The MITF paralog tfec is required in neural crest development for fate specification of the iridophore lineage from a multipotent pigment cell progenitor

Understanding how fate specification of distinct cell-types from multipotent progenitors occurs is a fundamental question in embryology. Neural crest stem cells (NCSCs) generate extraordinarily diverse derivatives, including multiple neural, skeletogenic and pigment cell fates. Key transcription factors and extracellular signals specifying NCSC lineages remain to be identified, and we have only a little idea of how and when they function together to control fate. Zebrafish have three neural crest-derived pigment cell types, black melanocytes, light-reflecting iridophores and yellow xanthophores, which offer a powerful model for studying the molecular and cellular mechanisms of fate segregation. Mitfa has been identified as the master regulator of melanocyte fate. Here, we show that an Mitf-related transcription factor, Tfec, functions as master regulator of the iridophore fate. Surprisingly, our phenotypic analysis of tfec mutants demonstrates that Tfec also functions in the initial specification of all three pigment cell-types, although the melanocyte and xanthophore lineages recover later. We show that Mitfa represses tfec expression, revealing a likely mechanism contributing to the decision between melanocyte and iridophore fate. Our data is consistent with the long-standing proposal of a tripotent progenitor restricted to pigment cell fates. Moreover, we investigate activation, maintenance and function of tfec in multipotent NCSCs, demonstrating for the first time its role in the gene regulatory network forming and maintaining early neural crest cells. In summary, we build on our previous work to characterise the gene regulatory network governing iridophore development, establishing Tfec as the master regulator driving iridophore specification from multipotent progenitors, while shedding light on possible cellular mechanisms of progressive fate restriction.


Introduction
Pigmentation is a conspicuous feature of animal diversity and has broad importance for behaviour and evolution (reviewed in [1]). Much is known about the development and cell biology of melanocytes but far less is understood about the genetic mechanisms underlying the diversity of pigment cell types in non-mammalian vertebrates. Zebrafish embryos display three neural crest (NC)-derived pigment cells: melanophores, melanin-producing cells homologous to the melanocytes of mammals, and often referred to simply as melanocytes; xanthophores, yellow-orange cells bearing pteridine and carotenoid pigments; and iridophores, shiny cells containing platelets composed of guanine (reviewed in [2], [3]).
Defining the fate specification mechanisms of these other cell-types is important for understanding their genetic control and their evolutionary origins.
The progressive fate restriction model proposes that neural crest cells (NCCs) become partially fate restricted as development progresses, giving rise to intermediate partiallyrestricted progenitors, each of which can generate a number, but not all of the NC derivatives [4]- [8]. Such a model has been strongly supported for the neural derivatives by a single cell transcriptional profiling study in mice, which surprisingly was unable to resolve pigment cell development [9]. Characterisation of the phenotypes of mutants affecting multiple NC derivatives has been widely used to infer the identities and potencies of these progenitors. In the pigment cell field, a progressive fate restriction model has been developed, with both a multipotent chromatoblast and a bipotent melano-iridoblast as identified partially-restricted intermediates ( [10] [11]- [14]). Studies of fate-specification mutants further permits elucidation of the gene regulatory networks (GRNs) governing diversification of these precursors ( [14], [15]).
To date, examination of zebrafish mutants provides evidence of complex genetic control of pigment cell development from multipotent NCCs [16]. Of the genes affected in these mutants, many have been shown to encode transcription factors that regulate fate specification of pigment cells from NCCs. Such factors may be required by either all three pigment cell lineages (e.g. colourless/sox10; [17], [18]) or only a single pigment cell lineage (e.g. nacre/microphthalmia-associated transcription factor a/mitfa, [19]). Consequently, genetic loss of several transcription factors affects one or more pigment cell types, indicating the existence of shared progenitors. For example, loss of sox10 function results in lack of all three zebrafish lineages, but also hinders the development of peripheral nervous system components ( [20], [21]). Intriguingly, a gene mutation affecting only the three pigment cell lineages has not been identified, thus the existence of a tripotent progenitor, exclusively generating chromatophore lineages [10], and the process of pigment cell diversification remain unclear.
As a first step towards understanding the complex GRN governing NCC fate restriction towards pigment cell lineages, it is important to define the key components involved in fate specification of individual lineages. Of the three zebrafish pigment cells, melanocytes are the currently best studied. In this lineage, Sox10, in conjunction with Wnt signalling, is required to activate and maintain mitfa transcription ( [15], [18], [22]- [24]). Like its mammalian counterpart, MITF, Mitfa has been proven necessary and sufficient to upregulate numerous melanocyte differentiation genes, including those controlling melanin synthesis (e.g. dct, silva and tyrosinase). Mitfa is thus dubbed the 'master regulator' of melanocyte fate choice ( [15], [19], [23]). However, in this role, Mitfa is supported by Tfap2, especially Tfap2a which acts as a key co-factor [25]. In contrast, Foxd3 mutants show a complex mutant phenotype that includes delayed melanocyte and xanthophore specification and differentiation, ultimately resulting in wild-type cell numbers, and reduced iridophore numbers ( [12], [26]- [28]). These phenotypes seem to reflect roles for FoxD3 in both lineage priming [29] and, in certain contexts, repressing melanogenesis promoting the specification of other fates ( [12], [13]).
Recent studies of key zebrafish iridophore mutants have begun to define the basic genetics of this cell type. In terms of their differentiation, heightened purine synthesis is central to the development of the guanine crystals that form the reflecting platelets. Thus, purine nucleoside phosphorylase 4a (pnp4a), which encodes an enzyme important in the biosynthesis of guanine, is a robust marker of mature iridophores ( [13], [14]). Additionally, mutations of several enzymes specific to differentiated iridophores have been shown to disrupt purine biosynthesis [30], while disruption of other proteins was found to impair iridophore survival ([31]- [33]). Furthermore, a signalling pathway crucial to fate specification, proliferation and differentiation of iridophores have been highlighted by mutations in the gene encoding the Leukocyte Tyrosine Kinase (Ltk; [11], [34]) receptor tyrosine kinase, with corroboration from targeted loss of function of its ligand ( [35], [36]). In shady/ltk mutants fate specification and proliferation of iridophores, but not other pigment cells, are impeded ( [11], [34]).
Nevertheless, Ltk signalling alone is not sufficient for iridophore specification, since iridophores are eliminated in sox10 mutants, even though ltk expression is strongly detectable by WISH ( [11], [14]). The function of Ltk signalling in specification of the iridophore lineage is, thus, likely analogous to that of Wnt signalling in generating melanocytes. This then leaves open the question of whether there is a 'master' transcriptional regulator of the iridophore lineage, analogous to the role of Mitfa in melanocytes.
The zebrafish "MiT" (Mitf/Tfe) family consists of six genes [37]; of these, mitfa and tfec are the only ones expressed in the NC. The distinctive expression pattern of tfec in cells along the dorsal and ventral midline in the trunk and tail and two patches over the yolk, is strongly reminiscent of differentiating iridophores [37]. Furthermore, Higdon et al. performed transcriptomic analysis to compare FACS-purified iridophores, melanocytes, and retinal pigment epithelium (RPE), and found tfec to be one of several transcription factor genes with enriched expression specifically in the iridophore lineage [38]. Together, these data lead to the hypothesis that Tfec might be the iridophore master regulator, equivalent to Mitfa in developing melanocytes.
In a recent detailed study, we demonstrated that, indeed, tfec serves as a robust marker of the iridophore lineage, allowing us to define the major stages of iridophore development [14].
Based on whole-mount in situ hybridisation (WISH) studies of tfec expression patterns throughout a developmental time-course, and co-expression analysis with both ltk and mitfa, we concluded that tfec is first expressed extensively throughout the premigratory NC progenitors of the trunk and tail at 18 hpf, and then exclusively in the developing iridophore lineage. Focussing on cells in the posterior trunk, we distinguished a subset of tfec-positive cells within the premigratory domain that have downregulated the early NCC marker, foxd3, but do not express definitive pigment cell markers; we refer to these as early chromatoblasts (Cbl early). At approximately 22 hpf, ltk and mitfa are upregulated in the tfec+; foxd3premigratory cells of the trunk ( [11], [14], [19]), indicating that they correspond to partially restricted progenitors, capable of generating pigment cells (Cbl late). By 24 hpf, tfec labels migrating progenitors which co-express mitfa and ltk markers, and which we consider fatespecified iridoblasts, but which likely retain at least melanocyte potential too (Ib(sp); [14]).
From 30 hpf, tfec+ cells co-express ltk, but not mitfa, and we now consider these to be either definitive iridoblasts (Ib(df), along the dorsal trunk and on the lateral migration pathway at 30 hpf), or mature iridophores (Iph; in iridophore locations from 42 hpf onwards) according to their position and state of visible differentiation. Thus, the established iridophore-associated expression pattern of tfec during zebrafish development reinforces the hypothesis that Tfec might act as the missing iridophore master regulator, but does not eliminate the possibility of additional earlier functions, either in multipotent early NCCs (eNCCs), or in partially restricted precursors with wider potencies.
In our previous study we used a preliminary assessment of tfec mutants to inform our derivation of a core iridophore GRN. Here, we describe in detail the generation and comprehensive characterization of the effects on NC development of mutations in tfec.
Following examination of the development of a wide variety of NC derivatives in tfec mutants, using early and late molecular markers, we conclude that, although neuronal and skeletal derivatives develop normally, specification of all pigment cell fates is delayed in homozygous mutants, suggesting a common early requirement for tfec in the GRN governing specification of all three chromatophore lineages, and providing support for a common chromatoblast precursor. Finally, our previous work identified the GRN governing maintenance of tfec in the iridophore lineage [14]. In the present study, we extend this work to define tfec as the iridophore master regulator and, importantly, to identify its upstream regulators in the multipotent premigratory NC, thus placing the transcription factor in context in the NCC specification GRN [39]. Together, these data shed light on the possible mechanism of progressive fate segregation of NCCs, and begin to elucidate the complex role for Tfec, being indispensable for iridophore development, but also playing subsidiary roles in specification of the other two chromatophores derived from the zebrafish NC.

tfec is a candidate for the iridophore master regulator
As we showed previously [14], tfec is co-expressed with the established iridophore marker, ltk [11], during iridoblast fate choice and iridophore differentiation. Although Higdon et al showed that tfec expression was prominent in the RNA-seq profiles of iridophores, they also detected low levels of expression in purified melanocytes [38]. Here we used WISH to detect tfec in individual embryos, following photographic documentation of their individual iridophore patterns, to show definitively its presence in differentiated iridophores (Fig. 1). tfec is maintained in differentiated cells (Fig. 1A-D). At these stages, consistent with our previous observations showing no overlap of tfec and the melanocyte marker mitfa in differentiated melanocytes [14], we do not detect expression in neighbouring differentiated melanocytes occupying the dorsal and ventral stripes ( Fig. 1A-D). Likewise, xanthophores, which are widespread under the epidermis of the flanks of the embryos, also do not show detectable tfec expression in these WISH studies (Fig. 1A-D). To confirm this, we used the xanthophore lineage marker, Pax7, detected via an immunofluorescence assay combined with simultaneous labelling of tfec transcript via WISH (Fig. 1E-G). In conclusion, at the detection threshold of WISH, tfec is a consistent marker of mature iridophores, but not of melanocytes nor xanthophores.

Loss of tfec function affects the development of all embryonic pigment cells
To assess the role of tfec in development, we induced mutations in tfec using CRISPR/Cas9 ( Fig. 2), selecting a target in the seventh exon of the gene, which encodes the second helix of the transcription factor's helix-loop-helix dimerization domain. Our two laboratories independently generated identical 6 base pair deletions (the tfec ba6 and tfec vc58 alleles) in two different wild-type strains, WIK and NHGRI-1, in addition to frameshifted alleles ( Fig. 2A). We reasoned that in this region of the gene it was likely that even indels that retained the correct reading frame (i.e., multiples of three) would likely be deleterious, because they would alter spacing of key residues and surfaces within this helix. Indeed, all of the generated alleles, when made homozygous, resulted in elimination of differentiated iridophores from the dorsal, ventral and yolk sac stripes, as well as from the lateral patches of the embryo ( Fig.   2F-I). In addition, iridophores were absent from the dorsal head (Fig. 2G,I) and the eye ( Fig.   2F-I) of homozygous mutants. Moreover, both injected (G0) fish raised to adulthood, as well as a single 'escaper' surviving F1 adult carrying biallelic frameshift mutations, lack iridophores in patches or in whole ( Fig. S1 A-C). All results presented here were produced using either the tfec ba6 or the tfec vc60 alleles, unless stated otherwise.
Quantification of iridophore numbers along the dorsal and ventral stripes of live embryos at 3 dpf illustrates the severity of the phenotype, with only very rare escaper iridophores present in homozygous tfec ba6 mutant embryos ( Fig. 2J; Table S1). The numbers of differentiated cells in heterozygous mutants are not significantly different from those in wild-type (WT) siblings (Table S1). The iridophore phenotype could be successfully rescued via injection of a Tol2 transposon-based plasmid containing 2.4 kb of the tfec promoter [40], guiding tissue-specific expression of full-length tfec (Fig. 3). Although the WT number of iridophores was not recovered, almost half of the injected mutant embryos presented with rescue of iridophores either on the eye, the trunk, or both of those domains (Fig. 3E). Successful iridophore rescue was also visible in injected fish raised to adulthood (Fig. S1 D).
In previous work, we showed that tfec is present in multipotent premigratory NCCs, which do not yet express definitive pigment cell markers (early NCCs, early Cbls; [14]). We further demonstrated that during early stages of specification and migration of NC derivatives, tfec expression transiently overlaps with that of mitfa in specified, but not definitive, iridoblasts (Ib(sp)). Here we report that melanogenesis is delayed in 30 hpf homozygous tfec mutant embryos when compared to WT or heterozygous siblings, supporting a functional role for Tfec during its transient expression in melanoblasts. Specifically, we observed a significant reduction in the numbers of differentiating melanocytes along the dorsal trunk, and the medial and lateral migration pathways ( Fig. 2B,C,K; Table S1). Melanocyte development recovers, and by 4 dpf mutant embryos present with the same number of melanised cells along their trunk as their WT or heterozygous siblings (Fig. 2D,E,L; Table S1). Furthermore, we see strikingly reduced melanisation of the retinal pigmented epithelium (RPE) of homozygous mutant embryos at 30 hpf, compared to WT or heterozygous siblings (Fig. 2B,C), suggesting an analogous role in these brain (not NC)-derived melanocytes. Surprisingly, we observed a mild, yet consistent and statistically significant increase in differentiated melanocytes on the dorsal head of mutant embryos at 4 dpf ( Fig. 2D,E,L; Table S1).
The delayed melanogenesis phenotype in these mutants might result from delayed specification of melanoblasts, or from slowed differentiation of normally specified melanoblasts. To distinguish between these two possibilities, we performed chromogenic WISH at 24, 30 and 48 hpf to detect expression of the melanocyte master regulator, mitfa.
Strikingly, mitfa expression was restricted to premigratory late Cbls in tfec mutants, whereas mitfa-positive melanoblasts occupied the medial and lateral migratory pathways in WT and heterozygous siblings (Fig. 4M,N). At 30 hpf, the delay was still detectable. The trunk was occupied by a relatively small number of mitfa-positive NC derivatives, whereas in the tail of mutants cells had still not entered the migratory pathways ( Fig. 4O,P). Consistent with the live phenotype, mitfa expression in mature melanocytes at 48 hpf was unaffected in the trunk and tail of tfec mutants, compared to WT siblings (Fig. 4Q,R). This early retardation of mitfa expression, coupled to absence of ltk expression (Fig. 4S,T; [14]), but normal migration of neural derivatives along the medial pathway (Fig. 5G,H), suggested that specification of the mitfa+; tfec+ Ib(sp) [14] from late Cbl, was hindered in the absence of functional Tfec. In addition, these data support our previous suggestion [14] that these Ibl(sp) retain melanocyte potential (i.e. they can be considered both specified melanoblasts as well as specified iridoblasts), and show that melanocyte fate specification is delayed in the tfec mutant. The subsequent recovery of normal melanocyte numbers makes clear that compensatory factors allow melanoblasts to be specified, albeit with a short delay.
We set out to investigate the early specification events that lead to the observed pigmentation phenotypes using WISH studies of iridophore markers. In homozygous tfec mutants, expression of the differentiated iridophore marker, pnp4a ( [13], [14]), was undetectable at 48  [14]). As the remaining tfec-positive cells do not express other iridophore markers, such as pnp4a (Fig. 4L) or ltk [14], we hypothesise that these correspond to early partially-restricted NC derivatives, perhaps early Cbls. Finally, we examine ltk expression in tfec mutants. ltk expression was completely lacking on the medial migration pathway in homozygous mutants at 24 hpf (Fig. 4S,T; [14]), consistent with an early defect in iridophore specification. However, we note that ltk expression is also missing in the premigratory Cbl domain, indicating a much earlier role for Tfec, in specification of the Cbl(late) from Cbl(early).
The tfec mutant embryos did not show obvious changes in the number and distribution of mature xanthophores (Fig. 3B). However, examination of early xanthophore specification markers by WISH showed that the developmental delay in producing melanoblasts is also true for generation of xanthoblasts (

Loss of tfec function does not affect the development of non-pigment NC derivatives
Given the unexpected role in non-iridophore pigment cells, we then asked whether loss of tfec function affected non-pigment NC derivatives. To examine cartilage development, we stained tfec mutant embryos and WT siblings with Alcian Blue but did not note any visible phenotypic distinction in homozygous mutants (Fig. 5A,B). To assess neural fates, we used a series of standard markers. The number and distribution of dorsal root ganglion (DRG) sensory neurons, as labelled by anti-Hu immunofluorescence, was unaffected by loss of tfec function (Fig. 5M,N). Both our anti-Hu assays and traditional WISH staining for phox2b expression ( To examine whether specification of either ectomesenchymal or neural derivatives from multipotent progenitors is delayed in tfec mutants, similar to non-iridophore pigment cell derivatives, we conducted additional WISH experiments at 30 hpf, when relevant early specification markers are detectable. We thus showed that expression of both dlx2a in migrating cranial NCCs (Fig. 5C,D), and of pou3f1 (previously oct6) in migrating precursors of the pLLn appears (Fig. 5E,F), appeared normal in tfec mutants, compared to known WT siblings. Furthermore, detection of sox10 expression in batches containing WT, heterozygous and tfec mutant embryos failed to detect alterations in either the distribution or the abundance of glial progenitors migrating along the medial pathway (Fig. 5G-H). Furthermore, these in situs provide evidence that premigratory NCCs appear to be present in the normal numbers and show unaltered timing of loss of sox10 expression (Fig. 5G,H). Note that in assays aiming to detect sox10 transcript at 30 hpf, we were able to confirm the presence of homozygous mutant embryos based on reduced melanisation of the RPE (Fig. 5G,H, insets).
For experiments from 48 hpf onwards (Fig. 5I-L), homozygous mutants were processed separately from their siblings.
In summary, although tfec expression is prominent in the majority of, if not all, early NCCs ( [14], [37]), we could not detect any changes at any stage of the development of NC-derived neurons, glial cells and skeletal components. This suggests that although tfec transcript is present at these early stages, Tfec is uniquely required for pigment cell fate specification, and essential for iridophore fate specification.

tfec is downstream of NC specifier genes in the NC progenitor GRN
To determine the upstream regulators of tfec in premigratory early NCCs and early Cbls of the dorsal trunk, we conducted WISH experiments on single and double mutants for the important vertebrate NC specifier genes foxd3, sox9b, sox10 and tfap2a [39]. Interestingly, at 18 hpf, all embryos from crosses of foxd3, sox10 and tfap2a mutant carriers showed identical expression of tfec, strongly suggesting that early expression of tfec is not strictly dependent upon any one of these genes (Table S2). In contrast, in 18 hpf presumed homozygous sox9b mutants, tfec expression does not extend towards the tail as far as in WT or heterozygous siblings (Fig. 6K,L; Table S2), which is likely attributable to delayed specification of early NCCs upon loss of sox9b function. At 24 hpf tfec expression in WT embryos is gradually downregulated from the majority of premigratory NCCs of the trunk in an anterior to posterior manner, strongly persisting only in Ib(sp) [14]. However, we observed a persistence of tfec expression in the anterior premigratory NC domain in sox10, sox9b, tfap2a and foxd3 mutants at this stage, consistent with retained premigratory late Cbls (Fig. 6A-D,M-N). This persistence differed in severity and duration between different mutants, but homozygous mutants of each genotype show highly consistent phenotypes across experimental replicates. Specifically, as was previously reported for a single time point [14], in sox10 mutants, where NC derivatives fail to become specified and to enter the migration pathways ( [11], [18]), tfec expression is maintained in trapped late Cbls, extending to the hindbrain/trunk boundary (Fig. 6A,B). Our results show that, at all time-points, tfec-positive premigratory progenitors persist in homozygous mutants (identified by their lack of tfec expression in Ib(sp) positions), until 36 hpf (Fig. 6A-B,E-F,I-J). In each of sox9b, tfap2a and foxd3 homozygous mutants at 24 hpf, tfec+ premigratory NCCs are detectable along the dorsal trunk, but not reaching the hindbrain/trunk boundary as in sox10 homozygous mutants (Fig. 6C,D,N; Fig. S2).
As members of the same SoxE group, it is not surprising that sox10 and sox9b have been shown to be functionally redundant in DRG sensory neuron development [20]. We asked whether this might also be true for tfec expression in early or late Cbls. Examination of embryo batches containing sox10;sox9b double mutants at 18 hpf did not reveal elimination of tfec expression in any of the assessed embryos (Table S2). However, tfec expression was completely eliminated from the NCC progenitor domain of genotyped tfap2a;foxd3 double mutants ( Fig. 6M-O), suggesting that both these genes act together to upregulate tfec expression in premigratory NCCs of the trunk. This effect was not observed in genotyped siblings, heterozygous for one or both alleles nor those homozygous for a single mutant allele.
In conclusion, our data show key roles for foxd3 and tfap2a in establishing expression of tfec in premigratory multipotent NCCs.

Mitfa represses tfec during melanocyte development
Both tfec and the melanocyte master regulator, mitfa, are transiently expressed in the multipotent NCC domain, in late Cbls, as well as in Ib(sp). We conducted WISH to assess the effects of loss of mitfa function on tfec expression. Interestingly, presumed homozygous mitfa mutants show ectopic expression of tfec along the dorsal trunk and in NC derivatives along both the medial and lateral migration pathways (Fig. 7A-D). This pattern of tfec expression in mitfa mutants resembles WT mitfa expression in developing melanoblasts, therefore it is likely that Mitfa represses tfec expression in NCCs that become biased towards the melanocyte lineage. This effect persists at 30 hpf and is also observed by a complementary, and more sensitive fluorescent WISH technique, RNAscope; we see persistence of tfec expression in cells migrating through the lateral pathway (Fig. 7E-F'). These cells co-express the definitive lineage marker, ltk ( [11], [14]), suggesting that they correspond to specified iridoblasts.

Discussion
Our previous work established tfec as a marker during NC development and iridophore fate choice ( [14], [37]). Interestingly, tfec was found to be co-expressed with mitfa in Ib(sp) cells, proposed to be able to at least give rise to melanocytes and iridophores, but not in mature melanocytes. Thus, Ib(sp) should also be considered as specified melanoblasts. The first aim of the present study was to determine whether iridophores are indeed the only embryonic differentiated pigment cell expressing tfec; to this end, we conducted analyses showing that tfec expression is maintained at detectable levels only in mature iridophores, but neither in melanocytes nor in xanthophores.
Considering the strong sequence conservation between Tfec and the melanocyte master regulator, Mitfa, we next asked whether tfec might have a function in iridophores analogous to that of mitfa in melanocytes; i.e. as the master regulator of the iridophore lineage. We generated mutations in tfec using a CRISPR/Cas9 approach, obtaining several alleles which displayed essentially identical phenotypes. Although tfec mutants have been generated independently [41], that report did not examine NC-related defects, focusing instead on deficiencies in hematopoiesis. As with the CRISPR-generated exon 3 allele reported by Mahony and colleagues, our mutants fail to inflate the swim bladder (which, along with the caudal hematopoietic tissue, is another site of tfec expression; [37]) and die after approximately 12 days, apparently from lack of ability to feed. Potential postembryonic roles of tfec remain unclear, but data from mosaic adults and a single adult escaper suggest that tfec is required for iridophores throughout the lifetime of the animal (Fig. S1).
Our mutants presented with a complete absence of iridoblasts and mature iridophores but also, surprisingly, delayed differentiation of both NC-derived and RPE melanocytes. While differentiation of xanthophores is also delayed in tfec mutants, we found that development of all non-pigmented derivatives (neurons, glia and skeletal components) is unaffected. This striking phenotype restricted to pigment cell specification is unique amongst the characterised zebrafish pigment mutants. Decades ago, it was proposed that melanocytes share a common origin with iridophores and xanthophores from a pigment-restricted precursor [10], which we would term a chromatoblast (Cbl). More recently, analysis of ltk expression in sox10 mutants indicated the presence of an ltk+ precursor of pigment cells, consistent with the chromatoblast [11]. The tfec loss of function phenotype thus provides further support for the existence of a Cbl, transiently localised within the dorsal trunk of embryos between 18 hpf and 24 hpf ( [14]; Fig. 8). indirect. In multipotent NCCs, Tfap2a and Foxd3 redundantly activate tfec expression (red arrows; this work), which is later maintained in the iridophore lineage by Sox10 (purple arrow; [14]). This activation occurs independently of previously described co-regulation of zebrafish sox9b and sox10 expression by Tfap2a and Foxd3 (black arrows; [46]), and is unaffected by potentially conserved activation of sox10 and foxd3 by Sox9b (blue arrows, as shown in chick; [45], [47]). (C) Following transition of Cbl early to Cbl late, tfec expression is supported by positive feedback interactions between Sox10, Ltk and Tfec, as described in Petratou et al., 2018 [14]. Sox10 directly activates mitfa expression (solid green arrow; [23]), which during early Cbl specification is co-expressed with tfec [14], inhibiting its expression to bias progenitors towards the melanocyte fate. (D) As the lineage progresses past the Cbl, into the ib(sp), ib(df) and iph stages, factor R (FR) mediates Tfec-dependent downregulation of Mitfa (dark blue edge), and expression of the marker pnp4a becomes prominent [14]. In (B-D) the black box outlines the core connecting the three networks. MiT, microphthalmia family transcription factors; PNS, peripheral nervous system. Based on our findings, tfec is the first zebrafish gene identified that specifically affects all chromatophore lineages without appearing to affect any non-pigment NC derivatives. While the key NC transcriptional regulator, sox10, is required for each of the three pigment cell lineages, loss of sox10 function additionally results in strong reductions or absence of all peripheral glia cells and NC-derived peripheral neurons ( [17], [20], [21], [23]). Furthermore, this requirement of pigment cells for sox10 manifests itself apparently in different ways; sox10 is required for mitfa expression in late Cbls, and mitfa can promote melanocyte fate in the absence of sox10 if misexpressed [23]. In contrast, sox10 mutants still strongly express tfec (as well as ltk) within trapped Cbls, yet fail to produce Ib(sp) [14]. We show here that in tfec mutants the ability to express the lineage markers of melanocytes and xanthophores is delayed but otherwise unaffected, whereas ltk expression is completely missing, even in Cbl(late). Thus, Tfec is necessary for both correct development of the Cbl(late) and for Ib(sp) specification from such a cell.
In order for tfec to be considered the iridophore master regulator, the gene must not only be necessary, but also sufficient for iridophore specification. Our data are strongly suggestive of tfec being required for iridophore development, as mitfa is for melanocytes (Fig. 8). Injection of wild-type tfec cDNA under its own promoter could rescue, albeit only partially, the iridophore phenotype in tfec mutants, with rescued iridophores in the eye observed more often than in the trunk. Whilst these results show that Tfec is sufficient to rescue iridophore specification in tfec mutants, fulfilling the conditions for a master regulator gene, gain of function experiments will be required to determine if the mechanism by which Tfec functions in iridophore specification is analogous to that of Mitfa, which can trigger melanocyte development in the absence of sox10 by independently triggering a feedback loop and expression of melanogenic genes ( [15], [23]). Further work to define direct and indirect transcriptional targets of Tfec, as well as any other transcriptional co-regulators, similar to the role of Tfap2a for Mitfa [25], is needed to understand the mechanistic details of iridophore fate choice. RNA-sequencing and ATAC-sequencing assays on iridophore precursors would be invaluable to determine, in an unbiased manner, appropriate candidates.
Another aim of the work presented here was to extend the GRN surrounding tfec beyond the iridophore lineage by examining its regulation in early NCCs ( [14], [37]). Initiation of tfec expression in the early NC is not affected by loss of tfap2a, foxd3 nor sox10 alone. Notably, loss of sox9b alone caused an apparent delay in induction of tfec expression in posterior NC yet, despite this alteration in the pattern, expression remained strongly present in the progenitor population.
Sox10 and Sox9b have previously been reported to act redundantly during zebrafish development [15], [20]. Our data, however, demonstrates that tfec expression in double sox10;sox9b mutant embryos is still strongly activated in the NC. It remains to be shown whether loss of function of a single or of both sox10 alleles modifies the degree of tfec expression delay noted in homozygous sox9b mutants. Double mutants for tfap2a;foxd3 have been shown to eliminate induction of NC [42].
Redundant activities of Tfap2a and Tfap2c are required for NC induction and development of other non-neural ectoderm derivatives in zebrafish embryos ( [42], [43]). Consistent with this, we found that foxd3 and tfap2a are redundantly required for induction of tfec expression in early premigratory NCCs (Fig. 8B). In this context, it is interesting that tfec was recently identified as a likely direct target of Tfap2a/2c, through analysis of gene expression changes in mutants combining different numbers of tfap2a and tfap2c mutant alleles [44]. The same study demonstrated the functional compensation of Tfap2a by Tfap2c, since in tfap2a mutants just a single copy of tfap2c was sufficient to maintain tfec expression at WT levels and rescue NC specification. Our data support the above finding, showing that transcriptional regulation of tfec via Tfap2 transcription factors is independent of them first activating sox10 or sox9b ([45]- [47]) (Fig. 8B). Moreover, our assays indicate that disruption of Foxd3 activity alongside Tfap2a counteracts functional compensation by Tfap2c. Further work will be required to assess whether this is due to a role for Foxd3 in Tfap2c expression.
Our data formally establishes tfec as a member of the GRN governing maintenance of NCC progenitors [39]. Furthermore, our data provides the first evidence for Tfec function in those early progenitors, since we show it is required to drive early expression of ltk (in Cbl(late)), as well as fate specification of the iridophore lineage from these multipotent progenitors.
Although tfap2a and foxd3 act redundantly to activate tfec expression in early NCCs, they also present with divergent ongoing effects in pigment lineages. Single mutations in tfap2a and foxd3 affect the melanocyte and iridophore lineages, respectively ( [26], [27], [48]), likely in a manner dependent upon distinct regulatory interactions with Mitfa. While tfap2a and mitfa work in parallel to promote melanocyte differentiation [25], foxd3 has been suggested to repress mitfa transcription ( [12], [49]), at least in some contexts [14]. In the absence of foxd3, iridophore numbers are reduced in a manner that is at least partially mitfa-dependent, and marker analyses and lineage-tracing experiments support the existence of a bipotent melanocyte-iridophore (MI) precursor, the fate of which is influenced by this foxd3/mitfa interaction [13]. Our findings that mitfa represses tfec expression in melanoblasts, and that tfec mutants have increased melanocytes, indicate that maintenance of tfec activity is also key to the melanocyte versus iridophore cell fate decision. Notably, Mitfa and Tfec have the potential to physically interact as a heterodimer ( [50], [51]), which adds an additional layer of complexity when attempting to elucidate the mechanism of cell fate choice. Interestingly, despite the similarity between tfec and ltk mutant iridophore phenotypes, ltk mutants do not show an analogous late increase in melanocytes [11].
Specification and differentiation of the third pigment cell type of zebrafish, the xanthophore, has been shown to depend upon the paired-box transcription factors Pax3 and Pax7a/b ( [52], [53]). Intriguingly, interactions between Pax3 and Mitf have been demonstrated in mammal melanocyte development [54]. At least some zebrafish xanthoblasts express mitfa [55] and we show that loss of tfec delays xanthoblast migration, raising the question of whether interplay between Mitfa and Tfec might be important for xanthoblast fate choice.
Furthermore, it will be of interest to examine potential interactions between not only Tfec and Mitfa, but also between MiT and Pax transcription factors during chromatoblast diversification.
To conclude, our study contributes to deepening our understanding of the molecular basis of NC and pigment cell development in zebrafish, as well as the process of progressive fate restriction of multipotent stem cells. Detailed assessment of the diversity of pigment progenitor states in these embryonic stages is needed to test the hypothesis of a tripotent chromatoblast. Furthermore, focused effort on the (redundant) roles of transcription factors in xanthophore development will be decisive in understanding pigment cell fate choice in zebrafish.

Ethics statement
This study was performed with the approval of the University of Bath ethics committee and in full accordance with the Animals (Scientific Procedures) Act 1986, under Home Office Project Licenses 30/2937 and P87C67227, and in compliance with protocol AM10125 approved by the Institutional Animal Care and Use Committee of Virginia Commonwealth University.

CRISPR mutagenesis
Target sequences for CRISPR/Cas9 mutagenesis were identified using version 1 of the CHOPCHOP webtool (http://chopchop.cbu.uib.no/; [60]). The template for synthesis of the guide RNA (gRNA) was generated using a previously described PCR method [61]. Primer sequences are provided in Table 1 Cat# AM1354) and purified using the miRvana miRNA Isolation Kit (Invitrogen; Cat# AM1560) as described by [62]. Capped Cas9 mRNA was generated from the plasmid pT3TS-nCas9n [62], a gift from Wenbiao Chen (Addgene plasmid; Cat# 46757) using the mMESSAGE mMACHINE T3 Transcription Kit (Invitrogen; Cat# AM1348). Cas9 mRNA and tfec exon 7 sgRNA were each diluted to 100 ng/µl for microinjection into one-cell embryos of the NHGRI-1 strain [63]. A fraction of the injected set was sacrificed for genomic DNA preparation [64] to evaluate the efficacy of the guide RNA using the primers cex7F and cex7R, followed by restriction digest with StuI, which cuts in the target sequence. The remaining injected embryos were raised to adulthood and intercrossed or mated to wild-type (NHGRI-1) partners. F1 carriers were identified using the PCR digest assay above, and undigested PCR products (representing mutant alleles) were purified and sequenced.

High resolution melt analysis
High Resolution Melting (HRM) Software V3.0.1 (Thermo Fisher Scientific) was used to detect and amplify differences between the melting temperature of 150-200 bp q-RT PCR amplicons generated from reference wild-type (WT) samples versus mutagenized embryos or adults. To perform q-RT PCR for HRMA, template genomic DNA was extracted using the KAPA Express Extract Kit (Sigma-Aldrich; Cat# KK7103) according to manufacturer's instructions, and was diluted to 8 ng/µl. Amplification reactions were set up as per manufacturer's instructions using KAPA HRM FAST reagents (Sigma-Aldrich; Cat# KK4201) and primers designed according to the relevant recommendations (Table 1). Following amplification, a continuous melt curve was generated by increasing the temperature from 60 ºC (1 min) to 95 ºC (15 sec) in 0.3 ºC /sec increments. To detect CRISPR/Cas9-mediated mutagenesis, at least 8 WT reference samples were included in the analysis.

Chromatophore counts
Melanocyte counts were performed at 30 hpf and at 4 dpf on anaesthetized or fixed embryos under transmitted light. Embryos at 4 dpf were treated with 2 µM melatonin directly prior to counting. To count iridophores, PTU-treated embryos were observed under incident light.
Pigment cell counts were made under a Zeiss Axio Zoom.V16 fluorescence stereo zoom microscope.

Cloning and rescue by plasmid microinjection
The full-length coding sequence of tfec was subcloned in-frame into the Gateway 3' Entry vector p3E-2A-FLAG-pA to make p3E-2A-FLAG-tfec-pA. A multisite Gateway LR+ cloning reaction was then carried out with this plasmid along with Tol2 Kit destination vector pDestTol2pA2 and entry vector pME-mCherry-no stop [65] and entry vector p5E-tfec2.4 [40].

Transcript detection in whole-mount embryos
Detailed information on the preparation of generic materials and the protocols for performing chromogenic WISH as well as multiplex fluorescent RNAscope can be found in Petratou et al.
Embryos were observed and imaged, and the Pearson's chi-squared test was used, as previously described [14], to process and statistically analyse results, to test the hypothesis that altered phenotypes correlated with homozygosed mutations. tfap2a and foxd3 mutant embryos were identified using previously described genotyping protocols ( [27], [70]) following imaging and preparation of genomic DNA [64].

Alcian blue staining and immunohistochemistry
Larvae from an intercross of tfec vc60 heterozygous adults were sorted at 4 days postfertilization based on the iridophore phenotype, and then were fixed in separate tubes overnight in 4% PFA. Alcian blue staining was then carried out essentially as described [71].
Samples were imaged using an Olympus SZX12 stereomicroscope with DP70 camera.
Immunohistochemistry was carried out as previously described [68]. Primary monoclonal antibodies against HuC/D (Molecular Probes) and Pax7 (Developmental Studies Hybridoma Bank) were used at 1:500 and 1:20 respectively, and goat anti-mouse secondary antibodies conjugated to Alexa 568 or Alexa 488 (Molecular Probes) were each used at 1:750 dilution.
For combined tfec WISH/Pax7 IHC, the Pax7 antibodies were added simultaneously with the anti-Fab fragments [72]. Samples were imaged on a Zeiss Axio Imager.M2 compound microscope with Axiocam 503 colour camera, processed using ZEN software and Adobe Photoshop CC 2018 and 2019.