Introduction

WD repeat domain 45 (WDR45) is located on Xp11.23 and encodes a protein that functions in the autophagy pathway, which is the major intracellular degradation system. Pathogenic variants of WDR45 cause neurodegeneration with brain iron accumulation 5 (NBIA5; OMIM:300894), mostly in female patients. In adulthood, patients often show progressive movement disorders, cognitive decline and iron accumulation in the globus pallidus and substantia nigra on brain MRI scans. These characteristic features can prompt a clinical diagnosis and targeted molecular testing. In childhood, patients with NBIA5 show diverse phenotypes, including developmental delay, intellectual disability (ID), Rett-like stereotyped movements and epileptic encephalopathies.1 These phenotypes are non-specific, making the diagnosis of NBIA5 based only on clinical symptoms and examinations difficult. However, an early diagnosis is important for the medical management of NBIA5 patients. Among undiagnosed paediatric patients with a wide range of neurological symptoms, WDR45 pathogenic variants may be responsible.2 3

Previously, we conducted whole-genome sequencing (WGS) of 45 patients with ID, the results of which suggested high genetic heterogeneity in ID.4 Patients with no apparent causal mutations in the initial analysis were subjected to reanalysis using an updated pipeline, and we found four patients with pathogenic WDR45 variants (4/45, 8.9%). These cases were reported in a parallel study of patients with Rett syndrome or Rett-like phenotypes,5 6 prompting us to hypothesise that WDR45 can be a major gene for female ID with a particular phenotype. Here, we set up a cohort of 32 female patients with ID from over 600 families from the Japanese ID collection and performed targeted WDR45 analysis. Furthermore, to delineate the X chromosome inactivation (XCI) status in a family with uninformative homozygous status at both the HUMARA and FRAXA loci, we developed an XCI analysis method based on methylation-specific PCR.

Materials and methods

Written informed consent was obtained from the guardians of all the patients.

Patient cohort

Japanese patients with ID and/or developmental delay referred to the Department of Child Neurology of NCNP hospital were enrolled to NCNP Biobank (Tokyo, Japan) with written informed consent for genetic studies. For this study, 32 female patients with ID were selected from the NCNP Biobank according to the following criteria: (1) followed up from childhood, (2) Sanger sequencing and microarray-based comparative genomic hybridisation excluded pathogenic MECP2 variants and copy number variations, respectively, and (3) not participated in previous studies. No clinical criteria other than ID was applied for selection. We obtained clinical information and extracted genomic DNA from the peripheral blood cells.

Sanger sequencing

We sequenced exons 3–12, which encompasses the entire coding region of WDR45 (NM_007075.3), using Sanger sequencing. The primers are listed in online supplemental table 1.

Whole-genome sequencing (WGS)

DNA extracted from the patient and her mother was subjected to WGS (Takara Bio, Shiga, Japan). DNA libraries were prepared using the TruSeq DNA PCR-free HT library prep kit, and samples were prepared according to the manufacturer’s instructions. WGS was performed using a NovaSeq6000 (Illumina, San Diego, California, USA) with 150 bp paired-end reads. The obtained FASTQ data were mapped to GRCh38 and variant calling was performed using Parabricks V.3.1.0 (Nvidia).

The filtering method for the obtained variants has been described previously, with some modifications.4 Briefly, exonic or canonical splice variants were filtered using an allele frequency of 0.5% as a cut-off in the gnomAD and Japanese National Research Hospital biobanks.7 Variants of genes associated with ID in OMIM (https://www.omim.org/) or HGMD Professional 2023 were evaluated using guidelines proposed by the American College of Medical Genetics and Genomics and the Association for Molecular Pathology (ACMG/AMP).8

XCI analysis

We sought to delineate the XCI pattern using a methylation PCR-based assay, with some modifications9 (online supplemental fig 1) because conventional PCR assays with methylation-sensitive restriction digestion at the HUMARA 10 and FRAXA 11 loci were uninformative due to allelic homozygosity at both loci. Instead of using short-fragment tandem repeats (STRs) at the HUMARA locus, we selected heterozygous variants on the X chromosome detected using WGS for allelic separation. We then searched for the methylation status in white blood cells using the UCSC genome browser to select those that were presumably active (ie, unmethylated). Next, we evaluated the actual methylation status of a selected locus by PCR in male and female DNA treated with methylation-sensitive restriction enzymes (online supplemental fig 2A). The variant NM_001256789.2:c.319G>A on CACNA1F gene was found to fulfil all conditions. Next, we designed primer pairs specific to either methylated or unmethylated DNA sequences after bisulfite modification (online supplemental figure 2B), which converts the unmethylated cytosine of CpG to uracil. Sanger sequencing of the PCR products from both the patient and control females was conducted (online supplemental figure 1). The peak ratios of the reference and alternative variants indicated the XCI pattern. We noted that the codon next to the 5′ side of the variant contains cytosine and converted thymine in methylated-specific sequence, whereas only cytosine exists in unmethylated-specific sequence.

ResultsClinical overview of the cohort

Thirty-two female patients were included in this study. A summary of the patients’ clinical information is presented in table 1. Epilepsy was observed in more than half of patients (19/32, 59%). Stereotypic movements and autism spectrum disorders were observed in 10 and 8 patients (10/32, 31%; 8/32, 25%), respectively, with three showing both phenotypes. Abnormal MRI findings include cerebral atrophy, band heterotopia, corpus callosum hypoplasia and delayed myelination.

Table 1

Clinical characteristics of the 32 patients

Overview of the identified WDR45 variants

We identified two pathogenic WDR45 variants, c.1051delG p.(Val351CysfsTer60) and c.868C>T p.(Gln290Ter), in two patients aged 2–4 years at the time of registration (2/32, 6.3%). The latter showed stereotypic movements and epilepsy and was previously reported.12

Case presentation with a novel c.1051delG variant

The patient was a 2-year-old Japanese female born to non-consanguineous parents at term as their first child after microfertilisation at registration. Her birth weight, height and head circumference were 3078 g (+0.69 SD), 50 cm (+0.77 SD) and 33.6 cm (+0.31 SD), respectively. The patient exhibited developmental delays from early infancy. She achieved rolling at the age of 4 months and head control at the age of 5 months; thereafter, her motor and psychomotor development remained unchanged. At 1 year and 6 months of age, she experienced a spasm and was diagnosed with epilepsy. Her developmental quotient scores on the Enjoji Infantile Developmental Scale and Kinder Infant Development Scale were 22 and 14, respectively. She began tube feeding at the age of 8 years after suffering from ketogenic hypoglycaemia and underwent scoliosis surgery at the age of 11 years. At the age of 12 years, she was bedridden and could not speak any meaningful words. Her height was 148.5 cm (−1.4 SD) and her weight was 34.8 kg (−0.9 SD), respectively. On laboratory findings, her serum iron at 2 years was 79 μg/dL, which was within optimal range. Other items regarding iron metabolism were not examined, but her haemoglobin was from 123 to 144 g/L, which did not suggest abnormalities in iron metabolism.

Brain MRI at the age of 2 years revealed brain atrophy and severely delayed myelination (online supplemental fig 3). A follow-up MRI at the age of 12 years revealed low signal intensity in the globus pallidus and substantia nigra on T2WI, which are typical features observed in NBIA5 patients (online supplemental fig 3).

Genetic testing

Trio Sanger sequencing confirmed the de novo nature of the heterozygous nucleotide deletion NM_007075.3(WDR45):c.1051delG, p.(Val351CysfsTer60), which presumably resulted in a frameshift and stop-loss variant (figure 1A). This variant was classified as ‘likely pathogenic’ according to the American College of Medical Genetics (ACMG) guidelines (PS2, PM2, PM4).8 WGS did not identify any other pathogenic variants, including those in two iron transporter genes, divalent metal transporter (DMT1) and ferroportin (FPN). The variant was submitted to ClinVar (accession number: SCV004697536).

Figure 1

WDR45 variant and X chromosome inactivation assay. (A) Sanger sequencing electropherograms of the de novo WDR45 variant in the patient. The arrow above the electropherogram shows the exact position of the frameshift variant. (B) The estimation of X inactivation pattern in our patient using her heterozygous variant in CACNA1F (NM_001256789.2:c.319G>A) on the X chromosome for allele separation. Methylated (upper) and unmethylated (bottom) sequences with bisulfite treatment are shown. The heterozygous variant is shown by a black arrow. In our patient, the peak heights of guanine and adenine are almost the same in both methylated and unmethylated sequence, suggesting a random X inactivation pattern. In the methylated-specific sequence, the codon next to the 5′ side of the variant contains cytosine and converted thymine. Contrariwise, in the unmethylated-specific sequence, only thymine exists (white arrows).

XCI analysis using methylation-specific PCR at the CACNA1F gene locus with genomic DNA from the peripheral blood estimated a random pattern in this patient (figure 1B).

Discussion

We identified two WDR45 variants in 2 of the 32 (6.3%) female patients with ID. Together with our previous findings of 4 WDR45 variants in 19 female patients (out of 45 total patients) in the re-analysed WGS study from the same ID cohort, cumulative WDR45 pathogenic alterations in female patients with ID occurred in 6 out of 51 (12%) individuals, apparently suggestive of WDR45 as the major causative gene for female ID. This finding is also in line with previous reports that in patients with Rett syndrome or Rett-like phenotypes who were negative for MECP2 variants, pathogenic WDR45 variants were identified in 2.5–9.5%.2 3 5 In the present study, one patient with the WDR45 variant showed stereotypic movements, whereas the other patient with early onset epilepsy did not. Therefore, WDR45 variants can cause ID in females with a wide range of clinical spectra.

Notably, a female patient with the novel frameshift WDR45 variant showed severe developmental delay from early infancy. Although patients with WDR45 pathogenic variants reveal various ranges of developmental delay, they mostly achieve independent walking before regression. An analysis of 123 patients with WDR45 pathogenic variants showed that approximately 90% of patients achieved rolling by the age of 1 year, 80% achieved independent walking by the age of 3 years and approximately 60% acquired a meaningful word by the age of 3 years.13 Compared with this common trend in children with WDR45 variants, the present case appears to show the most severe form of WDR45-associated phenotypes. Meanwhile, rare patients with similarly severe phenotype have been reported.13 Therefore, we consider that her particularly severe phenotype is still within the current clinical spectrum of WDR45-related disorders. The present case had no perinatal event that could have modified her clinical manifestation. In addition, no other pathogenic variants, including both SNV and CNV, were detected by WGS and microarray-based comparative genomic hybridisation.

The identified variant, c.1051delG, p.(Val351CysfsTer60), presumably causes a substitution of codon 351, elimination of the terminal 11 amino acid residues and putative stop codon, and translational extension into the 3′UTR with a stretch of 60 residues. In silico prediction analyses using NCBI Conserved Domains Search (https://www.ncbi.nlm.nih.gov/Structure/cdd/wrpsb.cgi), InterPro (https://www.ebi.ac.uk/interpro/) and PROSITE (https://prosite.expasy.org/) identified no potential functional domains within these newly gained polypeptides. Similar stop-loss variants have been reported in three other cases carrying mostly the same extended amino acid residues; in two of these cases, the clinical details are described (online supplemental table 2). One patient could walk independently and speak meaningful words at the age of 1 year and 6 months. On the other hand, the other patient showed mild motor clumsiness, developmental delay and autism spectrum disorder with the ability to walk, ride a bike, skate and attend school with special education,14 15 clearly showing a milder phenotype than that in the present case. These findings suggest that the clinical phenotypes of stop-loss variants with extensions vary.

Given the variable severity in cases with similar stop-loss WDR45 variants, it is unlikely that extended polypeptides acquire unique toxic gain-of-function or dominant-negative effects; rather, they may simply mitigate the autophagic activities of WDR45, similar to other mutants.

Another factor that can affect clinical severity is the skewing of XCI status, wherein unfavourable skewing may result in a severe phenotype. However, in the present case, XCI of peripheral blood DNA suggested a random pattern. An unfavourable skewed XCI pattern shown by the monoallelic expression of the mutant allele in lymphoblastoid cell lines has been reported in three patients with the WDR45 variant.15 However, their clinical manifestations, including developmental delay, were milder than those in the present case. Another patient with a truncating WDR45 variant showing severe developmental delay had a random XCI pattern.16 Taken together, skewed XCI may not be a major determinant of WDR45-associated phenotype severity in children, at least in the peripheral blood.

Deficient WDR45 impairs ferritinophagy, a ferritin-specific autophagy that contributes to the control of intracellular iron metabolism.17 Under dysregulated ferritinophagy, the active ferrous iron becomes too short to be used in critical biochemical reactions, thereby causing neurological symptoms. To compensate for the disordered ferritinophagy, WDR45 deficient cells change the levels of transmembrane proteins related to iron inflow and outflow, such as DMT1 and FPN. Therefore, the variable clinical severity of NBIA5 may be attributed to the intracellular level of residual active ferrous iron and the efficacy of compensatory mechanisms. In our case, exonic or canonical splicing variants of DMT1 and FPN were not identified, and the association of compensation mechanism for disordered ferritinophagy to the degree of clinical phenotypes remains unclear. It is also possible that yet unknown factors that influence the phenotypic variation in WDR45-associated disorders may be present.

In conclusion, we found that WDR45 is a major causative gene of ID in female children. Their clinical phenotypes can be variable and non-specific; therefore, genetic testing should be considered as the primary diagnostic examination. We identified a novel stop-loss variant of WDR45 in a female patient, who showed the most severe end of the wide clinical spectrum of WDR45 variants. At present, the molecular mechanisms underlying this severe phenotype are unclear, as at least two other cases with similar stop-loss variants gaining extended polypeptides did not show such a severe phenotype, and neither the XCI status nor alterations in iron transporter genes correlate with clinical severity.