Eda/Edar signaling guides fin ray formation with preceding osteoblast differentiation, as revealed by analyses of the medaka all-fin less mutant afl



Background: Ectodysplasin (Eda) signaling is essential for the morphogenesis of several ectodermal appendages. Results: Here, we report a medaka mutant, all-fin less (afl), which has a nonsense mutation in its eda gene. The adult afl fish displayed various abnormalities of its dermal skeleton, such as short and twisted fin rays, missing and abnormally shaped scales and teeth, and skull deformation. Focusing on the developing fin rays in the caudal region of afl larvae, we found that the fin rays did not elongate; although the initial formation of fin rays proceeded normally. Additionally, eda expression was lost, and the expression pattern of edar, the gene for the receptor of Eda, was different from wild-type one. In vivo imaging of the double-transgenic medaka expressing enhanced green fluorescent protein under control of the edar promoter and DsRed under control of the osterix promoter revealed that edar expression preceded that of osterix and that the edar-expressing cells migrated in the direction of fin ray elongation, indicating that the Eda/Edar signaling event precedes osteoblast differentiation. Conclusions: Our findings provide evidence that Eda signaling accompanied with the binding of Eda to Edar are essential for fin ray formation guided by cell migration. Developmental Dynamics 243:765–777, 2014. © 2014 Wiley Periodicals, Inc.


Ectodysplasin (Eda) signaling is crucial for the morphogenesis of ectodermal appendages, such as hair, mammary glands, teeth, feathers, and scales; and it is highly conserved from teleosts to mammals. The Eda pathway includes the ligand Ectodysplasin A1 (Eda), its receptor, Edar, and the adapter molecule Edaradd (Mikkola and Thesleff, 2003). Mutations in any of the genes encoding these proteins lead to a human disease called hypohidrotic ectodermal dysplasia (HED), which is typically characterized by missing and abnormally shaped teeth, sparse hair, absent or reduced sweating, and defects in several other glands (Mikkola, 2009). Mice deficient in Eda A1 (tabby) and Edar (downless) have been shown to be phenotypically similar to HED patients (Mikkola, 2008). Although these appendages vary considerably in form and function, their early stages of development are notably similar both morphologically and molecularly. The first morphological sign of embryonic development is a localized thickening in the surface epithelium that subsequently invaginates into the underlying mesenchyme to form a placode (Hardy, 1992; Cobourne and Sharpe, 2003; Veltmaat et al., 2003; Mikkola and Millar, 2006). During initial morphogenesis, Eda/Edar signaling, which regulates the function of the epithelial signaling centers, is associated with epithelial mesenchymal interactions (Laurikkala et al., 2001, 2002). Moreover, in medaka fish and zebrafish, the lack of the Eda pathway causes similar defects showing mainly reduced numbers of scales and teeth (Kondo et al., 2001; Harris et al., 2008), in which the Eda signaling organizes epidermal cells to act as signaling centers of the scale epidermal placodes (Harris et al., 2008). Regarding fin ray formation, zebrafish mutants defective in Eda signaling display abnormal fin rays; however, the molecular and cellular mechanisms associated with initial fin ray formation are not well understood.

Fin rays are categorized as being a part of the dermal skeleton by the morphological analyses and evolutional affinities, just like scales and teeth (Sire and Huysseune, 2003). The fin ray of the adult zebrafish is an excellent model for studies on vertebrate bone regeneration. The fin rays are covered by epidermis, and the space between the two hemirays is filled with fibroblast-like cells, in which blood vessels, nerves, and pigment cells are embedded (Knopf et al., 2011). A single layer of osteoblasts situated on the inner and outer surface of the hemirays forms bones without passage through a cartilaginous precursor (Laforest et al., 1998; Knopf et al., 2011). Each ray ends distally with a row of rigid but unmineralized elastoidin fibrils named actinotrichia, which internally line each hemiray (Laforest et al., 1998). During early fin ray development, these components surrounding the fin rays are completely formed (Laforest et al., 1998; Durán et al., 2011). Several factors important for bone formation in human and other mammals have been identified also in zebrafish and medaka fin rays. During zebrafish fin regeneration, bone morphogenetic protein 2b (bmp2b) is involved in the induction of dermal bone differentiation (Quint et al., 2002); and under the control of Bmp signaling, the crucial osteogenic transcription factor runx2 is activated (Smith et al., 2006). Expression analysis in zebrafish larvae shows that runx2 is required during the early stages of commitment and acts on osterix transcription (Li et al., 2009). Osterix is a zinc-finger transcription factor that plays important regulatory roles during the differentiation of preosteoblasts into mature osteoblasts. In medaka, osterix is expressed in early osteoblasts preceding bone mineralization (Renn and Winkler, 2009). These previous data suggest that the mechanistic principle common among all other vertebrates for bone formation also applies to fin ray formation.

Several mouse studies indicate that the NF-κB pathway is a major transducer of Eda signaling. Assessment of NF-κB activity by use of a transgenic reporter has revealed that in tabby and downless mice, the decreased Eda signaling in these mice effectively leads to a severe decrease in NF-κB activation (Schmidt-Ullrich et al., 2001, 2006).

Here, we describe the role of Eda signaling by means of NF-κB for fin ray formation in medaka. We identified a medaka mutant, all-fin less (afl), that has a nonsense mutation in the medaka orthologue of eda. This mutation caused abnormalities of the dermal skeleton, such as short and twisted fin rays, missing and abnormally shaped scales and teeth, and skull deformation. Focusing on the developing fin rays in the caudal region of afl larvae, we found that these rays did not elongate, even though their initial formation occurred normally. Furthermore, eda expression was lost; and the expression pattern of edar was different from that of the wild-type one. We also established double-transgenic medaka expressing enhanced green fluorescent protein (EGFP) under control of the edar promoter and DsRed under control of the osterix promoter, and found that edar expression preceded that of osterix. Moreover, edar-expressing cells migrated in the direction of fin ray elongation. Taken together, our findings provide evidence that Eda signaling initiated by the binding of Eda to Edar was essential for fin ray formation guided by cell migration.


afl Fish Exhibit Defects in Adult Dermal Skeletal Formation

To investigate osteogenesis in the vertebrate medaka, we performed a medium-scale screening of medaka mutants obtained by ENU mutagenesis, and isolated a medaka mutant that showed abnormal caudal fin rays at the larval stages. The homozygous mutant was not lethal during embryogenesis, and thus survived to become an adult fish. The homozygous mutant was physiologically, but not physically, fertile. The adult wild-type medaka has three median fins: dorsal, anal, and caudal and has two paired fins: pelvic and pectoral (Fig. 1A). In contrast, as the mutant displayed defects in all of these fins (Fig. 1B), we named this medaka mutant all-fin less (afl).

Figure 1.

Fin phenotypes of afl fish in the adult stage. Wild-type (A,C,E,G,I) and afl fish (B,D,F,H,J). A,B: Lateral view of adult medaka fish at 2 months post fertilization. Wild-type fish have three median fins and two paired fins (A). afl fish have lost all fins (B). C–J: Double staining of bone and cartilage by Alizarin red and Alcian blue. The fin rays, which were stained with red color in each fin, elongate in a straight line (C). In afl fish, the fin rays of the dorsal and anal fin are completely missing (arrows in “D”). Magnified view of the pectoral fin (E,F), the pelvic fin (G,H), and the caudal fin (I,J). The afl fin rays were short without elongation compared with the wild-type ones. Scale bars = 1 mm.

To reveal differences in structures between wild-type and afl fish, we first performed double staining with Alizarin red and Alcian blue, which stains are commonly used to detect calcified bone and cartilage, respectively. The three median fins consist of the endoskeletal structures that support the fins in the proximal region, and more specifically support the radials of the dorsal and anal fins, the hypurals of the caudal fin, and the exoskeletal fin rays. In afl fish, these endoskeletons were normally formed; however, the fin rays of the dorsal and anal fins were absent (Fig. 1C,D arrows). As to the pectoral and pelvic fins, the radial bones supporting these fins in the proximal region were normally formed; but the fin rays were almost lost, and small bones were observed at the proximal region of the rays (Fig. 1E–H). The caudal fin rays were very short and twisted (Fig. 1I,J). The caudal fin showed an upswept form, because afl fish swam with movement of the caudal region of their trunk instead of with that of their caudal fin (Fig. 1B,D arrowheads). In addition, defective formation was observed in other parts of the dermal skeleton: skull, teeth in jaws, pharyngeal teeth, and scales. The sagittal suture of the skull was closed tightly at the midline in wild-type fish (Fig. 2A), but afl fish showed misalignment of the suture (Fig. 2B, arrows). The teeth in the jaws (Fig. 2C,D) and the pharyngeal teeth (Fig. 2E,F) in afl fish were larger in size and smaller in number than those in the wild-type. The entire body of the wild-type fish was covered with scales, whereas the mutants showed a partial loss of scales. The remaining scales in these fish were nonuniformly deformed (Fig. 2G,H).

Figure 2.

Other dermal skeletal phenotypes of afl fish. Double staining of bone and cartilage by Alizarin red and Alcian blue in wild-type (A,C,E,G) and afl fish (B,D,F,H). A,B: Sagittal sutures of the skull in wild-type (A) and afl fish (B). Wild-type sagittal sutures have fused normally (A). afl fish have asymmetric sutures that did not meet at the site where they intersect the sagittal suture (arrows in “B”). C,D: Upper jaw in wild-type (C) and afl fish (D). A drastic reduction in the tooth number is noted in afl fish. E,F: Lower pharyngeal dentition in wild-type (E) and afl fish (F). afl fish showed loss of a large number of teeth, and these fish have thin pharyngeal bones. G,H: Scales in wild-type (G) and afl fish (H). Wild-type scales were uniformly-shaped and cover the entire body (G). afl fish showed a partial loss and nonuniform deformation of scales (H). Scale bars = 500 μm.

afl Larvae Show an Abnormality in the Initial Formation of their Caudal Fin Rays

The abnormal phenotypes of fin rays appeared in the early stage of fin ray development. To examine when this abnormal morphogenesis first appeared during embryonic development, we performed Alizarin red staining. At 5 days post fertilization (dpf; stage 38), the wild-type larvae began to form fin rays in the caudal region of the fin fold; and subsequently, mineralization and elongation of the fin rays occurred. Although afl larvae began to form caudal fin rays in the correct position at the normal developmental stage, the elongation rate of each fin ray differed according to its position. In the wild-type larvae, six fin rays were observed at 8 dpf (Fig. 3A). These fin rays were formed in order numbering from 1 to 6; therefore the length of fin ray 1 was the longest of all (Fig. 3A). In afl larvae, on the other hand, the length of fin ray 1 was the shortest of all; although no abnormal order of the six fin rays was found (Fig. 3B). In addition, the fin rays of afl larvae did not become thick like those of the wild-type larvae. Laser confocal microscope imaging revealed that the fibers in the fin fold, which are necessary to keep the fin fold straight, normally extended in both larvae. The fin rays elongated in the same direction as the fibers in both wild-type and afl larvae (Fig. 3C–E).

Figure 3.

Initial formation of caudal fin rays in afl larvae. A,B: Alizarin red-stained caudal fin at 8 dpf. Fin rays were formed in order from 1 to 6. In afl larvae, the fin rays were thinner and shorter than those in the wild-type fin rays. Although afl fin rays are also formed in order from 1 to 6, the number 1 ray was seriously deformed (B). C–E: Laser confocal microscope image of wild-type (C) and afl larvae (D, E). Actinotrichia were arranged radially in wild-type larvae (C). Red lines indicate actinotrichia orientation. In afl larvae, no abnormal actinotrichia were seen (D). Magnified view of caudal region, also showing normal actinotrichia in afl larvae (E). F–I: Cell proliferation assay by BrdU incorporation at 4 dpf (F, G) and 8 dpf (H, I). BrdU-positive cells were detected around fin rays in wild-type larvae at both stages (F,H). In contrast, the number of BrdU-positive cells was reduced in afl larvae (G,I). Scale bars = 250 μm in A-D, F-I; 150 μm in E.

To assess the influence of the mutation in afl larvae on cell proliferation, we performed a bromodeoxyuridine (BrdU) incorporation assay. The signal of green fluorescence shows the proliferating cells (Fig. 3F–I). Green fluorescence-positive cells were observed at the developing caudal region of the wild-type larvae at 4 dpf and 8 dpf (Fig. 3F,H). In the afl larvae, the number of these cells was significantly reduced at both stages (Fig. 3G,I). These results indicate that the afl mutation affected the length and thickness of fin rays coupled with a decrease in cell proliferation.

afl Fish Disrupt the Medaka eda Gene

To identify a genomic mutation in afl fish, we took a positional cloning approach. By using sequence-tagged site (STS) markers, we mapped the mutated genomic position to a 1.5-cM region on linkage group 10 (Fig. 4A). Subsequently we identified a candidate region containing the eda gene by using two closely linked markers (0 recombinations among 272 meioses). We then sequenced reverse transcriptase-polymerase chain reaction (RT-PCR) fragments of the eda gene from wild-type and afl larvae, and found a G-to-T nonsense mutation at the 301st base pair from the N-terminus in the open reading frame of the eda cDNA (Fig. 4B). Using both the alignment of vertebrate Eda amino acid sequences and the 5′ and 3′ rapid amplification of cDNA ends methods, we identified the medaka full-length eda cDNA sequence, which showed 1,095 base pairs encoding a 364 amino acid protein (Fig. 4C). The Eda protein contains four conserved domains: a transmembrane domain; a furin protease recognition sequence, which is necessary for proteolytic processing; a collagen domain, which is required for multimerization of Eda trimers; and a TNF homology domain, which effects specific interaction with Edar. In humans, two isoforms of Eda, Eda-A1 and Eda-A2, have been identified, differing only by two amino acids in the TNF homology domain binding to EDAR and XEDAR, respectively. In the search of the medaka genomic sequence, we found one medaka eda gene encoding a protein with high similarity to Eda-A1. In afl fish, the G-to-T transition introduced a stop codon between the transmembrane domain and the furin protease recognition sequence (Fig. 4C). As a result, the afl fish were expected to produce a truncated Eda protein lacking the main functional domains.

Figure 4.

afl fish have a disrupted medaka eda gene. A: Positional cloning of afl gene. The afl mutation was miotically mapped to a 1.5-cM interval on LG10 1.1 cM (3 recombinants/ 272 meioses) from 13.6 Mb distance marker and 0.37 cM (1recombinants/ 272meioses) from 15.4 Mb distance marker. B: Sequencing of cDNAs isolated from wile-type and homozygous afl larvae. Asterisk indicates the nucleic acid transition from G-to-T, resulting in an amino acid change from glutamine to STOP. C: A schematic of Eda protein in wild-type and afl fish, containing the N-terminal transmembrane domain (TM), furin protease recognition sequence (Fu), collagen domain (Coll), and TNF domain. D–G: Rescue from afl phenotypes by the DNA injection. A schematic of the rescue DNA construct, containing the 3-kb eda promoter, the genomic sequence coding for the Eda protein except for the 6-kb first intron, and crystallin promoter-GFP (D). Rescued construct-treated afl fish recapitulated the normal morphology, forming calcified bone stained with Alizarin red in all their fins (G). The inset in panel “G” shows the expression of GFP in the eyes, thus confirming transgene integration. Scale bars = 1 mm.

eda Is the Gene Responsible for the afl Fish Phenotypes

To confirm that the defects in afl fish were derived from the genomic mutation in the eda gene, we rescued fish from the afl phenotypes by introducing the wild-type eda gene under the control of the genomic eda promoter region into afl fish. To express the wild-type eda in the appropriate position and stage, we designed a DNA rescue construct (Fig. 4D). The construct contained the 3-kb upstream region from the first methionine of the eda gene. To analyze the genotype of the embryo having the rescue DNA construct, we cut off the first intron region of the eda gene, the size of which was approximately 6 kb. We then generated the construct involving the EGFP controlled by the crystalin promoter, which was used for monitoring the transgene integration into the chromosome followed by expression of green fluorescent signals in the eyes. Subsequently, we microinjected the rescue DNA construct into the fertilized eggs of mated heterozygous mutants and picked up embryos with EGFP-positive eyes. After their intercrossing, we obtained the afl homozygous fish harboring the rescue construct. This rescued afl fish showed a normal body structure, which was indistinguishable from that of the wild-type fish (Fig. 4E–G). Skeletal staining with Alizarin red revealed that the skeletal elements of afl fish were normally formed (Fig. 4G). These results indicate that the eda gene was indeed responsible for the afl fish phenotypes.

Caudal Fin Ray Formation Requires Eda/Edar Signal Transmission

To study the effect of Eda signaling on the morphogenesis of fin rays, we examined fin ray formation at the early developing stage of 6 dpf, and performed whole-mount RNA in situ hybridization (WISH) with eda and edar as probes (Fig. 5A–D,K–N). For comparison with the cellular localization of osteoblasts, we also examined the expression patterns of bmp2b, encoding a bone morphogenetic protein, which promotes bone formation by inducing osteoblast differentiation from undifferentiated mesenchymal cells (Fig. 5E,F,O,P), and runx2, which is a transcriptional factor gene essential for osteoblast differentiation and bone formation (Fig. 5G,H,Q,R).

Figure 5.

Expressions of medaka eda, edar, bmp2b, runx2, and shh at 6 dpf. A–J,U: WISH for eda (A,B), edar (C,D), bmp2b (E, F), runx2 (G,H), and shh (I,J). In the wild-type fish, eda expression was detected in every fin ray; and edar expression, along the fin rays (A,C). A schematic drawing shows the positional relationship between eda and edar expressions, and the arrow points to the fin ray 5 region (U). In afl larvae, no eda expression was detected (B), and the edar expression pattern differs from the wild-type one (D). bmp2b and runx2 were expressed along the fin rays in the wild-type larvae (E, G). These expression patterns are similar to the edar expression patterns except in the case of fin ray 5 region (arrows in C, E, G and U). bmp2b and runx2 expressions in afl larvae were reduced (F,H). shh expression was detected along the fin rays in wild-type larvae (I), but not in the afl larvae (J), which is a candidate for direct target gene of the Eda signaling. K–T,V: Cross sections of wild-type expressions. Red staining with PI shows nuclei. A schematic illustration shows these results (V). eda was expressed in mesenchymal cells (K,L), edar and shh were expressed in epithelial cells (M, N, S, T), and bmp2b and runx2 were expressed in the cell cluster next to edar-expressing cells (O–R). Scale bars = 200 μm in A–J; 20 μm in K–T.

The expressions of eda and edar were first detected at the head region in the wild type (data not shown). After that, they were detected at the caudal region of the developing fin fold (Fig. 5A,C). eda expression alternated with that of edar (Fig. 5U). bmp2b and runx2 were expressed in a limited area compared with edar (Fig. 5E,G). edar was expressed in areas from fin ray 1 to 5; in contrast, bmp2b and runx2 were not expressed in the area of fin ray 5 (Fig. 5C,E,G,U, arrows). These results suggest that edar expression preceded that of bmp2b and runx2. On the other hand, in afl larvae, no eda expression was detected (Fig. 5B). edar expression occurred in a narrower area relative to that for wild-type expression (Fig. 5D). bmp2b and runx2 were also expressed in a limited area compared with their wild-type expressions (Fig. 5F,H). Considering that the fin rays of afl larvae were shorter than wild-type ones (Fig. 3B), the defect in eda expression may have caused edar expression to remain in proximal regions and additionally restricted bmp2b and runx2 expressions to a narrow area. For investigating detailed expression patterns in wild-type larvae, cross sections of whole-mount stained tissues were prepared, which sections indicated that the expression of eda was restricted to the mesenchymal cells (Fig. 5K,L), that the expression of edar was restricted to the epithelial cells (Fig. 5M,N), and that the expressions of bmp2b and runx2 were localized around a cell cluster (Fig. 5O–R), which was observed in the sections for eda and edar (Fig. 5K–N). A schematic drawing shows these positional relationships (Fig. 5V).

Previous in vitro studies on mammals have suggested that Sonic hedgehog signaling is a crucial downstream component of the Eda pathway in the salivary gland (Häärä et al., 2011), and also a recombinant shh rescues branching of submandibular glands in the mouse edar mutant (Wells et al., 2010). Therefore, we investigated shh expression patterns. As a result, in wild-type larvae, shh was expressed in fin ray regions from 1 to 4 (Fig. 5I), and the cross sections showed that the shh expression was restricted in the epithelial cells (Fig. 5S,T). The expression patterns were similar to the edar expression, but it was unclear whether these two genes were expressed in the same epithelial cells. In afl larvae, no shh expression was detected (Fig. 5J). We also examined expressions of other candidate genes as target genes of the Eda signaling, i.e., dkk1, ccn2/ctgf, and follistatin; however, their expression patterns were not significantly different between wild-type and afl larvae (Fig. 6A–F). These results suggest that shh might be a crucial target of Eda signaling in the developing fin rays.

Figure 6.

Expression of medaka dkk1, ccn2/ctgf, and follistatin. A–F: The expressions of dkk1 (A, B), ccn2/ctgf (C, D), and follistatin (E, F), which are candidates for target genes of the Eda signaling, were not different between wild-type and afl larvae. Scale bars = 200 μm.

Eda/Edar Signaling by means of NF-κB Activation Regulates Fin Ray Elongation and Controls Cell Migration

As the Eda protein has a furin protease recognition sequence, shedding occurs through proteolytic cleavage that releases its extracellular Edar-binding region (Elomaa et al., 2001). Binding of Eda to Edar causes NF-κB activation, and activated NF-κB subsequently induces transcription of downstream target genes in the nucleus. We next investigated the role of Eda signaling during fin ray formation by performing edar expression analysis and established a transgenic medaka expressing EGFP under the control of the edar promoter.

edar expression was detected before fin ray mineralization at 4 dpf in both wild-type and afl larvae (Fig. 7A,B). In wild-type larvae, edar was detected in the fin ray 1 region and subsequently in order from fin ray 1 region to 7 region, which was the same order as for fin ray formation (Figs. 3A, 7A,C,E), and as the developing stages progressed, edar expression became progressively limited to the distal region in the fin fold (Fig. 7C,E). In afl larvae, although no alteration of initial expression regions and expression order from 1 to 7 was observed, each expression remained at the proximal region (Fig. 7D,F). EGFP of the edar transgenic line was detected in the same region as edar mRNA expression during fin ray formation. In vivo bone staining with Alizarin complexone (ALC) revealed that the EGFP-positive cells were located in the uncalcified area of the distal tip of the fin rays (Fig. 7G,I). In afl larvae, although EGFP-positive cells were located in the uncalcified area of the distal tip of fin rays as in the wild-type larvae, their fin rays did not elongate (Fig. 7H,J). These results suggest that while the trigger of edar expression was independent of the presence or absence of eda, movement of the edar-positive area was dependent on eda.

Figure 7.

edar expression precedes fin ray ossification. A–F: Expressions of edar during fin ray formation at 4, 6, and 8 dpf in wild-type larvae (A,C,E) and afl larvae (B,D,F). In the wild-type larvae, edar was expressed in the fin ray 1 and 2 region at 4 dpf (A), and subsequently the expression area expanded to the fin ray 5 region at 6 dpf (C) and to the 7 region at 8 dpf (E). These edar expressions seemed to have moved toward the distal tip of fin rays from 4 dpf to 8 dpf. In contrast, afl expressions were detected in the fin ray 1 and 2 region at 4 dpf (B); however, they remained at the proximal region of each fin ray at 6 and 8 dpf (D, F). G–J: In vivo analysis of ossification in edar-EGFP transgenic line. edar-EGFP transgenic wild-type (G, I) and afl larvae (H, J) were stained with Alizarin complexone. EGFP expressions were observed at the distal tip of fin rays in both wild-type (G) and afl larvae (H). Highly magnified views of the white-boxed region in “G” and “H” show that edar expression preceded the mineralization of fin rays in both wild-type (I) and afl larvae (J). Scale bars = 200 μm in A–H and 50 μm in I, J.

To examine whether the defective eda resulted in inhibition of NF-κB activation, we performed anti-p NF-κB antibody staining. This antibody was raised against a synthetic phosphopeptide corresponding to amino acids surrounding serine 276 of human NF-κB p65. Antibody staining in edar-EGFP transgenic medaka revealed that NF-κB was activated in mostly edar-positive regions (Fig. 8A,C,E). In afl larvae, however, NF-κB was not activated (Fig. 8B,D,F).

Figure 8.

Eda/Edar signaling activates NF-κB. Double staining with anti-GFP and anti-NF-κB antibodies against edar-EGFP transgenic line. A,B: GFP was expressed along the fin rays in wild-type (A) and afl larvae (B). C,D: In wild-type larvae, NF-κB has been activated at the distal tip of fin rays (C). However, in afl larvae, no activated NF-κB was evident (D). E,F: Merged images of wild-type (E) and afl (F) larvae showing the co-expression of edar and NF-κB. Note that NF-κB activation is mediated through Eda/Edar signaling. Scale bars = 100 μm.

To analyze the dynamics of fin ray formation, we performed long-time imaging of edar-EGFP and osterix-DsRed double-transgenic medaka. We observed these medaka for 15 hr starting at 5 dpf. At the start of imaging, osterix was only expressed in the fin ray 1 region, and edar was expressed in regions from 1 to 3 (Fig. 9A). In the fin ray 1 region, edar expression moved over time toward the edge of the fin fold along with expansion of osterix expression toward the edge of the fin fold (Fig. 9A–G). In the fin ray 2 region, no osterix expression was detectable at the onset of imaging (Fig. 9A,B). osterix expression became detectable 4 hr after the onset of imaging (Fig. 9C), and edar expression subsequently moved in the direction of fin ray elongation (Fig. 9D–G).

Figure 9.

In vivo imaging of edar-EGFP and osterix-DsRed double-transgenic medaka from 5 dpf to 6 dpf. A–G: Time-lapse imaging of the cells expressing edar with preceding mesenchymal differentiation into osteoprogenitor cells as seen by using edar-EGFP and osterix-DsRed double-transgenic medaka. Time is given corresponding to the start of the live-imaging observation. 0 h shows onset of imaging (A); and 15 h, the offset (G). B–F show images at 2, 4, 6, 8, and 10 h. To find the positional relation between body and expression sites, the bright-field image (middle views) was overlapped with the fluorescent image (upper views). Lower views are the magnified views of the upper views. At the initiation of this observation, edar was expressed in the fin ray 1, 2, and 3 regions, and osterix is only expressed in the fin ray 1 region (A). After 15 hr, edar was detected in the fin ray 1, 2, 3, and 4 regions, and osterix was detected in the fin ray 1, 2, and 3 ones (G). Focusing on the fin ray 1 region, the edar expression has changed its position toward the distal edge of fin fold, preceding osterix expression. Scale bars = 200 μm in upper and middle views, and 50 μm in lower views.

Furthermore, to investigate the behavior of the edar-positive cells at the cellular level, we performed high-resolution long-time imaging of an edar-EGFP and osterix-DsRed double-transgenic medaka for 5 hr with a confocal fluorescence microscope. The movie showed that edar-expressing cells migrated in the direction of fin ray elongation (Fig. 10A–F and G: Supp. Movie S1, which is available online). These cells changed their shape as having anterior–posterior polarity, indicating the typical phenotype of migrating cells.

Figure 10.

In vivo time-lapse imaging for migration of edar-positive cells. A–F: This series of images of a double-transgenic edar-EGFP and osterix-DsRed at 5 dpf was obtained by confocal microscopy. At the starting time (0 h), two edar-positive cells were observed (white arrowheads in “A”). As time passed, these cells migrated to the distal tip of the fin ray. All white arrowheads show the starting positions of the two cells at the initial starting time (0 h). We obtained this series of images taken over 5 hr at an interval of 2 min. B–F show images at 1, 2, 3, 4, and 5 h. G: A time-lapse movie was generated by connecting with this series of images (Supp. Movie S1). Focusing on the left edar-positive cell, the migration of this cell to the distal tip of fin ray can be seen. Scale bars = 5 μm.


In this study, we reported that afl fish showed a loss and change in the shape of their fin rays, deformation of the skull, a reduction in the number of teeth, and partial loss of scales. Previously, edar medaka mutants (rs-3) were characterized by reduced scale numbers (Kondo et al., 2001), and additionally they displayed a drastic loss of teeth in both of oral and pharyngeal dentitions (Atukorala et al., 2011). These abnormal phenotypes of teeth and scales are similar to those of the afl fish; however, in the fins of rs-3, the caudal fin is only lightly hypoplastic. It is likely that leaky expression of edar protein in rs-3 mutants induces fin ray formation, although a transposon inserted into the first intron of edar causes aberrant splicing. In zebrafish, various types of edar mutants have been reported (Harris et al., 2008). One of them has a molecular null mutation, showing loss of teeth, scales, and fins; and two others have missense mutations, resulting in amino acid changes near the C-terminal of the edar protein, show abnormalities of teeth and scales, although no abnormalities are observed in their fin rays. That zebrafish study showed that the phenotypic effect of the loss of edar is dose dependent and that scales and teeth are more sensitive to alterations in the level of edar transcription than fins (Harris et al., 2008). In our study, afl fish seemed to be the Eda null mutant because of the presumed protein structure of Eda and no activation of NF-κB, resulting the abnormal fin ray formation.

Expression experiments revealed Eda–Edar interaction between mesenchymal cells and epithelial cells. In the developing fin rays, eda and edar genes were expressed in mesenchymal and epithelial cells, respectively. We then detected cell clusters expressing bmp2b and runx2, which seemed to be osteoprogenitor cells. During cellular differentiation of osteoblasts, osteoprogenitor cells express runx2 in the presence of growth factors including bmp2, which expression is accompanied by mesenchymal condensation (Ducy et al., 1997; Stricker et al., 2002). According to our results, osteoprogenitor cells were localized adjacent to Edar-expressing cells and present among Eda-expressing ones. Several in vitro studies have demonstrated that Eda released from cells by proteolytic shedding binds to Edar through the TNF domain of the former (Elomaa et al., 2001; Schneider et al., 2001). Taken together, our results show that Eda protein released from mesenchymal cells bound to Edar in epithelial cells and that subsequently, activation of downstream transcription factors affected fin ray formation. afl larvae showed the normal positional relation between osteoprogenitor cells and edar-expressing epithelial cells without expression of eda at the initiation of fin ray formation. Therefore, fin rays of afl larvae were formed at the correct position. However, during fin ray development, the fin rays of afl mutants did not normally elongate and become thick. Our data revealed that this phenotype was due to the absence of the capacity of Edar-expressing cells to migrate in the correct direction. Time-course analysis of the expression patterns indicated that Edar-expressing cells in the wild-type larvae changed their position by moving toward the distal tip of the fin rays; in contrast, these cells in the afl larvae remained at the proximal region. Moreover, time-lapse imaging of edar-EGFP transgenic medaka showed that the Edar-positive cells migrated in the direction of fin ray elongation (see Fig. 10G: Supp. Movie S1). In afl larvae, this migration was not observed and may have been controlled by Hedgehog signaling. Drosophila studies have suggested that Hedgehog signaling is necessary in the germline for proper cell migration because misexpression of hedgehog induces germ cells to migrate to inappropriate locations (Deshpande et al., 2001). Likely, in afl larvae, the defect in shh expression may have inhibited cell migration. Several mouse studies have suggested that Shh is a transcriptional target of Eda signaling, which potently induces the expression of the shh gene (Schmidt-Ullrich et al., 2006; Pummila et al., 2007). Additionally, an earlier zebrafish study suggested that shh regulates cell proliferation, because the regenerating fin exposed to a hedgehog signaling inhibitor shows reduction in the proliferation of mesenchymal cells (Quint et al., 2002). Considering that cell proliferation in the afl larvae was decreased, shh may be a target gene of Eda signaling in fin ray formation.

Finally, we hypothesize the process of fin ray formation caused by Eda signaling. eda and edar genes are independently expressed in the mesenchymal cells and the epithelial cells, respectively, in the caudal region. The onset of fin ray formation is marked by the expression of Edar protein in epithelial cells. The binding of Eda to Edar causes activation of NF-κB, subsequently promoting the transcription of the downstream target genes, one of which may have been shh. The Edar-expressing cells then start migration in the direction of fin ray elongation with concomitant stimulation of mesenchymal proliferation, followed by mesenchymal condensation and differentiation of these mesenchymal cells into osteoprogenitor cells in the proximal region. In afl larvae, the onset of fin ray formation is marked by the expression of Edar in the epithelial cells; however, NF-κB is not activated because of loss of binding of Eda to Edar. Therefore, the Edar-expressing cells cannot migrate, remaining at the starting point. Because mesenchymal condensation and differentiation into osteoprogenitor cells subsequently occur, short and thin fin rays are thus formed.



The medaka (Oryzias latipes) strain Cab was used as the wild-type for all studies. The fish were maintained in an aquarium system with re-circulating water at 28.5°C. Under a photoperiod of 14-hr light and 10-hr darkness, medaka spawned daily. Naturally spawned eggs were harvested and incubated at 28°C. The eggs were maintained in medaka Ringer's solution (0.65% NaCl, 0.04% KCl, 0.011% CaCl2, 0.01% MgSO4, 0.01% NaHCO3) with continuous gentle shaking.

afl fish were obtained by mutagenesis using N-ethyl-N-nitrosourea (ENU), as previously described (Hibiya et al., 2009).

Embryos before hatching were dechorionated by use of hatching enzyme before using them. We prepared the hatching enzyme in the following way: five embryos at 6 dpf were immersed in 12 μl of medaka Ringer's solution and allowed to hatch by leaving them for approximately 2 hr in this solution at 28°C. After they had hatched out, the liquid was filtered and stored at −30°C. The liquid was used as a source of the hatching enzyme by applying it at the rate of 12 μl per eight embryos.

Positional Cloning

afl heterozygous fish was maintained on the southern Cab genomic background and mated with the wild-type northern HNI fish to generate F1 families. Embryos for the genetic mapping were obtained from inter-crosses of the F1 afl carriers. For establishment of the initial genetic linkage, we conducted bulked segregation analyses using expressed sequence tag genetic markers (M-marker 2003; Kimura et al., 2004). We narrowed down the genetic interval with additional sequence-tagged site (STS) markers (Naruse et al., 2000) and newly designed restriction-fragment-length polymorphism (RFLP) markers. The afl locus was genetically mapped within a interval of approximately 1.8 mega base pairs (mbps) on the medaka linkage group (LG) 10 by using 136 F2 homozygous afl larvae. The cDNAs of eda from wild-type and afl larvae were amplified by RT-PCR, and the sequences were verified. To directly confirm the linkage between afl locus and eda, we amplified a part of the eda cDNA, but did not find any restriction enzyme that could cleave either wild-type or afl allele. Next we used the polymorphic site in the first intron, amplified a part of eda genomic DNA by using the appropriate primers (5′-CCCAGCTGGAGTTACAGAGG-3′ and 5′-CTTTGCACAATGCAATGTCC-3′), and then digested the PCR fragments with the restriction enzyme BciT130 I (EcoRII, Mva I), which cleaved the wild-type, but not the afl, allele.

Skeletal Staining

Adult medaka fish were stained with Alizarin red and Alcian blue. All steps proceeded at room temperature. Before staining, the fish were fixed in formalin (10% formaldehyde/phosphate buffered saline [PBS]), washed in PBT (PBS/0.1% Tween) and distilled water, and placed in the Alcian blue solution (1% Alcian blue, 70% ethanol, 30% acetic acid) until the cartilages had been suitably stained. The fish were then destained in ethanol and washed in distilled water. They were next hydrolyzed by use of 1% trypsin in 30% saturated sodium tetraborate solution, washed 0.5% KOH, and stained with Alizarin red solution (20% Alizarin red saturated with ethanol, 0.4% KOH). Stained fish were washed in 0.5% KOH, which was substituted with glycerol by degrees, and stored in 80% glycerol.

Larval fish were fixed in 4% paraformaldehyde with 0.05 N sodium hydroxide at 4°C overnight, briefly washed in PBST, and stained with Alizarin red solution (4% Alizarin red, 0.5% KOH) at room temperature for 3 hr. After they were washed in 0.5% KOH, stained samples were stored in 80% glycerol.

In vivo skeletal staining was performed with ALC. Larvae were immersed in 0.005% ALC in distilled water for 2 hr. Thereafter, the stained larvae were rinsed with large volumes of distilled water, after which they were observed under a fluorescence microscope.

Whole-Mount RNA In Situ Hybridization and Tissue Section Analysis

WISH was performed with digoxigenin-labeled anti-sense RNA probes as previously described (Inohaya et al., 1995, 1999). Following WISH the specimens were rinsed in H2O and dehydrated in ethanol for 3 hr at 4°C. Using Technovit 8100 (Heraeus Kulzer, Wehrheim, Germany) resulted in an infiltration time that lasted for 10 min. The infiltration solution consisted of 50 ml of base-liquid Technovit 8100 and 0.3 g of hardener. For block preparation, 10 ml of infiltration solution and 0.3 ml of hardener II were mixed and agitated. The well-mixed embedding solution was poured into an embedding mold, and the specimens were properly placed in it. Then the mold was sealed hermetically with the cover foil and placed at 4°C. Afterward, the polymerized block was removed from the mold and fixed onto a block (Histoblock, Heraeus Kulzer) with an adhesive agent (Technovit 3040) and sectioned by using a microtome with a disposable knife. The section (5 μm) were stretched in water, mounted on a slide glass, dried, and stained with propidium iodide (PI) for nuclei.

Cell Proliferation Analysis

The cell proliferation analysis was conducted on larvae after having labeled them with BrdU for 8 hr at 28°C and subsequently performed in accordance with a previously described procedure (Yoshinari et al., 2009).

Whole-Mount Immunofluorescence

Larvae were fixed in 4% paraformaldehyde overnight at 4°C. After having been washed in PBSTx (PBS plus 0.1% Triton-X100), the larvae were dehydrated in a graded series of methanol (25–50–75%)/PBSTx and stored in 100% methanol at −20°C. They were rehydrated in a graded series of methanol (75–50–25%)/PBSTx and washed three times for 15 min each time in PBSTx. Following blocking with 2 mg/ml BSA and 2% lamb serum in PBSTx for 2 hr, the larvae were incubated with primary antibodies at 4°C overnight. The following antibodies were used: rabbit anti- NF-κB p65 (pSer276) polyclonal antibody (AnaSpec 29775-025; 1:1,000) and chicken anti-GFP polyclonal antibody (abcam ab13970; 1:1,000). The larvae were then washed six times in PBSTx for 15 min each time, and incubated with Alexa568-conjugated anti-rabbit IgG (1:1,000) and Alexa488-conjugated anti-chicken IgY (1:1,000) in PBSTx at 4°C overnight. After having been washed six times in PBSTx for 15 min, the larvae were stored in 80% glycerol and whole-mounted on glass slides for observation with a confocal microscope.

Transgenic Rescue

We used a 13-kb genomic fragment encompassing the eda gene as a transgene. This transgene had a 3-kb eda promoter region and a 7-kb region of genomic DNA, which lacked 6 kb of the first intron region. The genomic fragment was amplified in three parts (promoter-first exon fragment, second exon-XbaI fragment, and XbaI-3′ half fragment) from a BAC clone (NBRP Medaka; http://www.shigen.nig.ac.jp/medaka/) by using the appropriate primers (promoter-first exon fragment, 5′-CTCGAGCACATCGTCTGCCATTAGCC-3′ and 5′-CTGTCTGTTCTCTCCGCTGC-3′; second exon-XbaI fragment, 5′-GCAGCGGAGAGAACAGACAGGAAAGATTACAGTACAGAAATGACGGC-3′ and 5′-GCGGCCGCTTTAGAACTGACAGCACGATCC-3′; and XbaI-3′ half fragment, 5′-GGAGACAGAAATTGGTGTGGG-3′ and 5′-GCGGCCGCAGACATGTTCAGCTCTCGGC-3′). The amplified fragments were cloned into the TA cloning vector. The second exon-XbaI fragment primer was designed for containing the end sequence of the promoter-first exon fragment. These two fragments were assembled through overlapping their ends by PCR. We then digested this assembled fragment with XhoI and NotI, and subcloned it into the XhoI/NotI sites of an I-SceI backbone vector, which contains two I-SceI sites (Thermes et al., 2002). The XbaI-3′ half fragment, digested with XbaI and NotI, was cloned into XbaI/NotI sites of the assembled fragment. We then subcloned a reporter gene, which expresses EGFP under the control of a zebrafish αA-crystallin promoter (Kurita et al., 2003), into the Not I site on the 3′ side of the 10-kb transgene. This plasmid was digested with I-Sce I, and the fragments (20 ng/μl) were injected into the cytoplasm of one-cell stage embryos. After the injection, the embryos showing a transiently strong expression of the exogenous gene in their eyes were allowed to grow to adulthood. We then checked the EGFP expression in the next generation, and picked an embryo with stable integration of the injected construct as the transgenic line.

edar-EGFP Transgenic Medaka

We amplified a genomic fragment encompassing the 3-kb edar promoter region from the medaka genomic DNA by using appropriate primers (5′-CTCGAGAACGATGTATTAAAGTCTGTCTG-3′ and 5′-CCGCGGTTCCTTTCCCTGCTGGGGCAG-3′). The amplified fragment was cloned into the TA cloning vector. We then digested the amplified fragment with XhoI and SacII, and inserted the I-SceI backbone vector, which was then inserted into the EGFP/SV40polyA in pBSKI2 vector. This plasmid was digested with I-SceI, and fragments (20 ng/μl) were injected into the cytoplasm of embryo at one-cell stage. The embryos showing a transiently strong expression of the exogenous gene were allowed to grow to adulthood. We then checked the EGFP expression in the next generation, and picked an embryo having stable integration of the injected construct. To generate double-transgenic larvae, stable osterix-DsRed carriers were crossed with edar-EGFP transgenic medaka.

osterix-DsRed Transgenic Medaka

We cloned a 4.2-kb DNA fragment encompassing the osterix promoter region (Inohaya et al., 2010) into SacII/XhoI sites of the pDsRed2-1 vector (CLONTECH). This construct was designated as pDsRed-osterix. The circular pDsRed-osterix in medaka Ringer's solution was injected at a concentration of 2.5 ng/μl into the cytoplasm of one-cell stage embryos. For the establishment of transgenic lines, we used the fertilized eggs from the medaka Cab line.

Fluorescence In Vivo Imaging

Larvae used for imaging were anesthetized by using 3-aminobenzoic aced ethyl ester, immersed in 1% low-melting temperature agarose, and mounted on their lateral sides in glass dishes. Time-lapse analysis was carried out with an Olympus FV1000 confocal microscope equipped with a ×20 water objective and a Leica AF6000 fluorescence microscope equipped with a ×1.0 objective for the 488-nm and 543-nm laser lines.


We thank Dr. A. Kawakami for many helpful discussions. This work was supported by grants-in-aid for scientific research from the Japan Ministry of Education, Culture, Sports, and Technology of Japan.