The Potential Role of a Retrotransposed Gene and a Long Non-Coding RNA in Regulating an X-Linked Chromatin Gene (KDM5C): Novel Epigenetic Mechanism in Autism

Background: A growing body of evidence supports the potential role of the circadian system and chromatin remodeling genes in autism. Considering the heterogeneity in autism, and the complex nature of the epigenetic landscape, identification of biologically relevant epigenetic factors requires reducing heterogeneity using proper subtyping. Previously, we have reported that members of the KDM5 family of genes are involved in the circadian machinery. X chromosome inactivation (XCI) is one of the biological processes highly dependent on epigenetic steps. There is also a crosstalk between DNA methylation and gene regulatory processes, including alternative splicing. We have previously shown the trend of a higher degree of XCI skewness (XIS) in autistic females compared to controls.

Objectives: To investigate the role of circadian genes and their potential mis-splicing in autism.

Methods: In the present study, we used human blood samples (n=39) and a combination of whole genome (Exon array) and gene-targeted assays (PCR, TaqMan, Sanger sequencing) to characterize alternatively spliced KDM5 isoforms in controls and autism, stratified by XCI profiles (with or without XIS). Mouse brain samples were used to study isoforms circadian dependency.

Results: Several isoforms, including previously uncharacterized microexons and intron-retained variants, were identified for each gene. Overall, we found that the X-linked gene (KDM5C) undergoes a more complex splicing process than the autosomal genes (KDM5A and KDM5B). We also identified a KDM5C isoform (KDM5C-3’UTR-lncRNA) containing a novel 3’UTR originated from a retrotransposed gene of an autosomal methyltransferase (SUV39H2). This 3’UTR shows 83% sequence homology with lncRNAs and is located 32kb downstream of KDM5C. The KDM5C-3’UTR-lncRNA isoform was differentially expressed in autistic females with XIS compared with controls (Figure.1). Intriguingly, the XIS and noXIS autistic females presented different expression profiles, which indicates that such differences would have remained undetected without the subtyping. Since no differences were seen between the XIS and noXIS controls, the detected expression differences for KCM5C-3’UTR-lncRNA may be related to the underlying mechanism of autism, and not XIS itself.

This finding would ideally need to be studied in human brain samples collected at different circadian time-points, but it was impossible to do so. Therefore, we attempted to gather basic information about circadian dependent expression fluctuation of Kdm5 isoforms in wild type mouse brain samples (n=27). We characterized alternatively spliced isoforms of the Kdm5 genes and assessed their expression level in samples collected at different circadian time-points. In doing so, we showed that some Kdm5 isoforms follow a circadian oscillation pattern of expression (Figure.2).

KDM5C plays a crucial role in balancing histone H3K4 methylation states. The identified retro-SUV39H2 originated lncRNA also shows H3K4 marks. LncRNA RNA-seq data from UCSC browser confirmed the expression of lncRNA from the region encompassing the retro-SUV39H2 gene abundantly, in multiple tissues, including brain and blood.

Conclusions: This study provides the first evidence and a suggestive model for the potential role of retrotransposed elements in autism through linking methylases and demethylases, two functionally complementary components of chromatin remodeling, which may collectively contribute to disease etiology through lncRNAs.