Detection of Modified Histones from Oral Mucosa of a Patient with DYT-KMT2B Dystonia

Dystonia is a movement disorder characterized by sustained or intermittent muscle contractions causing abnormal movements and postures [Albanese et al., 2013]. The advent of next-generation sequencing has resulted in the discovery of novel dystonia genes, which improved our understanding of gene-associated phenotypes and the pathogenesis of inherited dystonia [Lange et al., 2021]. DYT-KMT2B, also known as DYT28 is an early-onset hereditary dystonia caused by the mutations of KMT2B [Zech et al., 2016]. Typically, the patients initially present with focal, lower-limb dystonia that subsequently evolves into generalized dystonia with prominent cranial, cervical, and laryngeal involvement. In addition, dysmorphism, short stature, intellectual disability, oculomotor abnormalities, and psychiatric comorbidities have also been reported [Cif et al., 2020]. KMT2B gene encodes a ubiquitously expressed epigenetic regulator that catalyzes histone H3 methylation at lysine 4 (H3K4) [Klonou et al., 2021]. H3K4 demethylation and DNA methylation potentiate histone remodeling, thereby causing transcriptional repression in target genes [Meissner et al., 2008]. As KMT2B transcription level is markedly decreased in the patients with DYT-KMT2B, a loss-of-function mechanism is postulated as the pathogenesis [Zech et al., 2016; Kawarai et al., 2018]. Hence, we investigated the decreased KMT2B resultant products using an oral mucosal sample from a patient with DYT-KMT2B. Furthermore, we investigated if H3K4 methylation was altered in patients by reanalyzing data obtained from whole genome bisulfite sequencing (WGBS).

Genomic DNA from the peripheral blood was captured using the Agilent SureSelect Human All Exon V4 or V5 kit and sequenced on an Illumina HiSeq 2500 with 101 bp paired-end reads. Obtained reads were aligned to the GRCh37 reference genome using NovoAlign (Novocraft Technologies, v3.02). Duplicated reads were removed using Picard (version 1.55), and local realignment and base quality recalibration were performed using a Genome Analysis Toolkit (GATK, version 2.7). The candidate of the causative mutation was confirmed by Sanger sequencing. An amino acid substitution was predicted to be pathogenic by in silico analysis tools, SIFT (https://sift.bii.a-star.edu.sg), and PolyPhen-2 (http://genetics.bwh.harvard.edu/pph2/).

Eating and drinking were prohibited for 30 min before the sample collection. After washing the mouth three times, oral mucosa was swabbed from two distinct sites from the buccal mucosa using NIPRO Sponge Swab^® (NIPRO, Osaka, Japan). Histone proteins were extracted from swabs that were separated and handled by acid extraction [Sugeno et al., 2016]. The profile of the modified histone profile was analyzed by Western immunoblotting using anti-H3K4me3 (ab8580; Abcam, Cambridge, MA, USA) and anti-H3 (ab1791; Abcam) antibodies. The study was reviewed and approved by the Ethics Committee of Tohoku University (certification number: 2021-1-748).

RNA sequencing dataset was obtained from the GEO dataset (GSE186805) [Gu et al., 2021]. FASTQ files were created using Fasterq-dump. Adapter trimming and quality control metric examination were processed on Trimm Galore (https://www.bioinformatics.babraham.ac.uk/projects/trim_galore/). The reads were mapped onto the GRCh38, and transcripts per kilobase million were counted using the RSEM package based on the STAR method [Li and Dewey, 2011]. Log10 transformed transcript per kilobase million values with a 0.1 correction constant were used for harmonic gene testing (MetaCycle package) [Wu et al., 2016]. Significant changes were determined as the gene presented both p value of <0.1 and a relative amplitude of >0.1.

WGBS dataset includes healthy controls and patients with DYT-KMT2B obtained from GEO dataset (GSE199836) [Lee et al., 2022]. Adapter trimming and quality control metric examinations were processed using Trimm Galore. Mapping onto GRCh38, deduplication, calculation of coverage and insert size metrics, extraction of CpG methylation values, and generation of genome-wide cytosine reports (CpG count matrix) were performed using Bismark software v.0.21 (http://www.bioinformatics.babraham.ac.uk/projects/bismark/). Data normalization, batch effect correction, and extraction of differentially methylated CpG sites were performed using the RnBeads package 2.10.0 [Muller et al., 2019]. Enrichment analysis for genomic region sets was conducted on Locus Overlap Analysis (LOLA), which was performed for the top 1,000 genome-tiling regions, with the latter being selected based on combined ranking score [Sheffield and Bock, 2016]. The following LOLA reference databases were utilized for the following analyses: cistrome_epigenome [Liu et al., 2011], codexes (https://codex.stemcells.cam.ac.uk), encode_tfbs, sheffield_dnase [Sheffield et al., 2013], and ucsc_features. The DMRichR, which utilizes the dmrseq and bsseq algorithms (https://github.com/ben-laufer/DMRichR), was used [Laufer et al., 2021], and regions containing at least five different CpGs within 1 kb with a minimum methylation difference of 10% were selected to identify differentially methylated regions.

We report a case of a 23-year-old Japanese male born to nonconsanguineous parents with a low birth weight (2,306 g at 39 weeks of gestation, between 3–10 percentile in the Japanese newborn population) by vacuum-assisted delivery. At age 6, his mother noticed abnormal posture of the right foot and increasing walking difficulty, and he was admitted to the pediatric clinic. Neither diurnal fluctuation nor joint contracture was observed. Gross intellectual and motor skill milestones (e.g., rolling over, sitting, and walking) were within normal limits, but his weight and height were consistently maintained at >2 standard deviations below the mean for age and did not increase until adulthood. No family history of neurological disorders was reported. Upon initial assessment, his motor symptoms were interpreted as focal dystonia, and the underlying etiology was extensively evaluated. Workup, such as routine blood tests, cerebrospinal fluid analyses, and endocrinological tests, did not provide any diagnostic markers. Cranial magnetic resonance imaging (MRI) and 18F-fluorodeoxyglucose positron emission tomography imaging were also normal (data not shown). Pharmacological intervention (e.g., levodopa, bromocriptine, trihexyphenidyl, and biperiden) has been tried, but none of them except for levodopa showed sustained, beneficial effects. An initial diagnosis of early-onset, focal limb dystonia was made based on these observations. Four years after the onset, disabling dystonia spread to the left lower limb, and in his early 20s, a weird tremulous movement occurred in his left hand while performing activities of daily life. To identify possible genetic predisposition, whole exome sequencing (WES) of proband-parent trios was performed, and finally, de novo heterozygous frameshift mutation c.5501_5502insT (p.His1835Profs*137) in KMT2B, a causal gene for DYT-KMT2B, was identified (shown in Fig. 1a, b).

At age 23, the patient was transferred from the pediatric department to an adult neurology division. On consultation, neurological examination revealed dystonic posturing of the right lower limb with foot inversion (equinovarus) accompanied by toe flexion. In addition, a dystonic tremor was induced by left elbow flexion (online suppl. video 1; see https://doi.org/10.1159/000530625). Shuffling gait and dragging of the foot during walking were observed. Other abnormalities (e.g., cognitive dysfunction, pyramidal sign, or cerebellar ataxia) were not detected. After cessation of dopaminergic and anticholinergic agents, mild rigidity and bradykinesia were observed in the limbs, and the walking difficulty worsened. Re-examination of cranial MRI did not show any abnormality (shown in Fig. 2a), but dopamine transporter imaging with 123I-β-CIT (2β-carbomethoxy-3β-[4-iodophenyl] tropane) single photon emission computed tomography (SPECT) revealed slightly decreased tracer uptake in the bilateral striatum, more pronounced in the posterior, left region (shown in Fig. 2b). Neopterin (13.80 nM; normal range: 7.3–31.6 nM) and biopterin (13.90 nM; normal range: 7.9–25.8) levels in the cerebrospinal fluid were normal.

Since KMT2B is widely expressed in a variety of tissues, easily accessible oral mucosa was selected to evaluate the KMT2B enzymatic activity. Trimethylated H3K4 (H3K4me3) is the end-product of KMT2B but is also produced by other histone methyltransferases. Among them, MLL1/KMT2A is known to act as a regulator of circadian gene oscillations [Katada and Sassone-Corsi, 2010]. Therefore, we first examined the possibility of daily H3K4 methylation modulator fluctuation. To this end, the RNA-seq dataset of oral mucosa from healthy volunteers was reanalyzed, and diurnal changes in expression levels of methyltransferase (writer) and demethylase (eraser) for H3K4 were evaluated [Gu et al., 2021]. Both six writers and six erasers did not show significant cyclic changes in the expression (shown in Fig. 3), indicating that the level of H3K4me3 does not depend on circadian changes and is not limited by the wall-clock time for sample collection. Following these preliminary experiments, histone proteins were extracted from the oral mucosa of healthy controls and the patient with DYT-KMT2B using a modified acid-extraction method and subjected to Western blotting [Sugeno et al., 2016]. A specific H3K4me3 band was successfully detected in all samples, but its expression level was considerably decreased in the patient (shown in Fig. 4). Although only one case was studied due to the rarity of the disease, the result suggests that methyltransferase activity of KMT2B tends to be decreased in DYT-KMT2B.

In terms of downstream epigenetic alteration caused by loss of function of KMT2B, direct evidence such as chromatin immunoprecipitation using KMT2B or H3K4me3 antibodies has not been investigated, but two studies using WGBS data from peripheral blood samples of patients with DYT-KMT2B showed an elevated DNA methylation level in the affected individuals [Lee et al., 2022; Mirza-Schreiber et al., 2021]. The dataset from one of these studies was publicly available and was reanalyzed with the standard Bismark pipeline, followed by a matching investigation in the LOLA database worked on RnBeads. Figure 5a shows the enrichment results of the 1,000 highest-ranking hypermethylated promoter regions in the patients with DYT-KMT2B compared with healthy controls. In the “cistrome_epigenome” category of the LOLA core database, several H3K4me3 profiles were in good agreement with hypermethylated regions in DYT-KMT2B. This H3K4me3-methylated DNA association is consistent with the hypothesis that reduced H3K4me3 leads to DNA hypermethylation. A subsequent enrichment analysis using genes with 500 highest-ranking hypermethylated promoters demonstrated that neuronal function was not associated with nominated gene ontology (GO) biological processes, but a known KMT2B-related assembly, “flagellated sperm motility,” was ranked highest (shown in Fig. 5b).

Hereditary dystonia has a wide range of clinical manifestations, and phenotypic overlap can occur even in patients with different genetic defects, making the diagnosis often challenging [Cif et al., 2020]. DYT-KMT2B is commonly associated with various morphological abnormalities and developmental delays/cognitive impairments, although these additional characteristics may not be obvious in some patients [Meyer et al., 2017; Zech et al., 2017; Kawarai et al., 2018]. Neuroimaging findings associated with DYT-KMT2B include signal changes in the globus pallidum on brain MRI and decreased striatal accumulation on DAT imaging, but these findings are not disease specific and are not observed in all patients [Feuerstein et al., 2021; Lange et al., 2021]. The responsible gene, KMT2B, maps to the chromosomal region 19q13.12 and comprises 37 exons that expand 8,148 base pairs without known hotspots of pathological variants [Zech et al., 2017]. Since most patients with DYT-KMT2B have arisen de novo, the variant pathogenicity is determined using a trio WES and a pathogenicity prediction algorithm. Therefore, if other methods could be employed in the screening test for DYT-KMT2B, the reliability of the WES results could be improved.

In this study, we investigated the usefulness of novel screening methods using swab samples of oral mucosa. The results obtained include several important points. First, a sufficient number of modified histones can be purified from the oral mucosa for analysis. Oral mucosa contains somatic DNA, and approximately 10 μg of nucleic acids can be obtained from a single swab [Hansen et al., 2007; Rogers et al., 2007]. Histones and DNA exist in a 1:1 mass ratio and are organized into the nucleosomes, by which a fundamental repeating unit forms chromatin [Cutter and Hayes, 2015]. When nucleosomes are treated under acidic conditions, nucleic acids are stabilized and histone proteins are eluted into the aqueous phase. Theoretically, 10 μg of histone proteins can be obtained from a single swab. Immunoblot analysis of extracted histones revealed both histone H3 and modified histone as visible bands (shown in Fig. 4). This technique could be used to detect other modified histones associated with diseases caused by genetic abnormalities of histone-modifying enzymes. Second, we confirmed less H3K4me3 modification in the patient compared to controls. KMT2B is an enzyme that catalyzes 1–3 methyl to the fourth lysine of histone H3. Previous studies using fibroblasts from patients with DYT-KMT2B showed a significantly reduced transcript level of KMT2B, indicating that the underlying mechanism of DYT-KMT2B is the result of loss of function [Zech et al., 2016; Kawarai et al., 2018]. However, direct evidence of reduced KMT2B enzymatic activity or H3K4me3 histone marks has not been demonstrated. Thus, direct evidence of KMT2B malfunction should be emphasized to be confirmed in patient-derived biological tissues.

Bioinformatic approaches have also corroborated the haploinsufficiency mechanism of DYT-KMT2B: two independent studies analyzed WBGS data obtained from healthy controls, and patients with DYT-KMT2B have led to a similar conclusion that methylated DNA profiles different from those of healthy individuals or other patients with hereditary dystonia [Mirza-Schreiber et al., 2021; Lee et al., 2022]. After reanalyzing these data, methylated DNA promoters were found to be associated with H3K4me3 histone marks. DNA and H3K4 methylation are known to be markedly anticorrelated [Ooi et al., 2007; Meissner et al., 2008]. The results obtained from reanalysis indicated that DNA methylation changes are largely influenced by the H3K4me3 status, and these H3K4me3 changes may be regulated by writer enzyme KMT2B. The biological process of GO using genes with methylated DNA failed to list neuron-related events, but the “flagellated sperm motility” ranked first. Although this GO term is far from the neuronal function, monovalent and bivalent H3K4me3 catalyzed by KMT2B is known to be required for the spermatogenesis [Tomizawa et al., 2018]. The pathogenesis of DYT-KMT2B is assumed to be a decrease in H3K4me3 due to KMT2B dysfunction [Zech et al., 2016; Faundes et al., 2018]. However, it is unclear why loss of function of the ubiquitously expressed KMT2B causes motor symptoms due to neuronal dysfunction [Klonou et al., 2021]. Although speculative, genetic abnormalities of the other five H3K4 methyltransferase members are also associated with neuropsychiatric disorders, suggesting that the central nervous system is more vulnerable to the altered H3K4me3 expression [Zech et al., 2019].

In summary, the expression level of H3K4me3 may be decreased in the oral mucosa of patients with DYT-KMT2B, and this result supports the haploinsufficiency mechanism of DYT-KMT2B. Evaluation of modified histone by immunoblotting using a biological sample is a useful diagnostic adjunct in predicting DYT-KMT2B. This study was conducted to better understand the DYT-KMT2B pathophysiology and other inherited dystonia to provide clues for the development of future therapies, including enzyme replacement for H3K4 methyltransferase-related disorders.

We thank all the neurology medical wards and department staff of Tohoku University Hospital.

All procedures performed in studies involving human participants were in accordance with the ethical standards of the Institutional and/or National Research Committee and with the 1975 Helsinki Declaration and its later amendments or comparable ethical standards. Written informed consent was obtained from the patient for publication of the details of the medical case and any accompanying images. This study was approved by the local institutional review board (Tohoku University Ethics Committees, certification number: 2021-1-748). Written informed consent was obtained from the patient for the publication of this article. We confirm that we have read the journal’s position on issues involved in ethical publication and affirm that this work is consistent with those guidelines.

The authors have no conflicts of interest to declare.

This work was supported in part by a Grant-in-Aid for Scientific Research (C) (Grant No. 20K07862 [to N.S.]; Grant No. 20K07896 [to T.H.]; Grant No. 21K07869 [to E.K.]; Grant No. 20K07907 [to S.M.]), a Grant-in-Aid for Young Scientists [Grant No. 19K16998] (to S.Y.), Japan Agency for Medical Research and Development (AMED) (Grant Nos. 22ek0109486, 22ek0109549, and 22ek0109493 [to N. Matsumoto]), and the Takeda Science Foundation (N. M.). The authors declare that there are no conflicts of interest relevant to this work.

Naoto Sugeno: conceptualization, modified histone analyses, bioinformatics analyses, and writing – original draft. Takafumi Hasegawa, Kazuhiro Haginoya, Takafumi Kubota, and Kensuke Ikeda: patient’s care; Eriko Koshimizu, Mitsugu Uematsu, and Miyatake Satoko: gene analyses; Takafumi Hasegawa and Takaaki Nakamura: data curation and investigation; Takafumi Hasegawa and Masashi Aoki: supervision; and all authors: writing – review and editing.

All data and materials supporting our findings are included in this article and its online supplementary material. The video is also openly available in “figshare” (DOI: 10.6084/m9.figshare.22256767). The code used for this study is available via GitHub (https://github.com/naosuge/DYT-KMT2B). Further inquiries can be directed to the corresponding authors.