Expanding the Spectrum of KDM5C Neurodevelopmental Disorder: A Novel De Novo Stop Variant in a Young Woman and Emerging Genotype–Phenotype Correlations

As a consequence of the implementation of NGS technologies, the diagnostic yield of neurodevelopmental disorders has dramatically increased during the past two decades. Among neurodevelopmental genes, transcription-related genes and chromatin remodeling genes are the most represented category of disease-causing genes. Indeed, the term “chromatinopathies” is now widely used to describe epigenetic disorders caused by mutations in these genes. We hereby describe a twenty-seven-year-old female patient diagnosed with moderate intellectual disability comorbid with other neuropsychiatric and behavioral issues carrying a de novo heterozygous stop variant in the KDM5C gene (NM_004187.5: c. 3847G>T, p.Glu1283*), encoding a histone demethylase that specifically acts on the H3K4 lysines. The gene is located on the X chromosome and has been associated with Claes–Jensen-type intellectual disability, an X-linked syndromic disorder. We discuss our case in relation to previously reported affected females harboring pathogenic mutations in the KDM5C gene with the objective of delineating genotype–phenotype correlations and further defining a common recognizable phenotype. We also highlight the importance of reverse phenotyping in relation to whole-exome sequencing results.


Introduction
Claes-Jensen type (OMIM#300534), an X-linked syndromic neurodevelopmental disorder, is caused by mutations in the KDM5C gene (OMIM*314690). Maternally transmitted or, more rarely, de novo pathogenic variants explain about 0.7-2.8% of X-linked intellectual disability [1]. KDM5C encodes a histone demethylase that specifically acts on lysine 4 of histone H3 (H3K4), regulating transcriptional repression and chromatin remodeling. The gene is intolerant both to loss of function and missense mutations. Indeed, pathogenic nonsense, missense, and frameshift variants have been identified in several patients [2]. To date, about 60 pathogenic variants and 40 likely pathogenic variants are included in the ClinVar database (https://www.ncbi.nlm.nih.gov/clinvar/ (accessed on 1 November 2022). Moreover, recent studies have shown that patients with pathogenic variants display a specific peripheral blood epi-signature [3] that could represent a marker of pathogenic KDM5C variants [4].
The KDM5C knockout male mouse displays cognitive impairment, dendritic spine morphological abnormalities, and transcription dysregulation of many neurodevelopmental genes. The female knockout mouse exhibits a milder cognitive phenotype characterized by memory deficits and learning disability [5,6]. Similarly, heterozygous carrier females are often asymptomatic or mildly affected compared to hemizygous carrier males who usually exhibit moderate to severe intellectual disability and other neurodevelopmental disabilities including language or behavioral impairments and epilepsy [2]. Physical anomalies reported in the OMIM database (https://omim.org/ (accessed on 1 November 2022) include short stature, small forehead, prognatism, micrognathia, maxillary hypoplasia, facial hypotonia, flat philtrum, thin upper lip, high narrow palate, small and deep-set eyes, and large ears. Visual problems such as strabismus, hypermetropia, myopia, and skeletal anomalies of the hands and feet have also been reported in some affected individuals.
The clinical variability among females, even within the same family [2], might be related to the X chromosome inactivation pattern, which can help to compensate for disruptive genetic variants. However, a recent study [7] highlights a new role for KDM5C demethylase in activating the expression of Xist lncRNA (which is normally required for stable X chromosome inactivation), suggesting an alternative molecular disease mechanism based on a cascade effect of altered XIST expression on other X-linked genes. Therefore, in the presence of a KDM5C heterozygous variant, the expression of Xist could decrease, and this in turn could impair appropriate X chromosome inactivation. In this scenario, in female carriers, nonsense variants leading to the loss of function of one KDM5C allele could have a stronger and more detrimental effect on the X chromosome inactivation than missense variants. This in turn could result in a more severe phenotype in females bearing variants that abolish the expression of one allele, compared to females bearing missense variants. In order to explore this hypothesis, we reviewed genotype-phenotype correlations in all the affected females reported in the literature to date with a thorough clinical description. Moreover, we included in this dataset a twenty-seven-year-old female evaluated in the Medical Genetics outpatient clinics of our hospital. She carries a new de novo nonsense variant (NM_004187.5: c.3847 G>T, p.Glu1283*) in the KDM5C gene. We also aim at delineating specific clinical features that can help the geneticist to suspect or even recognize the KDM5C-related syndrome. Recognition of these features could drive reverse phenotyping analysis and facilitate the interpretation of pangenomic results.

Analysis of the X Chromosome Methylation Status
Considering the severity of clinical manifestations, an assay for evaluating the methylation status (i.e., inactivation) of the X chromosomes (X chromosome inactivation, XCI) was performed on DNA from blood leukocytes of the patient. The XCI pattern was determined by evaluating the cytosine methylation of CpG dinucleotides within the polymorphic CAG repeat in the first exon of the androgen receptor (AR) gene, located on the X chromosome [9]. In brief, DNA was digested by methylation-sensitive restriction enzymes (i.e., HpaII). After digestion, DNA amplification can occur only in the presence of methylated restriction sites (inactive allele). Both a digested and an undigested DNA sample were then amplified by two different primers' pairs specific for the AR locus. PCR products were separated on an ABI3500 sequencer (Thermo Fisher Scientific, Waltham, MA, USA) and analyzed by GeneMapper 5 software (Thermo Fisher Scientific). All samples were analyzed in triplicate, and the average values were used in calculating the degree of X inactivation. In our heterozygous patient, the XCI skewing ratio was determined by comparing the ratio of allele peak heights in the digested sample (d1 and d2, for paternal and maternal alleles, respectively) with the ratio in the undigested sample (u1 and u2), according to the following proportion: XCI percentage of the paternal allele = [(d1/u1)/{(d1/u1)+(d2/u2)}] × 100, which ranges from 0% to 100% [10].

Retrospective Study
We introduced the following key words in PubMed for KDM5C or JARID1C genes: (a) name of the gene; (b) name of the gene AND language delay; (c) name of the gene AND neurodevelopment; (d) name of the gene AND intellectual disability. We selected 12 papers [4,[11][12][13][14][15][16][17][18][19][20][21] relative to patients carrying variants in KDM5C/JARID1C genes. Among them, we discarded papers without a deep clinical description and papers in which patients were only males. We ended up with four papers [4,17,19,20] concerning female patients carrying KDM5C mutations and with a thorough clinical description. We considered "thorough" to mean a clinical description where there was information concerning at least the intellectual status of the patient (developmental delay or ID), the physical description (peculiar physical anomalies), and information about language development and behavior. Most of the selected patients (Table 1) also had additional clinical information. We extrapolated genetic and clinical data in order to delineate possible genotype-phenotype correlations.

Clinical Presentation of the Patient
The proband came to our genetic counselling service for a genetic re-evaluation when she was twenty-seven years old (Figure 1). She had previously undergone conventional karyotype and array-CGH, which were both normal. The initial "gestaltic" diagnosis was that of Rubinstein-Taybi syndrome when she was in her childhood, but no test was available at that time. She was the second child of two apparently healthy unrelated parents and was born naturally at term (Figure 2a). Perinatal distress due to umbilical cord wrapping and an Apgar index of 8, at the first and fifth minutes, was reported by the mother. Due to the low birthweight (2400 Kg) and due to respiratory and cardiopulmonary distress, she was hospitalized in intensive care; length was normal (48 cm). Initial language and motor development were referred to as normal: at six months, she sat down without support. Regression started at eight months, immediately after an epileptic seizure. Both cognitive and motor development were delayed: some communication skills were acquired at 24 months, and walking alone occurred at 18 months. EEG examination showed slight but not significant anomalies, whereas brain MRI was normal. Menarche was premature and occurred when she was 9 years old. The proband came to our genetic counselling service for a genetic re-evaluation when she was twenty-seven years old (Figure 1). She had previously undergone conventional karyotype and array-CGH, which were both normal. The initial "gestaltic" diagnosis was that of Rubinstein-Taybi syndrome when she was in her childhood, but no test was available at that time. She was the second child of two apparently healthy unrelated parents and was born naturally at term (Figure 2a). Perinatal distress due to umbilical cord wrapping and an Apgar index of 8, at the first and fifth minutes, was reported by the mother. Due to the low birthweight (2400 Kg) and due to respiratory and cardiopulmonary distress, she was hospitalized in intensive care; length was normal (48 cm). Initial language and motor development were referred to as normal: at six months, she sat down without support. Regression started at eight months, immediately after an epileptic seizure. Both cognitive and motor development were delayed: some communication skills were acquired at 24 months, and walking alone occurred at 18 months. EEG examination showed slight but not significant anomalies, whereas brain MRI was normal. Menarche was premature and occurred when she was 9 years old. When she attended primary school, she received a diagnosis of mild intellectual disability and required special educational support for the entire school period. As a child, she was hyperactive with a short attention span, easy distractibility, and poor abilities in executive functions. Significant difficulties were also reported in expressive language, especially for lexicon and phonetics, as well as in calculation/arithmetic skills. When she attended primary school, she received a diagnosis of mild intellectual disability and required special educational support for the entire school period. As a child, she was hyperactive with a short attention span, easy distractibility, and poor abilities in executive functions. Significant difficulties were also reported in expressive language, especially for lexicon and phonetics, as well as in calculation/arithmetic skills.
Clumsiness and deficit in fine motor skills were also reported. All these symptoms persisted throughout high school. As an adult, she also displays issues in the self-regulation of emotional response (outbursts), with agitation when frustrated, deficits in social skills, occasionally aggressive behavior, and indolence. Very low self-esteem and full awareness of her condition were also reported by the mother. She is no longer hyperactive but sometimes shows oppositional behaviors. She lives with her parents and carries out voluntary work in the local church. The last psychodiagnostic evaluation, using WAIS-R battery, rated an IQ of 47 and an overall evaluation of the global functioning equal to 45 (VGF). She has been constipated since childhood; she is also tending towards overweight and shows a strong preference for sweets. Clumsiness and deficit in fine motor skills were also reported. All these symptoms persisted throughout high school. As an adult, she also displays issues in the selfregulation of emotional response (outbursts), with agitation when frustrated, deficits in social skills, occasionally aggressive behavior, and indolence. Very low self-esteem and full awareness of her condition were also reported by the mother. She is no longer hyperactive but sometimes shows oppositional behaviors. She lives with her parents and carries out voluntary work in the local church. The last psychodiagnostic evaluation, using WAIS-R battery, rated an IQ of 47 and an overall evaluation of the global functioning equal to 45 (VGF). She has been constipated since childhood; she is also tending towards overweight and shows a strong preference for sweets.
Morphological evaluation showed a short stature (145 cm), a V-shaped frontal hairline, prognathism, facial hypotonia, a flat philtrum, a prominent nose with a downturned and bulbous tip, thin and arched eyebrows, small eyelashes, small deep-set eyes, strabismus (alternate exoforia), a thin upper lip, a high narrow palate, small feet, and squared hands with short and thick distal phalanges and brachydactyly ( Figure 1). All these features have been described in patients with Claes-Jensen syndrome.

Genetic Findings
By WES, the proband was found to carry a nonsense variant in KDM5C (NM_004187.5:c.3847G>T; p.Glu1283*) in heterozygosity. The variant was classified as pathogenic since it is a null variant in a gene whose loss of function is a known mechanism of disease and is rare, being absent in from GnomAD 2.1 population database. By Sanger analysis, the presence of c.3847G>T was excluded in the DNA from peripheral blood of the parents, and thus the variant was assumed to be de novo. XCI analysis performed on DNA from proband's lymphocytes showed a skewed pattern of methylation with 87% methylation status of one X chromosome allele (Figure 2b). Morphological evaluation showed a short stature (145 cm), a V-shaped frontal hairline, prognathism, facial hypotonia, a flat philtrum, a prominent nose with a downturned and bulbous tip, thin and arched eyebrows, small eyelashes, small deep-set eyes, strabismus (alternate exoforia), a thin upper lip, a high narrow palate, small feet, and squared hands with short and thick distal phalanges and brachydactyly ( Figure 1). All these features have been described in patients with Claes-Jensen syndrome.

Genetic Findings
By WES, the proband was found to carry a nonsense variant in KDM5C (NM_004187.5:c. 3847G>T; p.Glu1283*) in heterozygosity. The variant was classified as pathogenic since it is a null variant in a gene whose loss of function is a known mechanism of disease and is rare, being absent in from GnomAD 2.1 population database. By Sanger analysis, the presence of c.3847G>T was excluded in the DNA from peripheral blood of the parents, and thus the variant was assumed to be de novo. XCI analysis performed on DNA from proband's lymphocytes showed a skewed pattern of methylation with 87% methylation status of one X chromosome allele (Figure 2b).

Genotype-Phenotype Correlations
The genetic and clinical characterization of twenty-three affected females carrying KDM5C variants is reported in Table 1. KDM5C pathogenic variants identified to date are summarized in Figure 3. There are thirteen females belonging to four families (family no. 4 with five females, family 10 with four females, and families 11 and 14 with two females each). Four females (17%) had no intellectual disability, whereas in the others (83%), mental impairment ranged from mild to moderate. No patients with severe intellectual disability were present in this cohort. These data confirm that females are less severely affected than males, as previously reported [2]. Language delay was reported in 18/23 females (78%), and in most cases, it was specific for expressive speech impairment (19). Overweight/obesity were common in older females (≥17 ys), affecting nine out of twelve patients (75%). In addition, specific behavioral problems, including aggressive behavior, low frustration tolerance, anxiety, and socialization impairment, seem to manifest mainly in adulthood (females ≥ 17 ys) and were reported in 7/12 patients (60%). Seven different types of missense variants (five in the JmjC domain, one in the JmjN domain, and one in the ARID domain) and seven types of frameshift (three), splicing (two), and nonsense (two) variants are reported in Table 1. Eleven females carry missense variants, and twelve females carry frameshift/truncating/splicing variants. In general, females carrying missense variants seem to be less severely affected than females carrying frameshift/truncating/splicing variants: normal development or mild intellectual disability was reported in 73% (8/11) females, whereas moderate intellectual disability was reported in 67% (8/12) females. In addition, if we consider only sporadic cases diagnosed by WES and not by segregation analysis as familial cases, the genotype-phenotype correlation becomes stronger: 83% (5/6, patients no. 1, 9, 10, 11, and 13) females diagnosed with moderate intellectual disability had stop/frameshift/splicing variants, and 100% (1/1, patient 2) of females diagnosed with mild intellectual disability or normal intelligence had missense variants (Tables 1 and 2). Variable clinical expressivity was observed in all the four families reported in Tables 1 and 2 and could be due to genetic background effects.
The genetic and clinical characterization of twenty-three affected females carrying KDM5C variants is reported in Table 1. KDM5C pathogenic variants identified to date are summarized in Figure 3. There are thirteen females belonging to four families (family no. 4 with five females, family 10 with four females, and families 11 and 14 with two females each). Four females (17%) had no intellectual disability, whereas in the others (83%), mental impairment ranged from mild to moderate. No patients with severe intellectual disability were present in this cohort. These data confirm that females are less severely affected than males, as previously reported [2]. Language delay was reported in 18/23 females (78%), and in most cases, it was specific for expressive speech impairment (19). Overweight/obesity were common in older females (≥17 ys), affecting nine out of twelve patients (75%). In addition, specific behavioral problems, including aggressive behavior, low frustration tolerance, anxiety, and socialization impairment, seem to manifest mainly in adulthood (females ≥ 17 ys) and were reported in 7/12 patients (60%). Seven different types of missense variants (five in the JmjC domain, one in the JmjN domain, and one in the ARID domain) and seven types of frameshift (three), splicing (two), and nonsense (two) variants are reported in Table 1. Eleven females carry missense variants, and twelve females carry frameshift/truncating/splicing variants. In general, females carrying missense variants seem to be less severely affected than females carrying frameshift/truncating/splicing variants: normal development or mild intellectual disability was reported in 73% (8/11) females, whereas moderate intellectual disability was reported in 67% (8/12) females. In addition, if we consider only sporadic cases diagnosed by WES and not by segregation analysis as familial cases, the genotypephenotype correlation becomes stronger: 83% (5/6, patients no. 1, 9, 10, 11, and 13) females diagnosed with moderate intellectual disability had stop/frameshift/splicing variants, and 100% (1/1, patient 2) of females diagnosed with mild intellectual disability or normal intelligence had missense variants (Tables 1 and 2). Variable clinical expressivity was observed in all the four families reported in Tables 1 and 2 and could be due to genetic background effects.

Discussion
Our study further contributes to outlining the specific phenotype in females with X-linked intellectual disability, Claes-Jensen type: variable degrees of developmental delay/intellectual disability (from absence to moderate), language delay, language impairment mainly in the expressive domain, and physical features (facial gestalt and short stature are the most frequent). Additionally, behavioral issues such as aggressive behavior, low frustration tolerance, anxiety, and social disability are often reported in young women, such as our patient. Our observations are in line with a recent caregiver report on the characteristics of patients diagnosed with KDM5C variants [1]: all females had developmental delay and language impairment, 60% had intellectual disability, 70% had short stature, 56% had aggressive behavior, and 44% had anxiety disorder. The clinical geneticist should be aware of these specific clinical features, especially when attempting the interpretation of more variants emerging from WES analysis [22]. Indeed, reverse phenotyping has become a very useful approach in clinical genetics following the routine implementation of pangenomic tests.
Our data support a possible genotype-phenotype correlation with a milder phenotype associated with missense variants and a more severe phenotype associated with variants leading to the loss of the expression of one KDM5C allele. Indeed, in vitro functional studies [23] have demonstrated that the frameshift variant V1075Yfs*2 and the N-terminally truncating variant p.M1_E165del (family 4, Table 1) completely abolish the KDM5C histone demethylase activity, whereas two missense variants (p.D402Y and p.P480L) reduce the activity. In agreement with our conclusions, Rujirabanjerd and colleagues (15) suggested a possible genotype-phenotype correlation in two unrelated families, each with three affected males. In the first family, males had a frameshift variant (p.K1087fs*43) and were all severely affected, whereas in the second family, males had a missense variant (p.P544T) and were mildly affected. In addition, males belonging to the first family had epilepsy, short stature, and hypereflexia/spasticity, which were absent in the males of the second family. Functional characterization of these two variants [15] supported the genotypephenotype correlation observed in this study since the missense alterations affected both the tri-and didemethylase activity of KDM5C, whereas the frameshift variant resulted in a premature termination codon and a protein lacking the C terminal end and the second PHD zinc-finger domain.
A specific peripheral blood genome-wide DNA methylation signature has been detected in patients with Claes-Jensen syndrome [3]. Nine genomic regions encompassing 1769 CpGs were found to be differentially methylated in patients carrying pathogenic KDM5C variants when compared to healthy controls. Furthermore, female carriers showed less pronounced but distinctive changes in the same regions. Some of the carriers had missense variants, and others had truncating variants. Unfortunately, no methylation differences have been analyzed between the missense and truncating mutations' carriers. Moreover, it is not clear whether the carriers were all completely healthy, as stated by the authors, or if some were affected. Indeed, the same females described in the study by Schenkel and colleagues [3] are reported in a previous study [14], and some of them were described as mildly affected (three of the four p.A77T carriers). In order to confirm the emerging genotype-phenotype correlation observed in this study, it would be very useful to perform genome-wide DNA methylation in affected females carrying different types of KDM5C variants.
Indeed, epigenome signatures might be used in the future as biomarkers for diagnosis, as well as in assisting the reclassification of variants of unknown significance, as has been recently done [4]. The skewed pattern of methylation we observed in our patient suggests a possible episignature of the X chromosome. This should be confirmed by genome-wide methylation studies involving more patients.
In summary, this study has shown a possible genotype-phenotype correlation in females carrying pathogenic KDM5C variants. This correlation should be confirmed in a larger cohort and could possibly include genome-wide DNA methylation analysis. Furthermore, it is advisable for clinicians to perform deep clinical characterizations of patients in order to facilitate the reverse phenotyping process.
Author Contributions: C.L. interpreted the genetic data and wrote the manuscript; I.B. performed the genetic analysis, the variant interpretation, and the drafting of the manuscript; R.S. performed the clinical work; F.G. performed the clinical work and revised the manuscript; A.A., I.C. and M.P.C. performed the genetic analysis; P.G. revised the final manuscript. All authors have read and agreed to the published version of the manuscript.