Emerging X-linked genes associated with neurodevelopmental disorders in females

Abstract

A significant source of risk for neurodevelopmental disorders (NDDs), including intellectual disability (ID) and autism spectrum disorder (ASD), lies in genes located on the X chromosome. Males can be particularly vulnerable to X-linked variation because of hemizygosity, and male-specific segregation in pedigrees has guided earlier gene discovery for X-linked recessive conditions. More recently, X-linked disorders disproportionally affecting females, with complex inheritance patterns, and/or presenting with sex differences have surfaced. Here, we discuss the genetics and neurobiology of X-linked genes that are paradigmatic to understand NDDs in females. Integrating genetic, clinical, and functional data will be key to understand how X-linked variation contributes to the risk architecture of NDDs.

Introduction

NDDs, including ID and ASD, are a group of conditions emerging during childhood. Male bias in diagnosis has been reported for ID (~1.4:1) and ASD (~4:1) [1,2], and as a result these disorders are understudied in the female population [3]. A significant portion of risk for adverse neurodevelopmental outcomes resides in genes located on the X chromosome. >80% of genes on the X chromosome are associated with ID, seizures, or language impairment [4] and ~6% of NDDs are estimated to arise from X-linked coding variation [2].

Here, we review the relevance of X-linked genetic variation in NDDs and then dive into 7 X-linked genes associated with NDDs (DDX3X, USP9X, KDM6A, KDM5C, SMC1A, HDAC8, and PCDH19) (Table 1). These genes were prioritized because of asymmetries in the prevalence and/or presentation of the associated clinical manifestations across sexes and/or new emerging evidence that illuminates the mechanisms of NDDs in females. Other more intensively studied X-linked genes associated with NDDs (e.g., MECP2, FMR1, CDKL5) are well covered elsewhere ([4] and references therein). We believe that these 7 genes offer a case in point for the complex intersection between sex and X chromosome biology and are paradigmatic to understand sex differences in neurotypical development and NDDs.

Table 1.

X-linked genes associated to NDDs discussed in this study. The list of clinical features includes the most frequently observed (please refer to the references and/or OMIM for a more comprehensive clinical description). DD/ID, developmental delay/intellectual disability.

Gene	Disorder	Clinical Features	Inheritance pattern	XCI escapee	Y homologs	References
DDX3X	DDX3X syndrome (MIM #300958)	ID/DD, ASD, hypotonia, seizures, motor delays, movement disorders, brain congenital malformations	X-linked semi-dominant	Yes	DDX3Y	14–24
USP9X	X-linked ID (MIM #300919)	ID/DD, hypotonia, behavioral abnormalities	X-Linked recessive	Yes	USP9Y	25–28
	Female-restricted X-linked ID (MIM #300968)	ID/DD, motor delays, congenital heart defects, pigment abnormalities	X-linked dominant
KDM6A	Kabuki syndrome (MIM # 300867)	ID/DD, neonatal hypotonia, feeding difficulties, distinctive facies, congen ital anomalies	X-linked dominant	Yes	KDM6C	29–33
KDM5C	Claes-Jensen syndrome (MIM #300534)	ID/DD, short stature, hyperreflexia, microcephaly, aggressive behaviors	X-linked recessive	Yes	KDM5D	34–37
SMC1A	Cornelia de Lange syndrome (MIM #300590)	ID/DD, facial dimorphisms, growth delays, limb defects, congenital heart defects	X-linked dominant	Yes	--	38–42
SMC1A	Developmental and epileptic encephalopathy (MIM #301044)	ID/DD, early-onset severe refractory seizures, holoprosencephaly	X-linked dominant	Yes	--	38–42
HDAC8	Cornelia de Lange syndrome (MIM # 300882)	ID/DD, growth retardation, dysmorphic facial features, distal limb defects, delayed fontanel closure, ocular hypertelorism, hooding of the upper eyelids, bulbous nasal tip, dental anomalies, nevus flammeus	X-linked dominant	No	--	40, 43, 44
PCDH19	Female-restricted developmental and epileptic encephalopathy (MIM #300088)	ID/DD, infant-onset epilepsy, autistic features	X-linked	No	--	45–48

Dynamics of X chromosome biology

In therian mammals, females have two copies of the X chromosome and males have one X and one Y chromosome. Harmonization of dosage for X-linked genes across sexes is achieved through X chromosome inactivation (XCI), a mechanism of epigenetic silencing of one of the X chromosomes in females (excluding the pseudoautosomal regions) [5]. The dynamics and molecular mechanisms of XCI differ across species, and silencing can involve the paternal X-chromosome (imprinted XCI) or either maternal or paternal chromosomes leading to tissue mosaicism (random XCI) [5,6]. In mouse, imprinted XCI ensues shortly after fertilization, followed by reactivation of the paternal X chromosome in the cells of the blastocysts that will give rise to the embryonic tissues, and then random XCI in the embryo proper [5,6] (Fig. 1A). In humans, the paternal X chromosome is not silenced pre-implantation and random XCI takes place in post-implantation embryos [5] (Fig. 1B). XCI is driven by genes on the X inactivation center, including the X inactive-specific transcript (XIST) locus. XIST encodes a long non-coding RNA that recruits in cis epigenetic regulators that induce stable silencing [5]. As hypothesized in humans, before embryo implantation there is biallelic expression of XIST (hence lack of imprinted XCI), which coats both X chromosomes, producing a dampening of expression of X-linked genes. After implantation, random expression of XIST in either the paternal or the maternal allele results in random XCI [5] (Fig. 1B). Notably, XCI status is highly unstable in cultured human embryonic stem cells (ES) or induced pluripotent stem cells (hiPSCs) and can undergo “erosion”, a phenomenon characterized by loss of XIST expression and coating, with consequent re-expression of silenced X-linked genes [7].

Figure 1. Mechanisms of X chromosome inactivation in mouse and human (XCI).

A.In the mouse after fertilization, the paternal chromosome is preferentially silenced through the upregulation of X inactive-specific transcript (XIST) (orange), which encodes a long non-coding RNA that binds to the chromosome and recruits epigenetic factors (blue) to induce stable silencing (imprinted XCI). Continuing from blastocyst to embryo, the maternal imprint (denoted in red), which represses the maternal expression of XIST, is removed and no preference is given to silencing of the maternal or paternal chromosome (random XCI). Xp: paternal X; Xm: maternal X; Xa: active X; Xi: inactive X. B. In humans, before implantation overall X chromosome dampening in both X chromosomes through the biallelic expression of XIST is hypothesized. After implantation, there is a transition to random XCI where XIST and the recruitment of epigenetic factors occurs randomly between the maternal and paternal chromosome C. This panel reports the genes discussed in this review, DDX3X, USP9X, KDM6A, KDM5C, SMC1A, HDAC8, and PCDH19, their relative location on the X chromosome, and their status as escapees are annotated.

XCI can be skewed, with the preferential inactivation of one X chromosome in >80% (even 90%) of somatic cells. Skewed XCI can be detected in asymptomatic mothers of male patients with X-linked recessive disorders (as a protective mechanism favoring the non-mutant allele) or in affected females with pathogenic mutations in X-linked recessive genes. The latter scenario has been detected in up to 13% of NDD cases in females that had remained genetically undiagnosed after testing for Fragile X syndrome, chromosomal microarray analysis, and trio exome sequencing [8].

~23% of human X-linked genes escape XCI and are expressed from both alleles in cells with XX karyotype [9]. Some escapee genes are constitutive while others show inter-individual variability or tissue and/or developmental specificity (facultative escapees) (Fig. 1C, Fig. 2). For example, data on 5,500 transcriptomes from 449 individuals across 29 human tissues [9] show how KDM6A, a known constitutive escapee, is expressed in female tissues at higher levels than male tissues ubiquitously, unlike HDAC8 (Fig. 2). The mechanisms behind XCI escape and the factors discriminating constitutive and facultative escapees are not fully understood. Constitutive escapees typically have homologues on the Y chromosome that may or may not equalize their dosage, also depending on the tissue and/or developmental window [10,11]. A screening across 8 mammalian species (from opossum to human) has identified 36 X-Y pairs, including 4 of the 7 that we discuss below [DDX3X/DDX3Y, KDM6A(UTX)/KDM6C(UTY), KDM5C/KDM5D, USP9X/USP9Y] [11]. As we will discuss, some of these Y homologues have lost part of their function [10] or acquired tissue-specific expression to execute male-specific functions (e.g., [12]). Considering that many genes on the Y chromosome were lost during evolution, the retention of Y homologues might provide mechanisms to offset the dosage unbalance and, in turn, force the X homologues to escape XCI [10].

Figure 2. Sex differences in the expression of the 7 genes discussed in this review.

The heatmap shows the fold change in female vs. male tissues, based on data on 5,500 transcriptomes from 449 individuals across 29 human tissues annotated in Supplementary Table 2 of Tukiainen et al [9]. DDX3X, KDM5C, KDM6A, SMC1A, and USP9X show higher expression in females, consistent with their status as escapee genes. Orange tiles, higher expression in females compared to males; purple tiles, higher expression in males compared to females; white tiles, comparable expression between sexes; and, gray tiles, lack of data for that gene in that tissue.

DEAD-box helicase 3 X-linked (DDX3X)

DDX3X encodes a DEAD-box ATP-dependent RNA helicase that promotes the unwinding of RNA structures [13–15] and the organization of RNA-containing phase-separated organelles via liquid-liquid phase separation [16], with broad implications for mRNA metabolism.

Mutations in DDX3X cause DDX3X syndrome (MIM #300958), an NDD that manifest with ID/developmental delay often in co-morbidity with ASD, hypotonia, seizures, motor delays, movement disorders, and brain congenital malformations [14,17]. DDX3X syndrome has a semi-dominant pattern of inheritance [2]: most patients are females with de novo mutations causing haploinsufficiency or missense mutations, and the few identified males have missense mutations typically inherited from asymptomatic mothers [14,18]. DDX3X escapes XCI in human [9] (Fig. 2) and mouse [6,19].

DDX3X has a homolog on the Y chromosome, DDX3Y [11]. Molecularly, DDX3X and DDX3Y can have redundant functions in mRNA translation in human cell lines [20], but with differing processivity, as DDX3Y has weaker ATPase activity and stronger propensity for liquid-liquid phase separation [16]. Whether DDX3Y compensates for DDX3X is debated and likely varies across species. For example, more severe disruptions of brain development have been reported in brain-conditional Ddx3x female null mice compared to their male counterparts, which also have upregulated expression of Ddx3y mRNA [15,21]. Additionally, in humans and primates (but not mice), DDX3Y is located on a specific region of the Y chromosome associated with male fertility, the azoospermia factor region. Furthermore, human DDX3Y has evolved alternative polyadenylation sites that are used specifically in testis and define the function of DDX3Y in male germ cells [12]. In fact, mutations or deletions in DDX3Y lead to infertility/subfertility, without reported brain disorders [22].

Ddx3x is fundamental for both embryonic and placental development [23]. Null male mice die in utero [19,23]. In line with the permanent imprinted XCI on the paternal chromosome in extraembryonic tissues [5,6], germline mutations on the maternal (but not paternal) chromosome are embryonic lethal in females [23]. During neural development, Ddx3x is critical for the development of the forebrain [23] and hindbrain [21]. Within the forebrain, Ddx3x regulates the proliferation of neural progenitors and their differentiation into glutamatergic neurons [14,15], by controlling the mRNA translation of genes critical for cell cycle [15], as well as the laminar identity of cortical neurons [19]. Ddx3x haploinsufficiency in mice causes developmental, sensory, and motor delays that then translate into adult deficits, including hyperactivity, anxiety-like behaviors, and cognitive impairments [19].

Ubiquitin Specific Peptidase 9 X-linked (USP9X)

USP9X encodes for is a deubiquitylating enzyme that targets proteins critical for neurodevelopment [24].

Mutations in USP9X have, to some extent, distinct presentations in females and males. Mutations in males follow an X-linked recessive pattern of inheritance (MIM # 300919), and they can be maternally inherited or de novo and lead to partial loss-of-function [25,26] contributing to developmental delays and ID. De novo mutations resulting in complete loss of function are detected exclusively in females [27] and can present with greater severity and/or with clinical features that are female-restricted, including congenital heart defects and pigment abnormalities [26] (MIM #300968). USP9X is an escapee gene (Fig. 2) [9,27]. However, alterations in skin pigmentations or facial asymmetries were noted in females with USP9X mutations [27], suggesting mosaic patterns of escape.

USP9X is in a X-Y pair with USP9Y [10,11]. Like DDX3Y (see above, [12,22]), USP9Y seems to have evolved male tissue specificity, as it is linked to infertility (MIM #415000) [24].

Importantly, Usp9x might not escape XCI in mouse (see [26] but also [6]), and studies have focused on male mice. Complete knockout of Usp9x is embryonic lethal, and forebrain-specific Usp9x^−/y male mice show behavioral deficits [24,26], loss of hippocampal neurons [26], and defects in synaptogenesis [24]. While these models demonstrate the role of Usp9x in brain development and behavior, sex differences have not been explored due to lack of female heterozygotes in the current models.

Lysine Demethylase 6A (KDM6A)

KDM6A, also known as UTX, encodes a histone demethylase that mediates the removal of repressive tri-methylation of lysine 27 of histone H3 (H3K27me3) to activate gene expression [28,29].

De novo (or more rarely inherited) mutations in KDM6A cause Kabuki Syndrome (MIM #300867), an NDD manifesting with neonatal hypotonia, feeding difficulties, developmental delay/ID, distinctive facies, and multiple congenital anomalies affecting heart, intestine, kidney, and skeleton [30]. KDM6A mutations follow an X-linked dominant pattern of inheritance, with no significant sex bias in prevalence. KDM6A escapes XCI (Fig. 2) [9]. This likely contributes to the greater phenotypic variability in females [30] and the higher frequency and severity of neurodevelopmental problems in males, who are on average born earlier, present more gastrointestinal problems, and have more profound ID [31].

KDM6A is in an X-Y pair with KDM6C/UTY [11], but there might be a functional divergence between these two proteins, as demethylase activity for KDM6C has yet to be demonstrated [10].

KDM6A is essential for neural development [28,29]. KDM6A-deficient neurons from human pluripotent stem cells show defective differentiation, decreased neurite complexity, electrophysiological defects, and alterations in gene expression [31]. In mouse, Kdm6a is also a constitutive XCI escapee and contributes to the reversal of imprinted XCI on the paternal chromosome [6]. Kdm6a is essential for embryogenesis: Kdm6a^−/− females die prenatally, and only a small fraction of Kdm6a^−/y male mice survive to adulthood [32]. In utero downregulation of Kdm6a in the cortex results in increased progenitor proliferation and reduced neurogenesis, and phenotypes are more severe in female than male mice [28,29]. Downregulating both Kdm6a and Kdm6c in males enhances some of the impairments observed in cortical development [28]. These data suggest that murine Kdm6c, perhaps via demethylase-independent functions, might partially compensate Kdm6a loss [29].

Lysine demethylase 5C (KDM5C)

KDM5C (also known as JARID1C) encodes a histone H3 lysine 4 di- and tri-methyl (H3K4me2/3)-specific demethylase.

Mutations in KDM5C cause Claes-Jensen syndrome (MIM #300534), an NDD characterized by ID, short stature, hyperreflexia, microcephaly, and aggressive behaviors. The disorder was initially described in males with maternally inherited mutations, but female carriers of heterozygous mutations have emerged to have learning disabilities, speech impairments, and behavioral problems [33]. In genomic analyses, KDM5C is enriched for de novo mutations in males compared to females [2].

KDM5C escapes XCI (Fig. 2) [9] and has a Y homolog, KDM5D [10,11]. Recent data unveiled that KDM5C enhances the expression of XIST in placental and marsupial mammals [34], thus contributing to XCI. Females with Claes-Jensen syndrome can have skewed XCI, but the relationship between skewing status and phenotypes remains inconclusive [33].

Studies in mice show that Kdm5c represses gene expression by regulating H3K4me3 levels both in the developing and adult brain, with impact on activity-regulated enhancers during neuronal maturation [35]. Also, Kdm5c^−/− females die in utero [34]. Kdm5c^y/− males or Kdm5c^+/− females have a growth delays and behavioral phenotypes reminiscent of Claes-Jensen syndrome, with more penetrant deficits in mutant males [36].

Structural Maintenance of Chromosome 1A (SMC1A)

SMC1A encodes a subunit of the cohesin complex, which regulates chromatin segregation during mitosis, genome organization, and transcription in post-mitotic neurons [37]. Mutations in SMC1A can cause two distinct X-linked dominant disorders.

First, missense and in-frame deletions/duplications contribute to ~5% cases of Cornelia de Lange Syndrome (CdLS, MIM #300590). CdLS is a clinically and genetically heterogenous NDD characterized by ID, facial dimorphisms, growth delays, limb defects, and congenital heart defects [38,39]. SMC1A mutations are associated with milder CdLS phenotypes when compared to mutations in the NIPBL gene that account for most cases [38,39]. SMC1A-associated CdLS can be diagnosed in both males and females [38,39], with a higher burden of de novo mutations in females [2].

Second, SMC1A mutations leading to loss of function cause a developmental and epileptic encephalopathy with or without midline brain defects (MIM #301044), which present with early-onset severe refractory seizures and can co-occur with holoprosencephaly (i.e., incomplete division of the embryonic forebrain) [38]. Patients with this condition are all females [38,40], suggesting that loss-of-function mutations are compatible with life only in females because SMC1A is an escapee (Fig. 2) [9]. Interestingly, Smc1a also contributes to the reactivation of the inactive X chromosome during the reprogramming of differentiated cells into pluripotent stem cells in mouse [41].

Histone Deacetylase 8 (HDAC8)

HDAC8 is a cohesin regulatory protein, as its deacetylases the cohesin subunit SMC3 to regulate the binding of the complex to chromatin during cell division [42].

Mutations in HDAC8 contribute to another 2–4% of CdLS cases and can present with either classic or non-classic CdLS manifestations (MIM #300882) [39,42,43]. HDAC8 mutations can be identified in both females and males [39,42], but with a higher burden of de novo mutations in females [2]. By virtue of HDAC8 undergoing XCI (Fig. 2) [9], affected females often display extremely skewed XCI and males tend to be more severely affected [42]. Also, asymptomatic mothers with skewed XCI can transmit HDAC8 pathogenic variants to their male offspring [42], although de novo mutations are most common [39]. The functional convergence on the cohesin complex underlies the overlapping clinical manifestations resulting from the various classes of mutations contributing to CdLS [39].

Protocadherin 19 (PCDH19)

PCDH19 encodes protocadherin 19, a transmembrane protein that mediates calcium-dependent cell-cell adhesion by homophilic binding through extracellular cadherin domains [44].

Mutations in PCDH19 cause a form of developmental and epileptic encephalopathy (MIM #300088) characterized by seizure onset in infancy, ID, and autistic features [44]. Unlike other X-linked disorders, including those we have reviewed above, this condition affects mainly females with heterozygous mutations and hemizygous male carriers are typically unaffected [45]. This unusual landscape is explained by cellular mosaicism generated by random XCI of PCDH19 (Fig. 2), where there is ‘cellular interference’ between cells expressing the wild-type PCDH19 and cells expressing the mutant PCDH19 [45]. This model is supported by genetic evidence, e.g., affected males with a postzygotic somatic mutation [46], and findings in cellular and animal models [44,47,48]. For example, heterozygous Pcdh19 female mice, but not hemizygous male mice, show deficits in trans-synaptic interactions between protocadherin-19 and N-cadherin, leading to impaired synaptic development and physiology of hippocampal mossy fibers and cognitive impairments [48]. Within the mosaic brains of heterozygous Pcdh19 females, neurons lacking PCDH19 display hyperexcitability, which might contribute to changes in neuronal network activity and connectivity and increased seizure susceptibility [48]. Protocadherin genes, including PCDH19, are under complex epigenetic regulation downstream to gonadal hormone receptors [49], which might contribute to sex differences in expression (e.g., Fig. 2).

Conclusions

Gene discovery for X-linked NDDs has historically focused on males due to the vulnerability of hemizygosity in XY karyotypes. However, disorders disproportionally affecting females have emerged, revealing a spectrum of inheritance modes, e.g., semi-dominant (e.g., DDX3X [2,17]), co-existing recessive and dominant (e.g., USP9X [25–27]), and unique patterns due to cellular mosaicism (PCDH19 [45,46]). These conditions have reaffirmed the complex evolution and regulation of the sex chromosomes, with variability in the timing and tissue specificity of XCI [5,6], aspects of XCI dynamics regulated by escapees [6,34,41], and X-Y pair balance [10]. Specie-specific differences in escapee status or dosage compensation from Y partners [11,12,26] (Fig. 1) are critical to consider when studying NDDs X-linked genes in rodents. On the other side, faithfully recapitulating XCI in female hiPSCs-derived models has been challenging, due to inactivation variability and erosion [5,7]. Efforts aiming at mitigating these limitations are ongoing (e.g., [7]) and are expected to propel the development of improved female experimental models.

We note that in this review we discuss sex in relationship to XX and XY karyotypes, but it is important to consider the intersection between sex assigned at birth and gender identity, as gender diversity has been reported at higher rates in individuals with ASD [50]. We also note that, while we cover here genetic variation on the X chromosome, sex differences in NDDs can also arise from gene-by-sex effects on the autosomes [1,51].

In conclusion, we argue that genetic, clinical, and neurobiological analyses of forms of X-linked NDDs prevalent in females may guide our understanding of sex differences during brain development and how they shape risk and manifestations for these conditions.

Highlights.

• Chromosome X is enriched for genes associated with neurodevelopmental disorders

• X chromosome inactivation evolved to harmonize gene dosage across sexes

• Genes can escape inactivation, contributing to sex differences in neurodevelopment

• X-linked neurodevelopmental disorders more prevalent in females have emerged