CHD7 mutations are implicated in a majority of cases of the congenital disorder, CHARGE syndrome. CHARGE, an autosomal dominant syndrome, is known to affect multiple tissues including eye, heart, ear, craniofacial nerves and skeleton and genital organs. Using a morpholino-antisense-oligonucleotide-based zebrafish model for CHARGE syndrome, we uncover a complex spectrum of abnormalities in the neural crest and the crest-derived cell types. We report for the first time, defects in myelinating Schwann cells, enteric neurons and pigment cells in a CHARGE model. We also observe defects in the specification of peripheral neurons and the craniofacial skeleton as previously reported. Chd7 morphants have impaired migration of neural crest cells and deregulation of sox10 expression from the early stages. Knocking down Sox10 in the zebrafish CHARGE model rescued the defects in Schwann cells and craniofacial cartilage. Our zebrafish CHARGE model thus reveals important regulatory roles for Chd7 at multiple points of neural crest development viz., migration, fate choice and differentiation and we suggest that sox10 deregulation is an important driver of the neural crest-derived aspects of Chd7 dependent CHARGE syndrome.
CHD7 is a chromodomain helicase DNA-binding domain (CHD) protein and is a member of the SWI-SNF superfamily of ATP-dependent chromatin remodelers that bind to DNA and modulate gene expression (1). CHD7 has been shown to bind to gene enhancer elements and promoters, functioning as a transcriptional co-regulator (2,3). In vitro assays have shown that CHD7, powered by ATP hydrolysis remodels chromatin and that disease-causing mutations identified in patients impair this function of the protein (4). CHD7 co-localizes with the histone acetyl transferase, P300 and the stem cell pluripotency factors OCT4, SOX2 and NANOG on enhancers to regulate transcription (5). The occupancy of CHD7 correlates both with activator as well as repressor functions on these enhancers (3,6), however, the functional relevance of these activities is still not completely clear.
Mutations in CHD7 cause CHARGE syndrome (OMIM ID: 214800), which is characterized by defects in multiple tissues ranging from eye, heart, craniofacial skeleton, vertebrae, genital organs, with neuronal disorders, deafness and general growth retardation (7–9). Mutations along the whole length of the protein have been associated with CHARGE syndrome (10) suggesting that most domains in the protein are functionally important. CHARGE syndrome models have been created using morpholino based antisense oligonucleotide knockdowns of chd7 in xenopus (11) and zebrafish (12,13), mutations and knock-outs in mouse (14,15) and drosophila (16) as well as by shRNA-based knockdown in human stem cells (11). A few of these models have identified defects in neural crest derived organs implying a conserved role for CHD7 in the neural crest.
Neural crest (NC) is a transient population of multipotent cells in the embryo that emerges from the ectoderm to migrate and populate a diverse set of tissues in the body (17). Abnormalities in many of these tissues, such as the craniofacial skeleton and peripheral nerves, are major contributors to CHARGE syndrome. In cultured human neural crest like cells (hNCLCs) CHD7 has been shown to be a part of the PBAF (polybromo-and BRG1-associated factor-containing) complex that binds to and activates enhancers of NC specific genes such as SOX9 and TWIST1 (11). CHD7 has been demonstrated to be important for maintaining multipotency of NC cells in mouse neural crest stem cells when complexed with BRG1 and other proteins (18). SOX2, a transcription factor with an important role in the epithelial to mesenchymal transition of NC cells (19), has been shown to physically interact with CHD7 to co-regulate gene expression (3). Microarray analysis of a mouse mutant for Chd7 revealed its role in the regulation of a number of genes that are important for NC migration (20). Also, NC-specific conditional deletion of Chd7 in mouse causes deformities in the craniofacial skeleton (21). However, the role of Chd7 in the various stages of neural crest development and differentiation has not yet been explored.
Here we investigate the effects of chd7 knockdown on the development and differentiation of zebrafish neural crest. We show that chd7 deficiency affects early NC gene expression as well as NC migration. We demonstrate that Chd7 is critical for fate determination of NC derivatives and that its deficiency adversely affects crest-derived precursor as well as differentiated populations of craniofacial cartilage, cranial ganglia and enteric neurons. We find that Chd7 is crucial for the differentiation of pigmented cells and myelinated Schwann cells, which has not been reported previously. In the chd7 morphants we also observed a deregulation of the transcription factor sox10, known for its role in neural crest. By reducing the levels of Sox10, we could rescue a number of different crest-derived phenotypes indicating the importance of sox10 deregulation in the pathophysiology of CHARGE.
chd7 RNA is inherited from parents and ubiquitously expressed in early zebrafish embryos
CHD proteins play important roles in development, stem cell function, adult physiology and diseases (1,22). To determine the expression of these genes in zebrafish, we analysed existing RNA sequencing data available in the public domain for different developmental stages and tissues of zebrafish [SRP009426 (23), SRP008845 (24), ERP000016 (Zebrafish transcriptome sequencing project, PRJEB1986), PRJNA207719 (25) and SRP017135 (26)] . We found expression data for 10 members of the chd family. Chd1-like, chd2, chd3, chd6 and chd9 had very low expression at most stages of zebrafish development, while chd1, chd8, chd4a and chd4b showed high expression in some tissues and stages (Fig. 1B). The most abundant chd mRNA in zebrafish was chd7 (Fig. 1B).
Chd7 expression was maximum in the early stages of development starting from 2 to 4 cells (Fig. 1A). In zebrafish embryos, the zygotic transcription starts between 1K cell stage and dome stage (3–4 hours post fertilization (hpf)) (27), thus mRNA present in 2–4 cell and 1K cell stages are indicative of parental contribution. The only adult tissue with comparable expression to the early embryonic stages was the ovary (Fig. 1A). Chd7 could be amplified from both adult zebrafish ovary and testis cDNA (Fig. 1C), suggesting that the contribution of the chd7 transcript to the zygote could be from both parents.
To confirm observations made from RNA sequencing analysis and to understand the spatial distribution of the chd7 RNA, we performed RNA in situ hybridization at various embryonic stages (Fig. 1D–J). Reinforcing the RNA sequencing data, chd7 was highly expressed from 2-cell stage onwards (Fig. 1D). Ubiquitous expression was observed at 2-cell, 8-cell, sphere and 10-somite stage (Fig. 1D–G). CHD7 is known as a positive regulator of ribosomal RNA biogenesis in stem cells (6) and might be essential in the first few hours when the translation of maternal RNA is essential for the rapid progression of embryonic development. By 24hpf, the expression becomes more restricted to the head, specifically in the eye and the mid-and hindbrain (Fig. 1H and I), as has also been shown previously (12,28). At 4 days post fertilization (dpf), chd7 expression was visible in the pineal complex, midbrain-hindbrain boundary, eye, jaw region and the gut tube (Fig. 1J) which supports previous studies showing the important roles for Chd7 in neurogenesis in the embryo as well as the adult (29,30).
Knockdown of CHD7 in Zebrafish
To address the role of Chd7 in zebrafish development, we designed a splice blocking antisense morpholino oligonucleotide (MO) against the exon8-intron8 boundary (MO1) (13) that is predicted to result in a 1447 aa Chd7 protein product (
Injection of 2.4ng chd7 MO1 caused developmental defects without causing lethality and all following experiments in this study were performed at this concentration of the MO1, unless otherwise mentioned. From hereon, we refer to chd7 MO1 as chd7 MO (and where MO2 has been used, it is explicitly stated). All the control embryos were injected with equal amount of a control MO with random scrambled sequence from Gene Tools®.
Injection of chd7 MO caused multiple morphological defects including small head and eyes, missing or reduced jaw and pericardial edema (Fig. 7B). In order to determine if the phenotypes are due to downregulation of Chd7, we co-injected 1.5ng of full-length human CHD7 RNA (11) with 1.2ng of chd7 MO. The expression of the human CHD7 RNA was determined by PCR of 24hpf injected embryos (13). Four-day old Tg(sox10:eGFP) larvae had GFP expression in the craniofacial cartilage stacks (
Chd7 Regulates Neural Crest Gene Expression and Migration
Many NC-derived cell types such as craniofacial cartilage and cranial motor and sensory neurons are affected in the CHARGE syndrome and Chd7 has been implicated in the development of NC derived cell types (11,18,21). However, no study has examined the role of Chd7 in the NC cells in detail. We set out to systematically dissect the effect of chd7 knockdown on NC in the zebrafish embryos, as a model for vertebrate development. We began by analysing 3-somite stage (ss) embryos; around the stage when NC cells are specified. Expression of one of the earliest NC specification markers snai2, is reduced in the chd7 morphants (Fig. 2A and B) while levels of tfap2a and foxd3, both early markers of NC cells, appear largely unperturbed in the morphants compared to the controls (Fig. 2C and 2F). sox9a, an SRY box gene that is crucial for NC specification and migration, is severely downregulated (Fig. 2G and H). Thus, the neural crest appears to be specified but not patterned correctly in the chd7 morphants, perhaps eventually influencing the downstream steps in NC development.
Since NC cells delaminate from the neural tube, we analysed the expression of sox2, a neural tube marker. Expression of sox2, is significantly downregulated in chd7 MO injected embryos (Fig. 2I and J). Human fetuses with CHD7 mutation have been known to have neural tube defects (31). It is possible that some of the effects on NC could be secondary to the defect in neural tube patterning.
We followed the expression of foxd3 into the 18-somite stage embryos. foxd3 is expressed in pre-migratory NC cells and in the somites in the 18-somite stage embryos (Fig. 2K and L) and there is an upregulation of both domains of expression in the chd7 morphants. At 24hpf, the NC-specific expression of foxd3 remains only in the caudal-most region and this appeared unperturbed in the chd7 morphants (Fig. 2M and P). Foxd3 is also expressed robustly in the somites and this expression was significantly upregulated in the morphants (Fig. 2M and P). Foxd3 has been implicated in regulation of myf5 (32), the myogenesis factor expressed in the somites as they differentiate to myoblasts. The foxd3 deregulation may suggest possible skeletal muscle-related complications in CHARGE. We performed quantitative RT-PCR of 24hpf embryos and found an average 3.13 fold upregulation of foxd3 mRNA (Fig. 2Q). A recent study compared the RNA profiles of wildtype and whirlgig (chd7wg/wg) mouse embryos during NC migration (9.5dpc) using microarray and they found a 2.5 fold upregulation of foxd3 in the chd7wg/wg mouse embryos (20). Our whole organism analysis suggests that the overall upregulation evident in the chd7 mutant/morphant background may be primarily attributed to the somitic expression.
By 24hpf NC cells are rapidly migrating in a rostral-to-caudal chronology. Thus, the cranial and vagal cells have completed their migration while the trunk NC may be seen in different stages of the migration as visualized by crestin, a NC marker (33). Trunk migration was severely impaired in the chd7 morphants and cells appear to be still residing in the dorsal position when compared to the control embryos (Fig. 2R and S). This data supports previous studies on human NC like cells (hNCLC) and Xenopus embryos, which demonstrated that NC cell migration is impaired upon CHD7 knockdown (11).
Chd7 Deficiency Causes Deregulation of sox10 Expression
A critical regulator of neural crest cell specification, migration and fate choice, is Sox10. We analysed the expression of sox10 in chd7 morphants from 3ss, when the NC is first evident, to 4dpf, when many NC derived cells are undergoing differentiation, in zebrafish embryo. We mapped sox10 expression in the 3ss (11hpf), 6ss (12hpf), 10ss (14hpf), 14ss (16hpf), 18ss (18hpf) and 22ss (20hpf) embryos. The embryos were staged by counting somites (till 22ss) to confirm the developmental stage. The 3ss control embryos expressed sox10 in the neural plate border cells while in chd7 morphant embryos this expression was not very apparent (Fig. 3A and B). By 6ss the control NC cells were migrating medially (also described in (34), however the NC in the chd7 morphants were still farther apart and there was no evidence of medial migration (Fig. 3C and D). As the vagal and cranial NC condense at 10ss, the medial migration was no longer evident in the control, but appeared to continue in the chd7 morphants (Fig. 3E and F). In lateral views of the embryos the NC, marked by sox10 expression, stretches caudally to more than half the length of the embryo, while in the chd7 morphants the crest was seen only in the anterior trunk region (Fig. 3G and H). In the 14ss (Fig. 3I and L), sox10 marks the ventrally streaming NC cells and these streams were absent or severely reduced in the chd7 morphants; this effect was also evident in 20hpf embryos (
At 24hpf, as the NC completes their migration, sox10 is downregulated in these cells. In the control embryos, sox10 expression decreased rapidly and was found only in the occasional dispersed cells and in the yet-to-migrate caudal crest (Fig. 3T). Schulz et al. as described in the previous section, in their microarray analysis on wildtype and whirlgig (chd7wg/wg) 9.5dpc mouse embryos also reported a 2.2 fold upregulation of sox10 RNA in the chd7wg/wg embryos (20).
To determine if the increase in sox10 RNA impacts the protein expression, we performed immunostaining with an anti-Sox10 antibody in 24hpf embryos. We found that the chd7 morphant embryos had more cells expressing Sox10 protein compared to the controls (Fig. 3M and N). We used the Tg(–4.9sox10:egfp)ba2 line to visualize the migrating NC in live embryos at 24hpf and observed a similar increase in the number of GFP positive cells in the morphants when compared to controls (Fig. 3O and P). We quantified this effect using flow cytometry and found a 1.7 to 3.5 fold increase in the number of GFP positive cells in the chd7 morphants (Fig. 3S).
The increase in sox10 expressing cells at 24hpf may be attributed to the delay in migration and the concomitant delay in downregulation of Sox10 in migrating NC cells. However, we observed that the elevation in Sox10 expression was not confined to the 24hpf morphant embryos. We observed ectopic and elevated expression of sox10 in 28hpf sox10:eGFP morphant embryos (Fig. 3Q). The chd7 morphant larvae did not have a very strong fluorescence in the spinal cord, however there was ectopic fluorescence in the trunk region (Fig. 3R). Thus, we conclude that chd7 morphant animals have an overall deregulation of sox10 expression from the early NC stages to later.
chd7 Knockdown Affects Migration of Pigment Precursors and Inhibits Differentiation of Pigment Lineages
Sox10 is a crucial transcription factor for pigment development in vertebrates (35). Zebrafish has two different kinds of pigment cells melanophores and iridophores that are derived from same precursor pool of NC cells and regulated by Sox10 (36). Since sox10 is deregulated in the chd7 morphant fish, we analysed the status of these pigments in 5dpf zebrafish larvae. Iridophores produce the iridescent pigment in the eyes, top of the head, body and in discrete spots on the trunk in 5dpf control larvae (Fig. 4A). We found an absence of the iridophores (or rather the iridescent pigment) in the whole body of chd7 morphant larvae (Fig. 4B) (previously reported in the eye (12)). Melanophores that contain the black pigment melanin are evident by 30hpf in zebrafish embryos (37) and we observed that 2dpf and 5dpf chd7 morphant animals had decreased pigmentation. At 5dpf, the melanophore shape also appeared to be abnormal in the morphants (Fig. 4C and D). We quantified the melanin content per embryo in 2dpf and 4dpf animals and found a significant decrease in the chd7 morphants compared to controls (Fig. 4G and H).
The differentiation of melanophores is initiated by the transcription factor mitf-a which is expressed in the precursors of melanophores as they migrate laterally from the crest (38). The mitf-a positive NC cells are marked by GFP in the Tg(mitf:GFP) transgenic line. At 24hpf, the chd7 morphant embryos have elevated levels of GFP (Fig. 4I and J) compared to controls. RNA in situ hybridization of mitf-a showed delayed migration and dorsal accumulation similar to that seen in sox10 (Fig. 4K and L). Melanophores express two enzymes that are important for melanin formation, tyrosinase (tyr) and dopachrome tautomerase (dct) (39). The expression of these genes is evident in distinct scattered cells throughout the embryo at 24hpf (Fig. 4O and S); this expression is severely downregulated in the chd7 morphants (Fig. 4P and T).
It is important to note that CHARGE patients have not been reported to have any pigmentation defects. Melanophore development is known to be very plastic and alternate routes of melanogenesis may become active even when the primary wave of melanogenesis is compromised (40). It is possible that the requirement for Chd7 by melanophores is species specific and is not necessary in the human melanocytes.
Knockdown of sox10 in Zebrafish Embryos
Since an upregulation of sox10 was noticed in the neural crest of the chd7 morphant (Fig. 3M and P) embryos and sox10 is a known regulator of mitf-a (41), we sought to downregulate Sox10 levels using a MO. We used two sox10 MOs in this study: MO1, a translation blocking MO (42) and MO2, a splice blocking MO (designed for this study, details in
Sox10 Downregulation Does Not Affect The Pigment Defect in chd7 Morphant Zebrafish
We examined the embryos after injection of the Sox10 MO for the pigmented melanocytes and pigment regulatory genes. As would be expected, Sox10 MO injected embryos exhibited a reduced pigment phenotype [as previously shown in (43);Fig. 4M,Q,U). sox10 MO2 elicited a similar but weaker effect (Fig. 4N). The double morphants did not show any recovery of the expression of the melanophore differentiation factors tyr or dct (Fig. 4R and V, 44). Thus, although Sox10 along with Mitfa is known to activate the differentiation genes dct and tyr (45) chd7 may also be essential for the induction of differentiation in the pigment lineage, independent of the deregulation of Sox10.
Loss of CHD7 Causes Reduction of Peripheral Neuronal Lineages
Neuronal defects are a hallmark of CHARGE syndrome. CHARGE patients have multiple defects such as defects in facial muscles, xvestibular functions, hearing, olfaction and swallowing difficulties (9). The peripheral nervous system consists of neurons and glia derived from the neural crest. We probed the different peripheral neuronal lineages in our model for CHARGE syndrome. Cranial neurons include both sensory and motor neurons that control functions in the head and face. We detected differentiated cranial neurons in 72hpf embryos using islet-2a (isl2a)(46) as a marker and discovered that knockdown of chd7 leads to a dramatic reduction of these neurons (Fig. 5A and B). This is also substantiated by a previous study in zebrafish (12). Since the cranial neurons are reduced in chd7 morphants, we probed the status of the cranial neuron precursors in the epibranchial ganglia marked by neuroD (47) at 36 hpf; this expression was severely reduced in the chd7 morphant embryos (Fig. 5C and D) suggesting that the defects begin in the precursor populations. The chd7 MO2 had similar effects on neuroD and isl2a expression (
Enteric nervous system innervates the gut. The enteric neurons are originally derived from the migratory vagal NC marked by crestin at 36hpf (48). Knockdown of chd7 reduced the expression of crestin in vagal NC cells (Fig. 5E and F) suggesting that the specification of these cells is compromised. This chd7 MO2 had similar effects on crestin expression (Fig. 5G and H). CHARGE patients have not been reported to have enteric neuronal defects, although there have been reports of chronic constipation in patients (49), which is a symptom of colonic agangliogenesis as seen in Hirschprung disease (50).
At 4dpf, the olfactory microvillous neurons are marked by sox10:eGFP expression (51). We visualized the effect of chd7 knockdown on these cells and found that the chd7 morphant larvae have a reduced number and structural disorganization of the olfactory microvillous neurons (Fig. 5I and J). Chd7 mutant mice also exhibit defects in the number and organization of olfactory sensory neurons (52). The lateral line hair cells are important mechanosensory neurons in the zebrafish. We used the fluorescent dye DASPEI to visualize the lateral line neuromasts and found that chd7 morphants had a significant reduction of these neurons (Fig. 5K and L). Thus, multiple sensory neuronal types are depleted in the chd7 morphant embryos.
To test whether the depletion of peripheral neurons is a result of of Sox10 deregulation, we co-injected embryos with chd7 MO and sox10 MO. We did not observe any rescue of the neurod, crestin or isl2a expression in the embryos co-injected with both chd7 MO and sox10 MO (
Loss of chd7 Expands Glial Precursors But Inhibits Myelination of Schwann Cells in A Sox10-Dependent Manner
Neural crest cells give rise to a number of different kinds of glia associated with the peripheral neurons viz. satellite glia and myelinated and nonmyelinated Schwann cells (17). At 24hpf, foxd3 is expressed in cranial glia associated with the preotic and postotic ganglia (53) and this expression was completely abolished in the chd7 morphant embryos (Fig. 6A and B). However, at 56hpf foxd3 expression in the trunk satellite glia associated with the dorsal root ganglia (DRG) (54) showed an increase in chd7 morphants compared to the controls (Fig. 6C and D). We also observed an increase in the trunk glia of 4dpf chd7 morphant larvae (Fig. 6G and H) by foxd3 RNA in situ hybridization; this effect was dramatically illustrated in the Tg(foxd3:GFP) transgenic line (Fig. 6E and F) that marks the lateral line glia in 4dpf larvae with GFP (55).
To determine if this increase in foxd3 positive glial cells also results in more myelinated Schwann cells we performed RNA in situ hybridization for myelin basic protein (mbp) mRNA. mbp, a specific marker of myelinated Schwann cells is prominently expressed in the lateral line and in the periocular region at 4dpf (56). Deficiency of chd7 causes an overall decrease in the expression of mbp, around the eye and on the lateral line (Fig. 6K and L). This was further confirmed by quantitative RT-PCR (0.4 fold,
We have thus uncovered a novel aspect of CHARGE syndrome in our zebrafish model. The severe reduction in the myelinated Schwann cells of the peripheral nervous system was accompanied by an expansion of satellite and other non-myelinated glia marked by foxd3 in the morphant. We speculate that glial defects may contribute to many of the peripheral nervous system deficits in CHARGE patients. Micucci and colleagues have recently reported an increase in glia in the subventricular zone in the brain in Chd7 mutant mice resulting from a neuron-to-glia fate switch (57). Very recent studies by He et al. have shown that Chd7 and Sox10 interact to specify central nervous system glial cells in mouse (58). In our study, chd7 deficiency caused an expansion of glial cells in the trunk but was detrimental to myelination of these cells.
Sox10, overexpressed in 24hpf and older chd7 morphant embryos is a known regulator of foxd3 (59), the marker of glial precursors (55). Sox10 is also a crucial regulator of glial fate choice and myelination (60). We asked whether the glial cell phenotype in chd7 morphants is mediated by Sox10. We knocked down sox10 in the chd7 morphants and observed the effects on glial fate. The expansion of foxd3 expression in the trunk of 4dpf chd7 morphant larvae was reversed by the combined reduction of chd7 and sox10 (Fig. 6G and J,Fig. 6K and N and
Sox10 is a known inducer of the myelination program in the glia; however a previous study in chick embryos has shown that the overexpression of Sox10 results in inhibition of differentiation of NC derivatives, including myelinated Schwann cells (61). We therefore suggest that the overexpression of Sox10 in the chd7 morphants is the cause of inhibition of myelination and that when this inhibition was relieved in double morphants, the foxd3 positive glial cells differentiated into myelinated Schwann cells efficiently.
chd7 Knockdown Leads to Defects in Craniofacial Cartilage Specification and Differentiation in a Sox10-Dependent Manner
Craniofacial skeletal abnormalities are a predominant feature of CHARGE syndrome (9). Our chd7 morphant larvae also exhibited a smaller head, smaller eyes and an absence of the lower jaw at 4dpf (Fig. 7A and B). Alcian blue staining of cartilage (Fig. 7F) in the 4dpf morphant larvae confirmed the severe reduction of craniofacial cartilage structures such as the Meckel’s, Ceratohyal, and Ceratobranchial cartilages. Craniofacial defects have been reported in zebrafish (13) and mouse models (21). chd7 MO2 caused similar but less severe jaw phenotypes (
The precursors of ectomesenchyme express transcription factors such as dlx2a, sox9a and sox9b that commit them to cartilage fate and promote the activation of cartilage-specific genes (62). We analysed the expression of dlx2a in 24hpf embryos, expressed in the migratory neural crest cells that form pharyngeal arches 1-4 (62,63), and found that the chd7 morphants showed a reduction of dlx2a in the arches (Fig. 7I and J). Sox9a and sox9b, two paralogs of SOX9 in zebrafish, regulate chondrogenesis in the head (62). The control embryos have robust sox9a and sox9b expression in the pharyngeal arches 1 and 2 and in the perichondrium (62) respectively. We found a severe reduction in the expression of both sox9a and sox9b genes in the chd7 deficient embryos (Fig. 7M,N,Q,R). It has been previously shown that the CHD7 protein can bind and activate the human SOX9 enhancer (11) and our present results suggest that sox9 might be an evolutionarily conserved target of CHD7 from zebrafish to humans.
We probed the precursor population of NC cells that give rise to the ectomesenchymal derivatives in the craniofacial region of the embryo. Migratory cranial NC cells are marked by crestin at 24hpf (64), and we observed an almost complete absence of crestin staining in the chd7 morphant embryos (Fig. 7E and H and Fig. 7G). Upon co-injection of chd7 MO with sox10 MO1 or sox10 MO2 a partial recovery of the cartilage structures were evident in the larvae (Fig. 7L,P,T and
CHARGE syndrome is a debilitating and often fatal congenital disorder. 65–70% of patients have a mutation in the chromatin remodeler CHD7 (9). Many of the tissues affected in CHARGE syndrome such as heart, auditory neurons, facial neurons and craniofacial bones (9) have their origins in the neural crest lineage. Earlier studies have also implicated a role for Chd7 in the neural crest cells (11,18,21). In this study, we aimed to systematically understand the neural crest related defects in a chd7 compromised state. Since NC is a unique vertebrate tissue type (65), we chose zebrafish the highly amenable vertebrate model organism to perform this study.
Using morpholino antisense oligonucleotides to knockdown zebrafish chd7 we generated a model for CHARGE syndrome. Using this model we identified certain novel aspects of Chd7 function and perhaps CHARGE manifestation. We discovered defects in pigmentation, enteric neurons and myelination of Schwann cells, all previously undocumented in CHARGE patients or CHARGE models. We suggest that some of these phenotypes such as abnormalities in enteric ganglia or myelination defects may have relevance in patients and may have been overlooked just because they are not obvious during gross examination in the clinic.
The morphant zebrafish showed a selective delay in NC development with striking abnormalities in the trunk NC migration. We found that NC cells marked by sox10, migrate efficiently into the head NC regions but crestin positive cells were absent in the head. We interpret this as a change of fate in the cranial NC from crestin expressing cells to Sox10 expressing cells. We also discovered that the craniofacial cartilage, that form from the cranial NC are compromised in the chd7 morphants. Sox10 downregulation was able to rescue these craniofacial defects indicating a role for Sox10 in the fate determination of the cranial NC.
We also documented a persistence or upregulation of Sox10 in the NC cells in the chd7 morphant embryos. This could be a direct consequence of the delay or defect in migration of the NC in the chd7 morphants, since Sox10 downregulation in normal NC cells is linked to their ventral migration. However, our study indicates that timely downregulation of Sox10 is an important event in the specification and differentiation of NC cells into their derivatives and that a delay in the downregulation of Sox10 can have consequences on fate choice and differentiation of NC cells. This hypothesis is borne out by the apparent rescue of craniofacial cartilage and myelination defects in chd7 morphants by downregulation of Sox10 protein.
Thus, we conclude that Chd7 is an important player in the specification, migration, fate-choice and differentiation of NC. We also conclude that Sox10 is an important mediator of function of Chd7 in NC. We show that by external regulation of sox10 expression, we can rescue some aspects of the neural crest derived phenotypes in the CHARGE model suggesting possible avenues of intervention in the future.
Materials and Methods
Zebrafish lines and maintenance
Zebrafish (Danio rerio) were bred, raised and maintained at 28.5 °C under standard conditions as described (66). Embryo staging was done by using both timing (hours post fertilization, hpf) and morphological features as previously described (67). For fluorescence imaging and in situ hybridization analysis of embryos older than 24hpf, embryos were raised in 1-phenyl-2-thiourea (PTU) (0.003% in Embryo medium), to inhibit pigment formation (68). Zebrafish handling was in strict accordance with good animal practice as defined and all experiments were performed using protocols approved by the Institutional Animal Ethics Committee (IAEC) of the CSIR-Institute of Genomics and Integrative Biology, India. The zebrafish lines used in this study are Tuebingen (TU), used as the wild-type strain, Tg(–4.9Sox10:egfp)ba2(69), Tg(foxd3:GFP)(55), Tg(mitfa:GFP)w47(70), and Tg(NBT-dsRed) (71).
RNA sequencing analysis
The publicly available data for RNA-seq was obtained from NCBI SRA. Developmental timepoints RNA seq data was obtained from SRP009426 (23), SRP008845 (24), ERP000016 [Zebrafish transcriptome sequencing project, PRJEB1986] PRJNA207719 (25) and SRP017135 (26), corresponds to adult tissues. The SRA data was processed into fastq format, followed by preliminary quality check (QC) by FastQC and adapter trimming using Trimmomatic (72) The data comprised of 10 developmental stages viz., 2-4 cell, 1Kcell, dome, shield, bud, 1dpf, 2dpf, 3dpf, 5dpf, 14dpf and 9 adult tissues viz. blood, brain, heart, liver, muscle, eye, spleen, ovary and testis .
Once the preliminary QC was completed the reads for respective stages and tissues were aligned on the reference genome of zebrafish (zv10) using TopHat (73). The alignment was followed by analysing the differential expression of the annotated transcripts (ens84). Differential expression of the transcripts across the 10 stages and 9 tissues of zebrafish was obtained using Cuffdiff module (73) for which Binary alignment map (BAM) files served as input obtained from TopHat. The differential expression was in terms of Fragment per Kilo base Exon per Million Reads (FPKM) scores which is used to give the relative abundance of the transcripts across the data points. The FPKM scores were further fetched for the chd transcripts and the heatmap for the same was plotted so as to visualize the expression of them across the stages and tissues of zebrafish.
Morpholino and RNA injections
The endogenous expression of genes was perturbed by injection of morpholinos. All morpholino antisense oligonucleotides (MOs) were designed and synthesized by Gene-Tools, USA. For all experiments, the minimum effective concentration was used as determined by titration experiments. Two different chd7 splice blocking MOs have been used in this study. The first chd7 splice blocking MO (MO1) was designed against the exon8-intron8 boundary of chd7 transcript (ENSDART00000135230) (13).
To determine the specificity of the phenotypes observed full length human CHD7 RNA was co-injected with chd7 MO. pCDNA plasmid containing full length human CHD7 was linearized with AvrII. The linear fragment was gel-eluted and transcribed using T7-ultra mRNA synthesis kit (Ambion-Thermo Fisher). The full length mRNA was purified using oligodT based mRNA purification kit (Invitrogen), run on an agarose gel for confirmation, quantitated and frozen in small aliquotes. As truncated or partially degraded CHD7 mRNA can function as dominant negative, thawed aliquotes of mRNA were used only once. Each embryo was injected with 1.5ng of human CHD7 RNA and 1.2ng of chd7 MO.
sox10 knockdown was performed using a previously reported translation blocking morpholino (42) and a splice blocking morpholino designed across exon2 and intron2 hereafter termed as sox10 MO2. Concentrations of the various MOs used in injections are: 2.4ng for both chd7 MO1 and MO2 and sox10 MO1, and 12ng for sox10 MO2. A standard control scrambled sequence MO obtained from Gene Tools® was used in all experiments as a control. All injections were done at one-cell staged embryos. Morpholinos were suspended in nuclease free water to form a stock solution of 1mM. Working solutions were made by dissolving in nuclease free water and used within 2–3 days . The list of MO sequences is given in
Polymerase chain reaction
Total RNA was extracted by homogenizing 15–20 pooled embryos using Trizol reagent (Invitrogen). RNA was further treated with DNase (Ambion) to remove genomic DNA contamination. Wildtype (zebrafish and human) as well as aberrant chd7 transcripts were detected using RT-PCRs. Quantitative real time PCR was carried out as described (74) on the Roche LightCycler® Real-Time PCR System. For quantification, the relative standard curve method was used (as described by the manufacturer) to generate raw values representing arbitrary units of RNA transcript. Each experiment was performed on three independent occasions in biological triplicates of pooled embryos (in every experiment). Statistical analyses on normalized data were performed using 2−ΔΔCT algorithm (known as the delta-delta-Ct or ddCt algorithm) (74,75). All genes were normalized against RPL13α unless mentioned otherwise. All primers used in this study are listed in
Embryos (staged between 1 and 6dpf) were monitored and scored for phenotypes using Zeiss (Stemi 2000C®) bright field microscope (with AxiocamICc1). For visualizing iridophores, reflected light with the black background was used. For fluorescent imaging, Zeiss AxioScope A1 microscope (with Axiocam HRc®) was used. Images were captured using Zeiss proprietary software and processed and analysed in Adobe® Photoshop®.
Wholemount RNA in situ hybridization and immunohistochemistry
RNA in situ hybridization was performed as described (76). Probes of the following genes were used: sox2, snai1b, Sox10, foxd3, tfap2a, sox9a, dlx2a, crestin, mitfa and dct (77); sox9b (62), neurod (78), isl2 (79), tyr (80), mbp (56) and krox20 (81).
To generate a probe for chd7 a 525bp fragment of the chd7 gene was PCR amplified from 24hpf zebrafish cDNA (Primer sequences in the
Immunostaining for Sox10 was performed broadly as described (82). The embryos were fixed in 4% paraformaldehyde and were dehydrated through a graded methanol series. Further, the embryos were incubated with anti-Sox10 Ab (1:100 dilution) (GTX128374, Gene Tex®) for 24hrs and goat anti-rabbit IgG secondary antibody conjugated with Alexa fluor 488 was used (1:500 dilution) (A-11008, Thermo Fisher Scientific®).
Fluorescence Assorted Cell Sorting (FACS) Analysis
At 24hpf, live embryos were dechorionated with Pronase and de-yolked in ice-cold Ringers solution using a micropipette tip. Single cell suspension of the de-yolked embryos was prepared using TrypLeTM Express (Thermo Fisher Scientific®). The cells were collected, washed with ice-cold Ringers solution and passed through a 70µm filter to avoid cell clumping. The cells were then collected and re-suspended in DPBS with 2.5% FBS. The cells were analysed for GFP + ve cells in BD FACS Aria III®. Uninjected sox10:eGFP positive embryos were used as controls. FACS data were analysed using BD FACSDiva™ software.
Alcian blue staining
Alcian blue stain was used to visualize the structure of the craniofacial skeleton of zebrafish as described previously (83).
DASPEI (2-(4-(dimethylamino)styryl)-N-ethylpyridinium iodide) stains hair cells within lateral line neuromasts of live zebrafish embryos. The staining procedure was carried out as described (84).
Melanin content was determined as described previously in (85). Embryo lysates were prepared for both control embryos and chd7 morphants, at 2dpf and 4dpf. The sample pellet was dissolved in 1 N NaOH at 100 °C for 50 min. Optical density was measured at 490 nm, and the sample results were compared with a standard curve of known concentrations (0–300 µg/ml) of melanin (Sigma). The experiment was performed on three or more independent occasions; with sample size, i.e. number of embryos ≥50, and same for both control and chd7 MO, each time. For relative comparison, melanin content per embryo (µg) was plotted against sample.
We thank Alessandro Mongera, Prateek Mahalwar and Soundhar Ramsay for generous help and for reagents. We thank Archana Vats and D. Ayyappa Raja for help with FACS analysis and Manish Kumar for confocal microscopy. We thank Vivek Natarajan and Mageshi Kamaraj for comments and discussion.
Conflict of Interest Statement. None declared.
This work was supported by Council of Scientific and Industrial Research (CSIR), New Delhi [BSC0118 to C. S., research fellowships to Z.A. and S. K. and BSC0403 for the microscopy facility], National Institutes of Health-National Institute of Dental and Craniofacial Research [R01DE024584 to R. B.] and CHARGE Syndrome Foundation Grant [R.B.].