Altered Gene-Regulatory Function of KDM5C by a Novel Mutation Associated With Autism and Intellectual Disability

Abstract

Intellectual disability (ID) affects up to 2% of the population world-wide and often coincides with other neurological conditions such as autism spectrum disorders. Mutations in KDM5C cause Mental Retardation, X-linked, Syndromic, Claes-Jensen type (MRXSCJ, OMIM #300534) and are one of the most common causes of X-linked ID. KDM5C encodes a histone demethylase for di- and tri-methylated histone H3 lysine 4 (H3K4me2/3), which are enriched in transcriptionally engaged promoter regions. KDM5C regulates gene transcription; however, it remains unknown whether removal of H3K4me is fully responsible for KDM5C-mediated gene regulation. Most mutations functionally tested to date result in reduced enzymatic activity of KDM5C, indicating loss of demethylase function as the primary mechanism underlying MRXSCJ. Here, we report a novel KDM5C mutation, R1115H, identified in an individual displaying MRXSCJ-like symptoms. The carrier mother's cells exhibited a highly skewed X-inactivation pattern. The KDM5C-R1115H substitution does not have an impact on enzymatic activity nor protein stability. However, when overexpressed in post-mitotic neurons, KDM5C-R1115H failed to fully suppress expression of target genes, while the mutant also affected expression of a distinct set of genes compared to KDM5C-wildtype. These results suggest that KDM5C may have non-enzymatic roles in gene regulation, and alteration of these roles contributes to MRXSCJ in this patient.

Keywords: KDM5C/SMCX/JARID1C; X-linked intellectual disability; autism spectrum disorders; chromatin; histone demethylase; mutation analysis; neuroepigenetics.

FIGURE 1

KDM5C R1115H mutation in family UM1. (A) Pedigree of family UM1. (B) Sanger sequencing of genomic DNA from lymphoblastoid cell lines generated from proband (UM1 III-3) and father (UM1 II-1). (C) Multi-species conservation alignment of KDM5C homologs. The following RefSeq sequences were used for alignment: human NP_001140174.1, orangutan NP_001125719.1, rhesus XP_014982969.1, mouse NP_038696.2, rat XP_008771368.1, dog NP_001041497.1, elephant XP_010598233.1, frog NP_001072719.1, fugu XP_003963594.1. (D) Conservation alignment of human KDM5 family proteins, KDM5A-D. The following RefSeq sequences were used for alignment: KDM5A NP_001036068.1, KDM5B NP_006609.3, KDM5C NP_001140174.1, KDM5D NP_001140177.1. (E) Schematic of human KDM5C protein and 26 reported mutations associated with KDM5C-XLID. Missense mutations are depicted above the protein, while nonsense and frame-shift mutations are depicted beneath the molecule. JmN, jumonji-N domain; ARID, AT-rich interacting domain; PHD, plant homeodomain box domain; JmjC, jumonji-C catalytic domain; ZF, zinc finger domain; PLU-1, PLU-1-like domain.

FIGURE 2

KDM5C R1115H has largely intact enzymatic activity. (A,B) In vitro demethylation assay. Full-length KDM5C wildtype (WT) or mutant (R1115H) were affinity-purified from Sf9 or Hi5 insect cells infected with baculoviruses and subjected to demethylation assay using synthetic histone peptide carrying (A) H3K4me3 or (B) H3K4me3 + K9me3. Relative fluorescence values of formaldehyde, produced by the demethylation reaction, are normalized by the amount of purified proteins in the reaction and plotted. Error bars represent SEM of a technical triplicate. (C) In situ demethylation assay. Expression constructs of strep-KDM5C WT, mutant R1115H, or catalytically inactive H514A were transiently transfected into U2OS cells and stained with antibodies for Strep (red) and H3K4me3 (green). Compared to untransfected cells, significant decrease of K4me3 was found in WT- and R1115H over-expressing cells. Demethylation activity was completely abrogated by the H514A mutation. Nuclei were stained with DAPI. (D) H3K4me levels in lymphoblastoid cell lines from proband (UM1-III-3, KMD5C R1115H) and father (UM1-II-1, KDM5C WT) were measured by quantitative Western blot analysis. H3K4me signals were normalized to pan-H3 signal (n = 3, Mean ± SEM). No noticeable change was found between the two cell lines.

FIGURE 3

KDM5C R1115H is stable in cells. Lymphoblastoid cell lines from father (UM1-II-1, KDM5C WT) and proband (UM1-III-3, KDM5C R1115H) were treated with a cycloheximide (CHX) time course from 0 to 28 h. The KDM5C levels upon treatment of the cell lines were measured by quantitative Western blot. Relative fluorescence unit normalized by GAPDH signals were plotted (n = 3, Mean ± SEM).

FIGURE 4

RNA-sequencing of primary cultured neurons expressing WT- and mutant KDM5C. (A) Schematic of experimental procedures. (B) Normalized expression values of all genes in Vector-WT, Vector-R1115H, and Vector-H514A comparisons. (C) Fold change of down-regulated genes by WT-, R1115H-, and H514A-overexpression plotted against significance. WT up (green) and down (purple) DE genes were plotted for each mutant condition. Horizontal line indicated significance cutoff of p < 0.01. (D) Overlap of significantly down- and up-regulated genes (p < 0.01).

FIGURE 5

Ontology analysis of KDM5C-regulated genes. Down-regulated genes by KDM5C-WT (A), KDM5C-R1115H (B), and KDM5C-H514A (C) were subjected to GO analysis using DAVID (Functional Annotation Bioinformatics Microarray Analysis). Representative two GO terms each from the top four most enriched annotation clusters were presented with p-values and cluster enrichment score. See Supplementary Table S2 for full list of GO-terms.