Histone Lysine Methylases and Demethylases in the Landscape of Human Developmental Disorders

Abstract

Histone lysine methyltransferases (KMTs) and demethylases (KDMs) underpin gene regulation. Here we demonstrate that variants causing haploinsufficiency of KMTs and KDMs are frequently encountered in individuals with developmental disorders. Using a combination of human variation databases and existing animal models, we determine 22 KMTs and KDMs as additional candidates for dominantly inherited developmental disorders. We show that KMTs and KDMs that are associated with, or are candidates for, dominant developmental disorders tend to have a higher level of transcription, longer canonical transcripts, more interactors, and a higher number and more types of post-translational modifications than other KMT and KDMs. We provide evidence to firmly associate KMT2C, ASH1L, and KMT5B haploinsufficiency with dominant developmental disorders. Whereas KMT2C or ASH1L haploinsufficiency results in a predominantly neurodevelopmental phenotype with occasional physical anomalies, KMT5B mutations cause an overgrowth syndrome with intellectual disability. We further expand the phenotypic spectrum of KMT2B-related disorders and show that some individuals can have severe developmental delay without dystonia at least until mid-childhood. Additionally, we describe a recessive histone lysine-methylation defect caused by homozygous or compound heterozygous KDM5B variants and resulting in a recognizable syndrome with developmental delay, facial dysmorphism, and camptodactyly. Collectively, these results emphasize the significance of histone lysine methylation in normal human development and the importance of this process in human developmental disorders. Our results demonstrate that systematic clinically oriented pathway-based analysis of genomic data can accelerate the discovery of rare genetic disorders.

Keywords: ASH1L; Developmental disorders; KDM5B; KMT2B; KMT2C; KMT5B; chromatin remodeling; histone lysine demethylase; histone lysine methyltransferase.

Figure 1

Variants in Histone Lysine Methyltransferases and Demethylases Are Frequent in Developmental Disorders and Haploinsufficiency Is Their Predominant Mechanism (A) The bar graph shows the proportions of postulated disease causing published heterozygous protein-truncating variants (PTVs) (in red) and protein altering variants (PAVs) (in blue) in known dominant developmental disorder (DD)-associated KMTs and KDMs. (B) A plot of the probability of being LoF intolerant (pLI) for all KMTs and KDMs. Red dots represent the pLI scores for known dominant DD-associated KMTs and KDMs, orange dots depict these scores from KMTs and KDMs that are candidates for involvement in dominant DDs, and green dots display the pLI scores for non-candidate KMT and KDM genes. The dotted line depicts the cut-off for defining the candidate genes (pLI > 0.9). (C) Proportion of canonical transcripts of known DD KMTs and KDMs from the total human exome (left donut graph), proportion of individuals with pathogenic variants in known KMT and KDM genes from the Deciphering Developmental Disorders (DDD) Study cohort (central donut graph), proportion of pathogenic, benign variants or variants of uncertain significance (VUS) in known KMT and KDM genes, and the percentage of variants in other KMTs and KDMs from the total number of KMT and KDM variants seen in the DDD cohort (right donut graph). The Venn diagram shows the distribution of the 120 rare high-quality variants that were detected in the DDD cohort in KMTs and KDMs not yet firmly associated with DDs. The green circle and the ellipse represent the number of variants according to their inheritance, the blue circle and the ellipse represent the number of variants according to their predicted protein effect, and the red circle and the ellipse represent the number of variants detected in candidate genes for dominant DDs and the other genes.

Figure 2

Variants of Interest Identified in This Study Locations of selected plausible candidate variants identified in this study are shown. Candidate genes for dominant DDs with DN PTVs are indicated in red font, and the other genes are in black font. The de novo PTVs in candidate KMTs and KDMs for dominant DD genes (KMT2B, KMT2C, ASH1L and KMT5B) (n = 8) are highly likely to be causal. We have also shown de novo protein-altering variants (PAVs) in candidate KMTs and KDMs for dominant DD genes (KMT2B, KMT2C, DOT1L, KDM3A, PRDM2, SETDB1) (n = 9) with limited evidence for causality at present (apart from those in KMT2B, which have been shown to cause early-onset dystonia). Inherited PTVs in candidate KMTs and KDMs (KDM3A and PRDM2) (n = 2) are shown. PTVs in these genes might cause non-penetrant phenotypes, or this might indicate that these genes tolerate haploinsufficiency, unlike what is suggested by their pLI scores. De novo PTVs in non-candidate KMTs and KDMs for dominant DDs (KDM5B and SETD1B) (n = 4) are also shown. These PTVs could be coincidental or might be acting as phenotype modifiers, or they could be non-penetrant in some individuals in the general population. Homozygous and compound heterozygous PTVs in KDM5B (n = 5) show that recessive histone-tail lysine-methylation disorders also exist.

Figure 3

Photographs from Individuals with Truncating Variants or Deletions of KMT2B, KMT2C, KMT5B, and KDM5B The numbers on each picture denote the corresponding individual in Table 1. Individual 1, with a KMT2B de novo PTV, has spare scalp hair, a large mouth, and no ear lobes; individual 2, with a KMT2C de novo PTV, has marked infra-orbital creases, down-slanting palpebral fissures, and a duplicated right thumb. Individual 3, with a KMT2C de novo PTV, has marked plagiocephaly and bilateral marked bulging just below the temporal region. Individual 8, with a KMT5B de novo PTV, has a broad and large forehead that has persisted over time. Individual 9, with a KMT5B de novo PTV, has a prominent forehead, thick ear lobes,a broad philtrum, an open mouth appearance, and synophrys, which is more noticeable in the more recent photograph. Individual 11, with a de novo KMT5B deletion, has a long and oval face, ptosis, prominent eyes, protruded ears, an open mouth, thick lips, and overlapping third and second toes. Individual 12, with a homozygous KDM5B PTV, has down-slanting palpebral fissures, a slightly bulbous nasal tip, low-hanging columella, a smooth philtrum, and thin upper and lower lips. He has bilateral camptodactyly of the fourth and fifth fingers. Individual 14, with a compound heterozygous KDM5B PTV, has a prominent metopic region, a high nasal bridge, a bulbous nasal tip, a smooth philtrum, thin lips, and a triangular ear with an absent superior crux of helix. He has also mild camptodactyly of the fourth and fifth fingers.

Figure 4

Comparison of Gene and Protein Properties between KMTs and KDMs That Are Known to Be Dominant for DDs or Are Candidate Genes for DDs and other KMTs and KDMs The comparisons were made with the data from UniProtKB, and a Mann-Whitney test in which an exact p value < 0.05 was considered to be significant was performed. The results are represented in dot plots as follows: (A) number of post-translational modifications (PTMs); (B) number of types of PTMs; (C) number of interactors; (D) length of canonical transcripts; (E) number of PTMs per 100 amino acids of canonical transcripts in KMTs and KDMs; and (F) the median reads per kilobase per million (RPKM) for candidate and non-candidate KMTs and KDMs in brain structures; significant differences across several stages are evident. Black (A–E) or colored (F) dots denote KMT and KDM genes that are known or are predicted to be candidates for dominant DDs. Unfilled dots (A–E) or colored triangles (F) denote KMT and KDM genes that are predicted not to be candidates for dominant DDs. The colors in (F) correspond to the legend for the brain structures provided in the figure. The longer horizontal lines in all the graphs represent the respective medians, the shorter horizontal lines indicate the inter-quartile ranges, and the p values are given at the top of each graph, where relevant.