One of my last post-doc projects is finally out! This was quite a journey, with many collaborators & institutions involved, and a lot of sequencing (>150 libraries, > 12 billion reads). Co-led with the twitterless Marc-Aurèle Chay. A thread. https://www.nature.com/articles/s41467-021-21365-3
Back in 2013, when it all started, our lab had been studying the role of epigenetics in long-term effects of child abuse on human brain function & psychopathology for several years.
While our initial studies had focused on a single epigenetic substrate, DNA methylation, here we had the opportunity to work with @McGillGenome & @IHEC_Epigenomes to investigate DNA meth at base-resolution (WGBS), but also 6 histone marks (ChIPseq) & gene expression (RNAseq).
At that time, inspiring recent work had suggested that a non-canonical, non-CG form of DNA methylation strongly accumulates in mammalian brains during neurodevelopment (especially the first 15 years of life in human). See for example: https://science.sciencemag.org/content/341/6146/1237905.long
We therefore:
1) Hypothesized that child abuse may disrupt patterns of non-CG methylation in the human brain
2) Used tissues & psychological autopsy data from the Montréal Brain Bank @douglas_bank
3) Focused on the amygdala, a critical site for emotional regulation
1) Hypothesized that child abuse may disrupt patterns of non-CG methylation in the human brain
2) Used tissues & psychological autopsy data from the Montréal Brain Bank @douglas_bank
3) Focused on the amygdala, a critical site for emotional regulation
We searched for differentially methylated regions (DMR) associated with child abuse, comparing 2 contexts: classical/CG vs non-canonical/CAC (where non-CG meth is most abundant). To our surprise, DMRs at CACs were as frequent as those at CGs, suggesting significant plasticity.
Strikingly, CAC- and CG-DMRs affected genomic regions that appeared distinct at every level of analysis, including genic or methylomic features, individual histone marks, chromatin states (defined as specific combinations of histone marks), and GO categories. In a nutshell:
Among many other results and analyses, we also identified child abuse-associated changes at the level of gene expression, individual histone marks and chromatin states, altogether pointing towards alterations of immune-related and small GTPase signaling pathways.
Overall, we believe this work provides insights into human brain genomic regulation as a function of child abuse, and unravels a potentially important role for a non-canonical form of epigenetic embedding of life experiences.
Of course, classical limitations of such retrospective studies should be kept in mind (psychological autopsy, differentiating effects of child abuse vs psychopathology, sample size, etc).
All raw & processed data are available on GEO & the @NatureComms website.
All raw & processed data are available on GEO & the @NatureComms website.
Our study complements recent wonderful studies in mouse models, for example on:
1) Sensory deprivation / seizure activity: https://www.cell.com/cell/fulltext/S0092-8674(17)31141-8
1) Sensory deprivation / seizure activity: https://www.cell.com/cell/fulltext/S0092-8674(17)31141-8
Or 2) positive experiences (environmental enrichment): https://www.nature.com/articles/s41467-017-02748-x
This was a massive effort, and we are very grateful to all collaborators & co-authors, incl. @mcgillu @douglas_bank @incicnrs @CNRS @unistra @DrTonyKwan @ChristmannIpek &, last but not least, @NMechawar & Gustavo Turecki.
NB: We also have a few supplementary figures.
NB: We also have a few supplementary figures.
Building on this, my lab in France now investigates non-CG meth & other epigenetic processes in psychiatric disorders, combining mouse genetic tools & behavioral models, NGS & investigations of peripheral multi-omic biomarkers in human. https://inci-en.u-strasbg.fr/?page_id=1969