ORIGINAL ARTICLE ONLINE FIRST Epigenetic Signatures of Autism Trimethylated H3K4 Landscapes in Prefrontal Neurons Hennady P. Shulha, PhD; Iris Cheung, PhD; Catheryne Whittle; Jie Wang, PhD; Daniel Virgil, BS; Cong L. Lin, BS; Yin Guo, MD; Andree Lessard, PhD; Schahram Akbarian, MD, PhD; Zhiping Weng, PhD Context: Neuronal dysfunction in cerebral cortex and Results: Subjects with autism showed no evidence for other brain regions could contribute to the cognitive and generalized disruption of the developmentally regu- behavioral defects in autism. lated remodeling of the H3K4me3 landscape that de- fines normal prefrontal cortex neurons in early infancy. Objective: To characterize epigenetic signatures of au- However, excess spreading of H3K4me3 from the tran- tism in prefrontal cortex neurons. scription start sites into downstream gene bodies and up- stream promoters was observed specifically in neuronal Design: We performed fluorescence-activated sorting and chromatin from 4 of 16 autism cases but not in controls. separation of neuronal and nonneuronal nuclei from post- Variable subsets of autism cases exhibit altered H3K4me3 mortem prefrontal cortex, digested the chromatin with peaks at numerous genes regulating neuronal connec- micrococcal nuclease, and deeply sequenced the DNA tivity, social behaviors, and cognition, often in conjunc- from the mononucleosomes with trimethylated H3K4 tion with altered expression of the corresponding tran- (H3K4me3), a histone mark associated with transcrip- scripts. Autism-associated H3K4me3 peaks were tional regulation. Approximately 15 billion base pairs of H3K4me3-enriched sequences were collected from 32 significantly enriched in genes and loci implicated in neu- brains. rodevelopmental diseases. Setting: Academic medical center. Conclusions: Prefrontal cortex neurons from subjects with autism show changes in chromatin structures at hun- Participants: A total of 16 subjects diagnosed as hav- dreds of loci genome-wide, revealing considerable over- ing autism and 16 control subjects ranging in age from lap between genetic and epigenetic risk maps of devel- 0.5 to 70 years. opmental brain disorders. Main Outcome Measures: Identification of genomic Arch Gen Psychiatry. loci showing autism-associated H3K4me3 changes in pre- Published online November 7, 2011. frontal cortex neurons. doi:10.1001/archgenpsychiatry.2011.151 UTISM SPECTRUM DISOR- higher cognition and social communica- ders comprise a group of tion, very little is known about the mo- complex and etiologically lecular pathology of PFC neurons in the heterogeneous illnesses. autistic brain. The genetic risk architec- Epigenetic dysregulation of DNA meth- Ature remains unknown for most patients, ylation and histone modifications could Author Affiliations: Program in and fewer than 10% of subjects on the au- play a prominent role in the pathophysi- Bioinformatics and Integrative tism spectrum harbor rare structural DNA ology of autism and related disease.4-11 This Biology (Drs Shulha, Wang, and variations and mutations with strong pen- hypothesis is in part based on the link be- Weng) and Brudnick etrance. Neurons residing in the prefron- tween autism and deleterious mutations Neuropsychiatric Research tal cortex (PFC) and other cortical asso- in genes considered of pivotal impor- Institute (Drs Cheung, Guo, ciation areas in autistic subjects are affected tance for regulation of chromatin struc- and Akbarian, Mss Whittle and by subtle defects in connectivity pat- ture and function, such as MECP27 and Lin, and Mr Virgil), University terns, cytoarchitecture, and other struc- other methyl-CpG-binding proteins,12,13 of Massachusetts Medical 1-3 14 School, Worcester; and tural alterations. While the role of such the histone deacetylase HDAC4, the his- Maryland Psychiatric Research disordered neural circuitry in the con- tone H3 lysine 9 (H3K9)–specific meth- Center, University of Maryland, text of neurodevelopmental disease is well yltransferase EHMT1/KMT1D/GLP,15 and Baltimore (Dr Lessard). understood, including impairment of the H3K4 demethylase JARID1C/KDM5C/ ARCH GEN PSYCHIATRY PUBLISHED ONLINE NOVEMBER 7, 2011 WWW.ARCHGENPSYCHIATRY.COM E1 Downloaded from www.archgenpsychiatry.com at University of Massachusetts Med School, on November 29, 2011 ©2011 American Medical Association. All rights reserved. SMCX.16 In particular, the trimethylated form of H3K4 ylation landscapes of terminally differentiated neurons across (H3K4me3) is primarily located at transcription start sites the lifespan, including developmental periods and disease con- (TSSs) and linked to the serine 5-phosphorylated, ini- ditions such as autism,26,27 without being affected by shifts in the tiation form of RNA polymerase II, thereby providing a neuron-glia ratio, which could greatly confound interpretation docking site at the 5Ј end of genes for chromatin remod- of conventional chromatin studies from tissue homogenates. We then digested chromatin with micrococcal nuclease, extracted eling complexes that mostly facilitate (but in some cases 17,18 the mononucleosome fraction, and performed ChIP with an repress) transcription. Epigenetic fine-tuning of antibody against H3K4me3 followed by deep sequencing. We H3K4me3 appears to be particularly important for neu- obtained genome-wide maps of the H3K4me3 mark for NeuNϩ ronal health. For example, neuronal differentiation is nuclei from the PFC of 16 subjects diagnosed as having autism dependent on H3K4 trimethylation mediated by the spectrum disorders (aged 2-60 years; mean age, 17.6 years; 4 fe- mixed-lineage leukemia (MLL) methyltransferase and males; supplemental Table S1, http://zlab.umassmed.edu/zlab transcriptional activation of RE-1 silencing transcrip- /publications/ShulhaAGP2011.html) and 10 age-matched con- tion factor (REST)–sensitive genes.19,20 Furthermore, trol subjects (aged 2.8-69 years; mean age, 19.7 years; 3 females). hippocampal learning and memory require MLL1- In addition, we included 4 control subjects younger than 2 years mediated H3K4 trimethylation at growth- and plasticity- (1 female) to investigate whether H3K4me3 profiles of autis- 21,22 tic subjects were more similar to those of normal infants than regulating genes. to those of older control subjects. Nine of the 14 control samples Recently, we presented the first neuronal-specific epi- were generated in our earlier study.23 We also sequenced an genomes from human brain and provided evidence that input library from the neuronal nuclei of 1 control subject, fol- nuclei of PFC neurons from normal infants (aged Ͻ1 year) lowing all the steps of ChIP with deep sequencing but omit- showed an excess of H3K4me3 at hundreds of loci com- ting the anti-H3K4me3 antibody. Altogether, 333 million 36- pared with older brains.23 This large-scale remodeling of nucleotide reads were obtained for 31 neuronal samples, among the H3K4me3 landscape in chromatin of young neu- which 283 million reads mapped to unique locations in the ref- rons speaks for epigenetic vulnerability of the immature erence human genome and an additional 32 million reads mapped to multiple locations in the genome. Furthermore, we PFC, an intriguing hypothesis when viewed from the neu- − rodevelopmental perspective of autism. Therefore, the aim performed ChIP followed by deep sequencing on NeuN chro- matin of 4 of the autism cases and 2 additional control sub- of this study was to compare, for the first time to our jects (supplemental Table S2). knowledge, the H3K4me3 epigenomes of the PFC neu- rons of autistic individuals with a panel of controls across a wide age range from infancy to 70 years. Our findings RESULTS indicate that a subset of autistic individuals are affected by loss or excess of H3K4me3 at hundreds of loci, in con- GENOME-WIDE CORRELATION OF H3K4me3 junction with dysregulated expression of transcripts im- OCCUPANCY IN ANNOTATED PROMOTERS plicated in neuronal communication and social and other higher-order behaviors. We propose a new disease model Consistent with the knowledge that H3K4me3 is a his- for PFC neurons that involves changes in the H3K4me3 tone mark sharply enriched around the 5Ј end of genes, landscape in hundreds of loci. These changes are highly a mean (SD) of 59% (9%) of reads in the 14 control variable between different patients and point to a com- samples mapped within RefSeq promoters (http://www plex interaction between the genetic and epigenetic risk .ncbi.nlm.nih.gov/RefSeq/). Similarly, a mean (SD) of 57% architectures that affect autism spectrum disorders. (11%) of reads in the autism samples mapped in proxi- mal promoters (supplemental Table S1). There is no sta- METHODS tistical difference in the percentage of tags mapping to promoters between autism and control samples, indicat- CHROMATIN IMMUNOPRECIPITATION ing that autistic patients do not show global displace- WITH DEEP SEQUENCING ment of H3K4me3 away from promoters. In sharp con- trast, fewer than 4% of reads from the input library mapped The eAppendix (http://www.archgenpsychiatry.com) has a de- to promoters. Postmortem confounds—tissue pH and tailed description of the methods and techniques used. Sorted postmortem interval—do not show significant correla- neuronal (NeuNϩ) and nonneuronal (NeuN−) nuclei were pro- tions with the percentage of TSS-proximal tags (corre- cessed for anti-H3K4me3 chromatin immunoprecipitation lation coefficients, −0.12 and −0.19, respectively; t tests, (ChIP), and resulting DNA libraries were deep sequenced with 2-tailed P=.55 and .31, respectively). an Illumina