Genetic analysis of fin formation in the zebrafish, Danio rerio

In the zebrafish, as in other fishes, larval fins differ from the fins present in the adult by both number and morphology. Larvae possess two pectoral fins (Fig. 1C) and a caudal fin fold (Fig. 1A). Pectoral fins develop from a fin bud that is homologous to the tetrapod limb bud in structure and gene expression (Krauss et al., 1993; Sordino et al., 1995). In contrast to the tetrapod limb bud, proliferation of the fin bud stops soon after its formation and the pectoral fin folds arise from an apical ectodermal ridge like structure.

The caudal fin fold, consisting of a dorsal part, a ventral part posterior to the anus, and an anterior ventral part below the yolk tube (Fig. 1A), develops from a structure very similar to the apical ectodermal ridge of the pectoral fin bud, at approximately 22 hours post fertilization (hpf). This apical ectodermal ridge-like structure is 6-9 cells wide and consists of a layer of epiderm covered by periderm. Fin epithelium is formed by folding of this primordium onto itself (Dane and Tucker, 1985).

At the hatching period (Kimmel et al., 1995), the pectoral and the caudal fin folds each consist of two layers of epiderm covered by periderm. In the space between the two layers of epiderm collageneous fibers, the so-called actinotrichia are present (Figs 1D, 2A). The epiderm is probably responsible for the formation of these fibers (Bouvet, 1974). Additional extra-cellular fibers are present, which cross the subepidermal space, and might be important for the maintenance of the folded structure (Dane and Tucker, 1985). While the pectoral fins are maintained throughout development, the single caudal fin fold is replaced by three distinct adult fins, the unpaired anal, tail and dorsal fins, during the first four weeks of post embryonic development (Fig. 1B). In the fin fold, at the positions where these fins will develop, mineralized segmented finrays are formed, the lepidotrichia. Lepidotrichia are part of the dermal skeleton of the fish, in contrast to the radial bones which are part of the endoskeleton (Fig. 1E,F). The larval fin fold between these definitive fins disappears. In addition, the paired pelvic fins, the homologues of the posterior tetrapod limbs, develop. The embryonic origin of the adult fin mesenchyme and the cells that form the lepidotrichia is not clear. It has been suggested that they derive from neural crest in the unpaired fins and from mesoderm in the paired fins (Smith et al., 1994).

In our screen we scored for larval fin and skin defects (Haffter et al., 1996). 118 mutants were identified in which fin development is affected in the embryo; in addition we isolated several dominant and recessive mutations affecting the development of the adult fin (Table 1). Ten mutants were identified that have defects in the formation of the embryonic skin. The screen for embryonic fin or skin mutants was performed between 24 and 48 hours of development using a 20-80× magnification on a dissecting scope (for details see Materials and Methods). The aim of this study is to present a brief overview of all skin and fin mutants, isolated in our mutational screen. For this reason we have also included short descriptions of mutants that have additional phenotypes and are described elsewhere in more detail (for references, see Table 1).

Fish maintenance and crosses were done as described in Mullins et al. (1994). For photography the following mutant alleles were used: nel^tq207, rfl^tr240, pif^tm95b, fyd^tj2b and bla^ta90 homozygotes, frf^tm317a/frf^tf5, lep^tq236/lep^tj222 transheterozygotes, uki^tc256d, dre^tm146d, ika^tm127c, krm^tc227d, dak^tw25e, box^te242, smp^td11b, ddf^ta50ames^tm105 homozygotes, pgy^ty40, alf^dty86, lof^dt2 and wan^dty127 heterozygotes.

Fish of the F₂ generation (families; for details, see Haffter et al., 1996a) were set up for spawning. These F₂ fish were visually inspected for fin and pigmentation morphology. Dominant mutations should affect 50% of the fish in a given F₂ family.

To identify zygotic recessive fin and skin mutations, embryos from individual F₂ sibling matings were inspected between 48 and 60 hpf. Using a DRC Zeiss dissecting microscope equipped with 0.8×, 2×, 4× and 8× magnification, the morphology of the caudal fin fold and the pectoral fin folds were examined in 12 embryos from each clutch (for details see Haffter et al., 1996).

In situ hybridization was done as described by Hammerschmidt et al. (1996). Skeletal staining was done as described by van Eeden et al. (1996).

Photographs of live and stained embryos were taken with a Zeiss axiophot microscope, on Kodak ektachrome 64T or 160T slide film. Pictures were scanned using a Nikon coolscan slide scanner, and composite images were made using the Adobe Photoshop software package on a Macintosh computer.

Mutations causing fin edema or fin degeneration were frequently identified in our screen. We kept such mutants if at least some mutant larvae had an air-filled swimbladder (43 out of an estimated 70). These 43 mutants can be grouped into six complementation groups, which can be subdivided into two phenotypic subgroups.

Mutants for the pinfin (pif), nagel (nel), fransen (fra) or blasen (bla) genes show bubbly fin folds at 24 hpf. Fluid accumulates in the fin folds, probably between the two epidermal layers. In contrast, mutants for rafels (rfl) and frayed (fyd) can be identified first at 48 hpf by the irregular fin fold edges and the presence of rounded-up cells in the fin folds. In the strongest alleles that we kept, the fin folds collapsed, and around 120 hpf all fin folds were severely reduced. An overview of the pheno-types is given in Fig. 2A-F. Mutants for all six genes are adultviable and fertile. Mutant adults have a subtle phenotype: fin rays are present and are of normal length, but their number is variably reduced. The degree of reduction varies even between siblings homozygous for the same mutant allele. For example, in one severely affected pif mutant fish only 8 instead of the average 19 segmented lepidotrichia were present in the tail fin. In such cases the tail fin is bar-shaped instead of triangular, leaving the periodicity of the fin rays unchanged. In contrast one other homozygous pif mutant from the same family had a normal number of lepidotrichia in the tail fin.

Four complementation groups have been defined that lead to undulations of the fin epithelium. In frilly fins (frf) and microwaved (med) mutants the distance from the base of the fin fold to the border is slightly reduced. The fin epithelium has curtain-like undulations (Fig. 2G,H), especially at the tail tip. Homozygous mutants for frf and med are viable. Med mutants do not show an adult phenotype, whereas frf mutants do: fins of these fish are reduced in length; adults are 30-50% shorter and have a kink at the head-trunk boundary.

Mutations in dreumes (dre), leprechaun (lep) and ukkie (uki) have a subtle effect on the caudal fin fold, causing slight undulations of the fin epithelium. The larval pectoral fins are most strongly affected in these mutants and show a clear increase in size (Fig. 3A,B), which at 120 hpf leads to an unusual fold in the dorsal part of the pectoral fins. These mutants also have a smaller pupil (Heisenberg et al., 1996) and an abnormally shaped ear in the embryo (Whitfield et al., 1996). Homozygous larvae for all three genes survive to adulthood, but remain small in size. In homozygous uki adults the number of lepidotrichia was found to be slightly increased (1-6 per fin, except the anal fin). Adult dre and lep have not been analyzed in detail. In both uki and dre, but not in lep, adults the anal fin is absent (Fig. 4A,B); it is not clear how this phenotype is related to the larval fin phenotype.

Three further genes have been identified that are required for all larval fins. A mutation in tutu causes deletions of the fin folds. These deletions are variable in size, and sometimes holes are present in the fin epithelium. Tutu mutants are semiviable and have variably reduced fins. The embryonic phenotype of u-boot (ubo) mutants is similar to rafels; the fin folds are reduced, slightly thicker and have irregular borders. Mutants for this gene also lack the horizontal myoseptum (van Eeden et al., 1996) and have a reduced number of chromatophores (Kelsh et al., 1996). The fin folds of moonshine (mon) mutants also have irregular edges. In addition this mutation leads to defects in blood formation and to an abnormal iridophore pigmentation of the larvae (Kelsh et al., 1996; Ransom et al., 1996). A phenotypic comparison of all fin mutants is given in Table 2.

Mutations in the ikarus (ika), sonic-you (syu) and chameleon (con) genes cause a variable reduction in the size of the pectoral fins. In all three mutants the degree of reduction frequently varies even between the left and the right fin fold in one embryo (Fig. 3C-E). Syu and con embryos in addition lack the horizontal myoseptum and have neural tube defects (Brand et al., 1996; van Eeden et al., 1996). Ika mutants do not have any other obvious defects and survive to adulthood. In ika adults the pectoral fin phenotype is also variable, ranging from absence to apparently normal pectoral fins (Fig. 4C,D,E). All other fins, including the paired pelvic fins, are normal in homozygotes.

Mutations in two genes, dackel and boxer, lead to a complex but very similar phenotype in the embryo. Mutants for the dackel gene (dak) (Fig. 3F) show a stronger phenotype in all aspects. Homozygotes form pectoral fin buds, but the pectoral fin does not develop further. As a result the pectoral fin fold is absent at 120 hpf. In addition to the fin phenotype, dak mutants have defects in the projection of retinal axons (Karlstrom et al., 1996; Trowe et al., 1996), in the formation of branchial arches (Schilling et al., 1996) and in ear morphology (Whitfield et al., 1996). Whole-mount in situ analysis with sonic hedgehog (shh) (Krauss et al., 1993), which is expressed in the posterior part of the fin bud, shows that shh expression is initiated but not maintained in dak mutants (Fig. 5). Sonic hedgehog encodes a secreted protein, which has been identified as a likely mediator of polarizing activity within the vertebrate limb bud (for a review, see Tabin, 1995). Mutations in the second gene, called boxer (box) (Fig. 3G), cause similar but somewhat weaker phenotypes. In this mutant the proximal part of the pectoral fins appears normal but the distal fin fold is absent, and rounded up cells are present on the tip of the fin. Both mutations are lethal, probably due to lack of an air-filled swimbladder.

krom mutants have curled pectoral fins at 72 hpf (Fig. 3H). In most homozygous adult krom fish the part of both pectoral and pelvic fins that contains the lepidotrichia is absent. The three unpaired fins appear unaffected. Sometimes this adult phenotype was not fully penetrant; in such cases single pectoral or pelvic fins were present unilaterally.

The stomp (sto) gene is defined by a single allele. The mutant phenotype of this allele is variable in expressivity and only partially penetrant (Fig. 3I). In smp mutant embryos the pectoral fin epithelium shows weak degeneration, whereas the epithelium of the caudal fin fold is normal. No obvious phenotype is visible in smp homozygous adult fish.

14 additional mutants have been identified where outgrowth of both the pectoral fins and the branchial arches is reduced. Homozygous mutants do not have any tissue in front of their eyes. Complementation analysis in this phenotypic group has defined at least five genes: hammerhead (ham), pekinese (pek), head on (hen), jellyfish (jef) and shallow chin (son), and six mutations are unresolved. These genes are probably required for proper cartilage differentiation (Piotrowski et al., 1996).

In our screen, two groups of mutations were isolated that affect the basic body plan of the embryo. They display dorsalization or ventralization phenotypes. The ventral tail fin is very sensitive to such mutations, and is therefore thought to be the ventralmost stucture of the embryo. These mutants are described in more detail elsewhere (Mullins et al., 1996; Hammerschmidt et al., 1996).

The dorsalizing mutants swirl (swr), somitabun (sbn), piggytail (pgy), minifin (mfn) and lost-a-fin (laf) show deletions of the ventral tail fin. For instance, heterozygous pgy mothers cause a dominant phenotype in heterozygous progeny: these embryos do not have the posterior part of ventral tail fin (Fig. 6A,B). Some heterozygotes survive to adulthood; they are often shorter than wild-type fish and lack the tail fin and sometimes part of the anal fin (Mullins et al., 1996).

Mutations in the ventralizing mercedes (mes) and dino (dio) genes result in a duplication or multiplication of the ventral tail fin, respectively (Fig. 6C,D) (Hammerschmidt et al., 1996). Mes homozygous embryos survive to adulthood and have partially duplicated tail and anal fins.

Three dominant mutations affecting fin development between larval and adult stages have been identified. Another long fin (alf, Fig. 4G) was found to cause formation of longer fins, similar to the previously described dominant long fin (lof) mutation (Fig. 4H) (Tresnake, 1981). In lof and alf heterozygotes fin rays are longer, but alf heterozygotes show in addition irregular segmentation and irregular bifurcation. A dominant mutation in wanda (wan, Fig. 4I), causes severe reduction of the fins in heterozygous fish. Distal fin rays, if present, appear unsegmented. In addition the fish have an abnormal body shape and an abnormal pigment pattern (Haffter, unpublished observations). Homozygous adults for the stein und bein (sub) gene lack the pelvic fins in approximately 75% of the cases, and some of these fish lack the unpaired dorsal fin as well (Fig. 4J). This mutation was originally identified by the variable absence of otoliths in homozygous embryos. At present it is unclear if the two phenotypes are caused by a single mutation or by two independent, linked mutations. We have not identified any crossover events in 25 meioses. In total we isolated viable mutations in three genes, which result in defects in the paired fins of the adult. ikarus affects the adult pectoral fins, stein und bein the pelvic fins and krom affects both.

Homozygous larvae for the finless (fls, Fig. 4K) gene were only recently identified among the progeny of another mutant stock. In fls embryos no obvious fin defects are visible, while fls adults lack all fins.

Although the analysis of the fin mutants that we have isolated is in an early phase, comparison of the different types of mutants enables some conclusions to be drawn about the process that leads to the formation of fins. We did not find any mutants that affect only differentiation of the caudal fin fold without also affecting differentiation of the larval pectoral fins, indicating that the process that leads to the formation of the folded fin epithelium is similar in the tail fin and the pectoral fin.

Most of the fin mutations that lead to a degeneration of all fin folds of the embryo, often cause surprisingly mild pheno-types in the fins of the adult. Thus, proper differentiation of the larval fin fold is not a prerequisite to forming adult fins. It appears that a different set of genes function during larval stages to replace the embryonic fin folds. In the larval pectoral fin a set of additional genes is required, defined by the mutations that specifically affect pectoral fins.

The finless mutation is an example of the reverse scenario: no effects are seen in the larval fin fold, whereas adults lack all fins. This substantiates the idea that the differentiation processes forming the larval fin fold and adult fins are largely independent.

In contrast, in dorsalizing or ventralizing mutants, such as swr, pgy or mes, the deletion or duplication of the larval caudal ventral fin fold is a consequence of an abnormal specification process. In dorsalizing mutants, the absence of specification is not compensated in the adult and these fish lack the ventral tail fin. Thus, at least in the ventral part of the tail, adult fins require the same tissue specification as the larval fin fold.

Dackel mutants show initiation of sonic hedgehog (shh) expression in the finbud, but fail to maintain it. shh is thought to be required for specifying the anteroposterior axis in the vertebrate limb (Riddle et al., 1993). How the phenotype that we observe in dak mutant embryos relates to the failure to maintain shh expression is not clear, since expression of shh is dependent on the expression of other patterning genes in the fin bud (Yang and Niswander, 1995). Thus, reduced expression of ssh in the later stages might reflect defects in other patterning genes required for proper ssh expression. In mouse there is evidence that the genes required for the patterning along the three axes of the tetrapod limb depend upon each other. For example, targeted disruption of the wnt-7a gene leads, in addition to dorsoventral patterning defects, to a reduction of the expression of shh and fgf-4 (Parr and McMahon, 1995). shh might play a role in anteroposterior patterning, and fgf-4 might be required for proximodistal patterning of the limb bud.

The boxer phenotype is similar to, but weaker than the dackel phenotype in all aspects. In boxer mutants only the most distal part of the larval pectoral fin seems to be affected, while the phenotypes of ika, syu and con mutants are best described as an overall reduction of the size of these fins without obvious differential defects along the proximodistal axis. shh expression in the pectoral fin buds appears normal in boxer mutants at 28 hpf. Since the defect in box embryos is restricted to the more distal parts of the fins, shh expression might be affected during later stages, which we did not examine. Preliminary analysis of syu and ika mutants at 28 hpf did not reveal any obvious defects in shh expression.

Might one of the mutants presented here be the result of a mutation in the shh gene itself? Chamaeleon and sonic you are amongst the most likely candidates, since they also seem to be involved in the process of patterning the neural tube and somites (Brand et al., 1996; van Eeden et al., 1996). All these processes are thought to be mediated by shh. Testing these mutations for linkage to the shh gene will soon answer this question.

The fin mutants described in this study will be of great value for studying various aspects of fin development. Those lacking the full complement of adult fins can be used to study the transition from an larval fin fold the to mature fin. Mutations affecting the pectoral fins buds can contribute to a genetic dissection of the steps by which pattern emerges in the fin/limb buds.

Mutations in least four genes, goosepimples (gsp), dandruff (ddf), bouillabaisse (bob) and penner (pen), cause defects in the embryonic skin (Table 3). Ddf, gsp and bob mutant embryos have similar phenotypes, while the pen phenotype is more subtle.

Mutant ddf embryos can already be recognized at the 12-somite stage by their abnormal head shape and the presence of rounded-up cells in the head region (Fig. 7A,B). At the pharyngula stage, loose cells can be seen floating around in the chorion and the entire surface of the embryo is covered with rounded-up cells. In addition to this skin defect, the head-trunk angle in mutant ddf embryos is abnormal. During the pharyngula stage, the head of wild-type embryos is curved around the yolk, but becomes straightened within the next 48 hours. Ddf mutant embryos, in contrast, still have a curved body axis at 72 hpf. Eventually about half of the embryos recover and develop an air-filled swimbladder (Fig. 7C,D). Homozygous ddf adults do not have an obvious phenotype.

Goosepimples mutants have a similar phenotype and like ddf, form a swimbladder. However, these homozygotes die within 2 weeks. bob mutants start to show general degeneration at the early pharyngula period and are completely lysed at 60 hpf.

Penner (pen) mutants can first be recognized at 96 hpf by an accumulation of round cells, predominantly in a region below the branchial arches and at the base of the pectoral fins (data not shown). Homozygous pen mutants die during larval stages.