New insights and advances in CHARGE syndrome: Diagnosis, etiologies, treatments, and research discoveries

Abstract

CHARGE syndrome is a multiple congenital anomaly condition caused, in a majority of individuals, by loss of function pathogenic variants in the gene CHD7. In this special issue of the American Journal of Medical Genetics part C, authors of eleven manuscripts describe specific organ system features of CHARGE syndrome, with a focus on recent developments in diagnosis, etiologies, and treatments. Since 2004, when CHD7 was identified as the major causative gene in CHARGE, several animal models (mice, zebrafish, flies, and frog) and cell-based systems have been developed to explore the underlying pathophysiology of this condition. In this article, we summarize those advances, highlight opportunities for new discoveries, and encourage readers to explore specific organ systems in more detail in each individual article. We hope the excitement around innovative research and development in CHARGE syndrome will encourage others to join this effort, and will stimulate other investigators and professionals to engage with individuals diagnosed as having CHARGE syndrome, their families, and their care providers.

1 |. HISTORICAL PERSPECTIVES

CHD7 (OMIM #214800) was discovered in 2004 as the main gene involved in CHARGE syndrome (OMIM #608892) (Vissers et al., 2004). Since then, more papers have been published each year on the phenotypic spectrum of the syndrome and its underlying pathogenic mechanisms. Basic studies on the function of CHD7 have also contributed to the understanding of this phenotypically complex and highly variable syndrome. The ability to molecularly confirm the diagnosis has enabled a more precise description of the full phenotypic spectrum (Bergman et al., 2011; Zentner, Layman, Martin, & Scacheri, 2010) and has improved genetic counseling on recurrence risk, recognizing that not all individuals are sporadic (Jongmans et al., 2008). With the introduction of fast and cost-efficient genome-wide variant screening techniques, more individuals with an atypical presentation of the syndrome are being identified, gradually changing the phenotypic spectrum and thus the clinical implications of the diagnosis (Hale, Niederriter, Green, & Martin, 2016). However, genome-wide screening has also revealed CHD7 variants for which clinical significance is difficult to predict, especially when the a priori chance of finding a pathogenic variant is low. This emphasizes the need for variant interpretation guidelines and functional studies (Bergman et al., 2012). The article in this issue by Hefner and Fassi provides a comprehensive discussion of topics related to genetic diagnosis and counseling in CHARGE syndrome.

There is some debate in the literature about who first recognized the combination of clinical features that, in 1981, were represented in the acronym CHARGE (coloboma—heart defect—atresia of the choanae—retardation—genital and—ear abnormalities) by Pagon, Graham, Zonana, and Yong (1981). Twenty years earlier, in 1961, Angelman (1961) described four children over a 10-year period with a primary diagnosis of coloboma and associated anomalies. Eighteen years later, this was followed by a paper by Hall (1979) describing 17 children with choanal atresia and associated anomalies, and a separate paper in the same year by Hittner, Hirsch, Kreh, and Rudolph (1979) confirming the observation by Angelman of children with syndromic ocular colobomata. This led to the early name of Hall–Hittner syndrome, later replaced by CHARGE association, reflecting the uncertainty that the combination of congenital anomalies seen were either due to a single underlying defect or a mere association based on, for instance, timing during embryogenesis. The discussion of whether CHARGE was an association or a syndrome abruptly ended in 2004, with the identification of deletions and truncating variants in the CHD7 gene in individuals fulfilling the contemporary clinical diagnostic criteria for CHARGE (Vissers et al., 2004). Later on, with the introduction of routine CHD7 sequencing in diagnostics, a new question arose which has yet to be resolved: is CHARGE syndrome a clinical or a molecular diagnosis? (Hale et al., 2016). A minority of classical individuals with CHARGE do not have an identifiable pathogenic variant or deletion in CHD7, and an increasing number of individuals with a (likely) pathogenic variant do not fulfil clinical criteria for the syndrome or only have a single feature (e.g., hypogonadotropic hypogonadism [HH]). Understanding the function of CHD7 should eventually help to distinguish between the different clinical aspects of the syndrome and thus guide this discussion. Functional studies may also reveal clues for treatment or early detection of comorbidities in subsets of individuals. See the article by Balasubramian and Crowley in this issue for a discussion of endocrine phenotypes and underlying mechanisms in CHARGE syndrome.

Recent insights into the pathogenesis and phenotypes associated with CHARGE syndrome and/or CHD7 pathogenic variants justify the publication of a new special issue on this intriguing syndrome. The last special issue was published already 12 years ago (Hartshorne, Hefner, & Davenport, 2005), and knowledge of the condition, its underlying genetic basis, and animal models exploring its pathogenesis have greatly expanded in the intervening time period. In this issue, contributing authors comprehensively present advancements in knowledge about (a) the function of CHD7 in specific organ systems; (b) newly available insights about the phenotypic consequences of Chd7 loss in these organs and tissues; and (c) how this information has led to progress toward new diagnostics and/or therapies. Tissues and organ systems emphasized are the central nervous system (e.g., see the articles by Whittaker et al. (2017) and de Geus et al. in this issue), craniofacial, auditory and vestibular structures (see the articles by de Geus et al. and Choo et al. in this issue), neural crest (see the article by Pauli et al. in this issue), heart (see the article by Corsten-Janssen and Scambler in this issue), immune system (see the article by Mehr et al. in this issue), endocrine system (see the article by Balasubramian and Crowley in this issue), gastrointestinal system (see the article by Blake and Hudson in this issue) and behavior (see the article by Hartshorne et al. in this issue). The richly detailed and complex organ specific information is compiled and highlighted in a comprehensive pheno-type-genotype study in a large French cohort of individuals with CHARGE syndrome (see the article by Legendre et al. in this issue).

2 |. ADVANCES IN DIAGNOSTICS

In clinically typical individuals with CHARGE syndrome, the tests of first choice are CHD7 Sanger sequencing and chromosomal microarray to screen for deletions and/or MLPA to test for exonic-deletions (Bergman et al., 2008; Lalani, Hefner, Belmont, & Davenport, 1993; van Ravenswaaij-Arts, Blake, Hoefsloot, & Verloes, 2015). A guideline to indicate when CHD7 diagnostics should be performed was proposed by Bergman et al. (2011). However, since the publication of this guideline, CHD7 pathogenic variants have been described in very mildly affected individuals, for example, individuals with isolated hypogonadotropic hypogonadism due to CHD7 missense variants (Balasubramanian et al., 2014). As a consequence, CHD7 has been added to next generation sequencing (NGS) gene panels for developmental delay, colobomata, heart defects, and other congenital malformations, as well as for Kallmann syndrome and normosmic HH. It is recommended that individuals with HH and a CHD7 variant be clinically screened for CHARGE syndrome features such as balance problems and deafness, amongt others (Bergman et al., 2012).

The majority of individuals with CHARGE syndrome have CHD7 variants that are easily detectable by Sanger sequencing, or by NGS techniques as long as the gene is sufficiently covered. One to two percent of individuals who test positive have an intragenic or whole CHD7 gene deletion that can be detected by microarray analysis, although for small exonic deletions, MLPA is preferred (Janssen et al., 2012). Copy number variations can also be detected by bioinformatics on whole exome (NGS) data (de Ligt et al., 2014). Advancements in these new techniques have made them more reliable (better coverage), easier to use (bioinformatics: variant calling algorithms) and cost-efficient, and has resulted in a growing number of individuals being diagnosed without a priori suspicion of CHARGE syndrome. This might be due to the syndrome being under-recognized by clinicians and/or by atypical or mild presentations of the syndrome. The advancements have led to two important challenges in clinical medicine: (1) how to best determine the pathogenic or polymorphic nature of a missense variant in an individual with a mild or atypical clinical presentation; and (2) how to identify the underlying etiology in typical individuals with CHARGE syndrome who lack a CHD7 variant or deletion.

Bergman et al. (2012) published an algorithm for the classification of missense variants. Their interpretation model was based on the combination of two computational algorithms (PolyPhen-2 and Align-GVGD), the prediction of a structural model of the chromo- and helicase domains of CHD7, and segregation and phenotype data. They showed that pathogenic missense variants are mainly present in the middle of the CHD7 gene. In the meantime, bioinformatic variant calling algorithms were enhanced, resulting in fewer variants being classified as of uncertain clinical significance (VUS).

Another aid in the classification of CHD7 variants is the free online available locus-specific database, www.CHD7.org (Janssen et al., 2012). At first publication, this database contained information on approximately 800 individuals, including clinical data when available. The current content of the database is based on approximately 1900 individuals, harboring 1000 pathogenic variants, 190 VUS and 220 benign variants. This database is useful for curating existing variants and adds additional evidence in favor of associations between specific phenotypes and genotypes. However, the gold standard to establishment of pathogenicity for missense variants remains functional tests that are, unfortunately, not readily available for CHD7.

A recent study may be instrumental to tackling this problem (Butcher et al., 2017). The authors first showed that a distinctive DNA methylation pattern was present in 19 individuals with CHARGE syndrome and a pathogenic CHD7 variant. Subsequently, they showed that this methylation signature differed between individuals with CHARGE syndrome and controls, and also differed between likely pathogenic and likely benign CHD7 missense variants. The information gained by results of this methylation study may prove helpful in distinguishing between individuals with a clinical diagnosis of CHARGE syndrome in whom CHD7 loss of function is not apparent by deletion or pathogenic variants in the gene. It is also well known that other clinical conditions may copy CHARGE syndrome, like Kabuki syndrome, 22q11.2 deletion syndrome and more rare chromosomal syndromes (see Table 1 in Janssen et al. (2012)). However, if high resolution SNP-array and whole exome sequencing do not provide clues for these overlapping syndromes, then unique methylation patterns may indicate whether or not CHD7 is involved. For example, pathogenic variants that disrupt promoter or regulatory intragenic regions may also lead to CHD7 deficiency and could present with similar methylation signatures.

TABLE 1.

Prevalence of clinical features in individuals with CHARGE syndrome with a confirmed pathogenic CHD7 variant

Feature	Lalani 2006 N = 64% (positive/observed)	Zentner 2010aN = 123% (positive/observed)	Bergman 2011 N = 280% (positive/observed)	Combined %
External ear anomaly	95 (59/62)	90 (95/106)	97 (224/231)	95
Semicircular canal anomaly	95 (21/22)	95 (37/39)	94 (110/117)	94
Coloboma	89 (55/62)	75 (85/114)	81 (189/234)	80
Choanal atresia	60 (34/57)	35 (39/113)	55 (99/179)	49
Cleft lip and/or palate	30 (18/60)	32 (35/108)	48 (79/163)	40
Cranial nerve dysfunction (VII, VIII, others)	92b (54/59)	85b (80/94)	99 (173/174)	94
Feeding difficulties			82 (90/110)	82
Facial palsy	64 (36/56)	35 (19/55)	66 (80/121)	58
Anosmia			80 (24/30)	80
Genital hypoplasia	55 (29/53)	57 (61/107)	81 (118/145)	68
Congenital heart defect	92 (54/59)	75 (86/115)	76 (191/252)	78
Tracheo-esophageal anomaly	18 (10/55)	22 (16/72)	29 (42/146)	25
Developmental delay			99 (147/149)	99
Intellectual disability		76 (64/84)c	80 (108/135)	79
Growth retardation		68 (65/96)	36 (35/94)	55

^aBased on a literature review of 28 papers, of which seven papers also contained individuals published by Lalani et al. (2006) and Bergman et al. (2011). The data of these seven papers have been excluded in this column.

^bIn Lalani et al. (2006) and Zentner, Layman, Martin, and Scacheri (2010) only information on hearing loss (nervus VIII) is given and not the combination with other affected cranial nerves.

^cIn Zentner, Layman, Martin, and Scacheri (2010), no clear distinction is made between intellectual disability and developmental (motor) delay.

Advancements in prenatal diagnostic techniques are also impacting CHARGE syndrome. Prenatally, a fetus with CHARGE syndrome may present with ultrasound abnormalities, for example, orofacial clefting, heart, renal, and limb defects (Busa et al., 2016). Subtler abnormalities like arhinencephaly, microphthalmia, and malformed outer ears may also be detected. A fetal brain MRI may support the clinical suspicion, but cannot exclude CHARGE syndrome if normal. Overall, a clinical suspicion of CHARGE syndrome will rarely be made on ultrasound abnormalities. Rather, the diagnosis can be made as a primary or secondary finding with high resolution chromosomal microarray showing a deletion or partial deletion of CHD7. Most CHD7 alterations causing CHARGE syndrome involve single nucleotides or small deletions and therefore cannot be detected by chromosomal microarray. By far the most frequent cause of CHARGE syndrome is de novo single nucleotide CHD7 sequence variants that are detectable by exome sequencing using child-parent trios (Talkowski et al., 2012).

Most individuals with CHARGE syndrome are sporadic, but recurrence has been documented (Jongmans et al., 2008). Parent-child transmission with a recurrence risk of 50% is predominantly seen in milder presentations of the syndrome, although intrafamilial variability is high and a mildly affected parent does not exclude a more severely affected child. If the pathogenic CHD7 variant of a proband cannot be detected in the parents, there remains a 2% recurrence risk due to germline mosaicism. Not surprisingly, performing invasive prenatal testing for a 2% recurrence risk may be a difficult decision for parents. Non-invasive prenatal testing (NIPT) provides an alternative wherein variants present in the fetus can be detected in blood from the mother (Bustamante-Aragones et al., 2012; Daley, Hill, & Chitty, 2014). In this case, however, low-level mosaicism for a pathogenic variant in the mother should first be excluded before NIPT can be reliably applied.

3 |. THE EVOLVING PHENOTYPIC SPECTRUM AND CLINICAL CRITERIA

Since publication of the first clinical diagnostic criteria for CHARGE syndrome by Blake et al. (1998), the clinical criteria have been adapted several times. First, because other clinical features such as abnormalities of the semicircular canals are often involved, Verloes proposed their inclusion in 2005 (Verloes, 2005). Second, identification of CHD7 resulted in broadening of the clinical phenotype, resulting in a proposition by Hale et al. to include results of CHD7 testing in the clinical criteria (Hale et al., 2016). For an overview of these clinical criteria see the article by Hefner and Fassi in this issue. It is known that in individuals with a molecularly defined diagnosis of CHARGE syndrome, the prevalence of specific features differs from those in cohorts of clinically diagnosed individuals (Bergman et al., 2011; Zentner, Layman, Martin, & Scacheri, 2010). The prevalence of the various clinical features in individuals with CHARGE syndrome and a proven CHD7 variant are summarized in Table 1. It should be kept in mind that the individuals presented in this table underwent CHD7 testing mostly because of a clinical suspicion of CHARGE syndrome. As a consequence, mildly affected individual are underrepresented. Moreover, new emerging CHD7-related clinical features are not in this table, but have been added to the clinical criteria recently (Hale et al., 2016). Examples are limb defects, as described by Van de Laar et al. (2007) and abnormalities of the cerebellar vermis recognized by Yu et al. (2013).

Several authors have tried to look at phenotype–genotype correlations. Thus far it only has been shown that individuals with a missense variant less often fulfill the clinical criteria and that especially congenital heart defects and choanal atresia are significantly less prevalent in individuals with a missense variant compared to those with a truncating variant (Bergman et al., 2012). As a consequence missense CHD7 missense variants are overrepresented in families with parent to child transmission of the disorder (Bergman et al., 2011) and in patients presenting with Kallmann syndrome or HH (see the article by Balasubramian and Crowley in this issue). However, the variability in clinical presentation is remarkable, even within families, and thus the phenotype cannot be predicted based on the genotype.

4 |. PROGRESS IN BASIC MECHANISMS OF CHD7 AND CHARGE SYNDROME

Identification of CHD7 as the causative gene for CHARGE syndrome in 2004 led to multiple avenues of inquiry to determine the pathogenesis of CHARGE-related features. First and foremost was the generation and establishment of mouse models for CHARGE syndrome. Karen Steel and colleagues in the United Kingdom were the first to report Chd7 mutations in mice, through a comprehensive study of nine different mouse strains that were first identified by screening mice generated by Ethyl-nitrosourea (ENU)-induced mutagenesis for their circling and hyperactivity behaviors (Bosman et al., 2005; Kiernan et al., 2002) (Table 2). These mice, named Whirligig (Whi), Flouncer (Flo), Tornado (Todo), Eddy (Edy), Leda (Lda), Orbitron (Obt), Cyclone (Cycn), Metis (Mt), and Dizzy (Dz) all exhibited structural malformations of the inner ear and variable features consistent with CHARGE syndrome, including choanal atresia, growth delay, cleft palate, cardiac malformations, and hypoplastic genitalia (Alavizadeh et al., 2001; Bosman et al., 2005; Hawker, Fuchs, Angelis, & Steel, 2005; Nolan et al., 1995; Pau, Hawker, Fuchs, De Angelis, & Steel, 2004; Pickard, Sollars, Rinchik, Nolan, & Bucan, 1995). Two additional strains, Carousel (Crsl) and Wheels (Whl), have features similar to other mice with heterozygous loss of function Chd7 variants, including inner ear defects and aberrant wheel running behaviors, but no confirmed Chd7 mutations (Alavizadeh et al., 2001). Wheels was found in an earlier study to map genetically to the region of mouse chromosome 4 that contains Chd7 (Bosman et al, 2005); however, the underlying pathogenic mutation for Wheels was never identified and the strain no longer exists.

TABLE 2.

Chd7 mutant mouse strains

Strain (s)	Allele	Protein effect	Reference(s)
Wheels (Whl)	Maps to chr 4	NA	Alavizadeh et al., (2001); Bosman et al. (2005); Nolan et al. (1995); Pickard et al., (1995)
Carousel (Crsl)	Not reported	NA	Bosman et al. (2005); Hawker et al., (2005)
Whirligig (Whi)	2918G-A,	W973X	Bosman et al. (2005); Hawker et al., (2005)
Flouncer (Flo)	IVS 27+ 2T-C	Donor splice site; S1864X	Bosman et al. (2005); Pau et al. (2004)
Tornado (Todo)	IVS3 + 2-C	Donor splice site; H549X	Bosman et al. (2005); Kiernan et al. (2002)
Eddy (Edy)	307C-T	Q103X	Bosman et al. (2005); Kiernan et al. (2002)
Leda (Lda)	3195T-A	Y1066X	Bosman et al. (2005); Kiernan et al. (2002)
Orbitron (Obt)	3945T-A	Y1315X	Bosman et al. (2005); Kiernan et al. (2002)
Cyclone (Cycn)	4286T-A	L1429X	Bosman et al. (2005); Kiernan et al. (2002)
Metis (Mt)	IVS 22-2A-G	Acceptor splice site; V1688X	Bosman et al. (2005); Kiernan et al. (2002)
Dizzy (Dz)	5536G-T	E1846X	Bosman et al. (2005); Kiernan et al. (2002)
Volchok (Vlk)	4705T-A	Y1459X	Lenz et al. (2010)
Looper	5690C-A	S1897X	Ogier et al. (2014)
Coa1	2155A-T	K719X	Jiang et al. (2012)
Ome	Exon 2–3 deletion	Loss of protein	Tian et al. (2012)
Chd7Gt/+	Gene trapped sequences inserted in intron 1	Loss of protein translation due to interrupted ribosomal entry site; expresses p-galactosidase from the Chd7 promoter	Hurd et al. (2007)
Chd7rrr	Gene trapped sequences inserted in intron 4	Reporter fusion	Randall et al. (2009)
Chd7xk	Gene trapped sequences inserted in intron 36	Reporter fusion	Randall et al. (2009)
Chd7flox	Floxed exon 2	Conditional loss of exon 2, containing the ATG start site for transcription	Hurd et al. (2010)
Chd7fl	Floxed exon 3	Conditional loss of exon 3	Feng et al. (2013)

Additional Chd7 mutant mice (Chd7^Gt/+) were generated by the Martin laboratory in 2007 using gene trapping technologies in embryonic stem cells (Hurd et al., 2007). Chd7^Gt/+ mice have the added advantage of expressing β-galactosidase as a Chd7 reporter, and have been useful for tracking sites of Chd7 expression during development and into adulthood. Two additional gene trap-derived Chd7 mutant lines have also been reported: Chd7^xk and Chd7^rrr (Randall et al., 2009), as well as four additional ENU-derived Chd7 mutant strains: Looper (Ogier et al., 2014), Ome (named for its otitis media phenotype) (Tian et al., 2012), Coa1 (named for its circling and ocular abnormalities) (Jiang et al., 2012) and Volchok (Vlk) (Lenz et al., 2010). Together with the spontaneous and ENU-derived alleles mentioned above, these Chd7 mutant mice have provided a highly valuable resource for investigators studying the various features associated with CHARGE syndrome, with important insights emerging about the underlying mechanisms of coloboma, cardiac development, hearing loss, olfactory dysfunction, delayed puberty, and brain malformation (Adams et al., 2007; Bergman, Bosman, van Ravenswaaij-Arts, & Steel, 2010; Gage, Hurd, & Martin, 2015; Hurd et al., 2011; Hurd, Micucci, Reamer, & Martin, 2012; Layman, Hurd, & Martin, 2011; Layman et al., 2009; Micucci et al., 2014; Randall et al., 2009).

As for other Chd7 null alleles, homozygous Chd7^Gt/Gt mice do not survive beyond E10.5, likely due to abnormalities in cardiovascular structures (Bosman et al., 2005; Hurd et al., 2007). To overcome the limitations of intrauterine lethality, researchers have also generated conditional Chd7 deletion mice (Chd7^flox/flox), which have been widely used to delete Chd7 in the heart (Randall et al., 2009), cerebellum (Feng et al., 2017; Whittaker et al., 2017), hippocampus (Jones et al., 2015), forebrain (Feng et al., 2013; Micucci et al., 2014), and developing oligodendrocytes (He et al., 2016). Conditional rescue mice have also been developed, in which deleted Chd7 can be restored upon Cre activation (Randall et al., 2009). These studies have uncovered critical roles for CHD7 in proliferation, differentiation, and survival of a wide variety of cell types, and help to explain the hearing loss, cardiac malformations, and central nervous system abnormalities that occur with CHD7 loss.

To date, 78% of pathogenic variants in human CHD7 are thought to be null alleles, while another 9% are missense and another 13% are due to splice site variants that may result in either a null allele (due to a frameshift) or a hypomorphic allele (due to exon skipping) (based on CHD7.org). Importantly, all currently available Chd7 mutant mice harbor loss of function alleles. Thus, it has not yet been possible to test whether missense variants or hypomorphic alleles exhibit phenotypes similar to those with Chd7 deficiency. A critical unanswered question is whether atypical or milder CHARGE-related phenotypes occur due to molecular mechanisms other than loss of function. With newly available CRISPR/Cas9 gene editing technologies, it is possible to more easily generate new mouse alleles such as point mutations that mimic those found in humans. Given their long gestation (19 days), however, mice are not an efficient model for high throughput functional studies of multiple alleles, and such studies will likely require use of zebrafish (Danio rerio) or flies (Drosophila).

In zebrafish, disruption of chd7 leads to abnormalities in otolith formation, cardiac development, and craniofacial structures (Jacobs-McDaniels and Albertson, 2011; Patten et al., 2012). CRISPR/Cas9 approaches were recently used to generate novel chd7 zebrafish alleles for use in functional studies with relatively high throughput (Balasubramanian et al., 2014; Prykhozhij, Steele, Razaghi, & Berman, 2017). Interestingly, concomitant reduction of fbxl10, a histone methyltransferase, partially rescues chd7 mutant phenotypes, suggesting that chd7 mediates its effects via activation of histone methylation (Balow et al., 2013). Downregulation of the transcription factor SOX10 also leads to rescue of neural crest related phenotypes in chd7 mutant zebrafish, suggesting that multiple convergent mechanisms involving histone methylation are critical for regulating proper organ and tissue development (Asad et al., 2016).

In Drosophila, the gene Kismet is an orthologue of all 4 class III mammalian CHD paralogous genes (CHD6, CHD7, CHD8, CHD9). Kismet mutant flies were originally described in a screen for dominant suppressors of the Polycomb class of homeotic gene repressors (Daubresse et al., 1999). Kismet flies exhibit homeotic transformations in the fifth abdominal segment and in the leg that may result from defects in transcriptional elongation and/or loss of interactions between Kismet and RNA polymerase at methylated histone Lysine 36 residues (H3K36me) in gene bodies (Daubresse et al., 1999; Dorighi & Tamkun, 2013; Srinivasan et al., 2005). Kismet also suppresses expression of hedgehog, a signaling factor and morphogen, in the wing imaginal disc (Terriente-Felix, Molnar, Gomez-Skarmeta, & de Celis, 2011).

In addition to mice, zebrafish, and flies, other metazoans are also useful for studying chd7 related functions. Xenopus laevis (frog) eggs can be easily electroporated, and have been used to study functional effects of chd7 mutations on neural crest cell development and to validate gene expression studies (Bajpai et al., 2010; Schulz et al., 2014). In chick (Gallus gallus), the Chd7 orthologue (cChd7) is highly expressed in the developing nervous system, including optic and otic placodes (Aramaki et al., 2007). To date, however, no studies have been published on the functions of chd7 in the chick, despite its widespread use in developmental biology and ease of manipulation by electroporation and in ovo monitoring of development (Aramaki et al., 2007). C. elegans worms have highly regulated and well-defined numbers of neurons and other cell types, their entire cellular lineages have been defined, and they also harbor a chd7 orthologue, chd-7, making them a model of choice for genetics of cellular differentiation and fate determination.

The plethora of animal models available to study CHD7 function and biology is truly remarkable, and will surely lead to important new discoveries relevant to CHARGE syndrome. Interestingly, unicellular model organisms may also prove beneficial. The first study examining the function of CHD genes was performed in the single cell organism Saccharomyces cerevisiae (baker’s yeast) (Flanagan et al., 2007; Woodage, Basrai, Baxevanis, Hieter, & Collins, 1997). Targeted deletion of ScCHD1 led to protection against the cytotoxic agent 6-azauracil, suggesting that ScCHD1 was a transcriptional inhibitor (Woodage et al., 1997). Dictyostelium discoideum, a single celled organism that forms multicellular fruiting bodies in the absence of nutrients, contains a gene called chdC, an apparent ortholog of mammalian class III CHD genes (Platt, Rogers, Rogers, Harwood, & Kimmel, 2013). Null mutants for chdC have severely impaired growth, cellular differentiation, and chemotaxis, suggesting broad functions for CHD proteins in early stages of development (Platt et al., 2013).

Single-cellular and multi-cellular animal models of CHARGE have provided essential insights about tissue and organ specific functions for CHD7. However, cell type-specific experiments can be technically challenging in animals, since specific cells must be accurately isolated either mechanically or by flow cytometry from surrounding cells and tissues. To overcome this limitation, cell-based models to study CHARGE syndrome have also been employed, including lymphoblastoid cell lines from individuals with CHARGE syndrome (Lalani et al., 2006), embryonic stem cells derived from human embryos, and embryonic stem cells from heterozygous and homozygous Chd7 mutant mice (Bajpai et al., 2010; Schnetz et al., 2009, 2010; Zentner et al., 2010). One study demonstrated that loss of CHD7 impairs the ability of human derived ES cells to migrate to neural crest after being implanted into frog embryos (Bajpai et al., 2010). Future studies of human induced pluripotent stem cells derived from individuals with CHD7 pathogenic variants and CHARGE syndrome should also reveal important roles for CHD7 in stem cell development and function.

Collectively, research using animal models and cell based assays has led to several proposed mechanisms for how CHD7 disruption leads to the clinical features observed in CHARGE syndrome. Zentner et al. proposed loss of ribosomal biogenesis as a unifying mechanism, based on CHD7 localization to ribosomes and loss of rRNA gene expression in Chd7 mutant tissues (Zentner et al., 2010). CHD7 binds to over 10,000 sites in the mammalian genome, with enrichment at methylated histone Lysine 4 (H3K4me) residues in gene promoter and enhancer regions (Schnetz et al., 2009, 2010; Zentner and Scacheri, 2012). Wysocka and colleagues proposed that CHD7 interacts with PBAF proteins to regulate downstream target genes which promote neural crest development (Bajpai et al., 2010). A similar paradigm was also observed in a recent series of experiments showing that CHD7 interacts with SOX10 to regulate oligodendrocyte development (He et al., 2016). Interestingly, CHD7 also regulates cerebellar granule cell development via activation of Reelin expression (Whittaker et al., 2017) and inner ear neurogenesis via activation of neurogenic genes (Hurd, Poucher, Cheng, Raphael, & Martin, 2010). Combinations of these and other currently undiscovered functions for CHD7 are likely to be involved in CHARGE syndrome.

5 |. BUILDING BRIDGES BETWEEN DIAGNOSIS AND THERAPIES

Individuals with CHARGE syndrome exhibit multiple structural malformations that affect development of the eyes, ears, craniofacial structures, heart, gut, skeleton, and genitourinary system. Given that these organs form early in development (before the 8th week of human gestation), fully effective interventions would likely need to occur very early during pregnancy, in a woman who knows her fetus is affected by a CHD7 pathogenic variant. Current standard prenatal screening for CHD7 variants is typically limited to familial cases, most often via chorionic villus sampling or amniocentesis at 10–12 and 18–20 weeks’ gestation, respectively. Since the vast majority of individuals with CHARGE syndrome have de novo CHD7 pathogenic variants, the paradigm often does not apply. Notably, newly available clinical tests in commercial laboratories use non-invasive prenatal testing (NIPT) of maternal blood to offer CHD7 as one of several candidate genes to be screened. Since NIPT can be performed as early as 10 weeks’ gestation, it is no longer impossible to conceive that a mother could learn about the CHD7 status of her fetus in time to start effective intervention, if it existed.

To date, children and adults with CHARGE syndrome receive a wide variety of supportive and corrective therapies, including repair of retinal detachments, surgical repair of cardiac and mediastinal/esophageal/gut malformations, gastrostomy tubes, tracheostomies, hernia repairs, limb corrective surgeries, and scoliosis surgeries, among others. In addition, therapies targeted toward augmentation of existing skills are highly beneficial. These include hearing aids, cochlear implants, BAHA (Bone Anchored Hearing Aid) devices, feeding pumps, orthotics, braces, walkers, walking sticks, wheel-chairs, communication devices, auditory amplification devices, visual amplification devices (magnifiers), and customized/adaptive bicycles. These therapies are vital for survival and favorable outcomes in CHARGE; however, none of these approaches is considered curative.

Several lines of inquiry have focused on potentially curative therapeutic modalities to treat CHARGE syndrome. Given that CHD7 regulates expression of other genes, much of the research to date has centered on the identification of key downstream targets of CHD7 that mediate major effects in the cells and tissues where CHD7 is expressed. The logic behind this approach is that once these genes are known, they may be amenable to manipulations targeting their up- or down-regulation to overcome the disruptions in expression that occur with CHD7 loss.

Another approach for discovering therapeutic targets for CHARGE syndrome is directed toward identification of the cellular and biochemical signaling pathways and mechanisms that are disrupted with CHD7 deficiency. One successful example of this has been a report showing that changes in retinoic acid signaling can partially rescue Chd7 loss of function phenotypes in the inner ear and brain (Micucci et al., 2014). Other recent examples include a study showing that inhibition of Topoisomerase rescues cerebellar defects in Chd7 mutant mice (Feng et al., 2017) and another which found that inhibition of Reelin, a glycoprotein target of CHD7, corrects Chd7 mutant mouse cerebellar granule cell abnormalities (Whittaker et al., 2017).

Clearly, there are many challenges that need to be overcome before any of these pre-clinical research discoveries can be translated into clinical trials. First, the accurate and effective delivery of replacement genes, corrective genes, or pharmacological agents requires that they not induce major off-target effects. Second, any gene replacement strategy would need to demonstrate lack of toxicity and high tropism for relevant cells and tissues. Third, therapies aimed at early embryonic timepoints would need to be administered in utero and lack adverse effects on the mother.

These myriad challenges are significant, and it is unclear whether replacement of lost CHD7 function later in development will be sufficient to overcome early defects. Thirteen years after the original discovery of CHD7 as the causative gene for CHARGE, however, (Vissers et al., 2004), scientists have performed a great deal of research and made major discoveries. Together, these successes paint an optimistic picture for the future of research and discovery in CHARGE syndrome. We hope that additional exciting breakthroughs will be made that ultimately translate to improvements for the lives of individuals with CHARGE syndrome and their families.

Biographies

C. van Ravenswaaij-Arts, MD, PhD is a consultant in Clinical Genetics and Professor in Dysmorphology at the Department of Genetics of the University Medical Center Groningen, Netherlands. Her research projects focus on neurodevelopmental syndromes, with a special interest in CHARGE syndrome and rare chromosomal dis orders. She coordinates an accredited center of expertise for both these disorders and is a member of the European Reference Network ITHACA.

D. M. Martin, MD, PhD is Professor of Pediatrics and Human Genetics and the Donita B. Sullivan Research Professor in Pediatrics at the University of Michigan Medical School in Ann Arbor, Michigan. She evaluates and cares for individuals with CHARGE syndrome and other genetic conditions. Her areas of expertise are human genetics, mouse genetics, and developmental disorders of the nervous system.