The neural crest lineage is a multi-potent, migratory population that is unique to vertebrate embryos and gives rise to a diversity of cell types including melanocytes, craniofacial cartilage and bone, smooth muscle, peripheral and enteric neurons and glia [1]. The specification, migration and differentiation of neural crest cells are tightly coordinated during development, with defects in these processes leading to a number of congenital diseases [2]. There is good evidence that during vertebrate embryogenesis a highly conserved neural crest gene regulatory network (GRN) orchestrates transcriptional events that are critical for various steps of neural crest development [3]. In addition to transcriptional regulation, mounting evidence supports a critical role for epigenetic regulation in neural crest development, most notably by controlling the timing of gene expression during this process [4].

Mediator is an evolutionarily conserved, multi-protein complex that is a key regulator of RNA polymerase II-mediated transcription [5]. In metazoan cells, multiple pathways that are responsible for homeostasis, cell growth and differentiation converge on Mediator through transcriptional activators and repressors that target one or more of the 30+ subunits of this complex [6]. Besides RNA polymerase II, Mediator interacts with and coordinates the action of numerous additional transcriptional regulators, including those acting at the level of chromatin remodeling [7, 8]. However, while in vitro data indicates the interaction between Mediator and chromatin remodelers, there is little in vivo data to support this hypothesis. Med14 is a subunit of Mediator that is essential for incorporation of the Tail module into Mediator [6, 9]. We have recently shown that Med14 plays an essential role in vertebrate embryogenesis and stem cell maintenance [10].

Working as a multi-subunit cellular machine that consumes ATP to modify DNA-histone contacts and modulate chromatin compaction, the BAF (BRG1/BRM-associated factors) complex plays a key role in many developmental processes by modulating gene expression. This further occurs via interaction of the BAF complex with transcription factors and other epigenetic readers at promoters and enhancers [11]. The BAF complex includes one of two core ATPases, Brm or Brg1, as well as a number of other subunits. Brm is dispensible for mouse development, whereas Brg1 (Smarca4) is essential for broad aspects of embryogenesis [12, 13]. Differential inclusion of other subunit variants can give novel functions to the BAF complex in processes including neuronal development and cardiogenesis [11, 14, 15]. Reports have demonstrated a role for the Mediator complex in recruitment of the BAF complex to promoters or enhancers of target genes [16, 17], however no in vivo evidence for this genetic interaction and its importance exist to date. Defects in neural crest cell-derived tissues has been noted in brg1 mutants [18] and recent work has shown that the BAF complex co-operates with CHD7 to orchestrate the expression of genes that regulate the migration of neural crest cells [19]. However, the mechanism underlying these roles in neural crest development has to date not been well characterized.

In the present study, we sought to determine the roles of med14 and brg1 during neural crest cells differentiation, and examine any possible genetic interactions. We found that med14 mutant zebrafish embryos demonstrated multiple neural crest cell-related defects. Further analysis indicated that specification and early migration of neural crest cells occurred normally in med14 mutants, with neural crest cells of the jaw subsequently failing to undergo terminal differentiation at their target sites. We further found that mutation of brg1 also resulted in similar abnormalities. Analysis of med14 and brg1 double mutant embryos revealed strong genetic interactions between the Mediator and BAF complexes. Based on transplantation analysis, we found that both med14 and brg1 function in neural crest cells differentiation in a cell-autonomous fashion. Taken together, our results indicate that the BAF and Mediator complexes play essential and overlapping roles in the terminal steps of neural crest differentiation.

In unrelated studies, we noticed that zebrafish log (a null allele for med14) [10] and young (a null allele for brg1) [14] mutants shared a common array of deficiencies in heart, eye, pectoral fin and pigment cell development (Fig. 1a–d and data not shown). This similarity suggested overlapping or common functions for med14 and brg1. To test this possibility, double mutants were generated. The med14; brg1 double mutant embryos displayed a much more severe phenotype compare to single mutants, including a curved body axis, smaller eyes, severe heart edema and loss of pigment (Fig. 1a). To further investigate the role of med14 and brg1 in development, as well as possible functional interactions between the Mediator and BAF complexes, neural crest cell-derived tissues were analyzed in various mutant backgrounds. In med14 and brg1 single mutants, the melanin in melanocytes showed a less even and spiky distribution compared to controls; whereas in med14; brg1 double mutants melanin distribution took on a small, rounded appearance (Fig. 1b–e). Quantification of melanocyte number on the dorsal surface of the trunk revealed no significant differences between controls and mutants (Fig. 1q, P = 0.94).

The development of the neurocranium and viscerocranium were next analyzed through Alcian blue staining at 96 h post-fertilization (hpf). In brg1 mutants, the neurocranium was dismorphic, and the size of Meckel’s cartilage, ceratohyal and palatoquadrate on the viscerocranium were greatly reduced, with the last two branchial arches absent (Fig. 1j–k and n–o). In med14 mutants, trabeculae formed, while most of the ethmoid plate and lateral parts of parachordal plate were not evident (Fig. 1l). Aside from relics of palatoquadrate, viscerocranium structures were absent in med14 mutants (Fig. 1p). In med14; brg1 double mutants, only posterior portions of trabeculae and parachordal plate were apparent, whereas viscerocranium was completely absent (Fig. 1m). It has been reported some posterior elements (part of trabeculae and parachodal plate) are derived from mesoderm [20]. As these elements remained in med14; brg1 double mutants, this suggested med14 and brg1 are only required for neural crest-derived cartilage.

To further explore genetic interactions between med14 and brg1, we examined cartilage defects in brg1−/−; med14+/− and med14−/−; brg1+/− embryos. In either case, further loss of one allele of med14 or brg1 in either brg1 or med14 null mutants resulted in more severe defects in facial cartilage formation as compared to single mutants (Fig. 1r–t). We also observed that med14 and brg1 mutants shared a common array of deficiencies in heart, eye and otic vesicle development. At 48 hpf, eye, otic vesicle and heart defects in med14 or brg1 mutant embryos were similarly exacerbated in med14/brg1 double mutants (data not shown). Taken together, these data showed that zebrafish med14 or brg1 mutants displayed defects in neural crest-derived cells and tissues such like craniofacial cartilage and melanocyte. Furthermore, med14; brg1 double mutants, or single mutants where one allele of the other gene was lost, displayed a more severe phenotype than single mutants, suggesting overlapping functions of the Mediator and BAF complexes.

As the phenotypes observed in med14 and brg1 mutants involved multiple tissues and developmental stages, we subsequently focused our analysis on jaw cartilage development to uncover the mechanisms of med14 and brg1 function in neural crest development. Neural crest specification was first analyzed via expression of foxd3, sox9b and snail1b, which are expressed in pre-migratory neural crest at the dorsal side of neural tube at 14 hpf [21]. No overt difference in expression was apparent between control and mutant embryos (Fig. 2a–l), indicating that neural crest cells were properly specified in med14, brg1 and med14; brg1 mutants. Subsequent migration of neural crest cells was studied via use of a neural crest-specific sox10:EGFP transgenic line [22]. Neural crest cells in all three mutant backgrounds dispersed and migrated around the eye and optic stalk to reach the oral ectoderm and formed primodia of branchial arches by 24 hpf; with the behavior of neural crest cells being similar between mutant and control embryos (Fig. 3a and d). In situ hybridization also showed that markers of migratory facial neural crest, dlx2a and twist1a, were expressed normally in mutants at 18 hpf (Fig. 3b–c).

To determine if postmigratory cranial neural crest was properly maintained in these mutants, we examined expression of sox9a, dlx3b and hand2, which function in neural crest differentiation [23, 24], at 32 hpf. The expression of sox9a and dlx3b were both down-regulated in med14 and brg1 single mutants, and almost abolished in med14; brg1 double mutants (Fig. 3e–f). At 30 hpf, hand2 clearly marks rings of ventral neural crest in the pharyngeal arches (Fig. 3g). In med14 and brg1 mutant embryos, the expression of hand2 was reduced (Fig. 3g). In med14; brg1 double mutants, the expression was largely abolished except for in a small patch posterior to the eyes, which was maintained. Notably, hand2 expression in the heart tube remained in double mutants, indicating that the regulation of hand2 by brg1 and med14 is neural crest-specific (Fig. 3g). During jaw development, neural crest cells undergo extensive proliferation, so it is possible the defects observed in mutants were due to impaired cell proliferation or enhanced cell death. Cell proliferation and cell death were therefore analyzed at 36 hpf, however no obvious differences between control and mutants was observed (Fig. 4, P = 0.94 for cell proliferation and P = 0.95 for cell death). Taken together, this data suggests that in mutants neural crest cells are formed and migrate properly, but are subsequently unable to initiate a differentiation program towards becoming skeletogenic ectomesenchyme.

Neural crest differentiation depends both on intrinsic gene regulatory programs and signals from the surrounding environment [25]. As such, deficiencies either in neural crest cells themselves or other tissues could be responsible for the defects we observed in brg1 and med14 mutants and compound mutants. As signals from the endoderm and notochord are indispensable for the migration and differentiation of pharyngeal neural crest cells [26, 27], we first analyzed expression of foxa1 and shh, which are markers of these two tissues, at 30 hpf. Both genes showed apparently normal expression in mutants and compound mutants, suggesting that these tissues were not grossly abnormal (Fig. 5a–d and e–h). Since endodermal pouches play important role in patterning and differentiation of pharyngeal arches [28], we analyzed expression of nkx2.3 and tbx1, which are expressed in both the pre-pouch endoderm and surrounding mesoderm [29]. At 32 hpf, expression of nkx2.3 and tbx1 appeared normal in both wild type and mutant embryos (Fig. 5 i to l and m to p), suggesting that induction of the endodermal pouches is not dependent on med14 or brg1 function.

To directly examine the cellular autonomy of med14 and brg1 function, we employed a transplantation approach. Cells were first taken from the animal pole of 4 hpf wild type sox10:EGFP donor embryos and transferred to the equivalent location in wild type or mutant 4 hpf host embryos (Fig. 6a). Based on their localization in the host embryo, a proportion of these naïve cells will normally take on a cranial neural crest fate in these experiments, as indicated by donor cell GFP expression. We observed that in all cases wild type donor GFP-positive neural crest cells in either wild type (n = 33) or mutant (med14−/−: n = 26; brg1−/−: n = 24; med14; brg1 double mutant: n = 11) host embryos migrated to oral ectoderm and formed cartilage clusters, (Fig. 6f–i). Strikingly, as assayed by cartilage staining, we found that wild type donor cells could rescue anterior neurocranium cartilage phenotypes in med14 (43 %, n = 34) and med14; brg1 double mutant embryos (31 %, n = 14) (Fig. 6j–m). When sox10:EGFP activity was compared to cartilage staining results from the same embryos, we observed that anterior neurocranium cartilage entirely matched to GFP-positive clusters (Fig. 6g and k), strongly suggesting that wild type donor cells acted as the source of the cartilage.

To further examine the fate of med14 and brg1 mutant neural crest cells, transplantation experiments were next carried out in which wild type or mutant sox10:EGFP donor embryos were injected with rhodamine-dextran to allow observation of all donor cells (regardless of neural crest fate) (Fig. 7a). In wild type hosts, GFP-positive single or double brg1 and med14 mutant donor cells were observed that migrated to oral ectoderm at 24 hpf (Fig. 7 c, g, k and o). At 72 hpf, in 54 % of control (wild type donor, n = 44) transplantation experiments, donor cells were observed contributing to cartilage (Fig. 7d–f). In the case of brg1 mutant donor cells, 11 % of transplants showed a contribution to cartilage (n = 31, Fig. 7h–j), with GFP-positive cells displaying abnormal cell shape (compared to the elongated cuboidal shape observed in control transplants (Fig. 7d and h). Strikingly, in none of the transplants where med14 (n = 37) or med14; brg1 (n = 21) mutant donor cells were used was contribution to cartilage noted (based on GFP expression). However, imaging of the donor cell lineage tracer (rhodamine-dextran) showed that mutant cells still survived in the jaw region of wild type hosts (Fig. 7l–n and p–r). These experiments indicated that both med14 and brg1 function cell autonomously in neural crest cells, to govern proper skeletogenic ectomesenchyme differentiation. Further, the persistence of the lineage label in mutant donor cells, despite the absence of cartilage formation, argues for a model where med14 and brg1 are required subsequent to migration of neural crest cells, where they are necessary to the initiation of chondrocyte differentiation. In the absence of this activity, neural crest differentiation to cartilage, and jaw development, is severely perturbed.

At different developmental stages, Mediator and BAF complexes influence a wide variety of biological processes. The nature of their activity is context dependent, and in many cases involves differential use of variant complex member subunits to confer different activities [11]. It will be interesting to investigate, specifically in the scenario of neural crest development, how these two complexes function in a tissue and developmental stage specific manner and how the genetic interactions revealed in this study are executed. An intriguing possibility is that subunit(s) of these two complexes act cooperatively to regulate the expression of genes critical for terminal differentiation of craniofacial neural crest cells. As neural crest contributes to a great diversity of additional cell types, and defects were observed in multiple neural crest-derived tissues in med14 and brg1 mutants, more specific analysis of defects in these cell types, and determining which genes are directly regulated by the Mediator and BAF complexes, will be of great interest. Further, while we have described defects in the maintenance of neural crest fate or terminal differentiation in this study, the actual fate of these cells is not clear. Or results suggest these cells are not lost via apoptosis. It will be of interest to determine if they adopt an alternative cell fate, and if so what this fate(s) and what mechanisms underlie this fate conversion may be.

In a previous report [19], knock down of brg1 by morpholino injection in frog embryos led to defects in neural crest migration. Our data, however, suggests that the initial specification and early migration of cranial neural crest occurs normally even in severely affected brg1; med14 double mutants, which is then followed by defects in skeletogenic neural crest differentiation from 30 hpf onwards. Furthermore, by using transplantation approaches, we have shown that med14 and brg1 act cell autonomously (in neural crest) to regulate differentiation of these cells. These discoveries reveal an unexpected mechanism in facial cartilage development, which has been previously ascribed to defects in migration of neural crest cells to the site of differentiation, and raises a possible mechanism underlying the symptoms of neural crest-related diseases, including CHARGE syndrome. It should be noted that these previous studies were largely based on use of morpholino oligonucleotides, whereas our current work uses genetic mutants for analysis. The brg1 mutant used in this study [14] is a strong null allele, and phenocopies other described brg1 mutants [18]. We have recently described three novel med14 mutant alleles [10], with the s231 allele used in this study having an equivalent phenotype to the m628 allele that results in pronounced loss of transcript by nonsense-mediated decay. It is possible the maternally deposited (wild type) Brg1 and Med14 protein or transcript mask early effects on neural crest development in zebrafish mutants. However, as initial neural crest specification and migration appeared roughly normal even in double brg1; med14 mutants, this seems unlikely.

Development of the neural crest is of exceptional medical importance, being affected in a large percentage of congenital defects. The results presented in this study and other recent publications show that the Mediator complex may contribute to a range of neural crest cell-related craniofacial abnormalities. The patients with CHARGE syndrome, a neural crest cell-related congenital anomaly, present a cluster of malformations in craniofacial, peripheral nervous system, ear, eye and heart [30]. Recent work has shown that the BAF complex orchestrates the development of neural crest cells via co-operating with CHD7 [19], which is mutated in the majority of CHARGE syndrome cases [31]. Attempting to identify mutations affecting Mediator complex components in CHARGE syndrome and other neural crest-related patients will be helpful to further illustrate the molecular etiology of these syndromes. A more detailed analysis of mutant models of neural crest-derived craniofacial (and other) defects is also clearly warranted. Generation and analysis of chd7 mutant zebrafish, for example, will be of great interest.

In this study we used genetic tools to analyze the function and interaction of Mediator and BAF complexes on neural crest development. We found that mutation of med14 or brg1 in zebrafish embryos led a cluster of neural crest cell-related defects. Further analysis revealed additive genetic interactions between med14 and brg1, resulting in more severe defects. These defects apparently arise not from an absence of neural crest at the site of cartilage development, but a failure of these cells to properly execute terminal differentiation into jaw cartilage elements. The genetic interaction between brg1 and med14 revealed in this present study indicates that Mediator complex function might act as an additional potential modifier in human congenital jaw defects.

Zebrafish were housed and handled as per Canadian Council on Animal Care and Hospital for Sick Children Laboratory Animal Services guidelines.