Zebrafish Endzone Regulates Neural Crest-Derived Chromatophore Differentiation and Morphology

The development of neural crest-derived pigment cells has been studied extensively as a model for cellular differentiation, disease and environmental adaptation. Neural crest-derived chromatophores in the zebrafish (Danio rerio) consist of three types: melanophores, xanthophores and iridiphores. We have identified the zebrafish mutant endzone (enz), that was isolated in a screen for mutants with neural crest development phenotypes, based on an abnormal melanophore pattern. We have found that although wild-type numbers of chromatophore precursors are generated in the first day of development and migrate normally in enz mutants, the numbers of all three chromatophore cell types that ultimately develop are reduced. Further, differentiated melanophores and xanthophores subsequently lose dendricity, and iridiphores are reduced in size. We demonstrate that enz function is required cell autonomously by melanophores and that the enz locus is located on chromosome 7. In addition, zebrafish enz appears to selectively regulate chromatophore development within the neural crest lineage since all other major derivatives develop normally. Our results suggest that enz is required relatively late in the development of all three embryonic chromatophore types and is normally necessary for terminal differentiation and the maintenance of cell size and morphology. Thus, although developmental regulation of different chromatophore sublineages in zebrafish is in part genetically distinct, enz provides an example of a common regulator of neural crest-derived chromatophore differentiation and morphology.


Introduction
The neural crest is a transient vertebrate embryonic cell population that gives rise to a wide variety of cell types, including chromatophores, craniofacial cartilage, and neurons and glia of the peripheral nervous system [1]. This array of neural crestderived cell types has long been of interest in studying the mechanisms of cell diversification among embryonic cell populations. The development of neural crest-derived chromatophores in particular has been studied extensively, and many important mechanistic insights have resulted from the analysis of mouse and zebrafish mutants [1][2][3][4][5].
Vertebrate chromatophore populations are readily observed, as they produce their own visible intrinsic markers. In addition, chromatophores are not strictly required for viability [5][6][7]. As a result, chromatophores have long been used to study developmental processes such as cell fate specification, proliferation, migration, differentiation, and survival. Mice and other mammals have a single chromatophore cell type termed melanocytes [8]. Hundreds of mouse coat color mutants have been identified, covering over 100 loci, which affect multiple cellular processes [4,5]. Further, many of these mutations in mice have proved to be medically relevant as models for human diseases involving the same genes [9]. Besides the melanocytes (melanophores) also found in mammals, zebrafish and other ectotherms possess neural crest-derived yellow xanthophores and iridescent iridiphores [10,11]. In addition to the isolation of several zebrafish chromatophore mutants that arose spontaneously [12,13], numerous mutagenesis screens have yielded over 100 mutations affecting various processes in the development of different combinations of the chromatophore types [6,[14][15][16][17].
Studies from several vertebrates, including zebrafish, have led to the extensive characterization of melanophore development, and to a lesser extent, xanthophore and iridiphore development [2,4,8,18,19]. Prior to overt differentiation, chromatophore precursors are referred to as chromatoblasts, and can be identified by expression of genes specific to one or multiple chromatophore sublineages. Sox10, mutations in which cause Waardenburg-Hirschsprung Syndrome in humans, is required for development of nonectomesenchymal neural crest derivatives, including all chromatophores, as well as many peripheral neurons and glia [20,21]. Sox10 has been shown to directly regulate expression of microphthalmia-associated transcription factor (mitf), which is both necessary and sufficient for melanophore development, and dopachrome tautomerase (dct), an enzyme in the melanin synthesis pathway [6,[22][23][24]. The receptor tyrosine kinase c-kit is also expressed by melanoblasts, and appears to be necessary for their migration and survival [25,26]. Similarly, the kit ortholog fms, which has no known role in mammalian melanocyte development, is required for the migration of embryonic xanthophores and the specification of a subset of adult melanophores in zebrafish [27,28]. fms is expressed by embryonic xanthoblasts and macrophages, which can be distinguished from one another based on location and cellular morphology [28][29][30]. Synthesis of yellow pteridine pigments, found in xanthophores, requires xanthine dehydrogenase (xdh), which is correspondingly expressed by xanthoblasts [28,31,32]. Both fms and xdh are co-expressed in a subset mitf + cells in the premigratory neural crest, which may represent uncommitted precursors of melanophores or xanthophores [28]. Neither of these genes is co-expressed with c-kit, however, and both have been used as specific diagnostic markers of xanthoblasts at migratory and post-migratory stages of chromatophore development [28,33]. The enzyme GTP-cyclohydrolase (Gch) is involved in the conversion of intermediates of both melanin and pteridine synthesis [34][35][36]. Accordingly, gch expression is observed in both melanoblasts and xanthoblasts [28]. A G protein-coupled receptor, endothelin receptor (ednr) b, is also expressed by neural crest-derived chromatophore precursors [37][38][39][40][41] Homozygous ednrb mutant mice are almost completely devoid of melanocytes [4,5]. In contrast, zebrafish ednrb1/rose mutants display defects in subsets of adult melanophores and iridiphores but lack an embryonic chromatophore phenotype [37]. In the zebrafish embryo, ednrb1 is initially expressed by all chromatophore sublineages, but by late embryonic/early larval stages, is restricted to iridiblasts and iridiphores [37].
Morphologically, differentiated melanophores and xanthophores are large and dendritic with many processes, while iridiphores are rounded in shape [8]. In ectotherms, considerable attention has been given to mechanisms of color adaptation, reversible changes in pigmentation brought on by prolonged exposure to either light or dark environments [42]. Extensive analyses, especially in a variety of fish species, have revealed that this occurs through relocalization of pigment organelles within cells, changes in cell morphology, and proliferation and apoptosis of chromatophores [42][43][44][45]. In adults, these processes appear to be under hormonal, as well as nervous control [43][44][45]. amelanophore-stimulating hormone (a-MSH) and melanin-concentrating hormone (MCH) appear to have mutually antagonistic effects on melanophores, with a-MSH enhancing melanin and melanophore development, and MCH promoting aggregation of melanosomes and downregulating secretion of a-MSH [46][47][48][49]. The effects of a-MSH are enacted in part through influence on mitf expression in mammals [50][51][52]. Rho GTPase-mediated cytoskeletal rearrangements may play roles in redistribution of pigment organelles and dendrite collapse [42].
Despite these and other data that have been amassed regarding chromatophore development in a variety of vertebrate systems, many questions remain. For example, relatively few genes have been identified that are required for the development of all embryonic chromatophores, and yet are specific to chromatophore sublineages within the neural crest. For example, Sox10 is necessary for development of chromatophore sublineages, but is also required for some crest-derived neurons and glia, and mammalian ednrb is necessary for both melanocyte and enteric neuron development [39,[53][54][55][56][57]. Further, although expressed by all three zebrafish chromatophore cell types, ednrb1 appears to be dispensable for their embryonic development [37]. While hormonal and neuronal influences have been demonstrated for pigment distribution, cell morphology, proliferation and survival of differentiated chromatophores, less is known about the downstream effectors governing these processes. Additionally, these studies focus on adults, and comparatively little is known about control of pigment cell morphology or color adaptation at embryonic stages.
We report here characterization of the zebrafish mutant endzone b431 (enz b431 ;enz), which was isolated based on the abnormal appearance of embryonic chromatophores. Our results indicate that all three chromatophore types are similarly reduced in cell size and number in enz mutant embryos. We show that enz is required specifically for chromatophore sublineages within the neural crest, and that this requirement is cell autonomous. Further, we report the identification of multiple enz alleles and progress toward the molecular identification of the enz locus. We suggest that enz is a relatively late cue required by pigment cell sublineages of the neural crest during embryogenesis, and is indicative of common requirements of chromatophores that are distinct from other neural crest derivatives.

Results
Live phenotype of endzone mutants enz was identified in a chemical (ethyl-nitrosourea, ENU) mutagenesis screen for mutations affecting neural crest derivatives as previously described [15]. The enz mutant was identified based on its altered pigment pattern. At 54 hours post-fertilization (hpf), the morphology of melanophores is dramatically altered being smaller and less stellate in enz b431 mutant embryos compared to wild-type siblings (compare insets, Fig. 1A, B). In addition, the pigmented retinal epithelium (PRE) of enz homozygotes, which is not neural crest-derived, is pale compared to that of wild-type siblings prior to 48 hpf. Unlike crest-derived melanophores, however, the PRE phenotype recovers rapidly ( Fig. 1 and data not shown). At 54 hpf, enz b431 embryos also lack the characteristic yellow cast caused by xanthophores concentrated dorsally in wild-type embryos ( Figure 1). In addition, iridiphores are reduced in number and size in 82 hpf and 6 days-postfertilization (dpf) enz mutant larvae compared to wildtype siblings ( Figure 1F-H, Table 1 and data not shown). Recently, we identified three additional ENU-induced enz alleles as determined by non-complementation of enz b431 and linkage analysis ( Figure 1 and data not shown). All four enz alleles are recessive. These alleles vary in the severity of the developmental defects in all three chromatophore phenotypes. enz os15 is less severe than enz b431 , enz os7 , and enz os18 , which are similar in expressivity. All data refer to enz b431 unless otherwise specified. Most enz homozygotes do not develop swim bladders, and subsequently fail to survive past early larval stages (Figure 2A, B). enz mutant larvae that do develop swim bladders survive, but are runted compared to wild-type siblings through adulthood. Qualitative differences in melanophore pigmentation persist through at least 30 dpf ( Figure 2). As adults, enz homozygotes are viable as both males and females, and are fertile (not shown), although they remain smaller than wild-type siblings.
Within the neural crest lineage, enz selectively affects chromatophore development Because of the visible defects in neural crest-derived chromatophore development in enz mutant embryos, we investigated whether the development of other cell types derived from the neural crest were affected by the enz mutation. Molecular markers indicated that neural crest-derived peripheral neuron and cranial glial populations develop normally in enz homozygotes ( Table 2). For example, cervical sympathetic neurons, enteric neurons, and neurons of the dorsal root ganglia are all present in qualitatively normal numbers and positions in enz mutant embryos ( Figure 3 and data not shown). Craniofacial cartilage, stained with alcian blue, was also found to be normal in terms of individual elements and their shapes ( Figure 3A, B). In addition, the pan-neural crest marker crestin was used to analyze neural crest populations at different embryonic stages. In wild-type embryos, crestin is first  . Some melanophores recover in size and morphology in enz homozygotes (D) compared to wildtype siblings (C). (E-L) Those enz homozygous larvae that do develop swim bladders survive, but are runted compared to wild-type siblings (stagematched wild-type and enz mutant larvae shown at the same magnification). Lateral (E, F) and dorsal (G-J) views of wild-type (E, G, I) and enz mutant (F, H, J) larvae at 21 dpf (E-H) and 30 dpf (I, J). Melanophores continue to be paler in enz mutant larva than in wild-type siblings through at least 30 dpf (see arrowheads in G-J). (K, L) Lateral views of 119 dpf wild-type (K) and enz homozygous (L) adults. Normal overall morphology and pigmentation of enz mutant adults, as well as nascent fin stripe formation (L, arrowheads), suggests generalized growth retardation in enz mutants compared to wild-type siblings. doi:10.1371/journal.pone.0002845.g002 expressed in neural crest cells at the boundary of neural and nonneural ectoderm during gastrulation. Expression continues in premigratory and migratory neural crest cells, and persists until slightly after overt differentiation of neural crest derivatives, such that by 24 hpf, crestin-expressing cells are found throughout the embryo [58]. crestin expression is normal in enz embryos at all stages, indicating that neural crest induction and migration are unaffected by the enz mutation and that neither the proliferation or survival of early neural crest cells are overtly compromised by the enz mutation ( Figure 4A, B and data not shown). Further, we examined the expression of genes that, within the neural crest, are expressed by precursors for melanophores, [23], xanthophores [28] and all three chromatophore cell types [37] at 24 hpf. Expression of each gene is qualitatively normal at this stage, suggesting that effects of the enz mutation on crest-derived chromatophores occur relatively late in the development of these cell populations ( Figure 4C-H). These data indicate that within the embryonic neural crest lineage, enz is required specifically for chromatophore development and only after neural crest dispersal and initial differentiation have largely occurred.

Reduced numbers of differentiated chromatophores in enz mutant embryos
To determine whether appropriate numbers of chromatophores are generated in enz mutant embryos, we counted differentiated melanophores and iridiphores in enz homozygotes and wild-type siblings. At 4 dpf, the number of melanophores in enz mutant embryos is significantly reduced compared to wild-type siblings (P,0.0001, Table 1). Likewise, iridiphore numbers are reduced in enz homozygotes compared to wild-type siblings at 6 dpf (P,0.0001, Table 1). In both cases, only about 75% of the wildtype complement of each class of chromatophores is present in enz mutants at the stages examined. Quantification of xanthophore numbers was precluded by indistinct boundaries and overlap between these cells. However, methylene blue staining (see below) at 3 dpf, as well as fms expression between 48 and 53 hpf, revealed that while many xanthophores are present in enz homozygotes, their numbers are reduced in a manner qualitatively similar to that of melanophores and iridiphores ( Figure 5 and data not shown). Thus, enz mutations appear to result in similar reductions in the numbers of differentiated chromatophores, while other neural crest-derived cell types are unaffected.

Chromatophore cell morphology is altered in enz mutant embryos
In wild-type zebrafish embryos raised at 28.5uC, neural crestderived melanophores normally begin to differentiate at approximately 25 hpf, while xanthophores and iridiphores begin to overtly differentiate at about 42 hpf and 72 hpf, respectively [59]. By 27 hpf, primarily in anterior regions, large, stellate, dark melanophores are present in wild-type embryos. In enz mutant embryos, as in wild-type siblings, melanophores differentiate at ,25 hpf (data not shown). At 27 hpf, enz melanophores are stellate, but pale compared to those of wild-type siblings ( Figure 5A, B). After 27 hpf, enz melanophores begin to transition from the initial pale, stellate appearance to a dark, punctate form, while wild-type melanophores remain stellate and dark. The transformation of melanophores occurs in a rostro-caudal wave, and is complete by about 48 hpf (Figure 5C-F and data not shown). We hypothesized that this apparent change in cell morphology of enz melanophores might be the result of either redistribution of melanosomes within cells or of a change in cell shape. To distinguish between these two possibilities, we performed in situ hybridization of melanized 36 hpf and 48 hpf wild-type and enz mutant embryos, using the melanophore sublineage-specific riboprobes c-kit [7] and dct [23]. In wild-type embryos, c-kit and dct mRNAs are distributed throughout the cell cytoplasm, including in the processes, reflecting stellate cellular morphology ( Figure 5G, H and data not shown). The distribution of dct and c-kit mRNAs is punctate in enz melanophores, similar to the distribution of melanin (Figure 5I, J and data not shown). This is consistent with a cell morphology change, rather than only relocalization of melanin or melanosomes within a stellate cell. Subsequently, we quantified the area of punctate melanophores in enz mutant embryos compared to stellate melanophores in wildtype siblings. At 2 dpf, wild-type melanophores at cephalic levels have a mean area of 282.1619.1 mm 2 , while enz melanophores at similar axial levels have a mean area of 21.961.7 mm 2 (P,0.0001, Table 3). The change in melanophore cell morphology in enz mutant embryos further suggested that the apparent absence of yellow pigmentation might be due at least in part to a reduction in size of xanthophores in these embryos. Individual xanthophores are difficult to distinguish, and even fms and xdh expression appear as diffuse staining over the dorsal aspect of the embryo, precluding the quantitative type of analysis performed on melanophores ( Figure 6A, B) [11,14,28]. However, qualitative observations of xanthophore morphology were made using methylene blue, which is taken up specifically by xanthophores and is concentrated around active pterinosomes, the organelles that produce pteridine pigments [60]. Methylene blue staining revealed that while some xanthophores are present in enz homozygotes, these appear much  smaller and less stellate than xanthophores in wild-type siblings at 3 dpf ( Figure 6C, D). Similarly, iridiphores are reduced in size in enz mutant embryos compared to wild-type siblings. In contrast to melanophores and xanthophores, iridiphores in the trunk of wildtype embryos have a rounded, rather than stellate morphology at 72 hpf (see Figure 1). While this is also true in enz homozygotes, overall iridiphore cell size, as measured by area, is reduced compared to wild-type siblings. At 6 dpf, the average iridiphore area in enz mutant larvae is reduced by ,40% compared to that in wild-type siblings (P,0.0001, data not shown). Together, these data indicate that mutations in enz similarly affect all three neural crest-derived chromatophore types with respect to cell size and also result in loss of the the typical stellate morphology of melanophores and xanthophores.
enz acts cell autonomously with respect to melanophore development Mosaic analysis can be used to determine whether a mutation acts cell autonomously or non-autonomously, and thus predict the mode of action of a gene product within developmental pathways. Genetic mosaics were created between wild-type, enz mutant, and golden b1 embryos. To facilitate the identification of donor-derived melanophores, particularly from enz donors, we utilized hosts homozygous for the golden b1 mutation, which results in pale melanophores that are distinct from wild-type melanophores [7,61]. Donor embryos were labeled with lysinated rhodamine dextran (LRD). Cells from donor embryos were then transplanted into unlabeled hosts, in which chromatophore development was subsequently observed. LRD-labeled cells from wild-type embryos formed dark, stellate melanophores at 28 hpf when transplanted into enz hosts (n = 6; Figure 7). In the reciprocal experiment, LRDlabeled cells from wild-type or enz mutant embryos were transplanted into unlabeled golden b1 homozygous hosts [7,61]. While wild-type cells were able to form dark, stellate melanophores (n = 13; Figure 7D), enz mutant cells formed punctate melanophores (n = 5; Figure 7E). Together, these data demonstrate that enz acts cell autonomously with respect to melanophore development, and inferentially, may act in a similar fashion in the development all chromatophores.
The enz locus is located on chromosome 7 As an initial step towards cloning the endzone gene, we placed allele enz b431 on the zebrafish linkage map [62,63]. Simple sequence length polymorphisms [64,65] between AB* and WIK zebrafish strains were used to map enz to chromosome 7. Parthenogenetic diploid embryos were generated by suppressing the second meiotic division using early pressure [66]. enz and wildtype embryos were identified by live phenotype, and PCR was performed on DNA from individual enz and wild-type embryos. The enz locus was initially localized to chromosome 7 in EP diploid embryos using three markers, z10441, z11652 and z6819 (see Methods). Subsequent recombination analysis employing DNA from 889 individual haploid embryos established a tentative interval in which enz lies proximal to z8252 and z10441 and distal to z55887 and z6819. Analysis of compound trpm7/enz mutants supports a late role for enz in chromatophore development To investigate the role of enz compared to a known gene, we generated a line of fish doubly heterozygous for enz b431 and trpm7 b508 , a gene previously shown to be required for development of the majority of neural crest-derived melanophores [67][68][69]. trpm7 b508 appears to affect melanophore development prior to overt differentiation, as most melanophores fail to appear in trpm7 b508 homozygotes. Further, there is evidence that some melanophore precursors undergo apoptosis in trpm7 b508 mutant embryos [67]. Those melanophores that do develop in trpm7 single mutant embryos have punctate cell morphology upon differentiation, in contrast to the transiently stellate cell morphology of enz melanophores. In clutches obtained from double heterozygote crosses, four classes of phenotypes were observed. As expected, three of these classes were wild-type, trpm7 single mutants and enz single mutants. The fourth phenotypic class, approximately 1/16 of the total number of embryos, exhibited defects resembling both trpm7 and enz mutant embryos, and is presumed to represent double mutants. trpm7-enz double mutants exhibit an early reduction in melanophore number as well initial punctate melanophore morphology that is characteristic of the phenotype of trpm7 single mutants. Iridiphores and xanthophores, on the other hand, are depleted as in enz homozygotes. At approximately 65 hpf, double mutants are touch insensitive, as are trpm7 mutants. Thus, double mutants show a combination of the trpm7 and enz phenotypes. Depletion of melanophores and initially abnormal melanophore morphology early in double mutants as in trpm7 homozygotes supports the inference that trpm7 is epistatic to endzone during melanophore development.

Discussion
Zebrafish enz is selectively required during neural crest development for the terminal differentiation and maintenance of cell morpholgy in late embryonic and larval chromatophores The zebrafish mutant enz was identified based on the reduced size and numbers of neural crest-derived melanophores, xanthophores and iridiphores. Analysis of the development of other neural crest derivatives indicates that the enz mutation selectively affects chromatophore sublineages, as non-chromatophore derivatives, based on diagnostic gene expression, are morphologically indistinguishable from wildtype embryos. Whether all nonchromatophore derivatives are functionally normal in enz mutants, however, is not known. We observed that melanophores begin to differentiate in enz mutant embryos at approximately 25 hpf, similar to wild-type siblings. However, these melanophores were pale compared to wild-type melanophores, and later transitioned to a punctate morphology due to a change in cell shape as opposed to only a change in melanosome distribution. Wild-type melanophores, on the other hand, remain highly dendritic. Iridiphores and xanthophores in enz mutants also begin to differentiate normally in enz mutants but subsequently, like melanophores, appear smaller or less dendritic, respectively, and less chromatic than the same cell types in wild-type embryos. In addition, expression of genes diagnostic of early neural crest cells, melanoblasts, xanthoblasts and all chromatoblasts was normal through 24 hpf. Thus, the specification, migration and early differentiation of neural crest-derived chromatophores is unaffected by the enz mutation. Double mutant analysis of enz with trpm7 b508 , a mutation previously demonstrated to affect melanophores just prior to overt differentiation [67], revealed that embryos homozygous for mutations in both genes resembled trpm7 b508 single mutants with respect to the melanophore phenotype. This places enz genetically downstream of trpm7 b508 and, together with phenotypic characterization, suggests enz function is required for the terminal differentiation of chromatophores, the establishment of correct cell numbers and subsequently for the maintenance of chromatophore cell morphology.

Potential functions of enz in zebrafish chromatophore development
The mechanism by which disruption of enz function affects chromatophore development is unclear. We have determined that it is unlikely that enz is required prior to overt differentiation. The fact that the numbers of differentiated chromatophores in enz mutants are reduced compared to wild-type embryos raises several possibilities. It is formally possible that incompletely differentiated chromatophores persist abnormally in enz mutants. That we have not detected comparatively more cells expressing chromatophore precursor markers and not pigment in enz mutants during embryogenesis argues against but does not rule out this possibility. On the other hand, embryonic chromatophores are believed to be mitotically active, although the proliferation rates of zebrafish chromatophores have not been determined. It is possible then, that enz mutations reduce chromatophore proliferation resulting in a smaller population of pigment cells. Conversely, enz mutations could result in an increase in chromatophore cell death that would result in reduced numbers of cells. Although TUNEL analysis at sequential stages of development between 16hpf and 30 hpf did not reveal detectable differences in melanophore cell death between mutant and wild type embryos (data not shown), there are several factors that prevent conclusions to be drawn from the experiments. First, TUNEL, as well as most other reliable methods to detect cell death in vivo are only useful in the instantaneous detection of apoptotic cells. Thus, if cell death were to occur at a low rate over an extended period, differences in cell death rate would be very difficult to detect. This would be especially true if, as is the case in enz mutants, the cell types of interest are only slightly reduced in number compared to control embryos. Nevertheless, it is potentially significant to note that multiple studies have shown that chromatophores differentiate and undergo apoptosis in response to factors that control chromatophore motile responses, that is, the redistribution of pigment organelles within chromatophores (see Sugimoto, 2002). Further, these factors are likely to utilize shared intracellular signaling pathways in different pigment cell types. Thus, the chromatophore phenotypic consequences and cell autonomous action of enz are at least consistent with a role for enz in such a pathway and the regulation of differentiation and apoptosis via such a pathway. Finally, in addition to the molecular identification of enz it will also be critical to determine whether any of the enz alleles we have identified result in loss of function. While the majority of chromatophores develop in enz mutants, the same situation is observed in most mutants and morphants affecting chromatophore development [4,14,15,70,71]. Further, the majority if not all chromatophores in more severe enz alleles exhibit altered morphology. Thus, it is possible that one or more of the enz alleles we have identified result in loss of function, but this will only be possible to ascertain by cloning the enz locus. While the vast majority of homozygous enz larvae die after 2-3 weeks of development, a minority of individuals does survive to adulthood but were found to be significantly smaller than wild-type siblings. Because neural crest-derived chromatophores are not required for viability and are not thought to regulate growth [6,7,55,72], enz must function in other cells required for growth and viability. Interestingly, there is abundant evidence demonstrating roles for pituitary hormone signaling systems in chromatophore development and growth control [42,46,73]. Notably, hormones known to regulate chromatophore morphology in teleosts specifically have been shown in mammals and fish to affect body weight and composition, as well [46]. Because our data indicate that enz acts cell autonomously, enz would be predicted to be a downstream effector if it acts in such a hormonal signaling pathway.
It is also potentially informative to note that recent reports have described mutants and morphants with live phenotypes similar to enz that disrupt genes involved in organellar biosynthesis and transport [70,71,74]. Although more in depth investigations into the nature of these pigmentation defects have not been described to date, such as whether organelle transport, chromatophore cell morpholgy or survival and proliferation are affected, morpholinomediated knockdown of zebrafish ATPase 6 subunit v0c (atp6v0c) and vacuolar protein sorting protein 18 (vps18) result in punctate melanophore appearance, apparent loss of xanthophores and reduced iridiphore numbers [71]. Similar phenotypes were observed in mutants for a variety of ATP synthase and vacuolar protein sorting genes generated in a viral insertion-based mutagenesis screen [70,74]. We noted a predicted vps35 ortholog assigned to chromosome 7 (http://www.sanger.ac.uk/), however, sequencing of the four enz alleles reported here did not reveal any molecular lesions in the vps35 coding sequence. This does not rule out the possibility that enz encodes vps35, as lesions may be present in non-coding regulatory regions of this gene in enz mutants. Rawls and colleagues have identified a mutation linked to chromosome 7 and SSLP marker z10441 that phenotypically resembles enz [17]. However, enz mutants complement this mutation. Further, the genomic region in which the enz gene is located contains several other genes involved in organellar function that may be good candidates for the enz locus (http://zfin.org). Thus, it is also possible that the enz locus encodes a protein required for normal formation or activity of cellular organelles. Specifically in the case of enz, such a gene would have to regulate chromatophore cell morphology instead of or in addition to organelle synthesis and distribution. The characterization of enz mutant embryos and larvae presented here may provide clues to the mechanisms behind phenotypes described for organellar biosynthesis mutants and morphants described previously. In any case, the potential relevance of the speculation discussed here will be resolved with the molecular identification of the enz locus.
Coordinated control of the late development of chromatophore sublineages Although many loci have been identified that are required for crest-derived chromatophore development, relatively few of them affect all three chromatophore cell types found in zebrafish [14,16,17,70]. Still fewer of these affect all chromatophore sublineages in the same way [14,70,74]. At the same time, other neural crest derivatives appear to develop normally in enz homozygotes. Together with our finding that the enz mutation acts cell autonomously with respect to melanophore development, this suggests that the enz locus does not encode a gene generally required by neural crest-derived cells, but rather a molecule intrinsic to chromatophores themselves. Our results strongly suggest that the differentiation and maintenance of cell morphology of all three zebrafish chromatophore cell types depends on enz and are thus at least partially coordinately regulated during terminal differentiation. However, since the development of all three chromatophore lineages is unaffected in enz mutants until late in their embryonic development and many mutants and identified genes selectively affect individual or subsets of chromatophore types [14,16,17,70,74], the regulation of earlier stages of chromatophore development is at least partially genetically distinct. Nevertheless, it will be interesting to determine when and to what degree the development of zebrafish chromatophore sublineages is regulated by common and distinct genes. The great numbers of zebrafish pigment mutants, including enz, will provide important contributions to ultimately determining the genetic regulation of chromatophore development and leading to potential insights into clinically relevant conditions in humans involving pigment cells. Further, many of these mutants are likely to provide valuable insights into the genetic regulation of color changes and light adaptation in fish and other organisms.

Zebrafish
Adult zebrafish and embryos were maintained in the Ohio State University zebrafish facility. Adults and embryos were reared at 28.5uC and embryos were staged based on morphological criteria, according to Kimmel et al., 1995. Mutant lines were maintained in the AB* and WIK backgrounds. Homozygous mutant embryos and wild-type siblings were obtained by crossing heterozygous carriers.

Cell size quantification and analysis
Melanophores in 2 days post-fertilization (dpf) wild-type and enz embryos were imaged on a Zeiss Axioplan compound microscope. Individual cells were outlined and areas calculated using Zeiss Axioplan software with appropriate scalings. Iridiphores in 6 dpf larvae were imaged on a Leica dissecting microscope. Individual cells were outlined and areas calculated using Spot Advanced software with appropriate scalings. A minimum of 3 melanophores or iridiphores in at least 3 wild-type and 3 enz mutant individuals were utilized for each set of measurements. Standard errors of mean were calculated for wild-type and enz cell areas, and mean numbers were also subjected to a one-tailed t-test to determine whether differences observed between wild-type and enz chromatophore areas were statistically significant.

Cell counts and statistics
Melanophores in 4 dpf larvae were counted in dorsal stripes. To better visualize distinct melanophores, wild-type and enz mutant larvae were placed in epinephrine (10 mg/ml) at 4 dpf for ,10 minutes, that results in redistribution of melanosomes to the center of the cell body in wild-type melanophores (Johnson 1995, Rawls 2000. Larvae were fixed in 4% paraformaldehyde at 4uC overnight, rinsed and stored in 1:1 PBS:glycerol. Larvae were then deyolked, mounted on single bridge cover slips, and viewed on a Zeiss Axioplan microscope. Melanophores in the dorsal stripes were counted from somite 5 to somite 14. Iridiphores in 6 dpf larvae were counted in the dorsal and ventral stripes. Live fish were viewed under incident light using a Zeiss dissecting microscope. Iridiphores were counted posterior of the hindbrain in the dorsal stripe, and caudally from the posterior edge of the yolk ball in the ventral stripe. Standard errors of mean were calculated for wild-type and enz cell counts. The mean numbers of cells in wild-type and enz embryos were subjected to a one-tailed T-test to determine whether the decrease in cell numbers in enz mutant embryos compared to wild-type siblings was statistically significant.

In situ hybridization
In situ hybridizations were performed as described by Thisse, et al. (1993) with minor modifications. A detailed protocol will be provided upon request. mitfa and c-ret cDNAs were provided by D. Raible [6,75]. cDNA clones of c-kit and fms were obtained from D. Parichy [7,28]. dct and ednrb1 cDNA clones were provided by R. Kelsh [23,37].

Immunohistochemistry
Antibody labeling was performed as previously described [76]. 7 dpf larvae were cryo-sectioned onto gelatin-subbed slides and stored at 220uC overnight. All neurons were detected with monoclonal antibody 16A11 that recognizes neuron-specific Hu RNA binding proteins [15,76], while DRG neurons and enteric neurons were subsequently identified by position within the larva. 16A11-immunoreactivity was detected using an Alexa red fluorescent secondary antibody (Molecular Probes). Sympathetic neurons were identified by detection of tyrosine hydroxylase (TH) using anti-TH polyclonal antibody (Pel-Freeze, Rogers, AZ). Neural crest-derived sympathetic neurons were subsequently assessed by location of TH + cells within whole-mount wild-type and endzone larvae.

Mosaic analysis
Genetic mosaics were produced using cell transplantation techniques [77]. Donor embryos obtained from AB* or heterozygous (enz +/b431 ) crosses were manually dechorionated and injected at the one to two cell stage with 2-5% lysinated rhodamine dextran (LRD, 10,000 MW, Molecular Probes) in 0.2 M KCl. To facilitate the identification of donor-derived melanophores, we utilized hosts homozygous for the golden b1 mutation, which results in pale melanophores that are distinct from wild-type melanophores [7,61]. Embryos were then allowed to develop to early blastula stages. 10-20 cells were transplanted from LRD-labeled donor embryos into unlabeled host embryos. Donor-host pairs were kept separate and allowed to develop to $32 hpf, then fixed in 4% paraformaldehyde at 4uC overnight. Donors and hosts were classified as either wild-type or mutant based on melanophore phenotype, and subsequently examined using a Zeiss Axioplan microscope with Nomarski optics, and a rhodamine filter to detect donor cells. Dark, punctate melanophores were scored as mutant, while dark, stellate melanophores were considered wildtype. Wildtype R wild-type and mutant R mutant transplants served as controls.
Genetic mapping enz alleles were maintained in the AB* background. For mapping purposes, enz carriers in this background were crossed to a wild-type WIK line, which is polymorphic to AB* [78]. enz carriers in the WIK background were then used to generate parthenogenetic diploid progeny by suppressing the second meiotic division with early pressure [66]. enz embryos and wildtype siblings were identified by live phenotype and used to obtain DNA. enz was initially placed on the zebrafish genomic map based on PCR amplification of simple sequence length polymorphisms (SSLPs) from diploid genomes [62][63][64][65]. enz b431 was initially demonstrated to be linked to SSLP markers near the centromere of chromosome 7 based on the MGH mapping panel (http:// zebrafish.mgh.harvard.edu/zebrafish/index.htm). enz was shown to cosegregate with z10441 (map position 36.7 cM), z11625 (map position 51.1 cM) and z6819 (map position 45.0 cM). Further linkage analysis was performed on enz using haploid genomes, or meioses [62]. This analysis confirms linkage to microsatellite marker z10441, and further defines a tentative interval for the enz locus distal to z6819 and z55887 and proximal to z8252 and z10441.

Author Contributions
Conceived and designed the experiments: BLA. Performed the experiments: BLA GRG. Analyzed the data: BLA GRG. Wrote the paper: BLA. Principal investigator: PH.