Aim:

To identify alleles which influence gene expression through a maternal or paternal parent-of-origin mechanism and to correlate these observations to growth and development phenotypes in offspring.

Background:

Parent of origin effects (POE) are genetic effects that deviate from traditional Mendelian inheritance rules by affecting the expression of a phenotype/trait in offspring in a manner that is specifically determined by the sex of the parent from which the locus is inherited. Both maternal and paternal effects may arise via epigenetic mechanisms like imprinting.

Genomic imprinting is a common mechanism of gene regulation achieved through epigenetic modifications, which result in uniparental gene expression. "Epigenetic" refers to DNA changes/modifications, which do not affect the sequence of the genome directly, but can still be inherited. An epigenetic change located in/near a promoter can alter chromatin structure and obstruct transcription. Epigenetic modifications which result in gene silencing include:

-DNA methylation at the 5' position of cytosine residues in CG dinucleotides [1]

-Histone modification through acetylation, phosphorylation, methylation and ubiquitylation [2]

-ATP-dependent chromatin remodelling [3]

Epigenetic modifications are removed during meiosis and then reinstated in the zygote, in imprinted regions these modifications depend on parental origin. It has been suggested that imprinting evolved as the result of genetic conflict between maternal and paternal genes in relation to nutrient transfer to the offspring [4]. Which affect foetal growth in a potentially antagonistic manner, paternally expressed genes generally enhance foetal growth while maternally expressed genes reduce it [5].

To investigate POEs affecting gene expression, publically available expression and genotype data from HapMap families was used to establish the presence of parent-of-origin e-QTL (POE-eQTL) which affect expression through cis-acting mechanisms. Stranger et al [6] measured expression of 47,294 probes in 30 CEU and 30 YRI trios previously genotyped as part of the HapMap project [7]. We were able to map 18,139 of these probes to a genomic location using the UCSC database. Of these, 1,078 probes with a gene expression standard deviation greater than 0.5 were analysed. For each probe, SNPs located 500Kb upstream and 50kb of the transcription start site were tested for parent-specific effects on gene expression using a linear regression model [8]. This analysis resulted in 392 SNPs (annotating 60 genes) with evidence for uniparental effects on transcript levels (p-uncorrectedless than 5 * 10-5). Of these, 89 SNPs can potentially create or destroy methylation sites (meth-SNPs), these meth-SNPs annotate 31 genes across the genome including DAPK1, RIN2, FANK1, PRL for which there is already some evidence of methylation and specific maternal expression [9-12]. From our full set of 392 SNPs, we have selected a total of 83 independent SNPs (33 of which are predicted to generate or destroy methylation sites).

To replicate these findings, we propose to study a subset of 150 children from the ALSPAC cohort, For which gene expression data has been collected using microarrays at the Sanger Institute in Cambridge and for which parental DNA has been collected. We would like to genotype the 83 SNPs selected within the n=57 samples for whom there is complete trio data available, there is genomewide data available and where genomewide transcript data will be available (Dermitzakis expression data collection in ALSPAC) (i.e. DNA from child, mother and father - buccal DNA from fathers). For genotyping, the amounts of DNA required will depend on the genotyping set-up at Kbioscience (currently on their website as 1.5ul of DNA at a concentration of 3.3ng/ul. For 83 SNPs, we will require ~260ng of DNA per sample). The regression model [8] used in the discovery phase of the project will be applied, and bisulphite treatment based technology will be used to establish methylation profile of the suspected meth-SNPs for which findings are replicated at this atage of the study. We expect that some noise caused by clonality issues suggested to affect immortalized cell lines [13] might be detected in our expression study. However, we are confident that our tiered approach and the larger size of the follow-up cohort will reduce the number of false positive results.

As well as evaluating and confirming the presence of POE-eQTLs, we wish to investigate the hypothesis that specific POEs affect growth and development. Therefore, we would like to investigate correlations between genotypes/alleles at SNPs for which we demonstrate POE-eQTLs and a number of growth phenotypes including: birth weight, and other markers of foetal development, growth rate, BMI and anthropomorphic/body composition measurements available in the ALSPAC database.

The results from this study, as well as, enhancing our knowledge and understanding of the control of gene expression through parent-specific mechanisms, have the potential to identify novel imprinted genes which could serve to explain some of the current observed associations to growth phenotypes and represent an appropriate use of the vast resources available through ALSPAC.