKN2A gene, CpG1 and 3 of the FEN 1 gene and CpG1 , CpG4, CpG5, CpG6 and CpG8 of the BACH1 gene. The genomic coordinates of these sites is provided hereinbelow, and was determined through the UCSC Genome Bioinformatics Group home page, at http://genome.ucsc.edu. More preferably, the one or more phenotypic characteristic is one or more of obesity, adiposity, total body fat mass, percentage body fat mass and/or body fat distribution, and wherein the one or more CpG dinucleotide is selected from the group comprising CpG5, CpG1 and CpG2 of the SLC6A4 gene, preferably, CpG5 of the SLC6A4 gene.

Also preferred, the one or more phenotypic characteristic is one or more of obesity, adiposity, total body fat mass, percentage body fat mass and/or body fat distribution and the one or more CpG dinucleotide is selected from the group comprising CpG1 , CpG2, CpG3, CpG4, CpG5, CpG6, CpG7, CpG8, and CpG9 of the CDKN2A gene locus.

Also preferred, the one or more phenotypic characteristic is one or more of obesity, adiposity, total body fat mass, percentage body fat mass and/or body fat distribution, and, wherein the one or more CpG dinucleotide is selected from the group comprising CpG1 and 3 of the FEN1 gene.

Alternatively, the one or more phenotypic characteristic is one or more of obesity, adiposity, total body fat mass, percentage body fat mass and/or body fat distribution, and, wherein the one or more CpG dinucleotide is CpG1 , CpG4, CpG5, CpG6 and CpG8 of the BACH1 gene.

Preferably, wherein the one or more condition is bone mineral content /or bone health and the one or more CpG dinucleotides are selected from the group consisting of CpG4, CpG5, CpG6, CpG7, CpG8, and CpG9 of the CDKN2A gene, CpG1 of the SLC6A4 gene, and CpG1 , CpG3, CpG4, CpG5, CpG7, and CpG8 of the BACH1 gene.

Preferably, wherein the one or more condition is cardiovascular disease and the one or more CpG dinucleotides are selected from the group consisting of CpG5 of the CDKN2A gene, CpG1 and CpG5 of the SLC6A4 gene, and CpG6 of the BACH1 gene.

Preferably, the method comprises determining the degree of methylation in genomic DNA from a tissue sample obtained from a patient.

Advantageously, the tissue sample is umbilical cord or another perinatal tissue sample including cord blood, placenta, hair, saliva, and buccal smear.

Alternatively, the tissue sample may be adipose tissue, for example. Preferably, the method of the present invention comprises comparing methylation status with the methylation status of an individual with said phenotypic characteristic at a selected older age.

Preferably, the method of the present invention comprises comparing the methylation status with the methylation status of an individual without said phenotypic characteristic at a selected older age.

A further aspect of the present invention provides the use of the methylation status of one or more SLC6A4 CpG dinucleotides for predicting the development one or more phenotypic characteristics in an individual, wherein the phenotypic characteristic is selected from the following:

(i) one or more of obesity, adiposity, total body fat mass, percentage body fat mass and/or body fat distribution;

(ii) bone mineral content and/or bone health;

(iii) cardiovascular disease.

Preferably, the one or more phenotypic characteristic is one or more of obesity, adiposity, total body fat mass, percentage body fat mass and/or body fat distribution, and wherein the one or more CpG dinucleotide is selected from the group comprising CpG5, CpG1 and CpG2 of the SLC6A4 gene, preferably, CpG5 of the SLC6A4 gene.

An alternative aspect of the present invention provides the use of the methylation status of one or more CDK2NA CpG dinucleotides for predicting the development one or more phenotypic characteristics in an individual, wherein the phenotypic characteristic is selected from the following:

(ii) bone mineral content and/or bone health;

(iii) cardiovascular disease.

According to a further aspect, the present invention provides the use of the methylation status of one or more FEN1 CpG dinucleotides for predicting the development one or more phenotypic characteristics in an individual, wherein the phenotypic characteristic is selected from one or more of obesity, adiposity, total body fat mass, percentage body fat mass and/or body fat distribution. According to a further aspect, the present invention provides the use of the methylation status of one or more BACH1 CpG dinucleotides for predicting the development one or more phenotypic characteristics in an individual, wherein the phenotypic characteristic is selected from the following:

(ii) bone mineral content and/or bone health;

(iii) cardiovascular disease.

According to a further aspect, the present invention provides the detection of ANRIL transcript for predicting the development one or more phenotypic characteristics in an individual, wherein the phenotypic characteristic is selected from the following:

(ii) bone mineral content and/or bone health;

(iii) cardiovascular disease.

Advantageously, said method comprises detecting ANRIL transcript in a sample, wherein the presence/absence of ANRIL transcript is indicative of the development of said phenotypic characteristic.

Preferably, said method comprises detecting ANRIL transcript in a sample, wherein the level of expression of ANRIL is indicative of the development of said phenotypic characteristic.

Preferably, the one or more phenotypic characteristic is adiposity.

According to a further aspect, the present invention provides the detection of p14^ARF transcript for predicting the development one or more phenotypic characteristics in an individual, wherein the phenotypic characteristic is selected from the following:

(ii) bone mineral content and/or bone health;

(iii) cardiovascular disease Advantageously, said method comprises detecting p14 transcript in a sample, wherein the presence/absence of p14^ARF transcript is indicative of the development of said phenotypic characteristic or characteristics, in said individual.

Preferably, said method comprises detecting p14^ARF transcript in a sample, wherein the level of expression of p14^ARF is indicative of the development of said phenotypic characteristic or characteristics, in said individual.

Preferably, the one or more phenotypic characteristic is adiposity.

According to a further aspect, the present invention provides the detection of p16^INK4a transcript for predicting the development one or more phenotypic characteristics in an individual, wherein the phenotypic characteristic is selected from the following:

(ii) bone mineral content and/or bone health;

(iii) cardiovascular disease.

Advantageously, said method comprises detecting p16 transcript in a sample, wherein the presence/absence of p16^INK4a transcript is indicative of the development of said phenotypic characteristic, or characteristics, in said individual.

Preferably, said method comprises detecting p16^INK4a transcript in a sample, wherein the level of expression of p16^INK4a is indicative of the development of said phenotypic characteristic, or characteristics, in said individual.

Preferably, the one or more phenotypic characteristic is adiposity.

Preferably, the method comprises comparing the level/presence/absence, with the level/presence/absence in an individual with said phenotypic characteristic at a selected older age. Preferably, the method comprises comparing the level/presence/absence with the level/presence/absence in an individual without said phenotypic characteristic at a selected older age. In a further aspect, the invention provides a method of predicting the development of one or more phenotypic characteristics in an individual, wherein the phenotypic characteristic is selected from one or more of:

(i) obesity and/or adiposity;

wherein the method comprises determining the expression of one or more genes in a sample, wherein said gene is selected from the group comprising ANRIL, p14^ARF and p16^INK4a or a combination thereof, and wherein the expression, or the level of expression, of the one or more genes provides an indication of the development of said phenotypic characteristic, or characteristics, in said individual.

Preferably, the method comprises comparing expression with the expression in an individual with said phenotypic characteristic at a selected older age. Preferably, the method comprises comparing the expression with the expression in an individual without said phenotypic characteristic at a selected older age.

Preferably, the method comprises determining the expression from a tissue sample obtained from a patient.

Alternatively, the tissue sample may be adipose tissue, for example.

A further aspect of the invention provides a method of managing one or more of obesity, adiposity, total body fat mass, percentage body fat mass and/or body fat distribution, bone mineral content and/or cardiovascular disease in an individual comprising the following:

(i) determining the methylation status of one or more CpG dinucleotides of a gene in a sample, wherein said gene is selected from the group comprising SLC6A4, CDKN2A, FEN1 and BACH1 or a combination thereof, according to the method of the present invention

(ii) managing the diet, behaviour or physical activity of the individual, wherein the methylation status indicates the development of said characteristic in the individual. The CpG sites of the present invention are listed in Table 1 , below. The sites are provided as coordinates in relation to the hg19 build of the human genome sequence. The relationship to selected phenotypic characteristics of each methylated CpG site is described in the right three columns, with 'inverse' indicating that the methylated site is associated with a negative trend, such as increased likelihood of obesity.

Table 1

It will be appreciated that it is generally more practical to assay for the presence of methylation on a selected gene, or selected sites of a gene, rather than on all of the genes relevant to the present invention. Accordingly, it is preferred to assay for methylation of CpG sites in one of CDKN2A, SLC6A4, BACH1 , and FEN1. It will also be appreciated that assaying for methylation of each, or a selection of, the genes will provide a more comprehensive prognosis of the possible development of undesirable phenotypic characteristics. In such cases, it will generally be preferred to analyse or assay the CpG sites of each gene separately, and to subsequently consider the results in the round.

From Table 1 , it can be seen that the CpG sites of the invention are not all associated each with the three adverse indication columns, with only CpG5 being associated, either positively or negatively, with all three. Accordingly, the methylation state of each site provides an incremental indication of the likelihood of the development of a phenotypic condition, so that if all of the sites labelled 'inverse' under obesity are found to be methylated, then the chances of the individual having weight problems later in life are substantially higher than if none is methylated, if no action is taken to compensate the possibility.

Where neither 'positive' nor 'inverse' is indicated, then the methylation state of the CpG site in question has no statistical correlation with the column in question.

Brief Description of the Figures

Figure 1 illustrates differential methylation at the CDKN2A gene; Figure 2 illustrates differential methylation at the SLC6A4 gene;

Figure 3 illustrates the correlation between perinatal DNA methylation for CDKN2A and SLC6A4 and DXA %fat age 6 years;

Figure 4 displays the results of luciferase and EMSA assays which examine the role of the CDKN2A region of interest in gene expression;

Figure 5 illustrates differential methylation at the BACH1 gene;

Figure 6 illustrates differential methylation at the FEN1 gene;

Figure 7 displays a graph showing the correlation between perinatal DNA methylation for BACH 1 and FEN1 and DXA %fat age 6 years;

Figure 8 illustrates the ponderal index at age 18 months in the GUSTO cohort according to umbilical cord methylation of CDKN2A CpG1 methylation; Figure 9 A - D display graphs showing p14ARF expression in (A) SW-873 (B) Saos-2 (C) A549 and (D) JAR cell lines;

Figure 9E displays the results of electrophoretic mobility shift assays showing binding to the CpG sites within the CDKN2A DMR.

Figure 10 illustrates the association between SLC6A4 CpG5 methylation total and percentage fatmass at 4 and 6 years in SWS cohort. Methylation has been divided into 4 equal groups according to rank; means and standard error of the mean are plotted for each group.

Figure 11 illustrates the association between SLC6A4 CpG5 methylation and tricep skinfold thickness from birth to 6 years in the SWS cohort. Methylation has been divided into 4 equal groups according to rank; means and standard error of the mean are plotted for each group.

Figure 12A illustrates the association between SLC6A4 CpG5 methylation and total fatmass at 4 years in the SWS cohort. Methylation has been divided into 4 equal groups according to rank; means and standard error of the mean are plotted for each group.

Figure 12B illustrates the association between SLC6A4 CpG5 methylation and child's tricep skinfold thickness at birth in the SWS cohort. Methylation has been divided into 4 equal groups according to rank; means and standard error of the mean are plotted for each group.

Figure 13A compares SLC6A4 CpG 5 methylation in adipose tissue from obese individuals and with lean individuals from the BIOCLAIMS cohort. Methylation from both lean and obese individuals is shown for CpG 5 (n=22). Whiskers show min to max. * p 0.01-0.05, ** p ≤0.01.

Figure 13B illustrates SLC6A4 transcript expression in adipose tissue from obese individuals and from lean individuals from an adult cohort. Whiskers show min to max. (n=63) * p 0.01- 0.05, ** p≤0.01.

Figure 14 illustrates the top two pathways enriched amongst the DMRs; (A) DNA replication, recombination and repair, and (B) cell death and survival, gene expression, and embryonic survival. The DMRs validated by pyrosequencing are indicated by the arrows. Figure 15A illustrates the concordance of methylation values with percent body fat for CDKN2A. The X-axis shows percent body fat and y-axis shows % methylation as estimated by BATMAN. Sample data points are shaded by percent body fat group groups (dark = low, lowest percent body fat; medium = medium, medium percent body fat; light = high, highest percent body fat).

Figure 15B illustrates the correlation between DNA methylation for CDKN2A CpG1 (β=-0.004; p=0.003), CpG3 (β=-0.005; p=0.007) and CpG6 (β=-0.004; p=0.01), and DXA %fat age 6 years.

Figure 15C illustrates the correlation between DNA methylation for CDKN2A CpG1 (β=-0.09; p=0.007), CpG3 (β=-0.07; p=0.1), CpG4 (β=-0.09; p=0.007) and CpG6 (β=-0.06; p=0.01), and DXA %fat age 4 years.

Figure 15E illustrates the association between CDKN2A CpG1 (p=0.015), CpG2 (p=0.047) and CpG3 (p=0.035) methylation at birth and Subscapular skinfold thickness at age 2 years in the GUSTO cohort (dark shading = female; light shading = male).

Figures 16A, B and D illustrate the concordance of methylation values with percent body fat for SLC6A4 (A), FEN1 (B) and BACH1 (D). The X-axis shows percent body fat and y-axis shows % methylation as estimated by BATMAN. Sample data points are shaded by percent body fat group groups (dark = low, lowest percent body fat; medium = medium, medium percent body fat; light = high, highest percent body fat).

Figure 16C illustrates the validation of the association observed between the DMR methylation of BACH1 at birth and % fat mass at 6 years of age (β=-0.009; p=0.038). Methylation has been divided into four equal groups according to rank; means and standard errors are plotted for each group.

Detailed Description

The current invention is based upon the surprising finding that the degree of methylation of particular CpG sites within the CDK2NA gene at birth correlate with the propensity of an individual to develop one or more of obesity, low bone mineral content and poor bone health and cardiovascular disease in later childhood. We have also discovered associations of the degree of methylation of the CpG loci within SLC6A4 and BACH1 with the propensity of an individual to develop one or more of obesity, low bone mineral content and poor bone health and cardiovascular disease, and of the degree of methylation of the CpG loci within FEN1 with the propensity of an individual to develop later obesity, thereby demonstrating that the early life environment is an important determinant of adiposity, bone health and the risk of cardiovascular disease, and may influence future health through the altered epigenetic regulation of gene expression.

In its broadest aspect, the present invention provides a method of predicting one or more of the following conditions:

(i) obesity or adiposity;

(ii) bone health; and

(iii) cardiovascular disease.

The method comprises determining the degree of methylation of one or more CpG dinucleotides of a selected gene in a sample, wherein the degree of methylation correlates with propensity for said condition. In a preferred embodiment, the condition is obesity, total body fat mass, percentage body fat mass and/or body fat distribution. In another preferred embodiment, the condition is bone mineral content and/or poor bone health. In a further preferred embodiment, the condition is cardiovascular disease.

A method also comprises detecting the transcript, or determining the expression, of a selected gene in a sample, wherein the presence or absence of the transcript, or the level of expression, correlates with propensity for said condition. In a preferred embodiment, the condition is obesity, adiposity total body fat mass, percentage body fat mass and/or body fat distribution. In another preferred embodiment, the condition is bone mineral content and/or poor bone health. In a further preferred embodiment, the condition is cardiovascular disease. In a still preferred embodiment the gene is ANRIL, p14^ARF and/or p16^INK4A

It will be appreciated that such a method of prediction is preferably applied to a neonate or young child, typically under the age of 12 months, and preferably which does not already exhibit one of, or the, phenotypic factor(s) being prognosticated. Samples are typically collected in the perinatal period, but the test may be done at a later age. If the test is performed later, the sample will typically be preserved, such as by freezing.

Where the sample used for DNA analysis is umbilical cord, or another perinatal tissue sample, it is preferred that the method further comprises a step of comparing the degree of methylation with the degree of methylation of variance of the condition in an individual at a selected older age. The current method may be used to manage propensity for the occurrence of undesirable phenotypic characteristics in humans. For example, such use of epigenetic markers enables the targeting of corrective strategies where propensity for obesity is identified.

The tissue to be used for determination is desirably a tissue readily available in early life, and may be obtained from a neonate, infant, child or young adult. Desirably, the tissue is a perinatal tissue sample containing genomic DNA of the individual, taken at, or soon after, birth, typically within 12 months of parturition. Preferably, the tissue may be umbilical cord, adipose tissue, blood (including umbilical cord blood), placenta, chorionic villus biopsy, amniotic fluid, hair, or buccal smears. Preferably, such tissue will be from a newborn or taken from an infant within a few weeks of birth, more preferably within a few days of birth (less than a week), although samples taken later, e.g. within 6 months, or longer, may prove useful.

The method of the present invention may be used to provide a targeted intervention to modify the effect of the condition on health is possible.

Obesity can be associated with type 2 diabetes, metabolic syndrome, and other conditions for which obesity is recognised to be a risk factor.

Bone mineral content prediction is of interest in identifying individuals who may be susceptible to impaired linear growth, and disease states linked with low bone mineral content such as osteoporosis in later life.

Once the propensity for exhibiting phenotypes has been determined in an individual, management of that individual's lifestyle (diet, behaviour, exercise, etc.) can be undertaken to reduce the risk of the actual occurrence of the undesirable characteristic such as obesity, low mineral content, or cardiovascular disease.

The methods of the present invention may be used to identify external factors that may contribute to the incidence of the characteristics in the individual or population in general. In this way, information may be obtained useful in directing initiatives aimed at reducing the incidence of the condition and thereby the associated disorders. The utility of this invention is that it allows selection of appropriate ways of managing an individual for optimal health and wellbeing. The method of the present invention is important, as obesity is also associated with metabolic disorders in later life, e.g. diabetes.

The accompanying Examples are for illustration only, and do not restrict the scope of the present invention.

EXAMPLES

EXAMPLE 1 Materials and Methods

Southampton Women's Survey: Participants

The Southampton Women's Survey (SWS) is a prospective cohort study of a large group of non-pregnant women aged 20 to 34 years living in the city of Southampton, UK. Comprehensive details of the study have been published previously (23). Women were recruited through General Practices across the city between April 1998 and December 2002. Each woman was invited to take part by letter, followed by a telephone call when an interview date was arranged; 12,583 women agreed to take part, 75 % of all women contacted. Trained research nurses visited the women at home and collected information about their health, diet and lifestyles, as well as taking anthropometric measurements. Women who subsequently became pregnant were followed up at 1 1 , 19 and 34 weeks gestation and their offspring were studied in infancy and childhood. Amongst women who became pregnant, smoking status in pregnancy was ascertained at the 11 and 34 week interviews. A total of 1981 women became pregnant and delivered a live-born singleton infant before the end of 2003. Six infants died in the neonatal period and two had major congenital growth abnormalities, which left 1973 mother-offspring pairs. Table 2 outlines participant characteristics.

Table 2

Characteristic group/number % or Median (InterQ range)

Qualification level none 0.9%

CSE 9.9%

O levels 27.0%

A levels 26.4%

HND 9.1 %

Degree 27.6% Social class Professional 6.4%

Management/technical 42.2%

Skilled non-manual 33.2%

Skilled manual 6.9%

Partly skilled 10.4%

Unskilled 0.9%

Current smoker No 77.0%

Yes 23.0%

Pregnancy First pregnancy 47.7%

Pregnant before 52.3%

Sex of children Male 52.8%

Female 47.2%

Birth order 47.7%

38.6%

3^rd or higher 13.6%

Woman's age at birth of child 352 30.7 (28.1 to 33.4)

Early pregnancy women's

279 24.8 (22.7 to 28.2) BMI

Birthweight (kg) 349 3.4 (3.2 to 3.8)

Gestation (wks) 352 40.0 (39.0 to 40.9)

Infant total fat (g) at birth 147 497.9 (364.3 to 644.3)

Offspring total fat (kg) at 4

283 4.0 (3.4 to 4.8) years

Offspring total fat (kg) at 6

264 4.5 (3.6 to 6.0) years

Infant percentage fat at birth 147 14.3 (1 1.6 to 17.3)

Offspring percentage fat at 4

283 27.8 (24.3 to 32.5) years

Offspring percentage fat at 6

263 24.2 (19.9 to 29.7) years

Height (cm) at 4 years 350 103.9 (101.2 to 106.9)

Height (cm) at 6 years 282 120 (116.7 to 124.3)

SWS Mother characteristics Body composition measurements in the SWS offspring

A sub-set of children from the SWS study with dietary data in infancy, who were born before 2003, were invited to take part in follow-up at age 4 and 6 years to assess their body composition. Adiposity measurements were made by dual-energy X-ray absorptiometry (Hologic Discovery, paediatric scan mode, Hologic Inc., Bedford, MA)(24).

Instruments were calibrated daily; coefficients of variation were <1 %. Follow-up of the children and umbilical cord sample collection/analysis was carried out under Institutional Review Board approval (Southampton and SW Hampshire Research Ethics Committee) with written informed consent. Clinical investigations were conducted according to the principles expressed in the Declaration of Helsinki. Umbilical cord is only collected in the immediate perinatal period; after collection it is typically frozen for later processing.

Growing Up in Singapore Towards Healthy Outcome (GUSTO): Participants

Umbilical cord specimens were collected from babies born at the KK Women's and Children's Hospital (KKH) and the National University Hospital (NUH), in Singapore. These hospitals are part of the GUSTO birth cohort study (25). Written parental consent to participate in the study was given and hard copies are stored by the GUSTO data team. Ethical approval for the study and the consent forms and contents was granted, by the ethics boards of both KKH and NUH, which are centralized Institute Review Board (CIRB) and Domain Specific Review Board (DSRB), respectively. We screened 3751 families and 2034 met eligibility criteria. Of the 1247 women (response rate 61.3 %) recruited, 1162 conceived naturally, while 85 conceived through in vitro fertilisation (IVF). At baseline, 55.9 % were Chinese, 26.1 % Malay and 18 % Indian. Mean maternal age at recruitment was 30.6 years (range: 18 to 46). Gestational age was defined from a dating ultrasound (10-12 weeks) followed by an additional scan at 18-22 weeks. The average birth weight in the GUSTO cohort was 3081 g, which is comparable to the average across a larger Singaporean sample of 3183 g for a term infant (unpublished data). Umbilical cord tissue samples were collected at the time of delivery, flushed with saline to remove foetal blood and flash-frozen in liquid nitrogen within 30 min of collection.

At age 18 months weight was recorded to the nearest gram using a calibrated scale (SECA- 334, SECA Corp. Hamburg) and recumbent crown-heel length measured to the nearest 0.1 cm using a SECA-210 Mobile Measuring Mat. For reliability, measurements were taken in duplicate. Ponderal index [weight/(length)³] was derived as a measure of adiposity. Table 3 outlines the characteristics of the children that participated in this study.

Table 3 GUSTO subject characteristics

The analysis in this table is based on those seen at age 18 months who had an umbilical cord sample available

Whole genome methylation analysis

Genomic DNA from umbilical cord samples was obtained from twenty-one children from the SWS cohort, chosen to represent a range of % fat values at 6 years (as measured by DXA scan) between the 5th and the 95th percentiles. The DNA from each subject was sheared by sonication to produce fragments of between 200 and 500bp. A proportion of this DNA (input DNA) was purified and the remaining DNA incubated with a His-tagged MBD2b (methyl binding domain of MeCP2) protein to form a complex which was then captured on nickel coated magnetic beads (MethylCollector kit, Active Motif). Beads were washed to remove unmethylated DNA fragments before the DNA was eluted using an elution buffer designed to simultaneously elute methylated DNA while degrading the MBD2b protein (26). To ensure the purity of the methylated DNA fraction, real-time PCR was carried out using primers specific to promoter regions of β-actin, which is unmethylated and the H19 Imprinting control region (H19ICR) which is known to be methylated.

Methylated and input samples were then hybridised to the Agilent Human Promoter Whole- Genome ChlP-on-chip array (G4489A)(27). The Agilent Human Promoter Whole Genome Chip on chip array (G4489A) contains probes (60mers) spaced every 100-300bp across the promoter regions of the best-defined human Refseq gene promoters (21 ,000) from -8kb to +2kb downstream of the TSS. Microarray hybridisation was carried out by Oxford Gene Technology (OGT, Oxford UK) in accordance with the company's quality control procedures using standard protocols for labelling, hybridisation and washing. Microarray slides were scanned at 5μΜ resolution using the extended dynamic range (High 100 %, Low 10 %). The slides were then feature-extracted using Agilent Feature Extraction Software (v.9.5.3.1). All arrays were normalised per spot and per chip using an intensity dependent normalisation (Lowess normalisation) using Genespring.

Data analysis

The log2 of Cy3/Cy5 values were obtained for each probe after background subtraction. The log2 ratios were then passed through Bayesian Tool for Methylation Analysis (BATMAN)(28). BATMAN is an algorithm that generates Bayesian distributions of % methylation values for every 100 nucleotide region covered on the array. Log2 ratios of the overlapping probes and CpG densities in the probe of 100nt of flanking genomic sequence are inputted to the likely beta value distributions. The mode of the distribution for each 100 nt region returned by BATMAN was used for further analysis. Sample quality control was performed by examining the frequency distribution of the BATMAN output as well as the raw log2 ratios, as suggested by (26). Sample quality control was performed by examining the frequency distribution of the BATMAN output as well as the raw log2 ratios, as suggested by (29) Identification of differentially methylated regions (DMRs)

Statistical tests to identify Differentially Methylated Regions

Percentage methylation values at each 100 nt region were subjected to robust regression analysis percent fat to correct for heteroscedasticity (30). Fisher Exact tests were performed to identify larger chromosomal regions (differentially methylated regions, DMRs) that tiled sequentially on the array. These regions were tested for significant enrichment of differential methylation amongst the 100nt regions, within their span. This principal is similar to that of (29) but is especially designed for the custom array design.

Test for robustness to failed probe signals

It was observed that a portion of the probes included in the raw data had a log2 value approaching 0. These probes were assumed to be failed probes and were removed from the data set and the modified dataset was subjected to the BATMAN algorithm. Any DMRs that were not robust between the two data sets were removed from consideration. Finally, to enrich for true positives and to identify biological pathways implicated in developmental programming of obesity, pathway enrichment analysis and de novo network discovery were performed using the Ingenuity network analysis suite.

Pyrosequencing

The levels of methylation of individual CpG dinucleotides in DMRs identified by BATMAN analysis were measured by sodium bisulfite pyrosequencing. Genomic DNA was prepared and bisulphite conversion was carried out using the EZ DNA methylation kit (ZymoResearch, Irvine, California, USA). The pyrosequencing reaction was carried out using primers listed in Table 4. Modified DNA was amplified using Hotstart Plus DNA polymerase (QIAGEN). PCR products were immobilised on streptavidin-sepharose beads (GE Healthcare UK Ltd., Amersham, Buckinghamshire, UK), washed, denatured and released into annealing buffer containing the sequencing primers. Pyrosequencing was carried out using the SQA kit on a PSQ 96MA machine (Biotage, Uppsala, Sweden) and the % methylation was calculated using the Pyro Q CpG software (Biotage). Within assay precision was between 0.8 and 1.8 % and detection limits were 2 - 5 % methylation. It will be appreciated that any suitable primer may be used to target CpG sites, provided that it binds the sequence with sufficient avidity under the experimental protocol used. Table 4: Primers

Pvrosequencing Primer Sequence SEQ ID NO.

CDKN2A 1-3 F AGTAGGAAAGGTGTATTTTAAGTATATTT 1

CDKN2A 1-3 S AGAATTATTGTTAATTATTTAAGTT 2

CDKN2A 4-9 F TGGGGAGAATTATTGTTAATTATTTAAGTT 3

CDKN2A 4-9 S TAGGAGAGTGGAGGA 4

CDKN2A 8-9 S GTAGGTAGAGATTTTTTGAAATGT 5

CDKN2A 1-3 R Bio TATCTCACCAATCCTCCACTCTCCTAAA 6

CDKN2A 4-9 R Bio AAAAACCCATTTCCCTATTAACTACA 7

APOL5 F AGTAATGGAATTGGTAGTTTTTAGAT 8

APOL5 P TGTTTTATTGTGTTTTTTTATTTAG 9

BACH 1 R1 ATCCCACCCCCAAAAAATACTA 10

BACH1 P1 ACCTAAAATTTAAACTTCATCC 1 1

BACH 1 F2 GGGATGAAGTTTAAATTTTAGGTGTTGTG 12

BACH1 P2 TGTATGAGGTTTTTTTTATATTGA 13

BACH F3 TTTTTTGGGGGTGGGATTTAGGAA 14

BACH P3.1 TTTTTTTTTTTTTTGAGATAGAGT 15

BACH P3.2 GATTTTAGTTTATTGTAA 16

BACH P3.3 GATTTTTTTGTGTTAGT 17

BACH P3.4 TGTTTGGTTAATTTTTTGTATTTTA 18

BACH P3.5 ATTATGTTGGGGTTTAGT 19

APOL5 R Bio CCCAAAATACAACAATAAAATACTATATTC 20

BACH 1 R2 Bio ATTCCTAAATCCCACCCCCAAAA 21

BACH 1 F1 Bio GTGGTTTTGGTTTAAAGTAAGAGAAT 22

BACH R3 Bio CATACCTATAATCCCAACACTTTCA 23

SLC6A4 F GGGATTTGTGTTTAATGAAATAGAGAATTG 24

SLC6A4 R ACCCTAAAACCTCAATAACTTCACAC 25

SLC6A4 S TGTTTAATGAAATAGAGAATTGT 26

FEN1 F GAGAAAATAAATTTTAGTGGTTTGGTAATG 27

FEN1 S GGGATGGATTTGGAAG 28

FEN1 R Bio TATTCTAATTCCCAACCTAACCCTACA 29

Realtime primer Sequence SEQ ID NO.

ANRIL unspliced F CAGTGGCTTCCTGTTCATGC 30 ANRIL unspliced R GGGCTTGACGTCTGATCTGT 31

ANRIL F1 TGAAAAACACACATCAAAGGAG 32

ANRIL R1 GATTCCACCACACCTAACAG 33

P14ARF F1 CCCTCGTGCTGATGCTACTGAG 34

P14ARF R1 GTGAGAGTGGCGGGGTCGGC 35

P16INK4a F1 CGAATAGTTACGGTCGGAGGC 36

P16INK4a R1 TCGGGTGAGAGTGGCGGGGTC 37

Primer sequence SEQ ID NO.

ANRIL F1 TTGGGTACCACCTCTAACTCACAAAGAAAGC 38

ANRIL R1 GCCAAGCTTTGGGAATGACTAAGACACAC 39

ANRIL F2 GAGAAGCTTCCCAGGATATTCGGGACTCA 40

ANRIL R2 TTACCATGGTGTCAGGTGACGGATGTAGC 41

Cloning of promoter regions

PCR was used to amplify the region containing the ANRIL promoter, and the region identified by the BATMAN analysis. Two sets of PCR primers were used, the first set amplified the ANRIL promoter (-926 to +20 relative to ANRIL TSS). A Hindlll restriction site was added to the forward primer, and an Ncol restriction site to the reverse primer for cloning into pGL Basic (Promega, UK) to create pGL -951. The second PCR primer pair amplified a region of interest (ROI), immediately adjacent to the already cloned ANRIL promoter region (-1281 to -925). A Kpnl restriction site was added to the forward primer, and a Hindlll restriction site to the reverse primer for cloning into pGL -951. The completed plasmid, pGL ANRIL, contained the full genomic sequence from -1281 to +20 of the ANRIL promoter with a Hindlll site inserted after the ROI to allow subsequent patch methylation. All PCR amplification was carried out using Hot Star High Fidelity DNA polymerase (QIAGEN). Primers are listed in Table 4 above. The base pair sequence of the cloned region was confirmed by sequencing (GATC, Germany).

Patch Methylation

A 351 bp fragment, containing the nine CpG dinucleotides identified in the BATMAN analysis was excised from pGL ANRIL by restriction digest with Kpnl and Hindlll (Thermo Scientific). The fragment was treated with 8 units of Sssl methylase (NEB UK) for 4 hours at 37°C. An additional 4 units of Sssl, along with further SAM, was then added and the reaction incubated at 37°C overnight, followed by heat inactivation. Complete methylation of the fragment was confirmed by restriction digest using the methylation sensitive restriction enzymes Acil and Hpall. A mock reaction was carried out in parallel, identical except for addition of water in place of Sssl.

The methylated fragment was then re-ligated into the pGL ANRIL vector using the rapid ligation kit (NEB UK), following the manufacturer's recommendations. The completed plasmid was purified by phenol/chloroform extraction followed by ethanol precipitation. 1 μΙ of purified plasmid DNA was restricted with Ncol and run on a TAE agarose gel for quantification.

Cell culture and transfection

SW872 cells were cultured In DMEM 4.5g Glucose +10% Fetal Bovine Serum + 1 % Penicillin/Streptomycin. Cells were cultured in 96-well plates for 24hrs prior to transfection. 100ng of prepared plasmid DNA was transfected per well for both the methylated and mock reactions, with six replicates per transfection. The pGL CMV Renilla plasmid (Promega UK) was co-transfected as a control. Transfections were carried out using FuGENE 6 (Promega UK) following manufacturers' guidelines. Transfected cells were cultured for 48hrs prior to addition of lysis buffer. Luciferase assays were carried out using the Dual-Luciferase® Reporter Assay System (Promega UK), on a VarioSkan Flash Luminometer (ThermoScientific).

Electrophoretic mobility shift assays

Nuclear extracts were prepared from human neuroblastoma cell line IMR-32. Cell pellets were re-suspended in 800μΙ of lysis buffer (10mM Hepes, pH 7.9, 10mM KCI, 0.1 mM EDTA, 1 mM DTT, 0.5mM PMSF, 1 mM leupeptin) and incubated on ice for 10 min. Then 50μΙ of NP-40 was added, the cells vortexed for 10 seconds and spun at 1 1 ,000g for 30 seconds. The supernatant was discarded and the pellet containing nuclei re-suspended in 50μΙ of nuclear extraction buffer (20mM HEPES, Ph 7.9, 0.4M NaCI, 1 mM EDTA, 1 mM DTT, 1 mM PMSF). This was then incubated on ice for 30 min and nuclear debris pelleted at 4C for 5 minutes at 1 1 ,000g. The protein concentration of nuclear extracts was determined with the BCA protein assay kit (Pierce) using the manufacturer's instructions. The nuclear extracts were stored at -80°C until use.

For EMSA, double-stranded DNA oligonucleotides (biomers.net) labelled with biotin at the 5' termini of both strands were used. Sense strand sequences were as follows: Single-stranded oligonucleotide probes were annealed by heating equimolar amounts of complementary strands to 95°C for 5 minutes and slowly cooling the reaction mixture to room temperature. Electrophoretic mobility shift assays were performed using LightShift Chemiluminescent EMSA Kit (Thermo Scientific). All binding reactions were carried out in the presence of 2pmol of the non-specific DNA Poly (dl*dC), x fmol of biotin-labelled probe, 1x binding buffer, 2.5 % glycerol, 0.05 % NP-40, 5mM MgClg (20μΙ final volume) and incubated on ice for 20 minutes. For reactions carried out in the presence of nuclear extract ^g) total protein was utilised. Competition was performed in the presence of 500-fold excess of unlabelled oligonucleotide (methylated or non-methylated) and incubating on ice for 30minutes prior to addition of labelled probe. Complexes were resolved on pre-run 4 % non-denaturing polyacrylamide gel in 0.5X Tris-borate-EDTA for 45 minutes at 100V followed by semi dry transfer (200mA 1 hr) to a positive nylon membrane and UV cross-linking. After incubation in blocking buffer for 15 min at room temperature the membrane was incubated with streptavidin-HRP conjugate for 15 minutes. The membrane was then washed and visualised with a chemiluminescent substrate (Thermo Scientific).

Statistical analysis

Statistical analysis was carried out using Stata (Statacorp) versions 1 1.2 and 12.1. Histograms of all continuous variables were plotted to check for normality. The distributions of some measures of adiposity were observed to be skewed and were therefore transformed using a log e transformation. DXA adiposity measurements were calculated without including the child's head in order to minimise the effect of head movement on the measurement. Regression models were built using child's adiposity measurement (at either 4 or 6 years) as the outcome and CpG methylation as the predictor. Models were adjusted for child's sex and age as appropriate for each time point. Adiposity was measured as %fat and total fat in grams, converted to Z-scores to facilitate interpretation of the effect size, and methylation measurements as %s; the regression coefficient can therefore be interpreted as the standard deviation change in %fat (or total fat) for each % change in methylation. Results are presented as regression coefficients (β) for each 10% change in methylation with their associated P-values.

Results

Identification of Differentially Methylated Regions at birth, associated with later adiposity

To identify perinatal epigenetic biomarkers associated with later body composition outcomes, we carried out a genome wide DNA methylation analysis using a methyl binding domain (MBD) array in umbilical cord, comparing DNA methylation levels at birth with differences in percentage fat mass in the children aged 6 years. Array data was analysed using the validated algorithm BATMAN (28), designed to handle MBD data. Genome wide DNA methylation data from high and low fat mass groups were compared at 100nt resolution, looking for regions of high differential methylation between these groups. This comparison identified 93 differentially methylated regions (DMRs) associated with percentage fat mass in the children at 6 years of age (Table 5).

Table 5

List of DMRs

GeneSymbol Fisher_Exact_p Chr Start End Alt Start Alt End

FEN1 1.00E-06 1 1 61561031 61561530

CCND1 1.00E-06 1 1 69456477 69456776

GAPDH 1.00E-06 12 6644416 6644715

NGB 1.00E-06 14 77736478 77736777

GUSB 1.00E-06 7 65445733 65445832

CCDC147 6.00E-06 10 1061 14677 1061 15476

ZNF17 6.00E-06 19 52075315 52075614

GATA6 2.40E-05 18 19747012 1974741 1

MPV17 2.90E-05 19 18301439 18301838

AKAP8 3.20E-05 19 15492244 15492743

TNFRSF19 4.20E-05 13 24154000 24154399

ABCA4 4.70E-05 1 94587826 94588425

TMEM1 16 5.50E-05 12 112450339 112450838 112446439 112446638

SETD4 5.60E-05 21 37432113 37432512

SEZ6 6.50E-05 17 27337616 27337915

ZNF56 7.60E-05 19 37178039 37178338

APP 9.30E-05 21 27541903 27542102

CREBL2 0.000121 12 12763984 12764683

ZSWIM1 0.000146 20 4450991 1 44510510

SP1 0.000228 12 53774930 53775429

FBX015 0.000281 18 71812958 71813357

OR4C1 1 0.000282 1 1 55371373 55371672

C17orf58 0.000297 17 65987793 65988192

SAA1 0.00037 1 1 18279872 18280171

SIAH1 0.000379 16 48418280 48418579

COPZ1 0.000479 12 54733063 54733521

LIN52 0.000655 14 74553677 74554676

C1 orf116 0.000813 1 207192230 207192629

SLC6A4 0.000894 17 23561381 28561680

ACTR6 0.000942 12 100594927 100595126

MED16 0.000948 19 893402 893901

C22orf39 0.000948 22 19435894 19436093

FYN 0.001036 6 112040878 112041277

MLEC 0.001079 12 121125691 121125990 ID1 0.001079 20 30190922 30191421 30190622 30190921

AMZ2 0.00129 17 66246576 66246875

FGF14 0.001357 13 102571875 102572274

BACH1 0.001549 21 30668863 30669262

EXD2 0.00176 14 69674108 69674407

CBLN4 0.001803 20 54578308 54578807

MEG8 0.001841 14 101379064 101379663

NDUFS8 0.002252 11 67799076 67799375

PCBP2 0.002439 12 53846447 53846946

SNORD17 0.002526 20 17948153 17948452 17945153 17945552

ZNF664 0.00283 12 124458896 124459195

OR11 H6 0.003012 14 20691887 20692386

OR4K5 0.003253 14 20384663 20385162

SLC39A6 0.003518 18 33707646 33708045

BRSK2 0.003586 11 1412220 1412419

PRRC1 0.003859 5 126858993 126859692

C14orf142 0.003905 14 93674364 93677363

APOL5 0.003905 22 36112644 36113243 36108844 36109443*

WWC2 0.004009 4 184023608 184024107

DDX10 0.004046 11 108534698 108534997

DPF3 0.004774 14 73365990 73366261

GNG4 0.004802 1 235812299 235812598

LAMA5 0.005051 20 60941631 60941930

DSCR9 0.005272 21 38593139 38593438

RNF43 0.005397 17 56433328 56434227

DGKZ 0.005776 11 46349163 46349562

TRIM13 0.006775 13 50567722 50568621

FAM174B 0.006871 15 93198365 93198664

TOP2A 0.007924 17 38573149 38573648

NGFR 0.007924 17 47575125 47575281

C10orf107 0.009278 10 63417842 63418241

PCNA 0.010237 20 5099301 5099900

PALB2 0.012353 16 23649185 23649584

DHX34 0.012353 19 47851748 47852047

CD1 E 0.013548 1 158324116 158326284

HSF5 0.014521 17 56565374 56565573

INTS7 0.014585 1 212205435 212205634

IFNK 0.014585 9 27518797 27519196

LOC 100130987 0.014747 11 67155334 67155633

ZNF420 0.014747 19 37567943 37568642

VN1 R4 0.016052 19 53775191 53775690

OR5D18 0.0181 11 55587297 55587596

TMEM109 0.018377 11 60682108 60682707

CDKN2A 0.018945 9 21993565 21993864

HLA-C 0.019681 6 31240679 31240978 LTBR 0.020236 12 6492667 6493066

TMEM61 0.022593 1 55441680 55441979

MTNR1 B 0.02273 1 1 92704083 92704582

CD40 0.025506 20 44742246 44742845

CLIC6 0.029338 21 36042842 36043341

MRPL16 0.032981 1 1 59577672 59578071 59575872 59576471

FAR1 0.03451 1 1 13689600 13689799

PSMD11 0.036848 17 30773381 30773979

GNAS 0.036848 20 57467617 57467916

CADM3 0.038332 1 159139108 159139407

TMEM49 0.04337 17 57916371 57916970

FGF14 0.045241 13 103052282 103052581

KRTAP 0.045241 21 31875320 31875619

OAS1 0.04992 12 113344320 113344754

Pathway analysis

Pathway enrichment analyses of the 93 DMRs was performed using the Ingenuity software programme to search for known interactions between the genes and to identify pathways enriched amongst the DMRs. The top pathways enriched amongst the DMRs were DNA replication and repair, cell death and survival and cell to cell signalling. Within the DNA replication and repair category were DMRs within Cyclin dependent Kinase 2A (CDKN2A), cyclin D (CDND1), flap structure-specific endonuclease 1 (FEN1), BTB and CNC homologyl gene (BACH 1), E3 ubiquitin-protein ligase (SIAH1), proliferating cell nuclear antigen (PCNA), A-kinase anchor protein 8 (AKAP8) (Figure 14B) and SLC6A4 (Figure 14).

Validation of Differentially Methylated Regions by pyrosequencing

To validate the association between the methylation of selected DMRs at birth with percentage fat mass in the children age 6 years, pyrosequencing-based bisulfite PCR analysis was carried out across the DMRs of CDKN2A, FEN1 , BACH1 and SLC6A4, on umbilical cord DNA from 247 SWS subjects for which 6 year DXA measurements of adiposity were available. DMRs were selected from the top two pathways enriched amongst the DMRs, namely DNA repair and replication, and gene expression and cell survival, based upon the level of differential methylation between the high and low fat mass groups, and CpG dinucleotide clustering.

The region of CDKN2A identified as a DMR lies within intron 1 of p14^ARF and 922 bp upstream of the TSS of the long non-coding RNA ANRIL, which is transcribed in the opposite direction from p14^ARF (Figure 1). Figure 1 shows differential methylation at the CDKN2A loci. Percentage adiposity values for SWS subjects were divided into three groups from low (group 1) to high (group 3) fat mass. The percentage methylation difference between group 1 and group 3 was then compared for each 100 nucleotide region to identify regions of high differential methylation.

An inverse association was observed on the MBD array between the methylation of the CDKN2A DMR at birth and %fat mass age 6 years (Figure 1 and Figure 15A). To validate this association, primers were designed across the 9 CpGs in the identified DMR and methylation analyzed by sodium bisulfite pyrosequencing on an extended sample set of subjects from the SWS cohort (n=247). This revealed that lower % methylation of CDKN2A CpGs 1 , 2, 3, 4 and 6 was associated with greater %fat mass at age 6 years (all p<0.05) (Table 6, Figure 3A and Figure 15B). As DXA measurements for the children at birth and 4 years of age were also available, the association between umbilical cord CDKN2A methylation and body composition outcomes of the children at these earlier time points was also examined. The methylation of CDKN2A CpGs 1 , 2, 4 and 7 were significantly associated with % fat mass at age 4 years (all p <0.05) (Table 6). There were also associations between the methylation of CDKN2A CpGs 1-9 with total fat mass ages 4 and 6 years (all p <0.05) (Table 6, Figure 3A and Figures 15C & 15D). No associations were observed between CDKN2A methylation and %fat mass or total fat mass at birth (data not shown).

Table 6

Pyrosequencing analysis of the DMRs associated with SLC6A4, FEN1 and BACH1 on the larger subset of SWS children also confirmed both the associations and direction of the associations between the methylation of the DMRs and %fat mass of the children age 6 years (Figures 2, 3B, 6, 7B, 5 and 7A, and Figure 16). Of the 5 CpGs measured within the DMR of SLC6A4 (Figure 2), higher methylation of SLC6A4 CpGs 1 and 2 was associated with higher %fat mass at age 6 years (p<0.004) (Table 7, Figure 3B). Figure 2 shows differential methylation at the SLC6A4 loci. Percentage adiposity values for SWS subjects were divided into three groups from low (group 1) to high (group 3) fat mass. The percentage methylation difference between group 1 and group 3 was then compared for each 100 nucleotide region to identify regions of high differential methylation. Figure 3 shows the correlation between DNA methylation for CDKN2A and SLC6A4 and DXA %fat at age 6 years in the SWS cohort. DXA fat measurements plotted against percentage methylation (fourths) illustrating (A) the negative correlation of increasing DXA %fat with increased methylation levels at CDKN2A CpG4. (B) the positive correlation of increasing DXA %fat with increased methylation levels at SLC6A4 CpG1.

A positive association between SLC6A4 CpG 2 methylation was also observed with total fat mass (p=0.013) at 6 years of age (Table 7). There were no associations between SLC6A4 methylation and %fat or total fat mass at birth or 4 years. For BACH1 , the methylation of CpGs within the DMR was measured (Figure 5); an inverse association was observed between BACH1 CpG1 methylation and % fat mass at age 6 (p=0.038) and 4 years (p=0.029) (Table 8, Figure 7A); there were trends for inverse correlation between BACH1 CpG4, CpG5 and CpG6 methylation and child's %fat and/or fat mass, and for a positive correlation between BACH1 CpG8 methylation and child's %fat and fat mass (Table 8). For FEN1 the methylation of 4 CpGs within the DMR was measured (Figure 6); of the four CpGs examined within the DMR, CpG1 showed association with % fat mass at 6 years (p=0.02) and CpG 3 methylation was associated with total fat mass age 4 years (p=0.045) (Table 9, Figure 7B). Figure 5 shows differential methylation at the BACH1 gene. Percentage adiposity values for SWS subjects were divided into 3 groups from low (group 1) to high (group 3) fat mass. The percentage methylation difference between group 1 and group 3 was then compared for each 100 nucleotide region across the BACH1 locus. Figure 6 shows differential methylation at the FEN1 gene. Percentage adiposity values for SWS subjects were divided into 3 groups from low (group 1) to high (group 3) fat mass. The percentage methylation difference between group 1 and group 3 was then compared for each 100 nucleotide region across the FEN1 locus. Figure 7 shows the correlation between DXA Fat and DNA methylation for BACH1 and FEN1 in the SWS cohort. DXA %fat measurements plotted against percentage methylation (fourths) illustrating (A) the negative correlation of increasing methylation levels at BACH1 CpG1 with increased DXA %fat at age 6 years. (B) The positive correlation of increasing methylation levels at FEN 1 CpG1 fat with increased DXA %fat at age 6 years. Table 7

Table 8

Association between CDKN2A methylation and ponderal index in the GUSTO cohort at 18 months and 24 months

Because of the strong associations found between methylation of specific CpGs within the DMR of CDKN2A and later fat mass, we sought to further replicate associations between umbilical cord CDKN2A CpG methylation and adiposity in a second independent cohort where body composition had been measured at age 18 months and 24 months. Lower methylation of CpG1 , 3 and 9 was associated with greater infant adiposity assessed by ponderal index at age 18 months (CpG1 ; p=0.015, controlling for infant gender; Figure 8 and CpG9, ponderal index at day 1 p=0.015, ponderal index at 18 months p=0.014). Figure 8 shows the ponderal index at age 18 months in the GUSTO cohort according to umbilical cord methylation of CDKN2A CpG1 methylation.

Methylation of CDKN2A Differentially Methylated Region of Interest (DM ROD is correlated with ANRIL expression in umbilical cord

The CDKN2A DMR lies within the first intron of p14ARF and less than 1 kb upstream of the transcriptional start site of ANRIL, transcribed in the opposite direction from the other genes in this locus. To determine if methylation of the DMR was associated with the expression of either p14ARF or ANRIL, real-time PCR was used to quantify their expression within umbilical cord samples from the GUSTO cohort. Real-time assays were designed that would detect the most commonly expressed splice variants of ANRIL as well as the unspliced form of the ANRIL transcript. Comparison of transcript levels with methylation at the CDKN2A DMR showed a significant inverse association between the methylation of CDKN2A CpGs 1 ,2, 4-8 with the spliced form of ANRIL (p <0.05) and between CpGs 4-9 with the unspliced variant (p<0.02, Table 10). There was also a positive association between CDKN2A CpG9 methylation and p14ARF expression levels (p=0.049) within the umbilical cord.

Table 10

Regression analysis of methylation level against qPCR Act for all amplicons. Only those tests with nominal p value≤ 0.061 are shown.

The CDKN2A ROI regulates ANRIL expression

Having shown associations between the methylation of specific CpGs sites within the CDKN2A gene locus at birth and later measurements of adiposity in infancy and childhood, we next investigated whether methylation of these CpG loci may have functional consequences for gene expression. The region of CDKN2A, identified through the MBD array, whose methylation status was associated with later adiposity in both the SWS and GUSTO cohorts, lies within intron 1 of p14ARF (273bp downstream of the 5' splice site) and 922bp upstream from the TSS of the ANRIL (Figure 1). Differential methylation of the DMR therefore has the potential to influence ANRIL and/or p14ARF expression. To determine whether the DMR identified within the CDKN2A locus is important for the expression of ANRIL, the promoter region of ANRIL (-951 to +20) plus or minus the DMR (-1281 to -950) was fused to the luciferase reporter gene in the vector pGL3basic and transfected into the SW-872 liposarcoma cell line. Deletion of the DMR led to a three-fold increase in ANRIL promoter activity (p=0.01) (Figure 4A). Figure 4 is a luciferase assay examining the role of the CDKN2A region in gene expression and EMSA examining protein complex binding within the CDKN2A DMR. (A) Comparison of luciferase activity for the pGL ANRIL plasmid containing the ANRIL promoter region including the ROI in its natural genomic context upstream of the ANRIL TSS, and the pGL 951 plasmid containing just the ANRIL promoter. (B) EMSA with methylated and unmethylated competitors for CpG 4-7 examining binding in liposarcoma cell extract.

The CPG sites within the CDKN2A DMR regulates ANRIL and p14^ARF promoter activity

To determine whether the CpG sites associated with later adiposity may play a role in regulating the expression of ANRIL and/or p14^ARF, the promoter region of ANRIL (-1281 bp to +20bp relative to TSS) and the 5' portion of the p14^ARF gene (-500bp to +1125bp relative to TSS) were fused to the reporter gene luciferase in the vector pGL3basic, and CpG sites 1 , 2, 3, 4 and 8, mutated (CpG>TpG). Mutation of CpGs 1-4, or CpG 8, led to a decrease in ANRIL promoter activity (all p≤0.008) in the liposarcoma cell line SW-873. In contrast only mutation of CpG 3 led to a decrease in p14^ARF promoter activity in SW-873 cells, while mutation of the other CpG sites had no effect on expression. Mutation of these five CpG sites also led to a decrease in ANRIL promoter activity in SAOS-2 cells (Figure 9B) and in the CpGs examined in A459 and JAR cells. However in A459, SAOS-2 and JAR mutagenesis of the sites CpG sites led to an increase in p14^ARF expression (Figure 9A-D).

Methylation of the CpG 4 blocks DNA binding

Having shown that CpG loci associated with later adiposity are important in regulating the level of ANRIL and p14^ARF promoter activity, we investigated whether transcription factors may bind to these CpG sequences, and whether DNA methylation affected transcription factor binding. Nuclear extracts from SW-873 liposarcoma cells showed a single complex bound to oligonucleotides containing either CpGs 1 or CpGs 8-9, while two complexes were seen bound to the oligonucleotides containing CpGs 2-3 and 4-7. The higher molecular weight complexes bound to CpG 2-3 and 4-7 seen in SW-873 cells, were not detected in SAOS-2 cells (Figure 9E), suggesting that binding in this region occurs in a tissue specific manner.

To determine whether methylation of the CpG loci within the DMR may affect the binding of these complexes to the DNA, the unmethylated labelled probe containing CpGs 4-7 was incubated with nuclear extracts from the liposarcoma cells with a 50, 100 and 500 -fold excess of either: the unmethylated competitor sequence, or a competitor sequence containing a methylated cytosine at position 4 (CpG4). While binding to the sequence was substantially reduced in the presence of 50 fold excess of unmethylated specific competitor, binding was only reduced in the presence of a 500 fold excess of the methylated competitor (Figure 4B).

Conclusion

We identified over 93 DMRs in umbilical cord samples taken at birth that are associated with adiposity in later life at 6 years of age. The DMRs identified in these studies were found in both the 5' promoter regions of genes as well as within gene bodies. 41 of the 93 DMRs identified from the genome wide analysis were associated with genes which play key roles in DNA replication and repair. In animal models investigating the developmental programming of obesity, persistent alterations to the methylation and expression of cell cycle control genes have been reported in the offspring. Such changes may allow an organism to adjust its developmental programme by regulating cell number, turnover and/or differentiation in response to environmental cues in early life during the process of developmental plasticity.

In particular, we have shown that methylation of CDKN2A, SLC6A4, FEN1 and/or BACH1 at birth is associated with later adiposity at 6 years of age.

Increased methylation of CpGs within the CDKN2A gene locus measured at birth was strongly correlated with lower adiposity in childhood. For CDKN2A, there was also an association seen with total fat mass at 6 years and % fat mass and total fat mass at 4 years. An association between the increased methylation of CpGs within the DMR of CDKN2A and lower ponderal index in children at 18 months was seen in the GUSTO cohort, a second independent mother offspring cohort. A negative association between the methylation of CpG1 to 3 and subscapular skinfold thickness at 24 months was also seen in the GUSTO cohort. Interestingly no associations were seen between the methylation of this region and measures of adiposity at birth in either cohort, suggesting that methylation of these CpG loci may be linked with the gain of fat postnatally, or act as a marker for this increasing adiposity. These results demonstrate that the early life environment is an important determinant of adiposity and may influence future health through the altered epigenetic regulation of gene expression.

A methylation array was used to identify a DMR within intronl of p14ARF and upstream of ANRIL. This region was shown to be important for the expression of ANRIL, and that its deletion leads to an increase in in ANRIL promoter activity, suggesting that the DMR acts to inhibit ANRIL expression.

There was also a strong inverse correlation between the methylation of these CpGs and the expression of ANRIL, supporting the experiments which show that the DMR is important for ANRIL expression and showing that methylation of CpGs within the DMR act to inhibit ANRIL expression. DMR methylation was also positively associated with p14ARF expression, demonstrating that methylation within the DMR has differential effects on these two genes and, in the case of p14ARF, enhances expression. The differential effects of methylation on ANRIL and p14ARF expression shown by these experiments may reflect the different location of the DMR with respect to the TSS of ANRIL and p14ARF, and show that lower ANRIL expression is associated with lower adiposity.

In liposarcoma cells, mutagenesis of CpG sites 1 , 2, 3, 4 or 8 all led to a decrease in ANRIL promoter activity, however only mutation of CpG 3 led to a decrease in p14ARF expression, while mutation of the remaining CpG sites had no effect on p14ARF expression. CpG mutagenesis was also seen to decrease ANRIL promoter activity but increase p14ARF expression in A459, SAOS-2 and JAR cells. The differential effects of CpG mutagenesis within ANRIL and p14^ARF may reflect the different location of the CpG sites with respect to the TSS of ANRIL and p14^ARF. Interestingly, an inverse correlation was seen between the methylation of CpGs within the CDKN2A DMR and the expression of ANRIL, and a positive correlation between DMR methylation and p14^ARF expression in the GUSTO cohort, consistent with the DMR having opposing effects on ANRIL and p14^ARF. The CDKN2A DMR may also mediate tissue specific effects, as the effects of CpG mutagenesis was tissue dependent and there were also tissue specific differences in protein binding to this region. Binding, at least in liposarcoma cells, was also affected by methylation. Methylation of CpG4, one of the CpGs most strongly associated with later adiposity, reduced the binding of a specific protein complex to this sequence, suggesting that CpG methylation within the DMR may alter the regulatory capacity of this sequence by disrupting transcription factor binding to this sequence.

The MBD array also identified a number of DMRs associated with later adiposity, including SLC6A4, BACH1 and FEN1 , the association of which with later adiposity was also confirmed by pyrosequencing. Interestingly, these are genes which have previously been linked to obesity or adipocyte function. SLC6A4 is a serotonin transporter, and serotonin plays an important role in the control of energy balance and body weight. Depletion of central serotonin results in hyperphagia and obesity, whereas agents that increase serotonin activity in the CNS inhibit food intake and promote weight loss. A promoter polymorphism in SLC6A4 has been associated with eating disorders, obesity and type 2 diabetes mellitus. Methylation of the SLC6A4 promoter is also associated with obesity in monozygotic (MZ) twins. The region identified in this screen was downstream of the differentially methylated region identified in the study that examined MZ twins. The role of the identified region is, at present, unknown and whether methylation affects the binding of a regulatory protein, or it is a marker of gene transcription is yet to be elucidated. BACH1 encodes a transcription factor that belongs to the cap'n'collar type of basic region leucine zipper factor family. BACH1 plays a key role in adipocyte differentiation; it also regulates heme oxygenase which plays a critical role in inflammation, insulin signalling diabetes and obesity. The FEN1 gene is located between the fatty acid desaturase 1 and 2 (FADS1 and FADS2) genes, on chromosome 1 1. A recent genome-wide association study for plasma polyunsaturated fatty acids (PUFAs) showed strong evidence for association with this region of chromosome 1 1 , most significant association between the SNP rs174537 (FEN1 10154G>T) near FADS1 and arachidonic acid (AA, 20:4ω6). AA is synthesised primarily in the liver and then mobilised to inflammatory cells via blood lipoproteins. AA is the precursor of prostaglandins, leukotrienes, and related compounds, all of which have important roles in inflammation. Thus, the FEN1 10154G>T SNP may represent a candidate gene of importance in obesity and cardiovascular disease.

Genome-wide association studies have suggested that genetic changes account for only a small proportion of the variations in fat mass. These findings suggest that the early life environment, through epigenetic mechanisms, may also make an important contribution to variation in adiposity. Without being bound by theory, it is believed that a 10 percent increase in perinatal CDKN2A methylation is associated with a decrease in %fat mass of around 0.20 standard deviations (95% CI -0.33, -0.06) at 4 years of age and 0.18 SD (95% CI -0.31 , -0.05) age 6 years, controlling for sex.

The data indicate that epigenetic measures at birth may have prognostic value and suggest that it is possible to detect epigenetic marks in readily available tissues such as cord at birth and use these marks as predictors of later phenotype in more disease relevant tissues.

The association between CDKN2A methylation at birth and later adiposity indicates that such epigenetic measurements provide useful markers of body composition in later life. This is important, as obesity is casually associated with insulin resistance in children and suggests that such epigenetic markers may be of long term clinical significance. CpG loci within the CDKN2A DMR are important for the regulation of the long non-coding RNA ANRIL and p14^ARF, indicating that epigenetic processes, as well as genetic alterations, are important determinants of future disease risk within the CDKN2A locus. EXAMPLE 2

CDKN2A and associations with bone mineral content and bone density measurements

The same populations used in Example 1 were also used in Examples 2 and 3.

• Relating the umbilical cord CDKN2A measurements to DXA measurements of bone health at ages 4 and 6 years, there were strong associations between methylation of CDKN2A CpGs 4-9 and bone area and BMC at 4 and 6 years of age.

At ages 4 and 6 years, there were numerous associations between the methylation of CpGs4-9 and bone area and BMC, strongest for CpG9, which was also associated with BMD.

Table 11 : Association of CDKN2A CpGs within the identified DMROI with total bone area, total BMC, total BMD, total size-corrected total area adjusted BMC at ages 4 years and 6 years.

4 yr DXA: Total bone 4 yr DXA: Total BMC 4 yr DXA: Total BMD area (cm sq), without (g), without heads, (g/cm sq), without heads, adjusted for sex adjusted for sex heads, adjusted for sex p- p- p- n b value n b value n b value

CDKN2A CpG1 263 -0.163 0.589 263 -0.145 0.609 263 0 0.798

CDKN2A CpG2 256 -0.272 0.381 256 -0.31 1 0.286 256 0 0.358

CDKN2A CpG3 231 -0.245 0.558 231 -0.444 0.259 231 0 0.188

CDKN2A CpG4 294 -0.654 0.027 294 -0.636 0.024 294 0 0.1

CDKN2A CpG5 292 -0.658 0.024 292 -0.621 0.025 292 0 0.096

CDKN2A CpG6 292 -0.381 0.177 292 -0.526 0.05 292 0 0.052

CDKN2A CpG7 290 -0.633 0.056 290 -0.61 0.052 290 0 0.157

CDKN2A CpG8 273 -0.621 0.047 273 -0.633 0.032 273 0 0.1 1

CDKN2A CpG9 263 -1.099 0.005 263 -1.037 0.005 263 -0.001 0.043

4 yr DXA: Total Size- 4 yr DXA: Total Area corrected BMC (g), adjusted BMC (g), without heads, without heads, adjusted for sex adjusted for sex

p- p- n b value n b value

CDKN2A CpG1 261 0.1 1 0.302 263 -0.016 0.906

CDKN2A CpG2 254 0.038 0.733 256 -0.096 0.487

CDKN2A CpG3 230 0.01 1 0.938 231 -0.251 0.177

CDKN2A CpG4 292 -0.028 0.803 294 -0.12 0.406

CDKN2A CpG5 290 -0.01 0.928 292 -0.102 0.469

CDKN2A CpG6 290 -0.035 0.743 292 -0.226 0.098

CDKN2A CpG7 288 -0.03 0.81 1 290 -0.11 1 0.487

CDKN2A CpG8 271 0.01 1 0.928 273 -0.143 0.348

CDKN2A CpG9 261 -0.047 0.754 263 -0.171 0.363

6 yr DXA: Total area 6 yr DXA: Total BMC 6 yr DXA: Total BMD

(cm sq), without heads, (g), without heads, (g/cm sq), without adjusted for sex and adjusted for sex and heads, adjusted for age age sex and age

p- p- p- n b value n b value n b value

CDKN2A CpG1 216 0.026 0.954 216 -0.088 0.86 216 0 0.7

CDKN2A CpG2 212 -0.332 0.475 212 -0.518 0.321 212 0 0.26

CDKN2A CpG3 191 0.135 0.832 191 -0.3 0.675 191 0 0.324

CDKN2A CpG4 241 -0.567 0.214 241 -0.909 0.078 241 -0.001 0.067

CDKN2A CpG5 240 -0.833 0.068 240 -1.195 0.02 240 -0.001 0.022

CDKN2A CpG6 240 -0.545 0.216 240 -0.908 0.068 240 -0.001 0.047

CDKN2A CpG7 238 -0.653 0.194 238 -0.986 0.081 238 -0.001 0.08

CDKN2A CpG8 221 -0.703 0.147 221 -0.983 0.07 221 -0.001 0.084

CDKN2A CpG9 212 -1.494 0.015 212 -1.818 0.008 212 -0.001 0.023

6 yr DXA: Total Size- 6 yr DXA: Total Area corrected BMC (g), adjusted BMC (g), without heads, without heads, adjusted for sex and adjusted for sex and age age

P- P- n b value n b value

CDKN2A CpG1 213 0.058 0.741 216 -0.117 0.583

CDKN2A CpG2 209 0.018 0.921 212 -0.177 0.425

CDKN2A CpG3 188 -0.075 0.763 191 -0.441 0.149

CDKN2A CpG4 238 0.013 0.944 241 -0.323 0.173

CDKN2A CpG5 237 -0.073 0.699 240 -0.338 0.153

CDKN2A CpG6 237 0.019 0.919 240 -0.346 0.129

CDKN2A CpG7 235 0.008 0.969 238 -0.313 0.229

CDKN2A CpG8 218 0.05 0.8 221 -0.26 0.292 CDKN2A CpG9

209 -0.063 0.804 212 -0.276 0.377

CDKN2A and associations with lean mass

• Relating the umbilical cord CDKN2A measurements to DXA measurements of lean mass at ages 4 and 6 years, there were associations for CDKN2A CpGs 1-9 with %lean mass at ages 4 and 6 years - these are likely to primarily reflect associations of these CpGs with greater fat mass.

At ages 4 and 6 years, there were no associations between cord CDKN2A methylation and total lean mass, apart from a borderline association for CpG6 & 6 year lean mass; there were however numerous positive associations between methylation of CpGs 1 -9 and % lean mass at ages 4 & 6.

Table 12: Association of CDKN2A CpGs within the identified DMROI with total lean mass and % lean mass at 4 years and 6 years.

4 yr DXA: Total lean (g), without 4 yr DXA: % lean, without heads, heads, adjusted for sex adjusted for sex

n B p-value n b p-value

CDKN2A CpG1 221 3.125 0.723 221 0.086 0.008

CDKN2A CpG2 215 -2.051 0.82 215 0.088 0.008

CDKN2A CpG3 192 -10.066 0.409 192 0.071 0.12

CDKN2A CpG4 247 -8.265 0.363 247 0.091 0.007

CDKN2A CpG5 246 -1 1 .345 0.195 246 0.054 0.097

CDKN2A CpG6 246 -12.245 0.148 246 0.059 0.062

CDKN2A CpG7 244 -5.788 0.555 244 0.076 0.037

CDKN2A CpG8 228 -12.051 0.181 228 0.056 0.096

CDKN2A CpG9 218 -15.929 0.162 218 0.073 0.085 6 yr DXA: Total lean (g), without 6 yr DXA : % lean, without heads, heads, adjusted for sex & age adjusted for sex & age

n B p-value n b p-value

CDKN2A CpG1 209 5.076 0.7 208 0.108 0.002

CDKN2A CpG2 205 1 .189 0.931 204 0.1 1 1 0.003

CDKN2A CpG3 184 -3.8 0.841 184 0.143 0.005

CDKN2A CpG4 232 -18.667 0.166 231 0.107 0.005

CDKN2A CpG5 231 -25.372 0.06 230 0.069 0.074

CDKN2A CpG6 231 -25.55 0.049 230 0.094 0.01 1

CDKN2A CpG7 229 -19.59 0.187 228 0.088 0.037

CDKN2A CpG8 213 -27.257 0.06 212 0.055 0.19

CDKN2A CpG9 204 -33.49 0.066 203 0.083 0.109

BACH 1 and associations with bone mineral content and density

Umbilical cord BACH1 CpG1 methylation was associated with bone area and BMC at age 6 years, with similar associations at age 4 years. CpG3 methylation was associated with bone area at age 4 years, but not age 6 years. A small number of other associations were of borderline significance.

Table 13: Association of BACH1 CpGs within the identified DMROI with total bone area, total BMC, total BMD, total size-corrected total area adjusted BMC at ages 4 years and 6 years.

4 yr DXA: Total BMD

4 yr DXA: Total bone 4 yr DXA: Total BMC (g/cm sq), without area (cm sq), without (g), without heads, heads, adjusted for heads, adjusted for sex adjusted for sex sex

p- p- p- n b value n b value n b value

BACH1 CpG1 315 1.556 0.097 315 1.817 0.041 315 0.001 0.058

BACH1 CpG2 324 0.825 0.553 324 0.715 0.585 324 0 0.758

BACH1 CpG3 323 1.879 0.042 323 1.409 0.107 323 0.001 0.346

BACH1 CpG4 261 -1.364 0.251 261 -1.881 0.09 261 -0.002 0.064

BACH1 CpG5 255 -1.098 0.093 255 -0.76 0.213 255 0 0.501

BACH1 CpG6 251 -0.8 0.477 251 -0.754 0.476 251 -0.001 0.508

BACH1 CpG7 233 -0.177 0.908 233 0.222 0.88 233 0 0.733

BACH1 CpG8 229 -0.265 0.839 229 -0.505 0.689 229 0 0.686

4 yr DXA: Total Size- 4 yr DXA: Total Area

corrected BMC (g), adjusted BMC (g),

without heads, without heads,

adjusted for sex adjusted for sex

p- p- n b value n b value

BACH1 CpG1 313 0.356 0.309 315 0.59 0.187

BACH1 CpG2 322 -0.289 0.574 324 0.065 0.922

BACH1 CpG3 321 -0.239 0.493 323 -0.072 0.872

BACH1 CpG4 259 -0.714 0.098 261 -0.805 0.137

BACH1 CpG5 253 0.037 0.874 255 0.105 0.724

BACH1 CpG6 249 -0.146 0.722 251 -0.123 0.812

BACH1 CpG7 231 0.359 0.537 233 0.361 0.629

BACH1 CpG8 227 -0.934 0.062 229 -0.296 0.646

6 yr DXA: Total area 6 yr DXA: Total BMC 6 yr DXA: Total BMD (cm sq), without heads, (g), without heads, (g/cm sq), without adjusted for sex and adjusted for sex and heads, adjusted for sex age age and age

P- P- P- n b value n b value n b value

BACH1 CpG1 258 3.624 0.01 258 3.564 0.025 258 0.002 0.11 1 BACH1 CpG2 262 0.894 0.692 262 0.04 0.988 262 -0.001 0.671 BACH1 CpG3 261 0.774 0.594 261 1.145 0.486 261 0.001 0.426 BACH1 CpG4 215 -1.484 0.418 215 -2.648 0.19 215 -0.002 0.148 BACH1 CpG5 21 1 -0.431 0.672 21 1 -0.115 0.92 21 1 0 0.888 BACH1 CpG6 207 -1.707 0.343 207 -2.388 0.235 207 -0.001 0.261 BACH1 CpG7 191 1.332 0.598 191 3.267 0.252 191 0.003 0.135 BACH1 CpG8 188 3.997 0.074 188 2.947 0.245 188 0.001 0.758

p- p- n b value n b value

BACH1 CpG1 255 0.15 0.795 258 -0.168 0.814 BACH1 CpG2 260 -0.586 0.521 262 -0.895 0.431 BACH1 CpG3 259 -0.514 0.381 261 0.354 0.629 BACH1 CpG4 212 -0.581 0.433 215 -1.105 0.223 BACH1 CpG5 208 0.256 0.531 21 1 0.329 0.518 BACH1 CpG6 204 -0.416 0.578 207 -0.618 0.495 BACH1 CpG7 190 1.742 0.096 191 1.888 0.147 BACH1 CpG8 187 -1.036 0.267 188 -1.181 0.308

BACH 1 and associations with lean mass

Umbilical cord BACH1 CpG1 methylation was associated with total lean mass and % lean mass at both ages 4 and 6 years.

Table 14: Association of BACH1 CpGs within the identified DMROI with total lean mass and % lean mass at 4 years and 6 years.

4 yr DXA: Total lean (g). 4 yr DXA: Percentage lean, without heads, adjusted for without heads, adjusted for sex sex

n b p-value n b p-value

BACH1 CpG1 267 68.1 12 0.025 267 0.245 0.03

BACH1 CpG2 274 43.779 0.317 274 0.053 0.744

BACH1 CpG3 274 10.585 0.713 274 -0.1 0.35

BACH1 CpG4 222 -40.348 0.251 222 0.174 0.171

BACH1 CpG5 217 -16.007 0.408 217 0.108 0.128

BACH1 CpG6 213 14.37 0.657 213 0.219 0.064

BACH1 CpG7 199 0.268 0.995 199 0.031 0.854

BACH1 CpG8 195 -6.566 0.87 195 -0.225 0.131

6 yr DXA: Total lean (g). 6 yr DXA: Percentage lean, without heads, adjusted for without heads, adjusted for sex and age sex and age

n B p-value n b p-value

BACH1 CpG1 249 93.605 0.026 248 0.254 0.03 BACH1 CpG2 253 -13.758 0.84 252 -0.105 0.582

BACH1 CpG3 252 54.497 0.217 251 -0.004 0.971

BACH1 CpG4 206 -42.821 0.431 205 0.163 0.274

BACH1 CpG5 202 5.463 0.856 201 -0.013 0.877

BACH1 CpG6 198 -9.762 0.862 197 0.186 0.229

BACH1 CpG7 182 71 .265 0.37 181 0.1 0.666

BACH1 CpG8 179 78.958 0.251 178 -0.288 0.156

EXAMPLE 3

Introduction

A recognised cause of increasing systolic blood pressure with age is aortic stiffness, which is a significant predictor of future cardiovascular risk and all-cause mortality. Pulse wave velocity (PWV) is an indirect measure of vascular stiffness, with greater PWV also a recognised marker of cardiovascular risk. Such risk is recognised to be set in part by early developmental factors such as unbalanced prenatal nutrition although the use of PWV to assess such risk has been little utilised.

Rationale

Pulse wave velocity (PWV) is an indirect measure of vascular stiffness and higher PWV is an established cardiovascular risk marker.

Objective

To examine the association between umbilical cord methylation status and vascular structure in the child at age 9 years using magnetic resonance imaging (M RI) measurements of aortic stiffness.

Method

Magnetic resonance imaging (MRI) measurement of aortic PWV was performed on 234 children aged 9 years who were participants in the Southampton Women's Survey (SWS), a prospective study of developmental influences on later health and disease. Ethical approval was obtained from the local research ethics committee. The participants and parent/guardian provided informed assent and consent respectively. MRI compatibility of the child and parent/guardian was determined as per local rules.

Aortic stiffness was assessed in the descending aorta. Flow measurements were made in the plane perpendicular to the long axis of the aorta on sagittal and coronal images, both in the proximal descending aorta at the level of the pulmonary trunk, and in the distal descending aorta above the bifurcation. A phase contrast flow mapping sequence was employed with free breathing and retrospective ECG gating. A velocity encoding gradient was applied in the through plane direction with a VENC of 150-200 cm/s. Images were acquired at 30 phases throughout the cardiac cycle. Velocity flow curves were generated using open source imaging software (Osirix) and PWV calculated in m/s using Matlab software (The Mathworks, Inc., Natick, MA) using the transit time method from Ad/ At where Ad is the distance between the two flow acquisition sites in the central aorta, and At is the transit time between the arrival of the systolic wave front at the two sites.

Right Brachial Blood Pressure was measured and recorded immediately following the flow sequence acquisitions with a MR compatible patient monitor (in vivo) using a paediatric cuff.

Results

PWV values were within previously reported childhood ranges. Umbilical cord CDKN2A CpG5 methylation and SLC6A4 CpG1 and CpG5 methylation were positively associated with MRI aortic PWV in childhood (β=0.00663, p=0.037; β=0.0301 , p=0.055; and β=0.0184, p=0.095, respectively); BACH1 CpG6 methylation was inversely with childhood aortic PWV (β=-0.0295, p=0.051).

Conclusions

Variations in epigenetic state may alter vascular development in utero, changing arterial structure with long-term consequences for cardiovascular risk in later life.

EXAMPLE 4

Association of SLC6A4 DNA methylation in perinatal tissues with adiposity in early childhood and with obesity in adults.

SLC6A4 and associations with adiposity in early childhood

Materials and Methods

Southampton Women's Survey (SWS): Participants

Table 16 outlines the participant characteristics.

Body composition measurements in the offspring

At age 4 and 6 years a sub-set of children from the SWS study were assessed for body composition by dual-energy X-ray absorptiometry (Hologic Discovery, paediatric scan mode, Hologic Inc., Bedford, MA). At birth, 12 months, 2, 3, and 6 years of age the children were invited for triceps skinfold thickness measurements.

Instruments were calibrated daily; coefficients of variation were <1 %. Follow-up of the children and sample collection/analysis was carried out under Institutional Review Board approval (Southampton and SW Hampshire Research Ethics Committee) with written informed consent. Clinical investigations were conducted according to the principles expressed in the Declaration of Helsinki.

Skin fold measurement method

To access subcutaneous fat deposition as a measure of adiposity, skinfold thicknesses at standard sites were measured using a skinfold calliper (Holtain) (Inskip, 2006). Measurements were repeated three times and the mean used in analyses.

DNA methylation analysis by Pyrosequencing

Genomic DNA was prepared from umbilical cord and cord blood by a standard high salt method, and from adipose tissue using the QIAamp DNA mini kit (Qiagen, Germany). Bisulphite conversion was carried out on ^g DNA (500 ng for adipose) using the EZ DNA Methylation-Gold kit (Zymo Research, USA) according to manufacturer's instructions. The Pyrosequencing reaction was carried out using primers listed in the following Table 15. Table 15

5' Biotin labelled primers are indicated. Amplicon length and number of CpG's within PCR amplicon are shown.

Bisulfite modified DNA was amplified using 1.25 U HotStar Taq Plus DNA polymerase (Qiagen). Pyrosequencing was carried out using the PyroMark Gold Q96 Reagent kit (Qiagen) on a PyroMark Q96 MD machine (Biotage, Sweden) according to manufacturer's instructions. % methylation was calculated using the Pyro Q CpG software (Biotage).

RT-PCR

Adipose tissue was processed using the RNeasy Lipid Tissue kit (Qiagen) according to manufacturer's instructions. 500 ng total RNA was used as a template for cDNA synthesis and incubated with 10 μηι random nonamers (Sigma-Aldrich, USA), 0.5 μΜ dNTP's and 200 U M-MLV Reverse Transcriptase (Promega, UK) according to manufacturer's instructions. cDNA was amplified for quantification using 5μΙ QuantiTect SYBR Green PCR Mix (Qiagen), 4μΙ cDNA and 1 μΙ SLC6A4 primer (Qiagen). Real time PCR was performed using the Roche LightCycler 480 Real Time PCR Detection System (Roche, Switzerland). Cycling conditions were as follows; initial denaturation of 94°C 15 minutes, followed by 40 cycles of 94°C 15s, 55°C 30s and 72°C 30sec.

Statistical methods

Statistical analysis was carried out using Stata (Statacorp) versions 13.1. Histograms of all continuous variables were plotted to check for normality. The distributions of some measures of adiposity and skinfold thickness were observed to be skewed and were therefore transformed using an appropriate transformation in order to satisfy the assumption of normality. DXA adiposity measurements were calculated without including the child's head in order to minimise the effect of head movement on the measurement. Regression models were built using either child's adiposity measurement (at either birth, 4 or 6 years), or child's skinfold thickness (at either birth, 6 months, 12 months, 2, 3 or 6 years) as the outcome and CpG methylation as the predictor. Models were adjusted for child's sex and age at each time point (and additionally for gestational age at birth). Adiposity was measured as %fat and total fat in grams, converted to Z-scores to facilitate interpretation of the effect size, and methylation measurements as %; the regression coefficient can therefore be interpreted as the standard deviation change in % fat (or total fat) for each % change in methylation. The skinfold thickness measurements used at each age were calculated using the mean of three skinfold measures taken by trained research nurses and were measured in millimetres. The regression coefficients can therefore be interpreted as the change in (transformed) skinfold thickness (mm) for each percentage change in methylation. All results are presented as regression coefficients (β) for each 1 % change in methylation with their associated P-values.

Statistical comparisons of gene expression, or CpG methylation with the lean and obese groups (based on BMI) from the BIOCLAIMS cohort was by Logistical regression, adjusting for age and sex.

Results

Cohort Characteristics

Genomic DNA was extracted from the umbilical cord of SWS cohort subjects (n=680) who had DXA (total and % fat mass) measurements at birth, 4 and 6 years of age and skinfold thickness measurements at birth, 6 months, 12 months, 2 years , 3 years and 6 years. The subjects had a median birth weight of 3.48 kg, median gestational age of 40.4 weeks, and 50% were female. The median maternal age at birth was 31.51 years with a pre-pregnancy median body mass index of 24.27. 48% were in their first pregnancy and 12% smoked in late pregnancy. 289 of these children also had DNA from cord blood available. These 289 children had a similar birth weight distribution to the full cohort with a median birth weight of 3.44kg, median gestational age of 40.06 with 47% being female. The median maternal age at birth was 30.67 years. They had a pre-pregnancy median BMI of 24.28 and 13% were smokers. DXA and skinfold thickness measurements showed a similar range in each group as illustrated by Table 16. Table 16

% or Median (5th,95th % or Median (5th, 95th percentile) for SWS percentile) for SWS 400

Characteristic

800 Cohort (Umbilical Cohort (Cord Blood Cord n=680) n=289)

Educational qualifications (%):

None 2.07 % 0.35 %

CSE 8.71 % 8.65 %

0 levels 26.44 % 26.99 %

A levels 31.91 % 27.34 %

HND 7.83 % 8.30 %

Degree 23.04 % 28.37 %

Social class (%):

Professional 6.01 % 7.02 %

Management and technical 40.09 % 43.16 %

Skilled non-manual 35.14 % 32.98 %

Skilled manual 6.76 % 7.02 %

Partly skilled 10.81 % 9.12 %

Unskilled 1.20 % 0.70 %

Age at birth, years 31.51 (24.6, 36.5) 30.67 (23.6, 35.5)

Smoking (during pregnancy) 12.98 % 13.19 %

BMI 24.27 (19.6, 34.8) 24.28 (19.6, 34.8)

Child

Female (%) 50.15 % 47.40 %

Birth order (%)

1st 47.94 % 48.10 %

2nd 38.24 % 37.37 %

3rd or higher 13.82 % 14.53 %

Birth weight, kg 3.48 (2.7, 4.4) 3.46 (2.7, 4.2)

Gestational age, weeks 40.14 (37.0, 41.9) 40.06 (37.0,41.9)

Child Fat mass Measurements:

Infant total fat at birth, g 510.3 (264, 979) 508.5 (277, 923)

Infant total fat at 4 years, kg 4.11 (2.8, 6.7) 4.05 (2.8, 6.7)

Infant total fat at 6 years, kg 4.88 (3.0, 8.8) 4.85 (3.0, 8.4)

Infant percentage fat at birth 14.39 (8.8, 22.5) 14.37 (9.8, 21.7)

Infant percentage fat at 4 years 28.89 (21.7, 39.1) 28.38 (21.7, 37.7)

Infant percentage fat at 6 years 25.19 (17.7, 35.8) 24.69 (17.8, 34.7)

Child Skinfold Thickness Measurements:

Infant triceps skinfold at birth, mm 4.67 (3.4, 6.3) 4.67 (3.6, 6.3)

Infant triceps skinfold at 6 months, mm 11.00 (7.8, 15.0) 10.97 (7.9, 15.1)

Infant triceps skinfold at 12 months, mm 10.92 (7.5, 15.2) 11.02 (7.0, 14.9)

Infant triceps skinfold at 2 years, mm 10.00 (6.9, 13.7) 10.10 (7.1, 14.0)

Infant triceps skinfold at 3 years, mm 9.95 (6.8, 13.9) 9.80 (6.8, 13.5)

Infant triceps skinfold at 6 years, mm 9.55 (6.3, 16.8) 9.87 (6.2, 16.8) Perinatal SLC6A4 CpG Methylation is associated with measures of childhood adiposity

In umbilical cord, SLC6A4 methylation of the 5 measured CpG's ranged from 79.54 % for CpG 5 to 87.49 % for CpG 4, with the 5^th-95^th centiles of methylation between subjects ranging from 13.74% for CpG 4 to 17.59% for CpG 5. This is illustrated in Table 17.

Table 17

Table 17 displays observed DNA methylation ranges for SLC6A4 CpG's within the different cohorts as measured by Pyrosequencing. Data is shown for the SWS cohorts (umbilical cord and cord blood) and the BIOCLAIMS cohort (adipose).

There was a negative association between SLC6A4 CpG5 methylation with % fat mass (β=- 0.005 p=0.013) and total fat mass (β=-0.012 p=0.0003) at 6 years and with % fat mass (β=- 0.003 p=0.044) and total fat mass (β=-0.007 p=0.007) at 4 years. This is outlined in Table 18 and illustrated in Figure 10. No associations were found between the methylation of CpG5 with total or % fat mass at birth.

To determine if the methylation of these CpG loci within SLC6A4 in cord tissue were associated with additional measures of adiposity in childhood, we investigated their relationship with triceps skinfold thickness in the children at birth through to 6 years. Here, higher methylation of CpG4 was associated with lower triceps skinfold thickness at birth (β=- 0.00 p=0.023) but there was no association at later ages. Higher methylation of CpG 5 was significantly associated with lower triceps skinfold thickness at birth (β=-0.003 p=0.013), at 6 months (β=-0.005 p=0.041), at 2 years (β=-0.006 p=0.0004), at 3 years (β=-0.007 p=0.0003) and at 6 years (β=-0.0001 p=0.013) (Table 19 and Figure 1 1). SLC6A4 CpG Methylation in cord blood is associated with measures of later adiposity in the SWS Cohort

The inventors sought to determine if an association would be found between SLC6A4 methylation and later adiposity in an alternative perinatal tissue such as cord blood. Pyrosequencing analysis of SLC6A4 CpGs 3 to 5 was performed on a subset of subjects from the SWS cohort where DNA from cord blood and measures of adiposity were available (n=289).

In cord blood, the median methylation ranged from 54.16 % for CpG 5 to 76.18 % for CpG 3. The 5^th-95^th centiles of methylation between subjects ranged from 14.04 % for CpG 3 to 18.88 % for CpG 5 (Table 20). The inventors also found associations between SLC6A4 methylation and measures of adiposity in cold blood. Cord blood methylation of CpG5 was negatively associated with total fat mass at age 4 years (β=-0.003, p=0.037) (Table 21), whilst weaker associations were also found between methylation of CpG 5 with total fat mass (β=-0.004, p=0.097) and % fat mass (β=-0.003, p=0.08) at birth. There were also associations between SLC6A4 methylation and skinfold thicknessJVlethylation of CpG3 was negatively associated with triceps skinfold thickness at birth (β=-0.0064, p=0.004), but positively associated at 2 years (β=0.0645, p=0.034) and 3 years (β=0.007, p=0.036). CpG4 (β=-0.0043, p=0.045) and CpG5 (β=-0.0042, p=0.013) were also negatively associated with triceps skinfold thickness at birth and CpG 5 weakly associated with triceps skinfold thickness at 6 years (β=0.006, p=0.049) (Table 22). For CpG 5, associations with triceps skinfold thickness at birth and total fat mass at 4 years were identified in both umbilical cord and cord blood (Figure 12A, 1 1 B).

Table 18: Associations between umbilical cord SLC6A4 CpG methylation levels with child's fatmass in the SWS cohort. Associations between percentage/total fatmass at birth, 4 & 6 years. * p 0.01-0.05, ** p≤ 0.01

Table 19: Associations between umbilical cord CpG SLC6A4 methylation levels with child's skinfold thickness in the SWS cohort.

Associations between triceps skinfold thickness at birth, 6 months, 12 months, 2, 3 & 6 years. * p 0.01-0.05, ** p≤ 0.01

Table 20: Observed DNA methylation ranges for SLC6A4 CpG's within the different cohorts as measured by Pyrosequencing. Data is shown for the SWS cohorts (umbilical cord and cord blood) and the BIOCLAIMS cohort (adipose).

Table 21. Associations between cord blood SLC6A4 CpG methylation levels with child's fatmass in the SWS cohort. Association between percentage/total fatmass at birth, 4 & 6 years. * p 0.01-0.05, ** p≤ 0.01

Table 22. Associations between cord blood SLC6A4 CpG methylation levels with child's skinfold thickness in the SWS cohort

Lower SLC6A4 methylation is associated with obesity in adipose tissue from adults.

SLC6A4 is known to function not just on the appetite control circuitry within the CNS but also has been shown to have an important regulatory role in adipose tissue. The current inventors investigated whether methylation of the CpG loci within SLC6A4 were altered in adipose tissue of obese individuals compared to lean.

Genomic DNA was extracted from the adipose tissue of BIOCLAIMS subjects (n=65) who had BMI measurements. The subjects had a median BMI of 31.74, median height of 1.65 cm and median weight of 82.8 Kg. 76.92 % of participants were female. Subjects were determined as lean or obese according to their BMI: lean participants had a BMI between 18.5 and 25, whilst obese participants had a BMI from 30-40 (Table 23).

Table 23

Pyrosequencing analysis of the 3 CpG's within the SLC6A4 DMROI in adipose tissue of BIOCLAIMS subjects revealed that median methylation ranged from 70.17 % for CpG 5 to 87.38 % for CpG 4. Values for the spread (5^th-95^th centiles) of methylation between subjects ranged from 10.63 % for CpG 4 to 16.29 % for CpG 5 (Table 20). Methylation of CpG 5 (p=0.042) was significantly reduced in adipose tissue from obese individuals compared to lean individuals (Table 24) (Figure 12A). There were no significant differences for CpG 3 or 4. The inventors performed real time PCR, in order to ascertain if the alteration in methylation in adipose tissue was accompanied by an alteration in gene expression of SLC6A4. Analysis of SLC6A4 expression by real time PCR demonstrated that its expression was significantly reduced in the obese individuals compared to lean (p=0.0168, n=63) (Table 24) (Figure 13B).

Table 24

Table 24 illustrates the associations between SLC6A4 gene expression or SLC6A4 CpG methylation and Obesity in the BIOCLAIMS cohort. * p 0.01-0.05, ** p≤ 0.01

Discussion

The above results demonstrate that altered methylation of specific CpGs 1 , 2, 4 and 5 within the gene body of SLC6A4 in perinatal tissue is associated with several measures of adiposity in early childhood. These epigenetic marks are not only present in both umbilical cord and cord blood, but also present in adipose tissue from obese adults, alongside alterations in gene expression.

SLC6A4 is a serotonin transporter known to play an important role in the control of energy balance and appetite, and the current data indicates that SLC6A4 methylation in perinatal tissues may be an epigenetic biomarker for obesity risk. Furthermore, given that the biomarker is present in adipose tissue of a separate adult cohort with gene expression changes, it may also contribute to the development of adiposity.

The current inventors have shown that for a specific CpGs 1 , 2, 4 and 5 within the first intron of the SLC6A4 gene, originally identified from an MBD array, that lower methylation in umbilical cord is associated with increased fat mass and triceps skinfold thickness at several time points during early childhood. This association was also found in cord blood. Next, using adipose tissue from a separate adult cohort, the inventors also showed that SLC6A4 is hypomethylated at the CpG of interest, CpG5, in obese subjects compared to lean.

The inventors also confirmed reduced expression of SLC6A4 in adipose tissue in obese subjects. More specifically, the inventors have shown that SLC6A4 is expressed in human adipose tissue, and is down regulated in obese individuals compared to lean.

The inventors speculate that the downstream affects may result in increased peripheral serotonin availability in the obese state. During development, before its role as a neurotransmitter, serotonin has a trophic role. Therefore any alterations in its signalling in early life may have an impact on brain development. It is therefore plausible that alterations in central serotonin availability induced by altered SLC64A methylation in early life may induce programming which has long term effects on energy balance. Two distinct serotonin systems exist, one which functions in the CNS and has well known effects on mood, sleep and energy balance and one which functions in the periphery. Interestingly, 98% of serotonin is synthesised in the periphery. Peripheral serotonin is thought to act as a hormone involved in intestinal motility, the immune system and vasoconstriction. As the central serotonin system is involved in energy regulation, it is quite plausible that the peripheral serotonergic system has effects on energy balance also, for example by acting on adipocytes.

Most studies looking at the methylation of specific genes in relation to obesity measure methylation and obesity at the same time, i.e. after the onset of obesity (HIF ref). In these situations, it is consequently unknown if altered DNA methylation is a cause or consequence of the obesity. It is therefore very important to distinguish between those modifications which are a result of altered bodyweight and those which may be potential biomarkers, occurring prior to the onset of obesity. For this reason, studies which look in tissue at early stages of development are of utmost importance. The inventors is the first study to define biomarkers of fat mass in perinatal tissue and replicate this result in the adipose tissue of adults with associated alterations in gene expression. The inventors have shown that the epigenetic biomarker, i.e. altered SLC6A4 methylation, is present before the onset of obesity, is present in several perinatal tissues, and also found in adipose tissue of an adult cohort alongside alterations in gene expression. Therefore, not only is this epigenetics marker predictive of fat mass and measures of obesity thus a potential biomarker, it is also highly conceivable that it's altered methylation may have functional consequences on energy balance, making it an appealing target for intervention. EXAMPLE 5

Expression of transcripts from the CDKN2A locus is also associated with measures of infant adiposity in the GUSTO cohort

Materials and Methods

Cohort

The GUSTO participants used in Example 1 and detailed in Table 3 were also used in Example 5

RNA Extraction

Umbilical cord tissue samples were collected at the time of delivery, flushed with saline to remove fetal blood and flash-frozen in liquid nitrogen within 30 min of collection. Umbilical cord tissue (300 mg) was first placed in a sterile Dispomix tube and homogenized for 55 s for 3 cycles in 3 ml of Trizol using the Dispomix (Medic Tools, AG, Zug, Switzerland). After spinning down the debris, the supernatants were divided equally into three 2 ml tubes. 200 μΙ of chloroform were added to each tube, vortexed vigorously and centrifuged for 15 min at 4°C. The aqueous phase was carefully transferred to a new tube containing 1 μΙ of linear acrylamide. An equal amount of isopropanol was added and mixed by inversion. After incubating at -20°C overnight to precipitate the RNA, the pellet was obtained by centrifuging at 13,200 rpm for 10 min at 4°C. The RNA pellet was washed twice in 70% (v/v) ethanol, air- dried and resuspended in RNase-free water. The isolated RNA was then purified using the RNeasy Mini Kit (Qiagen, Hilden, Germany). On-column DNase digestion was carried out before the first wash step according to the manufacturer's instructions. The purified RNA was then eluted in 30 μΙ of RNase-free water and stored at -80°C. RNA concentration and purity were measured using a nanodrop ND-8000 spectrophotometer (Nanodrop Technologies, Wilmington, DE, USA), and RNA integrity was determined using the Agilent 2100 Bioanalyzer and RNA 6000 Nano Labchips (Agilent Technologies, Santa Clara, CA, USA). RNA was reverse transcribed using a High Capacity cDNA Reverse Transcription Kit (Applied Biosystems Inc, ABI, CA, USA).

PCR

PCR reactions were prepared using 10 μΙ of Power SyBr Green PCR 2* Master mix (Applied Biosystems Inc, ABI, Foster City, CA, USA), 1 μΙ of each primer (2 uM), and 20 ng of cDNA in a total reaction volume of 20 μΙ. PCR for each sample was done in triplicates in 384-well plates using the ABI 7900 HT Sequence Detection System. Expression was compared to two endogenous control genes (GAPDH and beta-actin).

Realtime PCR was carried out using Qiagen SYBR Green. Amplification conditions were: 15min 95°C, 40 cycles of: 15s 94°C, 30s 55°C, 30s 72°C.

Primer pairs: p14ARF F:CCCTCGTGCTGATGCTACTGAG (SEQ ID NO: 34); p14ARF R:GTGAGAGTGGCGGGGTCGGC (SEQ ID NO: 35); P16INK4a F: AACGCACCGAATAGTTACGG (SEQ ID NO: 45); p16INK4A R:ACCAGCGTGTCCAGGAAG (SEQ ID NO: 46).

Amplification of ANRIL transcripts was carried out for unspliced RNA using custom designed primers as follows.

ANRIL F: CAGTGGCTTCCTGTTCATGC (SEQ ID NO: 30); ANRIL R: GGGCTTGACGTCTGATCTGT (SEQ ID NO: 31).

Amplification of ANRIL transcripts was carried out for ANRIL splice forms ANRIL14-5, 5-6, and 1-2 using primers from Burd et al 2010, as follows.

ANRIL14-5 F:GGAATGAGGAGCACAGTGATTA (SEQ ID NO: 47);

ANRIL14-5 R:CTGCTGTTGAATCAGAATGAGG (SEQ ID NO: 48);

ANRIL5-6 F:CCACCAGATATATGTTATCTGTGCTTA (SEQ ID NO: 49);

A N R I L5-6 R : GTACTG ACTC GG G AA AG G ATTC (SEQ ID NO: 50);

A N R I L1 -2 F : GCC ACG ACATTTCAAAGG ATTC (SEQ ID NO: 51);

ANRIL 1-2 R:GCACATACCACACCCTAACTAC (SEQ ID NO: 52).

Realtime PCR was carried out using Qiagen SYBR Green. Amplification conditions: 15min 95°C, 40 cycles of: 15s 94°C, 30s 60°C, 30s 72°C. Results

An association was found between p14ARF expression and total fat mass (p=0.027) and percentage fat mass (p=0.021) at birth, and with ponderal index (p=0.035) on postnatal day 7. The inventors also found an association between p16INK4A expression and percentage fat mass (p=0.024) and total fat mass (p=0.04) on postnatal day 7, and with triceps (p=0.031) and subscapular (p=0.019) skinfold thicknesses and % fat mass (p=0.035) at 18 months. Expression of ANRIL was found to be inversely associated with triceps skinfold thickness (p=0.045) and subscapular skinfold thickness (p=0.005) at postnatal day 7

Conclusion

The inventors have shown that expression of ANRIL transcript and the CDNK2A transcripts, p14ARF and p16INK4A, is associated with infant adiposity.

References

1. Frayling, T.M., et al., A common variant in the FTO gene is associated with body mass index and predisposes to childhood and adult obesity. Science, 2007. 316(5826): p. 889-94.

2. Loos, R.J., et al., Common variants near MC4R are associated with fat mass, weight and risk of obesity. Nat Genet, 2008. 40(6): p. 768-75.

3. Thorleifsson, G., et al., Genome-wide association yields new sequence variants at seven loci that associate with measures of obesity. Nat Genet, 2009. 41 (1): p. 18-24.

4. Wellcome Trust Case Control, C, et al., Genome-wide association study of CNVs in 16,000 cases of eight common diseases and 3,000 shared controls. Nature, 2010. 464(7289): p. 713-20.

5. Godfrey, K.M. and D.J. Barker, Fetal programming and adult health. Public Health Nutr., 2001. 4(2B): p. 611-624.

6. Lillycrop, K.A., et al., Induction of altered epigenetic regulation of the hepatic glucocorticoid receptor in the offspring of rats fed a protein-restricted diet during pregnancy suggests that reduced DNA methyltransferase-1 expression is involved in impaired DNA methylation and changes in histone modifications. Br.J.Nutr., 2007. 97(6): p. 1064-1073.

7. Gluckman, P.D., et al., Metabolic plasticity during mammalian development is directionally dependent on early nutritional status. Proc.Natl.Acad.Sci.U.S A, 2007. 104(31): p. 12796-12800.

8. Heijmans, B.T., et al., Persistent epigenetic differences associated with prenatal exposure to famine in humans. Proc.Natl.Acad.Sci.U.S A, 2008. 105(44): p. 17046- 17049.

9. Godfrey, K.M., et al., Epigenetic gene promoter methylation at birth is associated with child's later adiposity. Diabetes, 201 1. 60(5): p. 1528-1534.

10. Congrains, A., et al., Genetic variants at the 9p21 locus contribute to atherosclerosis through modulation of ANRIL and CDKN2A/B. Atherosclerosis, 2012. 220(2): p. 449- 55.

1 1. Pomerantz, J., et al., The Ink4a tumor suppressor gene product, p19Arf, interacts with MDM2 and neutralizes MDM2's inhibition ofp53. Cell, 1998. 92(6): p. 713-23.

12. Abella, A., et al., Cdk4 promotes adipogenesis through PPARgamma activation. Cell Metab, 2005. 2(4): p. 239-49. Zhang, Y., Y. Xiong, and W.G. Yarbrough, ARF promotes MDM2 degradation and stabilizes p53: ARF-INK4a locus deletion impairs both the Rb and p53 tumor suppression pathways. Cell, 1998. 92(6): p. 725-34.

Kim, W.Y. and N.E. Sharpless, The regulation of INK4/ARF in cancer and aging. Cell,

2006. 127(2): p. 265-75.

Matheu, A., et al., Anti-aging activity of the Ink4/Arf locus. Aging Cell, 2009. 8(2): p. 152-61.

Tzanetakou, I. P., et al., s obesity linked to aging?": adipose tissue and the role of telomeres. Ageing Res Rev, 2012. 11 (2): p. 220-9.

Melzer, D., et al., A common variant of the p16(INK4a) genetic region is associated with physical function in older people. Mech Ageing Dev, 2007. 128(5-6): p. 370-7. Cunnington, M.S., et al., Chromosome 9p21 SNPs Associated with Multiple Disease Phenotypes Correlate with ANRIL Expression. PLoS Genet, 2010. 6(4): p. e1000899. Grarup, N., et al., Studies of association of variants near the HHEX, CDKN2A/B, and IGF2BP2 genes with type 2 diabetes and impaired insulin release in 10, 705 Danish subjects: validation and extension of genome-wide association studies. Diabetes,

2007. 56(12): p. 3105-11.

Zhuang, J., et al., Methylation of p15INK4b and expression of ANRIL on chromosome 9p21 are associated with coronary artery disease. PLoS One, 2012. 7(10): p. e47193.

Zeggini, E., et al., Replication of genome-wide association signals in UK samples reveals risk loci for type 2 diabetes. Science, 2007. 316(5829): p. 1336-41.

Broadbent, H.M., et al., Susceptibility to coronary artery disease and diabetes is encoded by distinct, tightly linked SNPs in the ANRIL locus on chromosome 9 p. Hum Mol Genet, 2008. 17(6): p. 806-14.

Inskip, H.M., et al., Cohort profile: The Southampton Women's Survey. Int J Epidemiol, 2006. 35(1): p. 42-8.

Crozier, S.R., et al., Maternal vitamin D status in pregnancy is associated with adiposity in the offspring: findings from the Southampton Women's Survey. Am J Clin Nutr, 2012. 96(1): p. 57-63.

Soh, S.E., et al., Cohort Profile: Growing Up in Singapore Towards healthy Outcomes (GUSTO) birth cohort study. Int J Epidemiol, 2013. Palmke, N., D. Santacruz, and J. Walter, Comprehensive analysis of DNA- methylation in mammalian tissues using MeDIP-chip. Methods, 201 1. 53(2): p. 175- 184.

Weber, M., et al., Chromosome-wide and promoter-specific analyses identify sites of differential DNA methylation in normal and transformed human cells. Nat Genet, 2005. 37(8): p. 853-62.

Down, T.A., et al., A Bayesian deconvolution strategy for immunoprecipitation-based DNA methylome analysis. Nat.Biotechnol., 2008. 26(7): p. 779-785.

Jaffe, A.E., et al., Bump hunting to identify differentially methylated regions in epigenetic epidemiology studies. Int.J. Epidemiol., 2012. 41 (1): p. 200-209.

Barton, S.J., et al., Correction of unexpected distributions of P values from analysis of whole genome arrays by rectifying violation of statistical assumptions. BMC Genomics, 2013. 14: p. 161.

Claims

CLAIMS:

1. A method of predicting the development of one or more phenotypic characteristics in an individual, wherein the phenotypic characteristic is selected from one or more of:

(i) obesity and/or adiposity;

(ii) bone health, and

(iii) cardiovascular disease,

2. A method according to claim 1 , wherein the phenotypic characteristic is one or more of obesity and adiposity, and

wherein the one or more CpG dinucleotide is selected from the group comprising CpG5, CpG1 and CpG2 of the SLC6A4 gene, CpG1 , CpG2, CpG3, CpG4, CpG5, CpG6, CpG7, CpG8, and CpG9 of the CDKN2A gene, , CpG1 and 3 of the FEN 1 gene and CpG1 , CpG4, CpG5, CpG6 and CpG8 of the BACH1 gene, or a combination thereof.

3. A method according to claim 2, wherein the one or more CpG dinucleotide is selected from the group comprising CpG5, CpG1 and CpG2 of the SLC6A4 gene, or a combination thereof.

4. A method according to claim 2, wherein the one or more CpG dinucleotide is selected from the group comprising CpG1 , CpG2, CpG3, CpG4, CpG5, CpG6, CpG7, CpG8, and CpG9 of the CDKN2A gene, or a combination thereof.

5. A method according to claim 2, wherein the one or more CpG dinucleotide is selected from the group comprising CpG1 and 3 of the FEN1 gene, or a combination thereof.

6. A method according to claim 2, wherein the one or more CpG dinucleotide is CpG1 , CpG4, CpG5, CpG6 and CpG8 of the BACH1 gene, or a combination thereof.

7. A method according to claim 1 , wherein the phenotypic characteristic is bone mineral content and/or bone health, and

wherein the one or more CpG dinucleotide is selected from the group comprising CpG4, CpG5, CpG6, CpG7, CpG8 and CpG9 of the CDKN2A gene, CpG1 of the SLC6A4 gene, and CpG1 , CpG3, CpG4, CpG5, CpG7 and CpG8 of the BACH1 gene, or a combination thereof.

8. A method according to claim 7, wherein the one or more CpG dinucleotide is CpG4, CpG5, CpG6, CpG7, CpG8 and CpG9 of the CDKN2A gene, or a combination thereof.

9. A method according to claim 7, wherein the one or more CpG dinucleotide is CpG1 of the SLC6A4 gene.

10. A method according to claim 7, wherein the one or more CpG dinucleotide CpG1 , CpG3, CpG4, CpG5, CpG7 and CpG8 of the BACH1 gene, or a combination thereof.

1 1. A method according to claim 1 , wherein the phenotypic characteristic is cardiovascular disease, and

wherein the one or more CpG dinucleotide is selected from the group comprising CpG5 of the CDKN2A gene, CpG1 and CpG5 of the SLC6A4 gene and CpG6 of the BACH1 gene, or a combination thereof.

12. A method according to claim 11 , wherein the one or more CpG dinucleotide is CpG5 of the CDKN2A gene.

13. A method according to claim 1 1 , wherein the one or more CpG dinucleotide is CpG1 and CpG5 of the SLC6A4 gene, or a combination thereof.

14. A method according to claim 1 1 , wherein the one or more CpG dinucleotide is CpG6 of the BACH1 gene.

15. A method according to any of preceding claim, wherein the level of methylation is an increase in methylation, a decrease in methylation, the presence of methylation or the absence of methylation.

16. A method according to any preceding claim, further comprising

(b) comparing the degree of methylation determined in (a), with the degree of methylation of said one or more CpG dinucleotides in an individual without said phenotypic characteristic at a selected older age.

17. A method according to claim 16, wherein the degree of methylation determined in (a) is an increase in methylation if the level of methylation is greater than that of the level in an individual without said phenotypic characteristic at a selected older age.

18. A method according to claim 16, wherein the degree of methylation determined in (a) is a decrease in methylation if the level of methylation is less than or equal to that of the level in an individual without said phenotypic characteristic at a selected older age.

19. A method according to any preceding claim, wherein determination of an increase in methylation is indicative of the development of said phenotypic characteristic in said individual.

20. A method according to any preceding claim, wherein the methylation status of a CpG site is determined by pyrosequencing, bisulfite sequencing, single molecular real time sequencing, methylation-sensitive PCR amplification, sequenom, bisulfite conversion followed by primer extension, or a combination thereof.

21. A method according to any preceding claim, wherein the at least one CpG site is selected from

SLC6A4 CpG1 : Hg19 chr17: 28561601 ,

SLC6A4 CpG2: Hg19 chr17: 28561578,

SLC6A4 CpG3: Hg19 chr17: 28561505,

SLC6A4 CpG4: Hg19 chr17: 28561490,

SLC6A4 CpG5: Hg19 chr17: 28561468,

CDKN2A CpG1 : Hg19 chr9: 21993721 , CDKN2A CpG2: Hg19 chr9: 21993697,

CDKN2A CpG3: Hg19 chr9: 21993694,

CDKN2A CpG4: Hg19 chr9: 21993654,

CDKN2A CpG5: Hg19 chr9: 21993645,

CDKN2A CpG6: Hg19 chr9: 21993638,

CDKN2A CpG7: Hg19 chr9: 21993629,

CDKN2A CpG8: Hg19 chr9: 21993603,

CDKN2A CpG9: Hg19 chr9: 21993583,

BACH1 CpG1: Hg19 chr21: 30668941,

BACH1 CpG2: Hg19 chr21: 30668995,

BACH1 CpG3: Hg19 chr21: 30669026,

BACH1 CpG4: Hg19 chr21: 30669196,

BACH1 CpG5: Hg 19 chr21: 30669217,

BACH1 CpG6: Hg19 chr21: 30669219,

BACH1 CpG7: Hg19 chr21: 30669245,

BACH1 CpG8: Hg19 chr21: 30669259,

FEN1 CpG1: Hg19chr11: 61561170,

FEN1 CpG2: Hg19chr11: 61561183,

FEN1 CpG3: Hg19 chr11: 61561220 and

FEN1 CpG4: Hg19 chr11: 61561291.

22. A method according to claim 21, wherein the at least one CpG site is selected from the CDKN2A CpG sites identified therein.

23. A method according to claim 21, wherein the at least one CpG site is selected from the SLC6A4 CpG sites identified therein.

24. A method according to claim 21, wherein the at least one CpG site is selected from the BACH1 CpG sites identified therein.

25. A method according to claim 21, wherein the at least one CpG site is selected from the FEN1 CpG sites identified therein.