Introduction

Pigmentary mosaicism, or hypomelanosis of Ito (HI, MIM# 300337), is a rare neurocutaneous syndrome characterized by hypopigmented skin whorls along Blaschko’s lines. It is associated with nervous system abnormality and other ectodermal defects involving fingers, toes, and teeth (Kouzak et al. 2013; Sybert 1994). Hypopigmented lesions have a reduced number of melanocytes, which originate from neural crest cells (Carmignac et al. 2021; Kuster and Konig 1999; Lee et al. 1999). In some cases, it is difficult to determine whether the affected skin is hypopigmented or hyperpigmented (Nehal et al. 1996), and the term “pigmentary mosaicism” refers to patterned hypopigmentation or hyperpigmentation resulting from a clone of skin cells with altered melanin-producing ability (Kromann et al. 2018; Shaffer and Chernoff 2020). The prevalence of HI is unknown, but it is estimated to be between 1/10,000 and 1/8,500, making it one of the most predominant rare neurocutaneous syndromes (Barbel et al. 2015). It affects both sexes, with a female-to-male prevalence ratio of 2:1. It has long been considered that somatic mosaicism is involved in HI, because the differently pigmented skin areas correspond to the varying distribution of two mosaic cell lines in each individual. Up to 60% of individuals with HI present with multiple types of mosaic cytogenetic abnormalities, including numerical (e.g., trisomy 18 or 20, triploidy) and structural chromosomal aberrations (e.g., deletions, duplications, or translocations) (Kromann et al. 2018; Salas-Labadia et al. 2019; Shaffer and Chernoff 2020; Thomas et al. 1989). However, no medical treatments are currently available for such chromosomal abnormalities, even if they were detected. Thus, skin biopsies, only to confirm chromosomal abnormalities, have been less motivated.

Recently, a few monogenic causes of HI have been reported. Germline TFE3 (MIM* 314310) and somatic RHOA (MIM* 165390) variants were found in pigmentary mosaicism (Lehalle et al. 2020; Vabres et al. 2019; Yigit et al. 2020). Somatic or germline MTOR (MIM* 601231) variants have also been occasionally associated with pigmentary mosaicism (Gordo et al. 2018; Hadouiri et al. 2020; Handoko et al. 2019; Mirzaa et al. 2016) and a large cohort of MTOR-related HI has recently been reported (Carmignac et al. 2021). This study aims to further explore the monogenic causes of pigmentary mosaicism.

Materials and methods

Patients

Individuals who were clinically diagnosed with pigmentary mosaicism were recruited through the collaborative activity of the Japanese Society of Child Neurology between April 2018 and March 2021. A total of 11 patients with pigmentary mosaicism (clinically diagnosed as HI and predominant skin hypopigmentation) were enrolled (Fig. 1). Skin samples were obtained from the patients, whereas blood samples were collected from the patients and their parents. Fresh skin tissues were obtained by punch biopsies of hypopigmented skin. Genomic DNA was extracted from the skin and peripheral blood leukocytes. Informed consent was obtained from the patients’ parents. This study protocol was approved by the Institutional Review Board of the Yokohama City University Faculty of Medicine.

Fig. 1
figure 1

Flowchart of the study. DNA was extracted from blood leukocytes and skin lesion biopsies of patients with pigmentary mosaicism and analyzed by next-generation sequencing. Rare variants only in the skin (somatic variants) or in both the skin and blood (germline or somatic variants) were investigated. Candidate germline variants were confirmed by Sanger sequencing of trio-samples (from patients and their parents). Exome sequencing of 11 individuals with pigmentary mosaicism revealed four somatic MTOR variants, a somatic RHOA variant; de novo variants in USP9X (n = 2), TFE3 (n = 1), and KCNQ5 (n = 1); biallelic germline variants in GTF3C5; and one inherited germline variant in PHF6

Full size image

Exome sequencing (ES)

The flowchart of the study design is shown in Fig. 1. DNA was extracted from skin biopsies using proteinase K and sodium dodecyl sulfate with RNase and phenol/chloroform. DNA extraction from peripheral blood leukocytes was performed using the QuickGene DNA Extraction Whole Blood Kit L (Kurabo Industries Ltd., Osaka, Japan). ES was performed at a standard depth (approximately, 40–70×) as previously described (Sakamoto et al. 2021). In brief, genomic DNA was captured with a SureSelect Human All Exon V6 Kit (Agilent Technologies, Santa Clara, CA, USA) and sequenced on a Illumina Novaseq 6000 Sequencing System. Reads were aligned to the human reference genome (hg19) using Novoalign v3.02.13 (http://www.novocraft.com/).

Single nucleotide variant (SNV) and copy number variation (CNV) analysis

Germline variants in exons and at canonical splice sites (± 2 bp) were selected based on autosomal dominant and X-linked models considering de novo occurrence. Simultaneously, rare variants, assuming autosomal recessive inheritance were also considered. These variants were absent from our in-house exome database (n = 575) and the genome aggregation database (https://gnomad.broadinstitute.org/) (or at extremely low frequency if recessive). Somatic variants were extracted using MuTect2 (Cibulskis et al. 2013) and Varscan2 (Koboldt et al. 2012). We focused on variants detected by both programs. Candidate somatic variants were selected if they were covered by at least three or more reads. Known gene variants which have been previously reported in neurocutaneous syndrome-related phenotype on the Human Gene Mutation Database (HGMD) or the Catalogue of Somatic Mutations in Cancer (COSMIC) databases were prioritized among candidates with > 4% of variant allele frequencies (VAF). Candidate variants were validated by Sanger sequencing. Primers were designed using Primer3Plus, a web tool for designing primers (primer information is available on request) and PCR was performed using TaKaRa Ex Taq DNA Polymerase with 30 cycles of PCR. CNV was detected using eXome Hidden Markov Model (XHMM) and Nord’s program as described previously (Miyatake et al. 2015; Tsuchida et al. 2018; Uchiyama et al. 2021).

Results

ES analysis

ES identified plausible monogenic variants in all patients (11/11) (Fig. 1). The candidate variants information is summarized in Table 1 and Supplementary Table 1. The mean read depths of the protein coding sequences were covered with 40 reads or more. De novo somatic MTOR variants were identified in four patients [Patient 1, NM_004958.4: c.4448G>A, p.(Cys1483Tyr); Patient 2, c.5930C>T, p.(Thr1977Ile); Patient 3, c.6644C>T, p.(Ser2215Phe), Patient 4, c.7292T>C, p.(Leu2431Pro)]. Of these, three were exclusively detected in skin lesions, and one was identified in both skin and blood samples (Table 2). The somatic VAF of the MTOR mutant allele detected in skin in Patients 1, 2, 3, and 4 were 26.6% (20/70 reads), 21.3% (10/43 reads), 11.4% (6/62 reads), and 22.8% (8/35 reads), respectively. In Patient 4, the MTOR mutant read was also detected in the blood sample (5/28, 17.8%). All mosaic variants were confirmed to have signal peaks compared with wild-type variants in Sanger sequencing electropherograms (Supplementary Fig. 1). A somatic RHOA variant [NM_001664.4:c.139G>A, p.(Glu47Lys)] was found in the skin of Patient 5, with a VAF of 25.6%; this has been previously reported (Yigit et al. 2020). As for germline variants, de novo null variants in USP9X were found in two patients [Patient 7, NM_001039590.3: c.355dupT, p.(Arg121Profs*2); Patient 8, c.5834G > A, p.(Trp1945*)]. De novo TFE3 and KCNQ5 variants were identified in Patient 6 [NM_006521.6:c.560C>T, p.(Thr187Met)] and Patient 10 [NM_001160133.2:c.1477C>T, p.(Arg439Trp)], respectively. Patient 6 has been previously reported (Lehalle et al. 2020). A pathogenic PHF6 variant [NM_032458.3: c.821G>A, p.(Arg274Gln)] was maternally inherited in Patient 9, who presented hyperpigmented skin, rather than the usual hypopigmentation pattern. In Patient 11, we could neither identify de novo mosaic variants in the skin, nor any de novo germline variants. However, trio-based ES of blood DNA revealed biallelic GTF3C5 variants [NM_001286709.2: c.1160A>G; 1517G>T, p.(Lys387Arg); p.(Gly513Val)] as the candidates for this neurocutaneous phenotype.

Table 1 Genetic and clinical information of patients with hypomelanosis of Ito
Full size table
Table 2 Characterization of pathogenic variants
Full size table

Clinical data

Clinical photographs and images from brain imaging studies of patients are shown in Figs. 2 and 3, and clinical information is described in detail in Table 1 and in the Supplementary Information. Of the 11 patients, 9 (81.8%) exhibited intellectual disability. Patient 5 with a mosaic RHOA variant developed normally, whereas Patient 8 who had a germline USP9X variant had delayed developmental milestones (such as head control at 6 months, sitting at 11 months, walking at 17 months, and speaking a word at 2 years), but reached to the normal level of development by 6 years of age. Of the 11 patients, 7 (63.6%) had epilepsy, 6 (54.5%) exhibited brain abnormalities, 5 (45.5%) had dental anomalies, 3 (27.2%) experienced body/limb asymmetry, and another 3 (27.2%) demonstrated distinctive facial features. Macrocephaly was observed in only one case (Patient 4 with a somatic MTOR variant in blood and skin samples).

Fig. 2
figure 2

Skin lesion photographs and brain magnetic resonance imaging (MRI) studies of patients 1–4. a–f Patient 1 with a somatic MTOR somatic variant (c.4448G>A, p.Cys1483Tyr). a, c Hypopigmentation of the back and buttocks. b Axial view of T1-weighted MRI images of the brain. The brain is almost symmetrical, but the right hemisphere is slightly larger than the left. d Axial image of brain positron emission tomography (PET). The signals are significantly decreased in the right hemisphere (arrowhead). e, f Fontana–Masson stained skin biopsy of a depigmented lesion. Arrow indicates clusters of melanin-laden keratinocytes (brown areas), which are sparse in the hypopigmented skin compared to a normally pigmented skin lesion, although the number of melanocytes is not reducted (f). g Skin lesion of Patient 2. Hypopigmentation of the skin is seen only in the right side of the body. h, i T2-weighted coronal and axial images of brain MRI of Patient 2 with a somatic MTOR variant (c.5930C>T, p.Thr1977Ile). The right hemisphere is slightly larger than the left. The arrowheads show focal cortical dysplasia in bilateral occipital lobes, but predominantly in the right hemisphere. j Skin lesion of Patient 3 with a somatic MTOR variant (c.6644C>T, p.Ser2215Phe). k T2-weighted MRI brain finding of Patient 3 at day 21 after birth prior to surgery. Right hemimegalencephaly is observed in many slices. l, m Skin lesions and T2-weighted brain MRI of Patient 4 with a somatic MTOR variant (c.7292 T>C, p.Leu2431Pro) found in both skin lesion and blood leukocytes, respectively. Brain MRI of Patient 4 is normal and almost symmetrical

Full size image
Fig. 3
figure 3

Skin lesion photographs and brain magnetic resonance imaging (MRI) studies of patients 5–11. a Hypopigmentation in the left leg of Patient 5 with a somatic RHOA variant. b, c Saggital and axial views of T2-weighted brain MRI, respectively. No laterality is observed. Hyperintensity with mild dilatation of ventricles and small cysts are observed. d Skin lesion of Patient 6 with a germline TFE3 variant. e, f Skin lesions of Patients 7 and 8, respectively, both with a germline USP9X variant. g, h Skin lesion and sagittal view of T2-weighted brain MRI of Patient 9 with a germline PHF6 variant. The patient’s skin showed hyperpigmentation rather than hypopigmentation. Atrophy of the bridge capsules and superior cerebellar peduncles is observed. i Skin lesion of Patient 10 with a germline KCNQ5 variant. j, k Skin lesions and sagittal view of T2-weighted brain MRI of Patient 11. Brain MRI showed polymicrogyria, loss of white matter volume, and mildly enlarged lateral ventricles

Full size image

All of the four patients with somatic MTOR variants were diagnosed with epilepsy and developmental delay of varied severity after birth in addtion to skin hypopigmentation (Fig. 2 a, c, g, j, l). Patient 1 had a slightly enlarged right hemisphere on brain magnetic resonance imaging (MRI) (Fig. 2b). In a brain positron emission tomography scan, signals were significantly decreased in the right hemisphere (Fig. 2d). Patient 2 presented with unclear bilateral occipital cortexes (especially in the right) (Fig. 2i), but was able to walk and speak; therefore, his phenotype was considered relatively mild. Afebrile seizures occurred at 4 years, which were not improved with levetiracetam and lacosamide but were eventually well controlled with carbamazepine. Patient 3 had seizures 10 days after birth and exhibited severe developmental delay; thus, brain surgery was required to control refractory seizures. Brain MRI indicated that Patient 3 had apparent right hemimegalencephaly and focal cortical dysplasia (FCD) in the right occipital lobe, which was surgically corrected (Fig. 2k). Patient 4 showed macrocephaly, severe developmental delay, and epilepsy but without FCD or hemimegalencephaly on brain MRI (Fig. 2m), which is indicative of Smith–Kingsmore syndrome.

Patient 5 with a mosaic RHOA variant showed skin hypopigmentation (Fig. 3a) along with abnormally high MRI signals and cysts near the right ventricle (Fig. 3b and c). Almost all patients demonstrated skin hypopigmentation (Fig. 3a, d–f, i, j) except Patient 9, who had a PHF6 variant and demonstrated skin hyperpigmentation (Fig. 3g). The PHF6 variant was inherited from the patient’s mother who also had hyperpigmented skin. The familial pedigree of Patient 9 with genetic data is shown in Supplementary Fig. 2. Regarding a PHF6 variant, there was a report demonstrating grandparental genotyping was useful for interpreting ES results (Daum et al. 2020). We confirmed that the maternal grandmother did not possess a PHF6 variant. Unfortunately, a sample derived from the maternal grandfather was unavailable for confirmation. Neither grandparent exhibited hyperpigmentation or any other symptoms; therefore, the mother’s PHF6 variant may have occurred de novo. Patient 11 was born at 35 weeks of gestation by cesarean section because of oligohydramnios. He was noted to have HI after birth (Fig. 3j), refractory epilepsy at 1 year, and severe growth and developmental delay. The family pedigree of Patient 11 with genetic data is shown in Supplementary Fig. 2.

Discussion

We identified plausible monogenic causes of pigmentary mosaicism (mostly HI) in all 11 patients examined (Fig. 1). All causal genes identified in this study are highly intolerant to variation (high pLI, LOEUF, and z scores) except for the autosomal recessive GTF3C5 variants (Table 3). The high rate of monogenic cause of pigmentary mosaicism in this study was unexpected, but may be reasonable, considering that previous genetic investigations of pigmentary mosaicism were performed using only chromosomal analysis and/or array-based comparative genomic hybridization (without ES) (Kromann et al. 2018; Salas-Labadia et al. 2019). We propose that the exome-first strategy, preferably using blood- and skin-derived DNA, should be adopted for HI because it is a neuroectodermal disorder.

Table 3 Genes which can cause pigmentary mosaicism
Full size table

De novo somatic variants were identified in almost half cases (45%, five of 11 patients). Epidermal tissue contains both keratinocytes and melanocytes. It is well known that melanocytes are neuroectoderm derived. The neural tube (differentiating to neuronal cells in the central nervous system) and neuronal crest cells (to melanocytes) are derived from the neuroectoderm (Li et al. 2020). Thus, skin hypopigmentation could accompany the brain phenotype, depending on when somatic variants arise.

Postzygotic pathogenic variants in MTOR cause FCD (MIM# 607341), hemimegalencephaly, and Smith–Kingsmore syndrome (MIM# 616638). Among the 11 patients in this study, somatic MTOR variants were the most prevalent (n = 4, 36.3%), and among the four variants, three (Patients 1–3) were exclusively observed only in skin DNA, indicating that the variants may have occurred later in the postzygotic period (after epidermal differentiation), thus, only affecting the skin cells, while Patient 4 with Smith–Kingsmore syndrome had a variant in both skin and blood, indicating earlier occurrence in the postzygotic period (before epidermal differentiation). According to a review of Gordo et al. (2018), HI was observed in 14.8% (4/27) of patients with Smith–Kingsmore syndrome. Recently, MTOR-related HI in 14 (19.7%) of 71 patients with pigmentary mosaicism was reported (Carmignac et al. 2021), indicating that MTOR is the most common cause of pigmentary mosaicism. MTOR variants are often mosaic; therefore, skin biopsy is highly recommended for a conclusive diagnosis. All MTOR variants detected in this study have been previously known, and are all registered in COSMIC database as pathogenic somatic variants in cancers. The c.4448G>A, p.(Cys1483Tyr) variant in Patient 1 has been recurrently reported and is implicated as a cause of hemimegalencephaly and FCD or Smith–Kingsmore syndrome (D’Gama et al. 2015; Gordo et al. 2018). The c.5930C>T, p. (Thr1977Ile) variant found in Patient 2 has also been frequently reported as a cause of FCD and megalencephaly with pigmentary mosaicism (Carmignac et al. 2021; Handoko et al. 2019; Mirzaa et al. 2016). Interestingly, Handoko et al. (2019) confirmed that all four patients with c.5930C>T, p.(Thr1977Ile) showed cutaneous hypopigmentation along Blaschko’s lines. These two variants [c.4448G>A, p.(Cys1483Tyr) and c.5930C>T, p.(Thr1977Ile)] have been reported as either somatic or germline variant. The c.6644C>T, p.(Ser2215Phe) variant in Patient 3 has also been indicated as a cause of FCD or hemimegalencephaly (D’Gama et al. 2017; Moller et al. 2016), while the c.7292 T>C, p.(Leu2431Pro) variant in Patient 4 has never been reported as the cause of such condition (previously reported only in cancer). Somatic MTOR variants were first described as the cause of FCD requiring brain surgery to control epilepsy (Lim et al. 2015; Nakashima et al. 2015). Three patients in our cohort with MTOR somatic variants (except patient 3) did not require brain surgery, indicating that such variants can cause a relatively mild epilepsy phenotype. The VAF of MTOR variants in skin lesions was approximately 10–30%, which is consistent with that of a previous report regardless of the disease severity (Carmignac et al. 2021). The VAF of MTOR variants in skin lesions was considerably higher than that in FCD and hemimegalencephaly lesions, and it is usually less than 10% (sometimes 1–3% or less). This is advantageous for performing ES analysis of skin-derived DNA because deep sequencing may not be essential. As for patients with MTOR variants, mTOR inhibitors such as sirolimus and everolimus are potential treatment options.

Pigmentary mosaicism caused by somatic RHOA variants is usually associated with normal neurodevelopment, although apparent brain lesions can be detected on MRI, as in Patient 5 in this cohort (Vabres et al. 2019; Yigit et al. 2020). Previous reports have included patients with clinical manifestations compatible with this unique HI, but their genetic status was not confirmed (Inoue et al. 2015; Mateus et al. 2014). In two previous studies, the genetic cause of this type of HI in 9 of 11 patients (9/11) was a recurrent somatic RHOA variant, c.139G>A, p.(Glu47Lys), and is therefore called “RHOA-related mosaic ectodermal dysplasia” (Vabres et al. 2019; Yigit et al. 2020). RHOA variants are also frequently registered in the COSMIC database.

Pathogenic germline variants were found in three X-linked genes, TFE3 (transcription factor E3), USP9X (ubiquitin-specific protease 9, MIM*300072), and PHF6 (PHD finger protein 6, MIM*300414), which are consistent with the association of HI with structural abnormalities of the X chromosome. Germline or somatic TFE3 variants cause intellectual disability with pigmentary mosaicism and storage disorder-like features in both males and females (Lehalle et al. 2020). TFE3 functions in the mechanistic target of rapamycin (mTOR) complex 1 signaling in the PIK3–AKT–mTOR pathway, which can explain the association of these variants with pigmentary mosaicism similar to MTOR variants. Lehalle et al. (2020) found pigmentary mosaicism to be present in the majority of patients with TFE3 mutations (71%,  12/17), including 40% of the males and 83% of the females. USP9X and PHF6 variants cause distinct neurodevelopmental disorders with congenital anomalies in both sexes. X-linked recessive (MIM# 300919) and dominant (MIM# 300968) intellectual developmental disorders are caused by USP9X variants, whereas Borjeson–Forssman–Lehmann syndrome (BFLS, MIM# 301900) is caused by PHF6 variants. However, pigmentary mosaicism was only seen in female patients with abnormalities in these two genes (Jolly et al. 2020; Reijnders et al. 2016; Zhang et al. 2019). The rates of skin pigment complication in females related to USP9X and PHF6 variants were 65% (11/17) (Reijnders et al. 2016) and 77% (10/13) (Zhang et al. 2019), respectively. Skin lesions in females with aberrant USP9X and PHF6 manifest as hypopigmentation and hyperpigmentation along Blaschko’s lines, respectively. USP9X plays an important role associated with mTOR in neural development both in humans and mice (Agrawal et al. 2012; Murtaza et al. 2015; Reijnders et al. 2016), which may explain how this variant causes HI. PHF6 is implicated in chromatin remodeling by interacting with the SWI/SNF complex (Zhang et al. 2019; Zweier et al. 2013), the variants of which are occasionally found in patients with Coffin–Siris syndrome (MIM# 135900) (Zweier et al. 2013). In addition, pathogenic variants in USP9X and PHF6 are also found in cancers such as pediatric leukemia (Ma et al. 2018), which is similar to the reports of MTOR and RHOA variants being frequently detected in various cancers (Kilian et al. 2021; Mirzaa and Poduri 2014; Zou et al. 2020).

Germline KCNQ5 variants cause intellectual disability or epilepsy (Lehman et al. 2017), and only a few individuals with de novo KCNQ5 variants have been reported. Skin hypopigmentation in Patient 10 with a de novo KCNQ5 variant is indeed a novel phenotypic feature, which has not previously been reported in association with KCNQ5 abnormality. Therefore, it remains uncertain whether the KCNQ5 variant is related to the skin phenotype. GTF3C5 (MIM*604890) which encodes general transcription factor IIIC subunit 5 (GTF3C5), plays a crucial role in general transcription (Crepaldi et al. 2013); however, GTF3C5 abnormalities have never been reported in human diseases. Homozygous mutant Gtf3c5tm2a(KOMP)Wtsi mice exhibited complete preweaning lethality, while heterozygous mutant mice survived (Ayadi et al. 2012). The biallelic GTF3C5 variants may potentially cause skin hypopigmentation based on the high probability score (0.99) of gene intolerance to the biallelic loss of function (Lek et al. 2016). Further studies involving patients with GTF3C5 variants are required to confirm the association of GTF3C5 with this human ectodermal disorder. It is important to study and analyze more patients harboring KCNQ5 and GTF3C5 variants and collect further evidence before establishing any conclusion.

HI is thought to be caused by somatic chromosomal structural abnormalities, but these structural abnormalities have been inconsistent, and genetic testing for diagnosis has not been actively performed until recently. Moreover, HI has sometimes been considered as a benign phenotype rather than a disease (Ruggieri and Pavone 2000; Sybert 1994). However, the high discovery rate of monogenic causes of HI prompts us to perform ES for patients with pigmentary mosaicism, preferably by collecting skin samples, as this may lead to the optimal therapeutic approaches in the future. Although using skin tissue instead of brain tissue may allow regular ES with 40–70× read coverage for reliably detecting somatic mosaic variants with > 4% of VAF in pigmentary mosaicism, higher read coverage in the NGS analysis is better to avoid missing low-prevalent (≦ 4%) variants.