Authors: Esther R. Berko, Masako Suzuki, Faygel Beren, et al.
Publication Date: May 29, 2014
This week’s paper deals with trying to trace the biological causes of Autism Spectrum Disorder (ASD). The authors begin by noting that one of the causes associated with ASD is the age of the mother, although the reasons for this increased risk are unknown. What is known is that the eggs of older women are more prone to chromosomal abnormalities, and so this has been suggested as a likely reason for this association between parental age and ASD. However, as the authors write, “age is also associated with a loss of control of epigenetic regulatory patterns that govern gene expression,” which suggests epigenetic dysregulation as a second potential mechanism. Thus, for this paper the authors tested both possibilities.
This effort to distinguish between genetic or epigenetic causes of ASD is the first reason for the selection of this paper as the paper of the week; the second reason is the extensive descriptions the authors give about the methods they use to test between these two mechanisms. For anyone interested in epigenetics, this discussion of the cutting edge of the technical side of epigenetics research can only be helpful.
Genetic mutations have long been proposed as the predominant cause for ASD, but the explanation of epigenetic mechanisms has recently gained credence as a cause for ASD. The authors cite three recent studies in particular which support epigenetic dysregulation as a potential mechanism in the incidence of ASD. A 2012 paper reported the discovery of distinctive chromatin features in the brains of subjects with ASD. Authors of a paper published in 2013 tested blood leukocytes and found differences in DNA methylation between a monozygotic twin affected with ASD and their unaffected twin. A second paper published in 2013 found differences in DNA methylation from subjects with ASD and subjects without ASD.
To test whether ASD is the result of these genetic or epigenetic causes, the authors tested “homogeneous ectodermal cell types” from 47 individuals with ASD compared with 48 typically developing (TD) controls born to mothers of ≥35 years. Genome-wide tests were then performed on these cells to look for unusual chromosome numbers to test the hypothesis of genetic mutations; epigenome-wide analyses (EWASs) were used to test DNA methylation patterns in regards to the hypothesis of epigenetic dysregulation. This study, the authors note, “represents the largest epigenome-wide analysis to date testing a single cell type in ASD.”
The choice of this specific cell type served a couple of different purposes. First, the type of cell was selected because comes from the same developmental origin as brain cells. Second, the selection of this specific cell type minimizes the problem of “mixed cellularity” which has been already identified as a problem with EWASs.
The authors also addressed other such problems with EWASs, the biggest issue being “that the generally small changes in DNA methylation found may not be substantially in excess of the noise introduced by technical or biological effects influencing DNA methylation that have no relationship to the phenotype being tested.” To achieve what the authors call “the currently necessary level of stringency for EWAS studies,” the authors incorporated parallel SNP genotyping and “Surrogate Variable Analysis” (SVA) to account for these different possible sources of variability, as well as stringent pre-processing of the microarray data that was gathered and iterative use of this preprocessing data as means to focus in on only those regions of the DNA sequence that are being differentially methylated (i.e., to avoid ‘false positives’). Through these different efforts, the authors reduced the impact of these methodological issues due to “cell type and subpopulation heterogeneity, chromosomal aneuploidy, copy number variability, genetic polymorphism, age, sex and technical influences.” The authors go into considerable detail as to how these different approaches resolve these issues, so anyone interested should the relevant sections of the original paper.
Using these rigorous and intensive methods to test the competing hypotheses of mutations in DNA versus epigenetic dysregulation as increasing the risk for ASD, the authors discover 15 differentially methylated regions (DMRs) at 14 genes which distinguished the ASD and TD samples. From subsequent analysis of these DMRs, the authors conclude that “DNA methylation patterns are dysregulated in ectodermal cells” in individuals with ASD, but did not find evidence of chromosomal abnormalities in those same DMRs. In their own words, the authors conclude that “of the two mechanisms we originally proposed for AMA causing ASD, covert aneuploidy occurring at detectable levels (≥20%) is not as likely to be involved as epigenetic dysregulation.”
Interestingly, though, the genes in these DMRs are those already associated with ASD, which means that instead of genetic mutations this analysis reveals “a perturbation by epigenomic dysregulation of the same networks compromised by DNA mutational mechanisms.” In other words, the genes previously associated with ASD are still implicated in ASD, but through epigenetic dysregulation rather than through mutations in DNA sequences. However, the exact pathways of this epigenetic dysregulation are still unclear. Given the results of their study, the authors suggest aging parental gametes, environmental influences during embryogenesis, or mutations of the chromatin regulatory genes implicated in ASD as the most likely possible environmental factors in this epigenetic dysregulation.
Some additional points of interest about this study are worth mention:
First, mosaicism in this context usually refers to differences in chromosomal makeup between cell populations within the same individual (the “covert aneuploidy” just mentioned). As this was not found to be significant, the “mosaicism” in the title of this paper refers not to chromosomal differences between cell types but rather “the presence of a mosaic subpopulation of epigenetically-dysregulated, ectodermally-derived cells in subjects with ASD.”
Second, the authors note prior epigenetic studies of ASD had used mixed cell types, which may have limited the ability to detect the effects found by the authors and their use of homogeneous ectodermal cell types.
Third, the authors note that their study implicated the same gene in epigenetic dysregulation (OR2L13) as found in two previous studies associated with altered DNA methylation in individuals with ASD. This replication suggests this gene is “especially labile in ASD in terms of DNA methylation and expression.”
Fourth, the authors observe that while the epigenetic changes they observed from a cohort of subjects born to mothers with AMA may be the result of the aging of the mother’s egg, the sperm of the fathers—who are likely as old as the mothers—may also be experiencing epigenetic changes of their own which are contributing to the epigenetic dysregulation observed in their study; as this was not controlled for in their study, subsequent efforts should include such controls.
Finally, given the ages of the parents and the increased probability of mutational events, the authors allow that the observed epigenetic dysregulation may actually be a secondary effect of the mutations in the genes involved, and not actually the cause but rather a symptom of ASD. What is most interesting to me about this suggestion is the authors’ recommendation that “combined genetic and epigenetic analyses of the same subjects will be needed to test these possibilities.” I have written about the substantial benefits of this combination of epigenetics and genomics before in terms of the need to identify the impacts of both gene sequence and three-dimensional structure on gene expression in general, and as it pertains to expanding the scope of disease phenotypes which are amenable to drug discovery in particular, and this paper provides yet another concrete example of the need to combine both epigenetics and genomics.
 Shulha HP, Cheung I, Whittle C, Wang J, Virgil D, et al. (2012) Epigenetic signatures of autism: trimethylated H3K4 landscapes in prefrontal neurons. Arch Gen Psychiatry 69: 314-324.
 Wong CC, Meaburn EL, Ronald A, Price TS, Jeffries AR, et al. (2013) Methylomic analysis of monozygotic twins discordant for autism spectrum disorder and related behavioural traits. Mol Psychiatry.
 Ladd-Acosta C, Hansen KD, Briem E, Fallin MD, Kaufmann WE, et al. (2013) Common DNA methylation alterations in multiple brain regions in autism. Mol Psychiatry.