Genomic imprinting regulates the monoallelic expression of genes in a parent-of-origin specific manner. To date >90 imprinted loci have been identified in the mouse genome[1], many residing in co-regulated clusters with other imprinted genes, and their allele-specific expression controls a diverse range of functions including growth and development[1]. Of note, the imprinted insulin-like growth factor 2 (Igf2) gene located on mouse distal chromosome 7, along with other genes in the insulin and insulin-like growth factor regulatory pathway, has been shown to be integral for fetal growth and the development of many tissues including the brain[2–5].

In most somatic cells Igf2 and the neighboring H19 gene are reciprocally imprinted; Igf2 gene expression is silenced on the maternal allele, whereas H19 is silenced on the paternal allele. In mice this allele-specific expression is associated with allele-specific DNA methylation at several differentially methylated regions (DMRs) in the Igf2 gene and at the Igf2/H19 imprinting control region (ICR) located immediately upstream of the H19 promoter[6]. The ICR contains four methylation-sensitive CCCTC-binding factor (CTCF) binding-sites mediating the assembly of a chromatin insulator that blocks interactions between the Igf2 promoter and enhancers downstream of the H19 gene. On the unmethylated maternal allele, CTCF binds to the ICR silencing Igf2 expression and stimulating the transcription of H19 (Figure1). IGF2 is a major driver of prenatal growth – for example, placenta specific Igf2 transcripts control the growth of the placenta and supply of maternal nutrients to the developing fetus[7]. IGF2 also acts as part of the IGF signaling pathway to regulate postnatal growth of somatic tissues including the brain[5, 8]. The H19 gene encodes an untranslated RNA which acts as a trans-regulator of the imprinted gene network controlling embryonic growth in mice[9].

Despite being a classically imprinted region, with clear-cut patterns of allele-specific gene expression, there is evidence of considerable epigenetic heterogeneity at the Igf2/H19 locus. Population studies have shown that DNA methylation across the homologous region in the human genome is a normally distributed quantitative trait, which can be influenced by stochastic, genetic and environmental factors[10, 11]. For example, individuals conceived during famine show alterations in IGF2 methylation[12], whilst polymorphisms in the region show an independent and additive association with DNA methylation[10]. Even isogenic strains of mice raised in a controlled environment demonstrate evidence of epigenetic heterogeneity; allelic DNA methylation levels at the Igf2/H19 locus vary extensively in a tissue-specific manner during fetal development, primarily on the expressed paternal allele[13].

The cerebellum is a region of the brain with an important role in motor control and cognition[14]. The cerebellum has a protracted period of development[15] and ultimately accounts for ~12% of total brain weight[16]. Only a handful of studies have investigated Igf2 expression and its regulation by genomic imprinting in the mouse cerebellum. Early work revealed that the Igf2 gene is paternally expressed in cerebellar granule and glial cells during early postnatal development[17], with expression being spatially and temporally specific, indicating that the local synthesis of IGF2 is important for cerebellar development. A recent study showed that Igf2 regulates the proliferation of granule cell precursors, which ultimately determine the final size and shape of the cerebellum[18]. Together these results indicate that the regulation of Igf2 expression during early postnatal development likely plays an important role in the development of the cerebellum.

Previously we reported evidence of a strong, negative correlation between DNA methylation at the IGF2/H19 ICR and cerebellum weight in humans[11]. However, we observed no relationship between IGF2/H19 DNA methylation in the frontal cortex and net brain mass, suggesting that the association is cerebellum-specific. In the current study we comprehensively quantify cerebellar DNA methylation at all four CTCF binding sites spanning the murine Igf2/H19 ICR in an outbred population of Heterogeneous Stock (HS) mice. We report that epigenetic variation in the region shows an association with between-individual differences in cerebellar mass, reflecting the association observed in human post-mortem brain tissue.

These data provide further evidence that epigenetic variation at the Igf2/H19 ICR is associated with growth and development of the cerebellum. To our knowledge this is the first detailed investigation of DNA methylation across all four CTCF binding sites in the Igf2/H19 ICR in mouse cerebellum. Our results suggest that cerebellar DNA methylation across the 2nd and 3rd CTCF-binding sites, upstream of H19, shows a negative relationship with cerebellum mass, with ~7% of the variance in cerebellum mass being uniquely explained by DNA methylation at these loci. No significant relationship was observed between DNA methylation and overall net brain mass for either amplicon, indicating that the effect of cerebellar DNA methylation at these sites is likely to be cerebellum-specific. The results from this study concur with our previous data, which highlight a negative relationship between DNA methylation in the IGF2/H19 ICR in post-mortem human cerebellum tissue and cerebellum mass measured at autopsy. Although the human association was primarily mediated by altered DNA methylation at the 3rd CTCF binding site, it is hard to directly compare specific locations across species because there is incomplete sequence homology between the mouse and human genome at this region – for example the murine ICR contains four CTCF binding sites whereas the human ICR contains seven.

Epigenetic heterogeneity in the IGF2/H19 ICR is a normally distributed quantitative trait influenced by stochastic, genetic and environmental factors[19]. We observe considerably more inter-individual variation at this locus in human compared to mouse cerebellum, which may result from the fact that the HS mice used in the current study were maintained in a controlled laboratory housing environment and, although outbred, have limited genetic variation compared to humans. Indeed when we investigated cis-acting genetic factors we found no association with DNA methylation, suggesting that the reported DNA methylation differences are most likely to result from stochastic factors, trans-acting genetic factors, or unmeasured environmental factors. Of note, nutrition early in development is associated with decreased cerebellum mass and alterations to the IGF system within the cerebellum[20]. Although the HS mice in this study were given a standard controlled diet, they may have been exposed to subtle differences in their rearing – for example, maternal behavior and suckling prior to weaning – that could also have influenced DNA methylation.

The mouse cerebellum continues to grow for three weeks postnatally, at which point its circumference has reached adult size[21]. Corresponding expression data suggests that Igf2 mRNA levels are coordinated with specific growth patterns in the brain[17]. In this study we have looked at DNA methylation in adult mice to serve as a proxy for levels of gene expression during development. Investigations into the factors determining cerebellar growth and development are important as a growing body of work from human and animal studies suggests that the role of the cerebellum may extend beyond the regulation and coordination of motor function, to attention, perception, working memory and spatial orientation[14]. In humans, cerebellar abnormalities are among the most consistently reported structural findings in autism and attention deficit hyperactivity disorder[22], and progressive loss of cerebellar volume has been reported in childhood-onset schizophrenia and other types of psychosis[23].

The current study has a number of limitations. First, the absolute mean DNA methylation difference between the high and low cerebellar groups is relatively modest (~4%), but is consistent across the multiple adjacent CpG sites assessed in the CTCF2 and CTCF3 amplicons. Previous work has shown that Igf2 expression in the cerebellum is highly cell-type specific[17], and the small absolute difference we observe may result from much larger changes in a specific subpopulation of cells. Second, although the animals were maintained in a controlled environment, and we were careful to control for variables such as batch and net brain mass in our analyses, it is possible that our findings are biased by unmeasured variables or confounding influences such as maternal behavior. Third, the sodium bisulfite conversion method does not distinguish between methylated cytosines and hydroxymethylated cytosines, which may be particularly important here as hydroxymethylation is relatively enriched in the Purkinje cells of the cerebellum[24]. Fourth, we did not have access to RNA from the same samples, so we are unable to relate our epigenetic data to changes in steady-state mRNA levels. Our previous work[11] in humans, however, did not report an association between altered DNA methylation at this region and absolute levels of gene expression. It is possible that the observed DNA methylation changes result in allelic skewing of expression, rather than changes in absolute expression values or that DNA methylation may be a mark of gene expression changes occurring during development, rather than current gene expression levels. Fifth, we only analysed DNA methylation in the cerebellum itself, so we cannot assess the tissue-specificity of the findings. However, our previous work in humans looking at several brain regions[11] indicated that the association between IGF2/H19 DNA methylation and brain region weight was cerebellum-specific. Further work would be required to assess the tissue-specificity of the finding in mouse. Finally, we only assessed cerebellum mass and DNA methylation at one time point in adulthood, so could not examine changes occurring during development.

To conclude, we show that epigenetic variation across the Igf2/H19 ICR shows a significant negative relationship with cerebellar mass, with hypomethylation at specific CTCF binding motifs being associated with increased cerebellar mass, reflecting our previous work in human tissue. Future work will expand these analyses to look at brain-region and cell-type specific effects, to determine the genetic and environmental factors underlying the changes in Igf2/H19 DNA methylation and to explore the relationship between DNA methylation and gene expression and measures of behavior.